Urban Economics Arthur OSullivan

J MjHWjjyj'
Digitized by the Internet Archive

in 2012
http://archive.org/details/urbaneconomicsOOosul_0
Ml ECONOMIC
m HOHOHIO
Fourth Edition
Arthur S u 1 1 i v a n
DtPARTMENT OfCcONOMtCS
Oregon State University
Irwin
McGraw-Hill
Boston Burr Ridge, IL Dubuque, IA Madison. WI New York San Francisco
St. Louis Bangkok Bogota Caracas Lisbon London Madrid Mexico City
Milan New Delhi Seoul Singapore Sydney Taipei Toronto
McGraw-Hill Higher Education &y
\ I Hvision of The McGraw-Hill Companies
Urban Economics
Copyright © 2000. 1996. 1992. 1990 by The McGraw-Hill Companies. Inc. All rights reserved. Printed
in the United States of America. Except as permitted under the United States Copyright Act of 1976.
no part of this publication may be reproduced or distributed in any form or by any means, or stored in a
data base or retrieval system, without the prior written permission of the publisher.
This book is printed on acid-free paper:
domestic 12 3 4567890 DOC DOC 9098765432109

international 12 3 4 5 6 7 8 9 DOC DOC 90987654 3 2109
ISBN 0-25b-2b331-D
Editor-in-chief: Michael \Y. Junior

Publisher: Gary Burke
Executive editor: Paul Shensa
Editorial coordinator: Katherine Mattison
Editorial assistant: Shoshannah Flach
Senior marketing manager: Nelson W. Black
Project manager: Beatrice WUamder
Senior production supervisor: Richard DeVitto
Senior designer: Sabrina Dupont
Cover designer: Susan Newman
Cover art: Richard Diebenkom. Cityscape 1 . 1963. oil on canvas. 60 1/4 in. x 50 1/2 in.. San Francisco
Museum oj Modern Art
Compositor: Interactive Composition Corporation
Typeface: Times Roman
Printer: RR Donnelley & Sons Company
Library of Congress Cataloging-in-Publication Data
O'Sullivan. Arthur.
Urban economics / Arthur O'Sullivan.
p. cm.
Includes bibliographical references and index.
ISBN 0-256-2633 1
-0 talk, paper)
Urban economics.
I. I. Title.
HT321.088 1999
330.9 1 73'2—dc21 99-30311
CIP
International Edition ISBN 0-07-116974-1

(op\ right © 2000. Exclusi\e rights b\ The McGraw-Hill Companies. Inc.. for manufacture and export.
This hook cannot be re-exported from the country to which it is consigned by The McGraw-Hill
Companies. Inc. The International Edition is not available in North America.
http://www.mhhe.coni
To Professor Edwin whose Urban Economics (First Edition) remains my
S. Mills,
favorite textbook of all The book is full of Professor Mills' remarkable
time.
insights into urban phenomena and is written in a way that gave me and my fellow
students many opportunities to think for ourselves. Thanks, Ed.
About the Author
Arthur O'Sullivan is a professor of Economics at Oregon

State University. After receiving his B.S. in Economics at the
University of Oregon, he spent two years in the Peace Corps,
working with city planners in the Philippines. He received his
Ph.D. in Economics from Princeton University in 1981 and
^^ 4 ^^^^ then spent eleven years at the University of California. Davis.

^k i A where he won several teaching awards. At Oregon State Uni-
I ^k versity, he teaches introductory and intermediate microeco-
BaJI nomic and urban/regional economics. Professor O" Sullivan's

research explores economic issues concerning urban land use,
many economics
environmental protection, and public finance. His articles appear in
journals, including Journal of Urban Economics, Journal of Environmental Eco-
nomics and Management, National Tax Journal, and Journal of Public Economics.
REf ACE
(^7~ his book uses economic analysis to explain why cities exist, where they
c/ develop, how they grow, and how different activities are arranged within
cities. It also explores the economics of urban problems such as poverty, inadequate
housing, segregation, congestion, pollution, and crime.
The text is designed for use in undergraduate courses in urban economics and urban
affairs. It could also be used for graduate courses in urban planning, public pol-
icy,and public administration. All of the economic concepts used in the book are
covered in the typical intermediate microeconomics course, so students who have
completed such a course will be able to move through the book at a rapid pace. For
students whose exposure to microeconomics is limited to an introductory course or —
who could benefit from a review of the concepts covered in an intermediate microe-
—
conomics course I have provided an Appendix (Tools of Microeconomics) that
covers the essential concepts.
The book covers more topics than the typical urban economics text, giving instructors
several options for a one-semester course in urban economics. A course that em-
phasizes interurban location analysis would cover all five chapters in Part I (Market
Forces Development of Cities), while other courses might omit some of these
in the
chapters. A course emphasizing intraurban location analysis would cover all five
chapters in Part II (Land Rent and Urban Land-Use Patterns), while other courses
might omit Chapter 9 (General Equilibrium Land Use) or Chapter 1 1 (Land Use
Controls and Zoning). A course emphasizing urban problems would cover all four
chapters in Part III (Poverty and Housing) and most of the chapters in Part V (Ur-
ban Transportation) and Part VI (Education and Crime). The three chapters on local
government (in Part IV) could be omitted in a course emphasizing traditional urban
problems, but would be an integral part of a course emphasizing urban public finance.
What's New in the Fourth Edition?

There are many changes in the book for the Fourth Edition. The tables and charts
have been thoroughly updated with the most recent data. Throughout the book, I've
vii
viii Preface
added insights and facts from recent theoretical and empirical research. The policy
analysis has been updated to reflect recent changes in public policy and refinements
in the economic analysis of policy alternatives.
There are a couple of organizational changes. The material on poverty and public
policy has been thoroughly rewritten, with one chapter on the effects of racial seg-
regation in housing on poverty in the central city. A second chapter explores some
other reasons for poverty and discusses antipoverty policies, including the recent
overhaul of the welfare system, which requires most welfare recipients to work in
exchange for time-limited assistance. The second poverty chapter also has a new
section that explores the reasons for homelessness. I also regrouped the chapters in
the old Parts III and IV into four parts:
• Part III: Poverty and Housing

• Part IV: Local Government
• Part V: Urban Transportation
• Part VI: Education and Crime
Here are some highlights of the most important changes in the book, organized by
the six parts of the book.
Part I: Market Forces in the Development of Cities

• Incorporates many insights from the new field, "Economic Geography""
• Includes a formal model of labor-market pooling as an agglomerative
economy (Chapter 2
• Evaluates the prediction that innovations in telecommunications will cause
cities to disappear as workers disperse to work in "electronic cottages*' in rural
areas (Chapter 2
• Explores the effects of localization economies on the location patterns of firms
and the spatial concentration of industries (Chapter 3).
• Includes two new case studies of firms' location decisions: Japanese

automobile firms, and the Mexican garment industry (Chapter 3)
• Explores the forces responsible for the development of large primary cities in
developing countries, including trade, trade restrictions, infrastructure

investment, and politics (Chapter 5)
• Incorporates agglomerative economies into the model of labor demand and
supply (Chapter 6)
Part II: Land Rent and Urban Land-Use Patterns
• Explores the effects of Pittsburgh's graded property-tax policy (under which

land is taxed at a higher rate than structures) on urban development (Chapter 7)
• Contrasts the pattern of income segregation in the United States with the
pattern in other countries and explores the effects of cultural amenities on the
location choices of high-income households (Chapter 8)
Preface IX
• Includes new material on the development of employment subcenters

(Chapter 10)
• Explores the role of suburban subcenters in the metropolitan economy,
focusing on the relationship between subcenters and the central city
(Chapter 10)
• Documents the development of "edge cities" and explores the forces behind
their development (Chapter 10)
• Explains how the New Jersey legislature weakened the Mt. Laurel
exclusionary-zoning ruling by modifying the court's quotas for low-income
housing and allowing communities to buy and sell their quotas (Chapter 11)
Part III: Poverty and Housing

• Emphasizes the spatial aspects of poverty and the problem of poverty in
central cities, focusing on the effects of segregation in housing, segregation in
schools, differences in educational achievement, and the spatial mismatch

(Chapter 12)
• Includes a new section on the problem of homelessness (Chapter 13)
• Discusses the overhaul of the welfare system, including the replacement of
Aid to Families with Dependent Children (AFDC) with a system of block
grants to states that emphasize "workfare" (Chapter 13)
• Discusses recent changes in housing policy —
the shift away from supply-side
policies in favor ofdemand-side policies such as housing vouchers and the —
implications for the urban poor (Chapter 15)
Part IV: Local Government
• Includes the latest data on government taxes and expenditures

• Explores the connection between poverty and the fiscal problems of
central-city governments (Chapter 16)
• Discusses metropolitan government as a possible response to
interjurisdictional spillovers (Chapter 16)
• Explains why welfare reform — the switch from matching grants to block
grants — will reduce welfare spending (Chapter 18)
Part V: Urban Transportation

• Summarizes recent experiences with congestion pricing in the United States
and other countries (Chapter 19)
• Discusses a possible role for commercial paratransit — to fill the gap in the
urban transportation system between solo-rider taxis and large public buses
(Chapter 20)
• Provides new data on the costs of alternative transit systems (buses, light rail,
and heavy rail) (Chapter 20)

Preface
• Explores the efficacy of HOV lanes and HOT (high-occupancy toll) lanes in
controlling congestion (Chapter 20)
• Discusses the connection between land use and transportation and the reasons
for the apparent weakening of the connection (Chapter 20)
Part VI: Education and Crime

• Describes inequalities in education spending and the efficacy of equalization
programs (Chapter 2 1
• Describes the effects of equalization programs on central-city schools
(Chapter 21)
• Discusses the possible effects of education vouchers on segregation (with
respect to income, race, and ability) and achievement among students of
different abilities (Chapter 21)
• Discusses the sensitivity of crime rates to changes in police resources, arrest
ratios, imprisonment rates, and the wages of low-skill workers (Chapter 22)
• Explores the effects of the prison system on crime rates, focusing on relative
importance of incapacitation and deterrence (Chapter 22)
• Discusses the benefits and costs of three-strike laws (Chapter 22)
• Explores why crime rates are higher in large cities (Chapter 22)
• Explains how criminal activity changes when a criminal's legal status changes
from juvenile to adult (Chapter 22)
Acknowledgments
I am greatly indebted to my
two mentors in urban economics. As an undergraduate
at the I was taught by Ed Whitelaw, whose enthusiasm for
University of Oregon,
urban economics is apparently contagious. He used a number of innovative teaching
techniques that made economics understandable, relevant, and even fun. As a gradu-
ate student at Princeton University, I was taught by Edwin Mills, one of the founding
fathers of urban economics. He refined my mathematical and analytical skills and
also provided a steady stream of perceptive insights into urban phenomena. 1 hope
that some of what I learned from these two outstanding teachers is reflected in this
book.
I am also indebted to many people who read the book and suggested ways to im-
prove the coverage and the exposition. In particular I would like to thank those
instructors who participated in surveys and reviews that were indispensable in
the development of the Fourth Edition of Urban Economics. The appearance of
their names does not necessarily consitute their endorsement of the text or its
methodology.
Preface \\
Randall Bartlett Vijay Mathur

Smith College Cleveland State University
Charles Berry Steven Raphael

University of Cincinnati University of California —San Diego
Bradley Braun Stuart Rosenthal
University of Central Florida Virginia Polytechnic Institute and State
University
Jerry Carlino
University of Pennsylvania William Schaffer
Georgia Institute of Technology
Ed Coulson
Pennsylvania State University Steve Soderlind
Saint Olaf College
David Figlio
University of Oregon Mary Stevenson
University of Massachusetts —Boston
Andrew Gold
Trinity College John Yinger
Syracuse University
Kenneth Lipner
Florida International University
Arthur O'Sullivan
& ONTENTS in BRIEF
1 Introduction 1
Market Forces in the Development of Cities 15
2 Why Do Cities Exist? 17

3 Where Do Cities Develop? 47
4 The History of Western Urbanization 85
5 How Many Cities? 105
6 Urban Economic Growth 139
ART II
Land Rent and Urban Land-Use Patterns 179

7 Introduction to Land Rent and Land Use 181
8 Land Use Monocentric City 205
in the
9 General-Equilibrium Land Use 249
10 Suburbanization and Modern Cities 267
11 Land-Use Controls and Zoning 303
Part III
Poverty and Housing 333

12 Segregation and Poverty in the Central City 335
13 Other Dimensions of Poverty 351
14 Why Is Housing Different? 367
15 Housing Policies 409
XIV Contents in Brief
Part IV
Local Government 447

16 Overview of Local Government 449
17 Voting with Ballots and Feet 481
18 Local Taxes and Intergovernmental Grants 503
Urban Transportation 543

19 Autos and Highways 545
20 Mass Transit 585
'ART VI
Education and Crime 619
21 Education 621
22 Crime and Punishment 655
Appendix 699
Index 723
NTENTS
(2
1 Introduction 1
What Is Urban Economics? 2

Market Forces in the Development of Cities
Land Rent and Land Use within Cities 3
Spatial Aspects of Poverty and Housing 4
Local Government Expenditures and Taxes
What Is an Urban Area? 7
Census Definitions 7
Definitions Used in This Book 12
'ART I
Market Forces in the Development of Cities 15
2 Why Do Cities Exist? 17
A Model of a Rural Region 17

Comparative Advantage 19
Trade and Transportation Cost 19
Scale Economies in Transportation and Trading Cities 20
Economies in Production 21
Internal Scale
Scale Economies and the Cloth Factory 21
Scale Economies and an Urban Area 23
Examples of Urban Development 24
The English Woolen Industry 24
The Sewing Machine and U.S. Urbanization 25
xv
XVI Contents
Agglomerative Economies Production 26

in
Localization Economies 26
Urbanization Economies 32
Empirical Estimates of Agglomerative Economies 33
Innovations in Telecommunications and the Future of Cities 34
Agglomerative Economies in Marketing: Shopping Externalities 36
Imperfect Substitutes 36
Complementary Goods 38
Retail Clusters 38
3 Where Do Cities Develop? 47
Commercial Firms and Trading Cities 47

Transfer-Oriented Industrial Firms 48
Resource-Oriented Firms 49
Market-Oriented Firms 5
Intermediate Location 53
Implications for the Development of Cities 55
The Principle of Median Location 55
Location Choice with Multiple Markets 55
Median Location and Growth of Cities 57
Transshipment Points and Port Cities 57
Median Location and Cities 59
Finns Oriented toward Local Inputs 59
Energy Inputs 59
Labor 60
Amenities, Migration, and Labor Costs 61
Intermediate Inputs: Localization and Urbanization Economies 61
Local Public Services and Taxes 63
Several Local Inputs 65
Location Choices and Land Rent 65
Transport Costs versus Local-Input Costs 66
Trade-off between Transport Costs and Local-Input Costs 67
Change in Orientation: Decrease in Transport Costs 68
Location Orientation and Growth Patterns 69
Case Studies and Empirical Studies 70
The Semiconductor Industry 70
Computer Manufacturers 70
Japanese Automobile Firms 7
Mexican Garment Industry 71
A Case Study: The Location of GM's Saturn Plant 72
The Relative Importance of Local Inputs: Empirical Results 73
Effects of Taxes and Public Services on Business Location Decisions 75
Effects Wages and Unions on Business Location Choices
o\' 76
The Role of Government in the Location of Cities 77
Contents
4 The History of Western Urbanization 85
The First Cities 85

The Defensive City 86
The Religious City 86
The Defensive and Religious City 87
Greek Cities 87
Roman Cities 88
Feudal Cities 89
Mercantile Cities 91
Centralization of Power 91
Long-Distance Trade 92
Urbanization during the Industrial Revolution 92
Agriculture 92
Manufacturing 93
Intercity Transportation 93
Intracity Transportation 93
Construction Methods 94
Urbanization in the United States 94
From Commercial to Industrial Cities 96
The Postwar Era 97
Metropolitan versus Nonmetropolitan Growth 98
Population Losses in Some Metropolitan Areas 99
The Postindustrial City 100
5 How Many Cities? 105
The Analysis of Market Areas 106

Pricing with a Monopolist 106
Entry and Competition 108
Efficiency Trade-offs 109
Market Areas 110
Determinants of the Market Area: Fixed Demand 1 10
Market Areas and the Law of Demand 1 17
The Demise of Small Stores 1 1
Market Areas of Different Industries 1 19
Central Place Theory 119

A Simple Central Place Model 1 20
Relaxing the Assumptions 123
Other Types of Industry 124
Central Place Theory and the Real World 125
Urban Giants: The Puzzle of the Large Primary City 129
The Role of Trade 129
The Effect of Trade Restrictions 130
The Role of Infrastructure and Politics 131
XV111 Contenis
6 Urban Economic Growth 139
The Urban Labor Market and Economic Growth 140

The Demand for Labor 140
The Supply Curve 144
Demand and Supply Shifts 146
Equilibrium Effects of
Agglomeration Economies and Urban Growth 147
Public Policy and Economic Growth 149
Subsidy Programs 149
Urban Growth and Environmental Quality 150
Predicting Economic Growth 153
The Economic Base Study 153
Input-Output Analysis 157
Limitations of Economic Base and Input-Output Studies 160
Benefits and Costs of Employment Growth 162
Who Gets the New Jobs? 1 63
Employment Growth and Real Income per Capita 165
Other Benefits and Costs of Employment Growth 166
Composition of Changes in Employment 167
ART II
Land Rent and Urban Land-Use Patterns 179
7 Introduction to Land Rent and Land Use 181
Land Rent and Fertility 183

Competition and Land Rent 185
Land Rent and Public Policy 85 1
Land Rent and Accessibility 187

Linear Land-Rent Function 1 88
Factor Substitution: Convex Land-Rent Function 189
Decrease in Transport Costs and Land Rent 193
Two Land Uses 94 1
Market Interactions 196

The Corn Laws Debate 196
Housing Prices and Land Prices 196
The Single Tax 198
Urban Land Taxation in Pittsburgh 200
8 Land Use in the Monocentric City 205
Commercial and Industrial Land Use 206

The Bid-Rent Function of Manufacturers 206
The Bid-Rent Function of Office Firms 209
Land Use in the Central Business District 21
Location of Retailers 2 1
Contents XIX
Residential Land Use 215

The Housing-Price Function 216
The Residential Bid-Rent Function 219
Residential Density 223
Land Use in the Monocentric City 224
Relaxing the Assumptions 226
Changes in Commuting Assumptions 226
Variation in Tastes for Housing 227
Spatial Variation in Public Goods and Pollution 229
Income and Location 230
Trade-off between Land and Commuting Costs 230
Other Explanations 234
Income and Location in Other Countries 235
Income and the Residential Bid-Rent Function 235
Policy Implications 237
Empirical Estimates of Rent and Density Functions 237
Estimates of the Land-Rent Function 237
Estimates of the Population-Density Function 238
9 General-Equilibrium Land Use 249
General-Equilibrium Conditions 249

Numerical Examples of General-Equilibrium Analysis 251
Initial Equilibrium 251
Increase in Export Sales 253
A Streetcar System 256
The General-Equilibrium Effects of the Property Tax 259
10 Suburbanization and Modern Cities 267
Suburbanization Facts 267

Suburbanization of Manufacturing 269
The Intracity Truck 269
The Intercity Truck 271
The Automobile 272
Single-Story Plants 272
Suburban Airports 272
The Suburbanization of Population 272
Increase in Real Income 274
Decrease in Commuting Cost 274
Central-City Problems 276
Following Firms to the 277
Suburbs
Public Policy and Suburbanization 278
Gentrification: Return to the City? 279
Suburbanization of Retailers 279
Following the Consumers 279
XX Contents
The Automobile 280

Population Growth 280
Suburbanization of Office Employment 28
Suburbanization and Clustering of Offices 28
Communications Technology and Office Suburbanization 282
Suburban Subcenters and Modern Cities 284
Subcenters in Los Angeles 284
Subcenters in Houston 287
Subcenters in Metropolitan Chicago 287
Edge Cities 289
Subcenters and City Centers 290
Land Use in the Modern City: Urban Villages? 292
11 Land-Use Controls and Zoning 303
Controlling Population Growth 303

Urban Growth Boundary or Urban Service Boundary 303
Building Permits 305
Land-Use Zoning: Types and Market Effects 308
Nuisance Zoning 308
Fiscal Zoning 313
Design Zoning 316
A City without Zoning? 320
The Legal Environment of Land-Use Controls 32
Substantive Due Process 321
Equal Protection 322
Just Compensation 323
Part III
Poverty and Housing 333
12 Segregation and Poverty in the Central City 335
Poverty Facts 335

Who and Where Are the Poor? 335
The Working Poor 337
Poverty in the Central City 338
Residential Segregation 338
Segregation Facts 339
Reasons for Racial Segregation 340
Consequences of Segregation 341
Segregation. Education, and Poverty 342
The Spatial Mismatch 343
Poverty in the Central City and Public Policy 346
Contents XXI
13 Other Dimensions of Poverty 351
Economic Growth and Poverty 35

Evidence Concerning the Benefits of Growth 352
Changes in the Relationship between Growth and Poverty 352
Discrimination in Labor Markets 353
Theories of Discrimination 353
Facts on Earnings and Discrimination 354
Demographic Change: Female-Headed Households 356
Poverty and Public Policy 357
Employment and Job-Training Programs 357
Cash Payments and Payments in Kind 357
Welfare Reform 358
The Homeless 360
14 Why Is Housing Different? 367
Heterogeneity and Immobility 368

The Price of Housing: The Hedonic Approach368
Choosing a Dwelling 370
Segmented but Related Markets 37
Neighborhood Effects 372
Durability of Housing 372
Deterioration and Maintenance 373
The Retirement Decision 374
Abandonment and Public Policy 375
Durability and Supply Elasticity 377
The Filtering Model of the Housing Market 378
Why Do the Poor Occupy Used Housing? 378
Construction Subsidies 380
Summary: Filtering and Public Policy 382
The High Cost of Housing 383
The Cost of Renting 383
The Cost of Homeownership 384
Indifference between Renting and Owning? 384
Taxes and Housing 385
Inflation and Homeownership 389
Large Moving Cost 390
Income and Price Elasticities 391
Permanent Income and Income Elasticity 391
Price Elasticities 392
Racial Discrimination and Segregation 392
Housing Prices with Fixed Populations 393
Population Growth and Broker Discrimination 395
Neighborhood Transition 396
XX11 Contents
15 Housing Policies 409
Conditions of Housing and Neighborhoods 409

Housing Conditions and Rent Burdens 410
Neighborhood Conditions 411
Supply-Side Policy: Public Housing 413
Public Housing and Housing Consumption 413
Public Housing and Recipient Welfare 414
Living Conditions in Public Housing 418
The Market Effects of Public Housing 419
Supply-Side Policy: Subsidies for Private Housing 420
Demand-Side Policy: Consumer Subsidies 421
Rent Certificates: Section 8 421
Housing Vouchers 423
Market Effects of Consumer Subsidies 424
The Experimental Housing Allowance Program 427
Supply versus Demand Policies 428
Community Development and Urban Renewal 429
Urban Renewal 429
Recent Community Development Programs 430
Rent Control 43
Market Effects of Rent Control 432
Distributional Effects of Rent Control 434
Deterioration and Maintenance 435
Housing Submarkets 436
Controlling the Supply Response to Rent Control 437
Alternatives to Rent Control 439
Part IV
Local Government 447
16 Overview of Local Government 449
Local-Government Facts 449

The Role of Local Government 452
Natural Monopoly 453
Externalities 455
Characteristics of a Local Public Good 457
The Optimum Quantity of Good 458
a Local Public
Federalism and the Optimum Level of Government 460
Externalities versus Demand Diversity 461
Scale Economies versus Demand Diversity 462
Metropolitan Consolidation 463
Prevention of Suburban Exploitation 463
Exploitation of Scale Economies 463
Contents XX111
Equalization of Tax Rates 463

Equalization of Spending 464
Alternatives to Metropolitan Consolidation 465
Interjurisdictional Spillovers and Metropolitan Government 466
The Growth in Local Spending 466
Increases in Labor Cost: The Baumol Model 466
Other Reasons for Rising Labor Cost 468
Increases in Quantity 469
Taxpayer Revolt: Expenditure Limits 470
Fiscal Problems of Central Cities 471
Poverty and Fiscal Problems 472
Index of Fiscal Health 472
17 Voting with Ballots and Feet 481
Voting with Ballots: The Median-Voter Rule 481

Benefit Taxation 483
Budget Elections and the Median Voter 484
The Median Voter in a Representative Democracy 485
Implications of the Median-Voter Rule 486
Qualifications of the Median-Voter Model 488
Voting with Feet: The Tiebout Model 488
The Simple Tiebout Model 489
Efficiency Effects of the Tiebout Process 489
Tiebout Model with Property Taxes 490
Capitalization and the Tiebout Model 492
Evidence of Shopping and Sorting 492
Critique of the Tiebout Model 493
Metropolitan Consolidation and the Tiebout Model 494
Inefficiency from Spillovers 495
Effects of Consolidation 495
18 Local Taxes and Intergovernmental Grants 503
Who Pays the Residential Property Tax? 503

The Land Portion of the Property Tax 504
The Improvement Tax: The Traditional View 505
The New View of the Property Tax 507
Changing the Assumptions 509
Summary: Who Pays the Residential Property Tax? 512
The Local Effects of a Local Tax 512
The National Effects of a Local Tax 5 1
The Effects of a National Tax 5 1
Summary of Incidence Results 513

The Tax on Commercial and Industrial Property 514
The General Property Tax 514
Differential Tax Rates 515
XXIV Contents
The Tiebout Model and the Property Tax 516

Administering the Property Tax 5 1
Taxes on Income and Retail Sales 519

Municipal Sales Tax 519
Municipal Income Tax 521
Intergovernmental Grants 522
Why Intergovernmental Grants? 523
Unconditional Grants 524
Lump-Sum Conditional Grants 526
Matching Grants: Open-Ended 527
Matching Grants: Closed-Ended 529
Block Grants 530
Summary: The Stimulative Effects of Grants 530
Welfare Reform: Matching Grants to Block Grants 531
Capitalization and the Distributional Effects of Grants 533
Grants to Low-Income Communities 533
Grants to High-Tax Communities 534
Tax Expenditures: Implicit Grants 534
Deductibility of Local Taxes 534
Exemption of Interest on Local-Government Bonds 536
fART V
Urban Transportation 543
19 Autos and Highways 545
Congestion: Equilibrium versus Optimum Traffic Volume 546

The Demand for Urban Travel 547
The Private and Social Costs of Travel 548
Equilibrium versus Optimum Traffic Volume 549
The Policy Response: Congestion Tax 550
Benefits and Costs of Congestion Taxes 550
Peak versus Off-Peak Travel 552
Estimates of Congestion Taxes 552
Implementing the Congestion Tax 553
Experiences with Congestion Pricing 554
Alternatives to a Congestion Tax: Taxes on Auto Use 555
Gasoline Tax 556
Parking Tax 556
Alternatives to Auto Taxes 557
Capacity Expansion and Traffic Design 557
Subsidies for Transit 559
Highway Pricing and Traffic Volume in the Long Run 561
Average Total Cost for Two-Lane and Four-Lane Highways 561
Long-Run Average Cost and Marginal-Cost Curves 563
Contents xxv
Optimum Volume and Road Width 563

Congestion Tolls Pay for the Optimum Road 563
Who Pays for Highways? 565
Congestion and Land-Use Patterns 565
565
Partial-Equilibrium Effects
General-Equilibrium Effects 567
Autos and Air Pollution 569
The Economic Approach: Effluent Fees 570
Transit Subsidies 571
Auto Safety 571
The Effects of Safety Regulations 572
Small Cars and Safety 574
20 Mass Transit 585
Mass Transit Facts 585

Choosing a Travel Mode: The Commuter 589
An Example of Modal Choice 589
Transit Service and Modal Choice 592
High-Occupancy Vehicle (HOV) Facilities 593
Choosing a System: The Planner's Problem 594
Cost of the Auto System 595
Cost of the Bus System 596
Cost of BART 596
System Choice 597
Light Rail 597
Subsidies for Mass Transit 599
Justification for Transit Subsidies 599
Reasons for Transit Deficits 601
Policy Reform: Contracting and Paratransit 602
Contracting for Transit Services 603
Deregulation and Paratransit 604
British Experience with Deregulation 606
Transportation and Land-Use Patterns 607
Gasoline Prices 607
Fixed-Rail Systems and Land Use 608
'ART VI
Education and Crime 619
21 Education 621
The Education Production Function 622

Inputs and Outputs of the School 622
The Coleman Report 625
XXVI Contents
Other Studies of the Production Function 626

Alternative Approach: School Resources and Market Earnings 628
The Tiebout Process and Achievement 628
Inequalities in
Spending Inequalities and Public Policy 629
The Legal Mandate to Equalize Expenditures 630
Foundation Grants 632
Guaranteed Tax Base or Power Equalization 634
Effects of Equalization Plans on Spending Inequalities 636
Education Finance Reform and Central-City Schools 637
Desegregation Policy 637
Segregation and Public Policy 637
Court-Ordered Desegregation 639
Moving to Another School District 639
Switching to Private Schools 640
Tiebout-Based Segregation 643
Subsidies for Private Education: Education Vouchers 644
22 Crime and Punishment 655
Crime Facts 655

The Victims of Crime 657
Crime and the Price of Housing 657
The Costs of Crime 658
The Rational Criminal and the Supply of Crime 659
A Numerical Example of Burglary 659
The Crime-Supply Curve 661
Optimum Amount of Crime 662
Victim Cost and Prevention Cost 662
Marginal Cost and Variation in Optimum Crime Levels 664
Crime Prevention Activities 666
Increasing the Value of Legal Opportunities 666
Victims as Crime Fighters 667
The Police 669
The Court System 670
The Principle of Marginal Deterrence 673
The Prison System 676
Why Are Crime Rates Higher in Big Cities? 680
Illegal Commodities and Victimless Crime 682
Product Illegality and Production Costs 682
Product Illegality and Price 683
Drugs and Property Crime 684
Contents XXV11
Tools of Microeconomics 699
Model of Supply and Demand 701

The Market Demand Curve 701
The Market Supply Curve 703
Elasticities of Demand and Supply 704
Market Equilibrium 705
Shortage or Surplus 706
Market Effects of Changes in Demand or Supply 707
Theory of Consumer Choice 708
Constructing an Indifference Curve 709
An Indifference Map 712
The Budget Set and Budget Line 712
Maximizing Utility 715
Changes in Price and Income 717
Theory of Worker's Choice 7 1
Indifference Curves for Worker 718

Budget Line for Worker 718
Maximizing Utility 719
Theory of Producer Choice 7 1
Marginal Decision Making 721
Marginal Benefits and Costs 721
Using the Marginal Principle 721
Index 723
Chapter 1
£T NTRODIKTION
Cities have always been the fireplaces of civilization, whence light and heat
radiated out into the dark.
Theodore Parker
I'd rather wake up in the middle of nowhere than in any city on earth.
Steve McQueen
(^jf book explores the economics of cities and urban problems. The quotes
his
*~S from Parker and McQueen reflect our mixed feelings about cities. On the
positive side, cities facilitate production and trade, so they increase our standard of
living. In addition, they provide consumers with a wide variety of goods and ser-
vices. Unfortunately, cities also have serious problems such as poverty, congestion,
pollution, and crime. Although these problems are truly urban in nature, they could
be solved without abandoning our cities. One of the purposes of this book is to show
that policies that solve urban problems will increase the vitality of cities, causing
cities to grow, not shrink.
Another purpose of the book is to explain some broad changes in the size of
citiesand the fraction of the population living in cities. In 1990, over 75 percent of
the U.S. population lived in urban areas, up from only 6 percent in 1 800. This rapid
urbanization resulted in large part from the technological changes of the industrial
revolution. A number of innovations in production and transportation increased in-
dustrial output and trade. Because most firms are located in cities, growing output
and trade increased the size and number of cities. More recently, however, many
northeastern and north-central U.S. cities actually lost population, a result of migra-
West and the South and a shift away from the
tion to the traditional manufacturing
economy.
The book also explains some changes in the internal spatial structure of cities.
In the nineteenth century, the typical large metropolitan area was monocentric, with
Chapter 1 Introduction
the bulk of the city's employment in its central core area. In the typical mod-
ern metropolitan area, about half of the jobs are in suburban areas, and many
of the suburban jobs are in suburban subcenters. The suburbanization of employ-
ment was caused by a number of factors, including changes in transportation and
communications technology, the building of highways, and other government
policies.
As mentioned in the opening paragraph, this book explores the economics of
urban problems. The conventional list of urban problems includes poverty, segre-
gation, inadequate housing, congestion, pollution, inferior education, and crime.
This text provides three insights into the analysis of these problems. First, most
of these urban problems are related: many of the problems have common roots,
and some of the problems are exacerbated by the other problems. For example,
poverty contributes to the problems of inadequate housing and crime, and crime con-
tributes to neighborhood deterioration and thus worsens the problem of inadequate
housing. Given the common roots and interdependencies of many urban problems,
a comprehensive approach to the problems may be more effective than a piece-
meal one.
The second insight about urban problems is that the economic approach to these
problems often differs from the approaches adopted by policymakers. For example,
the problems of congestion and pollution occur because some resources (roads and
air) are improperly priced: the economic approach is to force travelers and polluters
to pay for the resources they use. In contrast, the policy response to these problems
often involves regulation or subsidization.
The third insight into urban problems is that most of the problems are affected
by land-use patterns and also influence land-use patterns. An understanding of the
spatial dimension of a particular urban problem is necessary to fully understand
( 1 )
the reasons for the problem and (2 predict the spatial responses to a particular public
)
policy.
The remainder of this brief introductory chapter addresses two questions. First,
what is urban economics? The answers to this question provide a preview of the
material covered in the book. Second, what is an urban area, and what is a city? The
economist's definition of a city differs from the definition used by the U.S. Census
Bureau.
What Is Urban Economics?

Urban economics is the study of the location choices of firms and households.
Other branches of economics ignore spatial aspects of decision making, adopt-
ing the convenient but unrealistic assumption that all production and consumption
take place at a single point. In contrast, urban economics examines the where of
economic activities. In urban economics, a household chooses where to work and
where to live. Similarly, a firm chooses where to locate its factory, office, or retail
outlet.
Chapter I Introduction 3
Urban economics explores the spatial aspects of urban problems and public
policy. Urban problems such as poverty, segregation, urban decay, crime, conges-
tion, and pollution are intertwined with the location decisions of households and
firms: location decisions contribute to urban problems, and urban problems in-
fluence location decisions. For example, the suburbanization of employment op-
portunities contributes to central-city poverty,which causes further suburbaniza-
tion as wealthy households flee the fiscal problems of the central city. An in-
formed discussion of alternative policy options must take these spatial effects into
account.
If urban economics is the study of location choices, why isn't it called location
economics or spatial economics? There are three reasons for the urban in urban
economics. First, most location decisions involve an urban choice: over three-fourths
of the U.S. workforce live in cities, meaning that much of location analysis involves
urban areas. Second, urban economics is also concerned with location choices within
cities. Finally, the most important problems caused by location choices occur in urban
areas.
Urban economics can be divided into four related areas that correspond to the
four parts of this book. These are ( 1 ) market forces in the development of cities, (2)
land rent and land use within cities, (3) spatial aspects of poverty and housing, and
(4) local government expenditures and taxes.
Market Forces in the Development of Cities

The first part of the book shows how the location decisions of firms and households
cause the development of cities. A firm chooses the location within a region that
maximizes its profit, and a household chooses the location that maximizes its utility.
These location decisions generate clusters of activity (cities) that differ in size and
economic structure. The questions addressed in this part of the book include the
following:
1 Why do cities exist?
2. What is the role of trade in the development of cities?
3. Where do urban areas develop?
4. Why did the first cities develop?

5. What is the connection between industrialization and urbanization?
6. Why do cities differ in size and scope?
7. What causes urban economic growth and decline?
8. How do local governments encourage economic growth?
9. How does employment growth affect a city's per capita income?
Land Rent and Land Use within Cities
The second part of the book discusses the spatial organization of activities within
cities. In contrast to the first part, which examines location choices from the regional
Chapter ] Introduction
perspective, the second part examines the location choices of firms and households
within cities. It shows how land-use patterns are shaped by the interactions between
different urban activities. The questions addressed in the second part include the
following:
1 What determines the price of land, and why does the price vary across
space?
2. How does public policy affect the price of land and land-use patterns?
3. Why does the price of housing van- across space?
4. Why are households in U.S. cities segregated with respect to income and
1
race'
5. Do the location decisions of households affect the location decisions of

firms?
6. What factors caused the suburbanization of employment and population?
7. Why was the monocentric city of the nineteenth century replaced by the
multicentric city of the twentieth century?
8. Why do local governments use zoning and other land-use controls to

restrict location choices?
9. How does zoning affect the equilibrium price of land?
Spatial Aspects of Poverty and Housing

The third part of the book examines the spatial aspects of two related urban prob-
lems, poverty and housing. Urban poverty has a distinct spatial dimension: the poor
are typically concentrated in the central city, far from the expanding employment
opportunities in the suburbs. The discussion of poverty addresses the following
questions:
1 Who are the poor, and where do they live?
2. What causes poverty?

3. To what extent do market forces diminish racial discrimination in
employment?
4. Does residential segregation contribute to poverty?
5. How does the federal government combat poverty?

6. What are the merits of reform proposals such as the negative income tax.
workfare. earned-income tax credits, and mandated child support?
The third part of the book also examines the spatial aspects of urban housing
problems. Housing choices are linked to location choices because housing is im-
mobile: household chooses a dwelling, it also chooses a location. The most
when a
important urban housing problems are segregation and the deca> o\ dwellings and
neighborhoods in the central city. The two housing chapters address the following
questions:
1 . What makes housing different from other commodities?

2. Why do the poor occupy used housing?
3. What causes racial segregation in housing?
4. Under what conditions will a neighborhood switch from being mostly
white to mostly black?
5. How does public policy affect the supply of low-income housing and its
price?
6. Should the government build new low-income housing or give money to

the poor?
7. How does rent control affect the supply of low-income housing?
Local Government Expenditures and Taxes

The last part of the book explores the spatial aspects of local government policies. Un-
der the fragmented system of local governments, most large metropolitan areas have
dozens of local governments, including municipalities, school districts, and special
districts. governments provide a wide variety of goods, including trans-

These local
portation systems, education, and crime protection. Firms and households "shop"
for the jurisdiction that provides the best combination of services and taxes: citizens
vote with their feet. Therefore, tax and spending policies affect location choices,
thus influencing the spatial distribution of activity within and between cities.
The process of shopping for a local government raises five questions about the
fragmented system of local government:
1 Is the fragmented system of government efficient?

2. What is the role of intergovernmental grants in the fragmented system?
3. How do taxpayers respond to local taxes on property, income, and sales?

4. Is the local property tax regressive or progressive?
5. Does the property tax encourage segregation with respect to income and
race?
One of the responsibilities of local government is to provide a transportation

system. Most cities combine an auto-based highway system with some form of
mass transit (bus, light rail, or heavy rail). Since activities are arranged within cities
to facilitate interactions between different activities, changes in the transportation
system that affect the relative accessibility of different sites affect land-use patterns.
The chapters on transportation address the following questions:
1 What causes congestion, and what are the alternative policy responses?
2. How does congestion affect land-use patterns?
3. What are the alternative policies for dealing with auto pollution?
4. Under what circumstances is a bus system more efficient than a fixed-rail

system such as San Francisco's BART or Washington's Metro?
Chapter I Introduction
5. What are the costs and benefits of light-rail systems?

6. Should transit be deregulated to allow private firms to compete with transit
authorities?
The last two chapters of the book deal with two local public goods: education
and crime control. The study of education fits naturally into the scheme of urban
economics because education affects location decisions: a household's choice of
residence depends in part on the quality of local schools and the taxes required to
support the schools. In addition, location choices affect the provision of education:
if households segregate themselves with respect to income, per pupil spending is
likely to be relatively low in poor districts, so poor children may receive inferior
education. In addition, a child's achievement level depends on (a) the income and
education level of his or her parents and (b) the achievement level of classmates.
Therefore, a child in a poor community may receive an inferior education even if all
schools spend the same amount per pupil.

The chapter on education addresses the following questions:
1 Which of the inputs to the education production function are most

productive? How do the home environment and the peer group affect
educational achievement? Do class size and teachers' qualifications matter?
2. Why does spending per pupil vary across school districts?
3. To what extent do intergovernmental grants decrease spending inequalities?
4. Why does educational achievement vary across school districts?
5. What are the effects of desegregation policies on enrollment in public
schools and educational achievement?

6. What are the merits of proposals to use vouchers or tax credits to subsidize
education in private schools?
The final chapter of the book explores the economics of crime. Like education,
crime affects location decisions: a household's choice of residence depends in part
on the local crime and the taxes required to support the criminal justice system.
rate
Location decisions affect the provision of crime protection: if households segregate
themselves with respect to income, spending on crime protection may be lower in
poor areas. In addition, because there is often a relatively large number of potential
criminals in poor areas, citizens in poor communities would experience higher crime
rates even if spending on crime protection were the same in all communities.
The crime chapter addresses the following questions:
1 Are criminals rational? Is it possible to decrease crime by changing the

expected benefits and costs of crime?
2. Why are crime rates relatively high in central cities?
3. What is the optimum amount of crime?

4. How do the police, the courts, and the prison system deter crime?
5. Do policies that control illegal drugs increase property crime?
What Is an Urban Area?

To an urban economist, a geographical area is considered urban if it contains a large
number of people in a relatively small area. In other words, an urban area is defined
as an area with a relatively high population density. For example, suppose that the
average population density of a particular county is 20 people per acre. If part of
the county contains 50,000 people in a 20-square-mile area (i.e., the population
density is 2,500 people per square mile), it is considered an urban area because
it has a relatively high population density. This definition accommodates urban
areas of vastly different size, from a small town to a large metropolitan area. The
economist's definition is stated in terms of population density because the urban
economy is based on frequent contact between different economic activities, and
such contact is feasible only if firms and households are packed into a relatively
small area.
Census Definitions
The Census Bureau defines a number of geographical areas. Since most empirical
work in urban economics is based on census data, a clear understanding of these
definitions is important.
Municipality or City. A municipality is defined as an area over which a municipal

corporation provides local government services such as sewerage, crime protection,
and fire protection. Another census term for municipality is city. In this case, city
refers to the area over which a municipal government exercises political authority.
When a census document refers to a city, it is referring to the political city, not the
economic city. To define the economic city, we would draw boundaries to include
all the activities that are an integral part of the area's economy. When using census
data, it's important to remember the distinction between the political city and the
economic city.
Urbanized Area. An urbanized area includes at least one large central city
(a municipality) and the surrounding area with population density exceeding 1,000
people per acre. To be an urbanized area, the total population of the area must be at
least 50,000. Figure 1-1 shows an urbanized area as the large circle centered on a
municipality. The urbanized area outside the central city is called the urban fringe.
In 1990, there were 396 urbanized areas in the United States, containing 63.6 percent
of the nation's population.
Metropolitan Area. A metropolitan area (MA) is defined as the area containing

a large population nucleus and the nearby communities that are integrated, in an
economic sense, with the nucleus. Each metropolitan area contains either ( 1 ) a central
city with at least 50,000 people or (2) an urbanized area. In the census definition, the
nucleus of a metropolitan area is either the central city or the urbanized area, and the
Figure 1-1 Urbanized Area and Metropolitan Area
Central city
(municipality)
•
•
Urbanized area
County
integrated communities are the surrounding counties from which a relatively large
number of people commute to the nucleus. To be designated a metropolitan area, the
area in and around an urbanized area must have either a central city with population
greater than 50.000 or a total population (including the integrated communities) of
at least 100.000 (75.000 in New England). This means that some small urbanized
areas are not metropolitan areas. For example, consider an urbanized area with
40.000 people in its central city. 60.000 people in the urbanized area, and another
30.000 people in the integrated counties, for a total of 90.000. Because the central
city has fewer than 50.000 residents and the entire area has fewer than 100.000
people, the urbanized area is not considered a metropolitan area. In 1990. there were
284 metropolitan areas in the United States, ranging in population from just over
50.000 to just over 18 million.
Figure 1-1 shows the simplest possible arrangement of a metropolitan area. The
MA is the county containing the urbanized area. This simple case shows one of the
problems with the MA

concept. To form a metropolitan area, the Census Bureau takes
an urbanized area (or a central city) and tacks on the rest of the county, so the MA
includes areas that are essentially rural. Metropolitan areas are defined by counties
because the county is the smallest geographical unit for which a wide range of data is
collected. In New England, where counties are relatively unimportant, metropolitan

areas include the urbanized area and towns that are sufficiently integrated with the
urban area.
Most metropolitan areas include more than one county. A central county is
defined as one in which ( 1 ) the majority o\' the population lives in the urbanized area
on 2) at least 2.500 of its residents live in the central municipality. An adjacent county
is included in the metropolitan area if it is sufficiently integrated with the central
county or counties. The degree of integration is measured in terms of commuting
flows and population density. For example, the county is included in the MA if
over 50 percent of workforce commutes to a central county and its population

its
density exceeds 25 people per square mile. As the commuting percentage drops, the
threshold population density increases. For the minimum commuting percentage
(15 percent), the county is included if the population density exceeds 50 people per
square mile.
Consolidated Metropolitan Statistical Area (CMSA). There are two types of

metropolitan areas: consolidated metropolitan (CMSAs) and metro-
statistical areas
politan statistical areas (MSAs). If the population of a metropolitan area exceeds 1

million, the metropolitan area may be divided into two or more primary metropolitan
statistical areas (PMS As). A PMSA consists of a large urbanized county (or a cluster
of counties) that is integrated, in an economic sense, with other portions of the larger
metropolitan area. When a metropolitan area is divided into two or more PMSAs,
the metropolitan area is called a CMSA.

In 1 990, there were 20 of these large metropolitan complexes, ranging in pop-
ulation from just over million (Hartford-New Britain-Middletown) to just over
1
18 million (New York-Northern New Jersey-Long Island). Figure 1-2 shows the
maps of two CMSAs, Chicago-Gary-Lake County and San Francisco-Oakland-San
Jose. Each of these CMSAs has six PMSAs.
Metropolitan Statistical Area (MSA). A metropolitan statistical area (MSA)

is CMSA. Most MSAs are
defined as a metropolitan area that does not qualify as a
designated as such because they have less than a million residents. Other MSAs have
more than a million residents, but do not have distinct areas within the metropolitan
area that can be classified as separate PMSAs. In 1990, there were 264 MSAs,
ranging in population from just over 50,000 to just under 2.4 million (Baltimore).
Figure 1-3 shows the maps of two multiple-county MSAs, Sacramento and
Atlanta. The urbanized areas are the darkened areas near the centers of the MSAs.
There are four counties in the Sacramento MSA, two of which stretch far to the east
through rural areas to the Nevada border, about 90 miles from the city of Sacramento.
The Atlanta MSA is more compact, with 18 counties surrounding an urbanized area
centered on the city of Atlanta.
The Census Bureau recently changed its terminology for metropolitan areas. Be-
fore 1983, metropolitan areas were called Standard Metropolitan Statistical
Areas (SMSAs), and groups of related SMSAs were called Standard Consolidated
Statistical Areas (SCSAs).
Urban Place and Urban versus Rural. An urban place is defined as a geograph-
ical area with at least 2,500 inhabitants in a relatively small area. The Census Bureau
defines the nation's urban population as all people living in urbanized areas plus
people outside urbanized areas who live in urban places. According to this definition,
in 1990, about three-fourths of the U.S. population lived in urban areas.
10 Chapter 1 Introduction
Figure 1-2 Consolidated Metropolitan Statistical Areas
Chicago-Gary-Lake County CMSA
Kenosha PMSA
Aurora ,
Elgin
PMSA
Gary-
Hammond
PMSA
San Francisco-Oakland-San Jose CMSA

Vallejo-Fairfield-
Napa PMSA
Santa Rosa-Petaluma PMSA
Mann
San Francisco PMSA Oakland PMSA
San Mateo " San Jose PMSA

Santa Cruz
Santa Cruz PMSA

Figure 1-3 Metropolitan Statistical Areas
Sacramento Metropolitan Statistical Area
Atlanta Metropolitan Statistical Area
Summary: What Is Urban? In collecting and reporting data, the Census Bu-
reau uses several definitions of urban. Table 1-1 shows, for three different urban
definitions, the percentage of the U.S. population that is considered "urban." The
percentage living in urban areas (75.2 percent) exceeds the percentage in urbanized
areas (63.6 percent) because the urban-area definition includes people who live in
relatively small places (urban places with populations between 2,500 and 50,000).
12 Chapter 1 Introduction
Table 1-1 Urbanization of U.S. Population under Different

Census Definitions
Percent of U.S. Population

Area Definition in Area in 1990
Urbanized areas: central cities and surrounding dense areas 63.6%

Metropolitan areas: CMSAs and MSAs 77.5
Urban areas: urbanized areas and smaller urban places 75.2
The percentage living in metropolitan areas differs from the percentage in urbanized
areas for two reasons. First, the metropolitan-area definition excludes people who
live in urbanized areas with either a relatively small central city (less than 50.000)
or a relatively small total population (less than 100,000 in the urbanized area and the
integrated counties). Second, the metropolitan-area definition includes people who
live outside urbanized areas (in the county containing the urbanized area or a county
adjacent to it). The exclusion of people in small urbanized areas is outweighed by
the inclusion of people outside urbanized areas, so the percentage of the population
living in metropolitan areas exceeds the percentage in urbanized areas.
Definitions Used in This Book

This book uses three terms to refer to spatial concentrations of economic activity:
urban area, metropolitan area, and city: These three terms, which will be used
interchangeably, refer to the economic city (an area with a relatively high population
density that contains a set of closely related activities), not the political city. When
referring to a political city, the book will use the term central city or municipality.
Discussion Questions
1. Suppose that you have the power to develop a new set of census definitions for
urban and metropolitan areas. How would you define (a) a metropolitan area,
(b) an urban resident, (c) a central city, and (</) a suburban area?
2. Consider a region in which all urbanized areas have populations of at least
100.000. Will the percentage of the region's population living in urbanized

areas be less than or greater than the percentage living in metropolitan areas?
3. Between 1980 and 1990. the gap between the percentage of the population
living in urbanized areas and the percentage of the population living in
metropolitan areas decreased, from 14.8 percentage points (76.2 percent-61.4
percent) to 13.9 percentage points (77.5 percent-63.6 percent). Provide an
explanation lot the narrowing of the gap.
4. Which of the three urban definitions used by the Census Bureau (urbanized area,
urban place, metropolitan area) is closest to the economist's definition of a city?
5. In 1990, 86 percent of the population in the western states lived in urbanized

areas, and 85 percent lived in metropolitan areas. In the same year, 79 percent of
the population in the northeastern states lived in urbanized areas, and 88 percent
lived in metropolitan areas. Why is the West more urbanized than the Northeast
under the urbanized-area definition but less urbanized under the
metropolitan-area definition?
References
U.S. Department of Commerce. "Appendix A: Area Classifications.** In 1990 Census of
Population and Housing: Summary of Population and Housing Characteristics.
Washington, D.C.: U.S. Government Printing Office, 1990. Defines all the
geographical entities and concepts that are used in the 1990 census. See pages A-8 to
A- 13 for definitions of cities and metropolitan areas.
Parti
J£ ARKET FORCES
DEVELOPMENT
in the
of (HIES
7-
^y n a market economy, individuals exchange their labor for wage income,
iS which is then used to buy other goods and services. How do these market
transactions affect cities? Chapter 2 discusses the fundamental reasons for central-
ized production and marketing and explores the implications for the development
of cities. Chapter 3 shows how the location decisions of firms cause the develop-
ment of cities in particular locations. Chapter 4 provides a brief historical sketch of
urbanization, using the concepts developed in Chapters 2 and 3 to explain the history
of cities. Chapter 5 takes a regional perspective, showing how competition among
firms leads to the development of a hierarchical system of cities. Chapter 6 explores
the reasons for urban economic growth.
Chapter 2
w UY DO
CITIES EXIST?
Man is the only animal that makes bargains; one dog does not change bones
with another dog.
Adam Smith
•O ities exist because individuals are not each of us produced

self-sufficient. If
O everything we consumed and we want much company, there would be

didn't
no need to live in cities. We aren't self-sufficient, however, so we exchange our labor
for other goods. Most of us live in cities because that's where most of the jobs are.
Cities also provide a rich mixture of consumer goods and services, so even people
who are not gainfully employed are attracted to cities. By living in cities, we achieve
a higher standard of living, but we also must tolerate more pollution, crime, noise,
and congestion.
This chapter explores three reasons for the concentration of jobs in cities. Com-
parative advantage makes trade between regions advantageous, and interregional
trade causes the development of market cities. Internal scale economies in pro-
duction allow factories to produce goods more efficiently than individuals, and the
production of goods in factories causes the development of industrial cities. Ag-
glomerative economies in production and marketing cause firms to cluster in cities,
and this clustering causes the development of large cities. This chapter focuses on
the market forces that generate cities, ignoring the social reasons for cities (e.g.,
religion, companionship, and politics). Some of these nonmarket reasons for cities
will be discussed later in the book.
A Model of a Rural Region

Under what conditions would there be no cities in a particular region? In other words,
what set of assumptions guarantees a uniform distribution of population? This section
17
18 Part l Market Forces in the Development of Cities
develops a model of a region that has a uniform distribution of population and thus
has no cities. The model of the rural region provides a list of assumptions that
together preclude the development of cities. When these assumptions are relaxed,
cities will develop.
The model of the rural economy captures some of the essential features of
England before the Norman Conquest of the eleventh century. The economy was
essentially rural, with only a few small market towns serving the local population.
Later in the chapter, the model will be used to explain the market forces behind the
development of English cities after the Norman Conquest.
Consider a region where only two commodities (wheat and wool cloth) are
produced and consumed. The regional economy has the following characteris-
tics:
1. Inputs. Wheat and cloth are produced with labor and land. Land is used to
grow wheat and raise sheep. Residents spin the raw wool into yarn and then
weave the yarn into cloth.
2. Equal productivity. All residents are equally productive at making cloth
and producing wheat. In one hour, every person can produce either one
bushel of wheat or one yard of cloth. Similarly, all land is equally
productive in the production of wheat and raw wool.
3. No scale economies. Production is subject to constant returns to scale: a
worker produces one yard of cloth per hour, regardless of how much cloth
is produced. Similarly, wheat production per hour is independent of the
volume produced.
4. Travel time. Travel within the region is by foot: residents walk at a speed
of four round-trip miles per hour.
These assumptions are strong enough to prevent trade. Every household in the
region will produce its own wheat and cloth. Because every household is equally
productive in producing both goods, there are no advantages from trade between
households. Because productivity is independent of the volume produced, there are
no advantages from centralized production. A factory worker would be no more
productive than a person working at home. The disadvantage of factory cloth is that
travel to the factory takes time, making the net cost of the factory cloth (price plus
travel cost) greater than the cost of homemade cloth. To summarize, there are no
advantages from trade or centralized production, so every household in the region
is self-sufficient.
The population will be distributed uniformly throughout the region. Every house-
hold will pick a plot of land and make its own wheat and cloth. Since all land is
equally productive, the distribution of population will be uniform. Since a city is
defined as a place with relatively high population density, a region with a uniform
distribution of population has no cities. There are no cities because the model's
assumptions are strong enough to eliminate the possibility of trade.
Chapter 2 Why Do Cities Exist ? 19
Comparative Advantage
One development of cities is comparative advantage. The model of the
factor in the
rural region assumes that all households are equally productive in the production
of cloth and wheat. If this assumption is relaxed, one part of the region may have
a comparative advantage in cloth production, and the other part may have a com-
parative advantage in wheat production. Under certain circumstances, comparative
advantage causes trade and the development of cities.
The notion of comparative advantage can be explained with a simple model of
tradebetween two parts of the region. Suppose that the western half of the region
more productive than the eastern half. In Table 2—1, the West has an absolute
is
advantage in producing both wheat and cloth: western residents produce twice as
much wheat per hour and six times as much cloth per hour. The differences in
productivity could be caused by differences in labor skills, weather, or soil quality.
The notion of comparative advantage is based on the principle of opportunity
cost. In a one-hour period, a western worker can produce either six yards of cloth or
two bushels of wheat. Therefore, the opportunity cost of cloth is one-third of a bushel
of wheat, and the opportunity cost of wheat is three yards of cloth. An eastern worker
can produce either one bushel of wheat or one yard of cloth, so the opportunity cost
of wheat is one yard of cloth, and vice versa. The West has a comparative advantage
in the production of cloth because the opportunity cost of cloth is one-third of a
bushel of wheat, compared to one bushel of wheat in the East. Similarly, the East
has a comparative advantage in wheat because the opportunity cost is one yard of
cloth instead of three.
Trade and Transportation Cost
Comparative advantage may lead to trade between East and West. To explain the
possible advantages of trade, suppose that all households in the region are initially
self-sufficient. Suppose that two households, one in the East and one in the West,
agree to exchange two yards of cloth for one bushel of wheat; that is, the price of
wheat is two yards of cloth. If the western household switches one hour of work
from wheat production to cloth production, it sacrifices two bushels of wheat to
produce six additional yards of cloth. If it exchanges the extra cloth for three bushels
of wheat from an eastern household, its net gain from trade is one bushel of wheat.
Table 2-1 Com iparative Advantage
Output per Labor Hoi ir Opportunity Cost of Production
East West East West
Wheat 1 2 1 Cloth 3 Cloth

Cloth 1 6 1 Wheat | Wheat
20 Part I Market Forces in the Development of Cities
If the easternhousehold switches one hour of work from cloth to wheat production
and exchanges the extra bushel of wheat for two yards of cloth, its net gain from
trade is one yard of cloth.
What about transportation costs? Trade between the two households is beneficial
only if the differences in productivity are large enough to offset the costs of shipping
wheat and cloth between the two areas. To continue the numerical example, suppose
that it takes 2 hours of travel time for the western household to execute the trade of
cloth for wheat. Is trade still beneficial? The net gain from trade is the gross gain
(one bushel of wheat) less the opportunity cost of the 2 hours spent executing the
trade. Since the western household could use the 2 hours to produce 4 bushels of
wheat for itself, trade is not beneficial.
Trade will be beneficial if transportation costs are small relative to the differ-
ences in productivity. If the net gains from trade are positive, western households
will specialize in cloth production and eastern households will specialize in wheat
production. For example, suppose that it takes only 6 minutes ( 1/10 of an hour) to
execute a trade, so the opportunity cost of the transaction for the western household
is only 0.20 bushels of wheat. In this case, the net gain is 0.80 bushels ( 1 .0 — 0.20).
so households trade to exploit their comparative advantages.
Scale Economies in Transportation and Trading Cities
Trade will cause the development of a trading city if there are scale economies in
transportation. To explain the importance of scale economies in transportation, sup-
pose that there are no scale economies: the transport cost per bushel of wheat (or
yard of cloth) per mile is independent of the volume shipped. In this case, house-
holds in the two regions engage in direct trade: each eastern household links up with
a western household to exchange cloth and wheat. If there are scale economies in
transportation, however, the cost per unit per mile decreases as the volume trans-
ported increases, so it is cheaper to transport wheat and cloth in bulk. By exploiting
these scale economies, trading firms can collect, transport, and distribute the goods
at a lower cost than would be incurred by households engaged in direct trade. The
trading firms locate at places convenient for the collection and distribution of the
goods, causing the development of marketplaces at crossroads, ports, river junctions,
and other transshipment points.
The location decisions of traders cause the development of market cities. People
employed by the trading firms live near the marketplace to economize on commuting
costs, and bid up the price of land. As the price of land increases, residents economize
on land by occupying relatively small lots. In other words, the population density
around the marketplace is higher than the population density in the rest of the region.
Since a city is defined as a place with a relatively high population density, the
combination of comparative advantage and scale economies in transportation causes
the development of a market city.
The market city develops because three conditions are satisfied. First, agricul-
tural productivity is high enough that agricultural workers produce enough wheat
and cloth for themselves and the urban traders. The agricultural surplus feeds and
Chapter 2 Why Do C 'tries Exist ? 21
clothes the urban workers. Second, the differences in productivity that generate com-
parative advantage are large enough to offset transportation costs, so trade occurs.
Third, there are scale economies in transportation, making large-scale trade and
central marketplaces efficient.
Internal Scale Economies in Production

A second factor in the development of cities is scale economies in production. One
of the assumptions of the model of the rural region is that there are constant returns
wheat and cloth. Therefore, there are no advantages
to scale in the production of both
from centralized production in factories, and all goods are produced in the home.
If this assumption is dropped, factory production may replace home production,
causing the development of industrial cities.
Scale Economies and the Cloth Factory

Suppose that there are advantages from the large-scale production of cloth. In other
words, as the volume of production increases, the labor required to produce one yard
of cloth decreases. Figure 2-1 shows the average labor requirements for different
amounts of factory cloth. The average labor time decreases from 1 hour for yard of 1
same time as home production) to 15 minutes for an output of 400 yards.

cloth (the
Figure 2—1 Scale Economies in Clothmaking
0.50
0.25 Average labor time
200 400 600
Volume of cloth (yards)
economies
If there are stale in clothmaking. the average time required to make a yard of cloth decreases as the
volume produced increases.
These scale economies arise for two reasons:
1 Factor specialization. In a large operation, each worker is assigned a

single task. The specialization of labor increases productivity because
(a) workers' skills increase with repetition and (b) workers spend less time
switching from task to task. The making of woolen cloth has several steps:
the raw wool is spun into yarn, the yarn is woven into cloth, and the raw
cloth is finished (cleaned, thickened, and dyed). One reason for scale
economies in cloth production is that individual workers can specialize in
different tasks. A group of 10 specialized workers can produce more than
10 times as much as 10 individuals working in isolation (each of whom
performs all the production tasks).
2. Indivisible inputs. An input to the production process is indivisible if the

input has a minimum efficient scale. If an indivisible input is cut in half, the
total output of the two halves is less than the output of the whole. The cloth
factory uses equipment (spinning machines, power looms, thickening
machines, dyeing vats) that cannot be efficiently scaled down for use by
individual clothmakers. As output increases, the factory uses more
indivisible inputs, increasing productivity.
If the cloth factory produces 400 yards of cloth per hour, it employs 100 workers
(400 times 0.25 hours of labor per yard of cloth).
What is the wage of The wage must be high enough to make
factory workers'?
cloth workers indifferent between working in the factory and working at home. If
consumers pay the factory workers 0.25 bushels of wheat for one yard of cloth, the
factory worker earns one bushel of wheat per hour of factory work. At this wage, the
worker is indifferent between working in the factory (earning one bushel of wheat
per hour) and producing his or her own wheat (at one bushel per hour).
A household buys factory cloth if the net cost of factory cloth is less than the
cost of home production. There are two components to the net cost of factory cloth.
The first is the time required to produce enough wheat to pay the factory worker
(0.25 bushels of wheat requires 0.25 hours of wheat production). The second is the
time required to travel to and from the cloth factory. If the trip to the factory takes
less than 0.75 hours, the net time cost of factory cloth is less than the time required
to produce cloth at home, so buy the factory cloth. To pay for the
the household will
factory cloth, the household stops making homemade cloth and uses the extra time
to produce more wheat.
Figure 2-2 shows the market area of the cloth factory. The market area is
defined as the area over which the factory underprices home production. The vertical
axis measures the net cost of cloth, the sum of the factory price (0.25 hours of wheat
production) and travel time to the factory. The horizontal axis measures the distance
to the factory in miles. If walking time is four round-trip miles per hour, the factory
will underprice homemade cloth for residents within three miles of the factory (within
0.75 hours walking time). Therefore, the factory's market area is a circle with a radius
of three miles.
Chapter 2 Why Do Cities Exist'.' 23
Figure 2-2 Market Area of Cloth Factory

Cost of factory cloth Cost of factory cloth
Cost of homemade cloth
Miles from factory
The market area of the cloth factory is the area over which the net cost of factory cloth (production time plus travel time) is less
than the production time of homemade cloth ( 1 hour).
Scale Economies and an Urban Area

A small urban area develops around the cloth factory. If the output of the factory
is400 yards of cloth, it employs 100 people. The workers live near the factory to
economize on commuting costs, and bid up the price of land near the factory. As the
price of land increases, workers economize on land by occupying small lots. In other
words, the population density around the cloth factory is higher than the population
density in the rest of the region. Since a city is defined as a place with a relatively
high population density, the factory causes the development of a small factory city.
— The factory city develops because two conditions are satisfied. First, the pro-
ductivity of agricultural workers (wheat growers) is high enough that they can feed
themselves and have enough left over to feed the cloth workers in the city. In other
words, the agricultural surplus feeds city workers. Second, scale economies are large
relative to travel costs, so the cloth factory can underprice homemade cloth.
The size of the city is determined by the number of factory workers, which in
turn depends on the total output of the factory. The factory can increase its output if
itmanages to underprice more home producers. Therefore, output increases as scale

economies increase (decreasing the factory price) and transportation costs decrease
(decreasing the net cost of factory cloth). Chapter 5 (How Many Cities?) provides a
more complete analysis of the factors that determine the firm's market area.
Examples of Urban Development

This section discusses the role of comparative advantage and scale economies in the
development of cities England and the United States. It uses the concepts of
in this
chapter to explain the timing and location of urban development.
The English Woolen Industry

Before the Norman Conquest in 1066. the making of wool cloth in England was an
entirely local affair. Individual households and villas raised their own sheep, spun
the raw wool into yarn, and wove the yarn into cloth. In other words, households
were self-sufficient in cloth production, and no cities were involved in the production
or trade of wool cloth. A small number of towns and cities were involved in local
trade.
The Norman Conquest increased the interaction between England and the con-
development of the woolen cloth trade. The barriers
tinent, setting the stage for the
to trade fell, and England was able to exploit its comparative advantage in the pro-
duction of wool cloth. Exports to the continent rose, causing the development of
small cities and towns based on the cloth trade.
In the twelfth century, there was a flourishing cloth industry in several English
cities, including Lincoln. Northampton. York, and Bristol. These cities developed as
both commercial and production centers. Although spinning, weaving, and finishing
were still performed by hand, there were some small scale economies associated
with labor specialization, so centralized production in cities was efficient.
Starting in the thirteenth century, the clothmaking industry moved from cities to
rural areas with water power. The fulling process involves cleaning and thickening the
wool cloth. Before the development of the fulling mill, workers stomped on wet cloth
to clean and thicken it. The water-powered fulling mill replaced foot-stomping, caus-
ing producers to move their finishing operations to sites along fast-moving streams.
The move to rural fulling sites caused the replacement of urban spinners and
weavers by rural workers organized under the putting-out system. The clothier dis-
Chapter 2 Why Do Cities Exist.' 25
tributed raw wool to rural households for spinning and weaving. The making of cloth
(using primitive spinning wheels and hand looms) was a part-time activity, typically
performed by the women of the rural household. The clothier collected the raw cloth
from his rural workers, brought it to fulling mills for cleaning and thickening, and
then shipped the cloth to his customers in other regions and countries.
The and a rural workforce caused the decline of the tradi-
shift to fulling mills
tional clothmaking and the development of new villages and small cities. The
cities
fulling mills provided a convenient collection point for the cloth produced by the ru-
ral households and caused the development of many small villages. Since most of the
production activity was performed by rural households, most of the cities involved
in the cloth industry served primarily as commercial centers. They differed from
the traditional clothmaking cities, which served as both commercial and production
centers.
The industrial revolution of the nineteenth century produced a number of inno-
vations that increased the scale economies associated with the production of wool
cloth. Mechanized spinning replaced hand spinning in the early 1800s, and power
looms replaced hand looms about 830. When the coal-fired steam engine was modi-
1
fied to power the spinning and weaving machinery, large factories developed on sites
accessible to coal deposits (e.g., the West Riding of Yorkshire). The cloth factory
integrated the production process, performing the tasks of spinning, weaving, and
finishing under a single roof. The scale economies in production allowed the factory
to produce wool cloth at a lower cost than home producers, so the putting-out system
was replaced by factory production. The development of large factories caused the
development of relatively large factory cities. The leading cities in the production
of wool cloth were Bradford, Halifax, Leeds, Huddersfield, and Wakefield.
This brief history of English clothmaking cities shows how changes in trade and
scale economies transformed rural areas into highly urbanized ones. The opening of
the European markets following the Norman Conquest allowed England to exploit its
comparative advantage in clothmaking, and the traditional clothmaking cities devel-
oped. The industrial revolution increased scale economies in clothmaking, causing
the development of clothing factories and large factory cities.
The Sewing Machine and U.S. Urbanization

The sewing machine, which was developed in the middle of the nineteenth century,
increased scale economies in the making of clothing and contributed to the devel-
opment of cities. Around the beginning of the nineteenth century, about four-fifths
of the clothing worn in the United States was hand-sewn in the home for a member
of the household. The other fifth was hand-sewn by tailors. The sewing machine
allowed factory producers to underprice home producers, and, by 1 890, nine-tenths
of U.S. clothing was being made in factories.As clothing factories sprang up in the
United States, new cities developed around the factories.

The sewing machine also contributed to the development of U.S. cities around
shoe factories. Before 1700, most shoes were produced in the home or the local
village. The cost of transportation was so high that local production was efficient.
26 Parti Market Forces in the Development of Cities
Over time, transportation costs decreased, allowing the exploitation of compara-

tive advantage in shoe production. The putting-out system was implemented in
the 1700s: shoe producers distributed raw materials to cottage workers, collected
their output, and finished the shoes in a central shop. As new shoemaking ma-
chines were developed, the number of operations performed in the central shop
increased. The McKay sewing machine, which mechanized the process of sewing
the soles to the uppers, increased scale economies in shoe production to the point
that the central shop became a genuine factory. Cities developed around the new shoe
factories.
Agglomerative Economies in Production

This section explores some of the reasons why cities are large. The analysis of scale
economies earlier in the chapter suggests that a city develops around a factory. This
section explains why most cities have more than one factory; that is. it explains the
development of large industrial cities.
The large industrial city develops because of agglomerative economies in pro-
duction. By locating close to one another, firms can produce at a lower cost. This
is an example of a positive externality in production: the production cost of a
particular firm decreases as the production of other firms increases. There are
two types of agglomerative economies: localization economies and urbanization
economies.
Localization Economies
Localization economies occur if the production costs of firms in a particular industry

decrease as the total output of the industry increases. The positive externalities
underlying localization economies are realized by firms in a particular industry. To
realize the localization economies, a firm must be located close to other firms in the
industry, so firms cluster to decrease their production costs. Localization economies
occur for three principal reasons: scale economies in the production of intermediate
inputs, labor-market pooling, and knowledge spillovers.
Scale Economies in Intermediate Inputs. Some clusters occur because firms in

a particular industry buy an intermediate input from the same supplier. Firms cluster
around a common input supplier if two conditions are satisfied:
1. The input demand of an individual firm is not large enough to exploit the
scale economies in the production of the intermediate input.
2. Transportation costs are relatively high. If demanders and supplier interact
in the design or fabrication of the intermediate input, face-to-face contact
between buyer and seller is necessary, and proximity to the input supplier is
important. Similarly, if the intermediate input is bulky, fragile, or must be

delivered quickly, proximity is important.
Chapter 2 Why Do Cities Exist? 27
Vernon (1972) uses the Manhattan dressmaking industry as an example of a

from scale economies in the production of intermediate inputs.
cluster that resulted
Because the demand for high-fashion dresses is unpredictable, a dressmaking firm
cannot commit itself to large-scale production of a given type of dress, but must be
prepared to quickly change to another type of dress if the first dress is a bomb. The
firm must employ inputs (capital and labor) that can ) produce a wide variety of
( 1
dresses and (2) make from one dress to another quickly. Specialized
the transition
labor and machinery are not feasible. The manager of the firm must carefully monitor
the dress market, the design process, and the production process. The operation must
be kept small enough that the manager can be involved in most decisions.
One of the intermediate inputs to dressmaking is buttons. Suppose that there are
scale economies in the production of buttons and it takes 10 dressmakers to generate
sufficient button demand to exhaust the scale economies in button production. By
sharing a buttonmaker, the 10 dressmakers can get buttons at a lower cost. Because
the scale economies in buttonmaking are large relative to the scale economies in
dressmaking, the dressmakers share a buttonmaker.
Why should dressmakers cluster around the button firm? They locate near the
button producer because they must supervise the production of buttons. Because
the buttons for a high-fashion dress are not standardized, they cannot be ordered
out of a catalogue. The buttons are designed to complement the other features of
the dress and the dressmaker participates in the design and fabrication of buttons to
ensure that the buttons match the dress. Given the importance of face-to-face contact,
dressmakers locate close to the buttonmaker.
There are many other examples of clusters that result from scale economies in
the provision of intermediate inputs.
1 Corporate headquarters. The corporate headquarters produces a wide

The executive may design an advertising campaign one
variety of outputs.
week, pick a location for a new plant the next week, and develop a strategy
to fend off a lawsuit the followingweek. Most corporations use outside
advertising firms because scale economies in the production of advertising
campaigns are large relative to the advertising demand of an individual
corporation. Corporate executives need to be close to the advertising firm
because the executives assist in the design of the advertising campaign; that
is, face-to-face contact is important. In general, the clustering of corporate
headquarters (usually in central business districts) allows corporations to
exploit scale economies in the production of intermediate goods
(advertising, economic consultants, legal services) for which face-to-face
contact is an important part of the production process.
2. High-technology firms. Firms producing new high-technology goods face
uncertain demand for their products. Instead of producing all of their own
components, they purchase electronic parts from firms that can exploit scale
economies in production. The firms interact with their suppliers in the
design and fabrication of the components, and must locate close enough to
facilitate frequent face-to-face contact. Although the high-technology firm
need not locate close to its supplier of nuts and bolts (standardized inputs
that can be ordered from a catalogue), it needs to locate closer to the
supplier of nonstandardized electronic parts. In addition, high-technology
firms exploit scale economies in product testing by sharing firms that
provide testing facilities. They locate close enough to tap the testing
facilities at top speed.
3. Publishing. Newspapers and magazines produce outputs (text and
illustrations) that change in unpredictable ways from day to day. In
producing their products, publishers call on a wide variety of experts,
including information sources and illustrators. If each publisher uses an
expert on Armenian history for one story per year, it would be inefficient
for each publisher to hire a full-time Armenian expert. Instead, the
publishers may share a single expert, sharing the cost of specialized
information. Similarly, publishers occasionally need special illustrations,
and call on graphic design firms to do the illustrations. Publishers cluster
around organizations that provide expert information (libraries, research
institutes, universities) and illustrations (graphic design firms).
What is the role of business services in the development of firm clusters? Some
industries require specialized financial services such as specialized banking and
insurance. There are scale economies in the provision of these services, and firms in
economies by sharing banks and insurance agencies.
a cluster can exploit these scale
Similarly, there are scaleeconomies in the provision of transportation services, and
firms can share a transportation firm. For example, a small firm can hire trucks from
a trucking firm instead of providing its own truck fleet.
Public services also play a role in the development of firm clusters. The public
sector is responsible for the development of the transportation network and the
sewage system. If firms in an industry require specialized transport networks or
ewage services, public-service costs are lower if the firms cluster. Th e prov ision of
public education also played a role in the dev elopment of firm cluste rs In England . ,
cpmp town-. ^t;i hlished spinning Tchools to provide specialized laborers for th e
te xtile industry. More recently, the universities and colleges in the Silicon Valley ar e
turning out a large number of engineers and computer scientists for the computer
industry.
Labor-Market Pooling. A second reason for localization economies is that a
cluster increases the efficiency of the labor market. Consider an industry with rapidly
changing production processes and product demands, such as the computer industry
Every year, computer firms introduce new products: some are big sellers, and others
fail. Eventually, the workers in the unsuccessful firms move to the successful ones.
A cluster of firms facilitates the transfer of workersbetween computer firms for
two reasons. First, workers in the cluster have relatively low search costs because
(1) information about job openings is spread through informal channels (casual
conversation at restaurants, bowling alleys, and baseball games) and (2) prospective
employers are nearby, making formal job searches relatively easy. Second, because
of the physical proximity of employers, moving costs are relatively low: workers can
easily switch to a different firm in the same city.Because search costs and moving
costs are relatively low, firms in the cluster of computer firms can draw from the
labor pool to quickly fill their job vacancies and quickly increase production.
The television industry also benefits from labor-market economies. In a given
season, some television programs are hits and others are failures. When it becomes
clear which programs will continue to be produced, actors and technicians move
from the unsuccessful programs to the successful ones. The concentrations of the
television industry in Los Angeles and New York facilitate the transfer of laborers
from one firm to another.
We can use Figure 2-3 to explore the implications of clustering and labor pooling
(Krugman, 1995). Consider a software firm that has two equally likely outcomes in
a particular year: good times or bad times. In good times the price of its product is
high, so the firm has a relatively high demand for labor, as indicated by the demand
curve labeled D g00 d- In bad times the price of its product is low, so the firm has a
relatively low demand for labor, as indicated by the demand curve labeled Dbad-
The left panel of Figure 2-3 shows what happens if the firm chooses an isolated
site, far from other software firms. The supply of labor available to the software firm
at the isolated site is fixed at 120 workers. In good times the firm pays a wage of
$30, but in bad times the firm pays a wage of only $10. In other words, an isolated
Figure 2—3 Benefits and Costs of Labor Pooling

(A) (B)
Isolated Firm Firm in Cluster
120 160
Number of workers Number of workers
An isolated firm pays different wages

in good times and bad times, but hires the same number of workers in
both cases. In contrast, a firm an industry cluster pays the same wage in both cases, but hires more workers
in
in good times. The benefit of clustering and labor pooling is that during good times the firm pays a lower wage
and hires more workers. The cost is that during bad times the firm pays a higher wage. The benefit exceeds the
cost because the firm hires more workers during good times than during bad times.
same number of workers in good times as in bad. but pays

firm hires the a higher
wage in good times.
The right panel of Figure 2-3 shows what happens if the firm locates in a cluster
of other software firms and draws on a pool of software workers. Suppose that in any
year, a good time experienced by one firm is balanced by a bad time experienced by
another firm, so the industrywide demand is stable. This means
for software workers
that the wage is constant over time at $20. A firm in the cluster will pay a wage of
$20 in good times and bad. but will hire 60 workers in good times, compared to
1
only 80 workers in bad times.

Consider the trade-offs associated with moving the firm from the isolated site
to the industry clusterwith the labor pool. The benefit of moving to the cluster is
that during good times, the firm would pay a lower wage and earn more profit. In
the right panel of Figure 2-3, the benefit of pooling is shown by the lightly shaded
trapezoid, which includes a rectangle and a triangle. The rectangle (a height of $10
($30 — $20) and a length of 120) shows the savings on labor costs for the first 120
workers hired: the firm saves $10 on each worker, so the total savings is $1,200. The
triangle (a height of $10 and a base of 40) shows the extra benefits associated with
hiring the 121st through the 160th workers. The demand curve shows the marginal
benefit (marginal revenue product) of workers, and the gap between the demand
curve and the wage line shows the net benefit from hiring a worker (the revenue
generated by a worker minus the wage). The area of the triangle is $200, so the area
of the trapezoid (the benefit of pooling) is $1,400.
The cost of moving from the isolated site to the cluster is related to the fact that
during bad times, the firm will pay a higher wage in the cluster than it would in an
isolated site. During bad times the firm would pay $20 in the cluster, compared to only
$10 at the isolated site. In the right panel of Figure 2-3. the darkly shaded trapezoid
shows the cost of labor pooling. The firm in the cluster pays a higher wage to the 80
workers it hires, generating a cost of $800 ($10 times 80). In addition, the firm hires
fewer workers and thus loses an amount shown by the triangle with height $10 and
base 40 workers ($200). Therefore, the cost of clustering and labor pooling is $ .000.
1
Are the benefits of clustering greater than the cost? Looking at the right panel
of Figure 2-3. the lightly shaded trapezoid is larger than the darkly shaded one, so
the benefits exceed the costs. In this example, the benefit is $1,400 and the cost is
$1,000, so the net benefit is $400. This will always be the case if the wage in the
cluster is the average of the two wages (for good times and bad times) at the isolated
site. Geometrically, the height of the two trapezoids is the same, but the base of the
benefit trapezoid is longer, so the benefit trapezoid will always be bigger than the
cost trapezoid. The economic explanation for the net benefits of clustering is that
the cluster allows the firm to benefit from good times by hiring more workers at the
same wage, but also allows the firm to cushion the effects of bad times by hiring
fewer workers.
Knowledge Spillovers. A third benefit of a cluster is that the cluster facilitates the
rapid exchange of information and diffusion of technology. Workers from different
firms exchange ideas about new products and new production techniques: the larger
C\mpier 2 Why Do Cities Exist? 31
the number of workers in an industry, the greater the opportunity to exchange ideas.
The opportunity to exchange ideas occurs in both formal and informal settings. A
cluster of computer makers produces a large concentration of computer scientists and
engineers who can exchange ideas while they work (e.g., the suppliers of intermediate
inputs interact with the designers of new products) and play (e.g., the workers from
different firms "talk shop" while eating or jogging).
As explained by Saxenian (1994), Silicon Valley — the conglomeration of elec-
tronics —
and computer firms centered south of San Francisco is a network of special-
ized companies, each of which focuses on a narrow niche within the industry. Each
new product uses inputs from many firms. This network system creates an atmo-
sphere that encourages collaboration, experimentation, and shared learning among
companies. In contrast, the firms along Route 128 in the Boston area are less inter-
dependent. This communication between firms in Silicon Valley played a big role
in the emergence of Silicon Valley as the center of the electronics industry.
Where Do the Clusters Develop? Localization economies cause the firms in some
industries to locate close to one another. Most of these clusters are not isolated from
other activities, but are typically in cities. As explained later in the chapter, some
intermediate inputs are shared by firms in different industries, causing firms from
different industries to cluster in cities. For example, the dressmaker cluster (around
the buttonmaker) occurs in Manhattan because dressmakers also use business ser-
vices (machine repair, banking, transportation services) utilized by nearby firms in
other industries. In the next chapter of the book, "Where Do Cities Develop?" we'll
explore the influence of localization economies on the location of cities.
The Incubator Process. Localization economies are responsible for the incuba-
tor process. A cluster of firms supplies a rich set of intermediate inputs, providing
a nurturing environment where firms in an immature industry "incubate," develop-
ing new products and production technologies. In the early life of many industries,
demand is unpredictable and production techniques are unsettled, so firms cluster
to (1) exploit scale economies in the production of intermediate goods. (2) realize
labor-market economies, and (3) exploit the opportunities to exchange ideas about
products and production processes. As the industry matures, it develops standardized
products and production processes, allowing large-scale production. Firms internal-
izemost of the production process, and move to locations that provide lower labor
and land costs. This is also known as the product cycle theory of industrial location:
products are developed by small firms in clusters; when the product is standardized
and produced on a larger scale, firms move to sites with lower labor and land costs.
As pointed out by Vernon (1972), one example of incubation comes from the
radio industry. In the 1920s, the market for radios was unpredictable, and firms
were experimenting with alternative production processes. Mass production was
impractical: the firms didn't know what kind of radio consumers wanted, and hadn't
developed efficient production techniques. Radio firms were small, agile, and ner-
vous, and clustered in New York City to exploit scale economies in the production
of intermediate goods. In the 1930s and 1940s, products were standardized, and
firms developed efficient methods ofmass production. Firms moved their assembly
Midwest, which had lower wages and better access to the national
facilities to the
markets. Tube producers migrated to the Northeast and the South, where labor costs
were lower.
The computer industry is also subject to the incubation process. In the 1970s
and early 1980s, the demand for computers was unpredictable, and production tech-
niques were being developed. Small computer firms clustered in the Silicon Valley
in California and along Route 128 in Boston to exploit localization economies. The
standardization of products and production processes allowed firms to set up large
production facilities outside the clusters. Why did the clusters survive the exodus of
production facilities? New products continue to be developed, and firms continue
to cluster to exploit localization economies. Most firms maintain their research and
development facilities in the clusters and set up branch plants for manufacturing and
assembly.
Urbanization Economies
Urbanization economies, a second type of agglomerative economy in production,

occur if the production cost of an individual firm decreases as the total output of the
urban area increases. Urbanization economies differ from localization economies
intwo ways. First, urbanization economies result from the scale of the entire ur-
ban economy, not simply the scale of a particular industry. Second, urbanization
economies generate benefits for firms throughout the city, not just firms in a partic-
ular industry.
Urbanization economies occur for the same reasons as localization economies.
Firms from different industries share common input suppliers, allowing the realiza-
tion of scale economies in the provision of business services (banking, insurance,
real estate, hotels, building maintenance, printing, transportation) and public services
(highways, mass transit, schools, fire protection). Large cities also provide city wide
labor-market pooling. If fluctuations in the labor demands of different industries are
not correlated, the large city provides a stable level of total employment. When a
job disappears in one industry, it is likely to be replaced by a new job in another
industry. Search costs and moving costs are lower in large cities, so firms are able
to increase or decrease theirworkforces more easily.
As explained by Jacobs (1960), the urban environment also encourages innova-
tion. Urban areas contain people with diverse backgrounds and interests. In such an
environment, the exchange of ideas between people with different perspectives can
lead to innovations in product design and production methods. There is evidence
that knowledge spillovers are important. Glaeser, Kallal, Sheinkman, and Shleifer
(1992) provide evidence that knowledge spillovers contribute to the growth of cities,
with the spillovers occurring mainly between industries rather than within a particu-
lar industry. Rauch (1990) knowledge spillovers increase productivity and
finds that
wages, and and Hernderson (1993) find that knowledge spillovers
Jaffe, Trajtenber,
encourage innovation, as measured by patent activity.
Chapter 2 Why Do Cities Exist.' 33
Empirical Estimates of Agglomerative Economies
How important are localization economies and urbanization economies? One ap-
proach to measuring agglomerative economies is to estimate the effects of changes
in industry output and city size on labor productivity. The hypothesized relationship is
q = f(k,e, Q,N) (2-1)
where
q = Output per worker in a particular industry
k = Capital equipment per worker

e = Education level of workers (a measure of labor skill and productivity)
Q= Total output of the industry
N= Total population of the metropolitan area
Output per worker should increase with capital per worker and the education level of
economies exist, output per worker also increases with
the workforce. If localization
Q (industry output), and if urbanization economies exist, output per worker increases
with N (population). Statistical analysis can be used to estimate the independent
effect of changes in Q (industry output) on output per worker, that is, the increase in
output per worker per unit change in Q, holding constant the other determinants of
labor productivity (k, e, and N). Similarly, one can estimate the independent effect
of changes in N (city size) on output per worker.
Henderson (1988) uses this approach to estimate both localization and urbaniza-
tion economies. He measures localization economies as the elasticity of output per
worker with respect to industry output, defined as the percentage change in output
per worker divided by the percentage change in industry output. For the electrical
machinery industry, the localization elasticity is 0.05, meaning that a 10 percent in-
crease in the output of the electrical machinery industry increases output per worker
by about 0.50 percent. His elasticity estimates for other U.S. industries range from
0.02 for the pulp and paper industry to 0.11 for the petroleum industry. In con-
trast, Henderson's results suggest that urbanization economies are very small: the
with respect to city size is close to zero.

elasticity of labor productivity
Henderson's results have important implications for urban development. It ap-
pears that an increase in city size, by itself, does not increase productivity. Larger
cities are more productive because they have large concentrations of specific in-
dustries (localization economies), not simply because they are large (urbanization
economies). The results of O'hUallachain and Satterthwaite (1992) are consistent
with the view that localization economies are more important than urbanization
economies.
Mun and Hutchinson (1995) study agglomeration economies in the office sector,
using data from the city of Toronto. They find that agglomeration effects in the
office sector are much more powerful than those in manufacturing. For example, a
10 percent increase in the number of office firms increases the productivity of office
firms by about 2.7 percent (i.e., the elasticity of average productivity with respect to
the size of the office industry is 0.27). The increase in productivity is larger in cases
when the increase in office employment occurs in central locations. In addition, the
agglomeration effects are largest in the area of the city where the increase in office
employment occurs.
Despite the empirical results of Henderson and others, it may be premature to
conclude that localization economies are more important than urbanization econo-
mies. In a study of 19 manufacturing industries in 68 metropolitan areas. Carlino
(1987) found that urbanization economies are present in 13 of 19 industries, while
localization economies are present in 5 of the 19 industries. In a study of the fac-
torsbehind urban growth, Glaeser, Kallal. Sheinkman. and Shleifer (1992) found
evidence of strong agglomeration economies. Given the conflicting results of these
empirical studies, it seems that the relative importance of localization and urbaniza-
tion economies is still an open question deserving additional empirical work.
Innovations in Telecommunications and the Future of Cities

Will improvements in telecommunications make cities obsolete? Cities provide
physical proximity and thus facilitate face-to-face communication among the pro-
ducers of goods and services. Recent developments in telecommunications have
generated new options for communication, including electronic mail, faxes, the in-
ternet,and teleconferencing. According to some noted futurists, because these new
technologies are substitutes for face-to-face communication, cities will shrink and
may even disappear. Alvin Toffler (1980) suggested that telecommunications tech-
nology will allow people to escape the city to live in "electronic cottages" in rural
areas. Nicolas Negroponte (1995) predicts that "the transmission of place itself
will start to become possible through new virtual reality and telecommunications
technology" resulting in a decline in the importance of "special places" like cities.
Marshall McLuhan (1984) suggested that the city "as a form of major dimension
must inevitably dissolve like a fading shot in a movie."
The prediction that cities will decline or disappear is based on the notion that
telecommunication is a true substitute for face-to-face communication. According
toGaspar and Glaeser (1998), however, the two forms of communication may be
complementary. Many production relationships involve both face-to-face interac-
tionsand telecommunications via the telephone, e-mail, faxes, or the internet. An
improvement in telecommunication has two effects on the number of face-to-face
contacts.
• In a given relationship, some face-to-face encounters will be replaced by

telecommunication: an investment banker could e-mail a client instead of
meeting her for a brief conversation; the banker could fax a contract instead
of delivering it by hand.
• Easier communication may increase the number of relationships: an
investment banker could consult with more people in the process of
evaluating the merits of a particular project: the banker could handle more
projects.
Chapter 2 Why Do Cities- Exist/ 35
As long as each relationship requires some face-to-face communication, the net

effect on the number of face-to-face contacts is ambiguous. For example, if the
investment banker reduces the number of face-to-face contacts per relationship from
five to four but increases the number of relationships from six to eight, the number
of face-to-face contacts will increase from 30 to 32.
There is some evidence communication and telecommunica-
that face-to-face
were substitutes, we'd expect that the widespread
tion are not pure substitutes. If they
use of faxes and electronic mail would decrease the demand for business travel as
business workers substitute electronic communication for face-to-face contact. Be-
tween 1985 and 1995, a period during which faxes and e-mail were first developed
and then became ubiquitous, the volume of business travel actually increased by
more than 50% (Gaspar and Glaeser, 1998). Of course, other changes during this
period may have contributed to the growth of business travel, and it is difficult to
discern the contributions of different factors. But the facts on business travel provide
suggestive evidence that the two forms of communication are complements rather
than true substitutes.
More evidence concerning the complementarity of face-to-face contact and
telecommunication comes from data on the use of telephones, the first telecommu-
nication device. If the two types of communication are complements, face-to-face
contact will increase the demand for telecommunication, and people who are physi-
cally closer —and presumably see each other more often— will call each other more
often. In the 1970s, more than 40 percent of phone calls were made to places within
a 2-mile radius, and more than 75 percent were made to places within a 6-mile
radius. In Japan, the shorter the distance between two points, the more telephone
calls between those points. Also in Japan, there is a positive relationship between
phone usage and urbanization: people who live in large cities (population exceeding
100,000) spend more time on the phone than people who live in small cities or rural
areas. One interpretation of these facts is that physical proximity and thus the ease —
of face-to-face contact —generates greater demand for telecommunications because
two forms of communication are complements.
A more subtle issue concerns the effect of telecommunication on the complexity
of interactions in cities. Easier communication may allow greater specialization
in design and production, increasing the need for face-to-face communication to
coordinate the specialized efforts of the various people involved. In addition, easier
communication may increase the complexity of the information and ideas transmitted
between people, requiring more face-to-face contact to communicate these ideas.
Gottman (1977) discusses the effects of the telephone (the first telecommunication
device) on cities:
The telephone provides, when needed, quasi-immediate verbal communication between all
the interdependent units at minimum costs. ... It would have been very difficult for all these
complex and integrated networks [in cities] to work in unison without the telephone, which
made possible the constant and efficient coordination of all the systems of the large modern
city. . . . The telephone helped to make the city bigger, more exciting.
The same logic applies to more recent innovations such as faxes, e-mail, and the
internet.
What are the facts on innovations in telecommunications and urbanization?

During the recent period of rapid development of telecommunications, urbanization
actually increased. Based on recent trends, the proportion of the global popula-
tion living in cities is expected to increase from its current level of 45 percent to
61 percent by the year 2025. During this period, the proportion of population liv-
ing in cities is expected to increase from 73 percent to 83 percent in Europe: from
74 percent to 85 percent in Latin America; from 34 percent to 54 percent in Africa;
from 76 percent to 85 percent in North America; and from 70 percent to 75 percent
in the Pacific region.
It appears that innovations in telecommunication technology will not cause
cities to disappear. Some interactions among producers require face-to-face contact,
so there will always be a need for cities and the physical proximity they provide.
The number of face-to-face contacts will be affected by innovations in telecom-
munication, but both the direction and the magnitude of the change are unknown.
Although telecommunication is a good substitute for some types of face-to-face
communication may increase the need for face-to-face contact as
contact, easier
(1) more relationships are developed, and (2) relationships become more complex.
If the net effect is more rather than less face-to-face communication, innovations in
telecommunications could cause cities to grow rather than shrink.
Agglomerative Economies in Marketing: Shopping Externalities

This section explains how agglomerative economies in marketing affect the devel-
opment of cities. The previous section of the chapter explains how agglomerative
economies in production cause industrial firms to cluster, resulting in the develop-
ment of large industrial cities. This section explains how agglomerative economies
in marketing cause trading firms to cluster, resulting in the development of small
market-based cities and retail concentrations within cities.
A shopping externality occurs if the sales of one store are affected by the
location of other stores. Suppose that an isolated drugstore relocates next to a gro-
cery store. If both stores experience an increase in sales, they generate shopping
externalities: each store attracts consumers to the retail cluster, generating benefits
for the other store. These shopping externalities cause firms selling related prod-
ucts to form retail clusters. Some retail clusters cause the development of market
cities.Other clusters occur within large cities, generating downtown shopping areas.
malls, and shopping centers. There are two types of products that generate shopping
externalities: imperfect substitutes and complements.
Imperfect Substitutes
Two goods are imperfect substitutes if they are similar but not identical. Suppose that
you've decided to buy a new sports haven't decided whether to buy a Ford.
car. but
a Chevy, or a Honda. The sports cars ottered by the three companies are similar, but
Chapter 2 Why Do Cities I. ast? 37
differ in performance, shape, color, and gadgetry. Because the differences between
the cars are subtle,you must do your comparison shopping in person, spending time
and money traveling to the three car dealers. If the three dealers form a cluster, they
decrease the cost of comparison shopping, attracting consumers to the cluster.
There are many goods that are imperfect substitutes. Some examples are clothes,
shoes, jewelry, and electronic equipment. For these goods, the clustering of firms
selling similar products decreases shopping costs and attracts potential buyers. Some
retailers cluster in the city center, while others cluster in shopping centers and
malls.
Figure 2^4- illustrates the notion of agglomerative economies from comparison
shopping. Suppose that there are two auto dealers in a particular city, and they are
initially located far from one another. Each dealer has a fixed supply of 50 cars per
week, so the supply curves at each location are vertical at 50 cars. In the initial
equilibrium, each dealer faces the demand curve D' and sells its 50 cars at a price
of $8,000. Suppose that dealer M moves to a site next to the other dealer. The total
supply of cars at the location of the original dealer doubles, from 50 cars per week
to 100 cars. If the customers of dealer M follow the dealer to its new location, the
demand for cars also doubles: the demand curve shifts to the right by 50, from D'
Figure 2—4 Agglomerative Economies from Comparison Shopping
50 100 1 10
Number of cars per week

In the initial equilibrium with two auto dealers located far from one another, each dealer supplies
50 cars at a price of $8,000. If dealer M
relocates next to dealer S, the supply of cars at location
S doubles, from 50 to 100. The demand more than doubles because the cluster attracts comparison
shoppers who would otherwise not patronize either dealer.
to D" . If the demand and supply curves shift to the right by the same amount, the
relocation of dealer M does not affect the market price of cars.
The shift of the demand curve from D" to D* incorporates shopping exter-
nalities. The clustering of auto sellers decreases the shopping costs of
consumers
who engage comparison shopping. Therefore, the cluster attracts consumers who
in
would otherwise not patronize either of the dealers. These additional consumers
cause another rightward shift of the demand curve, from D" to D'. In this example,
an additional 10 consumers patronize the two dealers. If the dealers continue to
supply only 50 cars each, the equilibrium price increases from S8.000 to S9.000.
Of course, another option is to increase the number of cars sold. In either case, the
extra shift of the demand curse from comparison shoppers generates benefits for the
dealers in the cluster, resulting in a higher price or a larger volume.
Complementary Goods
Clustering also occurs when firms sell complementary goods. Complementary goods
are often purchased on the same shopping trip. For example, if a consumer purchases
pants and shoes on the same trip, his travel cost will be lower if the pants store is near
the shoe store. The shoe from the presence of the pants store because
store will benefit
together they provide one-stop shopping for consumers. Because of the benefits of
one-stop shopping, firms selling complementary goods cluster in shopping centers,
malls, and city centers.
Retail Clusters
The agglomerative economies associated with shopping generate several types of

retail clusters. For some goods, retailers form retail strips along arterial streets, for
example, auto rows. Other retailers cluster in shopping centers, malls, and city cen-
ters.These clusters provide both a mix of imperfect substitutes (different jewelry
stores) and complements (food and drugs, clothing and shoes), allowing both com-
parison shopping and one-stop shopping. Retailers that choose isolated locations
instead of clusters sell goods that are not subject to shopping externalities.
Summary
1. Employment opportunities are concentrated in cities for four reasons:
comparative advantage, scale economies in transportation, internal scale
economies in production, and agglomerutive economies.

2. Comparati\e advantage from differences in productivity. Trade occurs if
results
the differences in productivity are largeenough to offset transport costs. Trading
firms (and urban areas) develop if there are scale economies in transportation.
3. Internal scale economies, which result from factor specialization and indivisible
inputs, generate a negatively sloped long-run average-cost curve.
a. If scale economies are large relative to transport costs, goods are produced in
factories.
b. Factory workers economize on travel costs by locating near factories,
causing the development of factory cities.
4. The English wool cloth industry illustrates the role of comparative advantage
and scale economies in the development of cities.
a.The opening of England to the continent allowed the country to exploit its
comparative advantage in clothmaking, causing the development of cities
involved in the production and trade of cloth.
b. The development of the water-powered fulling mill shifted production to
rural areasand caused the development of villages and small trading cities.
c. The industrial revolution produced mechanized spinning and weaving,
generating scale economies that caused the development of large factories
and factory cities.
5. The sewing machine increased scale economies in the production of clothing

and shoes, causing home production of clothes and shoes to be replaced by
factory production. Cities developed around the clothing factories and shoe
factories.
6. Localization economies occur if the production cost of a particular firm

decreases as industrywide output increases. There are three sources of
localization economies.
a. If there are scaleeconomies in the production of an intermediate good for
which transportation costs are relatively high, several firms share a supplier
of the intermediate input and form a cluster around the input supplier.
b. If output per firm varies from year to year, a cluster of firms facilitates the
transfer of workers from one firm to another (labor-market pooling).

c. A cluster of firms improves communication, facilitating the rapid exchange
of information and the diffusion of innovations (knowledge spillovers).
7. Urban areas, with their rich set of intermediate inputs, provide a nurturing
environment for immature industries (the incubator process).
8. Urbanization economies occur if the production cost of a particular firm
decreases as the total output of the urban area increases. Urbanization
economies result from the sharing of input suppliers (business services and
public services), labor-market economies, and communication economies.
9. A shopping externality occurs if the sales of a particular store increase as other
retailers move closer to the store. These agglomerative economies in marketing
cause the clustering of retailers.
a. Comparison shopping causes the clustering of firms selling imperfect

substitutes.
b. One-stop shopping causes the clustering of firms selling complementary
goods.
c. Retail clusters occur in downtowns, shopping centers, malls, and retail
strips.
Exercises and Discussion Questions

1 . The table below summarizes the productivity of workers in wheat and cloth
production in two parts of a region.
Output per Labor Hour Opportunity Cost of Production
East West East West
Wheat 1 12
Cloth 1 3
a. Complete the table by providing the opportunity costs of producing wheat

and cloth in the East and the West.
b. Assume that transport costs are zero and that the exchange rate is two
bushels of wheat for one yard of cloth. If a western household switches one
hour from cloth production to wheat production and exchanges half of its
additional wheat for cloth, will the household be better off?
c. Suppose that the time required to execute the trade in (b) is two-thirds of an
hour. Is trade still beneficial? At what transaction cost (time per trade)
would the net gain from trade be zero?
d. Under what conditions will the differences in labor productivity cause the
development of cities?
2. Consider a region that has a single trading city (a city that developed as a result
of comparative advantage and scale economies in transportation). There are no
shopping externalities (no imperfect substitutes or complementary goods), and
communication between households is costless. Suppose that a matter
transmitter is introduced into the region. The transmitter costlessly transports
goods, but cannot be used to transport people. The transmitter is cheap enough
that every household can purchase one at a relatively low cost.
a. Explain the effects of the matter transmitter on urban development.
Specifically, will the region's trading city grow, shrink, or disappear?
b. How would your answ er to (a ) change if cloth from one household is not a
perfect substitute for cloth from other households
3. Consider a country with two regions that are separated by a mountain range.
Initially, there are no cities in the country. Suppose that a tunnel is bored
through the mountain, decreasing travel costs between the two regions.
Under what conditions will the tunnel cause the development of trading
cities?
4. Consider a region that has a single factor)' city (a city that developed as a result
of scale economies in production). Suppose that the introduction of pump
sneakers increases the walking speed from four round-trip miles per hour to
eight. Assume that the new type of shoes sells for the same price as the old
type. How will the new footwear affect the size of the region's citj ? Will its
population increase or decrease?
5. Region F has a single factory city (a city that developed as a result of scale
economies in production). Suppose that a matter transmitter is introduced into
the region. The transmitter can costlessly transport goods, but cannot be used
to transport people. The transmitter is cheap enough that every household can
purchase one at a relatively low cost. Explain the effects of the matter
transmitter on urban development in the region. Specifically, will the city
grow, shrink, or disappear?
6. Consider a region that produces and consumes fruit and ice. All resources are
distributed uniformly throughout the region, and all people are equally
productive in producing fruit and ice. There are scale economies in the
production of ice, causing the development of an ice factory and a factory city.
Suppose that a small refrigerator is introduced and imported into the region,
providing an alternative to the ice purchased from ice factories. Explain the
effects of the refrigerator on (a the market area of the
) ice factory and (b) the
size of the city surrounding the ice factory.
7. Consider Retireland, a region where most residents are retired. Residents of

the region consume a single good (food), which is imported from another
region and sold in vending machines. Will there be any cities in Retireland? If
so, how will they differ from cities in regions where most people work?
8. Most of the dresses made in the United States are produced by large firms in
suburban areas, not by small firms in city centers. Is this consistent with the
notion of industry externalities and clusters? Hint: Is there a difference
between a KMart dress and a dress produced by a small dressmaker?
9. The conventional wisdom economic development is: "Don't put all
for urban
your eggs in one basket. Diversify the economy." To explain the idea of
diversification, consider old McDonald, who must carry a dozen eggs from the
barn to the house. The ground between the barn and the house is slippery, so
there is a 50 percent chance that McDonald will slip on a given trip and break
all the eggs in his basket. Consider two strategies: a one-basket strategy (a
single trip with all 12 eggs) and a two-basket strategy (two trips, with 6 eggs
per trip).
a. List all of the possible outcomes under each of the strategies.

b. What is the expected number of delivered (unbroken) eggs under each
strategy?
c. What are the trade-offs between the two strategies? If you were McDonald,
which strategy would you adopt?
d. What are the lessons for economic development strategies?
10. According to the conventional wisdom concerning urban economic
development, a city should develop a diverse economy with a large number of
industries. Evaluate the merits of the conventional wisdom in light of the
empirical evidence concerning the magnitudes of localization and urbanization
economies.
1 1 Consider a 10-firm industry that produces computer equipment, a set of goods
with rapidly changing demand and production technology. The industry has
the following characteristics:
i. The 10 computer equipment using labor and raw materials.

firms produce
ii. Raw all locations at the same price).
materials are ubiquitous (available at
Hi. Each firm produces one new product per year, and each product becomes
obsolete after a year.
iv. Only 3 of the 10 new products will be successful (sell more than a trivial
amount).
v. The monetary and time costs of switching a worker from one firm to
another are zero, regardless of the spatial distribution of firms.
a. Will the firms in the industry form a cluster? Why or why not?
b. How would your answer to (a change ) if workers incur moving costs when
they switch from one firm to another?
12. Suppose that the outputs of beauty shops and pet-grooming salons are
complementary, providing one-stop shopping for personal and pet
maintenance. Betty Beehive is thinking about moving her beauty shop from an
isolated location to a vacant building next to Peter's pet-grooming shop. In
making her decision, she makes the following assumptions:
/'.
If Betty moves, she will keep all of her current customers (20 people per
week), and attract 25 percent of Peter's current customers.
ii. Peter currently has 60 customers per week.
Hi. Excluding rent. Betty's profit per beauty treatment is $10.
iv. The weekly rent at the new location is $200 higher than Betty's current
rent.
a. If moves her beauty shop, will her profits increase or decrease?
Betty
b. Suppose that if Betty makes the move, 50 percent of her original customers
will switch from George's grooming salon to Peter's. If Peter's profit per
treatment is $8, how much would he be willing to pay Betty to make the
move? Will the payment be enough to induce Betty to make the move?
13. Consider a city with two auto sellers, a Toyota dealer and a Honda dealer.
Initially, the distance between the two sellers is three miles. The Toyota dealer
wants to relocate to a site adjacent to the Honda dealer and submits a rezoning
request to the city council. The Honda dealer responds to the rezoning request
with the following statement: "One of the lessons from Econ 100 is that an
increase in supply will decrease price. If the Toyota dealer moves to the site
adjacent to my dealership, the local supply of cars will increase and I'll have to
cut my prices to sell the same quantity of cars." Critically appraise the Honda
dealer's statement. If the statement is incorrect, what's wrong with the
reasoning? Illustrate your answer with a graph.
References and Additional Readings

History of the English Woolmaking Industry
Jenkins, D. T. The West Riding Wool Textile Industry. 1770-1835. Leeds: Munev & Sons.
1975. Describes 'he transition of the clothmaking industry from the putting-out system
to factory production
Ponting, Kenneth G. The Woolen Industry of South-West England. Bath: Adams & Dart.
1971. Describes the early history of the English woolmaking industry.
Localization Economies
Clapp, J. M. "Endogenous Centers: A Simple Departure from the NUE Model." Papers of
the Regional Science Association 54 ( 1984), pp. 13-24. Describes the process through
which contacts between firms cause the development of a CBD and subcenters.
Hall, Peter,and Ann Markusen. Silicon Landscapes. Winchester. Mass.: Allen & Unwin,
1985. Discusses the role of external scale economies in the location of high-technology
industries.
Hanson, Gordon H. "Agglomeration, Dispersion, and the Pioneer Firm." Journal of Urban
Economics 39 (1996), pp. 255-81. Describes the location patterns of the Mexican
garment industry, with marketing and design in industry centers but assembly in
low-wage regions.
Hoover, Edgar M. Location Theory and the Shoe and Leather Industries. Cambridge.
Mass.: Harvard University Press, 1937. Discusses the role of scale economies in the
location of shoe and leather industries.
Krugman, Geography and Trade. Cambridge, Mass.: MIT Press, 1991. A series of
Paul.
lectureson the topic of "economic geography," defined by Krugman as "the location of
production in space." Discusses the theory of localization economies and empirical
evidence of geographic concentration of industries.
Lyons, Donald I. "Agglomeration Economies among High-Technology Firms in Advanced
Production Areas: The Case of Denver/Boulder." Regional Studies 29 ( 1995). pp.
265-78. Explores the agglomeration advantages generated by high-technology firms in
the Denver/Boulder region.

O'Hara, D. J. "Location of Firms within a Square Central Business District." Journal of
Political Economy 85 (1977), pp. 1 189-207. Discusses how interfirm contacts cause
the development of a central business district.
Saxenian, AnnaLee. Regional Advantage: Culture and Competition in Silicon Valley and
Route 128. Cambridge, Mass.: Harvard University Press, 1994. Explores the recent
histories of two concentrations of high-technology firms and explains the different
experiences in terms of differences in industrial structure and culture.
Stephens, John D.. and Brian Holly. "City System Behavior and Corporate Influences: The
Headquarters Location of U.S. Industrial Firms 1955-1975." Urban Studies 18 (1981).
Tauchen, H., and Anne Witte. "An Equilibrium Model of Office Location and Contact
Patterns." Environment and Planning 15 (1983). pp. 131 1-26. Discusses how interfirm
contacts cause the development of a central business district.
Vernon, Raymond. "External Economies." In Readings in Urban Economics, ed. M. Edel
and J. Rothenberg. New York: Macmillan. 1972. A discussion of agglomerative
economies for industry and clusters.
Urbanization Economies
Carlino, Gerald A. Economies of Scale in Manufacturing Location. Boston: Martinus

Nijhoff, 1978.
Higgs, Robert. "American Inventiveness, 1870-1920." Journal of Political Economy 79
(1971), pp. 661-67.
Jacobs, Jane. The Economy of Cities. New York: Random House, 1960. Discusses why
many innovations are developed in urban areas.
Noyelle, Thierry, and Thomas Stanback. The Economic Transformation of American Cities.
Totawa, N.J.: Rowman and Allanheld. 1984.
Sveikauskas, L. "Interurban Differences in the Innovative Nature of Production." Journal of
Urban Economics 6 (1979), pp. 216-27.
Empirical Estimates of Localization and Urbanization Economies
Carlino, Gerald A. "Productivity in Cities: Does City Size Matter?" Business Review of the
Federal Resen'e Bank of Philadelphia (November/December 1987), pp. 3-12.
Estimates localization and urbanization economies for manufacturing.
Ciccone, Antonio, and Robert E. Hall. "Productivity and Density of Economic Activity."
American Economic Review 86 (1996), pp. 54-70. Empirical estimation of the
relationship between urbanization and productivity.
Glaeser, Edward L.; Hedi D. Kallal; Jose A. Scheinkman; and Andrei Shleifer. "Growth in
Journal of Political Economy 100 1992). pp. 126-52. Explores the role of
Cities." ( 1
knowledge spillovers in urban growth and suggests that urbanization economies

from these spillovers contribute to growth.
resulting
Henderson, V. "Efficiency of Resource Usage and City Size." Journal of Urban
J.
Economics 19 (1986), pp. 47-90. Estimates the magnitude of localization and

urbanization economies.
— . Urban Development: Theory, Fact, and Illusion. New York: Oxford University
Press, 1988. Chapter 5 provides estimates of localization economies and urbanization
economies for the United States and Brazil. Other chapters discuss the theory of
systems of cities and public policy implications.
Jaffe, A.; M. Trajtenber: and R. Hernderson. "Geographic Localization of Knowledge
Spillovers as Evidenced by Patent Citations." Quarterly Journal of Economics 107
(1993), pp. 577-98. Uses patent data to explore the effects of knowledge spillovers.
Moomaw, Ronald L. "Productivity and City Size: A Critique of the Evidence." Quarterly
Journal of Economics 96 (1981), pp. 675-88. Discusses statistical problems with the
Segal and Sveikauskas analysis of urbanization economies.
Mun, Se-il, and Bruce G. Hutchinson. "Empirical Analysis of Office Rent and
Agglomeration Economies: A Case Study of Toronto." Journal of Regional Science 35

(1995). pp. 437-55. Estimates the value of agglomeration economies, defined as
changes in productivity caused by increases in the number of office firms. Suggests
that the agglomeration economies in the office sector are much larger than those in
manufacturing.
O'hUallachain. Breandan, and Mark A. Satterthwaite. "Sectoral Growth Patterns at the
Metropolitan Level." Journal of Urban Economics 3 1992). pp. 25-58. Explores the
I (
clustering of firms in particular industries and provides evidence of localization

economies.
Rauch, J. "Productivity Gains from Geographic Concentration of Human Capital: Evidence
from the Cities." Journal of Urban Economics 34 (1993), pp. 380—100. Empirical study
of localization economies showing that knowledge spillovers increase productivity.
Segal. David. "Are There Returns to Scale in City Size?" Review of Economics and Statistics
58 (July 1976). pp. 339-50. Estimates the magnitude of urbanization economies.
C 'hapter 2 Why Do Cities Exist 45
Sveikauskas, L. "The Productivity of Cities." Quarterly Journal of Economics 89 (1975),

pp. 393-413.
Venables, Anthony J. "Economic Integration and Industrial Agglomeration." Economic and
Social Review 26 (1994), pp. 1-17. Reviews recent research on industrial location and
suggests that the lowering of trade barriers will encourage industries to agglomerate in
a few locations.
Telecommunications and the Future of Cities
Gaspar, Jess, and Edward L. Glaeser. "Information Technology and the Future of Cities."
Journal of Urban Economics 43 (1998), pp. 136-56. Explores the issue of whether
telecommunications and face-to-face contact are substitutes or complements.
Gottman, "The Telephone and the Structure of the City." In The Social Impact of the
J.
Telephone, ed. I. De Sola Pool. Cambridge, Mass.: MIT Press, 1977. Explains how the
telephone facilitates the coordination of activities within cities and suggests that the
telephone increased the size of cities.
Graham, Stephen. "Telecommunications and the Future of Cities: Debunking the Myths."
Cities 14 (1997), pp. 21-9. Attempts to debunk five myths concerning the effect of
telecommunications on cities.
McLuhan, H. Understanding Media — The Extension of Man. London: Sphere, 1964.

Suggests that telecommunications technology will cause cities to decline or disappear.
Negroponte, N. Being Digital: The Roadmap to the Information Superhighway. London:

Hodder and Stoughton, 1995. Predicts that innovations in telecommunication will
cause cities to decline.
Toffler, A. The Third Wave. New York: Morrow, 1980. Suggests that telecommunications
technology will allow people to escape the city to live in "electronic cottages" in rural
areas.
Agglomerative Economies in Shopping
Eaton, B. Curtis, and Richard G. Lipsey. "Comparison Shopping and the Clustering of
Homogeneous Firms." Journal of Regional Science 19 (1979), pp. 421-35. Discusses
the role of comparison shopping in the location of retailers.
Stahl, Konrad. "Differentiated Products, Consumer Search, and Locational Oligopoly."
Journal of Industrial Economics 31 (1982a), pp. 97-1 13. Discusses the effects of
consumer search costs on the location of retailers.
. "Location and Spatial Pricing Theory with Nonconvex Transportation Cost
Schedules." Bell Journal of Economics 13 (1982b), pp. 575-82. Shows that consumers
are attracted to retail sites providing many goods, explaining the clustering of retailers.
General
Toynbee, Arnold. Cities on the Move. New York: Oxford University Press, 1970a.
Chapter 3
w HERE DO
CITIES DEVELOP?
Let the river roll which way it will, cities will rise on its banks.
Ralph Waldo Emerson
•O hapter 2 showed that cities develop because of comparative advantage, scale

LJ economies in transportation and production, and agglomerative economies.
Cities develop around the concentrations of employment generated by firms, so the
location choices of firms play a role in the location of cities. This chapter explores
the location decisions of two types of firms: industrial firms and commercial firms.
raw
Industrial firms (e.g., sawmills, breweries, manufacturers, and bakeries) process
materials and intermediate inputs into outputs.Commercial firms trade goods rather
than produce them. The location decisions of industrial and commercial firms cause
thedevelopment of various types of cities. This chapter also discusses the role of
government in the location of cities.
Commercial Firms and Trading Cities

The location decisions of commercial firms cause the development of trading cities.
Trading firms collect goods from suppliers and distribute the goods to consumers,
providing employment for salespeople and other intermediaries. Trading firms typi-
cally locate at transshipment points (ports, crossroads, railroad junctions, river junc-
tions) because such locations offer convenient points for the collection and distri-
bution of goods. Transportation firms move goods from place to place, providing
employment at transshipment points for truckers, sailors, and longshoremen. Firms
providing business services (banking, insurance, bookkeeping, machine repair) lo-
cate close to the trading and transportation firms that use their services.
The American fur trade was an important force in the development of cities
along the Mississippi River. New Orleans, located at the mouth of the Mississippi
47
48 Pan I Market Forces in the Development of Cities
River, provided a convenient transshipment point for American furs destined for
European markets. from New Orleans established St. Louis,
In 1764. fur traders
located at the junction of the Missouri and Mississippi rivers, as a central substation
for the fur trade. By 1800. there were over 1.000 people in St. Louis, living in the
middle of an almost unpopulated continent.
The experience of New York City illustrates the importance of trade in the
development of cities. The city, which began as a fur-trading post for the Dutch, was
taken over by the British in 1664. The city grew rapidly, along with other eastern
cities, as a result of the flour trade with the West Indies. At the time of the War of
1812, New York and Philadelphia were about the same size. Following the war. New
York grew rapidly and became the country's dominant city. Why?
The rapid growth of New York was caused by increases in trading activity,
which were caused by two factors. First. New York State built the Erie Canal, con-
necting New York City with the Great Lakes. The Erie Canal opened vast market
areas to New York traders. Second, traders in New York developed innovative trad-
ing practices, one of which was the auction sale. A British manufacturer consigned
his goods to a U.S. auctioneer, who then sold the goods to U.S. merchants. The
auction sale decreased prices because it eliminated the British exporter and the
traditional U.S. importer. While Philadelphia and Boston passed laws that discour-
aged auction sales. New York allowed them. United States merchants responded
by buying more goods from New York traders, so the city grew at the expense of
Philadelphia and Boston. New York traders were also the first to establish regularly
scheduled shipping across the Atlantic, which attracted more traders and shippers
to the city.
Cities also developed as a result of the railroad. The spread of the railroad after
1830 gave merchants in Boston. Philadelphia, and Baltimore access to midwestern
markets, so these cities diverted some trading activity from New York. In 1868.
Kansas land speculators chose a site for a new city along the Chisholm Trail (the
route for cattle drives from Texas to Kansas railheads). They persuaded the railroad
company to extend its tracks to the site. By 1872. thousands of carloads of cattle
were being shipped from the city, and Wichita was a booming cow town.
Transfer-Oriented Industrial Firms

A transfer-oriented firm is defined as one for which transportation cost is the
dominant factor in the location decision. The firm chooses the location that minimizes
total transport costs, defined as the sum of procurement cost and distribution cost.
Procurement cost is the cost of transporting raw materials from the input source to
the factory. Distribution cost is the cost of transporting the firm's output from the
factory to the consumer.
The classic model of a transfer-oriented firm has four assumptions:
1 . Single output. The firm produces a fixed amount of a single good. The
output is shipped from the factory to a marketplace at point M.
Chapter .? Where Do Cities Develop.' 49
2. Single transferable input. The firm may use several inputs, but only one
input is shipped from an input source (point F) to the factory. All other
inputs are ubiquitous (available at all locations at the same price).
3. Fixed-factor proportions. The firm produces amount of output

its fixed
with a fixed amount of each input. In other words, firms do not substitute
between inputs in response to changes in relative prices, but use a single
recipe for producing the good.
4. Fixed prices. The firm is so small that it does not affect the prices of its
inputs or outputs.
Under these four assumptions, the firm maximizes its profit by minimizing its trans-
portation costs. The firm's profit equals total revenue (price times the quantity of
output) less input costs and transport costs. Total revenue is the same at all locations
because the firm sells a fixed amount of output at a fixed price. Input costs are the
same at all locations because the firm buys a fixed amount of each input at fixed
prices. The only costs that vary across space are procurement cost (input transport
costs) and distribution cost (output transport costs). Therefore, the firm will choose
the location that minimizes its total transport costs.
The firm's location choice is determined by the outcome of a tug-of-war. The
firm is pulled toward the input source by low procurement costs, and pulled toward
the market by low distribution costs. For the resource-oriented firm, the pull toward
the input source is relatively strong, so the firm locates near its raw material source.
For the market-oriented firm, the pull toward the market is relatively strong, so the
firm locates near its marketplace.
Resource-Oriented Firms
Table 3-1 shows the transport characteristics of a resource-oriented firm. The firm
makes baseball bats, using five tons of wood to produce three tons of bats. The firm
is involved in a weight-losing activity in the sense that the output is lighter than
the transferable input. The monetary weight of the input is defined as the physical
weight of the input (five tons) times the transportation rate ($1 per ton per mile), or
$5 per mile. Similarly, the monetary weight of the output is three tons times $1, or
$3 per mile. The firm is considered a resource-oriented firm because the monetary
weight of its transferable input exceeds the monetary weight of its output.
Table 3-1 Physical and Monetary Weights of a

Resource-Oriented Firm
Input (wood) Output (bats)
Physical weight (tons) 5 3

Transport rate (cost per ton per mile) $1 $1
Monetary weight (physical weight times rate) $5 $3
Figure 3-1 Total Transport Costs for a Resource-Oriented Firm
Market (M)
4 6
Forest 1F1 x (distance from forest)
Total transport cost (the sum of procurement cost and distribution cost) is minimized at the forest because the
monetary weight of the input ($5) exceeds the monetary weight of the output ($3). The weight-losing activity locates
at its source of raw materials.
Figure 3-1 shows the firm's transportation costs. If x is the distance from the
forest to the factory, procurement cost is
PC tt x (3-1)
that is. monetary weight of the input (weight it, times rate t, times the distance
the )
between and factory. The slope of the procurement-cost curve is the monetary
forest
weight of the input, so PC rises by $5 per mile, from zero at the forest to $50 at
the market 10 miles away. If Xm is the distance between the forest and the market,
distribution cost is
DC = w t (x M -a) (3-2)
thatis. the monetary weight of the output (weight »•„ times rate /„) times the distance
from the factory to the market. The slope of the distribution-cost curve is the monetary
weight of the output, so DC decreases by $3 per mile, from $30 at the forest (10
miles from the market) to zero at the market.
Total transport cost is the sum of procurement cost and distribution cost, as
shown in Figure 3—1. Total transport cost is minimized at the forest site ($30). If the
firm moves one mile away from the forest toward the market, its distribution cost
decreases by $3 (the monetary weight of the output) but procurement cost increases
by $5 (the monetary weight of the input), so total transport cost increases by $2.
Total cost isminimized at the forest site because the monetary weight of the input
Chapter 3 Where Do Cities Develop? 51
exceeds the monetary weight of the output. The resource-oriented firm locates near
its input source.
Weight-Losing Activities. The bat firm is resource-oriented because it is a weight-

losing activity, using five tons of wood to produce only three tons of bats. The cost
of transporting wood is large relative to the cost of transporting the finished output,
so the firm saves on transport costs by locating near the forest. In this case, the
tug-of-war is won by the input source because there is more physical weight on the
input side.
There are many other examples of weight-losing firms. Beet-sugar factories
locate near sugar-beet farms because one pound of sugar beets generates only about
2.7 ounces of sugar. Onion dehydrators locate near onion fields because one pound of
fresh onions becomes less than one pound of dried onions. Ore processors locate near
mines because they use only a fraction of the materials extracted from the ground.
Other Resource-Oriented Activities. Some firms locate near input sources be-
cause inputs are relatively expensive to ship. Suppose that the physical weight of the
transferable input equals the physical weight of the output, but the unit cost of ship-
ping the input exceeds the unit cost of shipping the output. The firm will locate near
its input source because the monetary weight of the input will exceed the monetary
weight of the output. The input will be more expensive to ship if it is more bulky,
perishable, fragile, or hazardous than the output. Hoover (1975) provides several
examples of such activities:
1 Cotton baling. The input (fluffy cotton) is more bulky than the output
(baled cotton). Since the cost of shipping a ton of fluffy cotton exceeds the
cost of shipping a ton of compacted cotton, the cotton baler is
resource-oriented and locates near the cotton fields.

2. Canning. The input (raw fruit) is more perishable than the output (canned
fruit). Since the cost of shipping one ton of perishable fruits (in refrigerated
cars) exceeds the cost of shipping canned goods, canners locate near the
fruit farms.
3. Skunk deodorizing. The input (fully armed skunks) is more fragile and
hazardous than the output (disarmed skunks). Since the cost of shipping a
ton of armed skunks exceeds the cost of shipping a ton of disarmed ones,
the skunk deodorizer will locate near the skunk source.
A firm with a relatively bulky, perishable, fragile, or hazardous input will locate near
its input source. The tug-of-war between input forces and output forces is won by the
input source, not because the input is heavier, but because it is more expensive to ship.
Market-Oriented Firms
For a market-oriented firm, the cost of transporting output is relatively large, so the
firm locates near its output market. Table 3-2 shows the transportation characteristics
Table 3—2 Physical and Monetary Weights of a Market-Oriented Firm
Input (sugar) Output I soft drinks)
Physical weight (tons) 4

Transport rate (cost per ton per mile) $1 $1
Monetary weight (physical weight times rate) $1 $4
Figure 3-2 Total Transport Costs for a Market-Oriented Firm
. Total transport cost = PC + DC
Distribution cost
Market I.U)
4 6
x (Distance from input source)

Sugar plantation (F)
{PC + DC) is minimized at the market because the monetary weight of the output ($4)
Total transport cost
exceeds the monetary weight of the transferable input ($1 ). The weight-gaining activity locates at its market.
of a market-oriented firm. The bottling company uses one ton of sugar and three tons
of water (a ubiquitous input) to produce four tons of soft drinks.
Figure 3-2 shows the firm's transportation costs. The monetary weight of the
output exceeds the monetary weight of the input by $3. As the firm moves toward
the market, total transport cost decreases by $3 per mile, dropping from $40 at the
sugar plantation (point F) to $10 at the market. The firm is market-oriented because
themonetary weight of the output exceeds the monetary weight of the input.
The bottling firm is market-oriented because it is a weight-gaining activity.
The firm adds three tons of local water to one ton of sugar to produce four tons
of soft drinks, so output is heavier than the transferable input (sugar). In this case,
the tug-of-war is won by the market because there is more physical weight on the
output side.
CluiptcrS Where Do Cities Develop? 53
Some firms locate near their markets because their output is relatively expensive
to ship. If the physical weights are equal but the output is more costly to ship, the
monetary weight of the output willexceed the monetary weight of the input, and the
firm will locate near its market. The cost of shipping the output will be relatively
high if the output is relatively bulky, perishable, fragile, or hazardous. Hoover (1975)
provides several examples of such firms:
1. Automobile assembly. The output (assembled automobiles) is more bulky

than the inputs (metal, plastic, and rubber parts). The cost of shipping one
ton of assembled automobiles exceeds the cost of shipping the ton of the
metal, plastic, and rubber that goes into the automobiles.
2. Bakery. The bakery produces a relatively perishable output: although flour

can be stored for months without spoiling, bread becomes stale after just a
few days. A bakery locates near its market because it would be expensive to
ship bread from a site near the flour mill to the marketplace in a timely
manner.
3. Weapons. A weapons producer combines harmless inputs into a lethal
output. The firm locates near its output market to avoid transporting the
hazardous (or fragile) output long distances.
A firm with a relatively bulky, perishable, fragile, or hazardous output will locate
near its market. The tug-of-war is won by the output, not because it is heavier, but
because it is more expensive to ship.
Intermediate Location
The analysis of the transfer-oriented firm suggests that a firm will locate either at its
input source (the resource-oriented firm) or market (the market-oriented firm).

its
Will a transfer-oriented firm ever choose a location between its input source and its
market?
A all the sites between the input source and the
firm will be indifferent about
market two conditions are met. First, the monetary weight of the input is equal to
if
the monetary weight of the output. Second, unit transport costs are independent of
the distance shipped. If these two conditions are met, the slope of the procurement-
cost curve will be equal to the slope of the distribution-cost curve, and the total-cost
curve will be a horizontal line. Total transport costs will be the same at all locations,
so the firm will be indifferent about all locations between its input source and the
market.
The possibility of an intermediate location is eliminated by scale economies in
transportation. Scale economies arise for two reasons:
1. Terminal costs. The fixed costs of a shipment are the costs of loading and
unloading the goods and the cost of paperwork for the shipment. Because
these costs are independent of the distance shipped, the average shipping
cost decreases as the distance shipped increases.
Figi re 3-3 Transport Cost with Declining Unit Costs
Market (A/)
v i Distance from input source)

Input source (F)
If there are terminal costs and declining unit transport costs, total transport cost ( PC + DC ) is minimized at either
the input or the market. Although the monetary weight of the input equals the monetary weight of the output, the
firm locates at either F or M, not somewhere between the two points.
2. Line-haul economies. The shipping cost per mile decreases as the distance
shipped increases, reflecting the efficiencies from using different modes.
Firms use trucks (relatively high cost per mile) for short hauls, trains (lower
costs per mile) for medium-length trips, and ships (low cost per mile) for
long hauls.
Figure 3-3 shows procurement cost and distribution cost when there are scale
economies in transportation. There is a fixed terminal cost per shipment, and the
unit costs of shipping (/, and ta ) decrease as the distance shipped increases. The
slope of the procurement-cost curve decreases as the firm moves away from the in-
put source, and the slope of the distribution-cost curve decreases as the firm moves
away from the market. The total-cost curve is no longer a straight line, but is bowed
outward in the middle, so total costs are lower at the endpoints (the input source or
the market) than at intermediate locations.
The presence of scale economies reinforces the tendency for firms to locate at
input sources and markets. If there are scale economies in transportation, it is more
efficient to make one long trip (shipping the input or the output the entire 10 miles
between the input source and the market) instead of two short trips (shipping the
input 5 miles and the output 5 miles).
Chapter 3 Where Do Cities Develop'.' 55
Implications for the Development of Cities
The analysis of the transfer-oriented firm provides a simple rule for predicting the
location choices of firms. If the monetary weight of the input exceeds the monetary
weight of the output, the firm will locate at its input source. If the reverse is true,
the firm will locate at its market. Of course, this rule is based on a simple model
in which the firm has only a single transferable input and a single output. What
about location choices involving a firm with several input sources and markets? The
general principles of this section apply to these more realistic cases. The higher the
cost of shipping inputs or outputs to a particular location, the greater the pull toward
that location.
The analysis of the transfer-oriented firm predicts the development of two types
of cities. Resource-oriented firms locate near their raw material sources, causing
the development of resource-based cities. Some examples are lumber towns, steel
towns, and food-processing towns. The analysis also suggests that as a city grows,
it will attract market-oriented firms. The location decisions of these market-oriented
firms often generate cities that serve as regional market centers.
The Principle of Median Location

The classic model of the transfer-oriented firm assumes that the firm has a single
input source and a single market. This section examines the location decisions of
firms with several input sources and several markets. The most important concept is
the principle of median optimum location
location. According to this principle, the
for a firm with several inputs and outputs is the median transport location, the
location that splits the total monetary weight of the firm into two equal halves. At
the median transport location, half the monetary weight comes from one direction,
and half the monetary weight comes from the other direction.
Location Choice with Multiple Markets
To explain the location choice of a firm with several markets, consider Ann, who
makes and delivers pizzas. In choosing a site for her pizza parlor, Ann must consider
the following:
1. Ubiquitous inputs. All inputs (labor, dough, toppings) are ubiquitous

(available at all locations for the same price), so input transport costs are
zero.
2. Pizza consumers. Ann's customers are located along a highway. The price
of pizzas is fixed, and every consumer demands one pizza per day.
3. Delivery costs. Ann delivers the pizzas to her customers at no charge,
making one trip per customer per day. The delivery cost is 50 cents per
pizza per mile.
Ann will choose the location that minimizes total delivery cost.
56 Part I Market Forces in the Development oj Cities
Figure 3-4 Pizza Delivery and the Principle of Median Location

W X Y S Z
I 1 1 1 1
Distance from IV (miles)
Number of consumers 2 8 1 10
Monetary weight $1 $4 SO. 5(1 $5

Assumptions
1 Every customer purchases one pizza.
2. The deliver) cost per pizza per mile is 5(ty.
3. The firm delivers one pizza at a time.
Ann >' because it is the median location: she delivers 10 pizzas

will locate her pizza parlor at point to consumers to the west of Y. and
10 toconsumers to the east of Y. A move from Y to S would decrease delivery cost for the 10 consumers at point E (savings of $5), but
would increase delivery costs for the consumers at points W. X. and Y (increase in costs of $5.50). A move in the opposite direction
1 1
would also increase total transport cost
Figure 3^4- shows the distribution of consumers along the highway. Distances
are measured from the western end of the highway (point W). There are 2 customers
at point W, 8 customers at point X (one mile from W), 1 customer at Y (two
miles from W), and 10 customers at Z (nine miles from W). Since each customer
buys one pizza and the delivery cost is 50 cents per pizza per mile, the monetary
weight of a particular location (the sales volume times the delivery cost per pizza
per mile) is half the number of consumers at that location.
According to the principle of median location, Ann will minimize total transport
costs at the median location. Since point Y divides the monetary weights into two
equal halves, it is the median location. The monetary weight of locations to the west
is $5 ($1 for W
plus $4 for X), and the monetary weight of locations to the east is
$5 ($5 for Z). The median location divides Ann's customers into two equal halves:
she has 10 customers to the east, and 10 to the west.
To show that the median location minimizes total transport costs, suppose that
Ann starts at the median location, and then moves to point 5, one mile east of Y.
As she moves to the east, there is good news and bad news. The good news is that
she spends less on delivery to point Z: she saves 50 cents per trip to Z, saving a
total The bad news is that westward delivery costs
of $5 in eastward delivery costs.
increase: she pays 50 cents more per trip to points W, X. and Y; since there are 1
customers to the west of westward delivery costs increase by a total of $5.50

S, her
($1 more for W, $4 more for X, and 50 cents more for Y). Since the increase in
westward delivery costs exceeds the decrease in eastward costs, the move from Y to
S increases total delivery costs. The same is true for a move in the opposite direction:
ifAnn moves from Y toward W, total delivery costs will increase.
Why does the median location minimize total transport costs? It is the minimum-
cosi location because it splits pizza consumers into two equal parts. At any location
CluipterS Where Do Cities Develop? 57
Figure 3-5 Median Location in the Large City

Locations S. S-, S-, SA
Demand 4 4 4 4
The median location is in the large city (Z,). Any move to the left of point L will increase travel costs for the
majority of consumers, increasing total transportation costs.
be more consumers to the west of the pizza parlor than

to the east of Y, there will
to the east.As Ann moves eastward away from Y, she moves further away from
1 customers, but moves closer to only 10 customers. Similarly, a westward move
1
will cause her to move closer to 10 customers, but further from 1 customers. In 1
general, any move away from the median location will increase delivery costs for
the majority of consumers, so total costs increase.
It is important to note that the distance between the consumers is irrelevant
to the firm's location choice. consumers were located 100
For example, if the Z
miles from W
instead of 9 miles from W, the median location would still be
point Y Total delivery costs would still be minimized (at a higher level, of course) at
.
point Y.
Median Location and Growth of Cities

The principle of median location provides another explanation of why large cities
become larger. Suppose that a firm delivers its product to consumers in five different
cities. In Figure 3-5, there is a large city at location L, and four small cities at
locations S\ , Sj, and S4. The firm
St,, sells 4 units in each small city, and 17 units in
the large city. The median location is in the large city, even though the large city is
at the end of the line. A one-mile move westward from L would decrease transport
costs by $16 (as the firm moved closer to consumers in the small cities), but would
increase transport costs by $17 (as the firm moved away from consumers in the large
city). The lesson from this example is that the concentration of demand in large cities
causes large cities to grow.
Transshipment Points and Port Cities
The principle of median location also explains why some industrial firms locate
at transshipment points. A transshipment point is defined as a point at which a good
istransferred from one transport mode to another. At a port, goods are transferred
from trucks or trains to ships. At a railroad terminal, goods are transferred from
trucks to the train.
Figure 3-6 shows the location options for a sawmill. The firm harvests logs from
locations A and B, processes the logs into lumber, and then sells the lumber in an
overseas market at point M. Suppose that because of scale economies in production,
a single sawmill is efficient. Highways connect points A and B to the port, and
ships travel from the port to point M
The sawmill is a weight-losing activity: the
.
Figure 3-6 Sawmill Locates at Port

Input source A
Output market Port
Monetary weights: A = $15

x Input source B
B = S15
M = $10
The firm locates its sawmill at the port ( ) median transport location. A move from P toward
P because it is the
either A or B would without affecting input transport costs. A move from
increase output transport costs by S 1
P toward M would increase input transport costs by S30 but decrease output transport costs by only S10.
monetary weights of the inputs are $15 for point A and S15 for point B. and the
monetary weight of the output is $10.
Where will the firm locate its sawmill? Although there is no true median location,
the port is the closest to a median location. If the firm starts at the port ( P). it could
move either toward one of its input sources or to its market.
1. Toward input source A or B. A one-mile move from P toward point A

will cause offsetting changes in the costs of transporting logs from A and
B. Because output transport costs increase by $10. the port location is
superior to locations between P and A. The same argument applies for a

move from P toward B.
2. To market (M ). Unless the firm wants to operate a floating sawmill, it
would not move to points between the port and the overseas market at M . It
could, however, move all the way to the market. A move from P to M
would decrease output transport costs by S10 (the monetary weight of
output) per mile of distance between and P and increase input transport M
costs by $30 per mile (the monetary weight of the inputs). Transport costs
would increase by $20 per mile, so the port location is superior to the
market location.
Although the sawmill is a weight-losing activity, it will locate at the port, not at
one of its input sources. The port location is efficient because it provides a central
collection point for the firm's inputs.
Chapter 3 When- Do Cities Develop.' 59
There are many examples of port cities that developed as a result of the loca-
1880 as a sawmill town: firms
tion decisions of industrial firms. Seattle started in
harvested trees in western Washington, processed the logs in Seattle sawmills, and
then shipped the wood products to other states and countries. Baltimore was the
nation's first boom town: flour mills processed wheat from the surrounding agricul-
tural areas for export to the West Indies. Buffalo was the midwestern center for flour
mills, providing consumers in eastern cities with flour produced from midwestern
wheat. Wheat was shipped from midwestern states across the Great Lakes to Buffalo,
where it was processed by rail, to cities in the eastern United
into flour for shipment,
States. In contrast with Baltimore, which exported its output (flour) by ship, Buffalo
imported its input (wheat) by ship.
Median Location and Cities
The median location explains the development of several types of cities.

principle of
If themedian transport location is an input source, a resource-based city will form
around the input source. If the median location is in the center of the region, a
regional center will develop. If the median location is the transshipment point, a
port city will develop. If the median location is in an existing city, the city will
grow.
Firms Oriented toward Local Inputs

An input is considered local if it cannot be efficiently transported from one location
to another. Examples of local inputs are energy, labor, intermediate goods, and local
public services. A firm that spends a large fraction of its total costs on a particular
local input is pulled toward locations where the price of the local input is relatively
low. For example, an energy-intensive firm is pulled toward a site with inexpensive
energy, and will locate at a site with cheap energy if the savings in energy costs are
large relative to any extra costs for other inputs and transportation.
Similarly, a labor-intensive firm is pulled toward a site with inexpensive labor,
and will locate at the low-wage site if the savings in labor costs dominate any extra
costs for other inputs and transportation.
Energy Inputs
In the first half of the nineteenth century, energy was a local input. The waterwheel
was the first device used to generate nonanimal mechanical energy. The waterwheel
was turned by waterfalls and fast-moving streams, providing power for production
facilities located along rivers and streams. Textile manufacturers set up factories
along small backcountry streams in New England and used waterwheels to power
their machinery. Among the cities that developed as a result of the waterwheel are
Lowell, Lawrence, Holyoke, and Lewiston.
The development of the steam engine in the second half of the nineteenth century
made energy a transferable input. The steam engine could be operated anywhere,
with the only constraint being the availability of coal to fuel the engine. Some
energy-intensive manufacturers located near the coal mines in Pennsylvania. Oth-
ers located along navigable waterways and shipped coal from the mines to their
factories. The steam engine allowed New England textile firms to shift from the
backcountry waterfall sites to sites along navigable waterways. Production shifted
to the Fall River-New Bedford area along the south coast of New England. After the
development of the railroad, firms were able to move to production sites connected
to the coal fields by rail.
The development of electricity affected the location patterns of manufacturers.
Because electricity can be transmitted over distances of several hundred miles, a firm
can use the energy generated from water power without locating along a backcountry
stream. Similarly, a firm can use coal resources without shipping the bulky fuel
from the mine to its factory. In general, the development of electricity decreased
the importance of energy considerations in location decisions. The activities for
which energy considerations are still important are energy-intensive activities such
as the production of aluminum and artificial abrasives. Such activities are attracted
to regions that offer inexpensive energy. The Pacific Northwest, with its abundant
and inexpensive hydroelectric power, attracts energy-intensive activities.
Labor
There are practical limits on how commute, so labor is a
far laborers are willing to
local input. A which labor costs are a large
labor-intensive firm, defined as a firm in
fraction of total cost, chooses a location with low labor costs. Note that the choice
is based on labor costs, not simply wages. A high-wage location will be efficient if
the labor productivity is high enough to justify the higher wages. In other words,
the firm bases its location decision on labor cost per unit of output, not simply the
hourly wage.
Labor costs differ across space for four reasons. First, firms must compensate
their workers for any undesirable features of the local environment. If an area has
relatively dirty air or foul weather, workers will demand relatively high wages.
Second, areas with strong labor unions generally have higher wages. Third, because
households are not perfectly mobile, some areas will have a plentiful supply of
labor and thus relatively low wages. In other words, there is disequilibrium in the
labor market. Fourth is the phenomenon of joint labor supply: households move
to locations that provide a job for the household's primary worker, increasing the
supply of secondary workers. In colonial Massachusetts, the area north of Boston
attracted men to sailing and fishing jobs. The resulting surplus of female workers
decreased the wages of jobs held by women. Shoemakers moved to the area to take
advantage of low wages.
its
There are many examples of firms that base their location decisions primarily on
labor costs. When advances in energy technology reduced their dependence on local
water power, many textile firms moved from the high-wage New England states
to low-wage southern states. More recently, apparel and shoe firms have moved
overseas or to sites near the Mexican border to take advantage of inexpensive labor.
The location decisions of these labor-intensive industries cause the development of
cities in areas where labor is inexpensive.
Amenities, Migration, and Labor Costs

Some firms are indirectly attracted to sites that provide amenities such as good
weather, high-quality schools, low crime rates, a clean environment, and cultural
opportunities. If workers migrate to locations offering these amenities, the resulting
excess supply of labor will decrease wages, attracting labor-oriented firms to the
area. In this case, the firm's location choice depends on the location choice of its
workforce. Instead of workers following firms, firms follow workers. Graves (1979)
and Porell (1982) provide empirical support for this phenomenon.
Amenities appear to be most important for firms that employ high-income work-
ers. Since the demands for these amenities are income-elastic, high-income workers
are attracted to locations with amenities, and the firms hiring these people follow.
For example, research and development firms employ engineers and computer sci-
entists, who place a high value on good weather, high-quality local schools, a clean
environment, and cultural opportunities. One explanation for the shift of employ-
ment from northern states to the southern and western states is that rising income
has increased the demand for certain amenities, causing workers to move to areas
that provide these amenities.
Intermediate Inputs: Localization and Urbanization Economies

Some firms are drawn to locations that provide intermediate inputs. As explained in
Chapter 2, localization economies generate clusters of firms in the same industry.
By sharing the suppliers of intermediate inputs, firms in the cluster exploit scale
economies in the production of specialized inputs. The cluster attracts both the
demanders of intermediate inputs (dressmakers, corporate headquarters, research
and development facilities) and the suppliers of the inputs (buttonmakers, adver-
tising agencies, law firms, testing facilities). Firms in the cluster also exploit scale
economies in the provision of specialized business services and local public services.
The other sources of localization economies are labor-market pooling (sharing a
pool of workers) and knowledge spillovers (dissemination of information about new
products and new production techniques).
Urbanization economies cause firms from different industries to cluster in cities.
The firms in cities exploit scale economies in the provision of business services
(machine repair, banking, insurance, transportation, employment agencies) and local
public services (transportation facilities, schools). Firms in cities may also benefit
from labor-market pooling and knowledge spillovers.
Krugman (1991) discusses the implications of localization economies for in-
dustrial location patterns. To illustrate the effects of labor-market economies, he
tells the story of how a wedding gift made by a teenager led to the emergence
62 Part I Market Fanes in the Development of Cities
of Dalton, Georgia, as the preeminent carpet manufacturing center of the United

States. In 1895. Catherine Evans used an outdated technique—known as tufting or
candlewicking — to make wedding gift. The gift was admired by
a bedspread as a
the recipient and other people who saw it. and in the next few years. Ms. Evans
made a few tufted items for her friends and even sold a few. After she discovered
a few production tricks such as a technique for locking the tufts onto the backing,
she and her neighbors launched a local handicraft industry, producing handmade
tufted items for sale. Immediately after World War II. a machine for producing
tufted carpets was developed, making tufted carpet a cheaper alternative to woven
carpet. New carpet makers sprung up in the Dalton area because that's where they
could find workers with knowledge and experience in tufting. Support firms located
nearby, supplying dyes, backing, and other intermediate inputs. Some of the old
firms that had produced woven carpets went out of business because they were un-
derpriced by tufting firms, while others moved from the Northeast to the Dalton
area and switched to the new technique. By the middle of the 1950s. Dalton was the
carpet capital of the nation. Of the top 20 carpet manufacturers in the United States,
six are located in Dalton. and thirteen others are located nearby.
Krugman also provides a measure of industrial concentration, which he calls
the "locational Gini coefficient." To compute the coefficient for a particular industry
(e.g.. glove manufacturing) we start by drawing a Gini curve for the industry. The
drawing requires four steps:
• For each region, compute ( 1 ) the region's share of total national

manufacturing employment and (2) the region's share of national
employment in the glove industry. For example, suppose region A has
20 percent of total employment but 60 percent of glove employment.
• For each region, compute the ratio of its share of total employment to its
share of glove employment. Region A has a ratio of 3.0 = 60/20.

• Rank the regions by this ratio. For example, if region B has 30 percent of the
total employment and 30 percent of glove employment, its ratio would be
1.0; if region C has the remaining 50 percent of total employment and the
remaining 10 percent of glove employment, its ratio would be 0.20.

• Going down the ranking, compute the cumulative totals for both the total
shares and the glove shares and draw the Gini curve (shown in Figure 3-7).
The 45-degree line shows an even distribution of glove employment across the
three regions. For example, point ; shows that with an even distribution of glove
employment, region A. with 20 percent of total employment, would have 20 percent
of glove employment. The closer the distribution of glove employment matches the
distribution of total employment, the closer the Gini curve will be to the 45-degree
line. A natural measure o\' localization is the area between the Gini curve and the
45-degree line. At one extreme, an industry that is concentrated almost entirely

in a relatively small region will have a Gini coefficient o\~ close to 0.50. At the
other extreme, an industry that is spread out in proportion to total manufacturing
emploj ment will have a coefficient of 0.
Figure 3-7 Locational Gini Coefficient
c
1 .00
_ —— /
0.90 '
" " " ~N /
/ /
//
>
'
/
/
/ !
/
//
i
< /
c /
/
'
u /
//
'
B ! /
>-. • /
3, // /
/
//
'
5 /
u /
//
i
a> a /
' /
0.60 ¥ /
5J3 /', i /
/, i /
o / i • /
1>
n /
/
:
y
/l
J= / / '
'S-
/ 1 /
1)
> / ' / !
/ ' / !
^ / ! s \
/ , / '
5 / i /
/ ' /
U / 1 / i
/ ! /
/ i /
/ ' /
'
/ ,
0.20 ..../ i' z :
/ / '
1 / ' !
1 ' \
/ / , •
/ / i
i
/ /
/ / 1
/ /
0.20 0.50 1.00
Cumulative share of manufacturing employment
The locational Gini coefficient is computed as the shaded area between the solid kinked line and the
45-degree line. Point a shows the shares of total and glove employment for region A, while point/) shows
the shares for regions A and B combined, and point c shows the shares for all three regions in the nation.
Table 3-3 shows locational Gini coefficients for the 10 most concentrated in-
dustries and a selection of industries from the top 50. To put these numbers in
perspective, the automotive industry, with half of its employment in the traditional
automotive district of southern Michigan and the nearby parts of Indiana and Ohio,
is ranked 49th in terms of geographic concentration, with a Gini coefficient of 0.30.
The fact that 48 industries have larger Gini coefficients suggests that geographic
concentration is very prevalent. The fact that many low-technology industries are
so concentrated suggests that there is much more to localization economies than
knowledge spillovers.
Local Public Services and Taxes

Local governments influence the location decisions of businesses by the provision
of public services. The public sector provides some of the inputs used directly by
64 Pari I Market Forces in the Development oj Cities
Table 3-3 Locational Gini Coefficients
Rank SIC Description Gini Main Slates
1 303 Reclaimed rubber 0.50 WY. \YI. WV

2 313 Boot and shoe 0.48 ME. MO. MA
3 315 Leather gloves and mittens 0.48 WI. NY. WY
4 222 Weaving mills, synthetics 0.4S GA. SC, NC
5 237 Fur goods 0.47 NY, WY. WI
6 223 Weaving and finishing mills, wool 0.45 ME. NH
RI.
7 221 Weaving mills, cotton 0.44 SC, GA, NC
8 319 Leather goods 0.44 TX. MA. CA
9 227 Floor covering mills 0.43 GA. SC. VA
10 228 Yarn and thread mills 0.43 CO. MN. OK
12 277 Greeting card publishing 0.43 AR. KS. MO
15 376 Guided missiles, space vehicle parts 0.41 CA, IT. AZ
19 235 Hats, caps, millinery 0.41 IA. MO. NY
22 396 Costume jewelry and notions 0.40 Rl. CT. NM
32 393 Musical instruments 0.34 MS. AR. IN
49 371 Motor vehicles and equipment 0.30 MI. OH. DE
50 395 Pens, pencils, office and art supplies 0.30 RI. IA. NJ
SIC stands for Standard Industrial Classification.
firms (water, roads, ports, and sewers) and provides goods consumed by the city's
workforce (education, parks, roads, transit systems, public safety). A city can attract
firms by providing a set of high-quality, efficient public services. In addition, indus-
trial firms require production sites that ( 1 ) are accessible to the intracity and intercity
transportation networks and (2) have a full set of public services (water, sewerage,
electricity). By coordinating its land-use and infrastructure policies to ensure an
adequate supply of industrial land, a city can attract new firms and encourage the
expansion of existing firms.
The public
sector also affects the location decisions of businesses through its tax
policies. Taxes on businesses increase production costs and discourage firms from
locating in the city. Taxes on residents increase the costs of living in the city and will
ultimately result in higher wages and production costs.
Many cities try to attract export firms by offering special subsidies for new firms.
Levy (1981) describes several types of subsidy programs, including the following:
1 . Tax abatement. In some cities, new firms are exempt from local property
taxes for a fixed period (e.g.. 10 years). Some cities offer tax abatements to
all new developments, while others offer abatements only to firms that are
particularly sensitive to tax differentials.
2. Industrial bonds. Some cities issue tax-free industrial bonds to finance

property developments. The local government uses the revenue from the
bonds to purchase the land, and leases the property to a private firm.
Because the interest income from industrial bonds is not subject to federal
Chapteri Where Do Cities Develop? 65
taxes, the bond buyer accepts a relatively low interest rate (e.g., 8 percent
instead of 12 percent). Therefore, the lessee pays less than the market
interest rate on the money borrowed to finance the project. The use of
industrial bonds was sharply curtailed by the Tax Reform Act of 1986.
3. Government loans and loan guarantees. Some cities loan money directly
to developers, and others guarantee loans from private lenders. In both
cases, developers borrow money at a relatively low interest rate: either the
city charges an interest rate below the market rate, or the city decreases the
risk associated with a private loan, allowing the developer to borrow private
money at a relatively low interest rate.
4. Site development. Some cities subsidize the provision of land and public
services for new development. The city purchases a site, clears the land,
builds roads and sewers, and then sells the site to a developer at a fraction
of the cost of acquiring and developing the site.
Several Local Inputs
How does a firm with several local inputs choose among the alternative locations?
The location choice is determined by the outcome of a tug-of-war between forces
pulling the firm toward its different local inputs (labor, energy, intermediate goods,
local public services). The strength of the pull toward a particular input source
depends on (1) the relative importance of the input to the firm and (2) the spatial
variation in the input price. The greater the importance of the input and the price
variation, the stronger the pull toward a particular input source.
Consider a firm that has two local inputs: labor and energy. The firm is pulled
toward city L (which has low wages) and city E (which has inexpensive energy).
The forces pulling the firm toward L will win the tug-of-war if 1 ) the firm's labor (
consumption is large relative to its energy consumption (the firm is labor-intensive)

and (2) the difference in the wage (controlling for productivity differences) is large
relative to the difference in the price of energy.
Location Choices and Land Rent
This section suggests that some types of firms choose locations that provide inex-
pensive local inputs. The labor-intensive firm chooses a location with relatively low
labor costs. The firm in an industry subject to localization economies locates in a
cluster of similar firms. The firm that depends on business services locates near
banks, law firms, and advertising agencies. How do these location choices affect the
land market?
If a particular location offers a desirable mix of local inputs, it will attract
input-oriented firms. The resulting increase in the demand for land increases the
equilibrium price of land. Therefore, the advantages of a particular site (inexpensive
labor, access to intermediate inputs) will be by higher land costs. For
offset, in part,
example, a firm in the city center can economize on intermediate inputs by locating
close to banks, law linns, and advertising agencies, but it also pays more for land.
Later in the book, the factors that determine the equilibrium price of land will be
examined in greater detail.
Transport Costs versus Local-Input Costs

This chapter divides industrial firms into two types: those that base their location deci-
sions exclusively on transport costs (transfer-oriented firms) and those that base their
location decisions exclusively on local input costs (input-oriented firms). Table 3—4
summarizes the results of the chapter, listing the various types of transfer-oriented
and input-oriented firms.
In the real world, things are not so tidy. Most firms cannot be easily classified
with respect to their locational orientation. For most firms, location decisions are
Table 3-4 Summary of Locational Orientation
Orientation Relevant Characteristic Example
Transfer Orientation Transport costs are large

fraction of total costs.
Resource orientation Weight loss Bat factory

Bulk loss Cotton baling
Perishability loss Canning
Hazard (fragility) loss Skunk deodorizing
Market orientation Weight gain Bottling

Bulk gain Auto assembly
Perishability gain Baking
Hazard (fragility > gain E\plosi\es
Local-Input Orientation Transport costs are small

fraction of total costs.
Energy Energj -intensive production Aluminum
Labor Labor-intensive operation Textiles
Intermediate inputs
Specialized inputs Localization economics Dressmakers
Business sen ices Urbanization economics Corporate HQ
Amenitv Orientation Workers are sensitive to R&D

weather and recreation.
based on both transport costs and local-input costs. Since transport costs and input
costs vary across space, the typical location choice requires a trade-off between these
two types of costs.
Trade-off between Transport Costs and Local-Input Costs

To explain the location choice of a firm that considers both transport and input costs
in its location decision, consider a decision about where to locate an auto-assembly
factory. In choosing a production site, the firm must consider the following:
1 Inputs and output. The input source is the same as the market: point B is
the source of transferable inputs and the market.
2. Transport costs. The monetary weights of inputs and outputs sum to $6.
As shown in Figure 3-8, total transport costs rise from zero at point B to
$60 at point M ( 10 miles from point B).
Figure 3-8 Transport Costs versus Labor Costs: Transfer Orientation
Point M
x (distance from B)
Input source
and market (B)
Total cost (the sum of transport cost and labor cost) is minimized at point B (the location of the input and
the output market). The transfer-oriented firm locates at B because the spatial variation in transport costs is
large relative to the variation in labor costs.

3. Local input. Labor costs vary across space. As the firm moves away from
B. labor costs decrease by $3 per mile, dropping from $40 at point B to $10
at point M
Total cost (transport cost plus labor cost) is minimized at point B because transport
costs are sizable relative to labor costs. As the firm moves from point B toward point
M . the savings in labor costs ($3 per mile) is less than the increase in transport costs
($6 per mile). Because the spatial variation in transport costs is relatively large, the
firm chooses point B. close to its input sources and market.
Change in Orientation: Decrease in Transport Costs
Figure 3-9 shows the effect of a decrease in transport costs on the firm's location
choice. Transport costs fall to one-fourth of their original level (from $6 per mile to
$ 1 .50 per mile), decreasing the slope of the transport-cost curve from 6.0 to 1 .5. The
slope of the total-cost curve is negative instead of positive: a one-mile move away
from point B generates a net savings of $1.50 (a $1.50 increase in transport costs is
combined with a $3 decrease in labor cost), so the firm will move to the low-wage
Figure 3-9 Switch from Transfer Orientation to Input Orientation
Total cost = I.C + TC
Point V
Input source
> (distance from B>
and market i/<i
\ decrease in transport costs causes the firm to switch from a transfer orientation to an input oin.-iH.uion. Total cost
(/< | rC) is minimized at > = 10 because the spatial variation in labor cost is great relative to the spatial
variation in transport cost.
Chapter 3 Where Do Cilies Develop'.' 69
location. In other words, the decrease in transport costs transforms the firm from a
transfer-oriented firm to an input-oriented firm.
In the last several decades, many industries have switched from a transfer orien-
Firms have moved from locations close to input sources
tation to an input orientation.
and markets to locations with inexpensive local inputs. The changes in locational
orientation resulted from innovations in transportation and production.
Innovations in Transportation. Improvements in transport technology decreased

the cost of transporting both inputs and outputs. For example, the development of
fastocean ships and container technology (shipping freight in large, standardized
containers instead of in small bundles) made ocean shipping more efficient, decreas-
ing the unit cost of transportation (cost per ton per mile). Similarly, improvements
and planes decreased unit freight costs. For some firms, the decrease
in rail, trucks,
in transport cost was large enough to cause a switch from a transfer orientation to a
local-input orientation. These firms then based their location decisions on access to
inexpensive local inputs rather than access to markets or inputs. A recent example is
the movement of the assembly operations of many U.S. manufacturers to sites along
the Mexican border.
Innovations in Production. Improvements in production techniques decreased

the physical weight of inputs. As the weight of inputs decreased, total transport cost
decreased, and some transfer-oriented firms became local-input firms. These firms
moved from sites with low transport costs to ones with inexpensive local inputs.
An example is the U.S. steel industry. Over the last several decades, the amount
of coal and ore required to produce one ton of steel has decreased steadily, a result of
improved steelmaking methods and the use of scrap metal (a local input) instead of
iron ore (a transferable input). As the physical weight of inputs decreased, transport
costs decreased relative to labor costs, and some firms became oriented to local
inputs. Many steel plants are now located in low-wage countries (Brazil. Korea.
Mexico), far from both raw materials and steel markets.
Location Orientation and Growth Patterns
Changes in transport costs affect economic growth patterns. As transport costs de-
crease, the relative attractiveness of a resource-rich region decreases, causing firms
to move to areas with inexpensive local inputs. The steel industry has moved from
the eastern United States, with its rich coal and ore deposits, to Japan and Korea,
which have neither coal nor ore deposits. The steel industry made the move because
the relative importance of transport costs decreased: steel firms moved to areas with
lower wages because the savings in labor costs exceeded the increase in transport
costs (for ore, coal, and steel). Similarly, manufacturers have moved from the United
States to Asia and Mexico, far from U.S. markets, because the savings in labor costs
dominate the increase in transport costs.
70 Pan 1 Market Forces in the Development of Cities
Case Studies and Empirical Studies

This section discusses some of the facts on the location choices of firms. The
first two
two branches of the computer industry,

parts discuss the location decisions of firms in
semiconductor manufacturers and computer makers. The third part discusses the
location choices of Japanese auto firms. The fourth part discusses the choices of
Mexican garment manufacturers. The fifth part summarizes a case study of the
decision to site GM's new Saturn plant near Nashville. Tennessee. The sixth part
summarizes an empirical study of location decisions in the electronic components
industry and also discusses some empirical evidence concerning the effects of taxes,
public services, wages, and unions on firms" location decisions.
The Semiconductor Industry

The semiconductor industry illustrates some of the complexities of locational choices
in modern industry. The industry's output is light and compact relative to its value
(i.e.. it has a low bulk-value ratio), so transport costs are relatively unimportant in
location decisions. What matters are localization economies and access to different
types of labor.
As explained by Castells ( 1988). the making of semiconductors involves several
distinct operations, requiring three different types of labor:
1 Research and development. Engineers and scientists design new circuits

and prepare the circuit for implantation into silicon chips.
2. Wafer fabrication. Skilled technicians and manual workers make the chips
holding the circuits.
3. Assembly into components. Unskilled workers assemble the chips into
electronic components.
Many semiconductor firms split their operations into three parts. Research and
development occurs in the Silicon Valley to exploit localization economies gener-
ated by the large concentration of semiconductor firms. The Silicon Valley is also
considered a desirable location by engineers and scientists. Advanced manufacturing
(wafer production) is typically located outside the Silicon Valley. For example. Na-
tional Semiconductor has manufacturing facilities in Utah. Arizona, and Washington
State: Intel has plants in Oregon. Arizona, and Texas: Advanced Micro Devices has
a plant in Texas. These facilities are located in areas that provide (Da plentiful
supply of skilled manual laborers. (2 l an environment attractive to engineers and
technicians, and (3) easy access, by air transportation, to the Silicon Valley. The
firm's assembly facilities are typically overseas, in locations such as Southeast Asia
that have a plentiful supply of low-skilled workers.
Computer Manufacturers
The splitting of manufacturing operations also occurs among computer makers.
Hekman ( 1985) studied the location decisions of firms producing large mainframe
Chapters Where Do Cities Develop? 71
computers. The five largest firms, which control more than 95 percent of the U.S.
market, have their main research and design facilities in large metropolitan areas, one
each in Boston, New York, and Philadelphia, and two in Minneapolis. Because of the
frequent design changes in mainframe computers, there is a great deal of nonroutine
communication between designers and manufacturers, so proximity between design
and fabrication activities is important. Four of the five mainframe makers assemble
their computers near their research and design facilities.
In contrast, peripheral equipment (terminals, tape drives, and printers) is pro-
duced in branch plants outside the home metropolitan area. Because peripherals are
not subject to rapid design changes, they can be easily produced far from the design
facilities, in Tennessee, Nebraska, South Dakota, Utah, North Carolina, Florida, and
Michigan. Similarly, individual computer components are standardized inputs, and

are produced in Mexico, Brazil, Hong Kong, Taiwan, and Puerto Rico.
Japanese Automobile Firms

A recent study of Japanese firms in the automobile industry (Smith and Florida,
1994) explores the relative importance of various locational attributes. Under the
just-in-time inventory system, a firm links the delivery of intermediate inputs with
its production schedule, the idea being that the inputs should arrive just in time for
production. This inventory system requires geographic proximity to input suppliers
to both save on shipping time and facilitate communication between firms. The study
suggests that the major factor in the location of Japanese-affiliated automotive sup-
pliers in the United States is proximity to Japanese-affiliated automotive assemblers.
In other words, agglomeration is the key feature in the location choices of such firms.
The study of Japanese firms provides some other insights into locational choices.
Japanese-affiliated automotive suppliers prefer locations with relatively large pop-
ulations, a high density of manufacturing activity, a more educated workforce, and
better transportation. Furthermore, these firms tend to pick locations with relatively
high wages, reflecting their willingness to pay higher wages in exchange for higher
labor productivity. Finally, Japanese-affiliated manufacturers tend to locate in areas
with relatively high concentrations of minority workers.
Mexican Garment Industry

Hanson (1996) describes Mexican garment
the locational patterns of firms in the
industry. Until the 1920s, Mexico were made by housewives and
most garments in
neighborhood tailors, who used factory-made cloth to produce custom garments for
family members and local patrons. The first garment factories, built by Lebanese and
Jewish immigrants, were located in downtown Mexico City. For several decades, the
industry was highly concentrated in the Federal District (the area containing Mexico
City), with about 60 percent of Mexico's garment jobs there.
During the 1970s and 1980s, garment employment in the Federal District de-
creased in relative and absolute terms. The district's share of national garment em-
ployment dropped from 60 percent in 965 to 33 percent by 985, with jobs moving to
1 1
states with lower wages (Aguascalientes, Jalisco, Mexico, Nuevo Leon, and Puebla).
Each state specialized in a particular type of garment, and garment employment in
each state is concentrated in one or two cities. For example, the city of Aguascalientes
is employment center for
the children's outerware. with over 44 percent of national
employment in that sector.
Although many garment jobs moved out of the Federal District, the district
retained its role as the center for marketing and design. In 1980. about 70 percent of
wholesale trade in garments, textile, and leather goods was conducted in the Federal
District, suggesting that the bulk of marketing and design still happens there. Most of
the jobs that moved to the outlying areas were low-skill assembly jobs. In other words,
there is a hierarchy of functions, with the activities subject to large agglomerative
economies (marketing and design) concentrated in one location, and activities with
smaller agglomerative economies (assembly) dispersed to outlying areas.
A Case Study: The Location ofGM's Saturn Plant

In 1985. General Motors announced that its new Saturn plant would be located in
Spring Hill. Tennessee, a crossroads community about 30 miles from Nashville. The
plant,which cost several billions of dollars to build, employs about 3.000 workers
and supports thousands of jobs in the Nashville area. Why did choose Nashville GM
over the dozens of other communities that sought the plant?
A case study by Bartik. Becker. Lake, and Bush (1987) suggests that the most
important factor in the site choice was labor costs. Transportation costs and taxes
also influenced the decision. A labor contract between GM and its unionized workers
stipulates that the Saturn workers will be paid the same wage regardless of the plants
location, so GM will not decrease own wage bill by locating in Tennessee, a state
its
with low wages. However, GM will purchase a large fraction of inputs from its
nearby suppliers, who will pay low wages and pass on the savings to GM.
The first step in the case study addressed the issue of transportation costs and
market access. The authors estimated the costs of shipping finished cars from hun-
dreds of alternative production sites to GM's markets in the continental United States.
Based on 1984 sales figures, the cost-minimizing location (the location offering the
best market access) was Indianapolis. Indiana. Based on the distribution of projected
sales in 2000. the location offering the best market access was Terre Haute. Indiana.
The shift from Indianapolis to Terre Haute reflects the shift in population and car
consumers to the southern and western areas of the United States.
The second part of the study incorporated labor costs and taxes into the anal-
ysis. The authors estimated labor costs and taxes for production sites that pro-
vided reasonable access to GM's markets (a total of 130 sites in seven states).
Table 3-5 shows the costs (expressed as costs per car) for one site in each of the
seven states.
1 . Transportation costs. The variation in transportation cost among the seven

sites was relatively small. The gap between the lowest-cost site (Terre
Haute) and the most expensive site (Kalamazoo) was only SI 7 per car.
Nashville was ranked fifth, with transport costs only $ 3 per car higher than
1
Terre Haute.
Table 3-5 Case Study of GM s Saturn Plant

Estimated Cost (doll ars per car)
Transport Labor Costs for State at \d Local Taxes Total Measured

Costs Local Suppliers (before subsidy) Cost
Nashville, TN 426 159 118 703

Lexington, KY 423 186 106 715
St. Louis, MO 419 172 134 725
Bloomington, IL 417 202 162 781
Kalamazoo, MI 430 244 116 790
Terre Haute, IN 413 209 168 790
Marysville, OH 427 219 169 815
Source: Timothy J. Bartik. Charles Becker. Steve Lake, and John Bush. "Saturn and State Economic-
Development," Forum for Applied Research and Public Policy (Spring 987 ). pp. 29-40.
1
2. Labor costs for local suppliers. The variation in labor cost was relatively
large. The difference between the lowest-cost site (Nashville) and the
highest-cost site (Kalamazoo) was $85 per car.
3. State and local taxes. Table 3-5

shows the tax cost per car in the absence
of special subsidies and exemptions. The difference between the
highest-tax site (Marysville) and the lowest-tax site (Lexington) was $63
per car. Nashville was ranked third, with tax costs per car $12 higher than
Lexington.
The case study suggests that in the absence of special tax treatment, Nashville was
the lowest-cost site. The sum of its transportation, local and tax costs was
labor,
$12 Lexington, the second-ranked site,
less per car than and $87 less than Terre
Haute, the minimizes transportation costs. Nashville's slight disadvantage
site that
in market access is offset by its lower labor costs.

Tennessee offered three types of inducements to GM. The first was a $30 mil-
lion highway project that connected the plant with Interstate 65. The second was
a subsidized job-training program for Saturn workers, worth about $4 per car. The
third was a program of property tax subsidies, worth about $30 per car. Although
these subsidies would not have been necessary in the absence of similar subsidies
from other states, it appears that Tennessee offered them to stay competitive with
other states bidding for the plant.
The Relative Importance of Local Inputs: Empirical Results

On a theoretical level, all local inputs (labor, energy, intermediate inputs) influence
location choices. Which local inputs are the most important? A number of recent
empirical studies of location decisions address this question.
Carlton Study: Birth Rates for Firms. Carlton (1979) examines the location
choices of firms in three input-oriented industries: plastics products, electronic trans-
mitting equipment, and electronic components. These industries have relativeh low
transport costs, so they are oriented toward local inputs, not toward output markets
or natural resources. Carlton hypothesizes the following relationship between the
number of new firms in a particular metropolitan area (the number of "births") and
the characteristics of the metropolitan area:
N= f(\Y. E.O.G.T.I.U) (3-3)
where
N= Number of new firms in a particular industry
W= Metropolitan wage
E = Unit energy cost
Q = Metropolitanwide output of the industry
G = Number of engineers in the metropolitan area
T = The tax level of the metropolitan area
/ = Measure of incentive programs offered by government
to new business (revenue bonds, subsidized loans, tax exemptions)

U = Metropolitan unemployment rate
In other words, the number of births depends on the cost of local inputs | W and
E). localization economies (as measured by O. the total output of the industry),
the technical expertise of the area (as measured by G. the number of engineers).
government tax policy (T and /). and the health of the local economy (as measured
by U, the unemployment rate).
Table 3-6 shows Carlton's results for the electronic components industry, re-
ported as the elasticities of the number of births w ith respect to the various location
factors. The elasticity of the number of births with respect to the wage is — 1 .07.
Table 3-6 Relative Strengths of

Location Factors
Elasticity of Births with

Respect to Variable
Wage -1.07
Energj cost -0.38 to -0.95
Output of industn 0.43
Number o( engineers 25
Tfcxes Close to zero
Government incentives Close to zero
Soi kci Dennis W. Carlton, The Location and Employment

Choices New Finns \n Econometric Model," in
ol
Interregional Movements and Regional Growth, ed William C.
\\ heaton Washington. DC: The Urban Institute, 1979).
I
A 10 percent increase in the metropolitan wage decreases the number of births by

10.7 percent. The elasticity for energy cost is lower, from —0.38 for natural gas to
—0.95 for electricity. The elasticity for industry output (0.43) suggests that local-
ization economies are important: the number of births increases as the size of the
industry increases. The technical expertise of the metropolitan area (measured by
the number of engineers) also affects the number of new firms.
Carlton's results suggest that tax policies do not affect the location decisions
of firms producing electronic components. The elasticities for the tax and incentive
variables are close to zero, implying that metropolitan areas with low taxes and
lucrative incentive packages do not experience higher birth rates. These results are
consistent with earlier studies of the influence of taxes on location decisions.
Carlton's results for the other two industries (plastics products and electronic
transmitting equipment) were similar. The most important location factors were
wages, energy costs (less important for electronic transmitting equipment), industry
output, and technical expertise.
Effects of Taxes and Public Services on Business Location Decisions
20 years, there have been dozens of empirical studies of firms' location

In the last
choices. Bartik( 1991 ) summarizes the results from many of these studies and draws
some general conclusions about the effects of taxes and public services on firms'
location choices.
There is solid evidence that local taxes have a strong negative effect on regional
business growth. In other words, a high-tax city will grow at a slower rate than a
low-tax city, everything else being equal. Similarly, if a city increases its taxes, it
will grow at a slower rate, everything else being equal. It's important to note that one
of the items included in everything else is public services: a high-tax city grows at a
slower rate because firms pay higher taxes without receiving better public services
in return. If a high-tax city were to use the extra tax revenue to provide better public
services, the city might grow at a faster rate than a low-tax, low-service city.
Empirical studies of business location decisions distinguish between two types
of decisions, intermetropolitan decisions (choosing a metropolitan area) and in-
trametropolitan decisions (choosing a site within a metropolitan area). One way to

summarize the effects of taxes on location decisions is to compute the elasticity
of business activity with respect to tax liabilities, defined as the percentage change
(or percentage difference) in business activity divided by the percentage change (or
percentage difference) For intermetropolitan location decisions, the
in tax liabilities.
elasticity is between —0.10 and —0.60: a 10 percent increase in taxes in a particu-

lar metropolitan area decreases business activity in the metropolitan area between
1 percent and 6 percent. For intrametropolitan location decisions, the elasticity is be-
tween — 1 .0 and —3.0. If an individual municipality increases its taxes by 10 percent,
business activity in the municipality decreases between 10 percent and 30 percent.
The elasticity for the intrametropolitan decision is larger because different loca-
tions within a metropolitan area are better substitutes than locations in different
metropolitan areas.
Two other results from recent empirical studies are worth noting.
First, it appears
that manufacturers are more sensitive than other types of firms to tax differentials
within and between metropolitan areas. This is sensible because manufacturers are
likely to be oriented toward the national market and thus have a wider range of
location options. Second, metropolitan areas with relatively high taxes on capital (in
the form of taxes on business property) tend to repel capital-intensive industries and
attract labor-intensive industries.
There is solid evidence that the provision of local public services has a strong
positive effect on regional business growth. If two cities differ only in the quality
of their local public services, the city with the better public services will grow at a
faster rate. Similarly, if a city improves its public services, it will grow at a faster rate,
everything else being equal. It's important to note that one of the items included in
everything else is local taxes: the high-service city grows at a faster rate because firms
receive better public services without paying higher taxes. The public services that
have the largest positive effect on business growth are education and infrastructure.
How would simultaneous increasesin taxes and spending on public services
affect location choices and business activity? Studies by Helms (1985) and Munnell
(1990) suggest that the effect of a tax increase depends on how the extra tax revenue
is spent. If the extra tax revenue is spent on local public services (infrastructure,
education, or public safety), the tax/expenditure program increases the relative at-
tractiveness of the city and promotes economic growth. In contrast, if the extra tax
revenue is spent on redistributional programs for the poor, the tax decreases the
and decreases the growth rate.
relative attractiveness of the jurisdiction
There are many anecdotes about firms shopping around for the best tax-
abatement deal. The classic example is of a firm that secretly decides to locate
in community B. but then goes to community C, an inferior site, and asks for a
tax-abatement package that would be lucrative enough for the firm to choose that
community. The firm then asks the preferred community (B) to match the tax pack-
age offered by community C. When community B matches the tax package, the
firm locates in that community, which it intended to do in the absence of any tax
incentives. A recent study of property-tax abatement (Anderson and Wasmer. 1995)
provides evidence that this sort of tax shopping occurs frequently. The authors sug-
gest that state policymakers should consider policies that discourage this sort of
competition between communities.
Effects of Wages and Unions on Business Location Choices

The recent evidence on the sensitivity of location decisions to wages is consistent
with Carlton's results. Bartik ( 1991 ) examines and summarizes the results of dozens
of studies and concludes that the long-run elasticity of business activity with respect
to the wage is between — 1 .0 and —2.0. This means that a 10 percent decrease in the
metropolitan wage will increase business activity between 10 percent and 20 percent.
How do unions affect location decisions and the volume of business activity?
One of the effects of unions is to increase wages, so a highly unionized metropoli-
tan area is likely to have relatively high wages. Since the elasticity of business
activity with respect to the wage is relatively large, to the extent that unions increase
wages, they decrease business activity. Unions may have other (nonwage) effects
that influence business location decisions. For example, unions may affect labor pro-
ductivity. According to Bartik ( 1 99 1 ), the empirical studies of the nonwage effects of
unions have generated mixed results: although most studies found that the presence
of unions decreases business activity, many studies suggest that the negative effects
are rather small.
The Role of Government in the Location of Cities

Not all cities develop as a result of the location decisions of profit-seeking firms.
Some cities owe their existence to government policies. Government affects the
location of cities in several ways. First, governments pick the sites for administrative
offices, causing concentrations of government jobs in capital cities. In the United
States, many state governments chose vacant sites for their capitol buildings, creating
instant capital cities. Second, the siting of public facilities such as universities and
prisons generates concentrations of jobs for guards and teachers.
Some cities develop as a result of differences in legal environments. During
colonial times, several cities along the eastern seaboard thrived on illegal trade with
the West Indies. Under the British Navigation Acts, trade with the French in the
West Indies was restricted. The British authorities made little effort to enforce the
trade regulations in the colonies, so smuggling between the colonies and the West
Indies flourished. British traders complained that because colonial traders ignored
the Navigation Acts, they had an unfair advantage over their British competitors.
Some state governments have exploited differences in legal environments to
stimulate growth. The state of Nevada has used liberal divorce laws and legalized
gambling to stimulate the growth of its cities near the California border. In 1900,
the state established itself as the state with the most liberal divorce laws and kept
one step ahead of other states as they liberalized their divorce laws. The divorce
industry was the primary force in the early development of Reno, 12 miles from the
California border. When the divorce traffic slowed during the Depression, the state
legalized gambling as a means of stimulating the state's economy. The gambling
industry contributed to the growth of Reno, Las Vegas, and Lake Tahoe. The cities
of Reno and Lake Tahoe cater to customers from Northern California, while Las
Vegas caters to customers from Southern California.
Summary
1 Cities develop around the concentrations of employment generated by firms,
so the location choices of firms play a role in the location of cities.
2. Trading firms collect goods from suppliers, distribute the goods to consumers,
and locate at transshipment points. The location of trading and transportation
firms explains the development of port cities and junction cities.
3. A transfer-oriented firm spends a relatively large amount on transport costs

and chooses the location that minimizes total transport costs.
a. A resource-oriented firm has relatively high transport costs for its inputs, so
it locates near its input source. The location choices of resource-oriented
firms cause the development of resource-based cities and port cities.
b. A market-oriented firm has relatively high transport costs for its output, so
it locates near its market. The location choices of market-oriented firms
cause the development of port cities and the growth of existing cities.
4. According to the principle of median location, a market-oriented firm locates

at its median transport location, defined as the location that splits its customers
into two halves.

a. This principle explains why big cities grow. The median location is often in
a large city, so many market-oriented firms locate in large cities.
b. This principle explains why some resource-oriented firms locate at
transshipment points (ports). The port is often the median location for the
firm because it provides a central collection point for inputs and a single
distribution point for the output.
5. A local input is defined as an input that cannot be transported long distances.

Examples are certain types of energy, labor, and intermediate goods.
a. A firm is oriented toward a local input if (1) it uses a local input whose
price varies across space and (2) it spends a large fraction of its budget on
the local input.
b. The input-oriented firm will locate near inexpensive local inputs, causing
the development of cities and regions near sources of inexpensive energy,
labor, and intermediate goods.
6. Firms that employ people who are sensitive to amenities (good weather,
high-quality schools, low crime rates, a clean environment, cultural
opportunities) are attracted to locations that offer these amenities. Instead of
workers following firms, firms follow workers. Amenities are most important
for firms that employ high-income workers.
7. One study of the location decisions of firms producing electronic components,
plastics products, and electronic transmitting equipment suggests that the most
important location factors are the wage and energy costs, industry output (a
measure of localization economies), and technical expertise. Taxes and
incentive programs do not seem to affect location decisions.
8. Until recently, most studies of location decisions suggested that taxes were not
an important locational factor. Recent studies cast some doubt on this
conclusion. One study suggests that the effect of a tax depends on how the tax
revenue is spent: an increase in taxes increases the relative attractiveness of a
jurisdiction if the money is spent on local public services (highways,
education, public health and safety), but decreases the relative attractiveness of
the jurisdiction if the money is spent on redistributional programs.
9. In the last several decades, the relative importance of transport costs has
decreased, a result of innovations in transportation (which decreased unit
Chapter 3 Where Do Cities Develop '.'
79
transport costs) and production (which decreased the physical weight of

inputs).
a. Many resource-oriented and market-oriented firms have been transformed
into firms oriented toward local inputs.
b. The change in locational orientation provides one reason for the shift in
economic growth from resource-rich regions (e.g., the United States) to
places with low labor costs (e.g., Asia).
10. Government decisions cause the development of administrative cities and

cities that host public facilities such as prisons and universities. Differences in
legal environments cause the development of border cities.

1 Comment on the following from the owner of a successful plywood mill:
"Firms don't use location theory to make location decisions. I chose this
location for my plywood mill because it is close to my favorite fishing spot."
2. Depict graphically the effects of the following changes on the bat firm's cost
curves (shown in Figure 3-1). Explain any changes in the optimum location.
a. The cost of shipping bats increases from $1 per ton to $4 per ton, while the
cost of shipping wood remains at $1 per ton.
b. The forest at point F burns down, forcing the firm to use wood from point
G, which is 10 miles west of point F (20 miles from the market).
c. The firm starts producing bats with wood and cork, using three tons of
wood and two tons of cork to produce three tons of bats. Cork is ubiquitous
(available at all locations for the same price).
3. Why do breweries typically locate near their markets (far from their input
sources), while wineries typically locate near their input sources (far from
their markets)?
4. The building of wooden ships was a weight-losing activity, as evidenced by

the piles of scrap wood generated by shipbuilders. Yet shipbuilders located in
ports, far from their input sources (inland forests). Why?

5. In the 1960s, the port of Oakland moved rapidly to develop facilities to take
advantage of the new labor-saving containerization technology. As a result, it
was able to take shipping business away from the San Francisco port (across
the bay). Under what conditions will the adoption of containerization
technology increase total employment in Oakland?
6. Consider a firm that delivers video rentals to its customers. The spatial
distribution of customers is as follows: 10 videos are delivered to location W,
10 miles due west of the city center; 50 videos are delivered to the city center;
25 videos are delivered to E, 1 mile due east of the city center; and 45 videos
are delivered to point F, 2 miles east of the city center. Production costs are
the same at all locations.
80 Part I Marker Forces in the Development of Cities
a. Using a graph, show where the firm should locate. Explain your location
choice.
b. Suppose that point W is in a valley and point F is at the top of a mountain.
Therefore, the unit cost of easterly transport (shipments from west to east)
is twice the unit cost of westerly transport. If production costs are the same
at all locations, where should the firm locate? Explain.
7. Figure 3^ shows the location choice of Ann's pizza firm. Discuss the effects
of the following changes on Ann's location choice.
a. A tripling of the distance between Y and Z (from 7 miles to 21 miles).
b. A tripling of the number of customers at point W. Instead of two customers
at W, there are six.
c. Ann stops delivery service, forcing consumers to travel to the pizza parlor.
8. In Figure 3-6, the weight-losing firm is located at point P (the port). If the
monetary weight of location B is $27 instead of $15, will the firm still locate at
point P?
9. There is some evidence that people have become more sensitive to air
pollution. In other words, people are willing to pay more for clean air. If this is
true, what influence will it have on the location decision of firms?
10. Consider a firm that uses one transferable input to produce one output. The
monetary weight of the output is $4. and the monetary weight of the input is
$3. The distance between M
(the market) and F (the input source) is 10 miles.
a. Suppose that production costs are the same at all locations. Using a diagram
like the one in Figure 3-1. explain where the firm will locate.
b. Suppose that the cost of land (a local input) increases as one approaches the
market. Specifically, suppose that the cost of land is zero at F. but increases
at a rate of $2 per mile as the firm approaches M. Depict graphically the
location choice of the firm.
1 1 Chapter 2 discusses the incubator process. When industries mature, they move
from single-activity clusters to areas with lower land and labor costs. Explain
this process in terms of changes in the orientation of firms as they mature.
12. Suppose that country L has a plentiful supply of labor (and low wages) but a
relativelylow supply of raw materials. In contrast. H has a plentiful supply of
raw materials, but a relatively low supply of labor (and high wages). The two
countries are separated by a mountain range that makes travel between the two
countries very costly. Suppose that a weight-losing product is initially
produced in H (close to the supply of raw materials). Suppose that a tunnel is

bored through the mountain, decreasing the costs of shipping raw materials
and output between the two countries. Assume that laborers do not migrate
from one country to the other.
a. How will the tunnel affect the location choices of weight-losing firms' 1
b. How w ill the tunnel affect wages in the two countries?
c. How might this analysis the shift in manufacturing
be used to explain 1 1 )
from the United States to East Asian countries and (2) the narrowing of
the wage differential between the United States and East Asian
countries?

Location Theory of Firms
Alonso. William. "Location Theory." In Readings in Urban Economics, ed. M. Edel and
J. Rothenberg. New York: Macmillan, 1972. Outlines the theory of firm location,
focusing on the role of transport costs. Contains a brief bibliography of the seminal
works in location theory.
Hoover. Edgar M. The Location of Economic Activity. New York: McGraw-Hill. 1963.
Discusses the economics of industry location, urban structure, and regional
development, focusing on the problems of locational change and adjustment.
Regional Economics. New York: Alfred A. Knopf, 1975. Chapters 2, 3, and 4 provide
.
a concise analysis of the location decisions of firms.

Krugman, Paul.Geography and Trade. Cambridge. MA: MIT Press, 1991. A series of
lectures on the topic of "economic geography." defined by Krugman as "the location of
production in space." Discusses the theory of localization economies and empirical
evidence of geographic concentration of industries.
Richardson, Harry W. Regional Economics. Urbana: University of Illinois Press. 1979,
pp. 37-80.
Location of People: Migration and Amenities
Graves, Phillip E. "A Life-Cycle Empirical Analysis of Migration and Climate by Race."
Journal of Urban Economics 6 ( 1 979), pp. 1 35-47. Estimates the effect of climate
variables on migration patterns in the 1960s.
Greenwood, Michael. "Human Migration: Theory, Models, and Empirical Studies." Journal
of Regional Science 25 (1985), pp. 521-44. Survey of the theoretical and empirical
on migration.
literature
Porell,Frank W. "Intermetropolitan Migration and Quality of Life." Journal of Regional
Science 22 (1982), pp. 137-58. Estimates the effects of economic and quality-of-life
variables on migration patterns.
Applications of Location Theory
Levy, John M. Economic Development Programs for Cities, Counties, and Towns. New
York: Praeger Publishers, 1981. Discusses the factors affecting the location choices of
firms.
Schmenner, Roger. "Energy and Location." In Energy Costs, Urban Development, and
Housing, ed. Anthony Downs and Katherine Bradbury. Washington, D.C.: Brookings
Institution, 1984. Discusses the role of energy costs in location decisions.
Storper, Michael. "Toward a Theory of Industrial Location." In Industrial

Structural
Location and Regional Systems, ed. John Rees, Geoffrey J. D. Hewings, and Howard
Stafford. Brooklyn, N.Y.: Bergin, 1981, pp. 17-40.
Industry and Case Studies of Location Decisions
Bartik,Timothy J.; Charles Becker; Steve Lake; and John Bush. "Saturn and State
Economic Development." Forum for Applied Research and Public Policy, Spring 1987,
pp. 29-40.
82 Part I Market Fares in the Development of Cities
Castells, Manuel. "The New Industrial Space: Information Technology Manufacturing and
Spatial Structure in the United States." In America's New Market Geography, ed.
George Sternlieb and James Hughes. New Brunswick. N.J.: Center for Urban Policy
Research. 1988. Examines the location patterns of information-technology industries
(semiconductors, computers, communication equipment, electronic automated
machines, and genetic engineering).
Hekman. John S. "Branch Plant Location and the Product Cycle in Computer
Manufacturing." Journal of Economics and Business 37 1985). pp. 89-102. Examines (
the influence of product development on the location decisions of computer

manufacturers.
Schmenner. Roger W. Making Business Location Decisions. Englewood. N.J.: Prentice
Hall. 1982. Discusses how firms make location decisions.
.The Manufacturing Location Decision: Evidence from Cincinnati and New England.
Cambridge. Mass.: Harvard-MIT Joint Center. 1981.
Empirical Studies of Business Location Decisions
Anderson. John and Robert W. Wassmer. "The Decision to Bid for Business':
E..
in Granting Property Tax Abatements." Regional Science and

Municipal Behavior
Urban Economics 25 1995). pp. 739-57.
(
Bartik. Timothy J. Who Benefits from State and Local Economic Development Policies?
Kalamazoo. Mich.: W. E. Upjohn Institute. 1991. Chapter 2 summarizes the results
from dozens of studies of firms' location decisions, focusing on the effect of public
policy on location choices.
Carlton. Dennis W. "The Location and Employment Choices of New Finns: An
Econometric Model." In Interregional Movements and Regional Growth, ed. William
C. Wheaton. Washington. D.C.: The Urban Institute. 1979. Estimates the effects of
wages, energy costs, industry output, and taxes on location decisions.
Helms. L. Jay. "The Effect of State and Local Taxes on Economic Growth: A Times
Series-Cross Section Approach." Review of Economics and Statistics 68 ( 1985 >.
pp. 574-82. Estimates effects of taxes on location decisions and growth rates;
concludes that the effect of a tax increase depends on how the tax revenue is spent.
Munnell. Alicia H. "How Does Public Infrastructure Affect Regional Economic
Performance?" New England Economic Review. 1990. pp. 1—33. 1
Smith. Donald F.. and Richard Florida. "Agglomeration and Industrial Location: An
Econometric Analysis of Japanese-Affiliated Manufacturing Establishments in
Automotive-Related Industries." Journal of Urban Economics 36 1994). pp. 23-41. (
Urban and Regional Growth Patterns

Mieszkowski. Peter. "Recent Trends in Urban and Regional Development." In Current
Issues in Urban Economics, ed. Peter Mieszkowski and Mahlon Straszheim.
Baltimore: Johns Hopkins. 1979. pp. 3-39. Describes the broad patterns of urban and
regional development since World War II. focusing on interregional migration, shifts in
employment, and changes in regional income differentials.
Norton. R. D. City Life Cycles and American Urban Public Policy. New York: Academic
Press. 1979. Explores the changes in the spatial pattern of growth in manufacturing and
discusses the implications for public policy.
Norton, R. D., and

J. Rees. "The Product Cycle and the Spatial Decentralization of
American Manufacturing." Regional Studies 13 (1979). pp. 141-51. Analyzes changes

in the spatial pattern of growth in manufacturing.
The Origins of American Cities
Boorstin, Daniel. The Americans: The Democratic Experience. New York: Vintage, 1974;
and The Americans: The National Experience. New York: Vintage. 1965. American
history. Discusses the influences of government on the location of cities (the siting of
state capitals, county seats, and universities; the influence of divorce and gambling
laws). Part Three, "The Upstarts: Boosters," of The Americans: The National
Experience discusses some of the reasons for the development of U.S. cities, focusing
on the role of local boosters.
Glaab, Charles, and A. Theodore Brown. A History of Urban America. 3rd ed. New York:
Macmillan, 1983. A textbook on American urban history. Discusses the origins of
many American cities.
Transport Costs and Industry Location
Chinitz, Benjamin. "The Effect of Transportation Forms on Regional Economic Growth."

Traffic Quarterly 14 (1960a), pp. 129-42.
. Freight and the Metropolis. Cambridge, Mass.: Harvard University Press, 1960b.
Chapter 4
&~ HE HISTORY of
WESTERN URBANIZATION
"
"Dost thou know how to play the fiddle?
"No," answered Themistocies, "but I understand the art of raising a little village
into a great city."
Motto on the masthead of the Emigrant Aid Journal
C~~7~ his chapter uses the concepts of Chapters 2 and 3 to discuss the historical
K-S development of cities in the Western world. The earlier chapters explained
that three factors contribute to the development of cities: comparative advantage,
scale economies, and agglomerative economies. Together, these three factors explain
why cities exist and where they develop. This chapter takes a historical perspec-
tive, discussing how changes in technology affected comparative advantage, scale
economies, and agglomerative economies, and how these technological changes af-
fected the development of cities. Changes in technology caused cities to grow and
shrink,and eventually caused the transformation of the earth from a rural world into
an urban one.
The First Cities

The first cities developed Near East around 3000 B.C.
in the fertile river valleys in the
The city-states that developed in the Mesopotamia Valley included Eiridu, Ur, La-
gash, and Kish. Ur was the largest of these cities, with a land area of about 150 acres
and a population of about 25,000. In 2000 B.C., the city of Babylon had a population
of about 50,000. In the Nile River Valley, the leading cities were Memphis, Helipolis,
and Thebes.
A necessary condition for the development of a city is an agricultural surplus.
A number of agricultural breakthroughs preceded the development of cities in the
Near East: grain was domesticated, irrigation systems were developed, and the plow
85
86 Part I Market Farces in the Development of Cities
replaced the hoe. Together, these innovations helped agricultural workers to produce
a surplus of food, allowing some people to pursue nonagricultural activities in cities.
The food surplus was relatively small. According to Davis (1976), between 50 and
150 farmers were required to support a single city inhabitant.
Why did the first cities develop? There is some uncertainty about the social
and economic features of early societies, so historians can only speculate about the
origins of cities. Most agree that the first cities served both religious and defensive
purposes. Can the development of defensive and religious cities be explained by the
same factors that explain the development of market-based cities? If scale economies
are responsible for the development of market-based cities, did the first cities develop
as a result of scale economies in defense and religious services?
The Defensive City
A farmer who generates an agricultural surplus will eventually use the surplus, either
consuming it himself or trading it for other goods. In either case, he must store the
surplus food for some period of time. The stored food provides a lucrative target for
thieves. If there are scale economies in food storage and protection, farmers may be
better off storing their surpluses in a central storage facility. The people working in the
fortified storage facility (managers and guards) will live near the facility, generating
a place with a relatively high population density, a small city. This is the theory of the
defensive city: the first cities developed because of scale economies in the storage
of the agricultural surplus. Support for the theory comes from archaeologists who
uncovered the remains of fortified storage facilities in the first cities.
How does this picture of the defensive city fit into the analysis of the market-
based city in Chapter 2? The agricultural surplus generated a new economic activity:
the storage and protection of the surplus. This activity is subject to scale economies,
meaning that a single fortified storage facility is more efficient than a series of facili-
ties,one for each farmer. Home storage was replaced by centralized storage, causing
the development of a specialized labor force of managers and guards, who exchanged
their services for a part of the agricultural surplus. In other words, cities developed
because of scale economies in the provision of a new commodity, grain storage. Just
as scale economies in cloth production caused the development of factory cities,
scale economies in storage services caused the development of defensive cities.
The Religious City

According to Mumford ( 1961 ), the development of the first cities coincided with the
development of large-scale religion. Before the development of cities, most people
worshipped in small groups, either in the home or in the local village. Around the time
that the first cities developed, the local earth gods were replaced by celestial gods,
who apparently demanded worship on a grander scale. If adoration and supplication
aremore effective on a large scale, a doubling of the size of the shrine more than
doubled the religious output. Large temples at central locations replaced the small
shrines in homes and villages. The temples employed chieftains, priests, and religious
Chapter 4 The History oj Western Urbanization 87
workers, causing the development of a place with relatively high population density,
a city. This is the theory of the religious city: the earliest cities developed because
of scale economies in the provision of religion. Support for the theory comes from
archaeologists who uncovered the remains of large temples in the first cities.
How does the theory of the religious city fit into the analysis of the market-
based city in Chapter 2? The switch from earth gods to celestial gods caused home
production of religion to be replaced by centralized production, and a specialized
labor force of religious workers exchanged their labor for a part of the agricultural
surplus. In other words, cities developed because of scale economies in the provision
of religion. Just as scale economies in cloth production caused the development
of factory cities, scale economies in religious services caused the development of
religious cities.
The Defensive and Religious City
The most prominent feature of the early city was a large temple at the city center. The
temple was a large monument with thick walls and extravagant decorations. The thick
walls were presumably used to both impress the gods and protect the agricultural
surplus stored in the temple. In other words, the temple served both religious and
defensive purposes, and early cities served as both religious and military centers.
What came first, the central storage facility or centralized religion? Perhaps the
central storage facility made worship in the central temple more convenient, causing
the switch to centralized religion. Alternatively, the development of a large temple
could have provided a convenient and defensible storage facility for the agricultural
surplus. In other words, therewere economies of scope in the provision of religion
and defense: the two services were provided more efficiently if they were provided
together. A third possibility is that the local chieftains used centralized religion as a
subterfuge to extract the agricultural surplus from local farmers.
The military played a major role in the early cities. The development of cities in-
creased the frequency and severity of human conflict for two reasons. First, the cities
worshipped different gods and fought wars to settle religious disputes. Second, the
urban societies accumulated wealth, providing lucrative targets that encouraged ma-
terialistic wars. Although cities encouraged aggressive behavior, they also provided
the most effective defense against aggression. Until the development of gunpowder
in the fourteenth century, the most effective defensive maneuver was a large but sim-
ple wall. In addition, a city contained enough people to defend itself against attack
from large forces. Once the simple wall was built, there was safety in numbers.
Greek Cities
The next stage of western urbanization occurred in Greece. In 500 B.C., there were
hundreds of independent city-states, ranging in population from a few hundred to
tens of thousands. Athens was the largest, with a population of about 150,000, and
Sparta had a population of about 40.000. In contrast with earlier cities, which were
controlled by chieftain-priests, most Greek cities were run by their citizens, using
public assemblies to make policy.
Athens was a market city, with trading activity centered in the Agora. Early
in its history, the city establishedindependent colonies, and traded its household
crafts and olive products for food and raw materials from the colonies. Trade was
facilitated by the stamping of gold and silver coins in the seventh century B.C. In this
early period. Athens fed itself through voluntary trade with other areas.
The Athenian empire developed in the aftermath of the successful war against
Persia in the fifth century B.C. After the Greek city-states repelled the Persian inva-
sion, theyformed the Delian League to carry the war into Asia Minor. By the end
of the successful campaign. Athens had assumed complete control of the league,
dictating policy to the other city-states. Athens took control of the treasury, trans-
forming the voluntary contributions of member city-states into payments of tribute
to Athens.
The Peloponnesian War between the Athenian Empire and Sparta (431 to
404 Because Athens continued to use the Delian League to
B.C.) devastated Greece.
demand homage and tribute from the lesser cities, it appeared that war was inevitable.
The war. which was precipitated by disputes between Athens and Corinth (an ally
of Sparta) over two Corinthian cities, ended in 404 when Athens renounced control
over its empire and demolished its defensive walls. The war caused large losses in
manpower, and also caused many intellectuals to lose faith in the democratic system.
Athens never regained its former power. Eventually. Philip of Macedonia was able
to take advantage of the weakened city-states, and expanded his kingdom to include
most of Greece.
Roman Cities
The next stage of western urban development occurred under the Roman Empire.
By the third century A.D.. Rome had a population exceeding 1 million. The Romans
set up colonies to the north and west, establishing colonial cities throughout Europe.
Rome fed its large population with a combination of trade and tribute, with tribute
playing a relatively large role.
The Roman were eventually overrun by marauders from rural areas. Ac-
cities
cording to Hohenberg and Lees 1985). the Roman economy emphasized the col-
(
lection of the agricultural surplus and neglected production activity. Instead of ex-
changing urban goods for agricultural products. Rome used conquest and tribute to
feed its population. In the fourth and fifth centuries. German tribes invaded from
the north, disrupting the Roman collection system. It appears that there was little
interest outside of Rome in restoring the "trade" routes, so the losses from succes-
sive invasions were cumulative. If Rome had relied to a greater extent on voluntary
exchange, the colonies might have been more interested in maintaining the exchange
network, and the Western empire might have recovered from the Germanic raids.
What are the lessons from the rise and fall of Athens and Rome? Early in
its history. Athens engaged in voluntary trade with other areas, exchanging urban
Chapter 4 The History of Western Urbanization 89
goods for food from the countryside. The city thrived under this system of voluntary
exchange. The Athenians eventually switched to a system of conquest and tribute,
resulting in war and the decline of the city. Mumford (1961) suggests that the city
of Rome should have been called "Parasitopolis" to indicate the extent to which its
population lived off the labors of outsiders. The decline of Rome was caused in part
by the disruption of its collection system by the Germanic raids. Perhaps the lesson
is that cities based on voluntary exchange are viable, but cities based on coercive
transfer payments are not.
Feudal Cities
In the first few centuries following the fall of the Roman Empire, cities in the West
declined. The Islamic conquest disrupted trading on the Mediterranean, causing the
decline of port cities. Waves of marauding barbarians continued to sweep across
Europe, making travel and trade dangerous. People sought safety inside city walls,
where, once again, there was safety in numbers.
Table 4-1 shows the largest cities in Europe between 1000 and 1900. In 1000,
most cities in western and central Europe were very small. The largest cities served
the Byzantine Empire and the Muslim areas of Spain. Venice was one of the few
large cities beyond the Byzantine and Muslim areas, but it depended heavily on trade
with the Eastern empire.
Trade on the Mediterranean increased during the eleventh and twelfth centuries.
The Italian city-states forged agreements with the Byzantine and Islamic rulers for
trade with North Africa and the East. The Europeans traded wood, iron, grain, wine,
and wool cloth for medicines, dyes, linen, cotton, leather, and precious metals. The
increased trade contributed to the growth of Venice, Genoa, and Pisa.
Table 4-1 European Cities, 1000-1900
1000 1400 1700 1900 1
Population Population Population Population

City (thousands) City (thousands) City (thousands) City- (thousands)
Constantinople 450 Paris 275 Constantinople 700 London 6,480

Cordoba 450 Milan 125 London 550 Paris 3,330
Seville 90 Bruges 125 Paris 530 Berlin 2.424
Palermo 75 Venice 110 Naples 207 Vienna 1,662
Kiev 45 Granada 100 Lisbon 188 St. Petersburg 1.439
Venice 45 Genoa 100 Amsterdam 172 Manchester 1.255
Regensburg 40 Prague 95 Rome 149 Birmingham 1,248
Thessalonika 40 Caffa 58 Venice 144 Moscow 1.120
Amalfi 35 Seville 70 Moscow 130 Glasgow 1.072
Rome 35 Ghent 70 Milan 124 Liverpool 940
SOURCE: P. M. Hohenberg and L. H. Lees, The Making of Urban Europe 1000-1950 (Cambridge, Mass.: Harvard University Press. 1985).
The feudal economy of the eleventh through the fourteenth centuries was based
on manorial estates and small walled cities. On the manor, the lord inherited his serfs,
who worked the lord's land and served in the military in exchange for the use of a
small plot of land. In the small towns, workers produced handicrafts (cloth, leather
goods, metal goods), which they exchanged for the agricultural surplus of the manor.
Cities in the feudal era were small and numerous. In the eleventh century, London
was the largest city in England, with a population of only 16,000. There were several
other English cities with populations of around 5,000. The largest of the many
German had a population of no more than 40,000.
cities
Between the eleventh and fourteenth centuries, the frequency of barbarian inva-
sions decreased, and the small defensive cities gradually became market cities. The
medieval cities specialized in commerce and handicrafts, and thus earned rather —
than robbed —
the agricultural surplus of their hinterlands. A small merchant class
developed, and marketplaces were established just outside the city walls. The mer-
chants were protected from plunder by the establishment of weekly market days:
once a week, theft was outlawed in places marked by the market cross. The market
peace was enforced by local chieftains, special courts with jurisdiction over traders,
and the church. The markets flourished, and the city walls were extended to include
the marketplaces. The urban market was primarily a place for the exchange of local
agriculture and handicrafts. The producers of handicrafts made up a large fraction
of the city's workforce.
Hohenberg and Lees (1985) use Leicester, in the East Midlands of England, as
an example of a medieval city in the fourteenth century. The city was surrounded by
walls on three sides and a river on the fourth side. The town ditch was just outside
the wall. The city was a regional marketplace, serving as a commercial center for
the surrounding county. The city produced staples such as beer and bread for local
consumption and produced woolen cloth for export. Occupations in the city included
butchers, shoemakers, tailors, mercers, weavers, and bakers. Over half of the city's
workers were employed in manufacturing, with the other half employed as food-
makers (25 percent), traders and merchants (10 percent), builders (5 percent), and
other occupations. The merchant guild controlled most economic activity in the city.
According to Davis (1976), the first market-based cities developed for two rea-
sons. First, power in the feudal system was decentralized. Unlike the Greek and
Roman cities, medieval cities could not simply dominate their hinterlands and de-
mand tribute, but had to produce something in exchange for agricultural goods.
Second, agricultural productivity was relatively low, so a city could not survive on
the output of its immediate hinterland, but had to trade with a relatively large area.
As a result, the city had to develop products that could compete with homemade
products in its hinterland and products produced in other cities.
Competition among the medieval cities caused innovations in production and
commerce. Urban producers developed new production techniques, allowing them
to underprice their competitors. Cities invested in secular education as a means ol~
promoting literacy and developing commercial skills. These early efforts to improve
the techniques of production and commerce set the stage for the industrial revolution
of the eighteenth and nineteenth centuries.

Table 4-1 shows the top 10 European cities in was a

1400. Paris, the largest city,
center for trade and education, and also the capital of the kingdom of France. Three
of the top six European cities were Italian city-states. Bruges and Ghent, two cities
in the Low Countries, grew rapidly in the fourteenth century, a result of the booming
woolen cloth industry.
A series of famines and plagues between 1350 and 1450 decreased Europe's
population by between one-third and one-half. Cities suffered more than rural areas
because cities had less sanitary living conditions and higher population density (an
urban sneeze transmits germs to more people). The associated disruptions of social
and economic activities caused most cities to stagnate or decline.
Mercantile Cities
Starting in the fifteenth century, large mercantile cities developed in Europe. Two
factors contributed to the development of large cities: the centralization of power
and the growth of long-distance trade.
Centralization of Power
and economic power was transferred from a large

In the fifteenth century, political
number of feudal lords to a relatively small number of princes, queens, and kings.
The shift in power was caused in large part by military advances that rendered the
traditional feudal defense maneuvers obsolete.
Innovations in warmaking increased the seriousness and scale of conflict. The
feudal lord used his serfs as part-time warriors. If the part-time warriors could not
defeat the attackers in the field, the feudal lord could always hunker down behind
his castle walls to wait out a siege. The professional army of the fifteenth century
combined infantry (armed with pikes, crossbows, and muskets) with siege cannons
and cavalry. The professional soldiers, with their sophisticated weapons, defeated
the feudal lord's part-timers in the field. When the feudal lord retreated to the castle,
the siege cannon blew holes in the castle walls. The professional armies defeated the
feudal lords, centralizing power in the hands of princes, queens, and kings.
The mercantile cities developed defensive maneuvers appropriate to the new
level of warfare. Defense was no longer a matter of building a simple wall and
dumping hot oil on the attackers. It required the building of intricate fortifications
and the hiring of professional soldiers. The larger cities were able to exploit the
increased scale economies in defense.
The centralization of power caused the centralization of administrative and mil-
itary functions in royal cities. After the sixteenth century, the most rapidly growing
cities were the ones that harbored a royal court. In a short period of time, more
than a dozen cities grew to a size attained by only a few medieval cities: London
had 250,000 people, Naples 240,000, Milan 200,000, and Paris 180,000; cities with
around 100,000 inhabitants included Rome, Lisbon, Palermo, Seville, Antwerp, and
Amsterdam.
Long-Distance Trade
Long-distance trade increased for two reasons. First, the consolidation of power
caused a lifting of many restrictions on trade. When power was centralized, the
imposed by feudal lords disappeared and trade increased. Second, the
local tariffs
development of ocean travel led to exploration and the discovery of new markets.
Cities developed along trade routes and at transshipment points. While cities with
river and sea ports flourished (Naples. Palermo. Lisbon, Liverpool), inland cities
like Florence declined. In Table 4—1, four new seaports appeared among the top 10
European cities in 1700: London, Naples, Lisbon, and Amsterdam.
Urbanization during the Industrial Revolution

Despite the rapid spread of cities between 3000 B.C. and 1800 A.D.. the world con-
tinued to be predominantly rural. Urbanization was limited by the relatively low
productivity of agriculture, the high costs of transporting goods (which limited the
volume of trade), and the relatively small advantages of centralized production. Un-
til the early part of the nineteenth century, the fraction of the world's population
living in cities was only about 3 percent. Between 1800 and 1970. the percentage
of the population in urban areas increased to 39 percent. In the United States, the
percentage of the population in urban areas increased from 6 percent in 1 800 to over
75 percent in 1990.
The rapid urbanization of the last two centuries was caused by the industrial
revolution, which started in the nineteenth century. The industrial revolution pro-
duced innovations in manufacturing and transportation that shifted production from
the home and the small shop to large factories in industrial cities. Table 4-1 shows
the effects of the industrial revolution on the top 10 cities in Europe. Between 1700
and 1900. Manchester, Birmingham. Glasgow, and Liverpool grew from tiny towns
into giant industrial cities. Most of the other cities in the top 10 combined political
functions with industrial development, causing dramatic growth. The innovations of
the industrial revolution can be divided into four areas: agriculture, manufacturing,
transportation, and construction.
Agriculture
The rapid urbanization of the industrial revolution was made possible by innovations
that increased agricultural productivity. Farmers substituted machinery for muscle
power and simple tools, increasing the output per farmer. For example, using a horse-
drawn reaper, two people could harvest the same amount of grain as eight people
using traditional harvesting methods. In addition, the development of agricultural
science led to innovations in planting, growing, harvesting, and processing. As pro-
ductivity increased, laborers were freed from food-raising responsibilities, allowing
them to pursue other activities. In the United States, the share of employment in
agriculture decreased from 69 percent in 1840 to 2.8 percent in 1988.

Manufacturing
Perhaps the most visible part of the industrial revolution involved innovations in man-
ufacturing. New machines, made of iron instead of wood, were developed for the pro-
duction of most goods. Manual production by skilled artisans was replaced by mech-
anized production using interchangeable parts, specialized labor, and steam-powered
machines. Output per worker rose, and scale economies in production increased.
The innovations in manufacturing caused the development of large industrial
cities. Mass production decreased the relative cost of factory goods, causing the
centralization of production and employment. In addition, many of the new produc-

tion processes were subject to agglomerative economies (localization economies and
urbanization economies), increasing the advantages of urban locations.
Intercity Transportation
Innovations in intercity transportation contributed to industrialization and urbaniza-

tion. The steamship and the railroad decreased the costs of moving goods between
cities, decreasing the delivered price of factory goods. Production became more
centralized, and factory cities grew. The new transportation modes also decreased
the costs of moving agricultural goods, allowing greater regional specialization in
agriculture. Agricultural regions were better able to exploit their comparative ad-
vantages, so agricultural productivity increased.
Before the railroad linked eastern cities with the western parts of the United
States, a large fraction of the food consumed in New England was produced by the
region's farmers. Similarly, a large fraction of the goods consumed in the West (e.g.,
shoes) were produced by local craftsmen. The coming of the railroad allowed eastern
factories to underprice western craftsmen, and allowed western farmers to underprice
eastern farmers. Trade increased because the railroad allowed both regions to exploit
their comparative advantages.
Industrial cities developed at points accessible to the new intercity transporta-
tion network. During the steamship era, cities developed along the rivers and lakes
served by the steamship. Firms that fueled their steam engines with coal located
along rivers served by coal shippers. Later, the development of the railroad caused
the development of cities along the rail lines. The cities that developed at trans-
shipment points had commercial jobs (transportation, trade, business services) and
manufacturing jobs.
Intracity Transportation
Innovations in intracity transportation increased the feasible size of cities. The size
of a city is limited by the cost of traveling within the city. One rule of thumb is that
the geographical area of the city should be small enough that the typical resident can
travel from the edge of the city to the city center in an hour. Before the innovations
in intracity transport, most people walked to workplaces and shops, so the radius
of the city could be no more than two miles. During the last half of the nineteenth
94 Par// Market Fonts in the Development of Cities
century, a series of innovations in public transit increased the speed of intracity

travel, increasing the radius of the city.
The Boston provides a good example of the effects of transit innovations

city of
on city size. In 1850. Boston was a "walking" city with a radius of about two
miles. In the 1860s, the horse-powered railroad (horses pulling cars along rails) was
introduced, and the radius of the city increased to 2.5 miles in 1872 and 4.0 miles
in 1887. By the 1890s. the horse railroad was replaced by the electric trolley, which
traveled at twice the speed and carried three times as many passengers. As a result,
the radius of the city increased to about six miles. In the 40 years during which
walking was replaced by the trolley, the radius of the city tripled and the land area
increased ninefold.
The introduction of the internal-combustion engine had similar effects on the
size of cities. This new engine allowed the development of relatively small vehicles
for transporting people and goods. The development of the automobile decreased the
cost of intracity travel, causing further increases in the feasible radius of the city. In
addition, it freed travelers from the old fixed-rail systems that were designed to deliver
people from the residential areas along the streetcar lines (the spokes) to the city
center (the hub). The automobile increased the accessibility of locations throughout
the city. Residents could live between the spokes, and firms could locate outside the
hub. As the set of feasible locations for residents and firms increased, cities grew.
The introduction of the truck had similar effects. The truck decreased the cost of
intercity transportation, increasing the advantages of centralized (urban) production.
In addition, the truck freed firms from their dependence on the railroad terminal and
the port, allowing firms to locate outside the central city. As the set of feasible
production locations increased, the feasible population of the city increased. The
development of the truck and the automobile also caused the suburbanization of
population within the urban area, a topic to be covered in Chapter 10.
Construction Methods
Innovations in construction methods also increased city sizes. The first skyscraper.
a 10-story building that housed the Home Insurance Company, appeared in 1885 in
Chicago. The building was revolutionary because its frame was made of steel instead
of bricks. Because the steel frame was relatively light, the steel-framed building
could be taller than the traditional brick building. The development of the elevator
decreased intrabuilding travel costs of tall buildings, increasing the feasibility of the
skyscraper. The skyscraper increased the intensity of land use. increasing the city's
productive capacity and its feasible population.
*5 rbanization in the United States

Figure 4-1 shows the trend in U.S. urbanization from 1800 to 1990. In 1800. only
about 6 percent of the population lived in cities. The share of population in cities
increased to 15 percent by 1850. 40 percent by 1900, 64 percent by 1950. and

Figure 4-1 Percent of U.S. Population Living in Urban Areas,

1800-1990
80
75
70
65 h
60
h 55
«
§ 50
I 45
J 40
c 35
| 30
25
20
15
10
o
1800 1820 1840 1860 If 1900 1920 1940 1960 1980 2000
Year
75 percent by 1990. Urbanization in the United States can be divided into three
periods. Between 1800 and 1930, the traditional port city eventually gave way to
the industrial city. Urbanization was rapid during this period: the percentage of
the population living in cities increased from 6 percent to 56 percent, and most
metropolitan areas grew rapidly. The 1930s and 1940s were times of depression
and war, and urbanization proceeded at a relatively slow pace: the percentage of
the population in cities increased from 56 percent in 1930 to 59 percent in 1950.
There are two figures for 1950. Under the old, more restrictive, census definition
of an urban area, 59 percent of the population lived in urban areas; under the new
definition, the figure is 64 percent.
During the early part of the postwar period, urbanization proceeded at a moderate
pace. The percentage of the population living in urban areas increased from 64
percent in 1950 to 73.6 percent in 1970. During the 1970s, there was a tiny increase
in the percentage of the population in urban areas (73.6 percent to 73.7 percent).
During the 1980s, the percentage of the population in urban areas increased by
1.5 percentage points, from 73.7 percent to 75.2 percent.
From Commercial to Industrial Cities
Table 4-2 shows the population trends for the 10 largest U.S. cities in 1930. In 1810,
only four cities (New York. Philadelphia. Baltimore, and Boston) had more than
10.000 residents. The rapid growth of these cities around the turn of the century was
caused in large part by the Napoleonic Wars, which tied up the English and French
merchant marine and caused a boom in the American shipping industry. In 1 850. the
four seaports were still the dominant cities, with the population of the smallest one
(Philadelphia) about 2.5 times the population of the next largest city (Pittsburgh).
The industrial revolution caused dramatic growth in U.S. cities. Between 1850
and 1890, the populations of New York and Boston more than tripled, the pop-
ulation of Baltimore more than doubled, and the population of Philadelphia in-
creased more than eightfold. Lurking behind these changes were large increases in
urban manufacturing activity. The share of employment in manufacturing more than
doubled in all of these cities, from about 9 percent of total employment to about
19 percent.
The most dramatic urban growth between 890 occurred away from
1 850 and 1
the eastern seaboard in Chicago, Detroit, Cleveland. Pittsburgh, and Los Angeles.
On average, the populations of these cities increased over 18-fold between 1850 and
1890. Again, manufacturing growth led the way. In Chicago and Detroit, the per-
centage of employment in manufacturing almost quadrupled, from about 5 percent
to around 19 percent. In Cleveland, the percentage in manufacturing more than dou-
bled, from 8 percent to 19 percent. Pittsburgh was already an industrial city by 1860.
so it experienced a relatively small increase in manufacturing employment (from
1 8 percent to 23 percent). Chicago was the first city to displace one of the eastern
seaport cities from the top four positions, moving into the second spot by 1890.
Table 4-2 Cities in the United States, 1810-1930
Population in Year (thousands)
City Rank in 1930 1810 1850 1890 1930
New York 1 120 696 2,507 6.930

Chicago 2 — 30 1,100 3.376
Philadelphia 3 54 121 1 .047 1 .95
Detroit 4 — 21 206 1.569
Los Angeles 5 — 2 50 1.238
Cleveland 6 — 17 261 900
Baltimore 7 47 169 434 S05
Boston 8 34 137 448 781
Pittsburgh 9 5 47 239 670
Washington 10 8 40 189 4S7
Si h ri i is Bureau of the Census. Census oj Population, Washington, D.C..

1972
Between 1890 and 1930, Philadelphia, Baltimore, and Boston were growing at
relativelyslow rates, allowing other cities to move ahead of them in the ranking.
Detroit moved to number four; Los Angeles moved to number five; and Cleve-
land moved to number six. By 1930, most of the largest cities were industrial
cities.
The Postwar Era

As shown in Figure 4-1, the pace of urbanization has slowed since 1950. Using
the new definition of an urban area, the percentage of the population living in cities
increased by 4.9 percentage points during the 1950s, 3.6 percentage points during
the 1960s, 0.1 percentage points during the 1970s, and 1.5 percentage points during
the 1980s.
Figure 4-2 shows the percentages of population living in urban areas for the
United States as a whole and for four regions in 1 990. The western region is the most
urbanized, with 86 percent of its population in urban areas. The western region has
five of the six most urbanized states: California (93 percent), Hawaii (89 percent),
Nevada (88 percent), Arizona (88 percent), and Utah (87 percent). The other highly
urbanized states are New Jersey (89 percent), Florida (85 percent), Massachusetts
(84 percent), and New York (84 percent). The southern region is the least urbanized,
with 69 percent of its population in urban areas.
Figure 4-2 Percent of U.S. and Regional Populations in Urban Areas in

1990
90
86
79
75
72
70
c 60
o
id
d. 50
S 40
U
30
20
10
U.S. Northeast Midwest South West

Figure 4-3 Growth Rates in Metropolitan and

Nonmetropolitan Areas
2.00
1.50 —
2 1.00
0.50
0.00
1960s 1980s
| United States
^] Metropolitan areas
J Nonmetropolitan areas
Metropolitan versus Nonmetropolitan Growth
Figure 4-3 provides a closer look at urbanization between 1960 and 1990. The
figure shows the population growth rates for metropolitan areas (areas included in
anMSA or a CMS A) and nonmetropolitan areas (counties that are not a part of an
MSA or a CMSA). During the 1960s, metropolitan areas grew more than four times
faster than nonmetropolitan areas ( 1.6 percent per year compared to 0.35 percent).
In contrast, during the 1970s, nonmetropolitan areas grew about 25 percent faster
than metropolitan areas ( compared to .04 percent During the
1 .29 percent per year 1 ).
1980s, metropolitan areas grew more than three times faster than nonmetropolitan
areas (1.29 percent per year compared to 0.39 percent).
Nonmetropolitan areas grew faster than metropolitan areas in the 1970s tor a
number of reasons.
1. Dispersion of manufacturing employment. Improvements in the

intercity highwaj system increased the accessibility of nonmetropol-
itan locations. Some labor-intensive manufacturers moved from
metropolitan areas to low-wage nonmetropolitan areas.
2. Slow decline in farm employment. Although total employment in
agriculture decreased during the decade, it decreased at a relative!) slow

rate. Fewer people left farms to move to cities.
3. Growth in extractive industries. Increased prices for energyand raw

materials increased employment in oil and mining industries. For
example, total mining employment increased from 626,000 in 1969 to
957,000 in 1979.
4. Growth in recreation and retirement activities. A large number of
elderly households moved to retirement communities in nonmetropolitan
areas, pulling retail and service jobs to retirement areas. Other urban
households moved to nonmetropolitan areas with outdoor recreational
opportunities.
5. Population spillovers. Some households moved to counties that the

Census Bureau inappropriately classifies as "nonmetropolitan." Some of
these counties are integrated, in an economic sense, with a nearby MSA.
Alonso (1980) suggests that this statistical problem causes a significant
overestimate of the nonmetropolitan growth rate. During the 1970s, many
of the fastest-growing areas were nonmetropolitan counties adjacent to
metropolitan areas (MSAs and CMSAs).
During the 1980s, the growth rate for nonmetropolitan areas decreased and the
growth rate for metropolitan areas increased, restoring the historical pattern of rela-
tively rapid metropolitan growth. According to Garnick (1988), growth rates in non-
metropolitan areas decreased for three reasons. First, farm employment decreased
more rapidly, a result of declining farm prices. Second, the employment in extrac-
tive industries decreased, a result of decreasing energy and resource prices. Third,
labor-intensive manufacturing employment in nonmetropolitan areas decreased, a
result of increased competition from abroad.
Population Losses in Some Metropolitan Areas

Another recent trend is the decline of some metropolitan areas. Table 4-3 shows the
population growth rates for the 25 largest CMSAs and MSAs. During the 1970s, 7
of the top 25 metropolitan areas lost population: New York, Philadelphia, Detroit,
Cleveland, St. Louis, Pittsburgh, and Milwaukee. The largest losses occurred in
Cleveland (0.57 percent per year), Pittsburgh (0.53 percent per year), and New York
(0.36 percent per year). In the 1980s, Detroit, Cleveland, and Pittsburgh continued
to lose population.
Why have some metropolitan areas decreased in size? First, people migrated
from the northern states to southern and western states, decreasing the growth rates
of northeastern and north-central metropolitan areas. Between 1980 and 1990, the
average annual growth rate was close to zero for northeastern and north-central
metropolitan areas, compared to over 2 percent for southern and western metropolitan
areas. Given this migration pattern, the best a northern metropolitan area could hope
for was a small positive growth rate. In some northern metropolitan areas, large
losses in manufacturing employment resulted in negative growth rates. Decreases in
employment in primary metals, motor vehicles, rubber, and nonelectrical machinery
industries caused population losses in Pittsburgh, Cleveland, and Detroit.
100 Part I Marker Font's in the Development of Cities
Table 4-3 Population and Annual Growth Rates for 25 Largest

Metropolitan Areas
Average Annual
Growth Rate % ( I
n/JuL'
Ixlilln. J)}
lit Population in
Metropolitan Area: CMSA or MSA 1990 1990 (thousands) 1970s 1980s
New York-Northern N.J.-Long Island CMSA 1 18.087 -0.36 0.31

Los Angeles-Anaheim-Riverside CMSA 2 14.532 1.42 2.37
Chicago-Gary-Lake County CMSA 3 8.066 0.20 0.16
San Francisco-Oakland-San Jose CMSA 4 6.253 1.22 1.54
Philadelphia-Wilmington-Trenton CMSA 5 5.899 -0.12 0.38
Detroit-Ann Arbor CMSA 6 4.665 -0.07 -0.19
Boston-Lawrence-Salem CMSA 7 4,172 0.08 0.49
Washington MSA 8 3.924 0.67 1.90
Dallas-Fort Worth CMSA 9 3.885 2 23 2.86
Houston-Galveston-Brazoria CMSA 10 3.711 3.64 1.82
Miami—Fort Lauderdale CMSA 11 3.193 3.43 1.90
Atlanta MSA 12 2.834 2.42 2.86
Cleveland-Akron-Lorain CMSA 13 2.760 -0.57 -0.26
Seattle-Tacoma CMSA 14 2.559 1.31 2.03
San Diego MSA 15 2.498 3.21 2.98
Minneapolis-St. Paul MSA 16 2.464 0.76 1.43
St. MSA
Louis 17 2.444 -0.22 0.28
Baltimore MSA 18 2.382 0.51 0.80
Pittsburgh-Beaver Valley CMSA 19 2.243 -0.53 -0.77
Phoenix MSA 20 2.122 4.51 3.47
Tampa-St. Petersburg-Clearwater MSA 21 2.068 3.85 2.51
Denver-Boulder CMSA 22 1.848 2.71 1.34
CMSA
Cincinnati-Hamilton 23 1.744 0.29 0.49
Milwaukee-Racine CMSA 24 1.607 -0.03 0.23
Kansas Citv MSA 25 1.566 0.43 0.89
Source: U.S. Bureau of the Census, Statistical Abstract of the United States: IWI. Washington. DC. 1991.
The Postindustrial City

What does the future hold for cities? One economic trend that will affect cities and
the urbanization process is the shift in employment from manufacturing to services.
Table 4—4 shows the changes in the distribution of employment from 1947 to 1994.
The percentage of jobs goods-producing sector (agriculture, mining, con-
in the
struction, and manufacturing) decreased from 45 percent in 1947 to 24 percent in
1994. The percentage in the service sector transportation, wholesale and retail trade,
|
finance, insurance, and real estate, services, and government increased from 55 per- )
cent in 1947 to 76 percent in 1994. The largest percentage increases in employment

occurred in retail trade. F.I.R.E. (finance, insurance, and real estate), services, and
government.
How will the decrease in the relativeimportance of manufacturing affect cities?
Most cities in the United States developed as manufacturing centers during the
Table 4-^4 Distribution of Employment by Industry,

1947-1992
Percent of Jobs in Activity
1947 1959 1969 1979 1994
Goods-Producing Sector
Agriculture 11.60 7.40 4.00 3.66 3.10
Mining 1.70 1.20 0.80 1.03 0.51
Construction 5.20 5.50 5.40 4.80 4.28
Manufacturing 26.90 25.60 25.40 22.65 15.65
Total: Goods-producing sector 45.40 39.60 35.70 32.14 23.54
Service-Producing Sector
Transportation, Commun., Util. 7.40 6.40 5.70 5.53 5.13
Wholesale trade 4.6 5.20 5.10 5.62 5.25
Retail trade 14.60 13.90 13.40 16.12 17.47
Finance, insurance, real estate 3.30 4.20 4.60 5.63 5.93
Services 13.00 14.70 16.90 18.06 26.33
Government 11.80 16.10 18.60 17.17 16.34
Total: Service-producing sector 54.60 60.40 64.30 67.86 76.46
Total 100.00 100.00 100.00 100.00 100.00
Sources: Daniel H. Garnick, "Local Area Economic Growth Patterns: A Comparison of the
1980s and Previous Decades," Urban Change and Poverty, ed. M. G. McGeary and L. E. Lynn
(Washington. D.C.: National Academy Press. 1988); U.S. Bureau of the Census, Statistical
Abstract of the United States: Washington, D.C., selected years.
industrial revolution. The shift to a more service-oriented economy will certainly

bring dramatic changes to some metropolitan areas. The cities that continue to depend
on traditional manufacturing activity are likely to decline in population.
Summary
1 The first cities provided religious services and defense.
a. Scale economies in the storage and protection of the agricultural surplus
caused the development of centralized storage facilities and cities.
b. According to some historians, cities appeared around the time that earth
gods were replaced by celestial ones, suggesting that cities developed to
exploit scale economies in the provision of religious services.
2. The development of cities encouraged aggression. Cities provided the best

defense against aggression (simple walls).
3. Athens started out as a market-based city, but evolved into a city based on
conquest and tribute. It declined after the Peloponnesian War with Sparta.
4. Roman cities were also based on conquest and tribute, and declined after the
Germanic raids disrupted the collection system.
5. Feudal cities started as defensive cities, and eventually became market cities,
specializing in commerce and handicrafts.
6. Mercantile cities developed as a result of the centralization of power caused by

military innovations that eliminated the defensive advantages of the feudal
castle. centralization of power caused the development of administrative
The
by breaking down the barriers to trade, increased trade. The
cities and.
development of efficient ocean travel led to exploration and the development
of new markets.
7. The rapid urbanization of the nineteenth and twentieth centuries was caused by
the industrial revolution and the associated innovations in agriculture,
transportation, and manufacturing.
a. Agricultural productivity increased as farmers substituted machines for
laborers, freeing laborers to pursue nonagricultural activities.
Transportation costs decreased, allowing greater regional specialization and
greater productivity.
b. The goods decreased because transportation costs
relative cost of factory
decreased and scale economies in production were realized. Factory
employment increased and cities developed around the factories.
c. Innovations in intracity transportation and construction techniques
increased the feasible size of cities.
8. Between 1800 and 1930. the percentage of the U.S. population in cities
increased from 6 percent to 56 percent. The pace of urbanization has slowed
since World War II. The population in cities rose from 64 percent in 1950 to
75 percent in 1990.
9. In the 1970s, nonmetropolitan areas grew faster than metropolitan areas. In the
1980s, the growth rate for nonmetropolitan areas decreased, restoring the
historical pattern of more rapid growth in metropolitan areas.
10. In the 1970s and 1980s, some northern metropolitan areas decreased in
population.

1 Critically appraise: "Innovation in agriculture during the industrial revolution
was a necessary, but not sufficient, condition for the urbanization of society."
2. One of the reasons for the decline of the feudal system was the development of
would
professional armies. If you were a feudal lord in the fourteenth century,
you have hired a professional army and equipped them with crossbows, pikes,
muskets, and cannons? Would you have been able to defeat the princes and
kings?
3. Some societies never developed any cities. Where did these societies develop?
How did they differ from the societies that developed cities
4. Suppose that you are a citizen of Rome when the first Germanic raid occurs.
You have been asked propose a strategy to respond to the raid (and the
to
expectation that future raids are likely). What's your strategy?
5. The Romans auctioned off the rights to tax their colonies to commercial
tax-collecting agencies. In other words, the Romans sold franchises to tax
collectors.Might this auction scheme have contributed to the decline of Rome?
Ifyou were in charge of the franchising system, what restrictions would you
place on the tax collectors?
References and Additional Reading

Alonso, William. "The Population Factor and Urban Structure." In The Prospective City,
ed. Arthur Solomon. Cambridge, Mass.: MIT Press, 1980. A discussion of recent
trends in urbanization.
Davis, Kingsley. "Urbanization." In The Urban Economy, ed. Harold Hochman. New York:
W. W. Norton, 1976. A brief history of urbanization.
Garnick, Daniel H. "Local Area Economic Growth Patterns: A Comparison of the 1980s
and Previous Decades." In Urban Change and Poverty, ed. Michael G. McGeary and
Lawrence E. Lynn. Washington, D.C.: National Academy Press, 1988. Discusses
growth trends in metropolitan and nonmetropolitan areas from 1959 to 1984, and
examines the growth rates of the 50 largest metropolitan areas (MSAs and CMSAs)
over this period.
Hohenberg, Paul M., and Lynn H. Lees. The Making of Urban Europe 1000-1950.
Cambridge, Mass.: Harvard University Press, 1985. Discusses the reasons for the
urbanization of Europe.
Hoselitz, Bert. "Generative and Parasitic Cities." Economic Development and Cultural
Change 3 (1955), pp. 278-94. Discusses the effects of parasitic cities on their
hinterlands.
Mumford, Lewis. The City in History. New York: Harcourt Brace Jovanovich, 1961. A
lengthy discussion of the history of urbanization.
Olmstead, Alan, and Eugene Smolensky. The Urbanization of the United States.
Morristown, N.J.: General Learning Press, 1973.
Pirenne, Henri. Economic and Social History of Medieval Europe. New York: Harcourt
Brace Jovanovich, 1961 Discusses the role of trade in the rise and decline of medieval
.
cities.
Rosenberg, Nathan, and L. E. Birdzell. How the West Grew Rich. New York: Basic Books,
1986. Discusses the transformation of the Western world from the anarchy and
autarchy of the Dark Ages to its present state.
Sjoberg, Gideon. "The Origin and Evolution of Cities." In Cities, ed. Dennis Flanagan. New
York: Alfred A. Knopf, 1965, pp. 25-39.
Chapter 5
sr OW MANY (HIES?
(^~~7~ his chapter examines urban development from the regional perspective. In
i^s contrast with earlier chapters, which explore the development of individual
cities in isolation, this chapter explains how cities develop as integral parts of a larger
regional economy. A region supports cities of different size and scope, causing the
development of a regional system of cities.
Table 5-1 shows the size distribution of urbanized areas in the United States in
1990. In general, the larger the size of the urbanized area, the fewer the urbanized
areas of that size. There were three urbanized areas with populations exceeding
6.4 million(New York, Los Angeles, and Chicago), but 172 areas with populations
between 50,000 and 00,000. Table 5-1 raises two questions about cities in a regional
1
economy. First, what are the effects of market forces on the equilibrium number of
Table 5-1 Size Distribution of U.S.

Urbanized Areas, 1990
Population of Urbanized Area Number of Areas
More than 12.8 million 1
6.4 million to 12.8 million 2

800,000 to 1.6 million 19
400,000 to 800,000 33
200,000 to 400,000 52
100,000 to 200,000 99
50,000 to 100,000 172
SOURCE: U.S. Bureau of the Census. Press Release,

August 16, 1991.
105
cities? Second, why are some cities larger than others? The answers to these ques-
tions are provided by central place theory.
The chapter is divided into three sections. The first section shows how firms in
a market-oriented industry cane a region into individual market areas, and why
some industries have larger market areas than others. The second section uses central
place theory to explain how merged
the location patterns of different industries are
to form a regional system of cities. The final section explores some possible reasons
for the development of giant cities in developing countries.
The Analysis of Market Areas

Market-area analysis was developed by Christaller (translated in 1966) and refined
by Losch (translated in 1954). A firm's market area is defined as the area over which
the firm can underprice its competitors. The firm's net price is the sum of the price
charged by the store and the travel costs incurred by consumers. To explain the notion
of a market area, consider a region where consumers buy compact discs (CDs) from
music stores. The region has the following characteristics:
1 Common store price. All music stores have the same production
technology and face the same input prices, so they charge the same price
for CDs.
2. Travel costs. Every consumer buys one CD per trip to the music store. The
travel cost (the monetary and time costs of travel) is 50 cents per round-trip
mile.
3. Shape. The region is rectangular. 60 miles long and 20 miles wide.
Because music stores sell the same product at the same price, they differ only in
location.
Pricing with a Monopolist
Suppose music store (owned by Bob) at the center of the region.

that there is a single
The net price per CD sum of
the store price and the consumer's travel cost.
is the
Figure 5-1 shows the net price of CDs for consumers living in different parts of the
region. If Bob charges $8 per CD. the net price rises from $8 for a household living
next to the music store, to $13 for a household living 10 miles from the music store.
to $23 for a household 30 miles from the store.
Figure 5-2 shows how the monopolist chooses the profit-maximizing output.
The monopolist's demand curve is negatively sloped: as the store price increases,
people buy fewer CDs. The negati\el\ sloped demand curve generates a negatively
sloped marginal-revenue curve. Profit is maximized where marginal revenue equals
marginal cost, so the monopolist produces qm units of output. If the monopolist sells
</,„CDs. the market-clearing price is Pm (the point on the demand curve above qm ).
Total profit (revenue less cost) is indicated by the shaded area.
Chapter 5 How Many Cities? 107
Figure 5-1 Net Price of CDs under a Monopoly
Distance from center of region (miles)
The net price of CDs is the sum of the store price ($8) and the cost of traveling to the music store (50 cents
per round-trip mile per CD).
Figure 5-2 Price and Quantity with a Single Music Store
Demand = Average
revenue
Marginal revenue
Quantity of CDs per store
The profit-maximizing output (q m is the quantity at which marginal revenue equals marginal
)
cost. The monopoly price is Pm generating profit equal to the shaded area.
,
Entry and Competition
Since Bob
the monopolist is making positive economic profit, other entrepreneurs
will set up new music stores. Some consumers (those who live near the new music
stores) will patronize the new stores, so Bob's demand curve shifts to the left: at ev-
ery price, he will have fewer customers. As his demand curve and marginal-revenue
curve shift to the left, his sales volume decreases and his profits fall. In other words,
entry causes competition, which decreases sales volume and profit per firm.
How long will entry and the associated decreases in profit continue? Equilib-
rium occurs when two conditions are satisfied. First, even firm is maximizing its
profit, producing the output at which marginal revenue equals marginal cost. Second,
economic profit is zero: the store price equals the average total cost of production.
Figure 5-3 shows the equilibrium situation. At an output marginal revenue equals <:/,..
marginal cost (firms are maximizing profit), and the store price Pe equals the average
production cost (profits are zero). At the equilibrium output, the marginal revenue
curve intersects the marginal-cost curve, and the demand curve is tangent to the
average production-cost curve.
This is the theory of monopolistic competition (covered in most intermedi-
ate microeconomics textbooks). Each music store is a monopolist within its own
Figure 5-3 Equilibrium Price and Quantity of CDs with Entry
Average
production cost
Entr) shifts the demand curve of the mdiv idiul -tore decreasing price and profit. Entrj
to the left,
continues until profit is zero. The equilibrium output is ./because marginal revenue = marginal cosl
iimns are maximizing protui and profit = average production cost iproiit equals aero)
( 'hapter 5 How Manx (. 'ities ? 109
territory, but its monopoly power is limited by entry and competition. Each store
faces a negatively sloped demand curve, and entry shifts the demand curve to the
point at which profits are zero. Like a perfectly competitive firm, a firm with a local
monopoly earns zero economic profit.
Efficiency Trade-Offs
There are some efficiency trade-offs associated with entry and competition. On the
one hand, entry decreases output per firm, so individual firms move upward along
their average cost curves. Scale economies are lost because output is divided among a
largernumber of firms, each of which produces at a higher average cost. On the other
hand, the increase in the number of stores decreases travel distances for consumers,
decreasing travel costs.
Figure 5—4 shows the trade-offs between production and travel costs. The average
total costof CDs is defined as the sum of average production cost (from Figure 5-3)
and average travel cost. Average travel cost is the travel cost incurred by the typical
CD consumer. As the output of the music store increases, the market area increases
and the typical consumer travels a longer distance to buy CDs. Therefore, the average
travel cost increases with the output of the store. Average total cost (production cost
Figure 5-4 Trade-Offs with Entry: Production Cost versus Travel Cost
Average
production cost
Entry increases the number of firms and decreases output per firm. As output falls, average production cost
increases and average travel cost decreases. In this case, entry goes "too far" in the sense that average total
cost is higher than the feasible minimum (C, ).
plus travel cost reaches ) its minimum point at q, far below the output that minimizes
.
average production cost (q c ).

Entry decreases output per firm, generating bad news and good news. The
bad news is that average production cost increases as firms move up the average
production-cost curve. The good news is that average travel costs decrease as firms
move down the average travel-cost curve. Starting from q m in Figure 5—1. the good
news dominates the bad news until output per firm reaches q,. From that point on.
further entry increases production costs by more than it decreases travel costs, so
average total cost increases. In equilibrium, the output per firm drops to q meaning t
.
that the entry process goes "too far" in the sense that average total cost is higher than
the feasible minimum.
Figure 5—4 shows only one possible equilibrium outcome. For different sets of
cost and demand curves, the equilibrium output per firm might be greater than q,
or equal to q,. In other words, unless one knows something about the demand and
cost curves. it"s impossible to predict exactly where on the average total-cost curve
the firm ends up. How ever, we do know that ( 1 ) entry decreases output per firm and
(2) in equilibrium, the firm does not produce at the minimum point of its average
production-cost curve, but at some point to the left of the minimum point.
Market Areas
Figure 5-5 shows the market areas for music stores under the assumption that there
are three stores in equilibrium. Tammy and Dick set up music stores 20 miles
from Bob and charge the same price for CDs ($6). Compared to Bob. Tammy
has a lower net price in the western third of the region, and Dick has a lower net
price in the eastern third of the region. The three music stores split the region into
three equal market areas, so every store has a circular market area with a 10-mile
radius.
The market arrangement in Figure 5-5 has two implications for CD consumers.
First, if the spaces between the circular market areas are ignored, the maximum net
price is SI 1. the sum of music stores (S6) and the maximum
the price charged by
travel cost ($5). Second, every household patronizes the music store closest to its
home. All stores charge the same price for the same product, so a household will
patronize the store with the lowest travel costs.
Determinants of the Market Area: Fixed Demand

What factors determine the market areas oi market-oriented firms? The size of the
market area depends on trav el costs, per capita demand, population density, and scale
economies. This section uses a simple algebraic model to identify the determinants
of market areas. At first we assume that the demand for CDs is perfectly inelastic:
later we drop this assumption.
An Algebraic Model of the Determinants of Market Area. An expression for

the si/e oi the market area can be derived with some simple algebra. The follow ing
Chapter 5 How Many Cities '.' Ill
Figure 5-5 Equilibrium Market Areas

Bob's net price
Dick's net price
Each store's market area is the area over which its net price is less than the net prices of other stores. Each store has a
circular territory with a 10-mile radius.
symbols represent the market for CDs in a region:
d= Monthly per capita demand (number of CDs)

e= Population density (people per square mile)
A = Land area of the region (square miles)
q = Output of the typical music store (CDs sold per month)
To simplify matters, per capita demand (d) is assumed to be independent of the net
price of CDs: the individual consumer demands a fixed number of CDs, regardless
of the net price. In other words, the individual demand curve is perfectly inelastic.
What is the equilibrium number of music stores? The total demand for CDs
equals the product of per capita demand (d), population density (e), and the land
Table 5-2 Numerical Example of Number of Stores

and Market Area
Variable Symbol Numerical Example
Per capita demand (number of CDs) d 4

Population density (people per square mile) e 50
Demand density (CDs per square mile) de 200
Land area (square miles) A 300
Output per store (CDs sold per month) q 1.000
demand
Total Q=de- A 60,000
Number of stores N= Q/q 60
Market area (square miles) M = q/Ul e) 5.0
area of the region (A):
Q= deA (5-1)
Table 5-2 provides a numerical example that shows how to compute the equilibrium
number of stores. Per capita demand four CDs and population density is 50 people
is
per square mile, so demand density (CD demand per square mile, equal to d times
e) is 200. The region has an area of 300 square miles, so total demand is 60.000 CDs
(200 times 300). The number of music stores in the region is total demand divided
by the output per firm:
N=— (5-2)
Output per firm is 1 .000 CDs. so there will be 60 stores in the region (60.000 divided
by 1.000).
How large is the market area of music stores? The size of the market area equals
the land area of the region divided by the number of firms:
4
M=— (5-s)
N
If the region has a land area of 300 square miles and there are 60 stores, the market
area of the typical store is 5 square miles. Using the expressions for N and Q. the
expression for M can be rewritten as
M= -?— (5-1)
d C
In other words, the size of the market area is determined by output per store (q). per
capita demand id), and population density (e). In Table 5-2. q = 1.000. d = 4. and
e = 50, so the market area is 5.0 square miles.
The algebraic model identifies three factors that determine the size of the firm's
market area: per capita demand, population density, and output per store. Output per
store is affected by both scale economies and transportation costs. How do changes
in each of these variables affect market areas'.'
Figure 5-6 Increase in Scale Economies Increases Output per Firm
Original average
total cost
Original average
production cost
New average
production cost
An increase in scale economies shifts the average production-cost curve downward, shifting the average total-cost
curve downward. The minimum points of both curves move to the right. If the music store produces at the mini-
mum point of its average total-cost curve, output per store increases from q to q 2
t
.
Market Area and Scale Economies. Figure 5-6 shows the effects of an increase
in scale economies on the firm's cost curves. An increase in scale economies means
that the average production-cost curve is negatively sloped over a larger range of
output. Average production cost is minimized at q$ instead of q^. Since average total
cost equals average production cost plus average travel cost, an increase in scale
economies also shifts the average total-cost curve. Average total cost is minimized
at qi instead of q\.
The shift of the cost curves increases output per firm. Although it's impossible
to predict exactly how much the firm will produce, it's safe to say its output will
increase. One possibility is the firm starts at q\ (the minimum point of the original
average total-cost curve) and moves to qj (the new minimum point). If the firm
starts with an output close to q\, it is likely to switch to an output close to <?2-
In general, an increase in scale economies increases the advantages of large-scale

operations. Starting from q\, the benefit of increasing output (savings in produc-
tion costs) will exceed the costs (increase in travel costs), so output per firm will
increase.
An increase in scale economies increases the market area of the typical music
store. As output per store increases, fewer stores are required to serve the region.
Using the numbers in Table 5-2, if output per store increases from 1,000 to 1,200,
the number of stores will decrease from 60 to 50 (60.000/1,200). The decrease in

the number of stores increases the market area from 5 square miles to 6 square miles
(300 divided by 50). Because an increase economies increases output per
in scale
store, every store needs a larger market area to exploit its scale economies. Note that
this result depends on the assumption that per capita demand is fixed, an assumption
to be relaxed later in the chapter.
Market Area and Travel Costs. How does a decrease in travel cost affect the
output per store? Suppose the development of a faster, cheaper travel mode decreases
the monetary and time costs of travel. In Figure 5-7. the gap between the average
total cost and average production is cut in half because travel costs fall by one-half.
The minimum point of the average total-cost curve moves to the right, increasing
the cost-minimizing output from q$ to q b . A decrease in travel costs decreases the
relative importance of travel costs, so the minimum point of the average total-cost
curve moves closer to the minimum point of the average production-cost curve.
The shift of the average total-cost curve increases output per firm. Again, it's
impossible to predict exactly how much the firm will produce, but it's safe to say
that output increases. If the firm starts with an output close to q$ (the minimum point
Figure 5-7 Decrease in Travel Costs Increases Output per Firm
Original average
total cost
Average
production cost
A decrease in travel costs shifts the average travel-cosl curve downward, -.lulling the average total-cost curve
downward. The minimum point on the VTC curve nunc-, to the right It the music store produces at the mini-
mum point oi us ATC curve, output per store increases from q^ to i/,.
ClwpierS How Many Cities'.' 115
on the original average total-cost curve), it is likely to switch to an output close to
<76- In general, a decrease in transport costs increases the advantages of large-scale

operations. Starting from <r/
5 , the cost of increasing output will be less than the benefit,
so output per firm will increase.
The decrease in travel costs increases the market area because every store needs
a larger territory to sell its output. Using the numbers in Table 5-2, if output per
store increases from 1,000 to 1,250, the number of stores will decrease from 60 to
48 (60,000/1,250). The decrease in the number of stores increases the market area
per store from 5 square miles to 6.25 square miles (300 divided by 48). Because a
decrease in travel cost increases output per store, every store needs a larger market
area to exploit its scale economies. Again, this result depends on the assumption that
per capita demand is fixed, an assumption to be relaxed later in the chapter.
Market Area and Per Capita Demand. How does an increase in per capita
demand affect the size of the market area? An increase in per capita demand increases
the demand density (the number of CDs sold per square mile, or d times e). If
output per store is fixed, each store needs a smaller market area to exhaust its scale
economies. For example, a doubling of per capita demand doubles total demand,
so it doubles the number of stores. If there are twice as many stores in the region,
the market area per store will be cut in half (from 5 square miles to 2.5 square
miles).
Market Area and Population Density. How does an increase in population den-
sity affect the market area? Like an increase in per capita demand, it increases
demand density. If output per store is fixed, each store needs a smaller market area
to exhaust its scale economies. For example, a quadrupling of population density
quadruples total demand for CDs, so it quadruples the number of stores. If there
are four times as many stores in the region, the market area per store decreases to
one-fourth its original size (from 5 square miles to 1.25 square miles).
Market Area and Income. How do differences in income affect the size of market
areas? Consider a region with the following characteristics:
1 There are two cities, a poor one and a wealthy one.

2. A CD is a "normal" good (positive income elasticity of demand).
3. Land is a normal good (positive income elasticity).
The city with the larger demand

density will have the smaller market area for
music stores. Since both CDs
and land are normal goods, the wealthy city has higher
per capita demands for both CDs and land: its residents consume more CDs per
capita, but also live in relatively low-density housing. For example, the wealthy
may live in single-family homes and the poor may live in apartment buildings. In the
wealthy city, demand for CDs (which increases demand density)
the larger per capita
is by a lower population density (which decreases demand density),
offset in part
so demand density may be either higher or lower. Therefore, the wealthy city may
have a smaller or larger market area. The relationship between income and market
area is ambiguous because income affects both per capita demand and population
density.
Under what circumstances will the wealthy community have a larger market
area? Suppose that the income elasticity of demand for land is large relative to the
income elasticity of demand for CDs. If so. the density effect (increased income
decreases population density) will dominate the demand effect (increased income
increases per capita demand), and the wealthy city will have a lower demand density
and a larger market area. Music stores will need a larger market area because although
each household consumes more CDs. there are many fewer households per square
mile.
Table 5-3 shows a simple example of how differences in income affect market
areas. The income elasticity of demand for where CDs is 0.50. In the wealthy city,
per capita income 20 percent higher, the per capita demand for CDs is 10 percent
is
higher. In the middle column, the income elasticity of demand for land is assumed
to be 1.0. Therefore, the wealthy city has a 20 percent higher demand for land and
a 20 percent lower population density (40 instead of 50). Given these assumptions,
the wealthy city has a lower demand density, so music stores need a larger mar-
ket area to exploit their scale economies (5.68 square miles instead of 5 square
miles).
This result depends, of course, on the numerical assumptions. If the demand
for land were relatively income-inelastic, the difference in population density would
be relatively small. The demand effect would dominate the density effect, so the
wealthy city would have a smaller market area. In the third column of Table 5-3. the
income elasticity of demand for land is assumed to be 0.25: population density in
the wealthy city is 47.5 (a 5 percent difference) instead of 40; 209 demand density is
instead of 176: and the market area is 4.78 instead of 5.68. In general, if the income
elasticity for land is small relative to the income elasticity for CDs. the wealthier
city will have a smaller market area.
Table 5-3 Income and Market Area with Different Income

Elasticities of Demand for Land
Wealthy City
High Income Low Income

Poor City Elasticity Elasticity
Income elasticity for land 1.0 1.0 0.25

Per capita income $1,000 $1,200 $1,200
Per capita demand tor CDs l</ ) 4 4.4 4.4
Population density (e) 50 40 47.5
Demand densit\ (d • e) 200 [76 209
Output per store U/l 1 .000 1.000 1.000
Market area in square miles (qfd e) 5 5.68 4.7S
Market Areas and the Law of Demand

Up to this point, the discussion of the market area has assumed that per capita
demand is fixed. In other words, the individual demand curve is perfectly inelastic.
What happens if consumers obey the law of demand, increasing the quantity of CDs
demanded as the net price decreases?
Travel Costs. To explain the effect of the law of demand on market areas, suppose
that travel costs decrease. As explained earlier in the chapter, a decrease in travel
costs increases the output per store because the minimum point on the average total-
cost curve moves to the right. The market area increases because each store needs a
larger territory to sell its output. This is the output effect of a decrease in travel costs.
The decrease in travel costs also decreases the net price of CDs: in Figure 5-7, the
net cost decreases from C5 to C^. If consumers obey the law of demand, per capita
demand will increase, so each firm will need a smaller territory to sell a given output.
This is the demand effect of a decrease in travel costs.
The net effect of a decrease in travel cost depends on the relative strengths of
the output effect and the demand effect. The expression for the size of the market
area is
M= -?— (5-5)
a e
Suppose that the output effect increases q from 1 ,000 to 1 ,500 and population density
(e) is 50. If the demand effect is relatively weak and d increases from 4 to 5, the
market area will grow from 5 square miles ( 1 ,000/200) to 6 square miles ( 1 ,500/250).
Alternatively, if the demand effect is relatively strong and d increases from 4 to 10,
the market area will shrink from 5 square miles to 3 square miles (1,500/500). The
strength of the demand effect depends on the price elasticity of demand for CDs:
the more elastic the demand for CDs, the stronger the demand effect. Therefore, a
decrease in travel cost is more likely to decrease the market area if the price elasticity
is large.
For an example of the effects of changes
in travel costs on market areas, consider
the effects of parcel post on farm communities. Before the introduction of parcel
post in 1913, most farmers purchased most of their goods, including clothing and
tools, in general stores in small farm communities. The introduction of parcel post
decreased the cost of shipping goods from big-city merchants to farmers. Mail-order
houses (Sears Roebuck, Montgomery Ward) underpriced the local general store, and
farmers started buying clothing and tools from the mail-order houses. In the year
following the introduction of parcel post, the sales of Sears and Montgomery Ward
quintupled, and many general stores disappeared.
The introduction of
parcel post decreased the cost of transporting clothes and
market area of the typical firm. The general store, with its small
tools, increasing the
market area, was replaced by Sears and Ward, with their large market areas. In this
case, the output effectwas stronger than the demand effect, so the market area grew.
Although farmers consumed more trousers and shovels as the net prices fell, the
increases in per capita demands were not large enough to offset the output effect.
Scale Economies. If consumers obey the law of demand, an increase in scale
economies has an ambiguous effect on the size of the market area. As explained
earlier in the chapter, an increase in scale economies increases the output per store,
increasing the market area required to exhaust scale economies (the output effect). If
the net price falls as a result of the increased scale economies, the per capita demand
for CDs will increase, generating a demand effect that provides downward pressure
on the market area: the firm needs a smaller territory to sell a given output. The
economies depends on the relative strengths of the
net effect of an increase in scale
output effect and the demand effect. If the demand for the good is relatively inelastic,
an increase in scale economies will increase the market area.
The Demise of Small Stores

The demise of the small store illustrates the effects of changes in population density
and transportation costs on market areas. Before the late 1800s. most goods in the
United States were purchased in small stores. People bought food in small "mom
and pop" grocery stores and purchased other goods in small general stores. The
traditional merchant was involved in every transaction, playing two roles:
1 Huckster. The merchant assisted customers in their choice of goods,

providing information on alternative products and persuading reluctant
customers to buy the good. The merchant was an effective huckster because
he knew the features of every good in the store and was a trusted source of
information about his products.
2. Haggler. The merchant negotiated with each customer over the price of the
good. The merchant was an effective haggler because she knew the cost of
every good in her store, and could quickly compute her profit margin as she
negotiated with the customer. In addition, the merchant was skillful in
assessing her customers' willingness to pay for various goods.
Since huckster and haggler skills were not easily transferred to low-skilled employ-
ees, the merchant was involved inevery transaction, and the traditional store was
small.
The replacement of the small general store with larger stores suggests that
the optimum store size increased. The optimum size increased for a number of
reasons, two of which are related to transportation costs. First, in the late 800s 1
and early 1900s. innovations in urban transit increased the speed and decreased
the monetary costs o( intracity travel. Second, rapid urbanization during this same
period increased population density, decreasing average travel distances (from home
to store). As travel costs decreased, merchants expanded to exploit scale economies
in marketing.
As merchants developed new marketing techniques.
larger stores, they adopted
New advertising and display techniques goods "sell themselves." Some
made the
stores, such as Macy's and Marshall Field's, used newspaper advertising to inform
and persuade consumers. Woolworth's used plate-glass windows to display goods
Chapter 5 Haw Manx Cities'.' 119
and attract customers. Grocery stores displayed their goods in such a way that con-
sumers could do their own comparison shopping. A grocery store named Piggly
Wiggly developed a floor plan that forced customers to walk through a maze from
the entrance to the exit. By following the prescribed path from the entrance to the
exit, customers were exposed to every commodity in the store, allowing them to
make their own choices. These marketing techniques freed the merchant from act-
ing as a huckster. Another new feature was the single-price policy, under which the
merchant charged each customer the same price. The single-price policy freed the
merchant from haggling.
These two new marketing techniques freed the merchant from being involved
in every transaction, allowing an increase in the size of the store. The merchant
hired low-skilled laborers to perform the simple tasks of stocking shelves, wrapping
goods, and collecting money. A single merchant could run a self-service store with
dozens of low-wage employees, and this store could underprice the small traditional
store.
Market Areas of Different Industries

Market areas vary from industry to industry, reflecting differences in travel costs,
per capita demand, and scale economies. If scale economies are large relative to per
capita demand, the industry will have a small number of firms, each of which has
a large market area. In contrast, if scale economies are small relative to per capita
demand, there will be a large number of firms with small market areas.
Consider the market areas of pizza parlors and Tibetan restaurants. Suppose that
both activities are subject to the same sort of scale economies, so the optimum output
for both types of restaurants is 200 meals per day. If the total demand for pizza is
10,000 meals per day, the region will have 50 pizza parlors (10,000/200). If the total
demand for Tibetan food is 200 meals per day, a single Tibetan restaurant will serve
the entire region. The market area is not determined by scale economies per se, but
by scale economies relative to per capita demand.
Central Place Theory

Central place theory, which was developed by Christaller (translated in 1966) and
refined by Losch (translated in 1 954), is used to predict the number, size, and scope of
cities in a region. The theory
based on a simple extension of market-area analysis.
is
Market areas vary from industry to industry, depending on scale economies and
per capita demand, so every industry has a different location pattern. Central place
theory shows how the location patterns of different industries are merged to form a
regional system of cities. The theory answers two questions about cities in a regional
economy:
1 How many cities will develop?

2. Why are some cities larger than others?
A Simple Central Place Model
Consider a region with three consumer products: CDs, pizzas, and jewelry. The
region has the following characteristics:
1. Population density. The initial distribution of population is uniform. The

total population of the region is 80,000.
2. No shopping externalities. As discussed in Chapter 2. shopping exter-
nalities normally occur with complementary goods (one-stop shopping)
and imperfect substitutes (comparison shopping). The simple central place
model assumes that there are no shopping externalities.
3. Ubiquitous inputs. All inputs are available at all locations at the same
prices.
4. Uniform demand. For each product, per capita demand is the same
throughout the region.
5. Number of stores. The three goods have different per capita demands and
scale economies:
a. Jewelry. Scale economies are large relative to per capita demand. Every
jewelry store requires a population of 80,000. so a single jeweler will
serve the entire region.
b. Compact discs. Scale economies are moderate relative to per capita
demand. Every music store requires a population of 20,000. so there will
be four music stores in the region.
c. Pizza. Scale economies are small relative to per capita demand. Every
pizza parlor requires a population of 5,000. so there will be 16 pizza
parlors in the region.
The central place model is a model of market-oriented firms, defined in Chapter 3

as firms that base their location decisions exclusively on access to their consumers.
Because all inputs are ubiquitous, the firms ignore input costs in their location
decisions.
The single jeweler will locate at the center of the region. Because production
same at all locations (all inputs are ubiquitous the jeweler will minimize
costs are the ).
its by minimizing its travel costs. According to the principle of median

total costs
location (discussed in Chapter 3), travel costs are minimized at the median location.
Because population density is uniform, the median location is the center of the
region. Therefore, the jeweler will locate at the center of the region.
A city will develop around the jewelry store. Jewelry workers will locate near
the store to economize on commuting costs. The population density near the jeweler
will increase, generating a city (a place of relatively high density) at the center of
the region. In Figure 5-8, a city develops at point L.
The music stores will carve up the region into market areas, causing the devel-
opment of additional cities. If the region's population density were uniform, music
firms would carve out four equal market areas. However, because there is a city
Chapter 5 How Many Cities'.' 121
Fie; IRE 5-8 The Central Place Hierarchy
© © f
©
r >>
f
/I L I
J J
©
CityL M Cities S Cities
1 Jeweler 1 Music store 1 Pizza parlor
2 Music stores 2 Pizza parlors
4 Pizza parlors
There are 1 1 cities in the region: one large city (L), two medium-sized cities (M). and eight small
cities (S). The larger the city, the greater the variety of goods sold.
surrounding the jeweler in the center of the region, there will be enough demand to
support more than one music store in city L. If city L along with the surrounding area
has enough people to support two music stores, the two other music stores will split
the rest of the region into two market areas. In Figure 5-8, two more cities develop
at the locations marked with an M.
The pizza up the region into market areas, causing the
parlors will also carve
development of more cities. Because the population density is higher in the cities
that develop around the jewelry store and the music stores, there will be more than
one pizza parlor in L and the two M
cities. Suppose that L will support four pizza
parlors, and each of the M

cities will support two pizza parlors. If so, a total of eight
pizza parlors will locate in cities L and M. The remaining eight pizza parlors will
divide the rest of the region into eight market areas, causing the development of
eight additional cities (the places marked with an S in Figure 5-8).
The rectangular region has a total of 1 cities. The large city at the center of the
1
region sells jewelry, CDs, and pizza. City L has a population of 20,000, meaning that
it is large enough to support four pizza parlors (5,000 people per pizza parlor). The
city sells CDs to consumers from the four surrounding S cities, so the total number of
CD consumers is 40,000 (20,000 from L and 5,000 each from four S cities), enough
to support two music stores. The two medium-sized cities sell CDs and pizza. Each
of the M
cities has a population of 10,000, meaning that each city is large enough
to support two pizza parlors. Each city sells CDs to consumers from two nearby
Scities, so the total number of CD consumers in each M city 20,000 (10,000 from is
M and 5,000 each from two S cities), enough to support one music store per M city.
The eight small cities sell only pizza. Each of the S cities has a population of 5,000,
meaning that each city can support one pizza parlor.
Figure 5-9 shows the size distribution of cities in the region. The vertical axis
measures and the horizontal axis measures the rank of the city.
city size (population),
The largest city (L) has a population of 20,000; the second and third largest cities
(M cities) have populations of 10,000; and the fourth through the eleventh largest
cities have populations of 5.000.
The simple model generates a hierarchical system of cities. There
central place
M
L (high order), (medium order), and S (low order).
are three distinct types of cities:
The larger the city, the greater the variety of goods sold. Each city imports goods
from higher-order cities and exports goods to lower-order cities. Cities of the same
order do not interact. For example, an M
city imports jewelry from L and exports
CDs to S cities, but does not interact with the other M city. Similarly, an S city
imports jewelry from L and CDs from either L or an M city, but does not trade with
other S cities. The system of cities is hierarchical in the sense that there are distinct
types of cities and distinct patterns of trade dominance.
Figure 5-9 Size Distribution of Cities with Simple Central Place

Model
20,000
g 10.000 -
5.000
I I
4 5 8 9 10 11
Rank
The simple central place model generates one large cu> (L) with a population of 20,000, two
medium-sized cities (Mi. each of which has a population of 10.000. and eight small cities (S), each of
which has a population of 5.000.
Three lessons can be learned from the simple central place model:
1 Diversity and scale economies. The region's cities differ in size and scope.
This diversity occurs because the three consumer products have different
scale economies relative to per capita demand, so they have different market
areas. To explain the importance of differences in relative scale economies,
suppose goods have the same scale economies relative to per
that the three
capita demand, so the region has 16 jewelers, 16 music stores, and 16 pizza
parlors. The market areas of the three goods would coincide, so the region
would have 16 identical cities, each of which provides all three goods. In
other words, if there are no differences in scale economies relative to per
capita demand, the region's cities will be identical.
2. Large means few. The region has a small number of large cities and a large
number of small cities. Why isn't there a large number of large cities and a
small number of small cities? A city is relatively large if it provides more
goods than a smaller city. The extra goods provided by a large city are those
goods economies. Since there are
that are subject to relatively large scale
relatively few stores selling the goods subject to relatively large scale
economies, few cities can be large. In the simple central place model, L is
larger than an M
city because L sells CDs, pizza, and jewelry. Since there is
only one jewelry store in the region, there is only one city larger than the M
cities.
3. Shopping paths. Consumers travel to bigger cities, not to smaller cities or

cities of the same size. For example, consumers from an M city travel to L
to buy jewelry, but do not travel to the other M city or an S city to consume
CDs or pizza. Instead, they buy CDs and pizza in their own city. Similarly.
consumers in S cities travel to larger cities for jewelry and CDs, but do not
shop in other S cities.
Relaxing the Assumptions
Several of the assumptions of the simple central place model are unrealistic. This
section relaxes some of the assumptions, addressing two questions. First, does a
more realistic model generate the same hierarchical pattern of cities? Second, does
a more realistic model have more or fewer cities?
Scale Economies, Market Areas, and the Number of Firms. Suppose that the
equilibrium number of pizza parlors decreases from 16 to 8. As explained earlier
in the chapter, this would occur if scale economies increased relative to per capita
demand, increasing the market area of pizza parlors. The decrease in the number of
pizza parlors would decrease the equilibrium number of cities in the region: there
would be two pizza parlors in city L and one parlor in each of the M cities, leaving four
pizza parlors for the S cities. The equilibrium number of cities would decrease from
1 1to 7 because there are fewer cities that provide only pizzas. The hierarchical pattern
of cities persists: each city still provides a subset of the goods sold in larger cities.
Shopping Externalities: Imperfect Substitutes. Suppose that consumers buy

used CDs instead of new CDs. Used CDs are imperfect substitutes: consumers
engage in comparison shopping, comparing the used CDs from several stores before
buying. Therefore, music stores will cluster to facilitate comparison shopping. If the
optimum cluster is four music stores, the stores will cluster at the center of the region.
The region will have only two types of cities, large ones (with jewelers, music stores,
and pizza parlors) and small ones (with pizza parlors). The presence of imperfect
substitutes reduces the equilibrium number of cities from 1 to 9 because music 1
stores compromise on their "ideal" (central place) locations to exploit the shopping
externalities. Although comparison shopping decreases the number of cities, it does
not disrupt the hierarchical pattern of cities.
Shopping Externalities: Complements. Suppose that pizza and CDs are comple-
mentary goods, so that the typical consumer purchases a CD and a pizza on the same
shopping trip. If so, music stores and pizza parlors will pair up to facilitate one-stop
shopping. If pizza parlors cannot survive without a companion music store, there
will be only two types of cities in the region: large (with jewelry, CDs, and pizza)
and medium (with CDs and pizza). Pizza parlors compromise on their ideal (central
place) locations to exploit the shopping externalities associated with one-stop shop-
ping, so the equilibrium number of cities decreases from 1 1 to 3. The presence of
complementary goods does not, however, disrupt the urban hierarchy.
Variation inDemand. One of the assumptions of the simple central place model
is demand does not vary with city size. Systematic variation in per
that per capita
capita demand may disrupt the urban hierarchy.
Consider first the possibility that the per capita demand for pizza decreases
as city size increases. In other words, per capita demand is large in small cities,
moderate medium-sized cities, and zero in large cities. Alternatively, pizza could
in
be replaced in the example with grits, a good for which per capita demand is higher
in small cities. Since there will be no pizza parlors (grits restaurants) in the large
city, the urban hierarchy will be disrupted: the medium-sized and the small cities
supply goods that are not available in the large city. The variation in demand may
also increase the equilibrium number of cities: if the region has the same total pizza
demand, all 16 pizza parlors will be outside the largest city (instead of 12). so there
will be more than 8 small (pizza-only) cities.
What if the per capita demand increases with city size? Pizza parlors will be
more concentrated in medium and large cities: there may be 6 pizza parlors in L
and 3 in each type-M city, leaving only 4 of the 16 pizza parlors for small cities.
Therefore, there will be a total of 7 cities instead of 11. This type of variation in
demand does not, however, disrupt the hierarchy of cities.
Other Types of Industry

Central place theory is applicable to market-oriented firms. The objective of a market-
oriented firm minimize the travel costs of its consumers. The market-oriented
is to
firm is not concerned about (1) the cost of transporting its inputs or (2) the costs of
Chapters How Many Cities? 125
local inputs. Central place theory predicts the pattern of cities that would result if
all firms were market-oriented.
As explained in Chapter 3, there are two other types of firms, resource-oriented
firms and input-oriented firms. For the resource-oriented firm (e.g., the producer of
baseball bats), the cost of transporting raw materials is relatively high, so the firm
locates near its input sources. For the input-oriented firm, the costs of local inputs
vary across space, and the firm locates near sources of inexpensive labor, energy, or
intermediate goods.
The location decisions of resource-oriented firms may disrupt the central place
hierarchy. Suppose that a bat factory is located near the coastal forests of the region.
Once the coastal city is established it may attract some market-oriented firms (e.g.,
pizza parlors). In addition to the three types of cities described above (large, medium,
and small) there will be a coastal city, with a bat factory and pizza parlors. The urban
hierarchy is disrupted because the coastal city has one activity (bat production) not
available in the largest city.
It is possible that the bat factory will not disrupt the region's urban hierarchy.
Suppose that the coastal city becomes large enough that it becomes the median
location for the jeweler. If so, the coastal city will have all four types of activities
(jewelry, bats, CDs, and pizza). The medium-sized cities will have two activities
(pizza and CDs), and the small cities will have only one activity (pizzas). In this
case, the hierarchy is preserved because the coastal city takes the place of city L.
The same arguments apply to input-oriented activities such as textile firms
(labor-oriented), aluminum producers (energy-oriented), corporate headquarters
(oriented to intermediate inputs), and firms engaging in research and development
(oriented to amenities valued by engineers and scientists). If the input-oriented firm
locates near its input source, the resulting city will have some goods that are not avail-
able in the largest city. If, however, the resulting city becomes the median location
in the region, it may replace city L as the largest city in the region.
Central Place Theory and the Real World
While central place theory is not literally true for many regions, it provides a useful
way of thinking about a regional system of cities. The theory identifies the market
forces that generate a hierarchical system of cities. It explains why some cities are
larger than others, and why the set of goods sold in a particular city is typically a
subset of the goods sold in a larger city. Exceptions to the hierarchical pattern result
from ( 1 ) systematic variation in per capita demand and (2) the location decisions of
firms oriented toward raw materials and local inputs.
Empirical Studies of Central Place Systems. Table 5^ shows the results of an

empirical study of central place theory. Berry and Garrison (1958) examined the
economic activities of urban places in Snohomish County, Washington. They con-
cluded that the dozens of communities could be divided into three distinct types of
urban places: town (high order), village (medium order), and hamlet (low order). The
largest urban place, a town, was about twice as large as a village, which was in turn
about twice as large as a hamlet. There were about half as many towns as villages,
126 Part I Market Farces in the Development of Cities
and about half as many villages as hamlets. As one moves down the hierarchy, the
number of establishments and the number of functions decrease, which is exactly
what is predicted by central place theory. Berry and Garrison concluded that the sys-
tem of urban places was indeed hierarchical, with individual communities exporting
to lower-order places and importing from higher-order places.
Table 5-4 The Urban Hierarchy in Snohomish County,

Washington
Type of Place
Town Village Hamlet
Number of places 4 9 20
Average population 2,433 948 417
Average number of establishments per place 149 54.4 6.9
Average number of functions per place 59.8 32.1 5.9
Average number of establishments per function 2.5 1.7 1.2
SOURCE: Brian J. L. Bern and William Garrison. "The Functional Bases of the Central
Place Hierarchy," Economic Geography 34 ( ls>58). pp. 145-54.
Figure 5-10A Size Distribution of 25 Largest U.S. Urbanized Areas
17 -r—
16
15 -f
14
13 -
12 --
y
g 11 -
l0
I
i 9 h
7 —
6 —
5 —
4 —
3
—
2 —
-
1
ID 15
n
Rank of Urbanized area
Most of the empirical studies of central place theory examine systems of small
towns in agricultural regions. Such regions have little most firms
industrial activity, so
are oriented toward the local market, not toward natural resources or local inputs.
What about the national economy? As explained earlier, the central place hierarchy
is disrupted by the location choices of resource-oriented firms and input-oriented
firms. Since a large fraction of national employment comes from resource-oriented
and input-oriented activities, it would be quite surprising if a study of the national
economy provided much support for the simple central place model.
Size Distribution of Cities: The Rank-Size Rule. Figure 5-10A shows the size
distribution of the 25 largest urbanized areas in the United States in 1990. Population
decreases from about 16 million in the largest U.S. urbanized area (New York) to
1.32 million in the 25th largest (Norfolk, Virginia). Figure 5-10B shows the size
distribution for the 26th through the 396th largest U.S. urbanized areas. (Note that
the vertical scale in Figure 5-1 OB is different from the scale in Figure 5-10A.)
Population decreases from 1 .27 million in the 26th largest metropolitan area (Kansas
City) to 50,066 in the smallest urbanized area (Brunswick, Georgia).
Figure 5-10B Size Distribution of 26th through 396th Largest U.S.

Urbanized Areas
100 200 300 396

Rank of urbanized area
How does the U.S. size distribution of cities compare to the size distribution
generated by the simple model of central place theory? To compare the two size
distributions, compare Figure 5-9 to Figure 5-10. Both figures show a small number
of large cities and a large number of small cities. In addition, as one moves down
in the ranking (to smaller cities), the size difference between two successive cities
decreases.
Geographers and economists have estimated the relationship between city size
and rank. The relationship is approximated by the rank-size rule:
Rank • Size = Constant (5-6)
This rule suggests that the product of rank and size is the same in all cities in a
region. For example, if the largest metropolitan area has a population of 16 million,
the rank-size rule suggests that the second largest city has a population of 8 million,
the third largest city has a population of 5.33 million, and the sixteenth largest city
has a population of 1 million. Figure 5-1 1 shows the rank-size relationship for U.S.
urbanized areas in 1990. The rank of the urbanized area is on the horizontal axis and
the product of rank and size (in millions) is on the vertical axis. The largest deviation
Figure 5-11 Rank-Size Relationship for U.S. Urbanized

Areas
100 200 300 396

Rank of urbanized area
from the rank-size rule occurs in cities ranked 10th through 80th; these cities are
much larger than would be predicted by the rank-size rule. In general, it does not
appear that the size distribution of U.S. urbanized areas satisfies the rank-size rule.
Empirical evidence concerning the rank-size rule comes from statistical studies
of city-size distributions. The hypothesized relationship between rank and size is
Constant
Size = =- (5-7)
Rank 8
The rank-size rule good approximation of the relationship between rank and
is a
size if B (the exponent of rank)
is close to .0. Rosen and Resnick ( 980) estimated
1 1
the rank-size relationship for 44 countries. They found substantial variation in B

across countries, with an average value of 0.88. This result suggests that population
is more evenly distributed between cities than would be predicted by the rank-size
rule. Rosen and Resnick conclude that the rank-size rule is only a first approximation
to the relationship between rank and size.
Urban Giants: The Puzzle of the Large Primary City

In many developing countries, the largest city has a relatively large share of the
nation's population. Table 5-5 shows and the national shares of
the populations
population for the largest cities in several countries. The New York metropolitan
area has only 6.5 percent of the population of the United States. In contrast, the
other metropolitan areas listed in the table have between 1 1 percent and 39 percent
of their national populations.
The Role of Trade

Economists have developed several models to explain the large concentrations
of population in primary cities. The simplest theory is based on the observation
that there are large economies of scale in trade activities, defined as the activities
Table 5-5 Population of Primary Cities as Share

of National Population
Share of Country's
City Population Population
Tokyo, Japan 19,037,361 15.76%

Mexico City. Mexico 16,465,487 20.97%
New York, USA 15.627,553 6.53%
Sao Paulo. Brazil 15,538,682 11.46%
Buenos Aires, Argentina 10.759,291 35.47%
Santiago. Chile 4.227.049 34.87%
Montevideo. Uruguay 1.157.450 39.36%
130 Part I Market Fortes in the Development of Cities
associated with importing and exporting goods. For example, the building of a port
requires substantial investments in infrastructure, generating large economies of
scale (declining average cost) for shipping and handling imports and exports. As a
result, a single port could be much more efficient than two or three ports, and a nation
which depends heavily on trade could have a few relatively large cities involved in
trade and commerce.
The experiences of London and Buenos Aires illustrate the role of trade in
urban growth and concentration. London's population rose from 55.000 in 1520
(about 2 percent of England's population) to 200.000 in 600. and rose to 475.000 in 1
1670 (about 10 percent of England's population). During the late sixteenth century,
international trade involving England grew rapidly, a result of military victories
against the Spanish, improvements in shipping technology, the discovery of huge
markets in Asia and the Americas, government policies promoting trade, and massive
emigration to the New World. The rapid growth in trade was a big factor in the rapid
growth of London. In the case of Buenos Aires, between 1 887 and 1914. exports from
Argentina rose by 400 percent, and the population of Buenos Aires the principal —
trading city —
increased by 1.1 million (265 percent).
The Effect of Trade Restrictions

The more complex theories of large primary cities require a more detailed analytical
framework (Krugman. 1995). Consider a nation that has two cities P and S and — —
a fixed workforce to divide between the two cities. The firms hire workers and sell
their output to workers and consumers. Let's take the perspective of a worker who
must pick a city to live in, either P or S. What are the trade-offs for the worker, given
that P is a large primary city, that is, most of the workers are in P?
1 . Bad news: higher commuting costs. A larger city has longer distances
between workers' homes and the production site, and this is obviously bad
news for the worker. The long commuting distance tends to pull the worker
toward city S.
2. Good news: higher nominal wages. If most workers are in city P. so are
most consumers, so firms in city P have lower transportation costs (for
shipping their output to their customers) than their counterparts in city S.
These lower costs could lead to higher profits in city P. but competition for
workers between firms will generate higher wages instead. The higher wage
will pull the worker toward city P.
3. Good news: lower net price of goods. Most of the goods consumed in city
P are produced in that city, while most of the goods consumed in city S are
produced in P and shipped to S. As a result, the net price of goods (price
plus transport costs) will be lower in city P, and this tends to pull workers
toward city P.
The worker will choose city P — the primary city — if the good news dominates the
bad news, that is. if the benefits of higher \\ ages and lower net prices dominate the
higher commuting costs. This is certainly possible, suggesting that the large primary
city is a feasibleand sustainable outcome.
What happens to the sustainability of this primary-city outcome if we introduce
international trade? From the worker's perspective, the presence of exports and
imports will make both bits of good news less powerful. If firms export some of
their output, the transportation costs associated with serving domestic consumers
will diminish in importance. Although firms in the larger city (P) will still have
lower transportation costs than firms in the small city and thus pay higher wages,
the wage premium will be smaller. In the extreme case, a firm that exports all its
output will pay the same wages in the two cities. On the import side, if consumers
buy import goods instead of some domestic goods, the gap between the net prices
in the two cities will become less important. In the extreme case, a consumer who
buys all imports won't care if net prices are lower in the primary city.
In this model, the introduction of trade makes the primary-city outcome less
likely. Trade weakens the forces pulling the worker toward the primary city but
doesn't affect the force pulling the worker toward the smaller city (lower commuting
costs), so the balance may be tipped in favor of the smaller city. Krugman (1995)
uses this logic to suggest that nations with liberal trade policies are less likely to
have large primary cities. Therefore, one possible explanation for a giant city in a
developing nation is that the nation has restrictive trade policies.
As is often the case in economics, there are conflicting theories concerning the
effects of trade on urban concentration. The simple theory of scale economies in
trade activities suggests that international trade contributes to the development of

urban giants, while the Krugman theory predicts that trade restrictions encourage
the development of such cities. Andes and Glaeser (1995) conclude that although
there is a negative relationship between trade and urban concentration, it may be that
urban concentration is causing low levels of trade, not that low levels of trade are
causing urban concentration. The ultimate resolution of this debate will come from
additional empirical work.
The Role of Infrastructure and Politics
How does the provision of transportation infrastructure affect urban concentration?

If a nation has a relatively inefficient intercity transportation network, the costs of
moving goods within the country will be relatively high, and this reinforces the
tendency to have large primary cities: the higher the cost of transportation, the larger
the wage premium and the price discount in the primary city, so the stronger the pull
toward the larger city. In many developing countries, there are substantial investments
in roads and telecommunication facilities in and around the capital city, but much
less in other areas. This sort of infrastructure policy generates high transport costs
within the country and encourages the development of large primary cities.
What is the role of politics in the development of large primary cities? Andes
and Glaeser (1995) suggest that nations run by dictators have larger primary cities
than democracies. One way for a dictator to stay in power is to take resources from
the hinterland (areas outside the capital city) and redistribute these resources to the
132 Pari I Market Forces in the Development of Cities
people who are most likely to overthrow the dictatorship, that is. the people in the
capital city. If a dictator pays off local agitators, the capital city will grow as people
migrate to the city to ( 1 ) provide goods and services to the folks receiving the payoffs,
and (2) get their own payoffs. Based on a study of cities in 85 countries, Andes and
Glaeser conclude that capital cities in countries with dictatorships are 45 percent
larger than capital cities in other countries.
The experience of Rome illustrates the role of redistribution policies on urban
concentration. In the period 130-50 B.C.. the population of Rome increased from
375,000 to 1,000.000, making Rome more than twice as large as any city up to
that point in history. Military successes during this period extended the empire into
Gaul and the eastern provinces of Asia, providing a large hinterland from which
to extract resources. At the same time, there was political unrest in Rome and the
surrounding areas. The traditional aristocracy responded by distributing free grain
to the residents of Rome, and perhaps just as important, staging the infamous and —
very expensive —
Roman circuses (50 per year at their height). These redistribu-
tion policies encouraged people to move to Rome. When Julius Caesar restored
stability and reduced the grain distribution around 45 B.C.. the growth of Rome
slowed.
Summary
1. The net price of a retail product is the sum of its price and consumers' travel
costs. The market area is the area over which the firm offers the lowest net
price.
2. Market-oriented firms are involved in monopolistic competition. Each has a

local monopoly, but its monopoly power is limited by entry and competition
from other firms.
a. In equilibrium, every firm makes zero economic profit.
b. The entry process decreases travel costs but increases average production
costs.
c. In equilibrium, each firm produces along the negatively sloped portion of its
average production-cost curve.

3. The size of a firm's market area depends on output per firm, per capita
demand, and population density. The market area increases as
a. Output per firm increases.
b. Per capita demand decreases.
c. Population density decreases.
4. Because an increase in scale economies increases output per firm (the output
and per capita demand (demand effect), it has an ambiguous effect on
effect)
market area.
5. A decrease in transportation costs increases output per firm and per capita
demand. The market area will grow if the price elasticity of demand is
relatively small.
Chapters How Many Cities? 133
6. Central place theory, which is based on market-area analysis, predicts that the
regional system of cities will be hierarchical, with a small number of large
cities and a large number of small cities. The set of goods sold in smaller cities
will be a subset of the goods sold in larger cities.
7. If the assumptions of the simple central place theory are relaxed to allow
imperfect substitutes (comparison shopping) and complements (one-stop
shopping), the equilibrium number of cities may decrease, but the urban
hierarchy will not be disrupted.
8. If the assumptions of the simple central place theory are relaxed to allow
demand, the equilibrium number of
variation in per capita cities may change
and the urban hierarchy may be disrupted.
9. Central place theory is not applicable to resource-oriented firms and firms
oriented toward local inputs. The introduction of such firms into the central
place model may disrupt the urban hierarchy.
10. The rank-size rule suggests that the product of city rank and city size
(population) is constant. A study of the rank-size relationship for 44 countries
suggests that population is more evenly distributed between cities than would
be predicted by the rank-size rule.

1. The discussion of the market area of music stores assumed that Bob, Tammy,
and Dick had the same production costs. As a result, each household
patronized the store nearest its home. Suppose that Bob discovers a new way
of marketing CDs that cuts his production cost (and store price) in half, from
$6 to $3. Tammy and Dick continue to sell CDs for $6.
a. How does the decrease in production cost affect Bob's market area?
b. What is the net price at the border between Bob's and Tammy's market
area?
c. Will each household still patronize the firm closest to its residence?
2. Consider the market area of food stores in a region described by the following
assumptions:
i. The per capita demand for food is 30 units.
ii. Population density is 40 people per square mile.
Hi. The land area of the region is 100 square miles.
iv. The output of the typical food store is 6,000 units.
a. How many food stores will there be in the region?
b. How large is the market area of the typical food store?
3. in two independent regions, Low
Consider the market areas of hardware stores
and High. The average production-cost (APC) curves of the two regions reach
their minimum points at the same level of output (1,000 units), but because
Low has lower input costs, the minimum point of its APC curve has a lower
average cost (the average cost of 1,000 units is $1 instead of $2). Will the
market area in Low be larger, smaller, or the same as the market area
in High?
Ifyou don't have enough information to answer the question, indicate what
additional information you need and how you would use it.
4. Consider a regional economy with two cities. The two cities have the same
population, but the average household income in city H exceeds that in city L.
The residents of H and L consume Y. The characteristics of Y are as follows:
i. The average production-cost (APC) curve is U-shaped.

li. The APC curve is the same at all locations.
iii. Y is a standardized product, with no close substitutes or complements.
iv. All households have the same tastes for Y.
v. Y is inferior (negative income elasticity).
a. How will the size of the average market area of Y in city H compare to the
size of the average market area in city L?
b. How would your response to (a) differ if the APC curve were a horizontal
line?
5. Suppose that you intend to purchase the franchise rights for a pizza parlor. The
franchiser has divided your region into two areas of equal size: H is a
high-income area, and L is a low-income area. Suppose that the income
elasticity of demand for pizza is zero: the consumption of pizza is independent
of income. The income elasticity of demand for land is 1.0. Your objective is
to maximize the quantity of pizzas sold.
a. Which of the two franchises will you choose?
b. How would your response to (a) change if the income elasticity of demand
for pizza is 1.5?
6. In the example based on the data in Table 5-3. the implicit assumption is that
travel costs were the same in the two cities. Suppose that the opportunity cost
of travel is higher in the wealthy city. How might this change the market areas
in the wealthy city?
7. Consider the city of Vidville. where initially all video consumers travel to
video outlets to rent videos. Suppose thatwhen the video outlets offer home
delivery, all video consumers switch to home delivery.
a. Under what circumstances is the switch to home delivery rational?
b. In what type of cities would you expect home delivery to be efficient ?
8. In 1930, most rural children walked to school. By 1980. most of them rode
school buses. Would you expect the average "market area" of rural schools to
increase or decrease as a result of the school bus?
9. The introduction of parcel post decreased transportation costs and increased

the market area of the typical firm. Can you think of any good for which a
decrease in transportation costs is likely to decrease the market area?
10. In the city of Metro, retail firms locate either in the central area of the city or in
one of the suburban subcenters. There are two retail goods sold in the city. A
and B. The per capita demands for the two goods are equal. The scale
economies associated with the production and sale of A are large relative to
the scale economies associated with B. Where in the city will the firms selling
A and B locate?
1 1 Consider a city with a uniform distribution of population where every
household consumes the same number of video rentals. The city is two miles
long and two miles wide. The mayor recently stated her policy for the location
of video rental outlets: "A video outlet reaches the minimum point of its
average production-cost curve at an output of 1.000 units. Since the total
demand for video rentals is 4,000 units, we should have four video outlets.
Since the distribution of population is uniform, the video outlets should be
distributed uniformly throughout the city, with every outlet at the center of a
one-square-mile market area." Comment on the mayor's policy. Will it lead to
an efficient distribution of video outlets?
12. In the city of Zone, the actual number of grocery stores is less than the number
that would occur in the absence of zoning. A recent survey of grocery prices
suggests that Zone consumers pay less for groceries than people in a similar
city without zoning. For example, the prices of Spam and Velveeta are 5
percent lower in Zone.
a. Is there any reason to doubt the validity of the survey?
b. Suppose that the survey is valid. Does the zoning policy make the residents
of Zone better or worse off?
13. A developer has requested a permit to build a drugstore in a rapidly growing
part of your city. Your mission is to figure out whether the new drugstore is
appropriate. What information do you need, and how would you use it?
14. Explain why poor areas of cities typically are served by small grocery stores,
not by large grocery-store chains.
15. Consider a region in which all firms are market-oriented (all employment is in
retail outlets). There are no shopping externalities, and all retail activities are
subject to the same degree of internal scale economies. In the last 10 years, the
region's rank-size curve (with city rank on the horizontal axis and city size on
the vertical axis) has become flatter.
a. What changes in market areas are lurking behind the flattening of the
rank-size curve?
b. Provide an explanation for the changes in the market areas lurking behind
the flattening of the rank-size curve.
16. Mr. Wizard, a regional planner, recently made the following statement: "If my
assumptions are correct, all cities in this region will eventually be identical.
They will be the same size and will sell the same set of goods."
a.Assuming that Mr. Wizard's reasoning is correct, what are his assumptions?
b. Are Mr. Wizard's assumptions realistic?
17. Some people claim that state capitals (e.g., Sacramento, California; Salem,
Oregon; Olympia, Washington) are boring cities. Specifically, it is claimed that
these cities have fewer goods than one finds in other cities of equal size. After
checking a map to see where each of these capital cities is located, use central
place theory to explain why they might be considered boring.

General Discussions of Central Place Theory
Beckmann. M. J. "City Hierarchies and the Distribution of City Size." Economic
Development and Cultural Change 7 (1958), pp. 243-48.
Christaller, Walter. Central Places in Southern Germany, trans. C. W. Baskin. Englewood.
N.J.: Prentice Hall. 1966. The classic piece on central place theory.
Hamilton, Bruce W. "Market Failure in the Land of Losch." Journal of Urban Economics
12 (1989). pp. 143-55. Discusses the efficiency properties of spatial competition.
Hoover, Edgar M. An Introduction to Regional Economics. New York: Alfred A. Knopf,
1975. Chapter 6, "The Location of Urban Places."
."Transport Costs and the Spacing of Central Places." Papers of the Regional Science
Association 25 (1970), pp. 255-74. Discusses the effects of changes in transport costs
on the size of the market area.
King, Leslie J. Central Place Theory: Beverly Hills, Calif.: Sage. 1984. A thorough
discussion of central place theory and its implications. Includes a careful discussion of
the origins of central place theory.
Losch, August. The Economics of Location. New Haven, Conn.: Yale University Press.
1954. A refinement of central place theory.
Mills, Edwin S., and Michael R. Lav. "A Model of Market Areas with Free Entry." Journal
of Political Economy 72 (1964), pp. 278-88. Discusses the shapes of market areas.
"The Nature of Economic Regions." In Regional Development and Planning: A Reader, ed.
John Friedman and William Alonso. Cambridge, Mass.: MIT Press, 1964. Discusses
the roles of scale economies and transport costs in determining the market area of the
firm.
Noyelle, Thierry, and Thomas Stanback. The Economic Transformation of American Cities.
Totowa, N.J.: Rowman and Allanheld. 1984. Discusses exceptions to the central place
framework.
Parr, J. B. "Growth Poles, Regional Development, and Central Place Theory." Papers of the
Regional Science Association 3 ( 1973(a)). pp. 173-212.
. "Structure and Size in the Urban System of Losch." Economic Geography 49
(1973b). pp. 185-212.
Parr, J. B., and K. G. Denike. "Theoretical Problems in Central Place Analysis." Economic
Geography 46 (1970), pp. 568-80.
Philbrick. Allen. "Areal Functional Organization in Regional Geography." Papers of the
Regional Science Association 3 ( 1957), pp. 87-98. Describes the hierarchy of urban
places.
Empirical Tests of Central Place Theory
Berry. Brian J. L. Geography of Market Centers and Retail Distribution. Englewood, N.J.:
Prentice Hall, 1967. Discusses central place theory, including a review of empirical
tests of the theory.
Berry. Brian J. L.. and William L. Garrison. "The Functional Bases of the Central Place
Hierarchy." Economic Geography 34 ( 1958). pp. 145-54. An empirical analysis of
central place theory.
Chapter 5 How Manx Cities' 137
Lloyd, Peter E., and Peter Dicken. Location in Space: A Theoretical Approach to Economic
Geography. New York: Harper & Row, 1977. Reviews empirical work on central place
theory.
Preston, Richard E. "The Structure of Central Place Systems." In Systems of Cities, ed. L. S.
Bourne and J. W. Simmons. New York: Oxford University Press. 1978, pp. 185-206.
Empirical analysis of central place theory in the Pacific Northwest.
Rosen, Kenneth T., and Mitchell Resnick. "The Size Distribution of Cities: An Examination
of the Pareto Law and
Primacy." Journal of Urban Economics 8 (1980), pp. 165-86.
Estimates the rank-size relationship and concludes that the rank-size rule is a poor
approximation to the true size distribution.
Central Place Theory and Intraurban Distribution of Activity
Carol, H. "The Hierarchy of Central Functions within the City." Annals of the Association
of American Geographers 50 (1960), pp. 419-38. Applies central place theory to the
location of activities within cities.
Proudfoot, M.J. "The Outlying Business Centers of Chicago." Journal of Land and Public
Utility Economics 13 (1937), pp. 57-70. Applies central place theory to the location of
activities within cities.
Weiss, Leonard. "The Geographic Size of Markets in Manufacturing." Review of Economics

and Statistics 54 (1972), p. 245.
Technology and the Urban Hierarchy
Berry, Brian J. The Basis of Developmental Filtering and

L. "Hierarchical Diffusion:
Spread in a System of Growth Centers." In Growth Centers in Regional Economic
Development, ed. N. M. Hansen. New York: Free Press. 1972. Discusses how
innovations are first utilized in large cities and then spread to urban areas lower on the
urban hierarchy.
Huang, J. C, and P. Gould. "Diffusion in an Urban Hierarchy: The Case of Rotary Clubs."
Economic Geography 50 (1974), pp. 333-40. Discusses how new organizations are
first established in large cities and then spread to urban areas lower on the urban
hierarchy.
Public Facilities
Dear, Michael. "Planning for Mental Health Care: A Reconsideration of Public Facility
Location." International Regional Science Review 3 (1978), pp. 93-1 1 1
Getz, Malcolm. The Economics of the Urban Eire Department. Baltimore: Johns Hopkins.
1979. Estimates the relationship between the density of stations and losses from fires.
The elasticity of losses with respect to the number of stations per square mile is —0.70.
Urban Giants and Primary Cities
Andes, Alberto F, and Edward L. Glaeser. "Trade and Circuses: Explaining Urban Giants."
Quarterly Journal of Economics (1995), pp. 195-227.
Krugman, Paul. "Urban Concentration: The Role of Increasing Returns and Transport
Costs." International Regional Science Review 19 (1996), pp. 5-30.
Chapter 6
R B A N ECONOMIC GROWTH
11
An economic forecaster is like a cross-eyed javelin thrower: he doesn't win
many accuracy contests, but he keeps the crowd's attention.
Anonymous
C> arlier chapters have shown why cities exist and where they develop. This chapter
(S> takes a closer look at the economy of a particular city, exploring the market
forces that determine the size of the economy. The first part of the chapter uses a
simple model of the urban labor market to describe the economic growth process. If
economic growth is defined as an increase in total employment, growth results from

an increase in the demand for labor (caused by an increase in the demand for the
goods produced in the city) or an increase in labor supply (caused by migration to
the city).
The second part of the chapter explores the effects of various public policies
on urban economic growth. The public sector can influence both sides of the urban
labor market. For example, the provision of local public services such as schools,
parks, and public safety increases the relative attractiveness of the city, causing in-
migration that increases the supply of labor. On the demand side, the provision of
industrial infrastructure such as roads and sewer systems decreases production costs
and attracts firms. A city's tax policy also affects the supply and demand for labor:
an increase in residential taxes decreases labor supply, and an increase in business
taxes decreases labor demand.
The third part of the chapter describes two techniques used to forecast urban
economic growth. Both forecasting techniques (the economic base study and input-
output analysis) suffer from a number of conceptual problems that limit their applica-
bility and accuracy. Nonetheless, they are the most frequently used and abused —
forecasting techniques, and a clear understanding of their strengths and limitations
is useful for both policymakers and firms. Local governments use these techniques
to generate economic forecasts that are then used to project the demands for local
139
public services, and firms use the same type of economic forecasts to project the
demands for their products.
The fourth part of the chapter explores the effects of employment growth on other
measures of economic welfare, including the city's unemployment rate, its labor-
force participation rate, and per capita income. This part of the chapter addresses
two questions. First, when a city grows, what fraction of the new jobs are filled by
newcomers, and what fraction are filled by original residents who would otherwise
not be employed? Second, how does employment growth affect real income per
capita?
The fifth
part of the chapter presents facts concerning the composition of changes
in totalemployment. The net change in total employment in a particular city can
be divided into four components: jobs gained from the opening of new plants, jobs
gained from the expansion of existing plants, jobs lost from plant closings, and jobs
lost from the contraction of existing plants. This part of the chapter addresses the
following question: do cities that grow rapidly have relatively large employment
gains (from openings and expansions) or relatively small losses (from closures and
contractions)?
The Urban Labor Market and Economic Growth

This part of the chapter describes a model of the urban labor market for a city that is
part of a larger regionaleconomy. The model is a long-run model in the sense that
it assumes that households and firms can move freely between cities in the region.
In the long run. each household lives in the city that maximizes its utility, and each
firm locates in the city that maximizes its economic profit. As explained later in
the chapter, the long-run mobility of households has important implications for the
effects of employment growth on the welfare of the original residents of the city.
Suppose that the size of an urban economy is defined in terms of total employ-
ment. Figure 6-1 shows the supply and demand curves for the city's labor market.
The equilibrium wage is $1,000 per month and equilibrium employment is 50,000
laborers. Economic growth, defined here as an increase in total employment, results
from either an increase in demand (a rightward shift of the demand curve) or an
increase in supply (a rightward shift of the supply curve).
The Demand for Labor

Labor demand comes from two types of activities. The export or basic sector
sells its products to consumers outside the city. Examples of export firms are steel
producers and computer manufacturers. In contrast, the local or nonbasic sector

sells its products to consumers within the city. Examples of local producers are
bakeries, bookstores, and local schools. Before discussing the city's labor-demand
curve, it will be useful to explain the relationship between these two employment
sectors.
Chapter 6 Urban Economic Growth 141
Figure 6-1 The Urban Labor Market
Labor supply
&> 1.000
25 50 75
Number of laborers in city (thousands)
The labor market is in equilibrium if supply equals demand. The equilibrium wage is $ .000 per month, and the
1
equilibrium number of laborers is 50.000.
Table 6-1 Increases in Local Spending from Increased

Export Sales
Increase in Round 1 Round 2 Round 3 Total
income $100,000 $60,000 $36,000 $250,000

consumption 60,000 36,000 21,600 150,000
ts 40,000 24,000 14,400 100,000
The Multiplier Process. Because export workers spend part of their income on
local goods, an increase in export sales increases local sales. Table 6-1 shows the
multiplier effects of a $100,000 increase in export sales. In round 1, the increase
in export sales is paid out to workers, capitalists, form of and landowners in the
wages, and land rent, so local income increases by $100,000. If 60 per-
interest,
cent of the increased income is spent on local goods, local consumption increases
by $60,000. The first-round increase in local consumption is the second-round in-
crease in local income: local producers pay the $60,000 to their workers, capital-
ists, and landowners. Because 60 percent of the increased income is spent on local
goods, the second-round increase in local consumption is $36,000. The spending and
142 Purr I Murker Forces in rhe Development of Cities
respending of income continues forever, but every spending round is smaller than
the previous one because 40 percent of income is spent on imports.
The increase in the city's total income exceeds the original increase in export
income. The marginal propensity to consume locally (in) income is the fraction of
spent locally (0.60 in Table 6-1 ). change in export sales is AX. the
If the original
increase in total income is the sum of the income changes from a series of spending
rounds:
ATotal income = A X + in AX + nr AX + m 3 AX + /» 4 AX +
- • • . . . (6-1
The change in total income is the sum of an infinite series, and can be rewritten as
ATotal income — AX •
(6-2)
1 — in
The income multiplier is defined as the change in total income per unit change in
export sales:
ATotal income 1
= (6-3)
AX 1-/?;
For example, if the marginal propensity to consume locally (in ) is 0.60. the multiplier
is
ATotal income 1
= 2.5 (6-4)
AX 1 - 0.60
For every additional dollar of export sales, total income increases by 2.5 dollars.
In Table 6-1. export sales increase by S 100.000. so local production increases by
$150,000 and total income increases by $250,000.
The multiplier process also affects employment. The employment multiplier
is the change in total employment per unit change in export employment. If there
is one job per $1,000 of sales, an increase of $100,000 in export sales will increase
export employment by 100 jobs. Similarly, the additional $150,000 in local sales will
increase local employment by 150 jobs. Total employment in the city will increase
by 250. meaning that the employment multiplier is 2.5.
The employment multiplier is the same as the income multiplier because the
sales volume is assumed to be the same in both sectors. If sales volume per local
job is less than sales volume per export job. the employment multiplier exceeds the
income multiplier because more local jobs will be supported b\ a given amount of
sales. For example, if sales volume per local job is $500 instead of $1,000. local
employment would increase by 300 instead of 150. and the employment multiplier
would be 4.0 instead of 2.5. Conversely, if sales volume per local job exceeds $1,000.
the emploj ment multiplier would be less than the income multiplier.
A second method for computing the employment multiplier uses data on the
composition of the workforce. If B is the number of jobs in export industries and L
is the number of jobs in local industries, total employment in the city is B + L. or T.
The ratio (L/ B) indicates how main local jobs are "supported" by each export job.
For example, if B = 20. 000 and L = 30.000. an economic "base"" oi 20,000 jobs
supports 30,000 local jobs. Each export job supports 1.5 local jobs. The employment
multiplier, defined as the change in total employment per additional export job, is
AT T
= - = 2.5 (6-5)
AB B
Every additional export job increases total employment by 2.5 jobs: in addition to
the export job, there are 1.5 new "support" jobs in local industry.
The Labor-Demand Curve. The city's labor-demand curve is negatively sloped

for two reasons. First, as the city's wage increases, both exporters and local producers
substitute capital for the relatively expensive labor. This is the substitution effect:
an increase in the wage causes factor substitution that decreases the quantity of
labor demanded. In other words, an increase in the wage causes firms to substitute
nonlabor inputs (capital, land, raw materials) for labor. Second, as the city's wage
increases, production costs increase, increasing the prices charged by the city's firms.
As the prices of exports increase, the quantity of exports demanded decreases, so
exporters need fewer workers. As the prices of local goods increase, city residents
substitute imports for the relatively expensive local goods, so local firms need fewer
laborers. This is the output effect: an increase in the city's wage increases prices and
decreases output, decreasing the quantity of labor demanded. The demand curve is
negatively sloped because an increase in wages generates both a substitution effect
and an output effect.
What determines the slope of the demand curve? In other words, how rapidly
does labor demand fall as the wage increases? The larger the substitution effect and
the output effect, the greater the responsiveness of labor demand to changes in the
wage, so the flatter the demand curve. If both the substitution effect and the output
effect are relatively large, a given change in the wage will cause a relatively large
decrease in the quantity of labor demanded, meaning that the demand curve will
be relatively flat. The substitution effect will be relatively large if firms can easily
substitute nonlabor inputs for labor. The output effect could be relatively large for two
reasons. First, if labor costs are responsible for a relatively large share of total costs,
the output price will be sensitive to changes in the wage: a small increase in the wage
will cause a relatively large increase in the output price and thus a relatively large
decrease in total production. Second, if the price elasticity of demand for the output
is relatively large, a given increase in the output price will cause a relatively large
decrease in total production. To summarize, the demand curve will be relatively flat
if (1) labor and nonlabor inputs are good substitutes, (2) labor costs are a relatively
large fraction of total costs, and (3) the demand for the city's output is relatively
elastic.
What causes the demand curve to shift to the right or the left? The following
factors determine the position of the curve:
1 . Demand for exports. An increase in the demand for the city's exports
increases export production and shifts the demand curve to the right: at
every wage, more workers will be demanded.
2. Labor productivity. An increase in labor productivity decreases

production costs, allowing export firms to cut prices and increase output.
Although firms need fewer workers to produce a given amount of output,
the decrease in price means that firms produce more output. If the increase
in output is relatively large ( if the price elasticity of demand for the city's
output is relatively large), the demand for export workers increases: the
demand curve shifts to the right. Similarly, an increase in labor productivity
allows local producers to underprice imports, increasing the demand for
local workers. One way to increase productivity is to improve the quality of
local public education.
3. Business taxes. An increase in business taxes (without a corresponding

change in public services) increases production costs and decreases output,
so the demand curve shifts to the left. The demand for labor decreases as
export firms lose customers to exporters in other cities and local firms lose
customers to importers. As discussed in Chapter 3. there is solid evidence
that an increase in business taxes decreases business activity and thus
decreases the demand for labor.
4. Industrial public services. An increase in the quality of industrial public-

services (without a corresponding increase in taxes) decreases production
costs and increases output, so the demand curve shifts to the right. The
demand for labor increases as export firms gain customers from exporters in
other cities and local firms gain customers from importers. As discussed in
Chapter 3, there is solid evidence that improvements in local infrastructure
increase business activity and thus increase the demand for labor.
5. Land-use policies. Industrial firms require production sites that (a) are
accessible to the intracity and intercity transportation networks and (b) have
a full set of public services (water, sewerage, electricity). By coordinating
its land-use and infrastructure policies to ensure an adequate supply of
industrial land, a city can accommodate (a) existing firms that want to
expand their operations and (b) new firms that want to locate in the city.
Figure 6-2 shows the multiplier effect of an increase in export sales. Suppose
that an increase in the demand for exports increases the demand for export workers
by 10.000. The city's right from D\ to D^. at a wage
demand curve will shift to the
of $1,000 per month, an additional 10.000 export workers will be demanded. This
is the direct effect of an increase in export demand. If the employment multiplier is
2.5, every export job supports 1 .5 local jobs, so the demand curve shifts to the right
by an additional 15.000 workers (from D: to D\). This is the multiplier effect of an

increase in export demand. Total labor demand increases b\ 25.0(H) (2.5 times the
increase in the demand for export laborers).
The Supply Curve

Consider next the supply side of the urban labor market. The supply curve is pos-
itively sloped because of the migration effect: an increase in the wage increases
Figure 6-2 Direct and Multiplier Effects of an Increase in Export Employment
1.000 Demand with direct and

multiplier effects
Demand with direct effect
50 60 75
If export employment increases by 10,000 jobs, the demand curve shifts to the right in two steps. The shift from D,
to Di is the direct effect of the increase in export employment. At a fixed wage, total employment rises from 50.000
to 60,000. The shift from Di to D t reflects the multiplier effect of the increase in export employment. If the employment
multiplier is 2.5. local employment increases by 1 .5 times the increase in export employment, increasing labor
demand by an additional 1 5.000 jobs.
the relative attractiveness of the city, causing the migration of workers from other
cities in the region. One of the assumptions underlying the supply curve is that the
number of work hours per laborer is fixed: changes in the wage do not affect the
number of work hours per laborer. This assumption is consistent with empirical evi-
dence concerning the elasticity of work hours with respect to the wage: in aggregate,
Another assumption underlying the supply curve is
this elasticity is close to zero.
by changes in the wage. In other
that the labor-force participation rate is unaffected
words, an increase in the wage does not increase the fraction of the city's population
in the workforce. Given these two assumptions, an increase in the wage increases
the supply of labor because more workers move to the city, not because existing
workers work more hours or because more of the city's current residents join the
workforce.
Why is the supply curve positively sloped, and what determines the slope of the
curve? An increase in total employment (and population) increases the total demand
for most goods, causing increases in the prices of land, housing, and other goods. A
growing city must offer a higher wage to compensate its workers for its higher cost
of living. The results of Hamilton and Schwab (1985) and Bartik (1991) suggest
that the elasticity of housing prices with respect to total employment is about 0.35:
a 10 percent increase in total employment increases housing prices by about 3.5
percent. According to Bartik (1991), the elasticity of the cost of living with respect
to city size (including all price changes) is about 0.20: a 10 percent increase in total
employment increases the cost of living by about 2 percent. This means that to keep
real wages constant, the city's nominal wage must rise by about 2 percent for every
10 percent increase in employment.
A number of empirical studies have estimated the responsiveness of wages to
total employment. The results of Treyz and Stevens (1985). Robak (1982), and Bartik
( 1 99 1 suggest
) that the wage for a given occupation increases at about the same rate
as the cost of living. In other words, the elasticity of the nominal wage with respect to
total employment is about 0.20: a 10 percent increase in total employment increases
the wage by 2 percent. Alternatively, the elasticity of labor supply with respect to
the wage is about 5.0: a 2 percent increase in the wage causes a 10 percent increase
in labor supply. It's important to note that this is the labor-supply elasticity for an
individual city. The elasticity is relatively large because workers migrate to cities that
offer relatively high wages. The national supply elasticity is much lower than the local
supply elasticity because there is less migration between nations than between cities.
What causes the supply curve to shift to the right or the left? The position of the
supply curve is determined by the following factors:
1 Environmental quality. An increase in the quality of the environment

(better air or water quality) increases the relative attractiveness of the city,
causing migration to the city. The supply curve shifts to the right: at every
wage, more workers are willing to work in the city.
2. Residential taxes. An increase in residential taxes (without a corresponding

change in public services) decreases the relative attractiveness of the city,
causing out-migration that shifts the supply curve to the left.
3. Residential public services. An increase in the quality of residential public

services (without a corresponding increase in taxes) increases the relative
attractiveness of the city, causing in-migration that shifts the supply curve
to the right. The results of Eberts and Stone (1992) suggest that labor
supply is responsive to changes in local infrastructure.
Equilibrium Effects of Demand and Supply Shifts
Figure 6-3 shows the effects of an increase in export sales on the urban labor market.
The labor-demand curve shifts to the right by 25.000 workers, reflecting both the
direct effect and the multiplier effect of an increase in 10.000 export jobs. As the
population of the city increases, the prices of housing and land increase, requiring
an increase in the wage to compensate workers for the higher cost oi' living. In other
words, the city moves up its supply curve. The equilibrium wage rises from SI. 000
per month to SI .300. and the equilibrium number of laborers increases from 50.000
to 66.000.
Figure 6-3 suggests that predicting the effects of an increase in export employ-
ment The simple approach is to use the employment multiplier to predict
is tricky.
the change in total employment from a projected change in export employment. In
the numerical example, the predicted change in total employment from this method
would be 25.000 (2.5 times 10.000). This approach provides an estimate of the
Figure 6-3 Equilibrium Effects of an Increase in Export

Employment
Labor supply
1.300
1,000
25 50 66 75
An increase in exports shifts the demand curve by 25,000. As the city grows, the
to the right
prices of housing and land increase, requiring a higher wage
to compensate workers for the
higher cost of living. The equilibrium wage increases from $1,000 to $1,300, and the
equilibrium number of laborers increases by only 16,000.
horizontal shift of the demand curve, not the change in equilibrium employment. To
accurately predict the change in total employment, one must also know the slopes of
the supply and demand The slope of the supply curve indicates how rapidly
curves.
the wage demand curve indicates how
increases with city size, and the slope of the
rapidly the quantity of labor demanded decreases as the wage increases.
Figure 6-4 shows the effects of a rightward shift of the supply curve. Sup-
pose that the city improves its residential public services. For example, the city
could improve its public-safety programs or alter its transportation system to de-
crease commuting costs. In Figure 6—4, the labor-supply curve shifts to the right:
at each wage, more people are willing to work in the city. The shift of the supply
curve increases equilibrium employment and decreases the equilibrium wage. Fig-
ure 6-4 is consistent with the empirical evidence provided by Eberts and Stone
(1992) concerning the effects of improvements in local infrastructure on wages and
employment. Workers accept lower wages because
total the city provides a superior
mix of local public goods.
Agglomeration Economies and Urban Growth
In this chapter, one reason cited for urban growth is an increase in the demand for
workers, represented by a shift of the labor-demand curve. Let's take a closer look
at what changes are causing the shift of the demand curve.
148 Part I Market Font's in the Development of Cities
Figure 6^4 Improvement in Public Services Shifts the Supply Curve

and Increases Total Employment
Original
supply
Supply with
improved public-
1,000
875
An improvement in local public goods increases the relative attractiveness of the city, shifting the supply curve to
the right. The equilibrium number of laborers increases to ?5,()()0. and the equilibrium wage falls to$875 per
month. Workers accept a lower wage because the city provides a more efficient set of local public services.
Suppose that in response to an increase in the demand for software, an en-

trepreneur starts a new software firm in the city. When the firm enters the market,
the labor-demand curve will shift to the right. For example, if the software firm
hires 100 workers, the initial shift of the demand curve (before multiplier effects
are considered) is 100 workers. If there are localization economies in the software
industry, however, this is not a complete story. As we saw in Chapter 2, there are
three sources of localization economies.
1 Scale economies in intermediate goods. The average cost of an

intermediate input is lower in a cluster of firms that share the suppliers of
the input. The start-up firm increases the total output of the city's software
industry, allowing the further exploitation of scale economies in the
production of intermediate inputs. The average cost of intermediate inputs

fall, decreasing the production costs of all software firms. These lower
production costs encourage other firms to expand their production and hire
more workers.
2. Labor-market pooling. A firm that shares a workforce with other firms
can easily expand its workforce when times are good and contract it when
times are bad. The larger the number of software firms, the larger the
benefits of associated labor-market pooling, and the higher the profits of
software firms. These benefits encourage new firms to join the cluster and
encourage existing firms to produce more output and hire more workers.
3. Knowledge spillovers. An industry cluster facilitates the rapid exchange of
information and the diffusion of technology. The new software firm
increases the number of software workers who can exchange information,
and this allows firms in the cluster to be more successful, so they will hire
more workers.
The lesson is that if there are localization economies, an increase in labor demand
by one firm may cause other firms more workers too.
in the industry to hire
The same logic applies to an increase in the quantity of labor demanded resulting
from a decrease in wages. Suppose a city's wage drops, allowing a firm in a particular
industry to underprice its competitors in other cities. As a result, the firm will hire
more workers and produce more output. If the industry is subject to localization
economies, other firms produce more output and hire more
in the industry will
workers too. Localization economies make the hiring decisions of firms in an industry
interdependent.
The same logic applies to urbanization economies. While localization economies
are efficiencies resultingfrom the growth of a particular industry, urbanization
economies result from the growth of the entire metropolitan economy. If there are
urbanization economies, a software start-up firm generates benefits for firms outside
the software industry (lower costs of intermediate inputs, benefits from labor pool-
ing, knowledge spillovers) and these benefits will tend to increase production and
labor demand in other industries.
Public Policy and Economic Growth

Public policy affects the equilibrium number of workers by shifting the city's supply
and demand curves. As explained earlier in the chapter, local government can shift the
demand curve to the right by improving the local education system, public services,
business infrastructure, and cutting business taxes. Local government can shift the
supply curve to the right by improving public services and residential infrastructure
and cutting residential taxes. This part of the chapter explores the effects of two
alternative policies: the subsidization of relocating firms and a tax on air pollution.
Subsidy Programs
Many cities try to attract export firms by offering special subsidies for new firms.
Levy (1985) describes several types of subsidy programs, including the following:
1 . Tax abatement. In some cities, new firms are exempt from local property
taxes for some period of time, often up to 10 years. Some cities offer tax
150 Part I Market Force* in the Development of Cities
abatements to all new developments, while others offer abatements only to

firms that are assumed to be particularly sensitive to tax differentials.
2. Industrial bonds. Some bonds to finance
cities issue tax-free industrial
property development. The local government uses the revenue from the
bonds to purchase the land, and leases the property to a private firm.
Because the interest income from industrial bonds is not subject to federal
taxes, the bond buyer accepts a relatively low interest rate (for example.
8 percent instead of 12 percent). Therefore, the lessee pays less than the
market interest rate on the money borrowed to finance the project. The use
of industrial bonds was sharply curtailed by the Tax Reform Act of 1986.
3. Government loans and loan guarantees. Some cities loan money to
developers, and others guarantee loans from private lenders to developers.
In both cases, developers borrow money at a relatively low interest rate:
either the city charges an interest rate below the market rate, or the city
decreases the risk associated with a private loan, allowing the developer to
borrow private money at a relatively low interest rate.
4. Site development. Some cities subsidize the provision of land and public
services for new development. The city purchases a site, clears the land,
builds roads and sewers, and then sells the site to a developer at a fraction
of the cost of acquiring and developing the site.
How do these subsidy programs affect the urban labor market? Any policy
that decreases production costs will increase labor demand and increase equilibrium
employment, everything else being equal. In the case of tax subsidies, not even thing
else is equal. Tax revenue supports local public sen ices, so a community w ith low-
taxes is likely to also have inferior public services. The empirical evidence cited
in Chapter 3 suggests that if a city cuts taxes and decreases its spending on public
sen ices (highways, education, public safety), it is unlikely to grow, and may in fact
shrink. In contrast, if the city cuts taxes and decreases its spending on redistributional
programs to the poor, the city is likely to grow. In other words, the effect of a tax cut
depends on what type of sen ices are cut along with taxes. If taxes are used to finance
public services used by businesses, a tax cut is unlikely to stimulate economic growth.
What are the fiscal implications of these subsidy programs? Another objective
o\ a subsidy program is to decrease the local tax burden. If a tax cut or a subsidy
makes the city more attractive to firms, the city's tax base will increase, increasing
total tax revenue. As the city grows, it w ill also spend more on local public sen ices
(roads, schools, police, fire protection). The subsidy program will be beneficial from
the fiscal perspective if the increase in tax revenue exceeds the increase in the cost
of public services.
Urban Growth and Environmental Quality

Is there a trade-off between environmental quality and economic growth'.' Sup-
pose that a city adopts a pollution-abatement program. Will the abatement program
increase or decrease the city's total employment?
Consider a city with two industries, a polluting steel industry and a relatively
"clean" assembly plant. If the city imposes a pollution tax of $100, steel producers
pay $100 for every ton of pollution they generate. The pollution tax affects both
sides of the urban labor market.
1. Shift of demand curve. The tax increases the production costs of steel
producers. In addition to paying for labor, capital, and land, a firm pays
$100 for every ton of pollution. The increase in production costs increases
the price of steel, decreasing steel production and decreasing the demand
for labor. In Figure 6-5, the demand curve shifts to the left: at every wage,
less labor is demanded.
2. Decrease in pollution. The tax decreases air pollution for two reasons.
First, steel producers will install pollution-control equipment as a means of
decreasing their pollution taxes, so the amount of pollution generated per
ton of steel will decrease. Second, the increase in the price of steel
decreases total steel production.
3. Shift of supply curve. The improvement of the city's air quality increases
the relative attractiveness of the city. People sensitive to air quality will
move to the city, shifting the supply curve to the right.
Figure 6-5 Pollution Tax Increases Total Employment
Initial labor
supply
Labor supply in
.000 cleaner city
687 Initial labor demand
A pollution tax increases production costs, decreasing the demand for labor. It also improves environmental
quality, increasing the supply of labor and decreasing the wage. The outward shift of the supply curve is large relative
to the inward shift of the demand curve (people are relatively sensitive to air pollution), so total employment
increases. The increase in employment in the clean industry more than offsets the decrease in employment in the
polluting industry.
152 Part 1 Market Forces in the Development of Cities
Figure 6-5 shows one possible outcome of the abatement program. Since supply
increases and demand decreases, the program decreases the equilibrium wage. Since
the rightward shift of the supply curve is large relative to the leftward shift of the
demand curve, equilibrium employment increases. The supply shift will be relatively
large if households are relatively responsive to changes in environmental quality,
meaning that a large number of households migrate to the city as air quality improves.
How does the abatement program affect the distribution of employment between
the polluting industry and the clean industry? As the wage falls, the production costs
of both industries decrease. For the steel industry, the decrease in the wage partly
offsets the increase in pollution taxes. The abatement program is likely to generate
a net increase in production costs, so the polluting industry is likely to decrease its
total workforce. In contrast, the clean industry will simply pay lower wages, so its
production costs will decrease and employment will increase. In Figure 6-5,
its total
the increase in employment in the cle in industry more than offsets the decrease in
employment in the steel industry, so total employment increases. This occurs because
households are relatively sensitive to pollution, so that migration to the cleaner city
causes a large decrease in the wage.
Figure 6-6 shows another possible outcome of the abatement program. In this
case, households are less responsive to changes in environmental quality, so the
Figure 6-6 Pollution Tax Decreases Total Employment

$
I abor Mipph in
cleaner cit)
.000
812
Initial labor demand
Number oi laborers in cit> (thousands)

A pollution tax decreases the demand for labor and increases the supplj ol labor. The outward shift oi the supplj
curve iv small relative to the inward shift of the demand curve (people are relativel) insensitive to air pollution),
so total employment decreases
increase in supply is small relative to the decrease in demand, and the equilibrium
number of laborers decreases. Because the supply curve shifts by a relatively small
amount, the wage decreases by a small amount. Therefore, the increase in employ-
ment in the clean industry is not large enough to offset the decrease in employment
in the steel industry.
The lesson from Figure 6-5 and Figure 6-6 is that an abatement program may
either increase or decrease totalemployment. Total employment will increase if the

supply response (migration that decreases wages) is large relative to the demand re-
sponse (the decrease in labor demand from the polluting industry). Total employment
will decrease if the demand response is relatively large.
Predicting Economic Growth

This section discusses two techniques that are used to predict future growth patterns.
The economic base study is a simple, low-cost method for projecting employment
growth. The input-output study provides a more sophisticated approach to economic
forecasting. Both techniques are designed to estimate the increase in labor demand
from an increase in export sales.
The Economic Base Study

According to economic base theory, the export industries form the economic base of
the urban economy. Like a household, a city earns its livelihood by producing goods
for others. Exports bring money into the local economy, increasing local income and
employment through the multiplier process. The ultimate purpose of an economic
base study is to estimate the increase in total employment generated by an increase
in export employment. In algebraic terms,
T
AT = Multiplier AB = - AB (6-6)
In other words, the predicted change in total employment (AT) equals the employ-
ment multiplier (T/B) times the change in export employment (AB). To predict the
change in total employment, the economic forecaster needs (1) an estimate of the
multiplier and (2) a projection of the change in export employment.
To compute the employment multiplier, the economic forecaster estimates how
much of the city's workforce is involved in export production. The most precise
method of estimating B is to collect information on the actual shipments of goods
to other cities. This approach is costly, and most cities use less expensive and less
precise methods to estimate export employment.
The simplest method is to classify each industry as either an export industry
or a local industry. For example, if the mitten industry is assumed to be an export
industry, all the employees of mitten producers would be counted as export workers.
If restaurant meals are assumed to be local goods, all the employees of restaurants
would be counted as local workers. The problem with this crude method is that some
mittens may be consumed locally and some meals may be eaten by nonresidents. In
other words, since few industries are exclusively exporters or local producers, it is
inappropriate to classify them as such.
Location-Quotient Approach. The location-quotient approach recognizes that

each industry produces partly for export and partly for local consumption. The trick
is to estimate what portion of each industry's output is exported. An industry's true
location quotient is
Mitten production in the citv

L >
= ^T
Mitten
~
consumption
— u
in the city
(6 ~ 7)
A quotient of 1.0 (the city's mitten production equals its mitten consumption) would
indicate that all of the city's production is consumed within the city, so none of the
mitten workers produce for export. If" Z, i
= 5.0. mitten production is five times local
consumption, so only one-fifth of mitten workers produce for the local market, and
four-fifthsproduce for export.
Data limitations prevent the direct estimation of the true location quotient. In-
stead, a proxy for the true location quotient is
City's mitten employment

City's totalemployment
2
—
Nation's mitten employment
Nation's total employment
The denominator (the share of national employment in mitten production) provides

a measure of how much local production is needed to satisfy the local demand for
mittens. For example, if percent of national employment is in mitten production
1
(the denominator is 0.01 ). the city is assumed to need percent of its workforce to 1
satisfy its local mitten demand. If the city actually employs 6 percent of its workers
in mitten production (if the numerator is 0.06). one-sixth of the workers are assumed
to produce for local consumption, and five-sixths are assumed to produce for export
There are three problems with the location-quotient approach. The first two are
related to the choice of the national economy as the reference point.
1. Uniform consumption patterns. The approach assumes that every city in

the nation has the same per capita demand for mittens. Of course, mitten
consumption depends on weather, fashion, and the tastes for outdoor winter
sports (skiing, ice fishing, snow shoveling), [fa city's residents consume a
relatively large amount of mittens, the location quotient Li will
overestimate the number of export workers. Conversely, if city residents
consume a relatively small amount of mittens. /. : will underestimate the
number of export workers.

2. National self-sufficiency. Another assumption of the location-quotient
approach is that the nation neither imports nor exports mittens, so the share
of national employment in mitten production is the share required for local
self-sufficiency. If the nation exports mittens, the share of national
Chapter?) Urban Economic Growth 155
employment in mitten production will overstate the share required for

self-sufficiency, so the location quotient will underestimate the number of
export workers. Conversely, if mittens are imported, the location quotient
will overestimate the number of export workers.
3. Industry grouping. The location quotient is typically computed for a set of
consumer goods, not for a particular product. For example, a city might
compute a location quotient for the apparel industry, which includes
mittens, pants, dresses, and children's clothes. Suppose that all the city's
apparel workers produce mittens for export: no other apparel products are
produced in the city. Suppose that the mitten industry is large enough that
the city's location quotient for apparel is 1 .0: the city uses the same share of
its workforce in mitten production as the nation uses in apparel production.
The city's location quotient suggests that the city does not export apparel
products, when in fact it exports mittens and imports other apparel.
Grouping goods into broad categories causes the location quotient to
underestimate export employment.
The results of Tiebout (1962) suggest that location quotients systematically

underestimate export employment. He compared estimates of export employment
derived from location quotients to the estimates generated by direct surveys of man-
ufacturers. Figure 6-7 shows For example,
his results for the city of Indianapolis.
the location quotient for the food industry suggests that 24 percent of food workers
produce for export, while the direct survey suggests that 63 percent of food workers
Figure 6-7 Estimates of Export Employment from Surveys

and Location Quotients
too
80
—
60
« 40
20
Food Chemicals Primary Nonelectrical Transportation

metals machinery equipment
Estimate from survey

J
Estimate from location quotient
SOURCE: Charles M. Tiebout, The Community Economic Base Study (New York: Committee for Economic
Development. 1962).
156 Purt 1 Market Forces in the Development of Cities
produce for export. For primary metals, the location quotient suggests that none of
the workers in the industry produce for export, while the direct survey suggests that
99 percent of the workers produce for export. In general. Tiebout's results suggest
that location quotients perform poorly in estimating export employment.
Estimating the Multiplier. The second step in an economic base study is to

compute the city's employment multiplier. For each industry, the location quotient
generates an estimate of export employment. If T, is total employment in industry i
and L, is the industry's location quotient, export employment in the industry is
B,
Li -
= -Lj — 1
T, (6-9)
For example, if L, is 4.0 and T

t
is 800. 5, is 600; three-fourths of the industry's work-
ers produce for export. The employment is estimated by summing
city's total export
the export employment across individual industries. The employment multiplier is
total employment divided by export employment.
Projecting Growth in Export Industries. The third step in an economic base

study is to project the growth of the basic industry, that is. to estimate AB in equa-
tion (6-6). Given the uncertainties associated with predicting future events, export
projection is more of an art than a science. One approach is to estimate the ( 1 )
national trend in the demand for a particular good and (2) assess the city's relative
attractiveness to the firms producing that good. The city's economic forecaster could
estimate the national demand for mittens for the next decade, looking at the trends in
weather, winter sports, and fashion. If the city's relative attractiveness is not expected
to change, the city's mitten production should increase at the same rate as national
mitten consumption: a doubling of mitten consumption should double the city's
mitten production and employment. If the city's relative attractiveness increases
over time, its mitten production and employment should increase at a relatively fast
rate.
Using the Economic Base Study. Economic base studies are used to project pop-
ulation growth and the demand for public services. Suppose that a city projects
that its mitten industry will grow by 1.000 jobs. If the employment multiplier
is 2.5. the economic base approach suggests that total employment will increase by
2,500 jobs. If there are three residents per job, the population of the city will in-
crease by 7,500 people, increasing the demand for all public services. For example,
if there is one school-age child for every five residents, total enrollment in schools
will increase by 1.500 students.
Local firms use the employment projections from the economic base study to
predict the demands for their products. Suppose that the city's economic base study
projects an increase in population of 7.500 people. If it takes 3.750 people to support
a barber, there will be room for two more barbers in the city. If the city grows by
7.500 people per year and it takes 30,000 people to support a shopping center, there
will be enough demand to support a new shopping center in four years.
( 'hapter 6 Urban Economic Growth 157
Local governments also use economic base studies to guide their growth-
management policies. If a local government can affect the sales and employment
of its basic industry, it can control the city's population. Suppose that the city has
a growth target of 15,000 additional people: it wants to increase its population by
15,000 people in the next 20 years. If there are three residents per job, the city will
meet its population target if it allows an additional 5,000 jobs. If the employment
multiplier is 2.5, the city will meet its employment and population targets if it allows
2,000 additional export jobs.
Input-Output Analysis
An alternative to an economic base study is an input-output study. Unlike the

economic base complete accounting of
study, the input-output study generates a
the transactions between firms and households in the economy. This approach has
two advantages over the economic base study. First, instead of assuming that every
industry has the same multiplier, it derives a multiplier for each export industry.
Second, instead of using the location-quotient approach to guess the volume of
exports, it measures exports directly.
To explain the workings of an input-output study, consider a city that produces
and consumes three goods: computers, electrical wire, and local consumer goods.
Computers and wire products are produced for both export and local consumption.
By definition, local merchants (restaurants, dry cleaners, grocery stores) produce
goods for local consumption.
Transactions Table and Input Coefficients. Table 6-2 shows the transactions
economy. The first column of numbers shows the input usage of firms
in the city's
in the computer industry. To produce computers, they buy $400 of electrical wire,
$1,000 of labor from city residents, and $600 of imported inputs (raw materials
and intermediate inputs). The sum of these input costs equals the total output of
the computer industry ($2,000). The next column shows the input usage of wire
Table 6- -2 Transactions Table from Input-Output Study
Producers
Computer Wire Local

Inputs Firms Producers Merchants Households Exports Total
Computers $ 300 $ 150 $ 180 $1,370 $2,000

Wire $ 400 600 1,000
Local 2,500 2,500
Labor 1,000 600 2,000 3,600
Imports 600 100 350 920 2.170
Total $2,000 $1,000 $2,500 $3,600 $1,970

Table 6-3 Input Coefficients Table
Producers
Computet Wire Local

Inputs Firnv, Producers Merchants Households
Computers 0.00 0.30 0.06 0.05

Wire 0.20 0.00 0.00 0.00
Local 0.00 0.00 0.00 0.69
Labor 0.50 0.60 0.80 0.00
Imports 0.30 0.10 0.14 0.26
Total 1.00 1.00 1.00 1.00
producers, who use computers, labor, and imported inputs to produce $1,000 of
wire. input usage of local merchants, who use computers,
The third column shows the
labor,and imports to produce $2,500 of output. The household column shows how
households divide their total income ($3,600) between computers ($ 1 80). local goods
($2,500), and imports ($920). The export column shows the exports of computers
($1,370) and wire ($600). The total column shows the sum of the items in each row.
Note that the figures in the total column match the figures in the total row.
Table 6-3, whichis derived from Table 6-2, provides a summary of the interac-
tions between firms and households. The first column of numbers shows the input
coefficients for the computer industry: for every dollar worth of computer produc-
tion, the computer industry uses 20 cents worth of input from the wire industry.
50 cents worth of labor, and 30 cents worth of imported materials (e.g., silicon chips
and plastic). The second and third columns show the input coefficients for wire
producers and local merchants. The fourth column shows that households spend
5 percent of their income on home computers. 69 percent on local goods, and 26
percent on imports.
The Multiplier Process. The information in Table 6-3 can be used to estimate
the multiplier effects of an increase in computer exports. Suppose that computer
exports increase by $ 00. The increase in computer sales will increase local spending,
1
precipitating a series of spending rounds. The first three spending rounds are shown in
Figure 6-8.
Round 1 The increase in computer sales increases wire production by $20

(0.20 times $100) and wages by $50 (0.50 times $100).
Round 2 The $20 increase in wire production increases computer sales by
an additional $6 (0.30 times $20) and wages by an additional
$12 (0.60 times $20).
Figure 6-8 Multiplier Effects of a $100 Increase in Computer Exports

Initial effect Round 1 Round 2 Round 3
$1.20 in wire
S6 in computers
$3.00 in labor
S20 in wire
SO. 60 in computers
SI 2 in wages
$8.28 in local
SI 00 in computer
exports
SO. 50 in wire
S2.50 in computers
SI. 25 in labor
S50 in wages
$1.73 in computers
$34.50 in local
$27.60 in wages
The first-round increase in wages ($50) increases home computer

sales by $2.50 (0.05 times $50) and local consumption by $34.50
(0.69 times $50).
Round 3 The increases in computer sales, wages, and local sales cause
additional increases in wire sales, wages, computer sales, and local
consumption.
Because of the multiplier process, an increase in computer exports increases spending

on all three goods.
The spending and respending of income continues forever, but every spending
round is smaller than the previous one. The multiplier process eventually peters out
because both producers and consumers spend some of their budgets on imports.
As shown in the fifth row in Table 6-3 (Imports), there is a leakage of 30 percent
in computer production, 10 percent in wire production, 14 percent in local goods
production, and 26 percent from consumers. These leakages weaken the multiplier
process, causing each spending round to be smaller than the previous round.
The input coefficients from Table 6-3 can be used to compute spending multi-
pliers for the two export industries. For a description of the techniques used to derive
the spending multipliers, see the references at the end of the chapter. In Table 6-A,
the column shows the multiplier effects of a $1 increase in computer exports.
first
Total computer sales increase by $1.23; wire production increases by $0.25; and
local sales increase by $1.17. The total multiplier effect is the sum of the effects on
the three industries: a $1 increase in computer sales increases total sales by $2.65.
Because the wire industry has a smaller import leakage (10 percent compared to
30 percent for the computer industry), it has a larger multiplier (3.04 versus 2.65).
Table 6^4 Spending Multipliers

from Input-Output
Examples
Added Sales per Dollar

of Additional Export Sales of
Computers Wire
Computers 1.23 0.52

Wire 0.25 1.10
Local 1.17 1 .42
Total 2.65 3.04
Limitations of Economic Base and Input-Output Studies
This section discusses five defects that limit the applicability of the economic
base and input-output approaches. There are two technical problems with the two
approaches. First, the income and employment multipliers are assumed to be con-
stant, regardless of city size. Second, the city's wage is assumed to be constant,
regardless of city size. In addition, these approaches are misleading for three rea-
sons. First, they suggest that the only way for a city to grow is by increasing its
exports. Second, they suggest that the only way to increase total employment is by
increasing the demand for labor. Third, they suggest that the city's fate is in the hands
of outsiders.
Constant Multipliers. The multipliers derived from both the input-output and
economic base approaches are assumed to be constant. In fact, a city's multipliers
can change for three reasons.
1 City size and consumer products. Multipliers increase with city size,
reflecting the ability of a larger city to support a wider variety of firms. As

explained in Chapter 5, a larger city has enough demand to support
activities with scale economies that are large relative to per capita demand.
For example, if a city grows to the point at which it can support its own
Mexican restaurant, city residents will patronize the local restaurant instead
of traveling to another city. A larger fraction of increases in income will be
spent locally, so the income and employment multipliers will increase.
2. City size and intermediate inputs. As the city grows, the demand for
intermediate inputs grows, and more of these inputs will be provided
locally instead of being imported. For example, if the computer industry
grows to the point at which the city can support its own silicon chip maker,
computer firms will spend less on imported inputs and more on locally
produced inputs. Therefore, a given increase in computer exports will
generate a larger increase in total income.
3. Changes in input prices. Changes in input prices cause factor substitution

that changes the interactions among the economic sectors. Suppose that the
city's wage increases, causing computer firms to substitute imported capital
equipment for laborers. If the input coefficient for labor decreases from
0.50 to 0.40, only $40 of a $100 increase in computer exports will be paid
to laborers. Because a smaller fraction of export income will be spent
locally, the multiplier effect of an increase in exports will decrease.
Constant Wage. The economic base and the input-output approaches assume
that the city's wage is fixed, regardless of city size. In other words, they estimate
the horizontal shift of the city's labor-demand curve, not the equilibrium change
in employment. Suppose that an economic base study computes a multiplier of
2.5 and projects an increase in export employment of 10,000 jobs. Based on these
numbers, the economic base approach would predict that total employment in the
city would increase by 25,000 jobs. This is incorrect because the city's labor-supply
curve is positively sloped, so an increase in labor demand increases the city's wage.
In Figure 6-3, total employment increases by only 16,000 jobs, not 25,000 jobs.
Because the economic base and the input-output approaches assume a fixed wage,
they overestimate the stimulative effect of increases in exports.
Focus on Exports as Source of Economic Growth. The economic base and

input-output approaches suggest that the only way grow is by increasing
for a city to
its exports. The notion that economic growth requires an increase in exports is
incorrect. Growth may also occur as a result of (1) a decrease in imports, (2) an
increase in labor productivity, or (3) an increase in trade within the metropolitan area.
A decrease in imports has the same stimulative effect as an increase in exports.
While an increase in exports increases the flow of money into the urban economy, a
decrease in imports decreases the flow of money out of the economy. The economy
can be stimulated by either increasing the inflow (exports) or decreasing the outflow
(imports). To explain the notion of import substitution, suppose that a city initially
spends $10 million per year on imported chairs. If consumers switch to locally
produced chairs, the $10 million that previously left the region will now be paid to
local producers and will be spent and respent in the local economy. The spending and
respending of the $10 million has the same effect as a $10 million increase in export
sales. If the spending multiplier is 2.5. the change in total income resulting from
import substitution is $25 million. The possibility of import substitution suggests
that economic development policies should consider decreasing imports as well as

increasing exports.
The most compelling counterexample to the export-oriented theories of urban
economic growth is the earth economy. The earth's economy has grown, despite
the absence of exports to other worlds. Economic growth has occurred for two rea-
sons. First, technological advances have increased output per person, increasing real
income per capita. Second, changes in production and transportation technologies
have increased trade between regions. Trade increases real income because it allows
each region to specialize in the production of goods for which it has a comparative
advantage. The same phenomena occur at the urban level: increases in labor produc-
tivity or trade within a metropolitan area increase per capita income, so the city's
economy can grow without increasing exports.
The Demand Side of the Urban Labor Market. Both the economic base and the
input-output approaches focus attention on the demand side of the urban labor mar-
ket. The idea is that growth occurs when the demand for the city's labor increases.
As explained earlier in the chapter, an increase in labor supply also increases equi-
librium employment, so growth can occur as a result of increases in either demand
or supply. There are a number of public policies that shift the supply curve and
increase equilibrium employment, including environmental policy (see Figure 6-5)
and infrastructure policy (see Figure 6^4).
Do Outsiders Determine the City's Economic Fate? The economic base and
input-output approaches incorrectly suggest that the city's economic fate is in the
hands of outsiders. In fact, there are several local public policies that can be used
to increase the city's equilibrium employment. On the demand side, the city can
shift the demand curve to the right by ) cutting its business taxes. (2) improving
( 1
its industrial infrastructure, or (3) increasing labor productivity by improving its

educational system. On the supply side, the city can shift the supply curve to the
right by (1) cutting its residential taxes, (2) improving its residential infrastructure.
(3) improving its residential services such as schools, recreation, and public safety.
or (4) improving the local environment.
It's important to note that public policy need not focus exclusively on increasing
the city's exports. As explained earlier, among the alternative strategies are import
substitution and increased trade within the metropolitan area. To the extent that public
policy encourages import substitution or intracity trade, it can increase employment
and income without increasing the city's exports.
Benefits and Costs of Employment Growth

This section of the chapter explores the trade-offs associated with increases in city
size, addressing two questions about the effects of employment growth on the welfare
of a city's residents. First, how many of the new jobs generated by economic growth
are filledby newcomers, and how many are filled by original residents of the city who
would otherwise not be employed? Second, how does an increase in total employment
affect real income per capita? As explained earlier in the chapter, changes in nominal
wages and prices usually offset one another, so the real wage for a given occupation
is usually unaffected by an increase in total employment. This section discusses
three other factors that affect real income: the unemployment rate, the labor-force
participation rate, and occupational rank.
Table 6-5 Long-Run Effects of Employment Growth

on Employment Rate
Initial Long-Run
Equilibrium Equilibrium Change
Unemployment rate («) 5.40% 5.33% -0.07%

Participation rate (p) 87.50% 87.64% 0.14%
Employment rate [e =(/—«) p] 82.78% 82.97% 0.19%
Potential workforce 120,809 121,737 928
Participating 105,708 106,687
Unemployed 5,708 5,687
Working 100,000 101.000 1,000
Not working 20.809 20.737
New jobs filled by newcomers 770
New jobs filled by original residents 230
Filled by unemployed 70
Filled by nonparticipants 160
SOURCE: Calculations based on Table 4.5 in Timothy J. Bartik, Who Benefits from State and Local
Economic Development Policies'.' (Kalamazoo. Mich.: Upjohn Institute, 1991 ).
Who Gets the New Jobs?

Bartik (1991) studied the effects of employment growth on unemployment rates,
labor-force participation rates, and migration rates 89 metropolitan areas.

in
Table 6-5 uses his results to show the effects of employment growth on the charac-
teristics of a hypothetical metropolitan area.
The first column of numbers shows the characteristics of the city's labor market
in the initial equilibrium. In the sample of cities examined by Bartik, the average
unemployment rate (the percentage of people who were unsuccessful in their search
forwork) was 5.40 percent, and the average labor-force participation rate (the per-
centage of working-age people who were actively involved in the labor market,
eitherworking or looking for work) was 87.50 percent. Suppose that the employ-
ment rate is defined as the percentage of working-age people who are employed.
The employment rate is computed by multiplying the participation rate by minus 1
the unemployment rate. In Bartik's study, the average employment rate was 82.78
percent. The hypothetical city has a potential workforce of 120,809 people; given
the participation rate (87.50 percent), 105,708 participate in the labor market; given
the unemployment rate of 5.40 percent, 5,708 of the participants are unemployed,
leaving 100,000 people who are employed.
The second column of numbers in Table 6-5 shows the predicted long-run effects
of a 1 percent increase in total employment (1,000 jobs). According to Bartik, the
long run is about six years: it takes about six years for all the markets to fully adjust
to the change in employment. Six years after the increase in total employment, the
unemployment rate is 5.33 percent, the participation rate is 87.64 percent, and the
employment rate is 82.97 percent. Migration to the city increases the city's potential
workforce by 928 people.
How many of the new jobs are filled by newcomers, and how many are filled
by original residents who would otherwise not be employed? Since there are 1,000
new jobs and only 928 new potential workers, there are at least 72 new jobs for the
original residents. If all the newcomers are employed (newcomers have a 100 percent
participation rate and a zero unemployment rate), there will be only 72 new jobs for
the original residents. If the newcomers have the same employment rate as the origi-
nal residents, however, only 770 of the newcomers will be employed (82.97 percent
of 928). so there will be 230 jobs left for the original residents. About a third of the
230 jobs will be filled who were previously unemployed, and
by original residents
the remaining two-thirds will be filled by original residents who previously did not
participate in the labor market. Figure 6-9 summarizes the results from Table 6-5:
77 percent of the new jobs are filled by newcomers, leaving 7 percent for original
residents who were unemployed and 16 percent for original residents who did not
participate in the labor market.
The simple lesson from Table 6-5 and Figure 6-9 is that employment growth
causes in-migration and population growth. There are three implications from this
lesson. First, if the city's workforce is just as qualified for the new jobs as workers in
other cities who have the option of migrating to the city, the original residents will
get about a quarter of the new jobs. Second, if the city's workers are. on average,
less qualified for the new jobs than workers in other cities, more of the jobs will
be by workers migrating to the city. Third, employment growth increases
filled
population and thus increases the demands for housing, land, and public services.
This suggests that local governments should coordinate their economic development
policies with their policies concerning land use. transportation, and infrastructure
investment.
Figure 6-9 Distribution of 1,000 New Jobs between Original

Residents and Newcomers
Newcomers: 770 jobs
who were
Original residents
unemployed: 70 jobs
Original residents who did not

participate: 160 jobs
Employment Growth and Real Income per Capita

How does employment growth affect a city's real income per capita? An increase in
total employment could, in principle, increase per capita income in several ways:
1 Increase in the real wage for each occupation. As explained earlier in the
chapter, employment growth causes offsetting changes in nominal wages

and living costs, so the real wage for a given occupation is unaffected by
employment growth.
2. Promotions. Bartik ( 1991 ) shows that employment growth hastens
workers' movement upward in the job hierarchy. It appears that an increase
in the demand for labor causes firms to promote workers to higher-paying
jobs more rapidly. Figure 6-10 shows, for different types of workers, the
average percentage movement up the job hierarchy (and up the pay scale)
from a percent increase in total employment. The average worker is
1
promoted to a job with a wage that is 0. 16 percent higher than the wage for
the old job. A less-educated worker (3 years less education than average) is
promoted to a job that pays 0. 196 percent more; a young worker (12 years
younger than average) is promoted to a job that pays 0.188 more; a black
worker is promoted to a job that pays 0.215 percent more.
Figure 6-10 Effect of Employment Growth on Occupational Rank

for Different Groups
0.250
r3 S 0.200 —
n. c
•-
D 0.150
O 1>
3 S 0.100
0.050
o-
£
0.000
Black
SOURCE: Timothy J. Bartik, Who Benefits from Slate and Local Economic Development Policies? (Kalamazoo.
Mich.: Upjohn Institute, 1991).
Figure 6-11 Effect of Employment Growth on Real Earnings per Capita

for Different Groups
0.500
a o
"5- 0.400
u
& 6
Q. u
u —
s 1
o 2
0.300
0.200 —
0.100
P E
0.000
Average Less Younaer Black
educated
SOURCE: Timothy J. Bartik. Who Benefits from State and Local Economic Development Policies'.' (Kalamazoo,
Mich.: Upjohn Institute. 1991).
3. Increase in employment rate. As explained earlier, employment growth

decreases the unemployment rate and increases the participation rate, so it
increases the fraction of the working-age population that is employed. In

Table 6-5. the 1 percent increase in employment increases the employment
rate from 82.78 percent to 82.97 percent.
Figure 6-1
1 shows the combined effects of changes in real wages, occupational
rank,unemployment rates, and participation rates on real income per capita. For the
average household, a percent increase in employment increases real income per
1
capita by 0.40 percent. In other words, the elasticity of real earnings with respect to
total employment is 0.40. The most important factors in the increase in income are
the increases in wages from being promoted to higher-paying jobs and the increases
in labor-force participation. The elasticities are larger for households that are less
educated, young, or black, because workers in these groups experience relatively
large benefits from promotion to higher-paying jobs.
Other Benefits and Costs of Employment Growth

There are a number of other benefits and costs from employment growth. On the
benefit side, a larger city is able to exploit scale economies in the production of
consumer goods and some public services. As explained in Chapter 5, a large city
can support have relatively large scale economies, so a larger city has
activities that
a wider variety of goods and services. In addition, there are scale economies in the
provision of some local public goods, so an increase in population decreases the
average costs of some public goods. On the cost side, a large city also has relatively
high commuting costs: as the land area of the city increases, the average commuting
distance increases, increasing the average commuting cost. In addition, larger cities
often have greater problems with pollution and traffic congestion. Finally, large cities
typically have higher crime rates.
It's important to note that the problems of congestion, pollution, and crime are
not caused by city size per se. They are caused by market failures that become worse
as the city grows. As explained later in the book, there are a number of policies that
address these problems directly. Therefore, the problems can be solved (or decreased
to their optimum levels) without decreasing city size.
Composition of Changes in Employment

This section of the chapter examines the composition of changes in total employment.
A city experiences an increase in total employment when the number of jobs gained
from the opening of new plants and the expansion of existing plants exceeds the
number of jobs lost from the closure and contraction of old plants. The question is
whether cities that experience rapid employment growth have relatively large gains
from plant openings and expansions or relatively small losses from plant closures
and contractions. Conversely, do cities that grow slowly or shrink have relatively
small gains from plant openings and expansions or relatively large losses from plant
closures and contractions?
Eberts and Stone ( 1 992) studied the composition of employment changes for 34
large metropolitan areas. In Table 6-6, the first column shows, for each city, the net
change employment as a percent of total employment. For example, Pittsburgh lost

in
8.19 percent of its total employment over this period. The second and third columns
show the gains from plant openings and expansions, and the sixth column shows
the total gross gain (the sum of gains from openings and expansions). For example,
Pittsburgh experienced an 11.46 percent gain from openings and a 5.82 percent
gain from expansions, for a total gross gain of 17.28 percent. The fourth, fifth, and
seventh columns show the losses from plant closures and contractions. For example,
Pittsburgh experienced a 3.76 percent loss from closures and an 11.71 percent loss
1
from contractions, for a total gross loss of 25.47 percent.

Three conclusions can be drawn from the data in Table 6-6. First, there is a
large range of changes in net employment, from —8.19 percent for Pittsburgh to
+ 19.36 percent for San Diego. Second, the gross changes in employment (gains and
losses) are large relative to the net changes. The average change from openings and
expansions is over 26 percent, and the average change from closures and contractions
is over 20 percent, compared to an average net change of about 6 percent. The fact
that a city has a relatively small net change in employment does not mean that its
economy is stagnant, but simply that the employment gains are close to the losses.
For example, the five cities that had positive net changes less than percent (Akron,
1
New York, Cleveland, New Orleans, and Detroit) had gross gains and losses between
20 and 23 percent.
Table 6-6 Employment Changes from Plant Openings, Expansi ons, Closings,
and Contractions, 1984-1986
Net
Employment Percentage Employment Change Originating from Percent Percent
Metropolitan Change from from
Area (percent) Openings Expansions Closings Contractions Open + Expand Close + Contract
Pittsburgh -8.19 11.46 5.82 -13.76 -11.71 17.28 -25.47

Houston -5.41 16.75 7.54 -20.91 -8.80 24.29 -29.71
Akron 0.46 16.20 6.43 -13.99 -8.18 22.63 -22.17
New York 0.61 12.64 8.33 -15.38 -4.98 20.97 -20.36
Cleveland 0.64 13.31 8.48 -15.17 -5.98 21.79 -21.15
New Orleans 0.86 15.41 6.03 -13.94 -6.65 21.44 -20.59
Detroit 0.97 13.20 8.96 -15.40 -5.80 22.16 -21.20
Miami 2.18 15.48 8.72 -16.39 -5.63 24.20 -22.02
San Francisco 2.26 13.48 8.81 -14.27 -5.76 22.29 -20.03
Buffalo 2.57 15.36 7.26 -14.63 -5.42 22.62 -20.05
Cincinnati 3.16 13.27 8.25 -13.54 -4.82 21.52 -18.36
Los Angeles 3.95 16.78 9.78 -17.10 -5.52 26.56 -22.62
Kansas City 4.28 16.26 8.24 -14.76 -5.46 24.50 -20.22
Milwaukee 4.36 13.57 7.96 -11.71 -5.45 21.53 -17.16
Portland 5.43 20.12 8.14 -13.15 -9.68 28.26 -22.83
Chicago 5.54 16.31 8.83 -13.67 -5.92 25.14 -19.59
Dallas 5.68 20.03 10.44 -16.68 -8.10 30.47 -24.78
Denver 6.16 20.18 8.87 -15.16 -7.72 29.05 -22.88
Philadelphia 6.16 15.54 10.01 -13.87 -5.52 25.55 -19.39
Anaheim 6.88 19.44 11.42 -18.26 -5.72 30.86 -23.98
Newark 7.18 16.36 8.88 -12.61 -5.45 25.24 -18.06
San Jose 7.50 19.19 10.74 -16.21 -6.21 29.93 -22.42
St. Louis 8.19 16.48 8.62 -11.72 -5.20 25.10 -16.92
Tampa 8.68 20.73 9.89 -17.25 -4.70 30.62 -21.95
Indianapolis 8.70 17.64 10.09 - 1 1 .05 -7.98 27.73 -19.03
Birmingham 8.73 21.12 10.25 -15.51 -7.13 21.37 -22.64
Baltimore 8.93 19.00 10.82 -15.62 -5.27 29.82 -20.89
Seattle 11.08 21.58 10.02 -13.87 -6.65 31.60 -20.52
Greensboro 13.24 17.21 9.34 -8.73 -4.57 26.55 -13.30
Minneapolis 15.29 19.42 11.22 -9.82 -5.53 30.64 -15.35
Atlanta 15.58 22.18 12.61 -14.65 -4.55 34.79 -19.20
Columbus 15.90 20.47 9.74 -9.56 -4.75 30.21 -14.31
Rochester 19.33 29.66 7.13 -12.12 -5.35 36.79 -17.47
San Diego 19.36 27.32 12.93 -15.63 -5.26 40.25 -20.89
Average 6.36 17.74 9.14 -14.30 -6.22 26.88 -20.52
SOURCE: Table 2.3 in Randall W. Eberts and Joe A. Stone. Wage and Adjustment in Local Labor Markets (Kalamazoo.
Mich.: Upjohn Institute. 1992).
Figure 6-12 Gross Gains and Losses in Employment in Selected Cities,

1984-1986
45 —
40 -
35 -
30-
$ 25 - — — —— - — — - -- -- --
u -
ft 20 -
15 - — --
10 -
5 -
I Openings and expansions
Closings and contractions
SOURCE: Table 2.3 in Randall W. Eberts and Joe A. Stone. Wage and Adjustment in Local Labor Markets
(Kalamazoo. Mich.: Upjohn Institute. 1992).
The third conclusion from Table 6-6 is that rapidly growing cities have relatively
large employment gains (from openings and expansions), not relatively small losses
(from closures and contractions). Figure 6-12 shows the gross gains and gross losses
for every third city in Table 6-6, arranged in ascending order of net employment
changes. There is very little variation in the gross loss rates across cities. Except for
Pittsburgh, all the cities have gross losses between 19 percent and 22.4 percent. In
contrast, the gross gains vary considerably across cities, from about 1 7 percent to
about 40 percent. The keys to rapid growth are openings and expansions: the three
most rapidly growing cities (San Diego, Atlanta, and Seattle) have about the same
gross losses as some of the slowest-growing cities (Pittsburgh, New York, Detroit,
and Buffalo), but much larger gross gains.
Summary
1 Urban economic growth is represented by various changes in the urban labor
market.
a. The demand for labor comes from exporters and local producers. An
increase in export sales increases local sales through the multiplier process.
The employment multiplier is the change in total employment per unit
change in export employment.
b. An increase in demand increases total employment and the city's wage.
The demand curve shifts as a result of changes in the cost of producing
exports and local goods.
c. The supply of labor comes from city residents. The supply curve is
positively sloped because an increase in city size increases housing and

land prices, forcing firms to pay higher wages to compensate for the
increased living costs. An increase in supply (caused by improvements in
environmental quality or public services) increases total employment and
decreases the wage.
2. Economic growth generates benefits and costs. The larger city can exploit
scale economies in production, retailing, and public goods. On the cost side,
the larger city has higher housing prices, longer commuting distances, and
more pollution, congestion, and crime.
3. Public policy can stimulate economic growth by decreasing the production
costs of exporters and local producers. Direct policies include public education
and training programs (which increase labor productivity) and the efficient
provision of public services and infrastructure. Indirect policies improve the
quality of life of the city, attracting workers and decreasing wages.
4. Some cities try to attract firms by offering subsidies (tax abatement, industrial
bonds, government loans, and subsidized site development).
a. Because taxes support local public services, a tax cut may decrease
spending on public services, making the city less attractive to firms.
b. Subsidy programs that attract new firms often generate relatively small
benefits for local workers because many of the new jobs are filled by
newcomers.
c. Subsidy programs often promote the growth of one city at the expense of
another, causing inefficient location choices.
5. A may increase total employment. The

pollution-abatement program
abatement program decreases labor demand and increases labor supply. If
people are relatively sensitive to pollution, the increase in supply will be large
relative to the decrease in demand, and the city will grow.
6. An economic base study predicts trends in urban income and employment. It is
based on the notion that a city must earn its livelihood by exporting. An
economic base study has three steps:
a. Estimate the amount of export employment. Location quotients provide
inaccurate estimates of export employment.
b. Compute the multiplier, equal to the ratio of total employment to export

employment.
c. Project growth in export sales or employment.
7. Economic base by local governments to predict the demands
studies are used
for local publicgoods (schools, sewerage, fire protection) and by firms to
predict the demands for private goods (groceries, shoes).
8. An input-output study is more sophisticated than the economic base study. It is
based on a complete accounting of the economy's transactions, so more it
accurately measures export and local activities. Input coefficients summarize

the interactions between different sectors and are used to derive income
multipliers for each industry.
9. The economic base study and the input-output study share a number of defects
that limit their applicability.
a. The assumed to be constant. The actual multiplier is likely
multipliers are
to increase as the citygrows and is likely to change with factor prices.
b. The two approaches assume that the city's wage is fixed. They estimate the
horizontal shift of the city's demand curve, not the change in equilibrium
employment.
c. Both approaches focus attention on the demand side of the economy,
suggesting that the city's economic fate is determined by outsiders. In fact,
local governments can stimulate growth through their provision of public
services.
10. Employment growth increases real income per capita by (a) hastening the
move up the job hierarchy and (b) increasing the labor-force participation rate.
The elasticity of real earnings with respect to total employment is 0.40. The
elasticity is larger for households that are less educated, young, or black,
because such households experience relatively large benefits from being
promoted to higher-paying jobs.
1 1. Rapidly growing cities have relatively large employment gains (from openings
and expansions), not relatively small employment losses (from closures and
contractions).

1. Suppose that a consulting firm has generated the following information about
the economy of Growville:
i. The current employment in export industries is 50,000.
it The current total employment in the city is 150,000.
iit Export employment is expected to grow by 10,000 jobs.
a. Is there enough information to accurately predict the effect of the increase

in export employment on total employment?
b. If you have enough information, predict the employment effect and
illustrate your answer with a graph.

c. If there is insufficient information, proceed with the analysis as far as you

can and list the additional information you need to complete the analysis.
Illustrate your answer with a graph.
2. Consider a self-contained city, that is. a city in which all workers live in the
city.The initial equilibrium wage is W°, and the initial equilibrium
employment is N .
a. Suppose that the demand for the city's exports increases. Use the
conventional supply-demand graph to show the effects of the increase in

export demand on the city's labor market. Label everything: the axes, the
curves, the new equilibrium wage \\"), and the new equilibrium total
(
employment (N').
b. Suppose that the city institutes a growth-control program that holds the
total housing stock (total square footage) at its initial level (before the
increase in export demand). same supply-demand graph to show
Use the
the effects of the increase in export demand under the growth-control
program. Label the equilibrium wage under the growth-control program
W* and the equilibrium total employment N*.
c. Explain the differences between the market equilibrium (a) and the
growth-control equilibrium (b).
3. Consider a city that uses a business tax to finance the provision of industrial
services (roads, water, sewers). If the city does not provide these services, the
individual firms must supply their own roads, water, and sewers. Suppose that
the city decides to cut business taxes and expenditures on industrial services
by the same amount. Evaluate the effects of the new tax-expenditure policy on
the city's labor market. Will total employment in the city increase or
decrease?
4. Consider two cities. Flexville and Rigid City, that have the same equilibrium
wage and the same equilibrium total employment. The export industries in the
two cities produce the same good. In Flexville, export firms produce with
variable factor proportions: the amount of labor per unit of output depends on
the relative price of labor. In Rigid City, export firms produce with fixed factor
proportions: the amount of labor per Suppose that both
unit of output is fixed.
cities find a way sen ices (for
to increase the quality of residential public
example, public schools, parks, libraries) without increasing taxes. Use two
graphs, one for each city, to show the effects of the improvements in public
services on (a) equilibrium employment and (/?) the equilibrium wage.
5. One of the assumptions lurking behind the city's labor-supply curve is that an
increase in the wage does not increase the fraction of the city's population in
the workforce. Draw two supply curves, one under the assumption of a fixed
participation rate and another under the assumption that the participation rate
increases with the wage. Explain any differences between the two supply
curves.
6. The section on environmental quality suggests that a pollution tax ma\

increase total employment. Suppose that instead of using a pollution tax to
reduce citywide industrial pollution by a total of 25 percent, the city requires

that each polluting firm decrease its pollution by 25 percent.
a. Is the uniform-reduction policy more or less efficient than the pollution-tax
program? In other words, will the total cost of abatement be larger or
smaller under the uniform-reduction policy?
b. Is the city more or less likely to grow under the uniform-reduction policy?
Demonstrate, using a graph.

7. Consider a city with the following characteristics:
i. The price elasticity of demand for the city's output is —3.0.
ii. The elasticity of labor supply with respect to environmental quality is
0.10.
Hi. Polluting firms produce with fixed factor proportions: labor per unit
output is fixed.
iv. Polluting firms initially employ 25 percent of the city's export workers.
Suppose that the city adopts a new environmental policy that increases its
environmental quality by 10 percent and increases the price of the polluting

industry by 2 percent.
a. Assume that the city's wage is fixed. Predict the percentage decreases in the
demand for workers in (/) the polluting industry and the export industry.(/'/')
b. By what percentage will the supply of labor increase?

c. Under what conditions (what values of the relevant elasticities) will the
equilibriumnumber of workers increase? Illustrate your answer with a
graph.
8. Consider a city that has decided to impose a pollution tax on its polluting
firms. Fill in the blanks to make the following statements correct, and then
briefly explain your responses. The city is more likely to grow as a result of the
pollution tax if
a. The elasticity of the supply of labor with respect to environmental quality is
relatively (the labor-supply curve shifts by a relatively

amount when environmental quality improves).
b. The elasticity of demand for labor (with respect to the wage) by the city's
nonpolluting industries is relatively
c. The elasticity of the supply of labor with respect to the wage is relatively
(the labor-supply curve is relatively ).
9. September 1989, a consulting firm estimated the

In a report issued in
economic impacts of moving the Los Angeles Raiders (a professional football
team) to Sacramento. The economic base study was based on the following
assumptions:
i. Total attendance at the Raider games will be 700.000 people per
year.
ii. The average ticket price will be $30.
Hi. The average fan will spend $10 on food, merchandise, and parking.
iv. Based on ii and Hi, the average fan will spend a total of $40.
v. Total "direct" spending will be $28 million per year ($40 •
700k).
Using a spending multiplier of 2.2. the consulting firm estimated the total
economic impact of the Raiders to be $61.6 million per year. Critically appraise
the methods used to compute the total economic impact of the Raiders.
10. There are two ways to increase total employment: increased exports and
import substitution. In what way and import substitution
are increased exports
similar? In what way are they dissimilar?
1 1 Consider the trade-offs associated with choosing a reference region for

defining a city's location quotient. The traditional approach [see equation
(6-8)] is to compare the city economy to the national economy. What if the
denominator of the location quotient was defined as the percentage of state
employment in mitten production? Would the measured location quotient be
closer to or farther from the city's true location quotient?

1 2. In Figure 6-2. the shift of the demand curve from Di to D;. which reflects the
multiplier process, is The implicit assumption is that the employment
parallel.
multiplier is constant. Draw the demand curves under the assumption that the
multiplier increases with city size.
13. Suppose that you have been given the following data on the transactions within
a city's economy:
Procedure
Computer Wire Local Total

Inputs Finns Producers Merchants Households Exports Output
Computers $ $350 $ S 300 S $2,000

Wire 500 500 1.000
Local goods 2.500
Labor 1.000 600 2.000 3.600
Imports 300 1 .650
Total $2,000 $ s S3.600 s S
a. Fill in the transactions table.

b. Will this city have a larger or a smaller export multiplier than the city with
the transactions listed in Table 6-2? Explain.
14. Suppose that wire production in the city with the input coefficients shown in
Table 6-3 increases by $100.
a. Compute the first, second, and third rounds of spending resultant from the
increase in wire sales, using the format of Figure 6-8 to report your results.
b. By how much does spending on w ire. computers, and local goods

eventually increase'.'
c. What is the ultimate increase in total sales?
15. Consider a city that estimates its export multipliers from the input coefficients
from a 1965 input-output study.
a. Will the city overestimate or underestimate the stimulative effects o\
exports?
b. If your answer to (a) is "It depends on ," make a list of questions that
. . .
you would ask Ms. Information, the most knowledgeable person in the city.
Note: She does not know the current input coefficients.
The Urban Labor Market

Bartik,Timothy J. Who Benefits from State and Local Economic Development Policies?
Kalamazoo, Mich.: Upjohn Institute, 1991 Chapters 5 and 6 explore the effects of
.
employment growth on housing prices and real wages.

Eberts, Randall W., and Joe A. Stone.Wage and Adjustment in Local Labor Markets.
Kalamazoo, Mich.: Upjohn Institute, 1992. Discusses how local labor markets adjust to
shifts in supply and demand.
Hamilton, Bruce, and Robert Schwab. "Expected Appreciation in Urban Housing Markets."
Journal of Urban Economics 18 (1985), pp. 103-18. Estimates the relationship
between city size and housing prices.
Henderson, J. V. Urban Development: Theory, Fact, and Illusion. New York: Oxford
University Press, 1988. Chapter 6 estimates the relationship between wages and city
size, environmental quality, weather, and crime rates.
Robak, Jennifer. "Wages, Rents, and the Quality of Life." Journal of Political Economy
90 (1982), pp. 1257-78. Estimates the relationship between wages and city size.
Rosen, Sherwin. "Wage-Based Indexes of Urban Quality of Life." In Current Issues in
Urban Economics, ed. PeterMieszkowski and Mahlon Straszheim. Baltimore: Johns
Hopkins, 1979. Estimates the relationship between wages and city size, environmental
quality, weather, and crime.
Treyz, George, and Benjamin Stevens. "The TFS Regional Modelling Methodology."
Regional Studies 19 (1985), pp. 547-62. Estimates the relationship between wages and
city size.
Public Policy and Economic Growth

Bartik, Timothy J. Who Benefits from State and Local Economic Development Policies'.''
Kalamazoo, Mich.: Upjohn Institute, 1991. Chapter 2 discusses the effects of various
public policies on economic development.
Harrison, Bennett, and Sandra Kanter. "The Political Economy of Job-Creation Business
Incentives." Journal of the American Institute of Planners 44 (October 1978),
pp. 424-35.
Levy. John M. Economic Development Programs for Cities, Counties, and Towns.
New York: Praeger Publishers, 1985. Detailed discussion of economic development
policies.
Peltz, Michael,and Marc Weiss. "State and Local Government Initiative for Economic
Development through Technological Development." Journal of the American Planning
Association 50 (1984), pp. 270-79.
Rasmussen, David W.; Marc Bendick; and Larry Ledebur. "Evaluating State Economic
Development Incentives from a Firm's Perspective." Business Economics 10 (1982),
pp. 23-29.
Schmenner. Roger W. Making Business Location Decisions. Englewood. N.J.: Prentice

Hall, 1982. Discusses how
make location decisions.
firms
Senia. Al. and George Weimer. "The War between the States for High Technology." Iron
Age, September 5, 1983. Discusses competition among states for high-technology
industries.
City Size and Environmental Policy

Henderson. J. Vernon. "Effect of Taxation of Externalities on City Size." In Urban Growth
Policy in a Market Economy, ed. George S. Tolley. Philip E. Graves, and John L.
Gardner. New York: Academic Press. 1979, pp. 91-97.
Izraeli.Oded. "Externalities and Intercity Wage and Price Differentials." In Urban Growth
Policy in a Market Economy, ed. George S. Tolley. Philip E. Graves, and John L.
Gardner. New York: Academic Press. 1979. pp. 159-94.
Infrastructure and Economic Growth

Aschauer. David A. "Rx for Productivity: Build Infrastructure." Federal Resene Bank of
Chicago Letter 13 (September 1988). Suggests that underinvestment in public
infrastructure during the 1970s and 1980s caused productivity to grow at a relatively
slow rate.
Blinder. Alan. "Are Crumbling Highways Giving Productivity a Flat?" Business Week,
August 1988. Discusses Aschauer's study of the connection between productivity
growth and investment in public infrastructure.
Eberts, Randall W.. and Joe A. Stone. Wage and Adjustment in Local Labor Markets.
Kalamazoo. Mich.: Upjohn Institute, 1992. Chapter 5 estimates the effect of changes in
public infrastructure on the urban labor market.
Economic Base Theory and Input-Output Analysis

DiPasquale. Denise. and Karen Polenske. "Output. Income, and Employment Input-Output
Multipliers." In Economic Impact Analysis: Methodology and Applications, ed. Saul
Martinus Nijhoff. 1980. pp. 85-1 13.
Pleeter. Boston:
Hoover, Edgar M.. and Frank Giarratani. An Introduction to Regional Economics.
New York: Alfred A. Knopf, 1985, Chapter 11. A discussion of economic forecasting
techniques (economic base analysis and input-output analysis).
Isserman, Andrew M. "Estimating Export Activity in a Regional Economy: A Theoretical
and Empirical Analysis of Alternative Methods." International Regional Science
Review 5 1980), pp. 155-84. Discusses location quotients and other methods for
(
measuring export employment.

Miernyk, William H. The Elements of Input-Output Analysis. New York: Random House,
1965. Describes the nuts and bolts of input-output analysis.
Pleeter. Saul. "Methodologies of Economic Impact Analysis: An Overview." In Economic
Impact Analysis: Methodology and Applications, ed. Saul Pleeter. Boston: Martinus
Nijhoff. 1980. pp. 7-31.
Richardson, Harry W. Input-Output Analysis and Regional Economics. New York: John
Wiley &
Sons, 1972, pp. 1-52.
Tiebout, Charles M. The Community Economic Base Study. New York: Committee for
Economic Development. 1962.
Critique of Economic Base and Input-Output Approaches

Chinitz, Benjamin. "Contrasts in Agglomeration: New York and Pittsburgh." In Readings in
Urban Economics, Matthew Edel and Jerome Rothenberg. New York: Macmillan,
ed.
1972. A discussion of the supply side of the urban economy.
North, Douglass C. "Location Theory and Regional Economic Growth." In Regional Policy:
Readings in Theory and Application, ed. John Friedmann and William Alonso.
Cambridge, Mass.: MIT Press, 1975. States the case for the export-driven model of
regional growth.
Tiebout, Charles M. "Exports and Regional Economic Growth." In Regional Policy:
Readings in Theory and Application, ed. John Friedmann and William Alonso.
Cambridge, Mass.: MIT Press, 1975. Critiques the standard export-driven growth
model.
Costs and Benefits of Employment Growth

Bartik,Timothy J. Who Benefits from State and Local Economic Development Policies?
Kalamazoo, Mich.: Upjohn Institute, 1991. Chapters 4 and 7 discuss the effects of
employment growth on employment rates and real income per capita.
Mishan, E. J. The Costs of Economic Growth. London: Staples Press, 1967. Discusses the
costs and benefits of regional growth.
Summers, Gene. The Invasion of Non-Metropolitan America by Industry: A Quarter
Century of Experience. New York: Praeger Publishers, 1976. Describes the effects of
new plants on small towns.
Composition of Changes in Employment
Armington, Catherine, and Marjorie Odle. "Small Business How — Many Jobs?" Brookings
Review, Winter 1982. pp. 14-17. A critique of Birch's work.
Birch, David. "Who Creates Jobs?" The Public Interest, Fall 1981. Discusses the role of
small business in economic development.
Eberts. Randall W., and Joe A. Stone.Wage and Adjustment in Local Labor Markets.
Kalamazoo, Mich.: Upjohn Institute, 1992. Chapter 2 presents facts on employment
changes from plant openings, expansions, closures, and contractions.
U.S. Small Business Administration. The State of Small Business: 1984. Washington, D.C.:
Government Printing Office, 1984. Chapters 1, 2, and 4 discuss the debate over David
Birch's work.
tan
AND RENT and URBAN
LAN D - USE PATTERNS
he first part of this book explained why cities exist and where they develop,
<^J7~
kS but not how activities are arranged within cities. This second part examines
the spatial structure of cities, taking a close look at land use within cities. It explores
the market forces and government policies that determine the equilibrium land-use
pattern.
This part is divided into five chapters. Chapter 7 develops the basic concepts of
land rent and land use: it explains the factors that determine the equilibrium price
of land and shows how land is allocated among alternative land uses. Chapter 8
discusses land rent and land use in the monocentric city, explaining why commercial
and industrial activity was concentrated in the central core area. It also explains why
the poor tend to locate in central cities. Chapter 9 uses general-equilibrium analysis
to explore the interactions between the residential land market, the business land
market, and the urban labor market. Chapter 10 describes land use in the modern
cities, explaining why the traditional monocentric city was replaced by the modern
multicentric city. Chapter 1 1 discusses the role of local government in the urban land
market, exploring the market effects of zoning and other land-use controls.
Chapter 7
£T NTRODUCTION
LAND RENT
to
ad LAND USE
The trouble with land is that they 're not making it anymore.
Will Rogers
{^~7~ his chapter introduces some basic concepts of land rent and land use, setting
^y the stage for the discussion of urban land use in Chapters 8 through 1 1 . This
chapter addresses three questions about the land market. First, what determines the
price of land? Second, who benefits from public policies that increase the fertility
or accessibility of land? Third, does the land market allocate land efficiently?
It will be useful to define two terms, land rent and market value. Like other
assets, land yields a stream of marketable services and thus a stream of income. For
example, agricultural land yields a stream of agricultural output (bushels of corn),
generating a stream of income for the farmer. Similarly, a parking lot in the city
yields a stream of parking services, generating a stream of income for the parking
firm. When a landowner grants the rights to use his land to another individual or a
firm, he charges land rent. If a farmer is granted the right to grow corn on a plot
of land, the rent might be $ 1 ,000 per acre per year. If a firm is granted the right to
operate a parking lot on a plot of land, the rent might be $5,000 per acre per year.
What determines The market value of land equals
the market value of land?
the present value of the stream of rental income generated by the land. To explain
the concept of present value, consider an asset that generates R of income each year
and is expected to generate this income for n years. If the market interest rate is i,
the present value of the stream of earnings from the asset is
R
7T
t=0
For example, if an asset is expected to generate $20 of net income per year, starting
today and lasting for a total of five years, and the interest rate is 10%, the present
181
182 Part II Land Rent and Urban Land- Use Patterns
value of the asset is
20 20 20 20
P V = 20 + + + + (7-2)
1.10 1.21 1.33 1.46
PV = 20+ 18.18+ 16.53+ 15.04+ 13.70 = $83.45 (7-3)
If the stream of earnings lasts forever, the equation for present value simplifies to
R 20
PV = - = = 200 (7-^)
/ 0.10
For example, if the $20 annual income lasts forever, the present value of the asset is
$200.
The present value is the maximum amount that an investor is willing to pay
for an asset, given an alternative investment that yields i percent per year. Suppose
that the alternative is a savings account that yields 10 percent per year. The investor
can either invest in an asset that yields $20 per year forever or invest in a savings
account that yields 10 percent per year. At a purchase price of $200, the investor
is indifferent between spending $200 on the asset and investing the same amount
in a savings account: in both cases, the annual income is $20. At a purchase price
less than $200, the investor prefers the asset to the savings account. For example,
if the price is $100, the investor can make $20 per year by investing $100 in the
asset, compared to $10 per year by investing the same amount in a savings account.
Similarly, for a purchase price exceeding $200, the savings account is more lucrative
than the asset.
The market value of land is the present value of the annual rental payments
from the land. Land used for residential, commercial, and industrial activities can,
in principle, yield a constant stream of rental income. In contrast with agricultural
land, which can deteriorate with use, developed land does not deteriorate. Therefore,
the market value equals the annual rent divided by the interest rate. For example, if
the annual rent on a plot of land is $5,000 per acre and the market interest rate is 10
percent, the market value of land is $50,000 per acre. The market value of land equals
the present value because the present value makes an investor indifferent between
buying the land (spending $50,000 to earn $5,000 per year in land rent forever) and
putting the $50,000 in a bank account with a 10 percent interest rate (earnings of
$5,000 per year).
This book uses land rent —
not market value —
as the price of land. Most of the
other relevant economic variables are defined as streams of revenue or costs. For
example, a household earns an annual income, and a firm computes its annual profits
as its annual revenue less its annual cost. To be consistent, ihe price of land is defined
as the annual payment in exchange for the right to use the land: the price of land is
synonymous with land rent. Given the simple relationship between rent and value,
it's easy to make the translation from land rent to market value: just divide the annual
rent by the market interest rate.

Chapter 7 Introduction to hind Rent and Land Use 183
Land Rent and Fertility

David Ricardo ( 1 82 1 ) is credited with the idea that the price of agricultural land is
determined by its fertility. The more productive the land, the more a tenant farmer
is willing to pay to use the land. Fertility analysis demonstrates some of the most
important concepts of land rent in a simple and compelling way.
Consider an agricultural county where tenant farmers use land of varying fertility
to grow corn. The characteristics of the local economy are as follows:
1 Fixed prices. The prices of the output (corn) and inputs (labor, seed,
fertilizer, capital) are determined in national markets, so local farmers take
the prices as given. The prices are the same at all locations in the county.
2. Zero economic profit. There is free entry into farming, so all farmers make
zero economic profits (normal accounting profits).
3. Fertility of land. There are three types of land: h (high fertility).

m (medium fertility), and / (low fertility).
4. Land to highest bidder. Landowners rent their land to the highest bidder.
5. Zero transport costs. Transport costs are assumed to be so small that they
can be ignored. Later in the chapter, this assumption will be relaxed.
Figure 7-1 shows the conventional cost curves for one-acre plots of the three
types of land. The marginal-cost curves (MC) are positively sloped, and pass through
the U-shaped average total-cost curves (ATC) at the minimum points of average-cost
curves. The cost curves include all the nonland costs of production, including the
costs of raw materials (seeds and fertilizer), capital (tractors), and labor. They also
include the opportunity cost of being a farmer, for example, the money the farmer
gives up by being a farmer instead of a steelworker.
The A farmer on
positions of the cost curves depend on the fertility of the land.
produce the same amount of corn with smaller quantities
relatively fertile land can
of the nonland inputs. Because the farmer spends less money on seeds, fertilizer,
tractors, and labor, his average-cost curves are lower. In general, the higher the
fertility, the lower the cost curves.
How much are farmers willing to pay for the three types of land? In Figure 7-1
the national corn market generates an equilibrium price of $10: supply intersects
demand at a price of $10. Farmers are price takers and maximize profit where price
equals marginal cost. The profit-maximizing output on the high-fertility land is 220
bushels per acre, generating profit equal to the shaded area. In this example, profit
equals $1,320 per acre per year (total revenue of $2,200 less a total cost of $880).
A farmer would be willing to pay up to $1,320 per year to use one acre of the high-
fertility land. Similarly, a farmer would be willing to pay up to $320 per year for the
medium-fertility land. For the low-fertility land, production costs are so high that
corn production is not profitable at a price of $10, so a corn farmer would not be
willing to pay anything for the low-fertility land.
• 3
st
s £ —
u O
o- —E
ra
-
-
c
t/1
-
=
u —
0) ^
3 ti
- —
D S
E B|
— H U
a -
u
u B
o J
</» s
£
-~
o o
184
( 'hapter 7 Introduction to Land Rent and Land Use 185
Competition and Land Rent

Competition among prospective farmers bids up the price of land to the point at
which economic profit is zero (accounting profit is normal). Farmers are willing to
pay up to $ 1 ,320 for the high-fertility land, and are forced by competition to do so: at
any rent less than $1,320 per acre, the landowner will be able to find another farmer
willing to pay slightly more to use the land. Similarly, the equilibrium rent on the
medium-fertility land is $320. Because the equilibrium land rents make economic
profits equal to zero, farmers are indifferent between different plots of land. Although
the high-fertility land has lower production costs, the savings in production costs are
offset by higher land costs.
In equilibrium, land rent equals the excess of total revenue over nonland costs.
This is the leftover principle: because of competition among farmers for land, the
landowner gets the leftovers. This principle assumes that individual plots of land
have unique characteristics, but farmers are all the same. Competition among a
large number of fanners, each of whom has the same cost curves, bids up the price
of high-fertility land to the point at which economic profit is zero. If the farmer
on the high-fertility land pays less than the excess of total revenue over nonland
cost, the farmer would be evicted and replaced with another farmer willing to pay
the leftovers (total revenue less nonland cost) for the opportunity to earn normal
accounting profits.
The leftover principle does not hold if there are restrictions on entry and compe-
tition. One restriction on entry comes from patents. If farmer Tom holds the patent for
a particular farming technique, he has lower production costs than all other farmers.
For example, suppose that Tom can produce an acre's worth of corn for a pre-rent
profit of $2,000, and other farmers, using inferior techniques, generate a pre-rent
profit of is unable to charge Tom a rent of $2,000 because the
only $500. The landlord
threat of eviction is a hollow one: there are no other farmers with the same production
costs, so there are no farmers willing to pay $2,000 per acre. Instead, Tom pays only
$500, allowing him to make an economic profit of $1 ,500. The landowner does not
get the leftovers because the patent restricts competition. Once the patent expires
and all farmers have access to the same technology, the landowner can increase land
rent and convert the economic profit into increased land rent.
Land Rent and Public Policy
Fertility analysis can be used to predict the effects of public policy on land rent.
Suppose that an agricultural county builds an aqueduct and provides free irrigation
to farmers. Who benefits from the irrigation project?
Consider first the possibility that the irrigation project does not affect the equi-
librium price of corn. The irrigation project decreases farmers' production costs,
shifting the cost curves downward, as shown in Figure 7-2. For all three types of
land, pre-rent profits increase: high-fertility land and medium-fertility land become
more profitable, and low-fertility land now generates a positive profit. As profit in-
creases, competition among farmers bids up land rent to the point at which economic
-
"g o
c .•-j ?:
• -a
>,
^ -rl-ss
Q.
— j=
r -= w
= .c
w
Ml
<
C 2 «
J "^c ^
-a m
H ^>
u
•n
a .i-
d — 3
-
* b "3
Ci r-
^
-3 ~ =-
Jl "3 J=
3 = £
3 u s
>> O
- M S «
— o i! o
j= § § Si
= S|
7^
o £
-
V
<->
s -
r u g
I l.s
'-
-3 »
S.S
-
.
^ c
u u c
i
U
'-J
—ii
—
"3
-
8, 1
aa-s
E 3 =
< - 2
186
Chapter 7 Introduction to Land Ren! and Land Use 187
profit is zero. The savings in production costs are paid to landowners in the form of
higher rent, so the benefits of the irrigation project go to landowners.
Will the price of corn be affected by the irrigation project? The project in-
creases the supply of corn for two reasons. First, the project shifts the marginal-cost
curves downward, increasing the profit-maximizing outputs of the high-fertility and
medium-fertility farms. Second, marginal land (low-fertility land) is brought into
production. For these two reasons, the supply curve shifts to the right, decreasing
the equilibrium price of corn. Therefore, corn consumers benefit from the irriga-
tion project. As the price of corn decreases, the pre-rent profits of farmers decrease,
decreasing land rent. In other words, consumers gain at the expense of landowners.
What determines between landowners and corn con-

the distribution of benefits
sumers? The general rule is that the smaller the geographical area covered by the
irrigation program, the larger the share of the benefits that go to landowners. Con-
sider first an irrigation project that decreases the production costs of a single 50-acre
plot of land. The project causes a trivial increase in supply and virtually no change
in the price of corn. Therefore, all the benefits go to the landowner. Consider next a
national irrigation project that decreases the production costs of all corn farmers. The
project causes a large increase in supply (existing land is cultivated more intensively
and more land is brought under production), so it decreases corn prices significantly.
In this case, a large share of the benefits goes to consumers.
The benefits of the irrigation project are capitalized into the market value of
land. Since the project increases the annual rent, it increases the present value of
the stream of earnings from land, increasing its market value. For example, suppose
that the annual rent on high-fertility land increases from $1,320 per acre per year
to $1,500. If the market interest rate is 10 percent, equation (1-4) suggests that the
market value of land rises from $13,200 per acre to $15,000 per acre.
Land Rent and Accessibility

This section uses a model developed by Von Thunen (in 1826) to explain why
land rent increases with the accessibility of land (Alonso, 1972). In the Von Thunen
model, accessibility replaces fertility as the determinant of land rent. The Von Thunen
approach will be used in Chapter 8 (Land Use in the Monocentric City) to discuss
urban land rent and land use.
Consider a county where all land is used to grow carrots. The characteristics of
the economy are as follows:
1 . Fixed prices. The prices of the output (carrots) and inputs (labor, seed,
fertilizer, capital) are detci mined in national markets, so farmers take the
prices as given. The prices are the same at all locations.
2. Marketplace. All carrots are transported from farms to a central

marketplace at a cost of t per ton per mile.
3. Competitive markets. There is free entry into carrot farming. In
188 Purr II Land Rent and Urban Land-Use Patterns
equilibrium, all farmers make zero economic profits (normal accounting

profits).
4. Fertility of land. All land is equally fertile, so production costs are the
same at all locations.
Linear Land-Rent Function
Suppose that carrots are produced with fixed factor proportions. In other words,
farmers do not engage in factor substitution as the relative prices of inputs change.
The typical farmer occupies one acre of land and produces Q tons of carrots, which
are sold at a price of P per ton. If the unit transport cost is t per ton per mile, total
transport cost for a farm located u miles from the marketplace is
TC = tQu (7-5)
For example, if unit transport cost (t is $4 per ton per mile and the farmer produces
)
20 tons per acre, transport cost is $80 for a farm one mile from the market. $160
for a farm two miles from the market, and so on. If the farmer spends C on nonland
production costs (labor, raw materials, and the opportunity cost of farming time) and
land rent is R. the profit per acre is
n = PQ-C-t-Qu-R (7-6)
For zero economic profit, the bid rent for land is
R = PQ-C-tQu (7-7)
Because land near the marketplace (it) has relatively low transport costs, the farmer
is willing to paymore for this land.
Figure 7-3 shows the bid-rent function of the typical fanner. The numerical
assumptions are listed below the graph. Total revenue per acre (shown by the hori-
zontal line) is the same at all locations because price and quantity do not vary across
space. Total cost is the sum of nonland cost per acre and transport costs. The total-
cost curve is positively sloped because transport costs increase with distance to the
market; the slope of the cost curve is (t Q). The farmer's bid rent for a particular
location is total revenue less total cost, so the bid-rent function is negatively sloped.
The bid rent is $250 at the marketplace and falls by $80 per mile.
The land-rent function shows the equilibrium land rent for different locations.
Farmers compete for land, bidding up land rent to the point at which economic profit
is zero at every location. If all fanners are identical they all have the same production
(
costs and transport costs), the land-rent function is the same as the bid-rent function
of the typical farmer. The land-rent function makes farmers indifferent among all
land in the county: as the farmer approaches the market, the savings in transport
costs are exactly offset by increases in land rent.
Chapter 7 Introduction to Land Rent and Lxind Use 189
Figure 7-3 Revenue, Cost, and Bid Rent for the Carrot Farmer
Total revenue per acre (P times Q)
300
$
\X
250 ^ Total cost
170
130
Land rent
90 ^T A ^>^
50
10 _1 L
I 1 1
Distance to market (miles)
Land rent equals total revenue less nonland costs and transport costs. The farmer's total revenue is $300 and
nonland cost is $50. Transport cost increases by $80 for every mile from the marketplace, so the slope of the
bid-rent function is $80.
Factor Substitution: Convex Land- Rent Function
The bid-rent function in Figure 7-3 is linear because the farmer is assumed to be
inflexible. The farmer produces 20 tons of carrots with one acre of land and
$50
worth of nonland inputs, regardless of the relative price of land. This section shows
that if farmers engage in factor substitution, the bid-rent function is convex instead
of linear.
Consider a flexible farmer, defined as one who changes input proportions as the
relative price of land changes. If the farmer moves to relatively expensive land, she
substitutes nonland inputs (capital, labor, raw materials) for land, so she can produce
20 tons of carrots on a smaller plot of land. The farmer's production isoquant, mapped
in Figure 7-4, shows the different input combinations that produce 20 tons of carrots:
the possibilities include points B (0.80 acres of land and $60 of nonland inputs) and
C (0.60 acres and $80 of nonland inputs).
Table 7-1 shows how to compute the farmer's bid-rent function. If the farmer
uses T acres of land, the new expression for profit is
P Q-C -t Qu- RT (7-8)

190 Part 11 Land Rent and Urban Land-Use Patterns
Figure 7-4 Isoquant for 20 Tons of Carrots
120 D( 0.40. $120)
100
80 C (0.60. $80)
B (0.80. S60)
60
A (1.0. $50)
40
20
0.20 0.40 0.60 0.80 1.00
Quantity of land (acres)
The isoquant shows the different combinations of land and nonland inputs needed to produce 20 tons of carrots. As
the farmer moves toward the market, the price of land increases, so the farmer moves up the isoquant. substituting
nonland inputs for land.
Table 7- -1 Costs and Land Rent for the Flexible Farmer
Distance
tn Market Farm Size Total Nonland Transport Pre-Renl Rent
1 miles) ( acres) Revenue Costs Costs Profit per Acre
0.4 $300 $120 $ $180 $450

1 0.6 300 80 80 140 233
2 0.8 300 60 160 80 LOO
3 1.0 300 50 240 10 10
Assumptions
Output = 20 tons
Price = $15
Transport cost = $4 per ton per mile
Pre-rent profit = Total revenue — Nonland costs •
Transport cost
Pre-rent profit
Rent =
Farm size
Chapter 7 Introduction to Land Rent and Ixmd Use 191
Economic profit is zero if the total payment for land is
R.T = P • Q-C-tQu (7-9)
Dividing each side by T gives the bid rent per acre:
P-Q-C-t-Q-u
R= (7-10)
At a location three miles from the market, the flexible farmer uses the same input
combinations as the inflexible farmer (one acre of land and $50 worth of nonland
inputs), and is willing to pay the same amount in land rent ($10). As the farmer
moves toward the marketplace, the price of land increases, causing movement up
the isoquant, from point B (two miles), to C (one mile), to D. For each location, the
pre-rent profit is total revenue less nonland costs and transport costs [the right-hand
side of equation (7-9)]. The bid rent per acre equals the pre-rent profit divided by
the size of the farm (in acres). For a farm two miles from the market, pre-rent profit
is $80 on 0.80 acres of land, so the farmer is willing to pay $100 per acre ($80/0.80).
The bid rent increases as the farmer approaches the market, rising to $450 for a farm
close to the marketplace.
Figure 7-5 shows the bid-rent functions of the flexible and inflexible farmers.
The flexible bid-rent function lies above the inflexible one for all locations except
u = 3; that is, the flexible farmer outbids the inflexible one for all land except at
u = 3. To explain the ability of the flexible farmer to outbid the inflexible one,
Figure 7-5 Bid-Rent Functions of Inflexible and Flexible Farmers
500 -
400
Bid-rent function of inflexible farmer
r 3oo -
200 -
100 -
Distance to marketplace (miles)
The bid-rent function of the inflexible farmer is linear because carrots areproduced with fixed factor proportions. In
contrast, the flexible farmer is more efficient, so the fanner has lower production costs and can pay more rent.
192 Part II Land Rent and Urban Land-Use Patterns
suppose that both fanners start at it= 3. At it =3, both farmers have the same
bid rent ($10) because they use the same input combination 1 acre of land and $50 (
of nonland inputs). Suppose that each farmer considers a one-mile move toward
the marketplace, which would decrease transport costs by $80. Since the inflexible
farmer does not change his input combinations, his bid rent would increase by $80.
In contrast, the flexible farmer would substitute nonland inputs for the relatively
expensive land (moving up the isoquant from point A to B in Figure 7-4). Given
the higher price of land at u = 2, point B is more efficient than point A. so the
flexible farmer has lower production costs. The decrease in production costs from
factor substitution increases the bid rent for land, so the flexible fanner can outbid
the inflexible farmer.
Figure 7-6 shows the general shape of the bid-rent functions of flexible and
inflexible farmers. Because the inflexible farmer uses the same input combination at
all locations, the bid-rent function simply reflects differences in transport costs: the
bid-rent function is linear, with a slope equal to transport cost per mile. In contrast,
the flexible farmer engages in factor substitution, generating savings in both trans-
portation costs and production costs as the fanner moves toward the marketplace. As
a result, the flexible bid-rent function isconvex, as shown by Rf. The two bid-rent
functions are tangent at the point at which the two fanners choose the same input
combination (h*).
Figure 7-6 Flexible versus Inflexible Bid-Rent Functions
Bid-rent function of flexible farmer
tion of inflexible farmer
The flexible farmei is more efficient than the inflexible one. and is therefore able to paj more for land. The rent
functions are tangent at the point at winch the inflexible fanner is luck) enough to choose the efficient input
combination.
Chapter 7 Introduction to Land Rent and Land Use 193
Will landowners rent land to inflexible farmers? For most plots of land, flexible
farmers outbid inflexible farmers. The inflexible farmer will get land only if the
farmer is lucky enough to choose the efficient input combination. In general, inflex-
ibility means inefficiency, so the competition for land eliminates inefficient farmers.
Decrease in Transport Costs and Land Rent

How does a decrease in transport costs affect the bid-rent function? In Figure 7-7,
the original bid-rent function is Rq and the radius of the carrot farming district is
«o miles. For land beyond u transport costs are so high that carrot farmers are not
,
willing to pay anything for land. Suppose that a new highway is built, decreasing
the cost of transporting carrots to the market. The decrease in transport costs per
mile decreases the slope of the bid-rent function. If the price of carrots is fixed, the
bid-rent function shifts from Ro to R\, so the radius of the carrot district increases
from wo miles to ii\ miles. In this case, the benefits of decreased transport costs go
to landowners: land within the original farm district commands a higher rent, and
marginal land (between uq and u\) now commands a positive rent.
The decrease in transport costs increases the supply of carrots for two reasons.
First, land in the original carrot district is used more intensively. As land rent rises.
Figure 7-7 Effect of Decreased Transport Cost on Bid-Rent Function
Bid-rent function with lower transport cost

and decreased carrot price
Bid-rent function with lower transport

cost and fixed carrot price
Oriainal bid-rent function
A decrease in transport costs decreases the slope of the bid-rent function. If the price of carrots is fixed, the
function shifts from R (l to R t
.If the supply of carrots increases and the market price drops, the bid-rent function
shifts downward from R< to /?•>.

farmers economize on land by substituting nonland factors for land. As farmers move
up their isoquants. output per acre increases. Second, marginal land (between u {)
and
ii\ ) is brought into production, increasing the total acreage of carrot production.
What are the implications of the increased carrot supply for the bid-rent function?
If the increase in supply is large enough to decrease the equilibrium price of carrots,
total revenue per acre decreases, decreasing the amount farmers are willing to pay
for land. As shown in Figure 7-7, the decrease in the price of carrots shifts the bid-
rent function downward from R\ to /?2- The downward shift of the bid-rent function
decreases the radius of the carrot district from u\ to »:•
The benefits of the highway are shared by landowners and consumers. Landown-
ers can charge a higher rent because their land is more accessible, and carrot con-
sumers face lower prices because there is a greater supply of carrots. As explained
between landowners and
earlier for the irrigation project, the distribution of benefits
consumers depends on the geographical extent of the public works program. If a sin-
gle county improves its road to increase the accessibility of its carrot-growing land,
the supply of carrots will increase by a trivial amount and the price will decrease by
a trivial amount. Therefore, most of the benefits of the highway go to landowners. In
contrast, a national highway program is likely to increase supply and decrease price
by relatively large amounts. Therefore, some of the benefits of new highways go to
consumers.
Two Land Uses

Figure 7-8 shows how land is allocated between two competing users. Because land
is allocated to the highest bidder, land within u miles of the market occupied by is
activity E, and land beyond it' is occupied by C. The equilibrium land-rent function
(the thick line in Figure 7-8) shows the market rent for different locations. Font less
than «', E outbids C. so the land-rent function is same as £"s bid-rent function.
the
For u greater than //', the land-rent function is the same as C's bid-rent function.
Activity E occupies the land closest to the market because it has a relatively steep
bid-rent function. The bid-rent function is negatively sloped because of transport
costs, so the higher the transport costs, the steeper the bid-rent function. Transport
costs are relatively high if the output is relatively heavy or costly to ship.
1. More output per acre: eggplant and cotton. Because eggplant farmers
produce more tons per acre than cotton farmers, they have higher transport
costs and a steeper bid-rent function (everything else being equal).
2. Higher unit transport costs: eggs and carrots. Although carrots can be
thrown in the in egg cartons. A ton oi'
back of a truck, eggs must be placed
eggs takes up more space than a ton of carrots, so eggs have a higher unit
transport cost. Egg producers have higher total transport costs and thus a
steeper bid-rent function (everything else being equal).
In general, the activity with higher transport costs (higher transport rate or greater
weight) occupies the central land.
Figure 7-8 The Activity with the Steeper Bid-Rent

Function Locates Near the Market
Bid-rent function of £
Bid-rent function of C
Because activity £'s bid-rent function is steeper than C's, £ occupies land closer to
the marketplace. £ has a relatively steep bid-rent function because it has relatively
high transport costs, a result of either higher unit transport costs (eggs versus carrots)
or more tons per acre (eggplant versus cotton). The equilibrium land-rent function is
the thick line (the E bid-rent function for u < u', and the C bid-rent function for
u > u'). Land is allocated to its "highest and best use."
Does the land market allocate land efficiently? The market allocates central
land to the activity with the highest transport costs, that is, the activity with the
most from proximity to the market. To explain, suppose that an eggplant
to gain
farmer a location one mile from the market and then swaps locations
starts at
with a cotton farmer three miles from the market. Since the eggplant farmer has
higher transport costs per acre of production, the land swap increases eggplant
transportation costs by more than it decreases cotton transport costs, so total trans-
port costs increase. In general, because the market allocates central land to the
activity with relatively large transport costs, it minimizes total transportation costs.
In the terms used by land developers, land is allocated to its "highest and best
use."
The conclusion that the land market allocates land efficiently rests on the as-
sumption that there are no externalities in land use. If there are externalities, the
market allocation is inefficient, and government intervention can be justified on ef-
ficiency grounds. The issue of land-use externalities will be examined in detail in
later chapters.
Market Interactions
The demand for land is derived from the demand for outputs (corn, carrots, housing,
retail goods, manufactured goods). This section examines the interactions between
the land market and the output market. The discussion addresses a sort of chicken-
and-egg question about the land market: is the price of land high because the price
of output is high, or is the price of output high because the price of land is high?
The Corn Laws Debate

The British Corn Laws of the 1 800s restricted grain imports to Britain. The decrease
in the supply of imported grain increased the demand for domestically produced
corn. Figure 7-9 shows the effects of the Corn Laws on the corn market and the land
market.
1 Corn market. The Corn Laws shifted the demand curve from d\ to ch.
The price of domestic corn increased from P\ to P^. and the quantity of
corn produced increased from C\ to Cj.
2. Land market. As domestic corn production increased, the demand for
land increased. In Figure 7-9, the increase in corn production shifted the
demand curve from D\ to Dt. Because the supply curve is
for land
making land any more), the increase
perfectly inelastic (they aren't in
demand increased the price of land from R\ to /?:.
To summarize, the price of land is high because the price of corn is high. The
Corn Laws increased the price of corn, which stimulated the production of corn and
the demand for corn-growing land. Landowners responded by increasing the price
of land to allocate the fixed resource among competing land uses. The lesson is that
high land prices are the result of high corn prices, not the reason for high corn prices.
Housing Prices and Land Prices

The lesson from the Corn Laws debate is applicable to the urban housing market.
Consider the following statement: "Greedy landowners in the San Francisco Bay
Area have increased the price of land, increasing the price of housing." In fact, the
price of land is demand for housing (and the demand for land on
high because the
which to build housing) As the demand for housing increases, the price
is high.
of housing increases, causing builders to buy more land to build more houses. The
increase in the demand for land increases the price of land. Or consider the following:
"The price of land Boston is so high that few people can afford to live there."
in
In fact, the price in Boston is high because so many people can afford to
of land
live there. The large demand for housing generates a large demand for land, which
causes a relatively high price of land. The high price of land is the result — not the
cause —of high housing prices.
Figure 7-9 The Corn Laws Debate

(A)
Corn market
Corn supply
C, C2
Quantity of corn
Quantity of land
The Corn Laws restricted grain imports, increasing the demand for
domestic corn and its price. Corn production increased from C\ XoCj,
increasing the demand for landfrom D] to £>2- Th e price of land
increased from /?] to Rj- The price of land is high because the price of
corn (and the production of corn) is high.
The Single Tax

In 1880, Henry George proposed a 100 percent tax on rental income. The proposed
tax was dubbed the "single tax" because it would have generated enough revenue
to support all levels of government at the time. The spirit of the single tax is best
expressed by George himself. The following is from an interview with David Dudley
Field (in the North American Review in 1885):
Field: Then suppose A to be the proprietor of a thousand acres on the Hudson.

chiefly fanning land, but at the same time having on it houses, barns, cattle,
horses, carriages, furniture; how is he to be dealt with under your theory?

George: He would be taxed on the value of his land, and not on the value of
his improvements and stock. The effect of our present system, which taxes
. . .
a man by his labor and capital, is to put a fine upon industry.

for values created
and repress improvement. The more houses, the more crops, the more
buildings in the country, the better for us all, and we are doing ourselves an
injury by imposing taxes upon the production of such things.
Field: Then you would tax the farmer whose farm is worth $1,000 as heavily
as you would tax the adjoining proprietor, who, with the same quantity of
land, has added improvements worth $100,000; is that your idea?
George: The improvements made by the capitalist would do no harm to
It is.
the farmer, and would benefit the whole community, and I would do nothing to
discourage them.
Field: A large landlord in New York owns a hundred houses, each worth, say,
$25,000 (scattered in different parts of the city); at what rate of valuation
would you tax him?
George: On his houses, nothing. I would tax him on the value of the lots.
Field: As vacant lots?
George: As if each particular lot were vacant, surrounding improvements
remaining the same.
Field: Well, what do you contemplate as the ending of such a scheme?
George: The taking of the full annual value of land for the benefit of the
whole people. I hold that land belongs equally to all, that land values arise
from the presence of all. and should be shared among all.
George proposed the single tax for both equity and efficiency reasons. On the
equity issue, George argued that land rent is determined by nature and society, not by
the efforts of landowners. As discussed earlier in the chapter, agricultural land rent is
determined by the fertility of the soil and its accessibility to markets. Similarly, the
urban land rent is determined by its accessibility to other activities. In George's time,
cities were growing rapidly, causing rapid increases in land rent and value. George
argued that landowners did nothing to deserve the increases in property value, so
that any windfall gains from urban growth should be taxed away.
On the efficiency issue, George argued that the land tax would eliminate the need
for taxes on improvements. The elimination of improvement taxes would stimulate
Chapter 7 Introduction to Land Rem and Land Use 199
investment in houses, crops, and buildings. The land tax would not affect the supply
of land because the supply of land is fixed. The replacement of the improvement tax
with the land tax would increase the total wealth of society.
The single tax has been criticized for three reasons. First, the single tax would
decrease the net return to the landowner (net land rent) to zero, making the market
value of land zero. In other words, the government would essentially confiscate the
land. This strikes many people if the net return on land is
as inequitable. Second,
zero, landowners will abandon government bureaucrats to decide
their land, leaving
who uses the land. Unlike the private owner, who receives more income if the land
is used efficiently, the bureaucrat has nothing to gain from the efficient use of land.
Therefore, the government land market is less likely to allocate land to its highest
and best use. The third criticism is that it is difficult to measure land rent (and the
appropriate tax). Most land has structures or other improvements, and it is difficult
to separate the value generated by the raw land from the value generated by the
improvements.
An alternative to the single tax is a partial land tax. Under a partial tax, land
is taxed at less than 100 percent of its value. A partial land tax would be less
confiscatory than the single tax: like conventional taxes on labor and capital, the
partial tax would confiscate only a portion of the taxpayer's resources. In addition,
because a partial tax leaves landowners with a positive net return, the land market will
continue to be run by those who have a private interest in allocating land to its highest
bidder.
Another alternative two-rate tax, or the split tax. Under
to a pure land tax is the
the conventional property tax, land and improvements are taxed at the same rate.
A 3 percent property tax is actually a 3 percent tax on land and a 3 percent tax on
improvements. Under a split tax, the tax rate on land may be 9 percent, while the tax
rate on improvements may be percent. The split tax is widely used in Australia and
1
New Zealand. It is also used in some cities in Pennsylvania: Pittsburgh implemented

a split tax in 1913, and six other cities, including Scranton and Harrisburg, have
adopted the split tax in recent years.
The replacement of the conventional property tax with the split tax would stim-
ulate capital investment. The switch to the split tax would decrease the tax rate
on capital, encouraging capital improvements. Suppose, for example, that Rhonda
would like to add a recreation room to her house. The new room would increase the
assessed value of her house by $20,000. Under a conventional 3 percent property
tax, her tax liabilities would increase by $600 per year. Under a split tax with a 1
percent rate on improvements, her tax liabilities would increase by only $200 per
year. The tax penalty from the home improvement would be lower under the split
tax, so Rhonda would be more likely to improve her house.
The same argument applies to investment in commercial and industrial property.
Under the split tax, the tax liability of the property owner is only percent of market
1
value, so investments that increase the market value of the property have smaller tax
penalties. The owner of an apartment building is more likely to install a new roof if
the roof (and the associated increase in assessed value) increases his tax liability by
a relatively small amount.
Urban Land Taxation in Pittsburgh

Many cities in Pennsylvania have a graded property-tax system under which land is
taxed at twice the rate as improvements. In the 1980s. Pittsburgh increased the tax
rate on land and decreased the rate on improvements, thus moving its graded system
in the direction of George's single tax. By 1991, the tax rate on land was more than
five times the rate on improvements. In the 1980s Pittsburgh experienced a boom in
commercial building, with most of the new construction involving office buildings
in the central business district. Is there a connection between Pittsburgh's tax reform
and its building boom?
Oates and Schwab (1992) explore the effects of Pittsburgh's tax reform on
building activity in the city. They conclude that the tilting of the graded-tax system
in the direction of higher land taxes stimulated building activity in the city. The
higher tax rate on land generated revenue that allowed the city to reduce the tax rate
on improvements. The decrease in the improvement tax decreased the penalty on
new construction and thus encouraged development. Although other factors played a
role in the city's commercial building boom of the 980s, the tilting of the graded-tax
1
system toward lower taxes on improvements was an important factor.
Summary
1 According to the leftover principle, the bid rent for land equals the difference
between total revenue and total cost. Competition for land ensures that the
landowner gets the excess of total revenue over total cost.
2. Land that is relatively fertile has relatively low production costs, so it
commands a higher rent.
3. A policy that increases fertility (e.g., an irrigation project) generates benefits for
both landowners and consumers.
a. The project decreases production costs, increasing land rent.
b. If the supply of the agricultural good increases, the price decreases,
generating benefits for consumers.
c. The distribution of benefits between landowners and consumers depends on
the geographical extent of the program: the smaller the geographical area
affected by the project, the smaller the price decrease and the larger the rent
increase.
4. The benefits of an irrigation project are capitalized into the market value of
land: the increase in rent increases the present value of rental income,
increasing the market value of land.
5. The bid-rent function shows how much a firm is willing to pay for land at
different distances from the market. The function is negatively sloped:

transportation costs are lower near the market, so rent is higher. The rent
function is convex — not —
linear if firms substitute nonland inputs for land as
the price of land increases.
6. A decrease in transport costs decreases the slope of the bid-rent function. If the
resulting increase in supply of agricultural goods decreases agricultural prices,

the bid-rent function shifts downward. The benefits of increased accessibility
are shared by consumers and landowners.
7. If there are two land uses, the activity with the higher transport costs has a
relatively steep bid-rent function and occupies the land closest to the
market.
a. The bid-rent function is relatively steep if (1) output per acre is relatively
large or (2) unit transport costs are relatively high.
b. The market allocates central land to the activity with the most to gain from
proximity.
8. The demand for land is derived from the demand for output (e.g., corn,
housing). The price of land is high because the demand for output is high.
Expensive land is the result —not the cause —of expensive output.
9. Henry George proposed the single tax, a 100 percent tax on rental income.

1. In the state of California, rice growers burn their field stubble to sanitize their
fields.The field burning causes serious air pollution. The alternative sanitizing
method costs $150 per acre. Consider a county where rice farmers are
currently willing to pay $500 per acre for land, and corn farmers (who do not
sanitize their fields) are willing to pay $300 per acre. The total output of the
county is small enough that the prices of rice and com are unaffected by events
in the county. Suppose that field burning is outlawed in the county, forcing rice
farmers to switch to the alternative sanitizing method.
a. How does the field-burning law affect rice consumers, corn consumers,
farmers, and landowners? In other words, who bears the cost of the
pollution-control program?
b. How would your answer to (a) change if the cost of the alternative method
were $250 per acre?
c. How would your answer to (b) change if field burning were outlawed in the
entire state of California?
2. Critically appraise the following statement:
I would like to clear the air with some facts about rice straw burning. Burning is the only
economical way to prevent stem rot in rice. This disease would drastically reduce the yield of
rice grown on the same land the next year. The California Department of Agriculture estimates
the cheapest alternative to rice straw burning, which involves baling and hauling it elsewhere,
would cost about $150 per acre. The opponents of straw burning suggest the savings ($150
per acre) go straight into the pockets of growers. Actually, straw burning decreases the prices
of Rice Krispies and other rice products, so the savings go to consumers.
3. Consider an agricultural economy with the following characteristics:

t All the land in the region is initially used by tenant farmers to grow
indigo.
it The price of indigo is determined in international markets.
Hi. The tenant initially pays the landowner 30 percent of the indigo harvest
as rent.
iv. Output per acre is 1.000 units per year and the price of indigo is S2 per
unit,
v. The interest rate is 10 percent per year.
a. Compute ( 1 ) nonland cost per acre per year and (2) the market value of land.
b. Suppose that the price of indigo drops to SI .90. Assuming that the tenant
continues to grow indigo on the land, compute the equilibrium rent ( 1 ) in
dollars. (2) in units of indigo, and (3) as a percent of the indigo harvest.
c. By how much does the market value of land drop as a result of the decrease
in the price of indigo (assuming the tenant grows indigo)?
d. How would your answer to (c) change if there is an alternative crop with the
same nonland costs and output per acre?
4. Suppose that Mr. Greengenes. a farmer and genetic engineer, develops a new
method for growing corn that decreases the cost of growing corn by S300 per
acre. Greengenes's landlord rejoices, saying. "According to the leftover
principle, you will pay me $300 more in rent." Is the landlord correct? If
not. is he applying the leftover principle incorrectly, or is the principle
wrong?
5. Consider a county where farmers produce with fixed factor proportions and
truck their output to a central marketplace. Draw the land-rent function under
the following circumstances:
a. Unit transport cost is a constant t per ton per mile.
b. Unit transport cost increases as the farmer approaches the marketplace i a
result of traffic congestion).

c. Unit transport cost decreases as the farmer approaches the marketplace, a
result of better roads closer to the market.
6. Using the information in Table 7-1, recompute the land-rent function for the
following events. Assume that the farm size at each location is fixed (0.4 acres
at u = 0. 0.6 acres at u = 1, and so on).
a. A herd of rabbits invades the county, stealing S50 worth of carrots per acre.
b. The unit cost of transport increases from $4 to $5.
c. The price of carrots increases from $15 to $40.
d. Is the assumption of a fixed farm size at each location realistic? If not. how
would your answer to (c) change if farm sizes could change.'
7. Consider a flexible carrot farmer. Complete the following table, assuming that
( the farmer produces 10 tons of carrots. (2) the price of carrots is $40 per
1 )
ton. and ( 3 ) transport cost per ton per mile is $5.

Distance
to Market Fann Size Total Nonland Transport Pre-Rent Rent
(miles) (acres) Revenue Costs Costs Profit per Acre
0.70 $66
1 0.80 52
2 0.90 40
3 1.00 30
8. Consider the input combinations listed in Table 7-1 . As the price of land
increases, the farmer substitutes nonland inputs for land. As the farmer uses
less and less land, the trade-off between land and nonland inputs becomes less
favorable.
a. In what sense does the trade-off become less favorable?
b. Why does the trade-off become less favorable?
9. Suppose that two activities, F and R, compete for land. Activity R produces a
good that requires fixed factors of proportion. Factor substitution is
impossible. Activity F engages in factor substitution: as the relative price of
land increases, the firm substitutes nonland inputs for land.
a. Draw a pair of bid-rent functions showing F occupying the land closest to
the market. In what sense is it F to occupy central land?
efficient for
b. Draw another pair of bid-rent functions showing F occupying land close to
the market and far from the market, and R occupying the intermediate land.
Inwhat sense is it F to occupy land close to and far from the
efficient for
market?
10. Pick the word in parentheses that makes the following statement correct: "As a
firm's isoquant gradually changes in shape from a straight line to an L-shaped
curve, the firm's bid-rent function becomes (less, more) curved." Note: Less
curved means closer to a straight line.
1 1 Consider Euphoric County, where a large share of the arable land is used to
grow M. The production of is illegal: M there are severe penalties imposed on
M growers, but no penalties imposed on M consumers. Suppose that M is a
competitive industry, with equilibrium profits equal to zero: total revenue
equals total costs. Included in the costs are the costs associated with engaging
in illegal activities (the opportunity cost of time spent in jail, legal costs,
concealment costs). Suppose that Euphoric County legalizes the production

ofM.
a. Depict graphically the effects of legalization on the equilibrium price and
quantity ofM. Explain your graph.
b. Depict graphically the effects of legalization on the price of land in
Euphoric County. Explain your graph.
12. The residents of mobile home parks own and rent land from
their dwellings
absentee landowners. Consider a city in which all currently occupied
land is
by mobile home parks. Suppose the city imposes a 50 percent tax on land, to
be paid (in legal terms) by the person who occupies the land (the tenant, either
a mobile home owner or some other user). Who actually pays the tax?
1 3. What would be the effect of a partial land tax ($100 per acre) on land rent, land
values, and corn prices?
14. As the flexible farmer approaches the marketplace, the farmer substitutes
nonland inputs for land. As a result, the land-rent function of the flexible
farmer is steeper than the land-rent function of the inflexible farmer. What
happens as the flexible farmer moves away from the marketplace? Is the
flexible land-rent function steeper or flatter than the inflexible function?
Explain.
References and Additional Reading

Alonso. William. Location and Land Use. Cambridge. Mass.: Harvard University Press.
1964.
"'A Theory of the Urban Land Market." In Readings in Urban Economics, ed.
.
Matthew Edel and Jerome Rothenberg. New York: Macmillan. 1972. Extends the
standard model of agricultural land rent to the urban land market.
George. Henry. Progress and Poverty. New York: Schalkenbach Foundation. 1954.
Discusses George's theory of land rent and explains the single-tax proposal.
Mills, Edwin S. "The Value of Urban Land." In The Quality of the Urban Environment, ed.
H. Perloff. Washington. D.C.: Resources for the Future. 1969. Traces the theory of land
rent from Ricardo's fertility analysis to the modern theory of urban land rent.
Oates. Wallace, and Robert Schwab. "The Impact of Urban Land Taxation: The Pittsburgh
Experience." Cambridge. Mass: Lincoln Institute of Land Policy. 1992.
Ricardo, David. Principles of Political Economy and Taxation. 1821. Reprint. London: John
Murray. 1886. Explains fertility analysis and Ricardo's views on other matters.
Chapter 8
AND USE in the
MONOCENTRIC CITY
The outcome of the city will depend on the race between the automobile and the
elevator, and anyone who bets on the elevator is crazy.
Frank Lloyd Wright
his chapter uses the concepts developed in Chapter 7 to discuss land rent
(^~J7~
c/ and land use in the monocentric or core-dominated city. The monocentric
city was the dominant urban form until the early part of the twentieth century.
In the monocentric city, commercial and industrial activity is concentrated in the
central core area. During the last 70 or 80 years, most large metropolitan areas have
become multicentric, with suburban subcenters that complement and compete with
the central core area. This chapter explains the market forces behind the development
of the monocentric city, and Chapter 10 discusses the market forces behind the
transformation of monocentric cities to multicentric ones.
Why study the monocentric city? Although few of today's large cities are mono-
centric, the analysis of the monocentric city is important for four reasons. First, the
monocentric city was the dominant urban form until the early part of the twentieth
century, so urban history is largely a history of the monocentric city. Second, many
of today's small- and medium-sized cities are still monocentric. Third, to understand
the transition from the traditional monocentric city to the modern multicentric city,
one must understand the forces behind the development of the monocentric city
in the first place. Fourth, many of the lessons from the monocentric model can be
extended to the modern multicentric city.
The discussion of the various land users in the monocentric city proceeds from
the city center outward. The first section derives the bid-rent functions of three busi-
ness sectors (manufacturers, office firms, retailers), and uses the bid-rent functions
to discuss land-use patterns in the central business district. The second section exam-
ines the location choices of households, deriving the bid-rent function for residential
land. The third explains why employment is concentrated in the city center, that
205
206 Part 11 Land Rent and Urban Land- Use Patterns
is, why monocentric. The fourth section derives the residential bid-rent
the city is
function under a more realistic set of assumptions than those of the simple mono-
centric model. The final two sections deal with empirical issues, addressing three
questions. First, why do poor households tend to locate in the central city, while
wealthy households tend to locate in the suburbs? Second, how rapidly does land
rent fall as distance to the city center increases'? Third, what is the relationship
between population density and distance to the city center?
The traditional monocentric city has the transportation technology of the nine-
teenth century. The monocentric model has four key assumptions:
1. Central export node. All manufacturing output is exported from the city
through a railroad terminal at the city center (a central export node).
2. Horse-drawn wagons. Manufacturers transport their freight from their
factories to the export node by horse-drawn wagons.
3. Hub-and-spoke streetcar system. Commuters and shoppers travelby
streetcar from the residential areas to the central business district (CBD).
The streetcar lines are laid in a radial pattern: the lines form spokes that
lead into the CBD (the hub).
4. Agglomerative economies. The office industry depends on face-to-face

contact: employees from different office firms meet in the city center to
transact business.
Under these assumptions, the city center is the focal point of the entire metropolitan
area: manufacturers are oriented toward the export node; office firms are oriented
tow ard the central market area: retailers are oriented to the hub of the streetcar system:
and households are oriented toward employment and shopping opportunities in the
central core area.
Commercial and Industrial Land Use

This section discusses land rent and land use in the central business district (CBD) of
the monocentric city. Three activities occupy the central core area: manufacturers,
office firms, and retailers.
The Bid-Rent Function of Manufacturers

Suppose that manufacturers in the monocentric city produce baseballs. The firms in
the baseball industry have the following characteristics:
1 Production. Firms produce baseballs with land, labor, capital, and raw
materials. Every firm produces B tons of baseballs per month.
2. Fixed prices. The prices of baseballs and nonland inputs (labor, capital,
and raw materials) are determined in national markets, so firms take these
prices as given. The prices are the same at all locations in the city.
Chapter 8 Land Use iii the Monocentric City 207
3. Competitive markets. There is free entry into the industry. In equilibrium,

each firm makes zero economic profits (normal accounting profits).
4. Baseball freight cost. Baseballs are shipped by horse-drawn wagon from
the factory to the central railroad terminal, where they are exported to other
cities.
5. Raw material freight cost. Raw materials are imported to the city by rail.
The intracity freight cost of raw materials is small enough to ignore.

6. Factor substitution. Baseball makers engage in factor substitution: as the
price of land increases, firms substitute nonland inputs for land.
For the purposes of choosing a location within the city, baseball firms are market-
oriented. The intracity freight cost of raw materials is assumed to be negligible, and
the costs of other inputs (capital and labor) are assumed to be the same at all locations
within the city. Because input costs are the same throughout the city, a firm's location
decision is based on access to its market. Once a firm decides to locate somewhere
in the city, the relevant market is the destination of baseballs within the city, that is,
the central railroad terminal.

The firm's profit equals total revenue less the cost of inputs, intracity freight,
and land. Total revenue is the price of baseballs (Pb) times the quantity produced
(B). C is the cost of all nonland inputs (capital, labor, raw materials). If freight cost
is t per ton per mile, total freight cost for a location w miles from the export node is
/ times B times u. If R is land rent per acre and T is the acreage of the factory site,
the profit at a location u miles from the export node is
7t = Pb B-C-tBu-RT (8-1)
All markets are perfectly competitive, so the firm's economic profit is zero. Accord-
ing to the leftover principle, the firm is willing to pay its landowner the excess of
total revenue over the cost of nonland inputs and freight cost. As shown in Chapter 7,
the expression for the bid rent is derived by setting ti = 0, adding R T to both sides
of the profit equation, and dividing by T. The firm's bid rent per acre is the pre-rent
profit divided by land consumption:
Pb
~ B-C-1 B • • u
R = ^ (8-2)
The bid-rent function indicates how much the typical firm is willing to pay
per acre for different production sites in the city. Table 8-1 shows how to compute
the bid rent for different distances from the city center, given values for Pi,, B, C, t,
T, and u. The bid-rent function, shown in Figure 8-1, is negatively sloped because
freight cost increases as the firm moves away from the city center.
The bid-rent function is convex because firms engage in factor substitution. As
the firm approaches the city center with its higher land cost, the firm substitutes
nonland inputs (capital and labor) for land, producing the same tonnage of baseballs
with less land and more of the other inputs. From Table 8-1, the firm uses one acre
of land and $1,400 worth of other inputs for a location three miles from the city
center, and 0.40 acres and $2,600 worth of nonland inputs for a location near the city
208 Part II Land Rent ami Urban Land- Use Patterns
Table 8- -1 The Bid-Rent Function of Baseball Manufacturers
Miles from Size of Site Total Nonland Freight Pre-Rent Rent

Export Node 1acres I Revenue Costs Costs Profit per Acre
0.4 $6,000 $2,600 $ $3,400 $8,500

1 0.6 6.000 2,000 900 3.100 5.167
2 0.8 6.000 1.600 1.800 2.600 3.250
3 1.0 6,000 1 .400 2,700 1.900 1.900
Assumptions:
1. Output = 6 tons of basebal Is
2. Price = $1,000 per ton
3. Transport cost = $900 per mile
4. Pre-rent profit = Total revenue — Nonland cost — Freight cost
Pre-rent profit
5. Rent =
Size of factory site
Figure 8-1 Bid-Rent Function of Baseball Firms

$
8.500
5.167
N. Baseball bid-rent function
3,250
1 ^\
I
1 .900
1 1
2 3
Distance to central export node (miles)
The bid-rent function of baseball firms is negatively sloped because freight costs increase as the distance
in the citj 's central export node increases. It is convex (not linear) because firms substitute nonland
lnpiit-s lot land as the price of land increases.
Chapter 8 Land Use in the Monocentric City 209
center. The bid-rent function of the typical baseball firm will be used to represent
the bid-rent function of the entire manufacturing sector.
The Bid-Rent Function of Office Firms

Although firms in the office sector provide a wide variety of goods and services, office
firms share two important characteristics. First they gather, process, and distribute
information. Because information becomes obsolete quickly, office firms must be
able to collect and distribute it rapidly. Second, office firms rely on face-to-face
contact in the collection, processing, and distribution of information. For example,
accountants explain and interpret the information in accounting reports. The loan
officers of banks meet with prospective borrowers to appraise their creditworthiness.
The investment advisors of finance firms meet with clients to assess their attitudes
toward risk and their investment inclinations. In general, office firms rely on speedy
face-to-face contact in collecting and distributing information. In contrast to baseball
makers, who throw their baseballs into the back of a horse-drawn wagon, office firms
transmit their output in the minds and briefcases of their employees.
Suppose that office firms in the monocentric city provide financial services. The
finance industry has the following characteristics.
1. The Every finance firm is based in an office. The nonland cost of

office.
and labor cost) is C per month. The "output" of the firm
the office (capital
is investment consultations, and each firm produces A consultations per
month.
2. Travel to city center. The manager of each firm travels from the office to
hub of the streetcar system) to consult with
the city center (the clients.
Every consultation requires one trip to the city center.
3. Fixed prices. The prices of financial advice and nonland inputs are
determined in national markets, so finance firms take the prices as given.
The prices are the same at all locations in the city.
4. Competitive markets. There is free entry into the industry. In equilibrium,

all firms make zero economic profits (normal accounting profits).
5. Factor substitution. Finance firms engage in factor substitution: as the

price of land increases, they substitute nonland inputs for land.
The travel cost of an office firm equals the opportunity cost of the manager's
travel between the office and the clients in the city center. Suppose that the manager
takes t minutes to walk one block, and the wage is W per minute. If the office is u
blocks from the city center, the travel cost per consultation is
TC = t W it (8-3)
For example, if t =3 minutes per block and W = $4 per minute, travel cost per
consultation per block is $12, so a firm located 10 blocks from the city center incurs
a travel cost of $ 1 20 per consultation. If the firm provides A consultations per month,
the monthly travel cost for a location u blocks from the city center is
TC = t- W Au (8-+)
If A — 200 consultations per month, the monthly travel cost is S2.400 for a location
one block from the city center. S4.800 for a location two blocks from the city center,
and so on.
The firm's total profit is total revenue less the cost of nonland inputs, land, and
travel. If the price of a consultation is Pa . rent per acre is R. and the firm occupies
T acres of land, profit is
7t = Pa A-C-RT-tWAu (8-5)
Since markets are perfectly competitive, economic profit is zero in equilibrium.

According to the leftover principle, the firm is willing to pay the landlord the excess
of total revenue over nonland cost. The bid rent per acre of land is the pre-rent profit
di\ ided by the amount of land consumed:
P A-C-t- W Au
—
a
R = (8-6)
T
The bid-rent function indicates how much the office firm is willing to pay for
different office sites. Table 8-2 shows how to compute the bid rent for different
distances from the city center, given values for P A. C. r. W, T. and //. The bid- tl
.
rent function, shown in Figure 8-2. is negatively sloped because travel cost increases
as the firm moves away from the city center.
The bid-rent function is convex because office firms engage in factor substitution.
As the firm approaches the city center, it substitutes nonland inputs (capital and labor)
for the relatively expensive land, producing the same number of consultations with
less land and more of its nonland inputs. In other words, office firms near the city
Table 8-2 The Bid-Rent Function of Office Firms
Blocks from Size of Site Total Nonland Freight Pre-Rent Rent

City Center lacn s
-
Revenue Cost Cost Profit per Acre
0.4 s^.600 $3,600 S $6,000 $15,000

1 0.6 9.600 2,400 2.400 4.800 8.000
: 0.8 9.600 1.800 4.800 3,000 3,750
3 1.0 9.600 1,500 7.200 900 900
\ssumptions:
1 . Output (A) = 200 consultations
2. Price = S4S per consultation
3. Travel time (/) =3 minutes per block
4. Opportunitx cost i U ' i — S4 per minute
5. Travel cost = t-A-W = $2,400 per block
6. Pre-rent profit = Total revenue — Nonland cost Travel cost
Pre-rent profit
7. Rent
Si/e of office site
Chapter 8 Land Use iii the Monocentric (
'ity 211
Figure 8-2 The Bid-Rent Function of Office Firms
15,000
c 8,000
Office bid-rent function
3,750
900
Distance to city center (blocks)
The office bid-rent function is negatively sloped because travel costs increase as distance to the
central marketplace increases. It is convex because linns substitute nonland inputs for land as the
price of land increases.
center occupy taller buildings. The bid-rent function of the typical office firm wil
be used to represent the bid-rent function of the entire office sector.
Land Use in the Central Business District
In the monocentric city, manufacturers and office firms are oriented toward the central
business district. Manufacturers are attracted by the central export node, and office
firms cluster around the city center to facilitate face-to-face contact. How is CBD
land allocated between the two activities?
Figure 8-3 shows the bid-rent functions of manufacturers (R m ) and office firms
(R )-Figure 8-3 also shows the bid-rent function of city residents (/?/,), which is
derived later in the chapter. Because land is allocated to the highest bidder, office firms
outbid manufacturers for land within u miles of the city center, generating an office
district with a radius of u miles. Manufacturers outbid office firms and residents for
land between u G and u m miles of the city center, generating a manufacturing district
with a width of (u m — u ) miles. The central area of the city is occupied by the
office industry because the office bid-rent function is steeper than the manufacturing
bid-rent function.
The office bid-rent function is relatively steep because the office industry has rel-
atively high transportation costs. Office firms rely on frequent face-to-face

212 Part 11 Lund Rent and Urban Land-Use Patterns
Figure 8-3 Bid-Rent Functions and Land Use in the Central Business District
) Office
(
J Manufacturing
The office industry has a relatively steep bid-rent function because the travel cosl ol people exceeds the travel cost
of freight I he office industr) outbids manufacturers for land near the citj center. Central land is occupied bj the
activity with the most to yam from proximity (decreased transportation costs).
Chapter X Land Use in the Monocentric City 213
contact, using high-priced financial consultants to transport their output to clients

in the city center. In contrast, the manufacturing industry ships its output by horse-
draw n wagon, so its transport costs are relatively low. In the numerical examples
shown in Tables 8-1 and 8-2. the baseball manufacturer has monthly freight costs of
$900 per mile, while the office firm has monthly travel costs of $2,400 per block.
The bid-rent functions are negatively sloped because of transportation cost, so the
larger the transportation cost, the larger the slope.
Does the land market allocate land efficiently? In the terms used by land devel-
opers, is land allocated to its "highest and best use"? The office industry, with its
higher transportation costs, occupies the land closest to the city center. This alloca-
tion is efficient because the office industry has the most to gain from proximity to
the city center. explain, suppose that a finance firm one block from the city center
To
swaps locations with a manufacturer three blocks (one-fifth of a mile) from the city
center. The land swap increases the finance firm's travel costs by $7,200 (three blocks
times $2,400 per block), but decreases the baseball firm's freight costs by only $180
(one-fifth mile times $900 per mile). Office travel costs increase by more than freight
costs decrease, so total transportation costs increase. The market allocation, which
gives central land to the office industry, minimizes total transportation costs.
Location of Retailers
Where in the monocentric city do retailers locate? Central place theory, which is
used in Chapter 5 to explain the regional distribution of retailers, is also applicable

to the intracity distribution of retailers. Retailers carve up the city into market areas,
with the size of the market areas determined by scale economies, per capita demand,
and transportation costs.
—
Large Scale Economies Glove Sellers. Consider first an activity for which scale
economies are large relative to per capita demand, for example, a glove store. The
glove market has the following characteristics:
1 . Single glove store. The scale economies of glove selling are exhausted
only at outputs that are large relative to the total demand for gloves, so
there is a single glove store in the city. The efficient size for the glove
seller is 5,000 pairs of gloves per month, and total demand for gloves is
5,000 per month.
2. Glove consumers. Consumers are distributed uniformly throughout the
city.
3. Perfect competition. Although there is a single glove seller, entry into the
glove market is not very costly. Given the threat of entry, economic profit is
zero.
The retailer's profit at a particular location is the excess of total revenue over
total cost. Suppose that the profit margin (price less average cost) is constant. If Pg
is the price of gloves, G is the quantity sold, and ACR is the average cost, pre-rent
214 Pan II Land Rem and Urban Land-Use Patterns
profit is
n =G • (Pg - ACg ) (8-7)
For example, if P is S9 and AC. is $5, the profit margin is S4. If G is 5.000. total
g
profit would be S20.000.
At what location will the glove store earn the most profit? If the profit margin
is constant, the firm maximizes total profit by maximizing sales volume (G). As
shown in Chapter 5. sales volume is maximized at the center of the market area (the
median location) for the simple reason that the central location is accessible to the
most consumers. Since the glove store sells to people throughout the city, profit is
maximized at the city center. The benefit of the central location is reinforced by the
hub-and-spoke streetcar system of the monocentric city, which delivers suburban
commuters and shoppers to the city center.
Because there is free entry into the glove-selling business, the glove store makes
zero economic profit. Competition for the best glove-selling site bids up the price of
land to the point at which economic profit is zero. If the glove store refuses to pay its
economic profits to the landowner, the landowner will rent the site to another glove
seller.
Moderate Scale Economies —Hat Sellers. Consider next an activity for which
scale economies are moderate relative to per capita demand, for example, hat stores.
The hat industry has the following characteristics:
1 . Five hat stores. The scale economies of hat selling are moderate relative to
the total demand for hats, so there are five hat stores in the urban area. The
efficient sales volume for each hat store is 4.000 hats per month, and the
total demand is 20.000 hats per month.
2. Hat consumers. Consumers are distributed uniformly throughout the
monocentric city.
3. Perfect competition. Entry into the hat market is not very costly. Given the
threat of entry, economic profit is zero.
The pre-rent profit of an individual store is total revenue less total cost. If H is the
number of hats sold. P/, is the price of hats, and AC/, is the average cost, total profit is
n =H (Ph - ACh ) (8-8)
For example, if Ph is $8 and AC h is $5, the profit margin is S3. If H is 4.000. total
profit would be SI 2.000.
According to the simple version of central place theory, hat sellers divide the
city into five equal market areas, and each hat seller locates at the center of a market
area. This result can also be expressed in terms of bid-rent functions. The bid-rent
function of a particular hat store depends on where the other stores locate. Locations
close to other hat stores have lower sales volume, so the hat seller is willing to pay
less in rent. The bid-rent function of the hat industry has five peaks, one at the center
of each market area.
Chapter <S ImiuI Use in the Monocentric City 215
Will hat sellers adopt the location pattern predicted by the simple version of
central place theory? The simple version of the theory assumes that unit travel cost
(the cost per mile) is same in all directions. In the monocentric streetcar city, this
the
assumption is violated: the hub-and-spoke streetcar system collects people along
the suburban spokes and delivers them to the central "hub." Travel along the spokes
into the city is cheaper than travel between the spokes, so a trip from a house in the
suburbs to a downtown hatter may be easier than a trip to a suburban hatter. If so.
most — if not all —of the hat sellers locate in the downtown core area.
The tendency for hat sellers to cluster in the core area is reinforced by shop-
ping externalities (explained in Chapters 2 and 5). If hats from different stores are
imperfect substitutes, hat consumers travel to several stores to compare hats, and
shopping cost is lower if the stores are clustered. Since the core area (the hub of the
streetcar system) is accessible to the entire urban area, hatters are likely to cluster
near the city center. If hats and gloves are complementary goods, consumers save
on shopping costs if hat stores are near the glove store in the city center. In terms of
compromise on their central place locations to exploit
central place theory, hat sellers
two types of shopping externalities: the externalities from comparison shopping and
the externalities from one-stop shopping. Given the hub-and-spoke streetcar system,
the retail clusters are likely to be in the downtown core area.
Residential Land Use

This section uses a simple model of the residential sector to explore residential land
use in the monocentric city. According to the leftover principle, the bid rent for
residential land equals the excess of total revenue over total cost. The first step in the
analysis of residential land rent examines the revenue side of housing production. The
housing-price function shows the relationship between housing prices and distance
to the city center.
The simple model of the residential sector has a number of simplifying assump-
tions. Later in the chapter, each of these assumptions will be dropped.
1. One member of each household commutes to a job in the central business

district (CBD).
2. Noncommuting travel is insignificant.
3. Public services and taxes are the same at all locations.
4. Air quality is the same at all locations.
5. All households have the same income and tastes for housing.
6. The opportunity cost of commuting time is zero.
The first four assumptions make the CBD the focal point of city residents. All jobs are
in the CBD, while all the other things that people care about (public services, taxes,
air quality) are distributed uniformly throughout the city. Given the fifth assumption,
the choices of the "typical" household can be used to represent the choices of all
216 Part II Laml Rent and Urban Land-Use Patterns
households in the city. The sixth assumption means that the simple model ignores
the time costs of commuting.
The Housing-Price Function

As explained in Chapter 14 (Why Is Housing Different?), the price of housing is
usually defined as the price per unit of housing service. For the purposes of this
chapter, the price of housing is defined as the price per square foot of housing
per month. If a household rents a 1 ,000-square-foot house for $250 per month, the
price of housing 25 cents per square foot ($250 divided by 1,000 square feet).
is
The housing-price function indicates how much a household is willing to pay for
dwellings at different locations in the city. There are two types of housing-price
functions, linear and convex.
Linear Housing-Price Function: No Consumer Substitution. Figure 8^1 shows

the housing-price function for a simple case described by the following set of
assumptions.
Figure 8-4 Housing-Price Function for a City with

Identical (1,000-square-foot) Dwellings
0.30
N. Housing-price function
0.18
0.06
j\
-1
Miles to city center
The price of housing drops from 30 cents per square foot at the city center to 6 cents
per square fool 2 miles from the city center. The price increases as commuting cost
1
decreases (as distance to the center decreases), making households indifferent

among all locations within the city.
1. Identical dwellings. Every dwelling in the city has 1.000 square feet of
living space.
2. Fixed budget. The typical household has a fixed budget of $300 per month
to spend on commuting and housing costs.
3. Commuting cost. The monthly costs of commuting are $20 per mile per
month: the household pays $20 per month in commuting costs for a
residence one mile from the city center. $40 per month for a residence two
miles from the city center, and so on.
How much is the household willing to pay for dwellings at different locations in
the city? At the city center, commuting costs are zero, so the household can spend its
entire $300 budget on housing. For a 1 ,000-square-foot house, the price is 30 cents
per square foot. At a distance of six miles from the center, commuting costs are
$120, so the household has $180 left to spend on housing (18 cents per square foot).
In Figure 8^1, the slope of the function is two cents per mile.
The negatively sloped housing-price function is necessary for locational equi-
librium. Locational equilibrium occurs when all households are satisfied with their
location choices, that is, no household wants to change its location.
To explain why the equilibrium housing-price function is negatively sloped,
suppose that the function starts out as a horizontal line. If the price of housing is
15 cents per square foot throughout the city, the household can get a 1 .000-square-
foot dwelling anywhere in the city for $150 per month. Suppose that a household
starts out in a dwelling 10 miles from the city center. Because a move toward the
city center decreases commuting costs without affecting rent, the household will
move closer to the city center. Other households have the same incentive to move
closer to the center. As the demand for housing near the city center increases, the
price of housing near the center increases; as the demand for suburban housing
decreases, the price of suburban housing decreases. In other words, the movement
of households toward the city center transforms a horizontal housing-price function
into a negatively sloped function.
The equilibrium housing-price function makes residents indifferent about all lo-
cations because differences incommuting costs are offset by differences in housing
costs. The good news from a one-mile move toward the city center is that commut-
ing costs decrease by t (commuting cost per mile). The bad news is that housing
costs increase by AP (the change in the price per square foot) times H 'hous-
ing consumption in square footage). The household will be indifferent between the
two locations if the decrease in commuting costs equals the increase in nousing
costs:
t = -APH (8-9)
If = $20 and H = 1,000 square feet, the household will be indifferent between
/
the two locations if the price of housing increases by two cents per square foot. In
Figure 8-4, the price of housing increases by two cents per mile as the household
moves toward the city center.
218 Pan II Land Rent ami Urban Land-Use Patterns
Convex Function: Consumer Substitution. The housing-price function in

Figure 8-4 is linear because the city's dwellings are identical: everyone lives in
a 1 .000-square-foot house, regardless of the price of housing. A more realistic as-
sumption is that housing consumption depends on the price of housing. In other
words, households obey the law of demand, decreasing the quantity demanded as
the price increases. As a household moves toward the city center, it pays a higher
price for housing, so it occupies a smaller dwelling. As the relative price of hous-
ing increases, the household substitutes nonhousing goods (pizza, hot dogs, stereo
equipment) for housing.
Figure 8-5 shows the effects of consumer substitution on the housing-price
function. The assumed pattern of housing consumption is shown below the graph.
Suppose that a household moves from a distance of 12 miles (where the price of
housing is 6 cents per square foot) to 9 miles. If housing consumption is fixed at 1 .000
square feet, the household would be willing to pay an additional S60 for housing (the
decrease in commuting cost), or 6 cents more per square foot. Because its housing
Figure 8—5 Housing-Price Function with and without Consumer

Substitution
Price function without

consumer substitution
0.30
Price function with

consumer substitution
0.12
0.06 -
6 9
Miles to cits center
As the household approaches the city center, the price of housing increases. If the household substitute-.
other goods for housing, the housing-price function is convex, not linear.
Assumed Consumption Pattern:

Distance to city center (miles) 3 6 9 12
Housing consumption (square feet) 400 600 750 1.000
consumption drops from 1,000 square feet to 750 square feet, the household is
willing to pay more than an additional 6 cents per square foot to offset the decrease
in commuting costs. In general, as a household moves toward the high-priced city
center, it occupies smaller dwellings, requiring progressively larger increases in
the price per square foot of housing to offset the fixed $20 per mile decrease in
commuting costs. The lesson from Figure 8-5 is that if consumers obey the law of
demand, the housing-price function is convex, not linear.
The slope of the housing-price function can be expressed in simple algebraic
terms. Since both the price of housing (P) and housing consumption (H) vary with
distance to the city center («), the trade-off between commuting and housing costs
can be rewritten as
Am -t = -AP(u)- H(u) (8-10)
At a given location (it), the change in commuting cost (the change in u times the
transport cost per mile) equals the change in the housing price times housing con-
sumption. The equation can be rearranged to show the slope of the housing-price
function:
AP(u)
—=
An
t
H{u)
(8-11)
In the numerical example, t is $20 and H{9) is 750, so the slope of the housing-
price function at u = 9 is $0.0267 (20/750), compared to a slope of $0.02 under the
assumption of fixed housing consumption. Since //(6) is 60, the slope at it is =6
$0,033 (20/600). As the household moves toward the city center, housing consump-
tion decreases, increasing the slope of the housing-price function.
How rapidly does the price of housing decrease as distance to the city center
increases? The housing-price gradient is defined as the percentage change in the
price of housing per mile. Dividing both sides of (8-1 1 ) by P,
AP/P
—=
An
t
H(u)-P(u)
(8-12)
In words, the housing-price gradient equals transport cost per mile divided by housing
expenditures. If the full commuting (including monetary and time costs) is
cost of
$1 per round-trip mile, the monthly commuting cost (for 20 workdays per month)
is $20 per round-trip mile. If the household spends $500 per month on housing, the
rent gradient is 4 percent per mile (20/500).
The Residential Bid-Rent Function

The residential bid-rent function indicates how much housing producers are willing
topay for land at different locations in the city. According to the leftover principle,
housing producers are willing to pay land rent equal to the excess of total revenue
over total cost. There are two types of bid-rent functions, one that occurs if housing is
produced with fixed factor proportions, and one that occurs if housing firms engage
in factor substitution.
The Bid-Rent Function with Fixed Factor Proportions. Consider first the pos-
sibility that housing is produced with fixed factor proportions. The characteristics
of firms in the housing industry are as follows:
1 Production. Each firm produces Q square feet of housing, using land and
other inputs. Once the firm erects a building, it can be used as a single
dwelling (with Q square feet of space), or divided into x units, each of
which has (Q/x) square feet of living space.
2. Nonland cost. Firms use K worth of nonland inputs for each building.
3. Fixed factor proportions. Each firm produces its Q square feet of housing
with T acres of land and K worth of other inputs, regardless of the price of
land.
4. Housing prices. The housing-price function is negatively sloped and

convex.
5. Perfect competition. The markets are perfectly competitive, so the firm
makes zero economic profits.
According to the leftover principle, the bid rent for land equals the excess of
total revenue over total nonland cost. Total revenue equals the price of housing (P)
times Q. and total cost is nonland cost (K) plus land cost {R times T). Since P
varies with the distance to the city center (u), the bid rent for land is
=
P(«)- Q- K
ft(") (8 - ,3)
j
If the price of housing decreases as u increases, the residential bid-rent function is
negatively sloped.
Figure 8-6 maps the residential bid-rent function. The horizontal line is nonland
cost per acre, assumed to be the same at all locations. Since the bid rent equals total
revenue less nonland cost, the bid-rent function below the revenue function, with
lies
the distance between the two equal to the cost of nonland inputs. At u* total revenue .
equals nonland cost, so the bid rent for land is zero. The bid-rent function is convex
because the housing-price function is convex.
The Bid-Rent Function with Factor Substitution. The bid-rent function shown
in Figure 8-6 is produced with fixed factor
based on the assumption that housing is
proportions. Housing firms use the same input combination at all locations, regardless
of the price of land. What happens if firms substitute other inputs for land as the
price of land increases?
Figure 8-7 shows the bid-rent functions for inflexible and flexible housing pro-
ducers. The inflexible firm uses the same input combination throughout the city.
In contrast, the flexible firm substitutes nonland inputs for land as the price of land
increases, building progressively taller buildings as it approaches the city center. The
flexible rent function lies above the inflexible rent function at even location except
u = 6. At this location, the input combination of the inflexible firm is, by chance, the
efficient combination, so the two builders use the same input combination. While the
Figure 8-6 Housing-Price and Bid-Rent Function
Total revenue = Ptu) times Q
Cost of nonland inputs
The bid rent of the housing firms equals total revenue per acre less the cost of nonland inputs.
Total revenue (the price of housing times square footage produced) decreases as the distance to the
city center increases because the housing-price function negatively sloped. The cost of nonland
is
inputs is the same at all locations. The bid-rent function convex because the housing-price
is
function (and the revenue function) is convex. At «*, the cost of nonland inputs equals total
revenue, so the bid rent equals zero.
inflexible firm's input ratio is efficient for u = 6, it is inefficient for other locations
(too low for locations closer to the city center and too high for locations farther from
the city center). For all locations except u = 6, the flexible firm produces housing
for a lower cost and thus outbids the inflexible firm.
Summary: The Convex Bid-Rent Function. There are two lessons from the
analysis of residential land rent. First, the bid-rent function is negatively sloped be-
cause the housing-price function is negatively sloped. Second, the bid-rent function
is convex because of both consumer substitution (which makes the housing-price
function convex) and factor substitution (which increases the convexity of the rent
function).
How rapidly does the price of residential land decrease as distance to the city
center increases? The rent gradient is defined as the percentage change in land rent
(or market value) per mile. The gradient depends on the housing-price gradient ( 1
)
and (2) the relative importance of land in the production of housing. In Table 8-3,
the housing-price gradient is 4 percent (a one-mile move away from the city center
decreases the market value of housing by 4 percent, from $150,000 to $144,000)
222 Pan 11 Land Rent and Urban Land-Use Patterns
Figure 8-7 The Residential Bid-Rent Function with Factor

Substitution
Bid rent with

factor substitution
Factor substitution (substituting nonland inputs for land as the price of land increases) increases the
convexity of the bid-rent function.
Table 8-3 Land-Rent Gradient
Location
/.'
Distance to city center (miles) 4 5

Market value of housing ($) 150,000 144.000
Land value ($) 30.000 24.000
Capital value ($) 120.000 120,000
Assumptions:
1. At location A, land value is 20 percent of the market
value of housing.
2. Housing-price gradient is 4 percent per mile.
and the value of land is assumed to be 20 percent of the total property value at
location A. Because the price of capital is the same at all locations, land absorbs
the entire $6,000 decrease in market value, dropping from $30,000 to $24,000, a
20 percent decrease. Since the market value is simply the present value of the annual
rental income (annual rent divided by the interest rate), the rent gradient (percentage
change in land rent per mile) is 20 percent, or five times the housing-price gradient.
The relationship between the housing-price gradient and the rent gradient can
be stated algebraically as
Rent gradient = •
Housing-price gradient (8-14)
Land's share of house value
The smaller theland's share ofhouse value, the larger tbe percentage decrease in land
rentneeded to absorb a given decrease in the price of h musing. For example, if land's
share of house value is 10 percent, the rent gradient is 10 times the housing-price
gradient.
Residential Density
How does population density vary within the monocentric city? Table 8-4 shows
how compute population density at different locations in the city. The first
to step
is compute the lot size (the amount of land occupied per household). Lot
to size
increases with distance to the city center for two reasons:
1 Consumer substitution. The price of housing decreases as the distance

to the city center increases, and households respond to lower housing
prices by consuming more housing. In Table 8^L housing consumption
rises from 1,404 square feet for a household 0.20 miles from the city center
(location A) to 3,000 square feet for a household four miles from the
center (location B).
2. Factor substitution. The price of land decreases as the distance to the city
center increases, and housing firms respond to lower land prices by using
more land per unit of housing. In Table 8-4, at a distance of 0.20 miles
from the city center, every square foot of living space comes with 0.33
square feet of land. In other words, people live in three-story apartment
buildings. At a distance of four miles from the center, the amount of land
per square foot of housing is 2.20: households live in one-story houses with
lot sizes 2.2 times the "footprint" of the house.
Table 8-4 Population Density at Different Locations
Location
A B
Distance to city center (miles) 0.2 4.0

Housing consumption (square feet) 1.404 3.000
Land per square foot of housing (square feet) 0.33 2.2
Lot size (square feet) 468 6.600
Lot size equals housing consumption (in square feet of living space) times the amount
of land per unit of housing. Because of consumer substitution and factor substitution,
the lot size increases as distance increases: a household located 0.20 miles from
the center uses only 468 square feet of land (sharing the 1,404 square feet under
the three-story apartment building with two other households), while a household
located 4 miles from the center uses 6.600 square feet of land. In this example,
residential density at a location 0.20 miles from the city center is about 14 times the
density 4 miles from the center.
Land Use in the Monocentric City

Figure 8-8 shows the land-use pattern of the monocentric city. The office bid-rent
function intersects the manufacturing function at a distance of u v miles from the
city center, so the office district is a circle with radius u miles. The manufacturing
bid-rent function intersects the residential function at a distance of u,„ miles from
the city center, so the manufacturing district is a ring of width (/<„, — u ) miles. The
residential bid-rent intersects the agricultural bid-rent function at a distance of w/,
miles, so the residential district is a ring of width (//;, — um ) miles. The retail bid-rent
function is omitted from this diagram in the interests of simplicity. As explained

earlier, most retailers congregate at the center of the monocentric city.
Activities are arranged according to their transportation costs: the higher the
transportation cost, the closer to the city center. As explained in Chapter 7, the
activity with relatively high transport costs has a relatively steep bid-rent function,
and thus locates closer to the marketplace. In the monocentric city, the market is
the city center, where office workers meet with clients and manufacturers load their
output onto ships or trains. The office sector, with the highest transport costs and thus
the steepest bid-rent function, occupies land closest to the center. Manufacturing,
with the next highest transport costs and thus the next steepest bid-rent function,
occupies the next ring of land. The residential sector, with relatively low transport
costs and thus a relatively flat bid-rent function, occupies the land farthest from the
city center.
This spatial arrangement has two interesting features. First, office firms occupy
the central area of the CBD. As explained earlier in the chapter, office firms have rela-
tively high transport costs and thus a relatively steep bid-rent function because office
output is transmitted by office workers, while manufacturing output is transported
by horse-drawn wagon.
The second feature of the monocentric city is that employment is concentrated in
the CBD. not distributed throughout the city. Why do all the manufacturers and office
firms locate in the CBD? To explain this monocentric location pattern, consider a
baseball firm that is considering a move from the CBD to a suburban location. What
are the trade-offs associated with a move to the suburbs?
1 . Higher freight costs. The firm will be farther from the central export node,
so it will pay higher freight costs.
Chapter H Land Use in the Monocentric City 225
Figure 8-8 Bid-Rent Functions and Land Use in the Monocentric City
Manufacturer bid-rent function
Residential bid-rent function
Agricultural bid-rent function
Distance from city center
^ Office district
Cj Manufacturing district
Residential district
Q_J
2. Lower wages. The firm will be closer to its workforce, so workers will
commute shorter distances. The wage compensates workers for commuting
costs: the longer the commuting distance, the higher the wage. When the
firm moves closer to its workforce, it decreases its workers' commuting
costs, so the firm can pay a lower wage.
The firm's location decision is determined by the outcome of a tug-of-war. On

one side is the central export node, which pulls the firm toward the CBD. On the
other side is its suburban workforce, which pulls the firm toward the low-wage
suburbs.
In the monocentric city, the tug-of-war was won by the CBD because the cost
of moving freight was large relative to the cost of moving workers. Freight traveled
by horse-drawn wagon, a relatively slow and expensive travel mode. In contrast,
workers traveled by streetcar, a relatively fast and inexpensive travel mode. Although
wages were lower in the suburbs, the wage differential was relatively small because
commuting by streetcar was fast and efficient: workers demanded a relatively small
premium to commute to CBD jobs. If a firm moved to the suburbs, the small savings
in wages would be more than offset by an increase in the cost of shipping freight to
the central export node. In the monocentric city, it was cheaper to bring the workers
from the suburbs to the central-city factory than to bring the output from a suburban
factory to the central export node.
The same analysis applies to office firms. Although a move from the CBD to
the suburbs would decrease wages, it would also increase the costs of travel between
the office and the central market area: managers would spend more time traveling
between the and clients in the city center. Given the frequency of travel to
office
clients in the central market area, the increase in travel costs would dominate the
savings in wages. In the monocentric city, it was cheaper to bring the workers from the
suburbs to a central-city office than to bring the output (in the minds and briefcases
of managers) from a suburban office to clients in the city center.
Relaxing the Assumptions

The simple monocentric model has a number of unrealistic assumptions about res-
idential land use. This section derives the residential bid-rent function under more
realistic sets of assumptions. The assumption that the city is monocentric is main-
tained: all employment is assumed to be in the CBD.
Changes in Commuting Assumptions

The simple model is based on a number of simplifying assumptions about commut-
ing. What happens if these assumptions are dropped?
1 . Time commuting. The simple model assumes that the only cost o\'
cost of
commuting a monetary cost, that is, money spent on cars (for gasoline and
is
maintenance) or public transit (bus tickets). In fact, commuting time comes

at the expense of work or leisure, so there is an opportunity cost associated
with commuting. The unit cost of commuting (/) is actually the monetary
and time costs per mile of travel. Studies of commuting behavior suggest
that most people value commuting time at between one-third and one-half
the wage rate. For a worker with a wage of $10, the time cost of commuting
Chapter X Land Use in the Monocentric City All
is between $3.33 and $5.00 per hour. Commuting costs are discussed in
greater detail in Chapters 19 (Autos and Highways) and 20 (Mass Transit).
2. Noncommuting travel: uniform distribution of destinations. The
simple model assumes that noncommuting travel is insignificant. This
assumption is unrealistic because households travel to different destinations
within the city for shopping and entertainment. Suppose that shopping and
entertainment destinations are distributed uniformly throughout the urban
area. For example, the household commutes northward to a job in the city
center and also travels north to amuseum, south to a grocery store, west to
a disco, and east to the shore. If the frequency and distance of travel to the
four sites are about the same, any change in residence causes a relatively
small change in total noncommuting travel time. If the household moves
south, the cost of museum travel increases, but the cost of the grocery travel
decreases. If the household travels in all directions for shopping and
entertainment, noncommuting costs usually offset one another, and it is
appropriate to focus on commuting as the primary factor in the location
decision.
3. Noncommuting travel: concentrated destinations. Consider next the

possibility that shopping and entertainment sites are concentrated rather
than dispersed. Suppose that members of a household travel to the city
center for work, shopping, and entertainment. As the household moves
it saves on travel costs for commuting, shopping,
closer to the city center,
and entertainment, so the savings in travel cost are relatively large and the
housing-price function is relatively steep. In general, the more frequent the
travel to the city center, the steeper the housing-price function and the
residential bid-rent function.
4. Two-earner households. The simple model assumes that a single person

from each household commutes to the city center. Suppose that all the
households in a city are suddenly transformed into two-earner households.
What happens to the housing-price function? If two members of each
household commute to the CBD and have the same commuting cost per
mile, a household that moves closer to the city center experiences double
the savings in commuting costs. In equilibrium, the housing-price function
and the bid-rent function will be steeper in the two-earner city, reflecting
the greater savings associated with living closer to central-city jobs.
Variation in Tastes for Housing

The simple model assumes that every household has the same tastes for housing.
The housing-price function of the '"typical" resident is used to represent the housing-
price function of the entire city. Suppose two types of households in
that there are
and small, and that all
the city, large households have the same income. The "tastes"
for housing are dictated by the number of children: the small household lives in a
small dwelling, and the large household lives in a large dwelling.
Where two types of households live? Since land is rented to

in the city will the
the highest bidder, the division of residential land between the two household types
is determined by the bid-rent functions of the two groups. As shown in Chapter 7, the
land user with the steeper bid-rent function occupies land closer to the city center.
Since the residential bid-rent function is determined by the housing-price function,
the user with the steeper housing-price function has a steeper bid-rent function.
Which household has the steeper housing-price function? The expression for
the housing-price function is
AP(u)
(8-15)
Aw H(u)
If the two households have the same commuting cost per mile (t), the small household
has a steeper housing-price function because it consumes less housing (smaller H).
Because the small household consumes a smaller amount of housing in square feet), (
it takes a larger change in the price of housing per square foot to compensate for an
increase in commuting cost.

Figure 8-9 shows the bid-rent functions for the two households. The bid-rent
function of the small household is steeper because its housing-price function is
steeper. The two functions intersect at a distance of four miles from the city center.
Figure 8-9 Bid-Rent Function and Family Size
Bid-rent function of the small household
Bid-rent function of the large household
Miles from city center
The slope of the bid-rent function decreases as housing consumption increases, so

large households, who live in large houses, have relatively Bat bid-rent functions.
Large families have the most to gain from low suburban housing prices, so they outbid
small households for suburban bousing.
Chapter 8 Land Use in the Monocentric City 2.2.9
so small households occupy dwellings within four miles of the city center, and large
households occupy dwellings outside the four-mile radius. Large households occupy
low-price suburban housing because they live in large houses and thus have more to
gain from inexpensive suburban housing.
Spatial Variation in Public Goods and Pollution

The simple model has a number of assumptions that make the city center the focal
point of the city. Except for jobs in the CBD, all the things that people care about
(public services, taxes, pollution, amenities) are distributed uniformly throughout
the city. What happens if these things are not distributed uniformly?
Public Goods and Taxes. Taxes and public services vary within a metropolitan
area. Suppose that the quality of public schools varies within the city, but the cost
of schools (tuition and taxes) is the same throughout the city. In equilibrium, the
price of housing is higher in the communities with better schools. Parents pay for
better public schools indirectly: instead of paying higher taxes, they pay more for
housing and residential land. Similarly, the prices of housing and land are higher
in communities with lower crime rates. The same argument applies to variation in
two communities have the same level of public services but one community
taxes. If
has higher taxes, the price of housing is higher in the low-tax community.
Pollution and Amenities. The simple model assumes that environmental quality
is the same at all locations in the city. To explain the effects of pollution on housing
and land prices, suppose a polluting factory moves into the center of a previously
clean city. If the smoke and smell from the factory are heaviest in the central area
of the city, the factory decreases the relative attractiveness of dwellings near the
city center, decreasing the price of housing. In addition, the factory increases the
relative attractiveness of more remote dwellings, increasing the price of suburban
housing.
Figure 8-10 shows the effects of the polluting factory on the housing-price
function. Pc is the price function in the absence of pollution (the clean city), and
P s is the price function with a small amount of pollution. The pollution from the
As a household
central-city factory decreases the slope of the housing-price function.
moves toward the city center, there are costs (more pollution) as well as benefits
(lower commuting costs), so in the polluted city, the price of housing increases less
rapidly as one approaches the city center. If the city has a high level of pollution,
the housing-price function may be positively sloped, as shown by Pj. In this case,
central-city pollution is so obnoxious that the advantages of a central-city dwelling
(lower commuting costs) are dominated by its disadvantages (greater exposure to
pollution). As a result, people are willing to live near the city center only if they are
compensated in the form of lower housing prices.
Changes in the housing-price function cause similar changes in the residential
bid-rent function. A relatively flat housing-price function (Ps generates a
) relatively
flat bid-rent function. Similarly, a positively sloped housing-price function (P,/)
generates a positively sloped residential bid-rent function.
Figure 8-10 Pollution in the Central City and Housing Prices
Price function without pollution
Price function with mild pollution
Price function w ith severe pollution
Air pollution from a central-city factory decreases housing prices near the city center and increases housing prices
far from the city center. The more severe the pollution, the greater the change in housing prices.
The same arguments apply to locations that have positive locational attributes
(amenities) such as scenic views or access to parks. If people get utility from scenic
views or park access, they are willing to pay more for dwellings that provide such
amenities.
Income and Location

In U.S. cities, the wealthy tend to locate in the suburbs, and the poor tend to locate near
the city center. In other words, average household income increases as one moves
away from Because the most expensive land is near the city center,
the city center.
this location pattern is puzzling: why should the poor occupy the most expensive
housing and land? There are several theories of this observed pattern of income
segregation. The first is based on the simple monocentric model, and the others are
based on extensions of the monocentric model.
Trade-off between Land and Commuting Costs
According to the simple monocentric model, a household chooses the location that
provides the best trade-off between land costs and commuting costs. One theory of
income segregation, developed by Alonso (1964) and Muth (1969). suggests that
Table 8-5 Trade-Off's between Land Cost and

Commuting Cost
Distance to
City Center Land Rent Decrease in Land Marginal Marginal
(miles) pel Acre Land Rent lucres) Benefit Cost
$3,800
1 3.100 $700 0.2 $140 $40
2 2. 500 600 0.2 120 40
3 2.000 500 0.2 100 40
4 1 .600 400 0.2 80 40
5 1,300 300 0.2 60 40
6 1,100 200 0.2 40 40
7 1.000 100 0.2 20 40
central locations provide the best trade-off for the poor, while suburban locations
provide the best trade-off for the wealthy.
Table 8-5 shows the trade-offs between land costs and commuting costs for a
household with the following characteristics:
1. The household takes the residential land-rent function as given. The second
column of the table shows the land rent (per month per acre) for different
locations, and the third column shows the changes in land rent for one-mile
moves away from the city center.
2. Land consumption by the household is 0.20 acres, regardless of location
column in the table).
(the fourth
3. Commuting cost is $40 per round-trip mile per month.
The marginal benefit of distance, defined as the decrease in the household's land
cost from a one-mile move outward, equals the decrease in land rent times land
consumption. For example, a one-mile move away from the city center decreases
land rent per acre by $700 and decreases land cost by $140 (0.20 times $700). The
marginal benefit decreases as we move down the table because land rent falls at a
decreasing rate: the land-rent function is convex. The marginal cost of a one-mile
move outward, defined as the increase in commuting cost, equals the commuting
cost per mile per month Suppose that the household tentatively decides to live
($40).
in the city center. Given the numbers in the table, a one-mile move outward would
decrease land cost by more than it would increase commuting cost ($ 40 versus $40), 1
so a central-city location is clearly inferior to a location one mile from the city center.
The optimum location is where the marginal benefit from a one-mile move
outward (the savings in land cost) equals the marginal cost (the increase in commuting
cost). In Table 8-5, the optimum location is six miles from the city center. At any
location closer to the center, the marginal benefit exceeds the marginal cost, so the
household will be better off at a more distant location. At six miles, the marginal
benefit equals the marginal cost.
232 Part 11 Land Rent and Urban Land-Use Patterns,
Fig IR = 8-11 Trade-Offs between Land Cost and Commuting Cost

$
140
Marginal savings in land cost
120 -
100 -
80
60
Marginal commuting cost ^>»«.
40
MC
MB
20
1 1 1 1 !
1 2 3 4 5 6
Distance to city center (miles)
The optimum location is benefit of distance (MB) equals the marginal cost (MC ). The
where the marginal
marginal benefit equals the decrease in land rent times land consumption. The marginal-benefit curve is
negatively sloped because the land-rent function is convex. The marginal cost equals the increase in
commuting cost per mile.
shows the benefit and cost curves from Table 8-5. The optimum
Figure 8-1 1
location where the marginal-benefit curve intersects the marginal-cost curve. The
is
position of the marginal-benefit curve is affected by the household's land consump-

tion: the larger the lot, the larger the benefit associated with lower land rent at more
remote locations. In Figure 8-1 1, an increase in land consumption shifts the benefit
curve upward, increasing the optimum distance. The position of the cost curve is
determined by commuting costs: an increase in commuting costs shifts the cost curve
upward and decreases the optimum distance.
Under what circumstances will the poor locate near the central city, while the
wealthy locate in the suburbs? Figure 8-12 shows the benefit and cost curves under-
lying the Alonso-Muth theory of income segregation. Consider the location choice
of a poor household with the following characteristics:
1 The household has one-fifth the income of the wealthy household whose
characteristics are shown in Table 8-5.
2. The poor household consumes one-fifth as much land as the wealthy
household (0.04 acres). In other words, land is a normal good, with an
income elasticity of demand (the percentage difference in land
consumption divided by the percentage difference in income) equal to 1.0.
Figure 8-12 The Alonso-Muth Theory of Income Segregation
MB for wealthy household
MC for wealthy household
3 4
If theincome elasticity of demand for land is large relative to the income elasticity of commuting cost, the gap
between the marginal-benefit curves will be larger than the gap between the marginal-cost curves. Therefore,
the poor live near the central city, and the wealthy live in the suburbs.
3. The commuting is 70 percent the commuting

cost of the poor household
cost of the wealthy household ($28 per month per mile). The poor
household has a lower commuting cost because it has a lower wage and
thus a lower opportunity cost of commuting.
In Figure 8-12, the optimum location for the wealthy household is six milesfrom
the city center, and the optimum location for the poor household is one mile from
the center.
What
are the assumptions underlying this theory of income segregation? The
benefit and cost curves in Figure 8-12 are drawn under the assumption that the
income elasticity of demand for land is large relative to the income elasticity of
commuting cost (the percentage difference in commuting cost divided by the per-
centage difference in income). Although both land consumption and commuting cost
increase with income, the increase in land consumption is relatively large. Therefore,
the gap between the two marginal-benefit curves is larger than the gap between the
marginal-cost curves, so the poor occupy central-city housing.
Wheaton (1977) provides empirical evidence that questions the validity of the
Alonso-Muth model of income segregation. His results suggest that the income elas-
ticity of demand for land equals the income elasticity of commuting cost. Therefore,
an increase in income shifts the benefit and cost curves upward by about the same
Figure 8-13 Location Choices Using Wheaton's Results
MB for wealthy household
MC for wealthy household
3 4
If the income elasticity of demand for land equals the income elasticity of commuting cost, the gap between the
marginal-benefit curves equals the gap between the marginal-cost curves (in percentage terms). Therefore, the
simple monocentric model predicts that location choices are unaffected by income: both households in the
example pick a location six miles from the city center.
amount (in percentage terms). In Figure 8-13, the poor household (with one-fifth
the income of the wealthy household) has half the land consumption and half the
commuting cost of the wealthy household. There is a 50 percent gap between the
benefit curves of the two households, and the same gap between the cost curves, so
the optimum location for both households is six miles from the city center. This re-
sult suggests the observed locational pattern (poor central-city residents and wealthy
commuting cost and land
suburbanites) cannot be explained by the trade-off between
Wheaton's results suggest that one must look beyond the simple monocentric
cost.
model to explain the observed pattern of income segregation.
Other Explanations
Some alternative explanations of income segregation are based on factors excluded

from the simple monocentric model.
1. New suburban housing. Suppose that the utility generated from a

particular dwelling decreases over time as the dwelling deteriorates and
becomes obsolete. In other words, people get less utility out of an older
house because it is less fashionable, has higher maintenance costs, and is
equipped with fewer modern gadgets. The wealthy, who demand

high-quality housing, occupy new housing instead of used housing. As an

urban area grows, expands outward, and developers build new housing
it
for high-income households in the peripheral areas. The poor are left with
old houses in the central city.
2. Fleeing central-city problems. As explained later in the book, poverty
contributes to three urban problems. First, crime rates are higher among the
poor, in part because the poor face a relatively low opportunity cost of
committing crime. Second, fiscal problems are more likely in a jurisdiction
with a large fraction of low-income citizens. Third, students from poor
families have relatively low achievement levels and pull down the
achievement levels of other students. To escape these problems, wealthy
households flee to the suburbs, leaving large concentrations of poor
households behind.
3. Suburban zoning. As explained in Chapter 1 1 (Land-Use Controls and
Zoning), suburban governments use zoning to exclude low-income
households. Therefore, only the wealthy have the opportunity to escape the
problems of the central city.
Income and Location in Other Countries

The U.S. pattern of income segregation — with high-income households tending
to live in the suburbs — is often reversed outside the United States. According to
Hohenberg and Lees (1985), "incomes rose with distance to the city center in Amer-
ica, whereas they typically fell in Europe." A good example of the European pattern
is the Paris metropolitan area, where the average income in the central city exceeds
the average income in the surrounding suburbs. The same pattern is observed in
many Latin American countries. Why do we observe a different income-location
pattern outside the United States?
Brueckner, Thisse, and Zenou (1996) suggest that differences in cultural ameni-
ties explain these different location patterns. Consider the difference between Detroit
and Paris. Paris has a rich mixture of museums, restaurants, parks, and a thriving
street life, and all of these cultural amenities make central Paris attractive relative
to the suburbs. If the demand for these cultural amenities increases rapidly with in-
come, the forces pulling the richtoward the central city (access to jobs and cultural
amenities) are more likely to dominate the forces pulling them toward the suburbs
(lower prices of land and housing). As a result, a relatively large fraction of high-
income households will locate in central Paris. In contrast, Detroit has few cultural
opportunities in the city center, so the amenity forces pulling the rich toward the
center are relatively weak, and fewer rich households will locate near the center.
Income and the Residential Bid-Rent Function

The issue of income segregation can also be explained with the housing-price func-
tion and the residential bid-rent function. As explained earlier in the chapter, the
activity with the steeper bid-rent function occupies land closer to the city center.
Figure 8-14 Bid-Rent Function and Income

s
Bid-rent function of the poor household
Bid-rent function of the wealthy household
Miles from city center
The wealthy have a flatter bid-rent function, so the poor occupy central land (land less
than u' miles from the city center). The wealth) have a flatter bid-rent function because
they are sensitive to crime, pollution, and the quality of schools, and the central cities
7
have more crime and pollution, and inferior schools.
Since the residential bid-rent function is determined by the housing-price func-

tion, the poor occupy central land if they have a steeper housing-price function.
Figure 8-14 shows two residential bid-rent functions, one for low-income housing
and one for high-income housing. The low-income function is steeper because the
poor have a steeper housing-price function.
Why do the poor have a steeper housing-price function? In the simple mono-
centric model, the expression for the slope of the housing-price function is
AP
(8-16)
Ah Hiu)
An increase in income increases both t (the opportunity cost of commuting) and H

(housing consumption), so rising income has an ambiguous effect on the slope of the
housing-price function. If the increase in H
in r (if the income
exceeds the increase
elasticity of demand income elasticity of commuting cost),
for housing exceeds the
the wealthy have a flatter housing-price function and a flatter residential bid-rent
function. This is the Alonso-Muth theory of income segregation.
Wheaton's results suggest that the slope of the housing-price function is inde-
pendent of income. If H and t increase at the same rate as income increases, the two
Chapter 8 Land Use in the Monocentric City 12TI
housing-price functions have the same slope, so the two residential bid-rent functions
have the same slope. Therefore, if the poor have a relatively steep bid-rent function,
it is not because of the trade-off betweencommuting costs and housing costs.
The income segregation suggest that the slope of
alternative explanations of
the residential bid-rent function is affected by other factors. Specifically, if central
cities have higher taxes, inferior schools, and more pollution and crime, households
are willing to pay more for housing and land in the suburbs. In other words, the
problems of the central city decrease the slope of the bid-rent function. If the income
elasticities of demand for safety, clean air, and education are relatively large, the
bid-rent function of wealthy households will be flatter than the bid-rent function of
poor households. In other words, if the wealthy are willing to pay much more than
the poor for safety, clean air, and superior education, wealthy households will outbid
poor households for land in areas that are relatively safe and clean and provide high-
quality education. Wasylenko (1984) summarizes the empirical evidence supporting
these alternative explanations of income segregation.
Policy Implications
These alternative theories of income segregation suggest that public policy can af-
fect the location choices of wealthy and poor households. A housing policy that en-
courages the renovation of central-city housing stock may cause some high-income
households to return to the central city. Policies that decrease poverty decrease crime
rates, reduce fiscal problems, and improve central-city schools, encouraging high-
income households to live in the central city. Similarly, policies that address the
crime and education problems directly increase the relative attractiveness of central-
city locations. Finally, policies that control exclusionary zoning allow the poor to
move to the suburbs.
Empirical Estimates of Rent and Density Functions

This section discusses empirical studies of land rent and land use in the monocentric
city. The studies are based on data from the early twentieth century, during the
heyday of the monocentric city. Chapter 10 (Suburbanization and Modern Cities)
discusses land-use patterns in modern cities. One of the key questions in Chapter 10
is: how "monocentric" are modern cities? In other words, are the patterns of land
rent and land use in modern cities roughly consistent with the patterns predicted by
the monocentric model?
Estimates of the Land-Rent Function
A number of researchers have estimated the relationship between land rent and
969) used data collected by Homer Hoyt to estimate
distance to the city center. Mills ( 1
the relationship between land value and distance. As explained in Chapter 7, land
Figure 8-15 Value of Land in Chicago in 1928
140 -
120 -
100
so
60
40
20 -
.20 1.4 2.6 3.8 5.0 6.2 7.4 8.6 9.8
Source: Edwin S. Mills, "The Value of Urban Land." in The Quality of the Urban Environment, ed. H. Perloff
(Washington. D.C.: Resources for the Future. 1969).
value is the present value of land rent, so it's easy to make the translation from value
to rent. Mills assumes the following relationship between value and distance:
V(u) = B e (8-17)
where
V(u) = Value of land u miles from the city center

B = parameter to be estimated from the data
e = Base of the natural logarithm
c = parameter to be estimated from the data
Figure 8-15 shows the estimated relationship for Chicago in 1928. when the
city was monocentric. The value of land drops from about $140, 000 per acre at the
city center to about $ 4,000 at mile from the center, to about $ 7.000 at 10 miles
1 1 1 1
from the city center. The value of land falls by 21 percent per mile, that is. the rent
gradient is 21 percent.
Estimates of the Population-Density Function
How does population density vary within the monocentric city? The density func-
tion describes the relationship between population density and distance to the city
Kh 8-16 Population Density in Baltimore in 1920
70
60
50
40
30
20
^\^^ Population density

10
I 1 1 1 1 1 1 1 1
1.4 2.0 2.6 3.2 3.8 4.4 5.0
Miles to city center (u)
The density function is negatively sloped because (1) housing consumption increases with u (a result of declining
housing prices) and (2) land per unit of housing increases with u (a result of declining land prices).
SOURCE: Edwin S. Mills, Studies in the Structure of the Urban Economy (Baltimore: Johns Hopkins. 1972).
center. Mills (1972) has estimated the density functions for 18 metropolitan areas
for different years. The assumed relationship between density and distance is
-gu
D(u) =A e (8-18)
where
D(u) = Population density u miles from the city center (people per square
mile)
A — Parameter to be estimated from the data
e — Base of the natural logarithm
g — Parameter to be estimated from the data
Figure 8-16 shows the estimated relationship for Baltimore in 1920, when the
city was monocentric. Population density drops from about 60,000 people per
square mile at a distance of 0.20 miles from the city center, to about 34,000 at
a distance of 1 mile, to about 4,200 at a distance of 4 miles. The density gra-
dient, defined as the percentage change in population density per mile, is about
70 percent.
240 Part II Land Rent and i 'rban Land-Use Patterns
Summary
1 The monocentric city has the following characteristics:
a. All manufacturers export their output through a central export node.
b. Manufactured goods are transported within the city by horse-drawn wagon.
c. Office workers travel by foot from offices to a central market area to
exchange information.
d. Commuters and shoppers travel on a hub-and-spoke streetcar system.
2. The manufacturing bid-rent function is negatively sloped because transport
cost is lower near the export node. It is convex because of factor substitution.
3. The office bid-rent function is negatively sloped because travel costs are lower
near the central market area. It is convex because of factor substitution.
4. Transport costs in the monocentric city are relatively high for office firms, so
the office bid-rent function is relatively steep and office firms occupy the
central area of the city.
5. The depend on scale economies, per capita

location choices of retailers
demand, and shopping monocentric city, the
externalities. In the
hub-and-spoke streetcar system makes central locations accessible to the entire
urban area, causing most retailers to locate there.
6. In the simple monocentric model, all employment is in the CBD. and other
things that residents care about (e.g.. public services, taxes, pollution) are the
same at all locations within the city.
7. The housing-price function shows the price of housing (per square foot of
living space) for different locations in the city. The function is negatively
sloped because commuting costs increase with the distance to the city center.
It is convex because of consumer substitution: as the price o\' housing rises.
consumers substitute other goods for housing.
8. The residential bid-rent function shows the amount housing producers are
willing to pay for residential land at different locations in the city. It is
negatively sloped because the housing-price function is negatively sloped. It is
convex because of consumer substitution (which makes the housing-price

function convex) and factor substitution.
9. The density function shows the number of people per acre for different
locations in the city. It is negatively sloped for two reasons.
a. Housing prices arc higher near the city center, so housing consumption is
lower (fewer square feet of living space per household).

b. Land prices are higher near the city center, so the amount of land per square
foot of housing is lower.
10. Activities in the monocentric city are arranged according to their transportation
cost: the higher the transportation cost, the closer to the city center.
a. The city center is occupied b\ the o\Y\cc sector rather than the
manufacturing sector because office output is transmitted by high-cost office
workers, while manufacturing output is transported b\ horse-drawn wagon.
b. Employment is concentrated in the CBD because the cost of commuting

(from the suburbs to the CBD factories and offices) is low relative to the
cost of moving output (from the suburbs to the city center). Freight travels
by horse-drawn wagon, a relatively slow and expensive travel mode. In
workers travel by streetcar, a relatively fast and inexpensive travel
contrast,
mode.
1 1 If the assumptions of the simple monocentric model are dropped, the
housing-price function and the residential bid-rent change.
a. The opportunity cost of commuting is the value of forgone work or leisure
time. The typical commuter values commuting time at between one-third
and one-half of his or her wages. The unit cost of commuting is the sum of
time and monetary costs.
b. Noncommuting travel to the CBD increases the slopes of the housing-price
and bid-rent functions because total travel costs increase more rapidly as
the household moves away from the CBD.
c. The two-earner household has a relatively steep housing-price function if
both earners commute to the CBD.
d. The prices of housing and land are higher in communities with superior
local public goods and low taxes.
e. The prices of housing and land are higher in communities with clean air,
scenic views, and access to parks.

12. The slopes of the housing-price and land-rent functions depend on housing
consumption. A household that occupies a relatively large house has a
relatively flat housing-price function and a relatively flat bid-rent function, so
the household lives relatively far from the city center. Such a household has
more to gain from inexpensive suburban housing.
13. In U.S. cities, the wealthy tend to locate in the suburbs, and the poor tend to
locate near the city center. One theory of income segregation suggests that the
location choices of wealthy and poor households are based on different
trade-offs between commuting and land costs. This theory has been refuted by
empirical evidence, suggesting that the observed pattern of segregation is
caused by other factors, such as the demand for new suburban housing, the
desire to escape central-city problems, and exclusionary zoning in the suburbs.

1 . In the city of Trekburg, manufacturers have two options for intracity freight.
They can use a conventional transportation system (the truck) or a matter
transmitter, which instantly transports the output from the factory to the central
export node ("Beam it over, Scotty"). A transmitter can be rented for C per
year, and running the machine is costless. The transmitter can transport output
up to a distance of two miles. All manufacturing output goes through the
export node.
a. Draw the bid-rent function for a firm that uses the matter transmitter, and
label it M.
b. On the same graph, draw the bid-rent function for a firm that uses the truck,
and label it T.
c. Will every manufacturer use the matter transmitter? If not. where will the
firms using the truck be located?
2. Consider a traditional nineteenth-century monocentric city with a CBD radius

of one mile. In 1869, buildings at the edge of the CBD are four stories tall. In
1870, all the buildings in the CBD are destroyed by an earthquake, and the
mayor of the city announces that the maximum building height in the rebuilt
city will be four stories. In addition, business development will be confined to
a circle with a one-mile radius; that the size of the CBD is fixed. Suppose
is.
that the city is smallenough that events in the city do not affect the equilibrium
prices of its export goods (office or manufacturing goods).
a. Draw the business bid-rent function before the earthquake and label it R .
Draw the business bid-rent function after the earthquake (with the height
restrictions) and label it R*.
b. Explain any differences between the two bid-rent functions.
c. Suppose that the city is large enough that events in the city affect the price
of its exports. Will the height restrictions increase or decrease the price of
the goods? What are the implications of the change in the price of goods on
the business bid-rent function?
3. Complete the following table, given the following assumptions:

The office firm produces 100 consultations per month.
i.
The consultation fee is $75.

/'/'.
Hi. Travel time is five minutes per block.

iv. The opportunity cost of travel time is $3 per minute.
v. Every consultation requires one trip to the city center.
Distanct to
City Center Size oj Sue Total Nonland Travel Pre-Rent Rent per
1 miles I (acres ) Revenue Cost Cost Profit Acre
0.40 $3,600
1 0.70 2.000
2 0.90 1.200
3 ! .00 900
Consider an office firm with the following characteristics: the wage o\'
executives is $120 per hour, and the executive takes lour minutes to walk one
block (eight minutes to make a round trip): the price of output is $150. and the
firm produces 50 consultations (requiring 50 trips to the city center): at a
location four blocks from the city center, the firm occupies a one-acre site and
spends $1,000 on nonland inputs.
a. What is the travel cost per block?

b. How much is the firm willing to pay for land four blocks from the city
center?
c. Given the available information, is it possible to compute how much the
firm is willing to pay for land one block from the city center? If not, what
additional information do you need?
5. Consider two monocentric cities: Rigid City, where office firms produce with
fixed factor proportions, and Flexville, where office firms produce with
variable factor proportions. In each city, the CBD is a circular area with a
radius of one mile, and all land in the CBD is used for office space. At the edge
of the CBD there is 5,000 square feet of office space per acre, and the bid rent
for office land is $20,000 per year. Suppose that each city imposes an annual
tax of $1 per square foot of office space. Assume that the cities are small
enough by events in
that the equilibrium price of office services is unaffected
draw the office bid-rent function before and after the
the cities. For each city,
new office tax. Provide numbers for the bid rents (pre-tax and post-tax) at the
CBD edges. Explain any differences in the effects of the tax in the two cities.
6. Depict graphically the effects of the following changes on the division of CBD
land between office firms and manufacturers:
a. The unit freight cost decreases.
b. The price of office output increases.
c. The opportunity cost of executive travel decreases.
7. Consider an industry that makes table tennis balls and competes with the
baseball makers for land near the central export node. Each table tennis ball
firm produces the same amount of output as a baseball firm (five tons of balls),
sells for the same price ($160 per ton), and has the same production isoquants.
Which activity will locate closer to the export node?
8. Consider a monocentric city in which the unit cost of commuting is $10 per
mile per month. A household located eight miles from the city center occupies
a dwelling with 1,200 square feet at a monthly rent of $600. Nonland cost per
dwelling is $200, and there are four houses per acre.
a. What is the price (per square foot) of housing at u — 8? What is the bid rent
at u = 8?
b. Assume that the demand for housing is perfectly inelastic. What is the price
of housing at u =5?
c. Assume that housing firms do not engage in factor substitution. What is the
bid rent at u =5?
d. How would your answers to (b) and (c) change if the demand for housing is
and firms engage in factor substitution? Would the prices of

price-elastic
housing and land be larger or smaller?
9. Choose one word in each set of parentheses to make the following statements
correct, and then explain your choice of words.
a. "The flatter the demand curve for housing, the {more, less) curvature in the
housing-price function. In other words, the flatter the housing demand
curve, the (more, less) convex the housing-price function."
b. "The flatter the demand curve for housing, the (more, less) curvature in the
residential bid-rent function."
10. Consider a region with two cities: Law land (L) and Violateville (V). The two
demand curves for housing: consumers in Lawland have

cities differ in their
negatively sloped demand curves: consumers in Violateville have positiveh

sloped demand curves (consumers in Violateville actually consume more
housing as the price of housing increases). Draw the housing-price functions
for the two cities (labeled PL for Lawland and P V for Violateville) under the
assumption that PL — P V at a distance of five miles from the city center.
Briefly explain any differences between the two housing-price functions.
1 1 Depict graphically the effects of the following changes on the equilibrium
housing-price function:
a. The workweek is shortened from five days per week to four days per week.
b. The workers in tw o-earner households start riding to work together.
1 2. Suppose that a city restricts the heights of residential structures. The maximum
height is four stories, the height that would normally occur at a distance of
five miles from the city center. Draw two residential bid-rent functions, one for
the city in the absence of height restrictions and one with height restrictions.
13. Between 1940 and 1965. the average household size increased dramatically.
Draw two housing-price functions (one for 1940 and one for 1965 and explain )
the differences between the two functions. Could the increase in household
size explain part of the suburbanization that occurred between 1940 and 1965?
1 4. Consider the example of the trade-offs associated w ith location choices in

Table 8-5. Suppose the poor household (with one-fifth of the income of the
wealthy household) consumes 0.15 acres of land and has a month!)
commuting cost of S 1 5 per mile.
a. What are the implied income elasticities of land consumption and
commuting cost?
b. What is the optimum distance for the poor household?
c. What are the implications for income segregation' 1
15. Suppose that the income demand for land is +0.75. The unit
elasticity of
commuting sum of monetary cost (30 cents per mile)
cost (cost per mile) is the
and time cost (opportunity cost). Suppose that the typical commuter earns a
wage of SI 2 and takes 30 minutes to commute 10 miles to work. Every worker
values commuting time at half of his or her wage. Can the observed pattern of
income segregation be explained by the trade-offs between commuting cost
and land cost? If there's not enough information to answer the question, what
additional information do you need, and how would you use it?
16. Suppose that the demand for housing H = square feet of housing space per
<
capita)and the demand for land (T = square feet of land per square foot of
housing space) are described by the following equations P = price of (
housing, and R = land rent per square foot):
H= 1,500-500- P
T = (15,000- R)
10,000
Compute land consumption per capita for the following locations:

a. Location A: P = 1.5 and R = 12,000
b. Location B: P = 1.0 and R = 10,000
c. Location C: P = 0.3 and R = 3,000
17. In Figure 8-5, the introduction of consumer substitution increases the price of
housing at all locations. Comment on "Something is wrong
the following:
here. As we move from a world without substitution to a world with
substitution, overall demand for housing decreases. Yet Figure 8-5 shows an
increase in the price of housing. How can price rise when demand falls?"

Location of Office Firms
Clapp, J. M. "Endogenous Centers: A NUE Model." Papers of

Simple Departure from the
the Regional Science Association 54 (1984), pp. 13-24. Describes the process by
David, Philip. Urban Land Development. Homewood, 111.: Richard D. Irwin, 1970.
Discusses the rent function of commercial firms.
O'Hara, D. J. "Location of Firms within a Square Central Business District." Journal of
PoliticalEconomy 85 (1977), pp. 1 189-207. Discusses how interfirm contacts cause
thedevelopment of a central business district.
Tauchen, H., and Anne Witte. "An Equilibrium Model of Office Location and Contact
Patterns." Environment and Planning A 15 (1983), pp. 131 1-26. Discusses how
interfirm contacts cause the development of a central business district.
."Socially Optimal and Equilibrium Distributions of Office Activity: Models with
Exogenous and Endogenous Contacts." Journal of Urban Economics 15 (1984),
pp. 66-86. Discusses two sources of market failure in the location decisions of office
firms: agglomeration economies and transaction externalities.
Transportation, Land Rent, and Land Use

Alcaly, Roger E. "Transportation and Urban Land Values: A Review of the Theoretical
Literature." Land Economics 52 (1976), pp. 42-53. Discusses the effects of changes in
transport cost on land rent.
Meyer, J.; John Kain; and M. Wohl. The Urban Transportation Problem. Cambridge,
Mass.: Harvard University Press, 1965. Discusses the effects of the trend toward
two-earner households and noncommuting travel on the residential rent function.
Housing-Price Function
Jackson. Jerry. "Intraurban Variation in the Price of Housing." Journal of I than Economics
6 (1979). pp. 465-79. Estimates the housing-price function, finding that housing prices
fall by about 2 percent per mile.
Kain. John F. and John M. Quiglev. "Measuring the Value of Housing Quality." Journal of
American Statistical Association 65 1970). pp. 532—38. Estimates the relationship
the (
between various housing characteristics (including location) and the price of housing.
King, Thomas. "The Demand for Housing: Integrating the Roles of Journe) to Work.
Neighborhood Quality, and Prices." In Household Production and Consumption, ed.
Nester Terleckyj. New York: National Bureau of Economic Research. 1975. Estimates
the relationship between various housing characteristics (including location) and the
price of housing.
Quiglev. John M. "Housing Demand in the Short Run: An Analysis of Polytomous Choice."
Explorationsin Economic Research 3 1976). pp. 76-102. Estimates the effects of
(
location on the implicit prices of different components of housing.

Stras/heim. Mahlon. An Economic Analysis of the Urban Housing Market. New York:
National Bureau of Economic Research. 1975. Estimates the effects of location on the
implicit prices of different components of housing.
Empirical Studies of Population Density and Land Rent
Mills. Edwin S. Studies in the Structure of the Urban Economy. Baltimore: Johns Hopkins.
1972. Chapter 3 estimates population and employment density functions for U.S. cities.
. "The Value of Urban Land." In The Quality of the Urban Environment, ed. H. Perloff.
Washington. D.C.: Resources for the Future. 1969. Estimates the land-rent function in
Chicago for 1836 to 1954.

Muth. Richard. Cities and Housing. Chicago: University of Chicago Press. 1969. A classic
studv of residential location decisions in Chicago. Chapter 7 estimates the density
function for several U.S. cities.
Income and Location

Alonso. William. Location and Land Use. Cambridge. Mass.: Harvard Universit) Press.
1964. Expands the Von Thunen model to the location decisions oi' households.
Brueckner. Jan K... Jacques-Francois Thisse. and Yves Zenou. "Whv Is Central Paris Rich
and Downtown Detroit Poor? An Amenity-Based Theory. Working Paper ( 1996).
Downs. Anthony. Urban Problems ami Prospects. Chicago: Markham. 1970. Suggests that
the building of new housing on the periphery of the metropolitan area contributes to the
suburbanization of high-income households.
Hohenberg. Paul M.. and Lyann H. Lees. The Making of Urban Europe 1000-1950.
Cambridge. Mass: Harvard Universit) Press. 1986.
LeRov. Stephen, and Jon Sonstelie. "Paradise Lost and Regained: Transportation
Innovation. Income, and Residential Segregation." Journal of Urban Economics 13
(1983). pp. 67-89. Discusses the effects of different transport modes on the location
patterns of different income groups.
Muth. Richard. Cities and Housing. Chicago: Universit) of Chicago Press. 1060. A classic
stud) of residential location decisions in Chicago. Chapter 2 models the location
decision, and Chapter 10 provides empirical evidence that suggests that the tendency
for higher-income households to locate farther from the city center is caused by
differences in the trade-offs between housing and commuting costs.
Wasylenko. Michael J. "Disamenities, Local Taxation, and the Intrametropolitan Location
of Households and Firms." In Research in Urban Economics, vol. 4, ed. Robert Ebel.
Greenwich, Conn.: JAI Press, 1984. Reviews the empirical evidence concerning the
effects of income on location. Also reviews the evidence concerning the
intrametropolitan location choices of firms.
Wheaton. William. "Income and Urban Residence: An Analysis of Consumer Demand for
Location." American Economic Review 67 1 977), pp. 620-3 1 Suggests that the
( .
income elasticity of demand for land is close to the income elasticity of time cost,
meaning that the tendency for the poor to locate near the city center cannot be
explained by the simple monocentric model.
Miscellaneous
Clawson, Marion. "Urban Sprawl and Speculation in Suburban Land." In Urban Economic-
Issues, ed. Stephen Mehay and Geoffrey Nunn. Glenview, 111.: Scott, Foresman, 1984,
pp. 47-52. Analyzes the market for undeveloped land and suggests that urban sprawl is
efficient.
"An Empirical Investigation of Intraurban Wage Gradients." Journal of

Eberts, Randall.
Urban Economics 10 1981), pp. 50-60. Estimates the variation in wages within a
(
metropolitan area.
Modeling the Monocentric City

Mills, Edwin S. "Planning and Market Processes in Urban Models." In Public and Urban
Economics, ed. Ronald Grieson. Lexington, Mass.: Lexington Books, 1976. A model
of urban land use that uses linear and nonlinear programming to allocate land to
alternative activities.
.
Studies in the Structure of the Urban Economy. Baltimore: Johns Hopkins. 1972.
Chapter 5 derives the conditions under which the monocentric city is more efficient
than a city with dispersed employment.
Wheaton, William. "Monocentric Models of Urban Land Use: Contributions and
Criticisms." In Current Issues in Urban Economics, ed. Peter Mieszkowski and
Mahlon Straszheim. Baltimore: Johns Hopkins, 1979. Discusses several renditions of
the traditional monocentric model.
Chapter 9
CNERAL-tQUILIBRIUM
LAND OSE
Fifty years hence . . . we shall escape the absurdity of growing a whole chicken to
eat the breast or wing, by growing these parts separately under a suitable medium.
Winston Churchill
(~JT~ his chapter uses general-equilibrium analysis to explore the interactions

K.S between different parts of the urban economy. In contrast with partial-
equilibrium analysis, which examines a single market in isolation, general-
equilibrium analysis recognizes that markets are interdependent and explores the
effects ofchanges in one market on related markets.
Urban general-equilibrium analysis explores the interactions between the res-
idential land market, the business land market, and the urban labor market. It is
used to predict the effects of changes in one part of the urban economy on land use
throughout the urban area. This chapter uses general-equilibrium analysis to explore
the effects of three changes in the monocentric city: an increase in export sales,
the introduction of a streetcar system, and an increase in the residential property
tax. Later in the book, general-equilibrium analysis is used to explore the land-use
effects of changes in technology and public policy. Chapter 10 (Suburbanization and
Modern Cities) explores the land-use effects of the truck and the automobile. Chap-
ter (Land-Use Controls and Zoning) examines the general-equilibrium effects of
1 1
various land-use controls. Chapter 19 (Autos and Highways) explores the effects of
highway congestion on land-use patterns.
General-Equilibrium Conditions
Figure 9-1 shows the equilibrium land-use pattern of a monocentric city. The bid-
rent function of the business sector (office firms and manufacturers) is negatively
sloped, reflecting the benefits of locating near the central market area and the central
249
250 Part II Land Rent and Urban Lund-Use Patterns
Figure 9-1 Equilibrium Land-Use Pattern in a Monocentric City
Bid-rent function of business
Bid-rent function of residents
Bid-rent function of farmers
Farming
Business firms outbid residents for central land, generating a CBD with radius »,.. The SRI ) is the area o\ er « Inch residents
outbid firms and farmers, thai is. a ring \\ ith width ui„ - u,.
I.
export node. The bid-rent function of residents is negatively sloped, reflecting the
benefits of locating near jobs in office and manufacturing firms. The central business
district (CBD) is the area over which firms (office firms and manufacturers) outbid
residents, so the radius of the CBD is n h miles. The suburban residential district
(SRD) is the area over which residents outbid firms and farmers, so the SRD is a
Chapter 9 General-Equilibrium Land Use 251
ring of width («/, — Ub) miles. For land beyond w/,, farmers outbid residents, so the
radius of the city is w/, miles.
The monocentric city is assumed to be one of hundreds of cities in a regional
economy. The city is "open" in the sense that households and firms can costlessly
enter or leave the city. The city is "small" in the sense that it is a trivial part of the
regional economy, so that changes in the city do not affect the common utility level
of the region's residents. Any change that increases the utility level of the city's
residents causes in-migration from the rest of the region. This migration bids up the
prices of housing and land (increasing the cost of living), decreasing the utility level
of the city's residents. Migration continues until the original utility level is restored.
The urban economy achieves general equilibrium when four conditions are
satisfied simultaneously:
1 Locational equilibrium for firms. If all firms make zero profits at all
locations, there is no incentive to change locations. This condition is
guaranteed by the business bid-rent function, Rf,: competition for land bids
up the price of land at locations with relatively low transportation costs, so
firms are indifferent among all locations in the city.
2. Locational equilibrium for households. If all households achieve the

same utility level at all locations in the city, there is no incentive to change
locations. This condition is guaranteed by the residential bid-rent function,
Ri, : competition for land bids up the price of land at locations with
relatively low commuting costs, so households are indifferent among all
locations in the city.
3. Competitive bidding. Land is allocated to the highest bidder. This

condition is satisfied if landowners, like other people in a competitive
economy, seek the highest return on their assets.
4. Labor-market equilibrium. The total demand for labor (from office firms
and manufacturers in the CBD) equals the total supply of labor (from
residents in the SRD).
To summarize, general equilibrium occurs when both the land market and the labor
market are in equilibrium at the same time.
Numerical Examples of General-Equilibrium Analysis

This section uses two numerical examples to explain the principles of general-
equilibrium analysis. An increase in export sales has a direct effect on the CBD land
market, and the introduction of a streetcar system has a direct effect on the SRD
land market.
Initial Equilibrium
Table 9-1 shows the numbers associated with the monocentric city depicted in Fig-
ure 9-1. In initial equilibrium, the CBD radius is two miles, and the city radius is
252 Part II Lund Ren: and Urban Land-Use Patterns
Table 9-1 General-Equilibrium Effects of an Increase in Export Price

and Export Sales
Initial Partial General

Equilibrium Equilibrium Equilibrium
Radius of CBD in miles (ub) 2.0 2.5 2.2

Area of CBD square miles)i 12.56 19.63 15.20
Width of SRD in miles («/, - u h ) 4.0 3.5 4.3
Area of SRD (square miles) 100.48 74.58 117.47
Radius of citj in miles in, i 6.0 6.0 6.5
Area of city (square miles) 113.04 113.04 132.67
Wage (S per hour) 10.0 10.0 13.0
Average labor density (workers per square mile) 20.000 22.000 21.000
Total labor demand (workers) 25 .2001 43 .750
1 319.150
Average household density (households per square mile) 2.500 2.500 2.717
Total labor supply (workers) 25 .2001 186.438 319.150
six miles. The land area of the CBD is computed with the formula for the area of
a circle:
CBD land area = it (u h )

:
(9-1
Because it), is 2 miles, the CBD land area is 12.56 square miles. The land area of
the city is
City land area =n {it/, (9-2)
Because it/, is 6 miles, the land area of the cit) is 1 1 3.04 square miles. The land area
of the residential district is the difference between the land areas of the city and the
CBD, or 100.48 square miles.
The city is in general equilibrium because total labor supply equals total labor
demand. Total labor demand equals the land area of the CBD times the average
number of workers per square mile:
Total demand = CBD land area •

Average employment density (9-3)
In Table 9-1 the average employment density

. is 20.000 workers per square mile, so
total labor demand is 251.200. Total labor supply equals the land area of the SRD
times average household density (assuming that there is one worker per household):
Total supply = SRD land area • Average household density (9—1)
In Table 9-1, the average household densit) is 2.500 per square mile, so total labor
supply is 251.200. Labor supply equals labor demand, so the land-use allocation
shown in Figure 9-1 is an equilibrium allocation.
Chapter 9 General-Equilibrium Lund Use 253
Increase in Export Sales
Suppose that the price of baseballs (the city's export good) increases. What are the
effects of the price increase on the city's land and labor markets?
Consider first the effect of the price increase on the CBD land market. In the
short run, the increase in price increases the profits of baseball firms. The demand
for CBD land increases as (1) existing baseball firms increase their output and (2)
new firms enter the baseball industry. The increase in the demand for land shifts the
bid-rent function for business land upward, as shown in Figure 9-2. The upward
shift of Rh is consistent with the leftover principle: at the higher baseball price,
firms have a greater excess of revenue over nonland cost, so they are willing to pay
more for land. As the business bid-rent function shifts upward, the CBD expands
at the expense of the residential area: the radius of the business district increases
from 2.0 miles to 2.5 miles, and the SRD shrinks from a width of 4.0 miles to
3.5 miles.
Figure 9-2 Partial-Equilibrium Effects of an Increase in Export Price

and Export Sales
Initial business bid-rent function
Business bid-rent function with higher export price
2.5 6.0
Distance from city center (miles)
An increase in the price of exports shifts the business bid-rent function upward. The CBD expands (the radius
increases from 2.0 miles to 2.5 miles) at the expense of the SRD (the width decreases from 4.0 miles to 3.5 miles).
254 Part U Land Rent anil Urban Land-Use Patterns
How does the upward shift of the business bid-rent function affect the urban
labor market? Total labor demand increases for two reasons:
1 CBD territory effect. As shown in the second column of Table 9-1, the
CBD increases from 12.56 square miles to 19.63 square
land area of the
miles.
2. Employment density effect. The increase in the relative price of land

causes baseball firms to substitute nonland inputs (capital and labor) for
land, increasing labor density (workers per square mile). In the second
column of Table 9-1, labor density has increased from 20,000 workers
per square mile to 22,000.
In combination, these two effects increase total labor demand from 251.200 to
431,750. The upward shift of the business bid-rent function also decreases total labor
supply: the CBD encroaches on SRD land, decreasing total supply from 251.200
to 186.438. Because labor demand increases and labor supply decreases, there is
excess demand for labor.
The increase in price increases labor demand while it decreases labor supply,
and the excess demand for labor increases the city's wage. The increase in the wage
increases the quantity of labor supplied for two reasons:
1 SRD territory effect. An increase in the wage increases the relative

attractiveness of the city. Laborersfrom the rest of the region move to the
city, bidding up the prices of housing and residential land. R/, shifts
upward, increasing the size of the SRD as residents outbid farmers (for land
near the city border) and firms (for land near the CBD border). As the
residential district expands, total labor supply increases.
2. Household density effect. As the price of housing increases, households

substitute nonhousing goods for housing, so housing consumption per
household decreases. In addition, as the price of residential land increases.
housing producers substitute capital for land, so the amount of land per unit
of housing decreases. In combination, these two changes decrease land
consumption per household: that is. even household occupies a smaller lot.
The decrease in land consumption increases household density (the number
of households and laborers per square mile), increasing total labor supply.
To summarize, the increase in the wage increases the quantity of labor supplied
because both the size and the density of the residential district increase.
How does the increase in the wage affect the demand side of the labor market?
The increase in the wage decreases the quantity of labor demanded for t\\ o reasons:
1. CBD territory effect. An increase in the wage increases production costs.

Some firms decrease their output, and other firms shut down. The decrease
in production activity decreases the demand for land, so the business
bid-rent function drops. As the CBD loses territory to the residential
district, total labor demand decreases.
2. Employment density effect. The increase in the wage decreases land rent.
As the relative price of labor increases, firms substitute land for the
relatively expensive labor. Labor density (the number of workers per square
mile) decreases.
To summarize, the increase in the wage decreases the quantity of labor demanded
because both the size and the density of the business district decrease.
These two changes in the labor market narrow the gap between labor supply
and labor demand. The wage continues to rise until general equilibrium is restored.
Figure 9-3 shows the new equilibrium allocation of land, and the third column of
Table 9-1 shows numbers behind the new allocation. The new equilibrium wage is
$13 per hour (up from $10). The CBD radius is 2.2 miles, and labor density is 21,000
workers per square mile, so total labor demand is 319,150. The SRD is 4.3 miles
wide and household density is 2,717 per square mile, so total labor supply is the
Figure 9-3 General-Equilibrium Effects of an Increase in Export Price

and Export Sales
Business bid-rent function

with higher wage
The increase in the export price causes excess demand for labor, increasing the city's wage. The increase in the
wage shifts the business bid-rent function downward and shifts the residential bid-rent function upward. The net
effect is a larger CBD (the radius increases from 2.0 miles to 2.2 miles) and a larger SRD (the width increases
from 4.0 miles to 4.3 miles).
same as total demand. In other words, general equilibrium is restored with a wage
of $13. For a lower wage, labor demand would exceed supply; for a higher wage,
supply would exceed demand.
There are four lessons from the general-equilibrium analysis of the increases in
the export price:
1 Market interactions. Changes in the CBD (the business land market)

affect the city's labor market and its residential land market. The upward
shift of the business bid-rent function increases the size and the density of
the CBD, causing excess demand for labor. The resulting increase in the
wage shifts the bid-rent function of residents upward, increasing the size
and density of the SRD.
2. Shifts in bid-rent functions. The increase in the export price causes an
upward shift of the business bid-rent function (a partial-equilibrium effect).
The resulting increase in the wage causes a downward shift of the business
bid-rent function (a general-equilibrium effect).
3. Land rent. The increase in the export price increases land rent throughout
the city. The increase in export production increases the demand for land in
both the CBD and the residential district, so landowners throughout the city
benefit from the increase in export sales.
4. Welfare effects of increased export sales. In general equilibrium, the
city's residents are no better off in the larger city. The increase in the wage
is offset by higher costs of housing and land, leaving the utility level
unchanged. The increase in export sales increases the number of residents,

not their utility.
A Streetcar System
As a second illustration of general-equilibrium analysis, suppose that the city installs

a streetcar system. The streetcar decreases the time and monetary cost of commuting,
causing a number of changes in the urban economy.
How does the streetcar system affect the residential bid-rent function? As ex-
plained in Chapter 8 (Land Use in the Monocentric City), the slope of the bid-rent
function is determined by commuting costs: the lower the unit cost of commuting,
the flatter the bid-rent function. In Figure 9—4, the streetcar system decreases the
slope of the bid-rent function and increases the width of the SRD from 4 to 6.2 miles.
The streetcar system makes peripheral areas accessible to CBD jobs, allowing res-
idents to outbid farmers for land near the city's edge. In addition, residents outbid
linns for the land near the CBD border. This is the partial-equilibrium effect of the
streetcar system.
The upward tilt of the residential bid-rent function increases total labor supply
for two reasons:
1 . SRD territory effect. The expansion of the residential district increases

labor supply. Table 9-2 shows an increase in the land area of the residential
district from 100.48 square miles in column one to 190.88 square miles in
column two.
Figure 9^4 The Partial-Equilibrium Effect of the Streetcar System
Residential bid-rent
function after streetcar
The system decreases commuting cost, decreasing the slope of the residential bid-rent function. The
streetcar
width of the SRD increases from four miles to six miles, and the city radius increases from six miles to eight
miles.
Table 9-2 General-Equilibrium Effects of a Streetcar System
Initial Partial General

Equilibrium Equilibrium Equilibrium
Radius of CBD in miles (ub) 2.0 1.8 2.3

Area of CBD (square miles) 12.56 10.18 16.61
Width of SRD in miles («;, — Ub) 4.0 6.2 4.8
Area of SRD (square miles) 100.48 190.88 141.68
Radius of city in miles («/,) 6.0 8.0 7.1
Area of city (square miles) 113.04 201.06 158.29
Wage ($ per hour) 10.0 10.0 6.0
Average labor density (workers per square mile) 20,000 20,000 22,000
Total labor demand (workers) 251,200 203,575 365,433
Average household density (households per square mile) 2,500 2,700 2,579
Total labor supply (workers) 251,200 515,380 365,433
2. Household density effect. The increases in the prices of housing and land
cause consumer substitution (consumers substitute nonhousing goods for
housing) and factor substitution (housing producers substitute nonland
inputs for land). The lot size per household decreases, increasing popula-
tion density. Table 9-2 shows an increase in household density from 2.500
in column one to 2,700 in column two.
To summarize, the streetcar system increases the quantity of labor supplied because it
increases both the size and the density of the residential district. The streetcar system
increases the relative attractiveness of the city, causing in-migration that bids up the
price of land. People migrate from other cities because workers in the streetcar city
have shorter commuting times, so they have higher net income (gross labor income
less the time cost of commuting).
The streetcar causes an excess supply of labor. The CBD shrinks, so labor
demand falls at the same time that labor supply increases. The excess supply of
labor decreases the city's wage, causing changes on both sides of the labor market.
On the demand side, the business bid-rent function shifts upward, increasing both the
territory and the employment density of the CBD. On the supply side, the residential
bid-rent function shifts downward, decreasing both the territory and density of the
residential district.
These two changes in the labor market narrow the gap between labor supply and
labor demand. The wage continues to drop until general equilibrium is restored. Fig-
ure 9-5 shows the new equilibrium allocation of land, and the third column of
Table 9-2 shows numbers behind the new equilibrium. The new equilibrium wage is
$6 per hour (down from $10). The CBD radius is 2.3 miles, and the SRD is 4.8 miles
wide. The equilibrium number of laborers is 365,433. There are three basic lessons
from the general-equilibrium analysis of the streetcar:
1 . Tilt and The streetcar affects

shift of the residential bid-rent function.
the residential bid-rent function in two ways. The upward tilt (or flattening)
of the function is the partial-equilibrium effect, which occurs because the
streetcar decreases commuting cost. The downward shift is the
general-equilibrium effect, which occurs because the streetcar causes an
excess supply of labor, decreasing the city's wage.
2. Land rent. The streetcar increases land rent in both the .SRD and the CBD.
The streetcar increases the accessibility of residential land, so residents are
willing to pay more for land. The streetcar also decreases wages and
production costs, so firms are willing to pay more for land in the CBD.
Landowners throughout the city benefit from the streetcar.
3. Welfare effects of the streetcar. In general equilibrium, the city's residents

are no better oil with the streetcar. Because the city is "open," the streetcar
causes in-migration from the rest of the region that decreases wages and
increases the prices of housing and land. The benefits of the streetcar (faster
and cheaper commuting) are offset by lower wages and a higher cost of
living.
Figure 9-5 The General-Equilibrium Effect of the Streetcar
Residential bid rent with

streetcar and lower wage
The streetcar causes excess supply of labor, decreasing the city's wage. The business bid-rent function shifts upward,
and the residential bid rent shifts downward. The CBD grows, and the SRD shrinks. General equilibrium is restored
with a larger SRD (the width increases from 4.0 miles to 4.8 miles) and a larger city (the radius increases from
6.0 miles to 7.1 miles).
The General-Equilibrium Effects of the Property Tax

Consider a city that initially finances all of its public services with a land tax. If
all the land in the city is owned by people living outside the city, the city's public
services are initially financed by a tax on absentee landowners. Suppose that the city
replaces its land tax with a tax on residential capital (improvements such as houses
and apartments) and the capital tax is paid by the city's residents. Under the new tax
policy, city residents pay more taxes but do not receive more public services.
How does the property tax affect the city's economy? There are two ways to
answer this question. First, a graph can be used to show the partial-equilibrium
effects of the tax. Second, a computer can be used to show the general-equilibrium
effects of the tax.
Figure 9-6 shows the partial-equilibrium effects of the residential property tax.
The property tax is a tax on both land and capital (improvements), so the tax liability
Figure 9-6 Partial-Equilibrium Effects of the Residential Property Tax
Distance from citj center (miles)
The residential property tax is on improvements, so u generates a larger tax burden on locations with
parti) a tax
large capital-land ratios. Therefore, the gap between the pre-tax and the post-tax residential bid-rent functions
increases as one approaches the citj center
increases with theamount of improvements. As explained in Chapter 8 (Land Use

in theMonocentric City), the capital-land ratio increases as one approaches the city
center because housing producers substitute capital for relatively expensive land.
Because the amount of improvements per acre is larger closer to the city center, the
tax liability per acre increases asone approaches the city center. For example, an
apartment building on a one-acre site two miles from the city center faces a higher
tax liability than a single-family home on a one-acre site live miles from the city
center. Therefore, the closer the site to the city center, the larger the downward shift
of the bid-rent function. This is shown in Figure 9-6: the gap between the pre-tax
and post-tax bid-rent functions increases as one approaches the city center.
What are the general-equilibrium effects of the residential property tax'.' In other
words, how does the tax affect the urban labor market'.' The downward shift of the
residential bid-rent function decreases the territorj and the density of the residential
district, causing an excess demand lor labor. The city's wage increases, causing
changes in the CBD and SRD that narrow the gap between supply and demand. In
Chapter y General-Equilibrium Land Use 261
the CBD, the increase in the wage increases production costs and decreases the bid
rent for business land. The resulting decrease in the size and the density of the CBD
decreases the quantity of labor demanded. In the SRD, the increase in the wage shifts
the residential bid-rent function upward. The resulting increases in the size and the
density of the residential district increase the quantity of labor supplied. The wage
continues to fall until general equilibrium is restored.
What will the city look like after general equilibrium is restored? One way to an-
swer this question is computer to generate two pictures of the urban economy,
to use a
one before the property tax is imposed and one after the economy has adjusted to the
new tax. An urban general-equilibrium computer model allocates different plots of
land in a hypothetical city to firms (labor demanders) and households (labor suppli-
ers). If a particular allocation does not satisfy the conditions for general equilibrium,
the computer tries a different allocation. This groping process continues until an
equilibrium allocation is found. The computer model can be designed to compute
a pre-tax equilibrium and a post-tax equilibrium. By comparing the "before" and
"after" snapshots of the urban economy, one can identify the general-equilibrium
effects of the property tax.
Sullivan (1985) used a computer model to simulate the effects of a 1.7 per-
cent residential property tax. The results of the study are listed in Table 9-3. The
Table 9-3 General-Equilibrium Effects of the Residential

Property Tax
Initial General
Equilibrium Equilibrium
Number of households 250.038 255.853
Territories
Radius of CBD (miles) 2.916 2.85
Width of SRD (miles) 3.582 3.378
Radius of city (miles) 6.498 6.228
Input Prices
Wage ($ per hour) 10.00 10.02
Median CBD land rent ($ per acre per year) 28,999 26,016
Median SRD land rent ($ per acre per year) 2,886 2,761
Density
Average labor density (workers per square mile) 9,360 8.851
Average household density (households per square mile) 2.359 2.344
Land Rent
CBD land rent ($ per week) 7,053. 341 6,258.973
SRD land rent ($ per week) 3,580,168 3.163.577
Total land rent ($ per week) 0,633,302 9,422,550
SOURCE: Arthur M. Sullivan, "The General Equilibrium Effects of the Residential Property Tax:
Incidence and Excess Burden." Journal of Urban Economics 18 (1985), pp. 235-50.
hypothetical city has an initial workforce of about 250.000 and a population of about
500,000. The property tax decreases the city's land area, population, employment
density, and population density. Although the tax applies only to residential prop-
erty, it ultimately affects land rent and land use throughout the city. Land rent at
the median CBD location decreases 1 1 percent (to $26,016 per acre per year), and
land rent at the median SRD location decreases 4 percent (to $2,761 per acre per
year). Employment density decreases 5.7 percent, and household density decreases
1 percent. The decrease in total land rent is about 160 percent of the total revenue
from the property tax. The computer model provides a comprehensive view of the
general-equilibrium effects of the property tax.
Summary
1. In the small, open city, the utility level of residents is fixed: any change that
increases the relative attractiveness of the city causes in-migration that increases
housing and land prices, decreasing utility to its original level.
2. The urban economy achieves general equilibrium when four conditions are met.
a. Firms make zero economic profit at all locations (guaranteed by the bid-rent
functions).
b. Households achieve the same utility level at all locations (guaranteed by the
bid-rent functions).
c. Land is rented to the highest bidder.
d. Total demand for labor (from firms in the business district) equals total
supply (from residents in the residential district).
3. An increase in the export price shifts the business bid-rent function upward,
increasing labor demand as the size and the density of the CBD increase.
a. The excess demand for labor increases the wage.
b. The increase in the wage shifts the residential bid-rent function upward,
increasing the quantity of labor supplied as the size and the density of the
SRD increase.
c. The increase in the wage also shifts the business bid-rent function
downward, decreasing the quantity of labor demanded as the size and the
density of the CBD decrease.
J. The wage continues to rise until labor supply equals labor demand.
4. The streetcar system decreases the slope of the residential bid-rent function,
tilting it upward. The size and the density of the residential district increase,
increasing total labor supply.

a. The excess supply for labor decreases the wage.
/?. The decrease in the wage shifts the business bid-rent function upward.
increasing the quantity of labor demanded as the size and the density of the
CBD increase.
c. The decrease in the wage shifts the residential bid-rent function downward,
increasing the quantity of labor supplied as the size and the density of the
SRD decrease.
d. The wage continues to drop until labor supply equals labor demand.
5. In general equilibrium, city residents are no better off with the streetcar.
Migration to the open city decreases wages and increases the prices of housing
and land, offsetting the benefits of the streetcar.
6. The property tax decreases the slope of the residential bid-rent function,
decreasing the supply of labor as the residential district decreases in size and
density. In general equilibrium, the property tax decreases the city's size,
density, and total land rent.

1 Suppose that the intracity truck is introduced into the traditional monocentric
city, decreasing the cost of intracity freight. Depict graphically the partial-
equilibrium and general-equilibrium effects of the truck.
2. Consider the Swampville, a city that recently drained a swamp near
city of
its and thus increased the supply of developable CBD land by
city center
10 percent. Predict the effects of the draining of the swamp on (a) land rent
in the CBD, (b) land rent in the residential district, (c) employment density,
(d) residential density, (e) the equilibrium wage, and if) total employment.
3. Consider a traditional nineteenth-century monocentric city with a CBD radius
of one mile. In 1 869, buildings at the edge of the CBD are four stories tall. In
1870. all the buildings in the CBD are destroyed by an earthquake. The mayor
of the city announces that in the rebuilt city (/) the maximum building height
will be four stories and (//') business development will be confined to a circle
with a one-mile radius: the radius of the CBD is fixed at one mile. Discuss the
effects of the height restriction on (a) the demand for labor, (b) the supply of
labor, (c) the equilibrium wage, (d) residential density, and (e) the city
radius.
4. Suppose a city restricts the height of its residential structures. The maximum
building height is four stories, the height that would normally occur at a
distance of five miles from the city center.
a. Draw two residential bid-rent functions, one for the city in the absence of
height restrictions and one with height restrictions.
b. Discuss the effects of the height restrictions on wages, total employment,
and land rent in the CBD.
5. Suppose a city imposes a gas tax of $1 per gallon and uses the increase in tax
revenue to decrease other taxes. The tax liability of the typical resident is
unchanged by the change in tax policy.

a. Draw two residential bid-rent functions, one before the gas tax and one
after the tax is imposed.
b. Discuss the effects of the gas tax on wages, total employment, and land rent
in the CBD.
6. Suppose that the government subsidizes the construction of radial highways.
How will such a policy affect land rent, land use, and wages in cities?
264 Pan II Land Rent and Urban Land-Use Patterns
7. Suppose that the government subsidizes the production of agricultural

products. How will such a policy affect land rent, land use, and wages in
cities?
8. Suppose a city builds a streetcar system and passes a law that prevents any
outsiders from moving to the city. In other words, the city keeps its population
at its pre-streetcar level, so it is a "closed" city instead of an open one. Depict
graphically the partial-equilibrium and general-equilibrium effects of the
streetcar in the closed city.
9. Chapter 6 (Urban Economic Growth) uses a supply-demand model to depict

the urban growth process.
a. Use the supply-demand model to depict the general-equilibrium effects of
the streetcar shown in Table 9-2. Be sure that your picture matches the
numbers in the table.
b. Compute the elasticity of labor demand with respect to the wage.
10. Use a graph to show the partial-equilibrium effects of a tax on business

property. Outline the general-equilibrium effects of the business property tax.
How do these general-equilibrium effects of the business property tax differ
from the effects of the residential property tax?
1 1. Consider a rectangular city with the following characteristics:
/'.
The city is two miles wide.
ii. The CBD is on a harbor, and the SRD stretches to the east of the CBD.
Hi. Manufacturers transport their output from factories to the CBD docks
by horse-drawn wagon.
iv. Labor demand is a fixed 2,000 workers per square mile (no factor
substitution),
v. Labor supply is a fixed 500 workers per square mile (no consumer or
factor substitution).
vi. In the initial equilibrium, the wage is $10. the CBD is one mile long,
and the SRD is four miles long.
a. Draw the bid-rent functions consistent with the assumptions above.
b. Show that the city is in equilibrium with a one-mile CBD and a four-mile
SRD.
Suppose that the intracity truck replaces the horse-drawn wagon. The
partial-equilibrium effect of the truck is to increase the CBD's length by
0.50 miles.
c. Depict graphically the partial-equilibrium effect.
d. What is the partial-equilibrium effect on the labor market: by how much

does labor demand exceed labor supply?
e. Will the wage increase or decrease?
/ The table below lists the length of the CBD and the length of the city for
different wages. What is the new equilibrium wage for the city?
Wage CBD Length SRI) Length City Length

<% per hour) 1 miles) i miles) 1 miles)
10.0 1.50 4.0 5.50

11.0 1.42 4.3 5.75
12.0 1 .33 4.7 6.00
13.0 1.25 5.0 6.25
14.0 1.17 5.3 6.50

Theoretical General-Equilibrium Models
Brueckner. Jan K. "Labor Mobility and the Incidence of the Residential Property Tax."
Journal of Urban Economics 10 ( 1982). pp. 173-82. Uses a theoretical
general-equilibrium model to explore the effects of the residential property tax on the
urban labor market.
Mills. Edwin S. "An Aggregative Model of Resource Allocation in a Metropolitan Area."
American Economic Review, Papers and Proceedings, May 1967. pp. 197-210.
Reprinted in Readings in Urban Economics, ed. Matthew Edel and Jerome
Rothenberg. New York: Macmillan, 1972. The first general-equilibrium model of the
urban economy.
Polinski, A. M.. and D. L. Rubinfeld. "The Long-Run Effects of a Residential Property Tax
and Local Public Services." Journal of Urban Economics 5 (1978). pp. 241-62.
A theoretical model that explores the interactions between a city's land markets and its
labor market. Discusses the general-equilibrium effects of the property tax on a
number of city characteristics.
Wheaton, William. "Monocentric Models of Urban Land Use: Contributions and
Criticisms." In Current Issues in Urban Economics, ed. Peter Mieszkowski and
Mahlon Straszheim. Baltimore: Johns Hopkins. 1979. Discusses several renditions of
the traditional monocentric model.
Computer Models of Urban Land Use

Birch. David; Reilly Atkinson; Sven Sundstrom; and Linda Stack. Patterns of Urban
Change. Lexington, Mass.: Lexington Books. 1974. A simulation model that predicts
the location patterns of housing and employment.
Mills, Edwin S. "Planning and Market Processes
in Urban Models." In Public and Urban
Economics, ed. Ronald Grieson. Lexington. Mass.: Lexington Books, 1976. A model
that uses linear and nonlinear programs to allocate land to different activities.
Sullivan. Arthur M. "The General Equilibrium Effects of the Residential Property Tax:
Incidence and Excess Burden." Journal of Urban Economics 18 (1985), pp. 235-50.
Uses a computational model of a monocentric city to simulate the general-equilibrium
effects of the residential property tax.
Computer Models of Residential Land Use

Amott. R. J., and J. G. MacKinnon. "The Effects of the Property Tax: A General
Equilibrium Simulation." Journal of Urban Economics 4 1977). pp. 389—4-07. Uses a
(
general-equilibrium model of the residential sector to compute the effects of the

property tax on residential land rent and land use.
Carlton, D. W. "The Spatial Effects of a Tax on Housing and Land." Regional Science and
Urban Economics 11 (1981), pp. 509-27.
Grieson. Ronald. "The Economics of Property Taxes and Land Values: The Elasticity of
Supply of Structures." Journal of Urban Economics 1 1974). pp. 367-81.
(
Leroy, S. F. "Urban Land Rent and the Incidence of Property Taxes." Journal of Urban
Economics 3 (1976). pp. 167-79.
Mills. Edwin S. Studies in the Structure of the Urban Economy. Baltimore: Johns Hopkins.
1972. A general-equilibrium model is used to simulate the effects of transportation
congestion on residential land rent and land use. The demand side of the urban labor
market is not included in the model.
Solow, Robert M. "Congestion Costs and the Use of Land for Streets." Bell Journal of
Economics and Management Science 4 (1973). pp. 602-1 8. Uses a general-equilibrium
model of the on
residential sector to simulate the effects of congestion externalities
residential land use and the use of land for streets.
."Congestion. Density, and the Use of Land in Transportation." Swedish Journal of
Economics 74 (1972), pp. 161-73. Uses a general-equilibrium model of the residential
sector to simulate the effects of congestion externalities on residential land use and the
use of land for streets.
Chapter 10
j" UBURBANIZATION and

M D t R N CITIES
A suburb is a place where a developer cuts down all the trees to build houses,
and then names the streets after the trees.
Bill Vauehn
his chapter explains the decline of the traditional monocentric city and the
(~J7~
t_x rise of the modern multicentric city. most
In the traditional monocentric city,
economic activity was concentrated in the central core area. The entire metropoli-
tan area was oriented toward the employment and shopping opportunities in the
central city. In the modern multicentric city, a large fraction of employment is in
suburban areas, with much of the suburban employment in subcenters. People who
live in metropolitan areas now depend less on the central city for employment and
shopping.
This chapter has seven parts. The first part discusses the facts on suburbanization,
focusing on the changes in the spatial distributions of population and employment in
the last several decades. The second through the fifth parts discuss the reasons for the
suburbanization of manufacturers, population, retailers, and office firms. The sixth
part discusses the development of suburban subcenters in Chicago, Los Angeles, and
Houston. The of the chapter focuses on land-use patterns in the modern
final part
multicentric city.
Suburbanization Facts
What are the facts on the suburbanization of employment and population?
Figure 10-1 shows the distribution of population and employment between central
cities and suburbs in 1948 and 1990. The percentage of the metropolitan population
in central cities dropped from 64 percent in 1948 to 39 percent in 1990, and the
267
Figure 10-1 Percent of Metropolitan Population and Employment in

Central Cities, 1948 and 1990
92
85
75
67
64
49 48
45
39
Population Manufacturing Wholesale trade Retail trade Selected services
SOURCES: 1948 data from John F. Kain. "The Distribution and Movement of Jobs and Industry." in The
Metropolitan Enigma, ed. James Q. Wilson (Cambridge. Mass.: Harvard University Press. 1%8): 1990 data
from U.S. Bureau of the Census, Journey to Work in the United States (Washington. D.C.: U.S. Government
Printing Office), 1994.
percentage of manufacturing employment dropped from 67 percent to 45 percent.

The losses in trade and service employment were even larger: the percentage of
wholesaling employment dropped from 92 percent to 49 percent.
in central cities
The percentage of retail employment dropped from 75 percent to 48 percent: the

percentage of service employment dropped from 85 percent to 52 percent.
It's important to note that part of the measured increase in the share of activity in
the suburbs results from urban growth with fixed central-city boundaries. As an urban
area grows, most growth occurs on the periphery. As the metropolitan area expands
outward, an increasing share of its population lives outside the fixed boundaries of
the central city. If the suburban population is assumed to be the population outside
the central-city border, growth causes suburbanization.

Table 10-1 provides a closer look at the distribution of employment between
central cities and suburbs in 1990. The employment sectors are listed in declining
order of importance. Three sectors had more than half of metropolitan employment
in central cities: public administration (62 percent), transportation, communications,
and utilities (56 percent), and finance, insurance, and real estate (58 percent). The
three service sectors (professional and related services, business and repair services,
and personal services) had between 52 percent and 57 percent of their employment
in central cities.
Chapter 10 Suburbanization and Modem Cities 269
Table 10-1 Distribution of Employment between Central Cities and

Suburbs by Type of Employment, 1990
Employment in Employment in Percent in Percent in
MSAs Central Cities Central Cities Suburbs
Total 93,117,895 47.861.224 51 49

Professional and related services 21.648.685 12.362.153 57 43
Manufacturing 15,586.909 6.938.010 45 55
Retail trade 15,351.455 7.352.682 48 52
Finance, insurance, real estate 6.993.936 4,066.519 58 42
Transportation, communication, utilities 6.748.723 3.767.546 56 44
Construction 5.649.399 2.514,248 45 55
Business and repair services 4.782.462 2.508.616 52 48
Other industries 4.672.853 1.919.649 41 59
Public administration 4,489.663 2.780.672 62 38
Wholesale trade 4,308,269 2.105.851 49 51
Personal services 2.885,541 1.545.278 54 46
Source: 1990 Census of Population and Housing. Journey to Work in the United Stales (Washington. D.C.: U.S.
Government Printing Office. 1994).
Suburbanization of Manufacturing
The share of metropolitan manufacturing employment in central cities decreased
from about two-thirds in 1948 to less than half in 1990. Mills (1972) provides
evidence that the suburbanization of manufacturing started long before 1948. What
caused the suburbanization of manufacturing employment?
The Intracity Truck

Moses and Williamson (1972) explain how the intracity truck encouraged the sub-
urbanization of manufacturing. In the monocentric city of the nineteenth and early
twentieth centuries, freight was shipped within cities by horse-drawn wagons. Start-
ing about 1910, manufacturers switched from the horse wagon to the truck. The truck
was both and cheaper: around 1 920. the truck was half as costly ($0.15 per ton
faster
per mile versus $0.33), and at least twice as fast. After the intracity truck was in-
troduced, its use spread rapidly: for example, the number of trucks in the city of
Chicago increased from 800 in 1910 to 23,000 in 1920.
As explained in Chapter 8 (Land Use in the Monocentric City), the location
decision of a manufacturing firm is determined by the outcome of a tug-of-war.
On one side is the central export node, which pulls the firm toward the city center
because freight costs are lower for firms near the export node. On the other side
is the suburb, which pulls the firm toward low-wage suburban locations. In the era
of the horse-drawn wagon and the streetcar, the cost of moving freight was large
relative to the cost of moving people, so manufacturers located in the city center.
It was cheaper workers from the suburbs to the central-city factory than
to ship the
to ship the output from a suburban factory to the export node. The intracity truck
decreased freight cost, weakening the pull toward the central export node, so the
tug-of-war was more frequently won by the suburb. Although a suburban location
was inefficient with the horse-drawn wagon, it was efficient with the truck.
How did the intracity truck affect the bid-rent function of manufacturers? The
bid-rent function is negatively sloped because of freight costs. The truck decreased
freight costs, decreasing the slope of the bid-rent function. This is shown in Fig-
ure 10-2: the bid-rent function tilts upward, allowing manufacturers to outbid resi-
dents for suburban land and increasing the radius of the manufacturing district from
u to it".
Figure 10-2 shows the partial-equilibrium effects of the intracity truck. As

explained in Chapter 9 (General-Equilibrium Land Use), the expansion of the man-
ufacturing district increases labor demand, causing excess demand for labor and an
increase in the city's wage. The increase in the wage shifts the residential bid-rent
function upward (increasing the quantity of labor supplied as the size and the den-
sity of the residential district increase) and shifts the manufacturing bid-rent function
downward (decreasing the quantity of labor demanded as the size and the density of
Figure 10-2 The Partial-Equilibrium Effects of the Intracity Truck
Bid-rent function with

horse-drawn waaon
Bid-rent function with

the truck
Residential bid-rent
function
Distance to export node (miles)
The replacement of the horse-draw n v. agon with the truck decreased freight costs, decreasing the slope
ol manufacturing bid-rent function. The radius oi the manufacturing district increased from u to u .
Chapter It) Suburbanization and Modern Cities 271
the manufacturing district decrease). These changes in the labor and land markets
continue until general equilibrium is restored.
The Intercity Truck
Two decades after the truck was first introduced, manufacturers started using the truck
Improvements in the truck made long-distance travel feasible,
for intercity transport.
and the expansion of the intercity highway system facilitated intercity truck traffic.
Eventually, the truck became competitive with the train and the ship for intercity
freight. As manufacturers switched from trains and ships to trucks, they were freed
from their dependence on the railheads and ports in city centers, and moved to sites
accessible to the intercity highways. The freight costs associated with suburban sites
decreased, allowing manufacturers to move closer to their suburban workers.
In 1956 the highway system was authorized by Congress, and the
interstate
bulk of the system was in place by the late 1970s. The highway system decreased
the relative cost of shipping by truck, causing more manufacturers to switch their
freight operations from ships and trains to trucks. More recently, cities have built
circumferential highways (beltways) that are connected to the interstate highway
system. Manufacturers locate close to the suburban beltways because they provide
easy access to the interstate system.
Figure 10-3 shows the bid rent for manufacturing land at different locations in
a beltway city. The figure is drawn under the assumption that the beltway circles
the city at a distance of 2.5 miles. In the beltway city, manufacturers can transport
their output to the beltway or to a central export node (a railhead or a port at the
city center). The bid-rent function forms a peak at the central export node (as in the
monocentric city) and a circular ridge centered on the beltway. Manufacturers are
likely to outbid other land users for land near the suburban beltway.
Figure 10-3 Manufacturing Bid Rent in Beltway City
Rent
The Automobile
The automobile contributed to the suburbanization of manufacturers. To explain the
effects of the automobile, consider a firm with a highly skilled workforce: only one
in 1.000 people has the skills required to work for the firm. Suppose that the firm
employs 100 people and the firm's workers are distributed uniformly throughout a
streetcar city of 100,000 people. Where in the city will the firm locate? Because
the firm draws laborers from the entire metropolitan area, it locates at a point that
is accessible to the entire urban area. In the hub-and-spoke streetcar city, the city
center (the hub) is the most accessible location.

Where will the firm locate in a modern automobile city? In the auto city, em-
ployees can drive their cars from their homes to any location in the urban area. While
the firm might still locate in the city center, it could also locate along an urban high-
way or a suburban beltway because such sites are accessible to residents throughout
the metropolitan area. In general, the automobile loosened the ties to the streetcar
nodes in the core area, allowing the suburbanization of firms with specialized labor
forces.
Single-Story Plants
Another factor in the suburbanization of manufacturing was the switch from the
traditional multistory plants of the nineteenth century to single-story plants. To ex-
ploitnew production technologies (assembly-line production and materials-handling
techniques such as the forklift truck), manufacturers built single-level facilities. As
land consumption increased, the forces pulling the firms toward the suburbs (low-
cost land and low wages) dominated the forces pulling the firms toward the central
city (lower freight costs), and many manufacturers moved to the suburbs.
Suburban Airports
The increased importance of air freight is another reason for the suburbanization of
manufacturing. A firm that transports a relatively large fraction of its output by air
experiences a relatively strong pull toward an airport. For some types of firms, the
suburban airport has replaced the old central export node (railhead or port facility)
as the point of orientation. As explained later in the chapter, many modern cities
have clusters of employment near suburban airports.
The Suburbanization of Population

The share of metropolitan population living in central cities dropped from about
two-thirds in 1948 to just under two-fifths in 1990. Mills (1972) provides evidence
of suburbanization as far back as 1 880. As explained in Chapter 8. the density func-
tion describes the relationship between population density and distance to the city
center. Mills estimated the density functions for four metropolitan areas (Baltimore.
Chapter 10 Suburbanization and Modern Cities 273
Milwaukee, Philadelphia, and Rochester) between 1880 and 1963. The assumed
relationship between density and distance is
D(u) = A-e~ 8 " (10-1)
where
D{u) = Population density it miles from the city center (people per square
mile)
A = Parameter estimated from the data
e = The base of the natural logarithm
# = Parameter to be estimated from the data
Table 10-2 shows the average density gradient (defined as the percentage change in
population density per mile) for the four cities between 1 880 and 963. The gradient
1
decreases over time, indicating that the population density function flattened out
over time. The third column of the table uses the estimated density gradients to
compute the percentage of the metropolitan population living within three miles of
the city center. Between 1880 and 1963, the percentage dropped from 88 percent to
24 percent.
The flattening of the density function over time is a worldwide phenomenon.
Between 1801 and 1961, London's density gradient decreased from 1.26 to 0.34,
meaning that the percentage of its population living within three miles of the city
center dropped from 88 percent to 28 percent. In Paris, the gradient decreased from
2.35 in 1817 to 0.34 in 1946. In cities throughout the world, population has been
shifting outward away from the city center.
Table 10-2 Average Density Gradients for Four

Metropolitan Areas, 1880-1963*
Average Percent of Population

Year Gradient i within Three Miles of City Center
1880 1.22 88%

1890 1.06 83
1900 0.96 78
1910 0.80 69
1920 0.69 61
1930 0.63 56
1940 0.59 53
1948 0.50 44
1954 0.40 34
1958 0.35 28
1963 0.31 24
*The metropolitan areas are Baltimore. Milwaukee. Philadelphia, and Rochester.

SOURCE: Edwin S. Mills. Studies in the Structure of the Urban Economy (Baltimore:
Johns Hopkins. 1972).
274 Part 11 Land Rent and Urban Land- Use Patterns
This section discusses five possible reasons for the suburbanization of popula-
tion:
1 Increase in real income

2. Decrease in commuting cost
3. Central-city problems: race, crime, taxes, education
4. Following the firms to the suburbs

5. Public policy
Increase in Real Income
Did income growth contribute to the suburbanization of population? As explained

in Chapter 8, an increase in income has an ambiguous effect on the relative attrac-
tiveness of the suburbs because it has two conflicting effects on the relative cost of
a suburban location.
1 Increase in commuting cost. An increase in real income increases the

amount of income sacrificed per minute of travel time: it increases the
opportunity cost of commuting. Therefore, an increase in income increases
the relative attractiveness of dwellings close to the workplace. For someone
who works in the CBD, an increase in income increases the relative
attractiveness of central-city housing. This is the opportunity-cost effect:
an increase in income increases the opportunity cost of commuting, pulling
households toward the central city.
2. Increase in land consumption. Because land is a normal good, an increase

in income increases land consumption (lot size). Income growth increases
the relative attractiveness of locations with low-priced land (the suburbs).
A household that occupies a large lot is willing to commute long distances
to occupy low-cost suburban land. This is the consumption effect: an
increase in income increases lot size, pulling households toward
inexpensive suburban land.
Given these conflicting effects, it is not possible to predict, on theoretical

grounds, whether income growth contributes to suburbanization. Income growth
causes suburbanization consumption effect dominates the opportunity-cost
if the
effect. This will occur income elasticity of demand for land exceeds the in-
if the
come elasticity of commuting costs. As explained in Chapter 8. the land elasticity is
close to the commuting-cost elasticity, so the two effects almost cancel each other
out. Therefore, it does not appear that income growth contributed very much to the
suburbanization of population.
Decrease in Commuting Cost

Over the last 140 years, technological innovations have decreased the monetary and
time costs of commuting. In the 1800s, the development of new transitsystems
decreased the cost of commuting from the suburbs to the central core. The most
important innovations were the horse-drawn streetcar (developed in the 1850s) and
the conventional streetcar (developed in the 1890s). In the last 60 years, improve-
ments automobile and the intracity highway network have decreased the cost
in the
of personal commuting. The automobile increased travel speeds, decreasing the time
cost of commuting.
Chapter 9 explains the partial- and general-equilibrium effects of the streetcar.
The same analysis applies to other changes that decrease commuting costs, such as
the introduction of the automobile. The sequence of changes in the urban economy
is as follows:
1 The slope of the residential bid-rent function decreases (the function tilts
upward), increasing the size and density of the residential district.
2. The increase in the supply of labor decreases the city's wage.
3. The decrease in the wage shifts the business bid-rent function upward,
increasing the quantity of labor demanded as the size and the density of the
CBD increase.
4. The decrease in the wage shifts the residential bid-rent function downward,
decreasing the quantity of labor supplied as the size and the density of the
residential district decrease.
5. The wage continues to drop until labor supply equals labor demand.
In equilibrium, the streetcar (or automobile) increases the population and land area of
the monocentric city. The increase in the relative accessibility of suburban locations
increases the share of population in the suburbs.
Figure 10^4 shows the initial bid-rent functions and the bid-rent functions after
the metropolitan area has fully adjusted to the lower commuting costs. Suppose
that the central city is a circular area with a radius of three miles: locations beyond
three miles from the city center are considered suburban. Before the streetcar (or
automobile), the CBD radius is one mile, the metropolitan radius is four miles, and
a relatively small fraction of the population lives in the suburban area. The decrease
in commuting costs shifts the bid-rent functions outward, increasing the radius of
the metropolitan area. The fraction of the population living outside the central-city
border increases for two reasons. First, the suburban area grows: the width of the
suburban ring increases from three miles to four miles. Second, the price of suburban
land increases, increasing population density within the suburban area.
In the late 1800s and early 1900s, many landowners profited from the introduc-
tion of streetcars. In a number of cities, speculators bought large tracts of undeveloped
land just outside the city's borders, and patiently awaited the extension of streetcar
service to their land. Other investors took a more active role. In Cleveland, the Von
Sweringen brothers bought a large parcel of undeveloped land from the local society
of Shakers. When the Cleveland State Railways refused to extend their streetcar ser-
vice to the undeveloped area, the Von Sweringen brothers built their own streetcar
line to the area, which became known as Shaker Heights. The brothers' land, which
was appraised at $240,000 in 1900, increased in value to $80 million by 1930.
Figure 10-4 Suburbanization Caused by a Decrease in

Commuting Cost
\
Rb
\R, Bid-rent functions

\ before streetcar
Bid-rent functions
--"I after streetcar
\\ 1
--I h
i
\ \
— -^
\
1
i 1 1
i 1 1 i
i 1 1 i
1.0 2.0 3.0 4.0 6.0

A change technology that decreases the unit cost of commuting streetcar or automobile)
in (
causes suburbanization: a larger fraction of the population lives heyond the central-city
border (three miles from the city centen.
Central-City Problems
Other theories of suburbanization focus on the negative locational attributes (dis-

amenities) of central cities. The following factors encourage households to move
from the central city to the suburbs:
1. Old housing. The deterioration of the central -city housing stock

encourages households to move to the suburbs, where newer housing is
readily available.
2. Race and income. Some households move to the suburbs to escape racial
move to avoid living near low-income households.
conflict; others
3. Central-city fiscal problems. Many central cities have relatively high

taxes, encouraging households to move to low-tax suburbs. The causality
goes both ways: fiscal problems cause suburbanization, and
suburbanization contributes to central-city fiscal problems.
4. Crime. Most central cities have relatively high crime rates, encouraging
households to move to the suburbs. The reasons for the relatively high
central-city crime rates will be discussed in Chapter 22 (Crime and
Punishment).
5. Education. Suburban schools are typically superior to central-city schools,

encouraging households to relocate in suburban school districts. The
reasons for the inferiority of central-city schools will be explained in
Chapter 21 (Education).
Empirical studies of the suburbanization process provide support for the theory
that central-city problems encourage suburbanization. The empirical literature is
reviewed by Wasylenko (1984). Bradbury, Downs, and Small (1982) tested various
theories of suburbanization, using a sample of 2 SMS As for the period 1 970- 975
1 1 1
They found that a metropolitan area experienced relatively rapid suburbanization if
its central city had(l)arelatively old housing stock, (2) relatively high taxes, and (3)
a relatively large black population. Another factor in suburbanization was the number
of suburban local governments: the larger the number of suburban governments to
choose from, the more rapid the suburbanization. Frey ( 979) found that metropolitan 1
areas with high taxes, high crime rates, and low educational expenditures experienced
relatively rapid suburbanization.
Following Firms to the Suburbs

As explained earlier in the chapter, manufacturing employment shifted from the
central city to the suburbs for a number of reasons, including the development of
the truck, the adoption of assembly-line production techniques, and the building of
circumferential highways. As explained later in the chapter, retailers and office firms
have also moved to the suburbs, forming suburban subcenters. The suburbanization of
employment contributed to the suburbanization of population because some workers
followed their employers to the suburbs.
In the polycentric city, workers commute to jobs along suburban beltways and
suburban subcenters. The first row of numbers in Table 10-3 shows commuting
patterns within all U.S. metropolitan areas. In 1990, about 40 percent of work-
ers commuted within suburban areas, more than twice the percentage that com-
mute from suburbs to central cities. The second line of numbers shows commut-
Table 10-3 Commuting Patterns within Metropolitan Areas, 1990
Central Citx Central City Suburb to Suburb to

Region to Central City to Suburb Central City Suburb
All U.S. metropolitan areas 32.5 7.6 19.8 40.1

All U.S. CMSAs 30.7 8.0 16.3 45.1
Northeast CMSAs 39.4 4.5 9.9 46.3
North central CMSAs 23.8 8.2 16.6 51.4
South CMSAs 20.6 9.6 28.6 41.2
West CMSAs 31.0 10.0 19.2 39.8
SOURCE: 1990 Census of Population and Housing. Journey to Work in the United States (Washington.
DC: U.S. Government Printing Office, 1994).
ing patterns within consolidated metropolitan statistical areas (CMSAs). In these

large metropolitan areas, a larger percentage of workers commute within suburbs
(45.1 percent) and a smaller percentage commute from suburbs to central cities
(16.3 percent). The other numbers show commuting patterns for the CMSAs in dif-
ferent regions. In the Northeast, the percentage commuting within the suburbs is
more than four times the percentage commuting from the suburbs to the central city.
How does the suburbanization of employment affect the residential bid-rent
function? Workers are willing to pay more for housing and land near beltways and
suburban subcenters because such locations have a relatively low commuting cost.
In the modern multicentric city, the residential bid-rent function has several peaks.
one at the city center and one for each concentration of employment. McDonald and
McMillen (1990) show that land rent increases as one approaches the employment
subcenters in the Chicago metropolitan area. Specifically, a one-mile move toward
the O'Hare Airport subcenter increases the price of land by 4.5 percent.
Public Policy and Suburbanization

A number of public policies affect the intraurban location choices of households,
encouraging households to live in the suburbs. Among the public policies to be
discussed later in the book are the following:
1. Subsidies for homeownership. The federal tax code allows the deduction
of interest paid on home loans, providing an implicit subsidy for home-
ownership. Federal mortgage programs (FHA guaranteed mortgages)
also decrease the cost of ownership. These housing subsidies encourage
suburbanization because they increase housing consumption and increase
the relative attractiveness of the suburbs, where housing is relatively
inexpensive.
2. Commuting externalities. Because commuting generates congestion

externalities and pollution externalities, the marginal private cost of
commuting is less than the marginal social cost. As explained in Chapter 19
(Autos and Highways), the internalization of these externalities would
increase the relative cost of suburban living, encouraging commuters to live
» closer to their workplaces.
3. Fragmented system of local government. Some households move to the

suburbs to flee central-city fiscal problems. The fragmented system of local
government encourages suburbanization because it provides alternatives to
the central-city government. Chapter 16 (Overview of Local Government)
explores the role of metropolitan consolidation (a switch to a metropolitan
government) on the location decisions of households. The issue is whether
consolidation would discourage suburbanization.
4. Highway construction. The building of the highway system decreased
commuting costs, increasing the relative attractiveness of the suburbs.
Chapter 10 Suburbanization and Modem Cities 2.1*)
Gentrification: Return to the City?
In the 1970s and 1980s, the popular press publicized the renovation of central-city
housing by wealthy households. To many people, the efforts of these households
signaled a change in residential location patterns. It appeared that many high-income
households were abandoning the suburban lifestyle to embrace an urban one. The
process is labeled gentrification because poor households are replaced by relatively
wealthy ones.
The most important fact about renovation and gentrification is that the number of
renovations is relatively small. In the late 1970s, renovation and gentrification in the
30 largest cities were concentrated in about 1 00 neighborhoods and involved less than
one-half of 1 percent of the housing stock in these cities (Frieden and Sagalyn, 199 1 ).
In most cities, the number of dwellings renovated each year is a small fraction of the
number of dwellings that are abandoned. Kern (1984) describes the characteristics
of renovators. The typical renovator is wealthy, young, highly educated, and either
single or married with less than two children. Such households are attracted to
central-city locations because they ( 1 ) patronize cultural establishments in the central
city, (2) have a relatively high commuting
cost, and (3) have relatively low demands
for housing and land. Most of the renovators moved from one part of the central
city to the area being renovated, not from the suburbs back to the central city. These
facts suggest that gentrification involves a relatively small number of households
and does not signal a fundamental change in residential location patterns.
Suburbanization of Retailers
The share of metropolitan retail employment in central cities decreased from about
two-thirds in 1948 to less than half in 1990. There are three principal reasons for the
suburbanization of retailing.
Following the Consumers
The suburbanization of population caused some retail activity to move to the suburbs.
According to the simple version of central place theory discussed in Chapter 5 (How
Many Cities?), a retail firm locates at the center of its market area. If scale economies
are small relative to demand density (per capita demand times population density),
market areas are relatively small. A retail firm with a relatively small market area is
likely to follow its consumers to the suburbs. In contrast, a retail firm with relatively
large scale economies would not necessarily follow its customers to the suburbs: if
there is a single store for the entire metropolitan area, the most accessible location
might still be in the city center.
The more sophisticated version of central place theory considers the effect
of shopping externalities on retailers' location choices. If there are benefits from
one-stop shopping and comparison shopping, retailers may compromise on their
ideal (central place) locations to exploit the externalities associated with comparison
shopping and one-stop shopping. The more sophisticated theory predicts that some
with moderate scale economies (relative to demand density) will not follow
retailers
their consumers to the suburbs, but will stay in the city center to exploit shopping
externalities. It also suggests that retailers will form clusters (shopping centers and
malls) in the suburbs.
The Automobile
The automobile, which replaced the hub-and-spoke streetcar system, loosened the
ties to the city center. Before the development of the automobile, a retailer with rel-
atively large scale economies was tied to the CBD because that's where the streetcar
delivered all the consumers. If shoppers use the auto instead of the streetcar, they
can easily travel from their homes to any point in the metropolitan area, so even
a suburban store can draw consumers from the entire metropolitan area. Retailers
no longer had to be in the city center to be accessible to consumers throughout the
metropolitan area, and many moved to the suburbs.
What about a retailer with moderate scale economies? In the hub-and-spoke
streetcar city, travel from the suburb to the city center (along the spokes) was fast and
inexpensive, while travel within the residential areas (between the spokes) was costly.
A suburban store could not survive because intrasuburban travel costs were high:
if it located at the geographical center of its suburban market area, few consumers
would show up. If shoppers use the auto instead of the streetcar, travel costs within
the residential area decrease, allowing the store to locate at the center of its suburban
market area.
Population Growth
The third reason for retail suburbanization is population growth. As the population
of a metropolitan area increases, the total demand for retail goods increases. As total
demand increases, the equilibrium number of retail stores increases. Some of the
new stores locate in the suburban areas.
To explain this idea, consider the demand for wigs. Suppose that scale economies
in wig retailing are exhausted with an output of 500 wigs per store per month, and
the per capita demand for wigs is 0.01 wigs per person per month. A city with a
population of 50,000 has a total demand of 500 wigs per month, and can support a
single wig store in the city center. If the city grows to 500.000, and the demand for
wigs per capita remains constant, the city supports 10 wig stores. If wig shopping is
not subject to shopping externalities (comparison shopping and one-stop shopping),
the wig stores divide the metropolitan area into 10 market areas, and each wig
store locates in the center of a market area. Because some of the market areas are
centered in the suburbs, some wig stores locate in the suburbs. If there are shopping
externalities, wig stores may compromise on their ideal (central place) locations, but
some wig stores are still likely to locate in the suburbs.
Suburbanization of Office Employment

Before the early 1970s, most office firms were still located in the city center be-
cause they depended on face-to-face contact with other firms, and the CBD pro-
vided a central meeting place for workers from different firms. Although some
office activities moved to the suburbs by the early 1970s, these were typically the
"back-office" operations of insurance firms and banks, operations involving paper
processing rather than face-to-face contact. For such activities, a CBD location was
unnecessary. For most office activities, the advantages of the CBD (opportunities
for timely contact with other firms) outweighed the disadvantages (high wages and
rents).
Since the early 1970s, suburban office space has grown at a rapid rate. Nation-
wide, suburban office employment grew over seven times faster than central-city
officeemployment during the 1970s (116 percent versus 15 percent). According to
Cervero (1986). the most rapid growth in suburban office employment occurred in
medium-sized metropolitan areas (with populations between 250.000 and million). 1
The suburban office boom continued into the 1980s. For example, suburban office
space in the Chicago metropolitan area more than doubled between 1980 and 1987,
increasing the suburban share of total office space from 29 percent to 38 percent.
Suburbanization and Clustering of Offices
Pivo (1990) discusses the suburbanization of office space in six metropolitan areas:
Denver, Houston, Los Angeles, San Francisco, Seattle, and Toronto. Some of his
results are shown in Table 10-4. The second column shows, for each metropoli-
tan area, the percentage of office space contained in the central business district.
In five of the six metropolitan areas, the CBD contained less than half of the total
Table 10-4 Characteristics of Suburban Office Clusters
Percent of Suburban
Cluster Space Close to
Metropolitan Percent of Space Number of
Area in CBD Clusters Freeway Interchange Rail Transit
Denver 27 63 73 36
Houston 20 160 70 29
Los Angeles 12 270 66 30
San Francisco 31 102 93 52 10
Seattle 45 30 93 69
Toronto 54 52 79 27 25
SOURCE: Gary Pivo. "The Net of Mixed Bead: Suburban Office Development in Six Metropolitan
Regions." Journal of the American Planning Association, Autumn 1990, pp. 457-69.
office space; in four of six, the CBD contained less than a third of the total office
space. At the national level, central business districts contained about 47 percent of
metropolitan office space.
The third column in Table 10—4 shows the number of suburban office clusters.
Pivo defines a cluster as two or more office buildings within a quarter-mile area.
Under this relatively permissive definition of a cluster, about 88 percent of the
suburban office space in the six metropolitan areas is clustered. The number of
clusters varies across metropolitan areas, reflecting differences in total suburban
office space: the larger the total office space in a metropolitan area, the larger the
number of suburban clusters. The size distribution of clusters was consistent across
metropolitan areas: the largest 10 percent of clusters contained about half of the total
office space; the largest 25 percent contained about two-thirds of the total space; the
largest50 percent of clusters contained about nine-tenths of the total space.

The fourth, fifth, and sixth columns of Table 10^4 demonstrate the importance
of access in the location of the clusters. The fourth column shows the percentage of
suburban (clustered) office space within a half mile of a freeway: the percentages
range from 66 percent (Los Angeles) to 93 percent (San Francisco and Seattle).
The fifth column shows the percentage of office space within a mile of a freeway
interchange: the percentages range from 29 percent (Houston) to 69 percent (Seattle).
The sixth column shows, for the two cities with rail transit systems, the percentage
of office space within walking distance of a rail transit station (a quarter mile):
10 percent for San Francisco and 25 percent for Toronto.
Communications Technology and Office Suburbanization

One factor in the suburbanization of office employment is the advances in communi-
cations technology (satellites, fiber optics, and solid-state electronics) that decreased
the cost of transmitting data, voices, and pictures. The new technology generated
two new methods of communication: electronic mail and teleconferencing. These
new communication methods made some types of office activities less dependent on
face-to-face contact. As an activity becomes less dependent on face-to-face contact,
the relative attractiveness of a CBD location decreases.
Electronic Mail. Office firms use electronic mail to send messages, documents,
and data over computer lines. Electronic mail decreases the need for face-to-face
contact because information can be sent over computer lines instead of being deliv-
ered by employees. Electronic mail allows the suburbanization of firms involved in
the rapid turnaround of documents and reports.
To explain the effects of electronic mail, consider Abby, who runs an accounting
firm. Abby gets 50 pages of data from her client (another firm) and condenses the
information into an income statement. The client demands that the income statement
be ready within 24 hours. In the absence of electronic mail, Abby would locate near
her CBD customers to ensure the timely pickup of her client's data and the timely
delivery of herincome statement. The development of electronic mail allows Abby
to move to the suburbs. Her inputs (the client data) and her output (the income
statement) can be sent over telephone lines, so she can do her accounting in the
suburbs, far from her CBD clients. Because Abby does not depend on face-to-face
contact, but on report-to-face contact, electronic mail allows her to escape the high
wages and high rents of the CBD.
Electronic mail has caused the decoupling of some CBD firms. Many firms
split their operations into suburban activities and CBD
Suppose that a activities.
firm has a large number of accountants, whose only task is to condense information
into reports: that is. they compute bottom lines. The accountants do not interact
with people outside the firm, but simply provide condensed information to the firm's
executives, who then use the information in their interactions with other firms.
In the absence of electronic mail, the firm's accountants and executives must be
in the same location to facilitate the rapid turnaround of information. Given the need
for face-to-face interactions among the executives of different firms, the firm locates
in the CBD. If the accounting information can be transmitted electronically, however,
the accounting division canmove to the suburbs, and the firm can send information
between the suburban accountants and the CBD executives over telephone lines.
Because electronic mail facilitates intrafirm communication, it allows the firm to
decouple its operations, moving its support staff to the suburbs while keeping its
executives in the CBD.
Teleconferencing. The second new communications method, the teleconference,

is essentially a video meeting. The teleconference is superior to the telephone confer-
ence because it allows both visual and verbal communication. The teleconference
decreases the need for face-to-face contact within and between firms: instead of
sending employees across town to another office, the firm can send them down the
hall to the video room.
Because the video camera does not transmit as much information as a face-to-
face encounter, the teleconference is not a perfect substitute for face-to-face contact.
While the video camera focuses on one or two people in a meeting, a person who is
physically present can watch everyone in the room, picking up verbal and nonverbal
cues from a large number of people. In addition, the video camera cannot transmit
tactile information. The tactile information provided by face-to-face encounters is
sometimes useful: a firm, enthusiastic handshake sends a message that cannot be

easily conveyed through a video camera.
Because the teleconference is not a perfect substitute for face-to-face contact,
it is unlikely to cause substantial suburbanization of office firms. Most of the face-
to-face contact that occurs in the CBD is between employees of different firms. The
subtle interactions between firms often require face-to-face contact. When people
from different firms meet, one of the purposes of the meeting is to establish trust.
Because it is easier to establish trust in a face-to-face encounter, a video meeting is
inappropriate for some types of interactions.
In most urban areas, officeemployment occurs in both the CBD and suburban
areas. The CBD contains activities for which face-to-face contact is still important
(negotiation, design, marketing, and research). In contrast, suburban areas contain
activities for which electronic communication is a good substitute for face-to-face
contact.
Suburban Subcenters and Modern Cities

Many of the firms that moved to the suburbs located close to other firms. The
clustering of firms in a few suburban locations caused the development of suburban
subcenters. This part of the chapter discusses the rationale for suburban subcenters
and presents some facts on subcenters in three metropolitan areas: Los Angeles,
Houston, and Chicago.
The clustering of firms in suburban subcenters results from agglomerative
economies in production. In a cluster of manufacturers or office firms, the total
demand for some intermediate inputs is large enough to support external suppli-
ers, who can exploit scale economies and thus produce the intermediate inputs at a
lower cost. For example, manufacturers in a cluster may save money by purchasing
business services (repair, maintenance, accounting) from a common supplier. Al-
ternatively, if there are scale economies in printing glossy brochures and reports, a
large cluster of office firms can support a printing firm, allowing each firm to save
on printing costs. The same principle applies to restaurants and hotels: one rule of
thumb is that a cluster of office firms with total office space of 2.5 million square
feet can support a 250-room hotel.
Figure 10-5 shows land rent at different locations in a hypothetical city with
a beltway (2.5 miles from the and two employment subcenters. The
city center)
land-rent surface has three peaks —
the highest one at the city center and two lower
—
ones at the subcenter locations and forms a circular ridge centered on the beltway.
Subcenters in Los Angeles
Giuliano and Small (1991) use data on the spatial distribution of activity in the Los
Angeles metropolitan area to show that there were 28 subcenters in Los Angeles
Figure 10-5 Land Bid Rent in Beltway City
Rent
Figure 10-6 Distribution of Employment and Population

between CBD, Subcenters, and Other Areas:
Los Angeles in 1980
90,- 90
80 -
70 - 66
60
50
40
30 -
20 -
10
CBD Subcenters Other CBD Subcenters Other
Employment Population
SOURCE: Genevieve Giuliano and Kenneth Small, "Subcenters in the Los Angeles Region,*
Regional Science and Urban Economics 21 ( 99 1 1 ).
County and Orange County in 1980. They define a subcenter as a zone where the
employment density is at least 10 workers per acre and total employment is at
least 10,000 workers. In contrast, employment density in the CBD (downtown Los
Angeles) is 36 workers per acre and total employment is 469,000. Figure 10-6
shows the shares of employment and population in the CBD (downtown L.A.),
the subcenters, and other locations. Total employment in the subcenters was about
twice the employment in the CBD (23 percent versus 1 1 percent). Two-thirds of the
metropolitan area's employment was outside the center and the subcenters. In terms
of population, 1 in 10 people lived in either the CBD or a subcenter.
Table 10-5 shows the characteristics of the CBD and the 28 subcenters. The
subcenters are listed after the CBD in descending order with respect to total employ-
ment. The subcenters vary in employment density (workers per acre). The average
employment density of the subcenters is 17.7 workers per acre, compared to 36.0
in the CBD. Giuliano and Small examined the relationship between employment
density and distance to the CBD and found a negative relationship between density
and distance. This model of the mono-
result is consistent with the predictions of the
centric city. In the case of Los Angeles, there is a negative relationship despite the
fact that a relatively small fraction of total employment is in the CBD.
The subcenters vary with respect to employment ratios (jobs per resident).
The average employment ratio of the subcenters is 1.58 jobs per resident, com-
pared to 1.47 in the CBD and 0.43 for the entire metropolitan area. Except for the
airport subcenters, which have few residents and thus large employment ratios, the
Table 10-5 Characteristics of the CBD and Subcenters in the Los Angeles
Metropolitan Area
Employment Workers Jobs Distance to

Rank Location < 1.000) per Acre per Resident CBD unites)
1 Downtown Los Angeles (CBD) 469 36.0 1.47 0.1

2 Los Angeles West 176 25.5 1.37 15.8
3 Santa Monica 65 16.9 1.11 16.7
4 Hollywood 64 21.4 0.73 7.3
5 Los Angeles Airport 59 16.7 4.32 18.8
6 Orange County Airport 48 16.1 1 .589.87 40.7
7 Glendale 43 15.5 1.07 12.3
8 Commerce 42 17.0 4.05 9.8
9 Vernon/Huntington Park 39 33.2 2.42 4.9
10 San Pedro 38 15.7 2.74 23.3
11 Santa Ana 38 17.3 1.51 32.9
12 Inglewood 37 14.6 1.24 14.7
13 Pasadena 36 25.3 1.73 12.1
14 Long Beach Airport 33 15.5 3.684.78 23.3
15 Marina Del Ra\ 32 11.4 1.28 14.0
16 Long Beach 30 18.0 0.84 25.3
17 Van Nuys Airport 28 12.6 2.04 22.1
18 Burbank Airport 26 28.4 10.86 16.5
19 Hawthorne 18 12.4 0.74 13.5
20 Canoga Park/Warner Center 17 11.2 1.21 27.4
21 Lawndale 17 17.1 1.36 20.5
22 L.A. East 16 37.3 2.30 6.8
23 Fullerton 16 11.4 4.97 27.3
24 Downey 15 17.3 2.3S 14.S
25 Santa Ana South 14 12.2 1.76 37.4
26 Sherman Oaks 13 11.9 1.04 18.6
27 Burbank S.W. 13 18.0 1.92 14.1
28 Anaheim/Orange/Garden Grove 11 11.3 1.06 30.2

29 Garden Grove/Stanton 10 12.9 5.60 26.6
All subcenters 994 (total) 17.7 (average) 1.58 (average)
I Genevieve Giuliano and Kenneth Small. "Subcenters in the Los Anceles Region," Regional Science mid Urban Economics 21
(1991).
employment ratios of the subcenters are relatively small. This suggests that the sub-
centers are not isolated employment centers, but instead provide a mixture of jobs and
dwellings.
Giuliano and Small suggest that the subcenters can be divided into five types:
mixed industrial, mixed service, specialized entertainment, specialized manufactur-
ing, and specialized service. Most o\' the mixed-industrial subcenters started out
as low-density manufacturing areas near transport nodes (airport, port, or marina)
and grew as they attracted other activities. Most of the mixed-service subcenters
are like traditional downtowns: they provide a wide range of services. Many of
these subcenters functioned as independent centers before they were absorbed into
the metropolitan economy. The specialized-manufacturing subcenters include areas
near airports that produce aerospace equipment and older manufacturing areas. In
the service-oriented subcenters, 90 percent of employment is in service activities
such as medical care, entertainment, and education.
There are four basic conclusions from this study of the Los Angeles area. First,
the subcenters differ in the mixes of goods and services they provide, suggesting that
the subcenters play diverse roles within the metropolitan economy. Second, many
of the subcenters are highly specialized, suggesting that there are large localization
economies (clustering of firms in the same industry to share input suppliers, save on
labor costs, and share information). Third, employment in the metropolitan area is
relatively dispersed: two-thirds of total employment is outside the CBD and the sub-
centers. Fourth, employment density decreases as distance from the center increases,
despite the fact that the center contains a relatively small fraction of total
employment.
Subcenters in Houston
Mieszkowski and Smith (1991) explore the land-use patterns in the Houston
metropolitan area and identify 10 subcenters. In 1985, total employment in the
individual subcenters was between 16,000 and 86,000, compared to total employ-
ment in the CBD of 156,000. In 1985, the CBD contained about 10 percent of the
metropolitan area's total employment, leaving 23 percent for the subcenters and
67 percent for areas outside the CBD and subcenters. The corresponding figures for
1970 were 14 percent for the CBD, 1 percent for the subcenters, and 75 percent for
1
areas outside the CBD and subcenters.

Like many other cities, Houston experienced rapid growth in its office space.
Between 1969 and 1989, total office space increased more than sevenfold, from 25
million square feet to 182 million square feet. The most rapid increases occurred in
the suburbs, both inside and outside subcenters, so the suburban share of office space
increased. Figure 10-7 shows the shares of office space in the CBD, the subcenters,
and other areas for 1969 and 1989. By 1989, over three-fourths of office space was
outside the CBD, and over half of the office space was in suburban subcenters.
Subcenters in Metropolitan Chicago
McMillen and McDonald (1998) explore the spatial distribution of employment

in the Chicago metropolitan area. They identify 20 subcenters, defined as a set of
nearby land tracts that each have at least 10 workers per acre and together have a total
of more than 10,000 workers. Table 10-6 shows the characteristics of the subcen-
ters, grouped into six categories based on age, location, and industrial composition.
The first six subcenters have been prominent since the turn of the century and are
dominated by manufacturing. The next six subcenters developed since World War II,
Figi re 10-7 Percentages of Houston's

OfficeEmployment in the
CBD, Subcenters, and Other
Areas, 1969 and 1989
- 54
55
50
45
40
35 -
30 -
25 23
20
15 -
10
-
5
-
CBD Subcenters Other
SOURCE: Peter Mieszkowski and Barton Smith. "Urban

Decentralization in the Sunbelt: The Case of Houston." Regional
s iem r and
( rban It onomics 21 1991
I 1 ).
and six are dominated by manufacturing, transportation, communication, utili-

all
ties,and warehousing. The next two subcenters are new concentrations of industry
and retailing. The next three subcenters are dominated by services and retail trade:
Burbank and Maywood specialize in health services, while Evanston. the location
of Northwestern University, specializes in education.
The last three subcenters arc edge cities, new concentrations of office
defined as
and retail activities outside the core areas of metropolitan areas.Most edge cities are
located where there was little or no development before 1960. Edge cities are typi-
cally centered on huge tracts of mixed-use office space, and provide jobs, residences,
shopping, and services for their inhabitants. The edge-city subcenters include a large
business-service sector and large volumes of office space.
Most of the subcenters listed in Table 10-6 grew during the l
c
)S0s. The excep-
tions are four of the industrial suburbs (Harvey. McCook. Franklin Park, and Niles).
and one old satellite city (Aurora). The new industrial/retail suburbs grew rapidly
during the 1980s: Northbrook grew by over 50 percent, and employment in Palatine
increased more than fivefold. All three of the edge cities grew during the decade
Table 10-6 Subcenters in the Chicago Metropolitan Area
Subcenter 1990 Percent Percent Percent Percent

Municipality Employment Manufacturing Retail Service TCUW
Old Satellite Cities
Aurora 10,689 59 8 15 8
Elgin 13.095 36 11 38 2
Waukegan 11,506 64 <1 1
Old Industrial Suburbs

Chicago Heights 10.862 62 2 22 5
Harvey 9,880 43 4 25 14
McCook 31,109 64 4 5 20
Post-World War II Industrial Suburbs

Addison 23.409 62 4 5 15
Des Plaines 27.653 31 3 4 51
Franklin Park 27,462 56 18 1 8
Niles 40.040 65 4 6 16
Norridge 10,457 43 17 21 6
O'Hare 32,681 21 5 27 31
New Industrial / Retail Suburbs

Northbrook 43,030 31 21 23 11
Palatine 19,385 19 45 9 8
Service and Retail Centers

Burbank 14,028 1 23 65 2
Evanston 19,551 4 13 69 3
Maywood 26,749 7 11 39 12
Edge Cities
Naperville 35.168 20 12 52 2
Oak Brook 1 1 1 .550 13 22 37 20
Schaumburg 40.295 13 36 26 12
'TCUW = communication, utilities, and warehousing.

transportation,
SOURCE: McMillen, Daniel P.. and John F. McDonald. "Suburban Subcenters and Employment Density in
Metropolitan Chicago." Journal of Urban Economics 43 (1998). pp. 157-180.
with employment quadrupling in Naperville. doubling in Oak Brook, and increasing

by 73 percent in Schaumburg.
Edge Cities
The most recent trend in urban location patterns is the development of edge cities,
defined as new concentrations of office and retail activities outside the core areas
of metropolitan areas. Edge cities are typically centered on huge tracts of mixed-
use office space, and provide jobs, residences, shopping, and services for their
inhabitants. Garreau (1991) identifies 123 existing edge cities and 77 emerging
edge cities in the 35 largest metropolitan areas in the United States. In many cases,
Table 10-7 Characteristics of Edge Cities
Planned Office Devel >ped Office Employment in

Space million
I Space (million 1993
Edge City square feet) square feet) 1 thousands
Irvine. CA 45 33 165
Tyson's Corner, VA 50 28 80
Hwy 180/284. NJ 15 5 75
Schaumburg. IL 20 11 71
Reston. VA 30 18 45
Research Triangle 22 14 40
Park. NC
Dearborn-Fairland, MI 8 4 20
Rariton Center, NJ 30 11 20
Park 10. Houston. TX 14 4 15
SOURCE: Henderson, Vernon, and Arindam Mitra. "The New Urban Landscape: Developers and Edge
Cities." Regional Science and Urban Economics 26 1996). pp. 61 3—643
(
a single development company or individual controlled the initial development and

building of the edge city.
Table 10-7 shows some of the characteristics of several edge cities (Henderson
and Mitra. 1996). The amount of developed office space ranges from 4 million to
33 million square feet, and most of the office space is of the highest quality (class
A). By comparison, the CBDs of Atlanta. Baltimore. Cleveland, Buffalo. Detroit.
Hartford, Indianapolis, and Milwaukee have between 2 million and 9 million square
feet of class A office space, and the CBDs of some of the largest metropolitan areas
(e.g.. Washington, Los Angeles, Boston, and Chicago) have between 17 million and
40 million square feet of class A office space.
Subcenters and City Centers
Based on recent studies of subcenters in several large metropolitan areas, we can

draw several conclusions about the nature and role of subcenters in the metropolitan
economy. Anas, Arnott, and Small (1998) provide a useful summary oi~ what we
know about subcenters in modern cities.
1. Subcenters are numerous in large metropolitan areas, both new and old.
The existing studies of large metropolitan areas (Los Angeles. San

Francisco, and Chicago) suggest that the largest metropolitan areas in the
United States contain at least 20 subcenters.
2. Most of the jobs in metropolitan areas arc outside centers (the main center
and subcenters). Centers contain only 47 percent of the jobs in the
San Francisco metropolitan area, and only about one-third of the jobs
Los Angeles metropolitan area.
in the In the Chicago metropolitan area,
suburban subcenters contain less than a quarter of the suburban jobs.
3. Subcenters have not eliminated the importance of the main center.

McMillen's study of land values in Chicago (1996) shows that there is a
clear and marked peak in land values at the CBD, despite the fact that
subcenters developed several miles away. In the San Francisco Bay Area,
the San Francisco CBD is by far the largest and densest employment center.
In the Los Angeles metropolitan area, downtown Los Angeles dominates
the surrounding subcenters.
What is the economic relationship between a central business district and the
surrounding subcenters? The central business district provides better opportunities
for the face-to-face contact required in the production of many goods and services,
and suburban office firms depend on the central city for many financial and pro-
fessional services. Swartz (1992) explores the corporate service linkages in New
York, Chicago, and Los Angeles. As shown in Table 10-8, for all five corporate ser-
vices studied (actuarial consulting, auditing, banking, investment banking, and legal
services), more than half the suburban firms patronize central-city firms. The per-
centage of suburban firms patronizing central-city firms was the highest for banking,
investment banking, and legal services. In contrast, few central-city firms purchase
corporate services from suburban firms.
The Swartz study suggests that suburban office activity depends on the central
city for some business services that require face-to-face contact. Swartz reaches the
Table 10-8 Percentage of Firms Using Corporate Services in the

Central City and Suburbs
Actuarial Investment Legal

Consulting Auditing Banking Banking Counsel
Percent of suburban 53 56 67 67 71
firms patronizing
central-city firms
firms patronizing
suburban firms
firms patronizing
firms outside
metropolitan area
Percent of central- 14 5 3 3 2
city firms
patronizing suburban
firms
SOURCE: Swartz, Alex. "Corporate Service Linkages in Large Metropolitan Area: A Study of New York,
Los Angeles, and Chicago." Urban Affairs Quarterly 28 (1992), pp. 276-96.
following conclusion:
Despite their ample supplies of office space, the suburbs of the nation's three largest CMS As
do not constitute a self-sufficient outer city economically autonomous from the central city.
Suburban office centers do not house the full range of business activities found within the cen-
tral city. ... A majority of these suburban companies remain dependent on their metropolitan
area's central cities for financial and professional sen ices. . . . These findings from a study of
intrametropolitan linkages do not. however, suggest that the suburbs are simply low -cost areas
for basic data-processing and other back-office functions. Except for the Chicago area, w here
suburban companies rely almost exclusively on the city of Chicago for corporate sen ices, a
significant portion —usually 15 percent to 35 percent —of suburban-based firms do depend

on suburban-based firms for their sen ice needs.
One interpretation of the Swartz study is that although suburban firms depend on
central-city firms for financial and professional services, the expansion and matura-
tion of the suburban office sector makes them less dependent.
Stanbach (1991) explores changes in central-city and suburban employment in
14 large metropolitan areas. He documents the growth in corporate services and
information processing in suburban areas, and reaches the following conclusion:
But the growth and maturation of the suburbs — especially insofar as the suburban development
process has been strengthened by the development of magnet centers —must, of necessity,
alter the relationship between central city and suburbs. In the new relationship, it would appear
that central cities may become more vulnerable to competition, at least in those activities for
which their comparative advantage is marginal.
In the new relationship between central cities and suburbs, the central city will
continue to have an advantage in providing goods and services for which face-to-
face contact is required.
Land Use in the Modern City: Urban Villages?

The modern city, with its suburban subcenters. has been called a system of urban
villages. In many metropolitan areas, suburban subcenters contain mid-rise office
buildings, hotels, shopping malls, and entertainment facilities. Therefore, people in
the area surrounding an employment subcenter have the option of traveling to the
subcenter for work, shopping, and play. In other words, the subcenter is the core of
an urban village. The urban-village idea was recently incorporated into the planning
process in Phoenix: for planning purposes, there are nine urban villages, with an
average population of 100.000. By the year 2000. planners expect to have a total of
12 urban villages, with an average population of 125.000.
Most cities in the United States are experiencing the urban-village phenomenon.
Subcenters are developing in the rapidly growing Sunbelt cities (Atlanta. Phoenix.
Dallas, Houston, and San Francisco) as well as in cities with less rapid growth (St.
Louis, Detroit, and Kansas City ). Subcenters are also developing in the nation's oldest
and most traditional cities (New York. Boston. Chicago. Baltimore). In most urban ar-
eas, the fraction of office and retail employment in suburban subcenters is increasing.
The development of urban villages results from the suburbanization and clus-
tering of retailers and office firms. Retailers moved to the suburbs to be closer to
their suburban customers, and many retailers clustered in malls and subcenters to
exploit shopping externalities. Office firms moved to the suburbs to get better access
to their suburban workforces and clustered in subcenters to exploit agglomerative
economies in the provision of business services, restaurants, and hotels.
Summary
1. The truck contributed to the suburbanization of manufacturing employment for
two reasons.
a. The intracity truck replaced the horse-drawn wagon, decreasing freight
costs relative to commuting costs, pulling firms toward the low-wage

suburbs.
b. The intercity truck replaced the boat and train for intercity freight, freeing
manufacturers from central export nodes. Firms moved closer to their
suburban workers.
2. The growing importance of air freight has caused some manufacturers to
locate close to suburban airports.
3. The automobile replaced the streetcar as the principal means of commuting,

making locations throughout the urban area accessible to workers. Firms with
specialized labor forces were able to move to the suburbs.
4. Because the income elasticity of demand for housing is close to the income
elasticity of commuting costs, it appears that income growth had a relatively
small effect on suburbanization.
5. The suburbanization of population occurred for five reasons.
a. Decreases in commuting costs (from the streetcar and the auto) increased
the accessibility of suburban locations.
b. Central-city problems (racial tensions, deteriorating housing, high taxes,
high crime rates, inferior schools) encouraged some households to flee to
the suburbs.
c. The suburbanization of employment opportunities contributed to the subur-
banization of population as households followed the firms to the suburbs.
d. The underpricing of commuting (commuting and pollution externalities)
decreases the relative cost of suburban locations.
e. The fragmented system of local government allows households to escape
central-city housing problems.
6. In the modern multicentric city, workers commute to jobs in both the CBD and
suburban subcenters. Workers are willing to pay more for housing and land
near employment centers (everything else being equal), so the land-rent
function has several peaks, one near the CBD and one near each subcenter.
7. Retailers moved to the suburbs for three reasons:
a. Some retailers followed consumers to the suburbs.
b. The replacement of the streetcar with the automobile increased the

accessibility of noncentral locations.
c. Population growth increased the equilibrium number of retailers. Some of
the new retailers located in the suburban areas.
8. During the 1970s and 1980s, suburban office space grew at a rapid rate.
Nationwide, suburban office employment grew more than seven times faster
than central-city office employment during the 1970s (116 percent versus
15 percent).
9. Advances in communication technology decreased the need for face-to-face

contact among office workers, allowing some firms to move to the low-wage
suburbs and allowing others to decouple their operations into CBD activities
and suburban activities.
10. The clustering of firms in suburban subcenters results from agglomerative

economies in production. Firms in the subcenters share input suppliers, a labor
force, and information.
1 1. The 28 subcenters in the Los Angeles metropolitan area contain about

23 percent of the area's employment and 9 percent of its population. The
subcenters are specialized: they differ in the mixes of goods and services they
provide.
12. The monocentric, core-dominated city has been gradually replaced with the
multicentric, suburbanized city. The land-rent surface has local peaks at the
city center and the subcenters. and forms a ridge centered on the beltway.
13. The development of suburban subcenters (urban villages) is explained by the
suburbanization and clustering of retailers and office firms. Retailers moved to
the suburbs to be closer to their customers and clustered in subcenters to
exploit shopping externalities. Manufacturing and office firms moved to the
suburbs to be closer to their workforces and clustered to exploit agglomerative
economies in production.

1 Consider a manufacturing firm that exports some of its output by train (through
a central export node) and the rest by truck (the truck exits the metropolitan
area using a suburban beltway). Suppose that production costs (including labor
costs) are the same The
unit cost of transporting output inward
at all locations.
(toward the central export node) and the unit cost of transporting output
is t,,
outward (toward the beltway) is t„. The volume of output going to the central
export node is it',, and the volume going to the beltway is w .
a. Under what conditions will the firm's bid-rent function be positively sloped?
b. How would your answer to (a) change if wages were lower in the suburbs'.'
2. Consider a metropolitan area with a 10-mile radius. Two members of the Dink
household work: Mr. Dink commutes to the city center (it = 0). and Ms. Dink
Chapter ID Suburbanization and Modem Cities 295
commutes to a suburban subcenter four miles due east of the city center.
Depict the household's housing-price functions for three cases:
a. Travel time per mile is the same in both directions (toward the city center
and away from the city center), and Mr. and Ms. Dink have the same
opportunity cost of travel time.
b. Because the bulk of workers commute to the city center, the travel time per
mile of inward commuting (toward the city center in the morning rush hour
and away from the center in the evening rush hour) exceeds the travel time
of outward commuting (away from the city center during the morning rush
hour and toward the center during the evening rush hour). Mr. and Ms. Dink
have the same opportunity cost of travel time.
c. Travel time per mile is the same in both directions, but Ms. Dink has a
higher opportunity cost of travel time.
3. Consider a city where a single firm sells pet rocks, a good with a large income
elasticity of demand. The firm is located in the central shopping area of the
city. Suppose that households in the city experience an increase in income.
a. Assume that the number of pet rock firms is fixed at one. What is the effect
of the increase in income on the firm's bid rent for land at the city center?
b. Is it realistic to assume that the number of pet rock firms will remain
constant?
c. If the number of firms is allowed to vary, what is the effect of the increase
in income on the retail bid-rent function?
4. Comment on the following statement: "Instead of investing money in

traditional radial mass-transit systems (hub-and-spoke systems), we should be
investing in circumferential transit systems. We should build a circular transit
system along the urban beltways."
5. The traditional core-dominated city was a segregated city in the sense that all
employment was in one area (the core) and all residents lived in another area
(the satellite). The modern city is less segregated in the sense that employment
is dispersed throughout the city, in subcenters and along suburban beltways.
Mr. Wizard recently made the following statement: "If my assumptions are
correct, land use in the typical American city will soon be completely
integrated. I predict that each manufacturer and each office firm will be
surrounded by its workforce: the metropolitan area will have dozens of
'company villages.' A firm will be at the center of each village, and the firm's
workers will occupy the land surrounding the factory or office. Every worker
will be employed in the firm at the center of the village. That's integration."
a. Assume that Mr. Wizard's reasoning is correct. What are his assumptions?
In other words, under what set of assumptions will a system of company
villages develop?
b. Are Mr. Wizard's assumptions realistic?
6. The development of the internal-combustion engine caused fundamental

changes in urban land-use patterns. The transformation from the core-
dominated city to the modern suburbanized city took only about 50 years.
296 Part II Lund Rent and Urban Lund-Use Patterns
Given the rapid pace of technological change, it seems likely that some future
innovation will cause another transformation of cities. Given your knowledge
of science fiction and fact, describe an innovation that would cause funda-
mental changes in the spatial structure of cities.
7. Table 10-5 shows the rank and total employment for the CBD and the 28 sub-
centers in the Los Angeles metropolitan area. Is the size distribution of the
employment centers consistent with a hierarchical system of centers? How
closely does the size distribution of employment centers fit the rank-size rule
(see Chapter 5)?
8. Telecommuting refers to a system under which employees work at home for

one or more days per week and use personal computers and telephone lines to
interact with people in the office. Consider an office firm that must decide
whether to locate in the CBD or along a suburban beltway (to be closer to its
suburban workforce). Discuss the effects of telecommuting on the relative
attractiveness of the CBD.
9. Consider the following statement: "The large traditional monocentric city of
the nineteenth century was a fluke, a result of a particular sequence of
technological transportation innovations over the nineteenth and twentieth
centuries. If the sequence of innovations had been slightly different, the large
monocentric city would have never developed. Instead, we would have gone
from the small cities of the eighteenth century directly to the large multi-
centric, suburbanized city that we see today."
a. List the innovations that caused the development of the large monocentric
city in the nineteenth century and the development of the multicentric,
suburbanized city of the twentieth century.
b. Explain how a reordering of the sequence of innovations would have caused
the development of large multicentric, suburbanized cities instead of large
monocentric cities.
10. Consider a modern city in which office employment has been steadily shifting
from the city center to suburban locations along beltways. Suppose that the
city improves its mass-transit system, decreasing the monetary and time costs
of radial travel. Assume that the improvement in the transit system does not
affect the spatial distribution of residents. Will the improvement in the transit
system speed up or slow down the movement of office firms to the suburbs?
1 1 Consider a modern city where firms are located in the central business district
or along a circumferential highway. Congestion is not a problem: everyone
travels at the legal speed limit. Suppose that in an attempt to internalize the
pollution externalities from auto use. the city imposes a S2 per gallon tax on
gasoline. Assume that (/) everyone continues to drive himself or herself to
work and (//') the tax does not affect the spatial distribution of residences. Will
the gas tax increase, decrease, or not affect the fraction of employment in the
central business district'.'

Chapter 10 Suburbanization and Modern Cities 29/
12. Consider two metropolitan areas, Anville and Diamondburg. The names
indicate the type of industry in each metropolitan area: export workers in
Anville produce anvils; export workers in Diamondburg cut diamonds.
Suppose that all output is exported through an export node at the center of the
Would you expect the two cities to have the same spatial
city. distributions of
employment? If not, what differences would you expect?
1 3. One of the major conundrums of urban spatial analysis is the occurrence of tall
residential facilities for the elderly (RFE) in areas of the city where other
residents occupy low-density housing (single-family homes or low-rise
apartment buildings). The RFEs defy the conventional logic that building
heights depend on the price of land. Provide an explanation for the occurrence
of relatively tall RFEs.
14. Consider a expected to grow in the following fashion: between
city that is
1990 and 1995, relatively wealthy households will move to the city; between
1995 and 2000, relatively poor households will move to the city. A developer
has asked the city planning department to allow "leapfrog" development.
Specifically, the developer wants to develop a site eight miles from the city
center at a time when there is plenty of vacant land within an eight-mile radius
of the city center.
a. Why does the developer want to engage in leapfrog development?
b. Is leapfrog development efficient?
15. Some retailers locate along major streets, forming commercial strips. Why
don't these retailers locate in the city center or in a suburban mall?

General Suburbanization
Chinitz. Benjamin. City and Suburb. Englewood, N.J.: Prentice Hall, 1965. A collection of
essays on the causes and effects of suburbanization.
Frieden, Bernard J., and Lynne B. Sagalyn. Downtown, Inc.: How America Rebuilds Cities.
Cambridge, Mass.: MIT Press, 1991. Discusses efforts to rebuild downtown areas.
Greenwood, Michael. "Metropolitan Growth and the Intrametropolitan Location of
Employment, Housing, and Labor Force." The Review of Economics and Statistics 62
(1980), pp. 491-501.
Harrison, B. Employment and Economic Development. Washington, DC: Urban Institute,
1974. Discusses the interdependence of households and firm location decisions,
showing that the suburbanization of residences contributed to the suburbanization of
retailers and service establishments.
Mills, Edwin S. Studies in the Structure of the Urban Economy. Baltimore: Johns Hopkins,
1972. Documents the suburbanization of employment and population between 1880
and 1963.
Stanback, Thomas, and Richard Knight. Suburbanization and the City. Montclair. N.J.:
Allanheld. Osmun and Company. Documents suburbanization trends and

1976. the
implications for the metropolitan economy.
Central-City Problems and the Suburbanization of Population

Bradbury. Katharine L.; Anthony Downs; and Kenneth A. Small. Urban Decline and the
Future of American Cities. Washington. D.C.: Brookings Institution. 1982. Discusses
the reasons behind the decline of central cities.
Frey. W. H. "Central City White Flight: Racial and Non-Racial Causes.** American
Sociological Review 44 ( 1979). pp. 425-88. Tests various theories of suburbanization.
focusing on the effects of central-city problems on suburbanization.
Grubb. W. N. "The Flight to the Suburbs of Population and Employment. 1960-1970."
Journal of Urban Economics 1982). pp. 348-67. Examines the role of taxes.
1 1 (
education, and crime rates on suburbanization.

Kern. Clifford R. "Upper Income Residential Revival in the City." In Research in Urban
Economics, vol. 4. ed. Robert Ebel. Greenwich, Conn.: JAI Press. 1984. Discusses the
facts on gentrification, including the characteristics of renovators.
Laska. Shirley, and Daphne Spain. Back to the City: Issues in Neighborhood Renovation.
New York: Pergamon Press, 1980. Case studies of gentrification.
Reschovsky, A. "'Residential Choice and the Local Public Sector: An Alternative Test of the
Tiebout Hypothesis." Journal of Urban Economics 6 (1979), pp. 501-20. Examines the
role of taxes, local expenditures, and school quality on suburbanization.
Wasylenko, Michael J. "Disamenities. Local Taxation, and the Intrametropolitan Location
of Households and Firms." In Research in Urban Economics, vol. 4. ed. Robert Ebel.
Greenwich. Conn.: JAI Press. 1984. Reviews the empirical evidence concerning the
effects of income on location.
Suburbanization of Manufacturers
Erickson. R.. and M. Wasylenko. "Firm Location and Site Selection in Suburban
Municipalities" Journal of Urban Economics 8 1980). pp. 69-85. Examines the(
influences of labor markets, land prices, transportation facilities, fiscal variables, and
agglomeration economies on the intrametropolitan location decisions ol firms.
Hoover, Edgar M., and Raymond

Vernon. Anatomy of a Metropolis. Cambridge. Mass.:
Harvard University Press. 1959. Discusses the suburbanization of manufacturers in the
New York area.
Kitigawa, Evylyn M.. and Donald Bogue. Suburbanization of Manufacturing Activity
within Standard Statistical Metropolitan Areas. Oxford. Ohio: Scripps Foundation.
1955. Discusses the suburbanization of manufacturing before World War II.
Moses. Leon, and Harold Williamson. "The Location of Economic Acti\ it\ in Chios'* In
Readings in Urban Economics, ed. Matthew Edel and Jerome Rothenberg. New York:
Macmillan. 1972. Discusses the role of the intraurban and interurban truck on the
suburbanization of manufacturing.
Struyk, Raymond. "Evidence on the Locational Activity of Manufacturing Industries in
Metropolitan Areas." Ixtnd Economics 48 ( 1972). pp. 377-82.
Wasylenko, Michael J. "Disamenities, Local Taxation, and the Intrametropolitan Location

of Households and Firms." In Research in Urban Economics, vol. 4, ed. Robert Ebel.
Greenwich, Conn.: JAI Press, 1984. Reviews the evidence concerning the
intrametropolitan location choices of firms.
Suburbanization of Office Firms and Services
Cervero, Robert. Suburban Gridlock. New Brunswick, N.J.:Center for Urban Policy
Research, 1986. Discusses the recent growth in office employment and its implications
for transportation planning.
Clapp, J. M. "Endogenous Centers: A Simple Departure from the NUE Model." Papers of
the Regional Science Association 54 ( 1 984), pp. 1 3-24. Describes the process through
Hughes, James W., and George Sternlieb. "The Suburban Growth Corridor." In America's
New Market Geography, ed. George Sternlieb and James W. Hughes. New Brunswick,
N.J.: Center for Urban Policy Research, 1988. Discusses the market forces behind the
recent office construction boom in suburbs.
Mills, Edwin S. "Service Sector Suburbanization." In America's New Market Geography,

ed. George Sternlieb and James W. Hughes. New Brunswick, N.J.: Center for Urban
Policy Research, 1988. Discusses the recent trends in the suburbanization of services
and compares the spatial distributions of workplaces and residences.
Pivo, Gary."The Net of Mixed Bead: Suburban Office Development in Six Metropolitan
Regions." Journal of the American Planning Association, Autumn 1990, pp. 457-69.
Discusses the suburbanization and clustering of office space between 1960 and the late
1980s.
Effects of Beltway
Bone, A. J., and Martin Wohl. Economic Impact Study: Massachusetts' Route 128.
Cambridge, Mass.: MIT Transportation Engineering Division, 1958.

Bureau of Population and Economic Research. The Socio-Economic Impact of the Capital
Beltway on Northern Virginia. Charlottesville, Va.: Bureau of Population and
Economic Research, University of Virginia, 1968. Discusses the effect of the beltway
in the Washington, D.C., metropolitan area.
Levitan, Don. "Massachusetts' Route 128: A Nonemulative Enigma." Transportation
Research Record, no. 583 (1976), pp. 45-54. Discusses the effects of Boston's beltway.
Mills, Florence. "Effects of Beltways on the Location of Residences and Selected
Workplaces." Transportation Research Record, no. 812 (1981), pp. 26-33. Compares
changes in population and employment in metropolitan areas with and without
beltways. Concludes that beltways did not have a significant effect on central-city
employment.
Payne-Maxie Consultants. The Land Use and Urban Development Impacts of Beltways:
Case Studies. Washington, D.C.: U.S. Department of Transportation and U.S.
Department of Housing and Urban Development, 1980. Discusses the effects of
beltways in eight different metropolitan areas.
300 Part II Lund Rent and Urban Land- Use Patterns
Telecommunications and Land Use
Chinitz. Benjamin. "The Influence of Communication and Data Processing Technolog> on

Urban Form." Research in Urban Economics, vol. 4. ed. Robert Ebel. Greenwich,
In
Conn.: JAI Press. 1984. pp. 67-77. Discusses the effects of new communication
technologies on the location decisions of office firms.
Moss. Mitchell. "Telecommunications: Shaping the Future." In America's New Market
Geography, ed. George Sternlieb and James W. Hughes. New Brunswick. N.J.: Center
for Urban Policy Research. 1988. Discusses the effects of new communication
technologies on location patterns.
Salomon. I. "Telecommunications and Travel Relationships: A Review." Transportation
Research, A 20A ( 1986). pp. 223-38. Discusses the effects of telecommunication
technologies on location patterns.
Suburban Subcenters and Modern Cities
Anas. Alex. Richard Arnott. and Kenneth A. Small. "Urban Spatial Structure." Journal of
Economic Literature 34 1998). pp. 1426-64. Discusses the historical changes in
(
land-use patterns in cities and the transition from the monocentric city of the
nineteenth century to the modern polycentric city with subcenters.
Bollinger. Christopher R.. Keith R. Ihlanfeldt. and David R. Bowes. "Spatial Variation in
Office Rents within the Atlanta Region." Urban Studies 35 1988). pp. 1097-1 118.
(
Shows that sites that provide better opportunities for face-to-face contact command
higher office rents. Suggests that technological advances in telecommunications have
not diminished the role of face-to-face agglomeration economies in determining office
rents. Provides evidence that office rents in the central city are lower in comparison to
the rest of the metropolitan area (suburban subcenters).
Garreau. Joel. Edge Cin: Life on the New Frontier. New York: Doubleday. 1991. A
description of the edge-city phenomenon with several case studies.
Giuliano. Genevieve, and Kenneth Small. "Subcenters in the Los Angeles Region."
Regional Science and Urban Economics 21 ( 1991 ). Discusses the characteristics of the
CBD and 28 subcenters in the Los Angeles metropolitan area.
Graham. Stephen. "Telecommunications and the Future of Cities: Debunking the Myths."
Cities 14 ( 1997). pp. 2 1-9. Identifies five "myths" concerning the effects of new
communication technology on cities and explains why the new technology will not
cause the demise of cities, but will actually cause cities to thrive.
Helsley. Robert, and Arthur Sullivan. "Urban Subcenter Formation." Regional Science and
Urban Economics 21 (1991 pp. 255-75. Explores the economic forces behind the
).
development of subcenters.
Henderson. Vernon, and Arindam Mitra. "The New Urban Landscape: Developers and Edge
Cities." Regional Science and I 'rban Economics 26 1996). pp. 613—43. Explores the
(
environment and decision-making process that lead to the development of edge cities
Ihlanfeldt. Keith R. "The Importance of the Central City to the Regional and National
Economy: A Rex iew o( the Arguments and Empirical Evidence." Cityscapes: A
Journal of Policy Development and Research 1 1 1995), pp. 125-50. Summarizes the
results of studies that explore the relationship between central cities and suburbs.
McDonald. John F. and Daniel McMillen. "Employment Subcenters and Land Values
in a Polycentric Urban Area: The Case of Chicago." Environment and Planning.
A 22 (1990), pp. 1561-74. Identifies four employment subcenters in the Chicago

metropolitan area and discusses the effects of subcenters on residential land values.
McMillen, Daniel P. "One Hundred Fifty Years of Land Values in Chicago: A
Nonparametric Approach." Journal of Urban Economics 40 (1996), pp. 100-24.

Explores the changes in land-rent functions over time. Shows that despite the
development of over a dozen subcenters, there is a marked peak in land values in the
CBD.
McMillen, Daniel P., and John F. McDonald. "Suburban Subcenters and Employment
Density Metropolitan Chicago." Journal of Urban Economics 43 (1998), pp.
in
157-80. Identifies employment subcenters in the Chicago area and explores the effects
of the subcenters on employment density.
Mieszkowski, Peter, and Barton Smith. "Urban Decentralization in the Sunbelt: The Case of
Houston." Regional Science and Urban Economics 21 (1991). Discusses the
characteristics of the CBD and 10 subcenters in the Houston metropolitan area.
Stanbach, Thomas M. The New Suburbanization: Challenge to the Central City. Boulder,
Colo.: Westview Press, 1991 Explores changes in employment in suburbs and
. central
cities and their implications for the competitiveness of the central city.
Swartz, Alex. "Corporate Service Linkages in Large Metropolitan Areas: A Study of New
York. Los Angeles, and Chicago." Urban Affairs Quarterly 28 (1992), pp. 276-96.
Explores the interactions between suburban and central-city firms involving financial
and professional services.
Future Location Patterns

Anas, Alex, and Leon Moses. "Transportation and Land Use in the Mature Metropolis." In
The Mature Metropolis, ed. Charles L. Leven. Lexington, Mass.: Heath, Lexington,
1978.
Macrae, Norman. "Tomorrow's Agglomeration Economies." In The Mature Metropolis, ed.
Charles L. Leven. Lexington, Mass.: Heath, Lexington, 1978. Predicts future
development patterns based on predictions about agglomeration economies.
Chapter 11
D
A N - U S C CONTROLS
and ZONING
(~J7~
his chapter discusses the government role in the urban land market. The
<_y discussions in Chapters 7 through 10 assume that land is allocated to the
highest bidder, that is, the government is not involved in the urban land market. In
governments use a number of policies to control land use. Most cities have
fact, local
zoning plans that limit the location choices of most activities. Some cities use zoning
and other land-use controls to limit population growth. This chapter addresses three
basic questions related to various types of land-use controls. First, why do cities
control land use? Second, what are the market effects of land-use controls? Third,
what are the legal foundations for zoning and other land-use controls?
Controlling Population Growth

Some local governments use land-use policies to limit population growth. The pur-
pose of a growth-control policy is to control the undesirable side effects of growth
(pollution, congestion, crime, the loss of a small-town atmosphere). This section
discusses two growth-control policies: urban service boundaries and limits on the
number of building permits issued.
Urban Growth Boundary or Urban Service Boundary

One way growth is to limit the land area of the city. This can be
to control population
done by outlawing development beyond an urban growth boundary. This
directly,
is the approach adopted by the city of Portland, Oregon. Alternatively, the city can
refuse to extend urban services (e.g., water, sewers, roads, schools, parks) beyond
an urban service boundary, an approach used by Boulder. Colorado. In both cases,
growth is limited to the area within the boundary.
Figure 1 1-1 shows the equilibrium land-use pattern in a monocentric city. The
residential bid-rent function intersects the agricultural bid-rent function at a distance
303
304 Part II Land Rem and Urban Land-Use Patterns
Figure 11-1 Partial-Equilibrium Effect of an Urban Service Boundary
Business bid-rent
function
Discontinuity with service

boundarv
Distance to cit) center mile-,

i i
In themarket equilibrium, the CBD radius in two miles and the cit) radius is eight miles An urban sen ice
boundary prevents residential development bevond six miles, so the residential bid rent drops to zero
between six and eight miles from the city center. This is the partial-equilibrium effect of the service
boundarv.
of eight miles from the city eenter. so the city radius is eight miles. An implicit
assumption of this model is that the city government provides urban services out to a
distance of eight miles from the city center. The business bid-rent function intersects
the residential bid-rent function at a distance of two miles from the cit) center, so
the equilibrium CBD radius is two miles.
Suppose that the cit) refuses to extend urban services beyond si\ miles from
the city center. Unless residents can provide their own roads, sewage systems, and
schools, residential development outside the boundar\ is impossible. Therefore, the
residential bid rent drops to zero at a distance o\' six miles from the city center. This
is shown in Figure 11-1 as a discontinuity in the residential bid-rent function at
a distance of six miles. This is the partial-equilibrium effect o\' the urban service
boundary: land beyond it = 6 is rendered uninhabitable, so the market rent on land
outside the boundary drops to the agricultural bid rent.
What are the general-equilibrium effects of the urban sen ice boundary? The
service boundary decreases the si/e of the residential district, decreasing the city's
Chapter II Land-Use Controls and Zoning 305
total labor supply and thus increasing the city's wage. The increase in the wage
increases the relative attractiveness of the city, causing migration that bids up
the prices of housing and land. The quantity of labor supplied increases for two
reasons:
1 . Increased population density. The increases in housing and land prices

decrease households occupy smaller dwellings, and housing
lot sizes:
producers use less land per unit of housing. The number of households
(and workers) per acre increases.
2. Increased residential territory. The upward shift of the residential
bid-rent function allows households to outbid businesses for land near the
CBD border. Therefore, the residential district expands inward toward the
city center.
The increase in the wage also affects the market for business land. Production costs
increase, decreasing the relative attractiveness of the city as a production site. The
demand for CBD land decreases, causing a downward shift of the business bid-rent
function. The quantity of labor demanded decreases for two reasons:
1 Decreased employment density. The increase in the relative price of labor

decreases the city's employment density (the number of laborers per acre).
2. Decreased business territory. The CBD shrinks as businesses lose land to
the residential district.
The city's wage continues to rise until general equilibrium (labor supply equals labor
demand) is restored.
Figure 1-2 shows the new equilibrium allocation of land. The CBD radius is
1
1.5 miles (down from 2 miles), and the city radius is 6 miles (the radius dictated
by the urban service boundary). The service boundary has different effects in the
city's two land markets. In the residential district, the service boundary increases
the prices of housing and land, thus increasing population density. In contrast, the
service boundary decreases both land rent and employment density in the CBD.
Who gains and who loses from the service boundary? People who own land
outside the service boundary lose: for land between u = 6 and u — 8, the market
rent drops to the agricultural bid rent, decreasing the market value of land. People
who own residential land within the service boundary win: the increased demand
for residential land (caused by the increase in the wage) bids up the price of land
within the boundary. People who own land within the CBD lose: the decrease in the
demand for CBD land (caused by the increase in the wage) decreases the price of
business land.
Building Permits
Some cities control residential growth by limiting the number of building permits
issued. By limiting the number of new dwellings built per year, a city can control its
growth rate.
Figirk 11-2 General-Equilibrium Effect of an Urban Service

Boundary
Business bid-rent function

with higher wage
Residential bid-rent function

with higher waste
Agricultural
bid-rent
function
Distance to city center miles)

i
The urban service boundary decreases the supph of labor, increasing ihe city 's wage. The increase in the
wage shifts the residential bid-rent function upward (increasing the quantity of labor supplied and shifts the
i
business bid-rent function downward (decreasing the quantity iA labor demanded!. General equilibrium is
restored with a smaller, less dense CBD and a more dense residential district.
Consider a city that initially has no limit on the number of building permits
issued. Figure 11-3 shows the city's market for new housing: the supply curve
AG intersects the demand curve at point F, generating an equilibrium quantity of
100 houses per year and an equilibrium price o\ $50,000 per house. The city issues
100 tree permits per year, which are used to build houses with a market value of
S50.000. Developers make zero economic profits (normal accounting profits), mean-
ing thai the cost of producing each house (the sum oi land, labor, and capital cost)
is $50,000.
Suppose that the city limits the number of building permits to 60 per year. The
new supply curve for housing is ACE: the maximum number of new bouses is 60. so
the supply curve becomes vertical at 60 dwellings. The new supply curve intersects
the demand curve at an equilibrium price of S70.000. In other words, the permit
policy increases the equilibrium price of housing by $20,000.
Chapter 1 1 Laiul-U.se Controls and Zoning 307
Figure 11-3 Market Effects of Building Permits
E
-^
^ Supply curve with 60
building permits
H
7(1
\. ^G
\. ^^^^ Initial supply
50
^\^^^^ curve
.S 40 \.
c
A""^ Demand
1
curve
60 100
New houses per year
If the city decreases the supply of building permits from 100 to 60, the supply curve for housing
shifts from H
AG to ACE. The new equilibrium is point (price of housing = $70,000). Point C
shows the cost of producing 60 houses ($40,000). The difference between the price and the cost
is the equilibrium price of the building permit ($30,000).
The permit policy also decreases the cost of producing housing. The permit
policy decreases the number of houses built, so it decreases the demand for land.
For example, if houses are built on quarter-acre lots, the permit policy decreases the
demand from 25 acres per year (100 times 0.25) to 15 acres per year. The
for land
decrease in the demand for land decreases the market price of land, decreasing
the cost of producing housing. This is shown by the housing supply curve: if 100
houses are built, the production cost is $50,000 per house; if only 60 houses are
built, the cost drops to $40,000 per house. Because the permit policy decreases the
number of houses built, it decreases the price of land, allowing developers to build
houses at a lower cost.
The city must decide how to allocate the 60 building permits among its de-
velopers. One option is to auction the permits to the highest bidders. What is the
monetary value of a building permit? A person with a building permit can make a
profit equal to the difference between the market price of a house ($70,000) and the
cost of producing the house ($40,000), so the monetary value of a permit is $30,000.
If the city auctions the permits, the market price would be $30,000.
Some cities allocate their building permits to development projects that promote
their development objectives. If the city is interested in high-density housing, it could
308 Part II Land Rent and I 'rban Land-Use Patterns
allocate the permits to a high-density housing project. Alternatively, the city could
allocate the permits to a project in an area targeted for development. Many cities
use sophisticated point systems to rate and rank alternative development proposals.
Among the criteria for which points are assigned are various design features (the
exterior appearance of the dwellings, the amount of open space within the project),
density, and the proximity of the project to existing public infrastructure (roads,
schools, sewer lines).
Land-Use Zoning: Types and Market Effects

A zoning plan designates a set of admissible land uses for each plot of land in a city.
In theory, the purpose of zoning is to promote public health, safety, and welfare.
The principal means of promoting health, safety,and welfare is the separation of
incompatible land uses.
The first zoning ordinances were used by the wealthy to exclude the poor. In the
1880s. the city of San Francisco passed laws to segregate its Chinese population.
When was declared unconstitutional, the city passed a zoning
explicit segregation
law that banned laundries from certain neighborhoods. The zoning law did not violate
the Constitution because it promoted public welfare by keeping an undesirable land
use (laundries) out of some residential areas. Because the Chinese operated most of
the city's laundries, the zoning law provided a legal means of segregation. In New
York City in 1916, the rapid growth of employment in offices and garment factories
increased the number of low-wage women on the streets of midtown Manhattan.
The owners of exclusive Fifth Avenue stores feared that the increased traffic of poor
women would be bad for business, and proposed zoning laws that limited building
sizes and building heights. The law limited the growth of office buildings and garment
factories, decreasing the number of clerical and garment workers walking along Fifth
Avenue. The zoning law was declared constitutional because it promoted public
welfare by controlling the nuisances created by tall buildings (shadows, blocked
light, disruption of views).
There are many types of zoning, each of which has at least one purpose. This
section divides zoning into three types. The purpose of nuisance zoning is to separate
incompatible land uses. Cities use fiscal zoning to exclude households that would
not pay their share of the costs o\' local government. Design zoning is a sort of
macroarchitecture: planners arrange activities within the city to promote the efficient
use of the city's infrastructure.
Nuisance Zoning
Nuisance zoning (or externality zoning) is the practice of separating land uses that
are considered incompatible. The classic example is the glue factory: one way of
dealing with the air pollution generated h\ the glue factor) is to move the factory into
an industrial zone, far from residential areas. Similarly, the externalities generated
by retailers (traffic congestion and noise) can be controlled by establishing a retail
zone. There are also externalities generated by high-density housing: apartment

buildings cause traffic and parking problems, and also block views and light. These
externalities can be controlled by establishing a zone for high-density housing.
Industrial Nuisances. Industrial firms generate all types of externalities, includ-

ing noise, glare, dust, odor, vibration, and smoke. Zoning separates residential and
industrial land uses, reducing the exposure to air and noise pollution. Zoning is
appealing as an environmental policy because of its simplicity: the easiest way to

limit exposure to pollution is to separate polluters from their potential victims. The
problem with zoning as an environmental policy is that zoning does not reduce the
total amount of pollution, it just moves it around. If one municipality's industrial
zone is far from its own residents but close to the residents of a nearby municipality,
industrial zoning may actually increase the total exposure to pollution. Because zon-
ing does not provide firms with the incentive to decrease pollution, it is less effective
than other environmental policies.
Alternative to Industrial Zoning: Spatial Effluent Fees. When economists hear

the word pollution, their knee-jerk response is "effluent fees." An effluent fee is a
tax on pollution. For efficiency purposes, the fee should be set equal to the marginal
external cost of pollution, that is, the cost to society of an additional unit of pollution.
For example, if one ton of sulfur dioxide generates a social cost of $2. the effluent fee
should be $2 per ton. The effluent fee "internalizes" the pollution externality, forcing
same way that they pay for labor, capital, and
polluters to pay for pollution in the
raw materials. The firm has an incentive to decrease pollution because abatement
decreases the firm's effluent bill. If the effluent fee equals the marginal external cost
of pollution, firms generate the optimum level of pollution.
In theory, a system of effluent fees would generate both the optimum level of
pollution and the optimum spatial distribution of pollution. If the effluent fee varies
across space, a firm would pay a higher effluent fee for pollution generated near
sensitive areas (e.g., residential areas). For example, the effluent fee might be $2 per
ton of pollution if the firm is 10 miles from the residential area and $5 per ton if the
firm 4 miles from the residential area. Under a system of spatial effluent fees, the
is
firm would base both its production decisions and its location decision on effluent
fees. If the effluent fees are set equal to the marginal external cost of pollution, the
firm would choose both the optimum location and the optimum level of pollution.
To explain the differences between zoning and effluent fees, consider a rectan-
gular city with the following characteristics:
1 Residents live in the western part of the city and commute eastward to a
polluting steel mill.
2. The longer the distance between the residential area and the mill, the higher
the wage: workers are compensated for commuting cost in the form of
higher wages. The firm's total labor cost equals the wage (dollars per
worker) times the total quantity of labor.
Figire 11^4 Spatial Effluent Fees and Optimum Location
Pollution cost
2 4.2 6 8 10
Distance from residential area miles ( i
Total cost is the sum of labor cost and pollution cost. As the polluting firm moves tow ard the residential
district, pollution cost increases (more residents are affected b) pollution, so the effluent fee increases),
but labor cost decreases (commuting distances decrease, so the wage decreases). Total cost is minimized
at a distance of 4.2 miles from the residential area.
3. The longer the distance between the mill and the residential area, the lower
the effluent fee. The firm's total pollution cost equals the effluent fee
(dollars per ton) times the amount of pollution.
4. The city starts with an industrial zoning policy under which the steel mill is
located 10 miles from the residential area.
Figure 1 1^4 shows the total cost of producing steel at different locations. The
positively sloped curve which increases from $20 for a location
shows labor cost,
adjacent to the residential district to $63 in the industrial zone (10 miles from the
residential area). The negatively sloped curve shows pollution cost, which decreases
from $60 at a location adjacent to the residential district to $5 in the industrial
zone. The variation in the pollution cost reflects the variation in the effluent fee. The
U-shaped curve shows total cost, the sum of labor cost and pollution cost, which
decreases from $80 for a location adjacent to the residential district to $53 for a
location 4.2 miles from the residential area, then increases to $70 in the industrial
zone.
Suppose the firm starts in the industrial zone. As the firm moves toward the resi-
dential district, there are benefits and costs. On the benefit side, labor cost decreases
Chapter 1 1 Land-Use Controls and Zoning 311
as the firmmoves closer to its workforce. On the cost side, the effluent fee increases
because the firm's pollution affects more people, so pollution cost increases. Given
the shapes of the labor-cost and pollution-cost curves, total cost is minimized at a
location 4.2 miles from the residential district. If the city switches from a zoning
policy to an effluent-fee policy, the steel producer would move from the industrial
zone to a location 4.2 miles from the residential district, decreasing its total cost
from $70 to $53.

The effluent-fee policy is more efficient than the zoning policy for two reasons.
First, the effluent policy allows the firm to choose the location that minimizes pro-
duction cost. From the social perspective, the new location is more efficient because
the savings in commuting cost (as indicated by the labor-cost curve) exceed the
increase in pollution cost (as indicated by the pollution-cost curve). Second, the
effluent-fee policy forces the firm to pay for pollution, so pollution decreases to its
optimum level.
Will all the polluting firms leave the industrial zone when the city switches
from industrial zoning to effluent fees? A firm will stay in the industrial zone if the
pollution-cost curve is steeper than the labor-cost curve. In this case, a move toward
the residential district increases pollution cost by more than it decreases labor cost,
so total production cost increases. Nonetheless, the effluent-fee policy is still more
efficient than industrial zoning because the effluent-fee policy decreases pollution
to its optimum level.
Why do cities use zoning instead of effluent fees to control industrial pollution?
There are two principal reasons. First, industrial zoning is simple compared to a
system of spatial effluent fees. To set the effluent fees, the city would have to estimate
the marginal external cost of pollution for different locations in the city, and to collect
the fees, the city would have to monitor the polluting firms.would certainly be
It
easier to put all the polluters in an industrial zone. Second, a switch from zoning to
effluent fees may increase pollution in some neighborhoods. Although the factory
produces less pollution under the effluent-fee policy, it may locate closer to the
residential district. Therefore, some neighborhoods may become more polluted. The
fact that pollution is at its optimum level is small consolation to those who breathe
the dirty air. The switch would improve efficiency and generate a
to effluent fees
net gain for society, so would be possible, in principle, to compensate those who
it
breathe dirtier air. In fact, compensation is rarely attempted, so local opposition to
effluent fees remains.
Performance Zoning. Traditional zoning is based on a zoning map and a land-use

list. The zoning map shows the zoning classification for each plot of land in the city,
and the list shows the admissible land uses for each zoning classification. If an activity
is designated heavy industry, everv firm involved in this activity must locate in the
heavy-industry zone, regardless of how much noise, odor, or smoke the individual
firms actually generate. The alternative to traditional zoning is performance zoning,
under which the city sets performance standards for each zone. For industrial uses,
performance standards typically set upper limits for the amount of noise, glare,
odor, vibration, gas, and smoke emitted by the firms. Performance zoning is a sort
of compromise between traditional zoning and effluent fees: although performance

zoning does not force firms to pay for pollution, it does encourage abatement because
it gives cleaner firms more location options.
Retail Nuisances. Retailers generate a number of externalities that affect nearby

residents. The traffic generated by retailers causes congestion, noise, and parking
problems. A traditional zoning plan establishes retail zones to decrease the exposure
For example, zoning prevents the invasion of quiet residential
to these externalities.
neighborhoods by shoppers and delivery trucks.
Under performance zoning, the city allows retailers in a particular zone as long
as they satisfy performance standards for parking, traffic, and noise. For each area,
the city may specify a minimum number of parking places, a minimum speed for
and a maximum noise level. These performance standards force commercial
traffic,
developments to provide off-street parking to control parking problems, signalization

and street improvements to control traffic problems, and berms and landscaping
to control noise problems. Performance zoning allows the mixing of commercial
and residential land uses because retailers take steps that shield residents from the
undesirable effects of commercial development.
Residential Nuisances. Most residential externalities are generated by high-

density housing. Suppose that a developer builds a four-story apartment complex in
a neighborhood with single-family homes. The apartment complex increases traf-
ficvolume, increasing congestion and noise, and increases the demand for on-street
parking, causing a shortage of parking spaces. In addition, the tall apartment building
is likely to deprive neighbors of scenic views and sunlight.
Under conventional zoning, the apartment building would be excluded from
the low-density neighborhood. Conventional zoning shields the residents of single-
family homes from externalities by excluding high-density housing.
The alternative to conventional zoning is performance zoning. Under a
performance-based zoning policy, the apartment complex would be allowed if the
developer took the following steps:
1 . Off-street parking. The developer provides enough off-street parking to

prevent parking problems.
2. Street improvements. The developer pays for street improvements that
prevent congestion problems.
3. Building design. The building is designed to prevent loss of views and
sunlight. The developer uses landscaping to establish a buffer between the
apartment building and the single-family homes.
The idea behind performance zoning is that the apartment building should be judged
on the basis of its actual effects on the neighborhood, not on the simple fact that it
is high-density housing. If a high-density project docs not generate externalities, it
would be allowed in the neighborhood.

Fiscal Zoning
The second type of zoning occurs because local governments finance public services
with the property tax. Under fiscal zoning, a city excludes households that would
impose fiscal burdens on local government. Suppose that a household consumes
$3,000 worth of local services but pays only $2,000 in local property taxes. Such
a household generates a fiscal burden because its tax contribution is less than the
cost of its local public services. Fiscal burdens may be generated by ( 1 ) households
living in high-density housing, (2) households living in the fringe areas of the city,
and (3) new commercial and industrial development.
High-Density Housing. As explained in Chapter 18 (Local Taxes and Intergov-

ernmental Grants), the most important source of local revenue is the property tax.
Because a household's tax liability increases with its housing consumption, a house-
hold in a small house or apartment pays a relatively small amount in taxes, and is
therefore more likely to impose a fiscal burden on local government. Local govern-
ments exclude households by zoning for low-density (high-value) housing.
Large-lot zoning establishes a minimum lot size for residential development.
Land and housing are complementary goods, so the city can use restrictions on
lot size to indirectly establish a minimum house value. Suppose that a city breaks
even on a house worth $125,000: the tax liability of such a house equals the cost
of supplying public services to the household living in the house. According to one
rule of thumb, the market value of land is about 20 percent of the total property
value (the value of the structure and the land). If the price of land is $50,000 per
acre, a half-acre lot ($25,000 worth of land) produces a house worth $125,000 (five
times $25,000). A minimum lot size of one-half an acre ensures that the government
breaks even on all new houses.
To explain the effects of large-lot zoning on the urban land market, consider
an unzoned city with 120 acres of undeveloped land. As shown in Figure 1-5, the 1
equilibrium price of land is $20,000 per acre. At this price, the total demand for land
equals the fixed supply, and land is divided equally between single-family homes
(SFHs) and apartments (60 acres in each).
Suppose that the city passes a zoning ordinance that limits apartment buildings
to a total of 40 acres of land. What happens to the equilibrium price of land? In
Figure 1 1-5, the city moves up the demand curve for apartment land from point B
to point E: since only 40 acres are available for apartments, developers arc willing
to pay $40,000 per acre for apartment land. At the same time, the city moves down
the other demand curve from point A to point C: there are now 80 acres available
for single-family homes, and the surplus of land decreases the price per acre from
$20,000 to $15,000.
This example shows the effects of large-lot zoning on landowners and housing
consumers. Zoning makes land for apartments a relatively scarce commodity, so the
owners of apartment land gain from large-lot zoning. In contrast, zoning makes land
for single-family homes a more plentiful commodity, so the owners of land destined
for single-family homes lose. What about housing consumers? The increase in the
Figure 11-5 Large-Lot Zoning and the Price of Land
Demand for
SFH land
20 40 60
Land for single-family homes (acres) Land for apartment buildings (acres)
A amount of land (120 acres) is divided between single-family homes and apartments. In the absence of large-lot zoning, the price of
fixed
land $20,000 per acre, and land is divided equally between apartments and homes (60 acres for each type of housing: points A and B).
is
If apartment acreage is limited to 40 acres (point E ). 80 acres are left for single-family homes (point C ). The price of apartment land rises
to S40.000, and the price of land for single-family homes drops to $15,000.
cost of apartment land increases the costs of apartments, so apartment dwellers lose.
In contrast, the land cost of single-family homes decreases, so home buyers gain.
Exclusionary and Inclusionary Zoning. Large-lot zoning is often labeled exclu-

sionary zoning because it excludes low-income households. The exclusion is both
direct and indirect. In the example above, large-lot zoning prevented the construc-
tion of 20 acres of apartments. Because low-income households are more likely to
occupy apartments, the decrease in the number of apartment buildings decreases the
number of low-income residents. In addition, the zoning policy increases the price
of apartments, indirectly excluding low-income households: the apartments that are
built are too expensive for many poor households.
In response to the exclusion problem, many cities have developed inclusionary
zoning policies. Under an inclusionary housing program, the local government forces
local developers to build dwellings for low-income households. The dwellings are
typically sold to low-income households for less than the cost of construction. By
providing subsidized dwellings for low-income residents, the local community partly
offsets the exclusionary effects of fiscal zoning. Most dwellings built for inclusionary
purposes are not typical low-income dwellings, but are medium-income dwellings
provided to the poor at the price of low-income dwellings.
Who pays for inclusionary zoning? The requirement of subsidized housing has
the same effect as a development tax. Suppose that a developer builds five inclu-
sionary units and loses $10,000 per unit, generating a total loss of $50,000. If the
developer is allowed to build 50 houses for the open market, the effective tax per
market dwelling is $1 ,000 ($50,000 divided by 50). The developer makes zero eco-
nomic profit (normal accounting profit) with or without inclusionary zoning, so the
implicit tax is passed on to consumers (housing prices increase) and landowners (the
price of vacant land decreases). In other words, housing consumers and landowners
pay for inclusionary zoning.
Fringe Land Use. New housing is usually built on the fringe of the metropolitan
area. If the costs of supplying public services are higher in fringe areas, the tax
new housing may be less than the cost of public
contribution of services. Therefore,
new housing may impose a fiscal burden on the city.
To explain the fiscal burden of fringe housing, consider a city with the following
characteristics:
1 The current average cost of public services is $500 per household per year.
2. The typical household owns a house worth $50,000 and pays a property tax
of $500 per year percent of market value).
( 1
3. A developer has proposed a new residential development on fringe land.

The market value of the new houses is the same as existing houses
($50,000).
4. The new development would require the expansion of the city's public
services (widening of streets, expansion of the sewage system, extension of
bus routes). The full cost of serving the new houses would be $600 per
house per year.
Given these assumptions, every new house generates a fiscal burden of $ 00 per year 1
($600 — $500). A simple way to prevent the new development and the associated
fiscal burden is to zone the vacant land for agricultural use. If the city sets a minimum
lot size of 20 acres, no additional houses will be built on the land, so the expansion
of city services is unnecessary.

An alternative to low-density zoning is a tax surcharge for new development.
Suppose that the city imposes an annual surcharge of $100 per new house. The
annual tax liability of new housing would be $600 ($500 in property tax plus the
$ 00 surcharge), and the occupants of new housing would pay the
1 full cost of the new
public services. The surcharge would eliminate the fiscal burden of new housing, so
the city would be more likely to approve the development project.
An alternative to an annual tax surcharge is a one-time impact fee or develop-
ment tax. By paying an impact fee, the developer compensates the local government
for the fiscal burden of new housing. If the city collects a $1,000 impact fee from
the developer and invests the money in the bank at a 10 percent interest rate, it can
use the annual interest earnings ($100) to cover the difference between the tax con-
tribution and the public-service cost of new housing. Who pays the impact fee? In a
competitive market, the developer makes zero economic profits, so the fee is passed
on to consumers (housing prices increase) and landowners (land prices decrease).
Lillydahl et al. (1988) describe impact-fee policies in four states: California,
Florida, Oregon, and Colorado. In California, local communities are allowed to
316 Part II Lciiul Rent and Urban Land-Use Patients
impose fees for transportation facilities (bridges, major thoroughfares, and free-
ways), drainage facilities, sewers, parks, and schools. In 1985. the average impact
charge was $3,527 per dwelling in the San Francisco Bay area and S9.500 in San
Diego. In Florida, the most common impact fees are for water and sewers: in 1 985. the
average impact fee was about S3.000 per dwelling. In Oregon, impact fees are called
systems development charges, and are typically used to support water, sewer, street,
and park During the 1 980s, impact fees averaged about $2,500 per dwelling
facilities.
in the Portland metropolitan area. In Colorado, cities along the Front Range (from
Colorado Springs to Fort Collins and including the Denver area) impose impact fees
most frequently for water, sewers, parks and recreation, drainage, and street facilities.
Commercial and Industrial Development. In some cities, commercial and in-

dustrial development generates fiscal burdens. If the city cannot pass on the cost of
new infrastructure to new firms and employees, it may restrict development by (1)
limiting the supply of commercial and industrial land and 2 restricting building
( )
heights.
The city of San Francisco recently adopted a zoning plan that restricts building
heights in its downtown area. The objective of the policy is to limit the number
of people working downtown. By limiting the number of downtown workers, city

officials hope to control the volume of traffic on the bridges into the city, all of which
are already congested. An alternative to the zoning policy would be to increase the
carrying capacity of the bridges, with the cost of the capacity expansion imposed
on new employers. Since this is impractical, the city has decided to use its zoning
policy to limit total employment.
Some cities impose impact fees on commercial and industrial developers, using
the revenue from the fees to expand local transportation networks. For example,
in the Westchester area of western Los Angeles, developers pay a one-time fee of
$2,010 for each additional rush-hour trip generated by new office buildings. The
revenue from the impact fee is used to widen the roads used by the employees of the
new office buildings. Impact fees can reduce the fiscal burden of new development,
decreasing the opposition to development.
Design Zoning
The third type of zoning is a form of macroarchitecture. Just as an architect designs
an individual house, the planner designs a city, arranging activities to promote the
efficient use of the city's infrastructure (streets, sewage systems, water systems).
Residential and employment growth is directed to areas where infrastructure can be
efficiently provided. Design zoning is also used to preserve open space.
Directed Development. Some cities use zoning to direct residential development

to particular areas. To explain the effect of such zoning, consider a metropolitan area
with the following characteristics:
1. There are 50 acres of vacant land: 25 acres to the north of the city and 25
acres to the south.
2. The city wants to accommodate 500 new households on its vacant land.
3. Under the initial zoning ordinance, both areas are zoned for 10 dwellings
per acre: 250 households would live in the north and 250 would live in the
south.
Suppose that the city changes its zoning policy. It decides to preserve the north
area and accommodate 500 households in the south area. It could do so by zoning
all
the north area for low-density use and the south area for 20 dwellings per acre. What
are the effects of the rezoning on land prices? The rezoning decreases the market
price of northern land because the landowner has fewer land-use options. In contrast,
the southern landowner has more options, so the price of southern land increases.
The change in zoning policy is clearly inequitable: southern landowners gain at the
expense of northern landowners.
An alternative to rezoning is a system of transferable development rights
(TDRs). Under a TDR policy, the city establishes a preservation zone (the north
area) and a development zone (the south area). The south area is zoned for 10
dwellings per acre, giving Mr. South (the southern landowner) the right to build
a total of 250 dwellings. Ms. North (the northern landowner) is not allowed to
develop her land, but is instead issued 250 development coupons. These development
coupons can be used to override zoning restrictions in the development zone. If South
wants to build 20 dwellings on one acre of land, he must purchase 10 development
coupons from North. The TDR policy gives each landowner development rights (the
right to build 250 houses) and allows the owners of preserved land to transfer their
development rights to other areas of the city. When the northern landowner sells her
coupons, she is at least partially compensated for the losses in property value caused
by the rezoning of her land.
Figure 1 1 -6 shows the effects of zoning and TDRs on the prices of southern and
northern land. In the initial equilibrium (both areas zoned for 10 dwellings per acre),
land sells for $50,000 per acre. Suppose that northern land is zoned for preservation,
but southern land is still zoned for 10 dwellings per acre. The rezoning of northern
land decreases its price to $30,000 because the landowner has fewer development
options. When northern land is taken off the residential market, the resulting shortage
of residential land increases the price of southern land to $60,000. Under a TDR pol-
icy, northern landowners receive development coupons that can be used to override
the density limits (10 dwellings per acre) in the development zone. The relaxation
of density limits gives southern landowners more development options, increasing
the market value of their land from $60,000 to $75,000. Southern landowners are
willing to pay up to $15,000 for an acre's worth of development coupons, so the
market value of an acre's worth of transferable development rights is $15,000.
The TDRs diminish the inequities caused by preservation zoning. In the absence
of the transferable rights, the owner of the preserved land suffers a loss of $20,000
per acre ($50,000 - $30,000). In contrast, if the allowable density is 20 dwellings
per acre, the southern landowner receives a gain of $25,000 per acre. Under the TDR
policy, the owner of the preserved land sells her development rights to the landowners
in the development zone for $15,000 per acre, so her net loss from the zoning policy
is only $5,000. In contrast, the landowner in the development zone pays $ 5.000 per 1
Figure 11-6 Market Effects of Rezoning and Transferable

Development
Z 60
Southern land Northern land

development /one preservation /one
In the initial equilibrium, the price of land is S50.000 per acre. If the north area is re/oned for
preservation, the demand for southern land will increase (increasing its price to S60.000). and
the demand for northern land will decrease (decreasing its price to S30.000). Under a TDR
policy, northern landowners receive development coupons that can be used to override densitx
limits in thedevelopment zone. The relaxation of density limns increases the market \alue of
southern land from $60,000 to S75.000. so southern landowners are w illing to pay up to $15,000
for an acre's worth of development coupons.
acre for the rights to build more than 10 dwellings per acre, so his net gain from the
zoning policy is only SI 0.000 per acre.
What determines the market value of the development coupons? The coupons
have a positive price because they allow landowners in the development zone to
exceed the normal density restrictions. depends on the power of

The market price
the coupons coupons allow a 100 percent
to override density restrictions: if the
increase in density, they are obviously more valuable than they would be under a
15 percent density bonus. The market price also depends on the underlying demand
for high-density housing in the development zone. If households in the development
demand low -density housing, a developer won't need the development coupons, and
the market price of the coupons w ill be relatively low. In contrast, if there is a strong
demand for high-density housing in the development zone, the market price of the
coupons will be relatively high.
Open-Space Zoning. Some cities zone parcels of land as "open space.'" "green
belts." or "agricultural preserves." This type of zoning provides citj dwellers with
open space by denying landowners the full use of their land.
( 'hapter 1 1 Land-Use Controls and TLoning 319
Figure 11-7 Inefficiency of Open-Space Zoning
Marginal benefit of
open space
Marginal cost
of open space
60 LOO
Acres of open space preserved
The opportunity cost of open space is the value ofland in an alternative use (housing). The opti-
mum amount of open space is 60 acres (where the marginal-benefit curve intersects the marginal-
cosl curve).II the government does not compensate landowners for open-space zoning, it will
preserve 100 acres (the point at which the marginal benefit of open space is zero).
Is open-space zoning efficient? The alternative to zoning is the outright purchase

of land for open space. If the city were to purchase land instead of zoning it for open
space, the city would preserve land only if the marginal benefit of open space exceeds
it chooses the optimum
the marginal cost. If the city faces the full cost of open space,
amount of open space. In contrast, if the city simply zones the land for open space,
it is likely to consume too much open space.
Figure 1-7 shows the costs and benefits of open space. The marginal-benefit
1
curve is negatively sloped, reflecting the diminishing marginal utility of open space.
The marginal cost of open space is the opportunity cost of using the land for open
space instead of residential development. For example, if housing developers are
willing to pay $20,000 for undeveloped land, the marginal cost of open space is
$20,000 per acre. If the city purchases land for open space, it will outbid developers
for 60 acres of land. This is an efficient allocation of land: for the first 60 acres of
open space, the value of land as open space exceeds the value of land in housing.
For the 61st through the 100th acres, the value of land in housing exceeds the value
as open space, so 40 acres are developed.
Suppose that the city simply zones land for open space. From the perspective
of city officials, the marginal cost of open space is zero. The city preserves all land
for which the marginal benefit (from the demand curve) is positive, so it zones for
100 acres of open space instead of 60. This is inefficient because the 61st through
the 100th acres are more valuable as housing land.
A City without Zoning?

What would an unzoned city look like? Would glue factories and pizza parlors invade
quiet residential neighborhoods? Would land use be disorderly and ugly?
Some tentative answers to these questions come from Siegan's (1972) analysis
of Houston, the only metropolitan area in the United States without zoning. Land
use in the city is controlled by restrictive covenants, voluntary agreements among
landowners that limit land uses and structures. The covenants governing residential
subdivisions (over 7,000 in number) are typically more strict than conventional
zoning. They have detailed restrictions on architectural design, external appearance,
and lot maintenance. The covenants for industrial parks limit the activities that can
locate in the park.
How does Houston compare to zoned cities? Although a rigorous compari-
son of land-use patterns may be impossible, some tentative observations can be
made:
1. Dispersion of industry. The spatial distribution of Houston's industrial

firms is similar to that of zoned cities. As in other cities,Houston's
industrial firms locate close to the transportation network (near railroads
and highways) and tend to cluster to exploit localization economies.
2. Retailers. Like retailers in most cities, most retailers in Houston locate
along major thoroughfares in strip developments and shopping centers. Few
retailers locate in quiet residential neighborhoods. This is the same pattern
observed in zoned cities. Retailers locate along major thoroughfares
because such locations provide large volumes of foot and auto traffic. Many
retailers are subject to shopping externalities (from the benefits of one-stop
shopping and comparison shopping), so they cluster in shopping centers,
malls, and downtown areas.
3. Strip development. Houston appears to have more strip development (retail
and commercial establishments along arterial routes) than do zoned cities.
4. Apartments. Low-income housing is more plentiful and relatively

inexpensive. There is a wide range of densities in apartment projects: the
projects occupied by the wealthy have more open space and lower density,
while the projects occupied by the poor have higher density.
5. Single-family homes. In some areas, Houston homes have relatively small
backyards; in other areas, lot sizes are similar to those in zoned cities.
There are two lessons from Houston's experience without zoning. First, in the
absence of zoning, landowners have the incentive to negotiate restrictions on land

use. It seems that neighborhood externalities are large enough to justify the cost
of developing and enforcing restrictive covenants. This is the Coase solution to
externalities (named after R. H. Coase): the parties affected by externalities negotiate
a contract to solve the externality problem. Second, in the absence of zoning, most
industrial firms cluster in locations accessible to the transportation network, and
most retailers cluster in shopping centers and retail strips. If a city were to drop its
zoning plan, it is unlikely that glue factories and pizza parlors would invade quiet
residential neighborhoods.
The Legal Environment of Land-Use Controls

Local governments, which are creatures of state governments, derive their power to
control land use from the states. In most states, enabling legislation for zoning is
patterned after the Standard State Zoning Enabling Act, which was developed by
the U.S. Department of Commerce in 1926. Section 1 of the Enabling Act states:
Grant of Power. For the purpose of promoting health, safety, morals, or the general welfare of
the community, the legislative body of cities and incorporated villages is hereby empowered
to regulate and restrict the height, number of stories, and size of buildings and other structures,
the percentage of the lot that may be occupied, the size of yards, courts, and other open spaces,
the density of population, and the location and use of buildings, structures, and land for trade,
industry, residence, or other purposes.
The model legislation grants local governments broad powers in the regulation of
land use. Zoning is considered a legitimate exercise of the police power of local
government if it promotes the public health, safety, and welfare.
Current zoning laws are the result of over 60 years of legal decisions. In the
last six decades, individuals affected by specific zoning laws have sued local gov-
ernments, forcing state and federal courts to rule on the constitutionality of zoning
ordinances. If a particular type of zoning is declared unconstitutional, all cities get
the message from the and rewrite their zoning ordinances to drop the illegal
courts,
practices. On the other hand, if a zoning practice is upheld as constitutional, the prac-
tice spreads to other local governments. In other words, zoning law is evolutionary.
Early court decisions established three criteria for the constitutionality of zoning:
substantive due process, equal protection, and just compensation.
Substantive Due Process
The case of Euclid v. Ambler (1924) established the standards for substantive due
process. According to the due-process criterion, zoning must be executed for a
legitimate public purpose using reasonable means. In the early 1920s, the city of
Euclid. Ohio, enacted a zoning ordinance that restricted the location, size, and height
of various types of buildings. Ambler Realty had purchased some property between
the railroad tracks and a major thoroughfare, and expected to sell the land to an
industrial developer. When the city zoned its land for residential use, Ambler sued,
claiming that the separation of industrial and residential land uses did not serve a
legitimate public purpose. The Supreme Court ruled against Ambler, concluding
that the zoning ordinance satisfied the standards for substantive due process because
it had some "reasonable relation" to the promotion of "health, safety, morals, and
general welfare." In other words, nuisance zoning (separating different land uses) is a
legitimate use of the city's police power because it promotes public health and safety.
One interpretation of the Euclid decision is that a zoning ordinance is constitu-

tional as long as it generates some benefit for the local community. The court did not
say that the benefits of zoning must exceed its costs, but that the benefits of zoning
must be positive. Fischel (1985) calls this benefit analysis, as opposed to benefit-cost
analysis. The court defined the possible social benefits from zoning in broad terms,
to include monetary, physical, spiritual, and aesthetic benefits.
Equal Protection
The second criterion for the constitutionality of zoning is equal protection. The
equal-protection clause of the Fourteenth Amendment requires that all laws be ap-
plied in an impersonal (nondiscriminatory) fashion.

Recent court cases have tested whether fiscal zoning violates the principle of
equal protection. Fiscal zoning excludes households who do not pay their share
of the cost of local public services. Local governments have been sued by people
outside the city, who claim that fiscal zoning systematically excludes some types
of people from the city. The plaintiffs argue that zoning laws are not applied in an
impersonal manner, but instead treat some people differently than others: rich and
poor are treated differently, as are whites and blacks.
The federal courts have upheld the constitutionality of exclusionary zoning. In
the Euclid decision, theSupreme Court suggested that although a zoning ordinance
must generate some benefit for insiders (citizens of the community), the effects of
zoning on outsiders are unimportant. In Worth v. Selden 1975). the court dismissed (
the claims of outsiders because they did not prove that the zoning ordinance caused
specific personaldamage. In Village of Arlington Heights v. Metropolitan Housing
Corporation (1977). the court dismissed the claims of outsiders because they did not
prove discriminatory intent on the part of zoning officials. In Ybarra v. Town of Los
Altos Hills, the court ruled that although zoning laws that discriminate on the basis
of race are unconstitutional, zoning laws that discriminate on the basis of income
are legal. In general, the federal courts have adopted a noninterventionist approach
to exclusionary zoning.
Some state courts have adopted a more activist role. In Southern Burlington
County NAACP v. Mount Laurel ( 1975). the New
Supreme Court ruled that Jersey
Mount Laurel's exclusionary zoning harmed low-income outsiders. The court di-
rected the city to develop a new zoning plan under which the city would accom-
modate its "fair share" of low-income residents. The court established quotas for
communities to provide enough housing for low- and moderate-income workers to
live within reasonable commuting distances of their jobs. The effects have been
minor, in part because the state legislature modified the quotas and even allowed
communities to buy and sell up to half their quotas (Mills and Lubuele. 1997):
A nearby high-income community could sell up to half its quota to Trenton at a negotiated
price. The result was presumably to improve the housing, if not the schools and security,
of low-income residents, hut it hardly contributed to integration of suburbs or to increase in
affordable housing close to suburban workplaces.

Chapter II Land- Use Controls and Zoning 323
Other states have ruled on exclusionary zoning. The implication from Associated
Home Builders Inc. v. Supreme Court, 976) is that the
City ofLivermore (California 1
courts will judge zoning on the basis of on both insiders and outsiders. If a
its effects
zoning ordinance does not represent a reasonable accommodation of the competing
interests of insiders and outsiders, it may be declared unconstitutional. In Oregon,
state law requires municipalities to plan and zone land for a diversity of housing types
and income levels. According to Fischel (1985), few state courts have followed
the lead of the New Jersey court, so the Mount Laurel decision has not affected
exclusionary zoning practices in many states. Given the decisions from federal courts,
exclusionary zoning is still considered legal.
Just Compensation
The third criterion for the constitutionality of zoning is just compensation. The Fifth
Amendment states ". . . nor shall private property be taken for public use, without
just compensation." This isgovernment converts land from
the taking clause: if the
private to public use, the landlord must be compensated.
How is the taking clause applied to zoning? Most zoning ordinances do not
actually convert land to public use, but merely restrict private use. For example,
nuisance zoning prevents a landowner from building a factory in a residential area,
and large-lot zoning prevents a landowner from building high-density housing. By
restricting the use of private land, zoning decreases the market value of the prop-
erty. The policy issue is whether landowners should be compensated for the loss
of property value caused by zoning. For example, if large-lot zoning decreases a

landowner's property value by $5,000, should the local government pay $5,000 in
compensation?
According to Fischel (1985), the courts have provided mixed and confusing
signals to local zoning authorities. The courts routinely uphold zoning laws that
cause large losses in property values, suggesting that as long as the landowner is left
with some profitable use of his land, compensation is not required. The courts have
developed four rules to determine whether compensation is required.
1. Physical invasion. Compensation is required if the government physically

occupies the land. The invasion rule is applicable only when the
government actually occupies the land. It does not apply to most zoning
actions, in which the government merely restricts private land use.
2. Diminution of value and reasonable beneficial use. The origin of this

rule is Pennsylvania Coal v. Mahon 1922), in which Justice Holmes states,
(
". . . while property may be regulated to a certain extent,

if regulation goes
too far it will be recognized as a taking." In other words, compensation is

required if zoning decreases the property value by a sufficiently large
amount. Unfortunately, the courts have not indicated how far zoning must
go before compensation is required. A related rule is reasonable beneficial
use: if zoning leaves the landowner with options that provide a reasonable
rate of return, no compensation is required.
324 Part II Land Rent and Urban Land- Use Patients
Balancing means (benefit-cost). According to this rule, compensation is
required if the cost of a zoning ordinance (the loss in property value)

exceeds the benefit to the local community. Because this rule requires
compensation for inefficient zoning ordinances, it discourages inefficient
zoning practices. For example, suppose that a zoning ordinance decreases
property values by $50 and generates a S30 benefit to the community.
According to the balancing-means rule, compensation must be paid.
Taxpayers will oppose the ordinance because the cost of compensating
landowners exceeds the benefit generated by the ordinance. If the
balancing-means rule were used, voters would approve only efficient
zoning ordinances.
Harm prevention. According to this rule, compensation is not required if
the zoning ordinance prevents a harmful use of the land. In other words,
zoning is not a taking if it prevents the landowner from using land in ways
that are detrimental to the general public. The harm-prevention rule
suggests that a landowner, like a car owner, has limited property rights. The
car owner has the right to drive her car. but she must stop at red lights.
Should the driver be compensated for the opportunity cost of time spent
waiting for the light to turn green? Since the traffic lights prevent a harmful
use of the car. compensation is not required. Similarly, landowners have
limited property rights: if zoning prevents the landlord from building a
polluting factor^' in a residential district, compensation is unnecessary
because nuisance zoning prevents a harmful use of the land. Most zoning
ordinances are judged by a broad interpretation of the harm-prevention
rule: if an ordinance promotes public health, safety, or welfare,
compensation is usually not required.
Summary
1. Some cities use land-use regulations to control population growth.
a.An urban service boundary controls population growth by decreasing the
supply of land available for development. Land prices within the boundary
increase, and prices outside the boundary decrease.
b. A limit on building permits decreases the number of houses that can be
built, increasing housing prices and decreasing land prices.
2. A zoning plan designates a set of admissible land uses for each plot of land in
the city. There are three types of zoning: nuisance zoning, fiscal zoning, and
design zoning.
3. Nuisance zoning decreases the exposure to pollution, noise, and congestion
externalities.
a. An alternative to industrial zoning is a system of spatial effluent fees.
b. An alternative to industrial and commercial zoning is performance zoning,

under which the city develops performance standards for noise, glare, odor.
smoke, parking, and traffic.
c. High-density housing produces congestion, noise, and parking problems,

and sometimes blocks light and views. Performance zoning can be used to
shield low-density housing from these effects, allowing high-density
housing to locate near single-family homes.
d. In the absence of conventional residential nuisances (traffic, blocked views
and light) and fiscal problems, the wealthy still have the incentive to
exclude the poor. By doing so, they may increase school achievement,
decrease crime rates, and protect property values.
4. Fiscal zoning excludes households who impose a fiscal burden on the city.
Fiscal burdens may be generated by high-density housing, fringe housing, and

new commercial and industrial development.
5. Local governments use large-lot zoning to exclude high-density (low-income)
housing.
a. Zoning decreases the supply of high-density housing, excluding the poor
in two ways: there are fewer low-income dwellings and the available
dwellings are more expensive.

b. The price of land zoned for high-density housing increases, and the price of
land zoned for low-density housing decreases.
6. Under an inclusionary zoning policy, developers provide subsidized new
dwellings to low-income residents. The cost of inclusionary housing is borne
by housing consumers (higher housing prices) and landowners (lower land

prices).
7. Fringe development often imposes a fiscal burden on the city: the taxes from
new development fall short of the cost of extending public services to the
urban fringe.
a. Cities prevent fringe development by zoning vacant land for low-density
use (e.g., agriculture).
b. The alternative to fiscal zoning is a system of impact fees or development

taxes.
8. Commercial development sometimes generates fiscal burdens. If the city
cannot pass the cost of new infrastructure on to new firms and employees, it
may limit the supply of land and restrict building heights. Some cities use
impact fees to pay for new infrastructure.
9. Design zoning is a form of macroarchitecture: zoning promotes the efficient
use of infrastructure and provides open space.
a. Some cities use zoning to direct development to particular areas and use
transferable development rights to prevent horizontal inequities.
b. In the absence of compensation, open-space zoning produces an inefficient
amount of open space.
10. The market value of a development coupon is determined by (a) the power of
the coupon to override density restrictions and (b) the underlying demand for
high-density housing in the development zone.
1 1 The zoning authority of local governments comes from state governments.
There are three criteria for the constitutionality of zoning: due process, equal
protection, and just compensation.
a. Zoning satisfies the due-process criterion if it promotes public health,

safety,and welfare.
b. Federal courts have rejected the claims that fiscal (exclusionary) zoning
violates the equal-protection clause of the U.S. Constitution. In contrast,
some state courts have declared fiscal zoning unconstitutional.
c. Zoning is one of the police powers of local government, so compensation is
generally not required.

1. In Figure 1 1-3, the equilibrium price of building permits is $30,000. Suppose
that the demand curve shiftsdown and intersects the supply curve at a quantity
of 50 dwellings per year. What is the equilibrium price of permits?
2. In the analysis of the building-permit policy (Figure 1 1-3), the decrease in the
number of building permits (from 100 to 60) increased the equilibrium price of
housing but decreased the price of land. Critically appraise the following state-
ment from a landowner who recently read Chapter 7 of Urban Economics:

"According to the leftover principle, the landowner gets the leftover (the
excess of total revenue over total cost), so an increase in the price of housing
should increase — not decrease— the price of land. Either the leftover principle
is wrong or the analysis of the permit policy is wrong."
3. Consider the building-permit policy depicted in Figure 1 1-3, with an
equilibrium price of permits equal to $30,000. Suppose that the city announces
on January 1 that 300 days later (October 28) it will give the 60 permits to the
first 60 licensed building contractors through the planning office door. Because
the police department expects a line to form outside the planning office, the
police chief announces that the following queuing rules will be enforced:
i. No cuts: when a person joins the queue, he or she goes to the end of the
queue.
ii. No substitutions: no one can reserve a place in line for anyone else.
To receive a permit, a licensed contractor must be one of the first 60 people in
line and must remain in the line until October 28. Therefore, instead of an
equilibrium price for the permits, there is an equilibrium waiting time (time
spent in line). Suppose that 25 of the city's 100 licensed contractors have an
opportunity cost of $ 50 per day; 25 have an opportunity cost of $300 per day;
1
25 have an opportunity cost of $600 per day; and 25 have an opportunity cost
of $1,000 per day.
a. What is the equilibrium waiting time?
b. Suppose that the city eliminates the no-substitution rule. Would you expect
the equilibrium waiting time to increase, decrease, or not change?
4. Use Figure 1 1^4 to show the effects of the following changes on the optimum
location of the polluting factory. Do the changes cause the optimum location to
move closer to or farther from the residential area?
a. The unit commuting cost (cost per mile) increases.

b. Residents become more sensitive to pollution.
c. The polluting firm discovers a new method of pollution abatement that cuts
abatement cost in half.
5. Consider a city where a polluting industry is separated from the residential

area by a 5-mile buffer. The following table lists the wages and effluent fees
for different locations within the buffer zone: the closer the firm to the
residential district, the lower the wage and the higher the effluent fee. Suppose
that the firm uses 450 hours of labor per week and generates 20 units of
pollution per week.
Commute Distance Effluent Fee Wage

(miles) ($ per ton) <$ per hour)
5 1.0 10.0
4 2.5 9.9
3 4.5 9.8
2 7.0 9.7
1 10.0 9.6
a. Assume that the quantities of output and input are constant. What is the
optimum location for the firm?
b. What is the social cost of the segregation zoning policy? In other words, if
the city prevents the firm from locating at its optimum location, what is the
cost to society?
6. Suppose that a city eliminates all zoning. What types of retailers would you
expect to move from the retail zones to the residential zones?
7. Suppose that all local taxes are eliminated and all funding for local public
services comes from state governments. Would you expect any changes in
local zoning practices?
8. Suppose that you are a member of the zoning board in a growing city. A
residential developer appears before the board with a proposal to develop a
parcel of land at the city's edge. The proposed development would increase the
population of the city by 5 percent. The developer makes the following
statements:
i. Because the development will increase the population of the city by

5 percent, it will increase tax revenue by 5 percent.
ii. Because the development will increase the population of the city by
5 percent, it will increase the total cost of public services by 5 percent.
Hi. Given (/) and (//'). the development will not create a fiscal burden on the
and therefore should be approved.
city,
a. How would you respond to statements (/) and (//')? Are the developer's
calculations correct?
b. What additional information would you need to evaluate the fiscal effects of
the proposed development?
9. In a certain city, all new housing is built on the city'sedge (10 miles from the
city center). The city institutes the following policy: every new house will be
subject to a $1,000 impact fee or development tax. The fee is paid by the firm
that builds the house.
a. Who will actually pay the development tax?
b. Will the development tax affect the radius of the city? Explain, using a
graph.
c. Would you expect the implementation of the development tax to change the
city's zoning practices?
10. In Figure 1 1-5, large-lot zoning increases the price of apartment land and
decreases the price of land for single-family homes. Under what circumstances
(what values of the relevant elasticities) will the zoning policy increase the
total value of land in the city?
1 1. Consider a large metropolitan area in which there is a large number of
suburban municipalities, each of which has some vacant land. In each
municipality except municipality Z. high-density housing (apartments) is
explicitly outlawed. Suppose that Z starts out with no controls on residential

land use. and its vacant land is expected to be divided equally between
high-density housing (apartments) and single-family homes (SFHs). The price
of land for apartments equals the price of land for SFHs. Suppose that Z
institutes a large-lot zoning policy that limits high-density housing to
one-quarter of its available land.
a. Depict graphically the effects of Z's zoning policy on the prices of
apartment land and SFH land within the municipality.
b. Will the zoning policy increase or decrease the total value of land within
municipality Z?
12. Suppose that a city has recently proposed the rezoning of a 10-acre parcel of
land, from low-density use (1 dwelling per acre) to high-density use (20
dwellings per acre).
a. Who will support the zoning change ? Who will oppose the rezoning'.'
b. Describe a policy under which all parties will agree to the rezoning if in fact
the rezoning is efficient.
13. Ollie recently bought the Notel Hotel in downtown Portland and intends to
demolish the hotel and replace it with an office building with 100.000 square
feet of office space. When the city planner hears about Ollie's plans, he
quickly rezones the hotel site as a historical preserve, preventing the
demolition of the hotel.
a. You have been asked to estimate Ollie's loss from the rezoning of his land.
What information would you collect, and how would you use it?
b. The city offers Ollie development coupons (TDRs) that permit the
development of 100,000 square feet of office space somewhere else in the
city. What determines the market value of the coupons?
c. Under what circumstances will Ollie be worse off as a result of the zoning
and TDR policy?
14. Consider the North/South TDR example in the section on design zoning.
Suppose that instead of issuing 10 development coupons per acre (250 in
total), the government issues 5 per acre (125 total).
a. Will the new policy increase or decrease the equilibrium price of TDRs?
b. Under what conditions (what values of the relevant elasticities) will the
typical TDR recipient be better off with 5 coupons instead of 10?
15. Suppose that city U establishes an urban service boundary five miles from the
city center. People who own land outside the boundary object to it and demand
compensation. Design a practical scheme under which those who are harmed
by the service boundary will be compensated for their losses. Be specific. How
will the moneycompensation be raised? Who will actually pay for the
for
compensation package? How will you determine the amount of compensation
per landowner?
16. Using Figure 1 1-7, compute the total welfare loss caused by a switch from the
optimum policy (purchase of land for open space) to open-space zoning.
17. Suppose that a city provides open space by purchasing land from landowners,
not by zoning land for open space. If someone suggests that the city buy
another plot of land for open space, the mayor appears on local television and
solicits contributions to the city's open-space fund. The city will purchase the
land for open space if the total contributions exceed the market price of the
land. If the contributions do not cover the cost of the land, the city will not buy
the land and will return the contributions to the individual donors. Will the city
provide the optimum amount of open space?
18. Consider the following quote: "Depending on the variable one controls, a
zoning policy may either increase or decrease the price of undeveloped land
within and around the zoned city." Explain this statement, using examples of
the variables that are controlled by the various zoning policies.
19. Consider two of the compensation rules developed by the courts: the
diminution-of-value rule and the balancing-means rule. Will these rules
promote horizontal equity (the equal treatment of equals)?
20. Comment on the following statement: "The courts should not be involved in
overseeing local zoning practice. Such matters are the responsibility of the
legislative branch of government. Let democracy work."

General Readings on Land-Use Controls and Zoning
Bosselman, Fred; David Callies; and John Banta. The Taking Issue. Washington. D.C.:
Council on Environmental Quality, 1973. A history of zoning that focuses on the issue
of compensation.
Delafons, John. Land-Use Controls in the United States. Cambridge, Mass.: MIT Press,
1969. A history of zoning and other land-use controls.
Fagin. Henry. "Regulating the Timing of Urban Development." Management and Control of
Growth, vol. 1. Washington. D.C.: The Urban Land Institute. 1975. Discusses the
reasons for design zoning.
Fischel. William. The Economics of Zoning Laws. Baltimore: Johns Hopkins. 1985. A
comprehensive economic analysis of land-use zoning, including discussions of the
legal foundations of zoning and the effects of zoning on housing prices and location
patterns.
Timothy V. Ramis: Antero Rivasplata: and Steven R.
Lillydahl. Jane H.; Arthur C. Nelson:
"The Need for a Standard State Impact Fee Enabling Act." Journal of the
Schell.
American Planning Association 54 (Winter 1988). pp. 7-17. Describes impact-fee
policies in four states (California. Florida. Oregon, and Colorado) and argues for a
national standard for impact fees.
Mills. Edwin. "Economic Analysis of Urban Land-Use Controls." In Current Issues in
Urban Economics, ed. Peter Mieszkow ski and Mahlon Straszheim. Baltimore: Johns
Hopkins. 1979. A discussion of the use of zoning as a means of controlling
externalities. Includes a historical sketch of land-use controls.
Patterson. T.Land Use Planning: Techniques of Implementation. New York: Van Nostrand
Reinhold. 1977. Chapter 2 describes the basic features of zoning policies.
Siegan. Bernard. Land Use without Zoning. Lexington. Mass.: D. C. Heath. 1972. A
discussion of land use in Houston, a city without zoning.
Growth Controls
Alonso. William. "Urban Zero Population Growth." In The Urban Economy, ed. Harold
M. Hochman. New York: W. W. Norton. 1976.
Cooley. Thomas F. and C. J. LaCivita. "A Theory of Growth Controls." Journal of Urban
Economics 12 ( 1982). pp. 129—15. Explores the factors that lead communities to
control population growth, including rising costs of public sen ices and limitations on
the property tax.
White. Michelle. "Self-interest in the Suburbs: The Trend toward No-Growth Zoning."
Policy Analysis 4 ( 1978). pp. 185-204. Discusses the reasons for no-growth zoning and
suggests public policies that might reverse the no-growth trend.
Nuisance Zoning and Performance Zoning

Kendig. Lane. New Standards for Nonresidential I ses. Report 405. Chicago: American
Society of Planning Officials. 1987. Show s how performance zoning can be used to
control externalities from nonresidential land uses.
Performance Zoning. Chicago: American Planning Association. 1980. Discusses
.
performance zoning as a means of controlling residential nuisances.

Meshenberg. Michael J. The Administration of Flexible Zoning Techniques. Report 318.
Chicago: American Society of Planning Officials. 1976. Discusses various t>pes of
flexible zoning, including planned unit development, special permits. Boating /ones.
overlay zoning, contract zoning, and transfer of development rights.
Moore. Tern. "Win Allow Planners to Do What The) Do?" American Institute of Planners
Journal 44 1978). pp. 387-98. Describes the reasons for planning and zoning in terms
i
of market failure.
Porter. Douglas R.; Patrick L. Phillips: and Terry J. Lassar. Flexible Zoning: How It Works.
Washington. D.C.: Urban Land Institute. 1988. Contrasts traditional zoning with more
flexible approaches developed in recent years. Discusses the effects of flexible zoning
in seven communities.
Fiscal Zoning and Inclusionary Zoning

Brooks, Mary E. "Housing Trust Funds: Lessons from Inclusionary Zoning." In
Inclusionary Zoning Moves Downtown,ed. Dwight Merriman, David J. Brower. and
Philip D. Tegeler. Chicago: American Planning Association, 1985, pp. 7-22. Describes
housing trust funds, which use money collected from office and commercial developers
to subsidize new housing.
Davidoff, P., and L. Davidoff. "Opening the Suburbs: Toward Inclusionary' Controls."
Management and Control of Growth 1 ( 1975), pp. 540-50.
Dowall, David E. "The Effects of Tax and Expenditure Limitations on Local Land Use
Policies." Perspectives on Local Public Finance and Public Policy 1983). pp. 69-87. 1 (
Suggests that fiscal limitations (e.g.. Proposition 13) will make fiscal zoning more
prevalent. Discusses the implications of fiscal zoning for housing costs and the
imbalance between jobs and residents.
Franklin, H. M.; D. Falk; and A. Levin, In-Zoning: A Guide for Policy Makers on
Inclusionary Land Use Programs. Washington, D.C.: Potomac Institute. 1974. Part I
discusses the legal reasons for inclusionary zoning.

Mills. Edwin. S.. and Luan Sende Lubuele. "Inner Cities." Journal of Economic Literature
35 (1997), pp. 727-56. Documents some of the problems experienced by central cities
and explores the reasons for these problems.
Schwartz, Seymour I., and Robert A. Johnston. "Inclusionary Housing Programs."
American Planning Association Journal 49 (1983). pp. 3-21. Describes inclusionary
programs in several California cities.
Transferable Development Rights
Roddewig, Richard J., and Cheryl A. Inghram. Transferable Development Rights

Programs: TDRs and the Real Estate Marketplace. Report 401 Chicago: American .
Society of Planning Officials, 1987. Case studies of TDR programs in four cities (New
York City, Denver, Seattle, San Francisco) and two counties. Also explores the legal
foundations of TDRs.
Schnidman. F "TDR: A Tool for More Equitable Land Management?" Management and
Control of Growth 4 (1978), pp. 52-57.
The Administration of Zoning

Babcock, Richard R. The Zoning Game. Madison: University of Wisconsin Press. 1966.
Discusses the political and legal aspects of zoning.
Babcock. Richard R., and Charles Siemon. The Zoning Game Revisited. Boston:
Oelgeschlager, Gunn, and Hain, 1985. Accounts of 1 1 zoning conflicts.
Hinds. D.; N. Cam; and O. Ordway. Winning at Zoning. New York: McGraw-Hill. 1979.
Gives practical tips for dealing with zoning authorities.
Market Effects of Zoning

Babcock. Richard. "The Spatial Impact of Land-Use and Environmental Controls." In The
Prospective City, ed. Arthur Solomon. Cambridge. Mass.: MIT Press. 1980.
Courant, Paul N. "On Zoning on Land and Housing Values." Journal of
the Effect of Fiscal
Urban Economics 3 (1976). pp. 84-94. Explores the effect of fiscal zoning
on total
land value, reaching conclusions different from Ohls. Weisberg. and White.
Dowall, David E. "Reducing the Cost Effects of Local Land Use Controls." Journal of the
American Planning Association 47 (1981). pp. 145-53. Discusses the effects of
land-use controls on the prices of housing and land. Proposes a monitoring system to
control the inflationary effects of land-use controls.
. "The Suburban Squeeze: Land-Use Policies in the San Francisco Bay Area." Cato
Journal 2 ( 1982). pp. 709-33. Discusses local land-use policies and their effects on the
regional economy.
Grieson, Ronald E., and James R. White. "The Effects of Zoning on Structures and Land
Markets." Journal of Urban Economics 10 ( 198 1 ). pp. 27 1-85. Explores the effects of
three types of zoning (restrictions on density, lot size, and allowable uses) on the prices
of housing and land.
Lustig. Morton, and Janet Rothenberg Pack. "A Standard for Residential Zoning Based
Upon the Location of Jobs." Journal of the American Institute of Planners 40 ( 974). 1
pp. 333-45. Proposes that residential zoning be used to match labor supply (residential
land use) with labor demand (commercial and industrial land use).
Ohls. James C; Richard C. Weisberg; and Michelle J. White. "The Effect of Zoning on
Land Value." Journal of Urban Economics 1 (1974). pp. 428-44. Discusses the effect
of zoning on (1) the value of land zoned for different uses and (2) total land
fiscal
value. Also discusses the trade-offs associated with nuisance zoning.

.
"Welfare Effects of Alternative Models of Zoning." Journal of Urban Economics 3
(1976), pp. 95-96. Discusses the reasons for the different conclusions of Courant.
White. Michelle J. "The Effect of Zoning on the Size of Metropolitan Areas." Journal of
Urban Economics 2 (1975), pp. 279-90. Explores the effect of large-lot zoning on the
land area and total land value of a metropolitan area, showing that the effects of zoning
on both variables are ambiguous.
Partm
V t RT Y ad HOUSING
(~J7~
his section of the book discusses two persistent problems in urban areas:
<_x poverty and inadequate housing. The federal government uses a number
of policies to combat poverty, including cash assistance, education and training
programs, and payments in kind (food stamps, medical care, housing assistance).
The problem of inadequate housing is related to the poverty problem in the sense that
an increase in income would allow an ill-housed family to move to a better dwelling.
Most housing policies, like antipoverty policies, are redistributional in nature.
The federal government is responsible for income redistribution because local
redistributional efforts will be weakened by the mobility of taxpayers and transfer
recipients. Suppose that a city imposes a tax on its wealthy citizens to finance transfer
payments to the poor. Some wealthy households would leave the city to escape the
tax, and some poor households would enter the city to get the transfer payment. In
combination, the flight of the wealthy and the migration of the poor would weaken
the city's redistribution program: there would be less money to transfer to more poor
households. A national redistribution program will be more successful because there
is less mobility between nations than between cities.
Under the welfare-reform program enacted in 1996, the federal government

will provide block grants to states, and states will be responsible for designing and
implementing their own welfare plans. Although the federal government has imposed
some restrictions on how the federal funds are to be spent, the states are assuming a
much bigger role in redistributional policies.
There are four chapters Chapter 12 discusses the spatial reasons
in this section.
for poverty, exploring the connection between racial segregation in housing and
poverty in central cities. Chapter 13 explores the nonspatial causes of poverty and
discusses various antipoverty policies —
including the policies implemented under
the new welfare-reform The chapter also discusses the problem of homeless-
plan.
ness, focusing on what causes people to become homeless. Chapter 14 explains why
housing is different from other goods. Chapter 15 takes a detailed look at several
types of housing policies, including public housing, housing vouchers, and commu-
nity development programs.
Chapter 12
j- CGREGATION and POVERTY

in the CENTRAL (ITY
A local citizen said to me:

You '11 lower the price of property
Move along, get along, move along, get along
Go, Move, Shift
Christy Moore
rf lthough there are poor people in every segment of society, the incidence of
^/w poverty is greatest among women, children, racial minorities, and central-
city residents. This chapter explores the problem of persistent poverty in central
cities and examines the effects of racial segregation in housing on urban poverty.
There are two big questions. First, why are poverty rates so much higher in central
cities? Second, what sort of public policies would be effective in reducing poverty
in the central city?
Poverty Facts
The U.S. government defines a poor household as one with total income less than the
amount required to satisfy the "minimum needs" of the household. The government
estimates a minimum food budget for each type of household, and multiplies the
food budget by three to get the official poverty budget. A household with income
lower than the official poverty budget is considered poor. In 1997, the poverty budget
was $8,183 for a single person, $12,802 for a three-person household, and $16,400
for a four-person household.
Who and Where Are the Poor?

Table 12-1 shows the characteristics of the poor population in 1997. For each type
of person, the poverty rate equals the percentage of people of that type who live in
335
336 Part III Poverty and Housing
Table 12-1 The Poverty Population in 1997
Persons in Poverty Rate

Poverty (1,000) (percent)
All persons 35.574 13.3
Race and Hispanic Origin

White 24.396 11.0
Black 9.116 26.5
Asian and Pacific islander 1.468 14.0
Hispanic origin 8.308 27.1
Age
Over 65 3.376 10.5
Under 18 14.143 19.9
White 8.994 16.1
Black 4.229 37.2
Hispanic origin 3.975 36.8
Residence
Outside metropolitan areas 8.301 15.9
Inside metropolitan areas 27.273 12.6
Inside central cities 15.018 18.8
Outside central cities 12.255 9.0
Type of Family
Married couple 2.821 5.2
Female-headed 3.995 31.6
White 2.305 27.7
Black 1 .563 39.8
Hispanic origin 836 47.6
Education Level
No high school diploma 2.776 24.1
High school graduate 2.113 9.8
Some college 1.222 7.2
College graduate 403 2.4
Soi kci U.S. Bureau of the Census, Poverty in ike United States, 1997
:
(Washington, D.C.: U.S. Government Priming Office, 1998).
households with incomes lower than the official poverty budget. Several facts are
shown in the table.
1 . Race. Although the number of poor whites is more than twice the number
of poor blacks and about four times the number of poor Hispanics. the
poverty rates for blacks and Hispanics are much higher than the white
poverty rate. For Hispanics. poverty rates vary between groups with
different cultural backgrounds: poverty rates are highest among those of
Puerto Rican and Mexican heritage (36 percent and 3 1 percent.
( 'hapter 12 Segregation and Poverty in the Central City 337
respectively) and lowest among those of Cuban heritage (18 percent). The
poverty rate among Asians and Pacific Islanders is about 14 percent, close
to the national average.
2. The aged. One of the successes in the war on poverty was the decrease in
poverty among the aged: their poverty rate dropped from 35 percent in
1959 to 10.5 percent in 1997, largely as a result of increased Social
Security benefits.
3. Children. About one-fifth of children lived in poverty in 1997, with much
higher poverty rates for racial minorities.
4. Residence. The poverty rate outside metropolitan areas is about one-third
higher than the metropolitan rate. The poverty rate inside central cities is
more than twice the rate outside central cities.
5. Type of family. The poverty rate for female-headed households is about

six times the rate for households headed by a married couple. The poverty
rate for female-headed households is higher because (a) a woman in a
single-parent household must juggle employment and child care responsi-
bilities, (b) on average, women earn lower wages than men, and (c) many
households do not receive child support from absent fathers.

6. Education. Poverty rates are negatively related to the level of education.
The poverty rate of high school dropouts is more than twice the rate for high
school graduates and more than ten times the rate for college graduates.
The Working Poor

Figure 12-1 shows the distribution of work effort by poor household heads. More
than half of poor households had someone who worked at least part-time, and one
had a full-time, year-round worker. A four-person household with a full-time
in five
worker is poor if the worker's hourly wage is less than $8.20 per hour. In other
Figure 12—1 Work Effort of Poor Household Heads

Worked full-time,
year-round
19%
Worked part-time
39%
SOURCE: U.S. Bureau of the Census, Poverty in the United States. 1997 (Washington, D.C.: U.S.
Government Printing Office, 1998).
Figi're 12-2 Poverty Rates in Central Cities and Suburbs

35.0
31.1
30.0
25.0
g 20.0
P 15.0
12.0
10.0
5.0
Central city Suburb Central citj Suburb
White Black
Source: Mills, Ed« in S.. and Luan Sende Lubuele. "Inner Cities." Journal of Economic literature 35 i
1997 1,
pp. 727-56.
words, a full-time job may not he enough to escape poverty. Either the wage must
be quite a bit higher than the minimum wage more than
or the household must have
one worker. In fact, many households whose primary worker earns less than $8.20
per hour escape poverty because there is a second worker in the household.
Poverty in the Central City
As shown in Table 12—1, the poverty rate for central cities is about twice as high as
the poverty rates for suburban areas. In some metropolitan areas, the differences in
poverty rates are even larger. For example, in Chicago and Philadelphia, the central-
city povertj rates are about four times the suburban rates. Figure 12-2 shows the
poverty rates in central cities for both white and black residents. For both groups.
povert) is more prevalent in the central city.
Residential Segregation
An important feature of cities in the United States is residential segregation with
respect to race. More than two-thirds of the blacks who live in metropolitan areas
reside in central cities, while only one-third of metropolitan whiles live in central
Chapter 12 Segregation and Poverty in the Central City 339
cities. This means that only one-third of metropolitan blacks live in the suburbs,
compared to two-thirds of whites. As we'll see, the concentration of blacks in central

cities, far from suburban job opportunities, is one reason for high poverty rates
in central cities.
Segregation Facts
One way to quantify the degree of racial segregation is the index of dissimilarity.
This index shows the proportion of one racial group that must relocate to achieve a
completely integrated spatial configuration, one in which each census tract replicates
the racial composition of the metropolitan area as a whole. In 1990, the average
dissimilarity index for U.S. metropolitan areas was 69 percent, indicating that to
achieve complete integration, 69 percent of blacks (or whites) would need to relocate
(U.S. Census Bureau, 1998). As shown in Figure 12-3, the degree of segregation
varies considerably among metropolitan areas, with greater segregation in larger
metropolitan areas. The figure also shows the degree of segregation for the nation's
Figure 12—3 Dissimilarity Indexes for Different MSAs

100
90 86 85
83 8^
80 77
73 74
69
X 70 64
auc
60 57
50
g 40
5 30
20
10
SOURCE: U.S. Bureau of the Census, Residential Segregation, Summary Tables and Residential Segregation,
Detailed Tables (Washington. DC: U.S. Government Printing Office, 1998).
most segregated cities, most of which are in the northern central states. In the last
several decades, the degree of segregation did not change very much.
Reasons for Racial Segregation

What causes residential segregation? There are several reasons for segregation, and
it is difficult to disentangle their separate effects.
1 Racial attitudes and tension. Whites and blacks have different desires
concerning the racial composition of neighborhoods (Kain. 1985. Clark.
1991 ). While a majority of blacks would prefer to live in integrated neigh-
borhoods, a majority of whites prefer segregated neighborhoods. Whites
and blacks also what they consider an integrated neighborhood: for
differ in
the typical black household, an "integrated" neighborhood is one that is
equally divided between whites and blacks: for the small number of whites
who prefer integrated neighborhoods, an "integrated" neighborhood is one
in which 80 percent of the residents are white and only 20 percent are
black. Racial tensions in mixed suburban communities may deter all but the
thick-skinned black households from moving there (Mills and Lubuele.
1997).
2. Income differences. On average, black households have lower income than

white households, and housing is less expensive in the inner city (Mills and
Lubuele. 1997). Although some black households choose low-cost housing
in the central city instead of more expensive housing in the suburbs, this
phenomenon does not fully explain the degree of segregation we observe.
A study by Gabriel and Rosenthal (1989) suggests that a black household
with the same income and other characteristics as the average white sub-
urban household is much less likely to live in the suburbs. Kain ( 1985)
posed the following question: if location decisions were based strictly on
nonracial characteristics (income, family type, age of household head), how
many black households would live in the suburbs? His answer was that
3.6 million black households would have lived in the suburbs, about twice
the actual number of black suburban households at that time.
3. Racial discrimination in real estate markets. Under a fair-housing audit
program, two individuals who are identical except for race (same income,
education, household size) go to real estate brokers to inquire about
housing. Yinger (1998) shows that these audits give conclusive evidence
that blacks and Hispanics are treated differently than otherwise identical
whites: they are shown fewer dwellings, steered into certain neighborhoods,
and given less advice and assistance. For example, among black renters,
one in ten was excluded altogether from housing made available to white
renters, and one in four learned about fewer vacant dwellings.
4. Exclusionary zoning in suburban communities. As explained in

Chapter 1. suburban governments use zoning to exclude low -income
1
households from their jurisdictions. Among the exclusionary practices are

minimum-square-footage requirements for lots and dwellings, prohibition
of multifamily units, maximum densities, requirements for two-car garages,
and development fees. Exclusionary zoning is motivated in part by fiscal
concerns: if low-income households do not pay their "fair share" of the
costs of local public services, suburban governments have an incentive to

exclude them. Since black households have lower income, on average, than
whites, exclusionary zoning has a larger effect on black households.
5. Housing policies. Historically, federal housing policies have encouraged

the poor to live in poor central-city areas. Most public housing was built in
poor areas, and until recently, housing vouchers (coupons used to help pay
where the recipient lived
for private housing) could be used only in the city
when the voucher was government made most
issued. Recently, the federal
housing vouchers portable, so more recipients can use them to rent
suburban housing.
Consequences of Segregation
Are black households better or worse off in cities that are more racially segregated
than other cities? A recent study (Cutler and Glaeser, 1997) provides evidence that
young black adults (people in their twenties) are much worse more segregated
off in
cities: they earn less income, have lower high school graduation rates, are more likely
to be idle (neither in school nor working), and more likely to become single mothers.
Figure \2-<\ summarizes the relationships between segregation and various ad-
verse outcomes for black households.The chart shows the elasticity of each adverse
outcome with respect to the degree of segregation, where the elasticity is defined
as the percentage change in the outcome divided by the percentage change in the
degree of segregation. For example, the elasticity for high school dropout status is
0.51, meaning that a 10 percent increase in the degree of segregation increases the
probability of being ahigh school dropout by 5. 1 percent. For the cities in the sample,
the average probability of dropping out of high school is 23.4 percent. An elasticity
of 0.51 means that the probability of dropping out is about 25.8 percent in a city
that is 20 percent more segregated than the average city. As shown in Table 12-1,
the incidence of poverty among high school dropouts is very high, meaning that
segregation increases poverty.
The other elasticities shown in Figure 12^ concern idleness and single parent-
hood. The elasticity for idleness is 0.88, indicating that each 10 percent increase in
segregation increases the probability of being idle by 8.8 percent. The elasticity for
single motherhood is 0.29, meaning that the probability of being a single mother in-
creases by 2.9 percent for every 10 percent increase in segregation. These elasticities
suggest that segregation has powerful effects on the economic prospects of black
households, leading to lower employment rates and higher rates of single parenthood.
Why does segregation generate these adverse outcomes? One reason is that
young people living in a highly segregated, low-income environment have fewer
342 Pari III Poverty and Housing
Figure 12—4 Effects of Segregation
1.00
-
- 0.88
| 0.90
5 0.80 -
u
•g 0.70
o
6 0.60 -
u
0.49
£ 0.50 -
£ 0.40 -
S 0.30 - 0.29
~ 0.20
Dropping out of Idlenes Single

hich school motherhood
SOURCE: Authors calculations based on results in Cutler. David M.. and Edward L. Glaeser. "Are Ghettos
Good or Bad.''" Quarterly Journal oj Economics I 1997). pp. 827-72.
contacts with positive role models —educated and successful people —and are less
likely to be successful themselves (Cutler and Glaeser. 1997). The lack of contact
with positive role models increases the frequency of adverse outcomes: the less
frequent the contact with educated adults, the higher the high school dropout rate
and the incidence of single motherhood, and the lower the share of black youths
attending and completing college.
In the next two parts of the chapter, we'll explore two other connections between
segregation and poverty in the central city.
• Low educational achievement. Educational achievement is lower in
central-city schools, and low achievement increases the likelihood of being

poor.
• Spatial mismatch. Confining black households to the central city, far from
suburban employment opportunities, results in poor access to job markets
and thus lower employment rates and wages.
Segregation, Education, and Poverty

Recall from Table 12-1 that one cause of poverty is low educational achievement.
The poverty rate for high school dropouts is about 2.5 times the poverty rate for
high school graduates and 10 times the poverty rate for college graduates. A study
of black inner-city youths from the National Bureau of Economic Research showed
that increases in the quantity of education (time spent in school) and the quality of
education (measured by grades) increased both wages and hours worked (Freeman
and Holzer, 1986). If the residents of central cities have lower educational achieve-
ment than their suburban counterparts, that would explain part of the difference in
poverty rates between central cities and suburbs.
The facts on central-city education are grim. Students in central-city school
districts are twice as likely as suburban students to drop out of high school. In terms
of the quality of education, the high school education received by black central-
city students is roughly equivalent to the education received by a white high school
dropout (Schiller, 1995). In high-poverty urban school districts (defined as districts
where at least 40 percent of the residents are below the poverty line), scores on
standardized tests are very low. At grade four, only 3 percent of students in high-
poverty urban districts are proficient in math, and only 5 percent are proficient in
reading (Institute on Race and Poverty, 1998). The performance at higher grades is
as follows: for grade eight, 6 percent proficiency in math and 9 percent proficiency
in reading; for grade twelve, 6 percent proficiency in math and 9 percent proficiency
in reading.
Why is educational achievement lower in central-city schools? One reason is
poverty among the families served by the schools (Orfield, 1998):
High-poverty schools have to devote far more time and resources to family and health crises,
security, children who come to school not speaking standard English, seriously disturbed
children, children with no educational material in their homes, and many children with weak
educational preparation.
There is some circularity concerning the problems of poverty and education: if the
problems associated with poverty spill over into the classroom, it is more difficult to
educate children, which makes it less likely that they will acquire the skills needed
to escape poverty themselves.
Somenatural experiments have demonstrated the differences between central-
city and suburban schools. The Gautreaux program in Chicago gives residents of
public housing in the central city the opportunity to move to affordable housing in
the suburbs. Among the children who moved to the suburbs, the high school dropout
rate was one-quarter of the rate for their central-city counterparts, and the college-
attendance rate was twice that of the central-city students (Orfield, 1998). In Norfolk
Virginia, black students in integrated schools had much higher test scores than black
students in more segregated schools (Meldrum and Eaton, 1994).
The Spatial Mismatch

How does racial segregation affect the job opportunities and earnings of urban
blacks? As discussed in Chapter 10, about half of the jobs in manufacturing,
retailing, and offices are now in suburban areas. Because only about a third of
black workers live in the suburbs, they have limited access to suburban employment
opportunities. According to the spatial mismatch theory, the segregation of blacks
Figure 12-5 Changes in Employment by Neighborhood Racial

Composition
30% r
Black > 209! >

Black 40<7r
70',
Total Manufacturing, transportation.

communication, utilities.
Manufacturing construction, public administration
Source: Authors calculations based on Rafael. Steven. "The spatial Mismatch Hypothesis and Black Youth
Joblessness: Evidence from the San Francisco Baj Area." Journal q) (rh<m £< onomics 43 1998), pp. 79-1
( 1 1.
from suburban employment opportunities, contributes to black

in the central city, far
poverty. The firstin support of this theory is from Kain (1968). who es-
evidence
timated that racial segregation in Chicago decreased employment opportunities for
blacks by more than 20,000 jobs.
A recent study suggests that employment growth is occurring in areas that are not
very accessible to black workers (Rafael, 1998). Figure 12-5 shows the percentage
changes in employment during the 1980s in the San Francisco-Oakland-San Jose
Consolidated Metropolitan Statistical Area (CMSA). For the area as a whole, total
employment increased by 21 percent. In areas near mostly white neighborhoods
(black population <20 percent), total employment increased by 23 percent, while
in areas near neighborhoods where the population is at least 20 percent black, total
employment increased by only 2 percent. Manufacturing employment increased

by percent near mostly white neighborhoods, but decreased by 32 percent near
1
neighborhoods that were at least 20 percent black. Over half of low-skilled black
workers are employed in manufacturing, transportation, communication, utilities,
construction, or public administration. Employment in these sectors increased by
4 percent in the CMSA as a whole and by 9 percent near mostly white neighborhoods,
but decreased by 14 percent near neighborhoods that were at least 20 percent black.
Residential segregation means that blacks commute relatively long distances.
For the working poor, the average one-way commute in 1990 was 26 minutes for
white workers and 30 minutes for black workers (O' Regan and Quigley. 998). After 1
controlling for various factors such as earnings, housing prices, and neighborhood
amenities. Gabriel and Rosenthal (1998) conclude that black high school graduates
commute roughly 22 percent longer than comparably skilled white workers. Longer
commuting distances decrease the net wage of black workers and reduce the time
available for work and other activities.
Segregation — and the resulting longer commuting distances — also decrease the
employment rates of black workers. Using data from the Philadelphia metropolitan
area, Ihlanfeldt and Sjoquist (1990) show that higher commuting costs discourage
work all youths, black and white, the likelihood of being employed drops
effort. For
as the commuting distance increases. One reason for the lower employment rate for
black youths in Philadelphia (27 percent versus 49 percent for white youths) is that
the average commute time for blacks was 26 minutes, compared to only 19 minutes
for whites. In fact, the differences in commute distances explain between a third and
a half of the differences in employment rates.
Although black central-city residents can commute to suburban jobs and many —
—
do this sort of reverse commuting is costly and time-consuming for two reasons.
First, most transit systems are designed to deliver suburban residents to the city
center, and are ill-suited for bringing central-city residents to suburban jobs. Second,
a relatively large number of poor households don't own cars: 45 percent of black
workers living in central cities don't have access to a car (O'Regan and Quigley,
1998). Having a car cuts travel commuting time significantly: for the typical black
worker who commutes to the suburbs by public
central-city transit, switching to an
automobile would save about 19 minutes each day.
The lack of access to a car also affects the likelihood of employment (O'Regan
and Quigley, 1998). A survey of low-skilled workers in Detroit found that those with
cars searched for jobs over a much wider area and discovered more job opportuni-
ties. A study of a job-training program for the fathers of welfare-recipient children
concluded that auto ownership was an "important prerequisite" to participation in
the program, completion of job training, and getting a job. There are some important
policy implications for welfare policy, in particular the recent efforts to move people
from welfare to employment (O'Regan and Quigley, 1998):
If potential commute patterns of people coming off public assistance are similar to those of
people currently in poor working households, government policy must pay more attention
to auto ownership opportunities. ... So programs that help job takers obtain a used car —
secured loan for purchase, a leasing scheme, a revolving credit arrangement —may offer real
promise, particularly in less dense and less centralized urban areas.
What do we know about the relative importance of the spatial mismatch?

Ihlanfeldt and Sjoquist (1991) tested the mismatch hypothesis in 50 metropolitan
areas. Their principal conclusions are as follows:
1 employment opportunities explains between

Overall, inferior access to
24 percent and 27 percent of the gap between black and white employment
rates and between 29 percent and 34 percent of the gap between Hispanic
and white employment rates.
2. The spatial mismatch is more important in larger metropolitan areas. In
small cities, only 3 percent of the gap between black and white employment
rates is explained by inferior access; in medium-sized cities, inferior access

is responsible for 14 percent of the gap: in large cities, inferior access is
responsible for about a quarter of the gap.
This and other studies suggest that the spatial mismatch is one of several factors that
lead to higher poverty rates in the central city.
Poverty in the Central City and Public Policy

The problem of persistent poverty in central cities has many causes. An important fac-
tor is racial segregation, which decreases workers" access to suburban employment
opportunities, decreases poor youths' contact with positive role models, decreases
educational achievement, and increases the incidence of single parenthood. The ob-
vious policy response is to eliminate the barriers that prevent mam poor central-city
residents — especially racial minorities —from moving to suburban areas.
• Racial discrimination in the real estate market. This barrier could be

lowered by more rigorous enforcement of the nation's fair-housing laws.
• Exclusionary zoning. This barrier could be lowered by encouraging local
governments to eliminate practices that discriminate against housing for
low-income households.
• Public housing policies oriented toward the central city. This barrier
could be eliminated by making all housing assistance portable.
The most recent wave of welfare reform, which will force most welfare recipients
into private employment, would increase the payoff from eliminating these barriers.
There is an old debate among urban policymakers concerning the wisdom of
encouraging the suburbanization of central-city black households. One concern is
that only the most educated and skilled blacks will move to the suburbs, and their
departure will worsen conditions for those who stay. To address this concern, policies
should be designed to give all central-city residents the option of moving closer
to suburban employment centers. The Cutler and Glaeser stud) o\ the effects of
segregation (1997) suggests that the poorest and least educated blacks are harmed
the most b\ segregation, and thus have the most to gain from moving into a less
segregated environment.
One alternative to a dispersal policy is a development policy in which firms are
encouraged to locate in the central city. Under the black-capitalism approach, the
government encourages the development of black businesses. The experiences with
black capitalism have not been encouraging (Bates and Bradford. 1979). Another
approach is enterprise zones, areas of the city where firms paj lower taxes.
( 1 )
(2) receive subsidies for worker training, and (3) are exempted from man) local
regulations. The limited experience w ith enterprise zones suggests that they are not
very effective in luring firms to central cities (Boarnet and Bogart. 1996: Papke.
1903 and 1994: Dow all. 1996).
There are two basic problems with the development approach. First, for many
types of activities the central city is not a profitable location. The suburbanization of
jobs was caused by fundamental changes in the economy, for example, the develop-
ment of the automobile and the truck, changes in production technology, innovations
in information technology. Given the advantages of a suburban location, a firm in
the central city may be unable to compete with its suburban counterpart. Second,
poor households in central cities may not have the skills required for jobs available
in the central city. As a result, policies that attempt to foster development of central
cities are unlikely to reduce poverty.
What is the role of public policy in reducing poverty in central cities? From the
perspective of an urban economist, the first step is to eliminate the barriers to sub-
urbanization. As explained in this chapter, there is ample evidence that segregation
adversely affects the economic prospects of black households. In the next chapter,
we'll look at some other, nonspatial reasons for poverty and explore other policies
that could decrease poverty in the central city as well as the suburbs.
Summary
1 Poverty rates are highest among racial minorities, children, female-headed
households, high school dropouts, and central-city residents.
2. Residential segregation is prevalent in U.S. cities, with an average dissimilarity
index of 69 percent, indicating that to achieve complete integration, 69 percent
of blacks would have to relocate.
3. Segregation is caused by racial attitudes, discrimination in real estate markets,
exclusionary zoning, and housing policies oriented toward central cities.
4. Black households are much worse off in more segregated cities: blacks have
lower high school graduation rates, are more likely to be idle, earn less income,
and are more likely to become single mothers.
5. Educational achievement (measured as high school completion rates and scores
on standard tests) is lower in central cities.
6. There is a spatial mismatch between jobs and residences: a disproportionate

number of black workers live in the central city, far from suburban employment
opportunities.

1. Consider a city with 10,000 jobs in the central city and 20,000 jobs in a
suburban jurisdiction. Suppose that black workers are confined to the central
city, but whites can live in either jurisdiction. Suppose further that travel costs
are so high that workers live and work in the same jurisdiction.
348 Part 111 Pen fin and Housing
a. Under what circumstances will racial segregation not affect the wages of
white and black workers?
b. Suppose that the circumstances you identify in (a) occur in 1950. Would you
still expect them to occur in 2000?
Comment on the following statement:
"Black central-city residents are poor because they do not have access to
suburban employment opportunities. I propose that we supply each central-city
household with a car. Such a plan will reduce poverty at a relatively low cost."
Consider a segregated city in which all blacks live in the central city. In the
suburb (eight miles from the central city), manufacturing firms in industry X
employ black workers.
a. Use the numbers shown below to compute the black suburban wage.
i. The unit cost of commuting (including monetary and time costs) is 40^
per mile.
it The wage for blacks in the central city is $ 10 per hour, and blacks work 8
hours per day.
b. Use the numbers shown below to predict the effects of racial segregation on
black suburban employment. In other words, predict the percentage increase
in total black employment that would result from a decrease in the black
commute distance from eight miles to zero.
hi. The price elasticity of demand for X is —2.0.
iv. Labor is responsible for 80 percent of production costs,
v. Firms produce X with fixed factor proportions.
Poverty Facts and General References

Ellwood. David T. Poor Support. New York: Basic Books. 1988. A detailed description and
analysis of poverty among two-parent families, single-parent families, and residents of
inner-city neighborhoods.
O'Hare, William P.: Kelvin M. Pollard: Taynia L. Mann: and Mary M. Kent. "African
Americans in the 1990s." Population Bulletin 46 (July 1991 ).
Schiller. Bradley. The Economics of Poverty and Discrimination. 6th ed. Englewood Cliffs,
N.J.: Prentice-Hall. 1995.
U.S. Bureau of the Census. Current Population Reports. Series P60-201, Poverty in the
United States, 1997. Washington, D.C: U.S. Government Printing Office, 1998.
U.S. Bureau of the Census. Statistical Abstract of the United States, 1997. Washington,
D.C: U.S. Government Printing Office, 1997.
Poverty in the Central City
Bates. Timothy, and William Bradford. "Social Problems and the Urban Crisis: Can Public
Policy Make a Difference?" In The Urban Economy, ed. Harold M. Hochman. New
York: W.W. Norton, 1976.
Cutler.David M.. and Edward L. Glaeser. "Are Ghettos Good or Bad?" Quarterly Journal
of Economics (1997). pp. 827-72.
Freeman. Richard B.. and Harry J. Holzer. "The Black Youth Employment Crisis: Summary
of Findings." In The Black Youth Employment Crisis, ed. Richard B. Freeman and
Harry J. Holzer. Chicago: University of Chicago Press, 1986.
Instituteon Race and Poverty. Examining the Relationship between Housing, Education,
and Persistent Segregation: Final Report. Minneapolis: Institute on Race and Poverty.
1998.
Meldrum, Christina, and Susan Eaton. "Resegregation in Norfolk, Virginia: Does Restoring
Neighborhood Schools Work?" Report of the Harvard Project on School
Desegregation (1994).
Mills, Edwin S., and Luan Sende Lubuele. "Inner Cities." Journal of Economic Literature
35 (1997), pp. 727-56. Interprets and generalizes studies and data related to economic
and social issues in inner cities of U.S. metropolitan areas.
Ricketts, Erol R.,and Isabel V. Sawhill. "Defining and Measuring the Underclass." Journal
of Policy Analysis and Management 1 1988), pp. 316-25.
(
Wilson. William Julius. "The Urban Underclass in Advanced Industrial Society." In The
New Urban Reality, ed. Paul Peterson. Washington. D.C.: Brookings Institution, 1985.
Segregation
Clark. William A. "Residential Preferences and Neighborhood Racial Segregation: A Test

of the Schelling Segregation Model." Demography 28 (1991). pp. 1-19. Describes the
differences in preferences for integrated neighborhoods between whites and blacks.
Kain, John. "Black Suburbanization in the Eighties: A New Beginning or a False Hope?" In
American Domestic Priorities, ed. John Quigley and Daniel Rubinfeld. Berkeley and
Los Angeles: University of California, 1985. Describes trends in the suburbanization
of blacks.
Orfield, Gary. "Desegregation, Resegregation, and Education." Part I of In Pursuit of a
Dream Deferred: Linking Housing and Education. Minneapolis: Institute on Race and
Poverty, 1995.
Yinger, John. "Evidence on Discrimination in Consumer Markets." Journal of Economic
Perspectives 12, no. 2 (Spring 1988), pp. 23^0.
Spatial Mismatch
Ellwood, David "The Spatial Mismatch Hypothesis: Are There Teenage Jobs Missing in
T.
the Ghetto?" In The Black Youth Employment Crisis, ed. Richard B. Freeman and
Harry J. Holzer. Chicago: University of Chicago Press, 1986, pp. 147-85. Provides
evidence from Chicago that questions the validity of the spatial mismatch hypothesis.
Gabriel. Stuart A., and Stuart S. Rosenthal. "Household Location and Race: Estimates of a
Multinomial Logit Model." Review of Economics and Statistics, 1989, pp. 240-49.
Shows that the location choices of black households are relatively insensitive to
changes in income and other socioeconomic characteristics.
Job Accessibility and Youth Employment Rates." Policy
Ihlanfeldt, Keith R. "Intra-Urban
Research Center Paper No. 18. Atlanta: Georgia State University, 1991. Provides
evidence from 50 metropolitan areas that relatively low employment rates for black and
in part, by inferior access to employment opportunities.
Hispanic youths are caused,
and David L. Sjoquist. "Job Accessibility and Racial Differences in
Ihlanfeldt, Keith R..
Youth Employment Rates." American Economic Review 8 (1990), pp. 267'-76.
Provides evidence from Philadelphia that inferior access is responsible for between a
third and a half of the gap between black and white employment rates.
Kain. John "Housing Segregation. Negro Employment, and Metropolitan

F.
Decentralization." Quarterly Journal of Economics 82 (1968). pp. 175-97.

O'Regan, Katherine M. and John M. Quigley. "Cars for the Poor." Access 12 (Spring
1998). pp. 20-5.
Rafael. Steven. "The Spatial Mismatch Hypothesis and Black Youth Joblessness: Evidence
from the San Francisco Bay Area." Journal of Urban Economics 43 (1998).
pp. 79-111.
Development versus Dispersal and Enterprise Zones

Banfield. Edward C. "Why Government Cannot Solve the Urban Problem." In The Urban
Economy, ed. Harold M. Hochman. New York: Norton. 1976. pp. 257-72.
Bateman, Worth, and Harold M. Hochman. "Social Problems and the Urban Crisis: Can
Public Policy Make a Difference?" In The Urban Economy, ed. Harold M. Hochman.
New York: Norton. 1976. pp. 283-93.
Boarnet. Marlon, and William T. Bogart. "Enterprise Zones and Employment: Evidence
from New Jersey." Journal of Urban Economics 40 1996). pp. 198-215. (
Brimmer. Andrew F. and Henry S. Terrell. "The Economic Potential of Black Capitalism."
In The Urban Economy, ed. Harold M. Hochman. New York: Norton. 1976.
pp. 239-56. Suggests that the black-capitalism approach is unlikely to substantially

improve the economic position of blacks.
Dowall. David. "An Evaluation of California's Enterprise Zone Programs." Economic
Development Quarterly 10 1996). pp. 352-68.
(
Edel. Matthew. "Development versus Dispersal: Approaches to Ghetto Poverty." In

Readings in Urban Economics, ed. Matthew Edel and Jerome Rothenberg. New York:
Macmillan. 1972. pp. 307-25.
Harrison. Bennett. "Ghetto Economic Development." Journal of Economic Literature 1974.
pp. 1-37.
Kain, John F and Joseph J. Persky. "Alternatives to the Guilded Ghetto." Public Interest
(Winter 1969). pp. 74-87. reprinted in The Urban Economy ed. Harold M. Hochman.
New York: Norton, 1976, pp. 21 1-25.
Orfield. Gary. "Ghettoization and Its Alternatives" In The New Urban Reality, ed. Paul
Peterson. Washington. D.C.: Brookings Institution. 1985.

Papke. Leslie. "Tax Policy and Urban Development: Evidence from the Indiana Enterprise
Zone Program." Journal of Public Economics 54 1994), pp. 37— c).
(
Papke. Leslie. "What Do We Know About Enterprise Zones?" In Tax Policy and the
Economy, ed. James Poterba. Cambridge. Mass.: MIT Press. 1993.
Chapter 13
& THtR DIMENSIONS
of POVERTY
Now, too much of nothing

Can make a man ill at ease.
One man 's temper might rise
While another man 's temper might freeze.
Bob Dylan
he previous chapter discussed the spatial aspect of urban poverty, exploring

^~J7~
c/ the connection between residential segregation and central-city poverty. In
thissecond chapter on poverty, we explore some nonspatial causes of poverty, in-
cluding slow economic growth or economic decline, racial discrimination in labor
markets, and increases in the number of female-headed households. We'll also exam-
ine some and discuss the features of welfare reform
traditional antipoverty policies
adopted in 1996. The main thrust of the reform is to change the welfare system into
one that requires work in exchange for time-limited assistance. The last part of the
chapter explores the problem of homelessness, focusing on what causes people to
become homeless.
Economic Growth and Poverty

One reason for povertyis slow economic growth or economic decline. According
to the queuing theory of unemployment, prospective workers form an employment

line (a queue) outside the factory or office building. A worker's position in the
line is determined by his productivity: the lower his productivity, the further down
the line he is positioned. Because employers draw workers from the front of the
line, the least productive workers are hired when economic growth increases the
demand for labor, causing down the employment queue.
firms to hire workers further
351
Because economic growth generates jobs for marginal (low-productivity) workers,

poor people experience relatively large benefits from growth and relatively large
losses from decline.
Evidence Concerning the Benefits of Growth

The results of Bartik*s ( 1991 ) study of urban employment growth are consistent with
the notion that poor households experience relatively large benefits from economic
growth. When a city's total some of the new jobs are filled
employment increases,
by city residents who were previously unemployed or did not participate in the
labor market. The increase in employment increases real income per capita because
it increases the labor-force participation rate and speeds up the promotion of city
residents to higher-paying jobs. Bartik predicts that a 1 percent increase in total
employment increases income by 0.40 percent for the population as a
real per capita
whole, 0.43 percent for less educated workers, and 0.44 percent for black workers.
A study of the black youth unemployment problem by the National Bureau of
Economic Research (NBER) found that cities with relatively strong labor markets
had relatively low black teenage unemployment rates. For example, the employment
rate in Boston (a city with a relatively healthy economy) was 10 percentage points
above the employment rates in Chicago and Philadelphia Freeman and Holzer. 986)
( 1
Several studies have estimated the effects of recessions on poor households.

Gramlich and Laren 1984) estimated the effects of changes in unemployment rates
(
on the earnings of different types of households. The earnings of both poor and
middle-income households decrease as the unemployment rate rises, but the loss
experienced by poor households was about three times the loss of middle-income
households. In terms of racial differences, a one-point increase in the unemployment
rate decreased the income of white males by 1 .2 percent, but decreased the earnings of
nonwhite males by 2.0 percent. The implications are clear: since recessions impose
relatively large costs on the poor, policies that promote macroeconomic growth
reduce poverty.
Changes in the Relationship between Growth and Poverty
The relationship between economic growth and poverty has changed in the last sev-
eral decades (Gottschalk, 1997). During the 1960s and 1970s, poverty consistently
increased during recessions and decreased during expansions. Because the increases
in poverty during recessions were more than offset by decreases in poverty during
economic expansions, a general trend of rising income decreased the poverty rate,
from 22.4 percent in 1959 to 11.1 percent in 1973. However, since the early 1970s
the declines in poverty during expansions have not offset the increases in poverty
during recessions. Although real per capita income increased by 3 1 percent between
1973 and 1994, the poverty rate increased from 11.1 percent to 14.5 percent.
The combination of rising income and increasing poverty resulted from changes
wages of low-skilled and high-skilled workers. Over the last two decades, the
in the
demand for more educated and experienced workers increased, pulling up their
Chapter 13 Other Dimensions of Poverty 353
wages. At the same time, the demand for less skilled workers decreased, pulling
down their wages. For example, between 1 979 and 994,
1 the real earnings of college
graduates increased by 5 percent, while the real earnings of high school graduates
decreased by 20 percent.
What changes in the economy are responsible for these changes in relative
wages? A spirited debate has generated two potential explanations. First, advances
in technology allowed firms to replace many low-skilled workers with "smart" ma-
chines and at the same time increased the demand for workers who have the education
and skills required to produce the new technology and use it. The second reason is that
increased international trade (more imports and exports) has benefited high-skilled
workers and harmed low-skilled workers. The United States has a comparative ad-
vantage in producing goods using high-skilled labor and a comparative disadvantage
in producing goods using low-skilled labor. As a result, increases in international
trade will increase the demand for high-skilled workers (as exports increase) and
decrease the demand for low-skilled labor (as imports increase).
Discrimination in Labor Markets

Another possible reason for poverty is racial discrimination in labor markets. Racial
discrimination takes two forms: employment discrimination and wage discrimina-
tion. Employment discrimination refers to the practice of systematically hiring one
group over the other, despite the fact that both groups are equally productive and are
paid the same wage. Wage discrimination refers to the practice of paying different
wages to equally productive workers. Both types of discrimination have the potential
to increase poverty among certain groups of workers because either workers are not
hired or are hired at a lower wage.
Theories of Discrimination
The traditional theory of discrimination suggests that market forces limit the extent
of wage discrimination (Becker, 1957). Consider a city with two types of workers,
black and white, who have the same marginal revenue product (MRP), defined as the
marginal physical product of labor times the market price of the product produced.
Suppose that white workers are paid their MRP ($6) but blacks are paid only $5. If
you were a profit-maximizing firm, what would you do? If you hire black workers,
you will have lower production costs than your all-white competitors so your profits
will increase. What's good for you is good for other capitalists, so other firms will
hire the undervalued black workers. As the demand for black workers increases and
the demand for white workers decreases, the gap between black and white wages
narrows. In equilibrium, whites and blacks have the same wage. Similarly, we would
expect gender discrimination (lower wages for women) to disappear.
Under what conditions would wage discrimination persist? If a white firm can
keep other firms from entering its industry, it can protect itself from the competition
of color-blind firms. However, bigotry is not costless. The white monopolist pays
relatively high wages to its white workers, and its higher production costs force its
consumers to pay a relatively high price. In addition, profits are lower, so stockholders
pay for bigotry in the form of lower dividends and stock prices. Job discrimination
may persist if (1) bigoted firms are protected from competition and (2) consumers
and stockholders are willing to pay for their bigotry. Similarly, gender discrimination
may persist.
The theory of statistical discrimination suggests that the black-white wage dif-
worker productivity
ferential will persist if there is imperfect information about
(Arrow, 1973, and Phelps, 1972). Suppose that an employer must hire one of two
workers, a black or a white. If the firm knew which worker had the higher MRP, it
would obviously hire the more productive worker. Such information is not gener-
ally available, however, so the employer must guess which worker has the higher
MRP. Suppose that because of differences in education and work experience, the
average black is less productive than the average white, although some blacks are
more productive than some whites. If the employer has no other information on the
productivity of the applicants, the best choice, in a statistical sense, is the white
worker. White workers may be chosen over equally productive (or more productive)
black workers because black workers are less productive on average. In this case,
discrimination occurs because of imperfect information, not bigotry.
Facts on Earnings and Discrimination

Tabic 13-1 shows average annual earnings by race and educational level for all
workers (part-time and full-time). The two main conclusions from the table are as
follows:
1 . White workers earn more than black and Hispanic workers. For
example, black males earn 70 percent as much as white males, and black
females earn 89 percent as much as white females. The racial gaps arc
much smaller for college graduates: black male graduates earn 77 percent
as much as white graduates, and black female graduates earn 95 percent as
much as white graduates.
2. Males earn more than females. Overall, females earn 58 percent as much
as males, and the earnings gap doesn't vary much with the level of
education.
These earnings gaps result from several factors, including differences in number
of hours worked (part-time versus full-time), age, work experience, location, the
quality of education, and discrimination related to race and gender. The questions
are: how much of the earning gap is caused by racial or gender discrimination, and
how much is caused by other factors?
Consider the racial differences first. Based on recent studies of discrimination,
Schiller 1995) concluded that only about a quarter of the earnings differences are
(
caused by discriminatory practices. Darity and Mason 19 l)8) suggested that black
(
Chapter 13 Other Dimem iores of Poverty 355
Table 13-1 Annual Earnings by Race and Education
All Not a Some College

Education High School High School Time, No Bachelor's
Levels Graduate Graduate Only Degree Degree
All workers 26.792 14,013 21.431 22.392 36,980

All males 33,251 16.748 26.333 28.458 46,111
All females 19,414 9,790 15,970 16,152 26,841
White workers 27,556 14.234 22,154 22,898 37,711
White males 34,276 17.023 27,467 29.206 47,016
White females 19.647 9.582 16,696 16.125 26.916
Black workers 20,537 12.956 17,072 20,275 29,666
Black males 23.876 14.877 19,514 24.894 36,026
Black females 17,485 10.739 14,473 16,627 25,577
Hispanic workers 18,262 13.068 18.333 18,903 30,602
Hispanic males 20,312 14.774 20,882 21.705 35,109
Hispanic females 15.310 9.809 14.989 15.699 25,338
Black male as percent 70% 87% 71% 85% 77%

of white male
Black female as percent 89% 112% 87% 103% 95%

of white female
Female as percent of male 58% 58% 61% 57% 58%
SOURCE: Statistical Abstract of the United States. 1997, Table 246.
men experience a loss in earnings of 1 2 percent to 1 5 percent as a result of labor-

market discrimination. Heckman (1998) noted that the black-white earnings differ-
ence decreased substantially over the last several decades, and is now small enough
so that "most of the disparity in earnings between blacks and whites in the labor
market of the 1990s is due to the differences in skills they bring to the market, and
not to discrimination within the labor market." It appears that there is not a consensus
among economists concerning the magnitude of the discrimination penalty.
Consider next the gender differences. The first issue concerns part-time ver-
sus full-time work. In 1995, the annual earnings of full-time working women was
71 percent of the annual earnings of men (compared to the 58 percent figure for all
workers in Table 13-1). When adjustments are made for differences in education,
job experience, and other factors, the earnings gap shrinks further. There is strong
evidence that women with children earn less than otherwise-equivalent women with-
out children (Waldfogel, 1998): among workers 24 to 45 years old, women without
children earn 81 percent as much as men, while women with children earn only
73 percent as much as men.
An important factor in the gender gap is occupational crowding by gender. To the
extent that women are concentrated in low-income occupations, they will earn lower
wages than men. In 1990, the index of occupational dissimilarity was 53 percent,
indicating that to achieve equal gender representation in all occupations. 53 percent
of women (or men) would have change occupations (Bianchi and Spain. 1996).
to
Although the index decreased in recent years (dropping from 68 percent in 1970).
it is clear that there are substantial gender differences in occupational employment
that are not related to educational attainment, and this contributes to lower earnings
for women (Darity and Mason. 1998).
We can summarize our discussion of racial and gender discrimination in labor
markets as follows. The large differences in earnings between blacks and whites and
men and women are due in large part to differences in number of hours worked,
education, experience, and location. Part of the difference in earnings results from
racial and gender discrimination in the labor market. The size of this discrimination
penalty has been shrinking over time, and there is some disagreement about the
actual size of the penalty.
Demographic Change: Female-Headed Households

Another reason for poverty is the prevalence of female-headed households. As shown
in Table 12-1 (in Chapter 12) the poverty rate of female-headed households is more
than six times the poverty rate of married-couple families. Female-headed house-
holds are more likely to be poor for three reasons. First, most of these households
are single-parent households, so the woman must juggle employment and child
care responsibilities. For many women, especially those with preschool children,
full-time work is women earn lower wages than
not feasible. Second, on average,
men. so a given amount of work generates less income. Third, many female-headed
households do not receive mandated child-support payments from absent fathers.
The number of female-headed households has increased rapidly in the last few
decades. Between 1960 and 1997. the percentage of families headed by a female
increased from 10 percent to 18 percent for white families, from 28 percent to
47 percent for black families, and from 7 percent to 23 percent for Hispanic families.
1
The increase in the number of female-headed households means that fewer children
are raised in two-parent households. Between 1960 and 1997. the percentage of white
children living in two-parent households dropped from 91 percent to 77 percent,
while the percentage of black children living in two-parent households dropped
from 67 percent to 37 percent.
What caused these dramatic changes in family structure '.'
One popular but in-
correct theory is that welfare policy is responsible. The idea is that because single
women can rely on the welfare system for support, the) have less incentive to be
married. Although welfare policy does support single parents and thus decreases the
financial consequences of being a single parent, the available evidence indicates that
welfare policy does not have a significant effect on the number of single parents and
nonmarital births (Hoynes. 1996: Moffitt. 1994: Hoynes. 1995: Ellwood and Bane.
1984: Acs. 1995).
The reasons for the dramatic changes in family structure are still unknown,
at least to economists. Ellwood and Crane ( I WO) suggest that changes in family
structure resultfrom complex interactions of social, cultural, legal, and economic

factors and that it is difficult to sort these factors to determine which are the most
important and worthy of attention by the public sector. Since poverty rates are highest
for single-parent households, our lack of knowledge about the reasons for changes
in family structure is troublesome.
Poverty and Public Policy

This part of the chapter provides a brief overview of policies designed to reduce
poverty, including job-training programs, cash assistance, and in-kind transfers such
as food stamps and medical care. In 1996, the overhaul of federal antipoverty pro-
grams ended decades of policy based on the notion that poor families are entitled to
cash and in-kind assistance. The federal government now provides block grants to
the states, and the states are responsible for designing and implementing programs
to move most poor adults from welfare dependence to employment.
Employment and Job-Training Programs

The government uses employment and job-training programs to combat
federal
poverty. Employment programs operate on the demand side of the market, provid-
ing direct employment for the poor. Job-training programs operate on the supply
side of the labor market: they increase the labor skills of the poor, making them
more attractive to private industry. Currently, the emphasis is on training rather than
direct employment. In 1997, total spending on training programs for economically
disadvantaged workers was $14 billion per year.
There are dozens of benefit-cost studies of job-training programs (Bassi and
Ashenfelter, 1986; Friedlander, Greenberg, and Robins, 1997). and the studies gen-
erated three basic conclusions. First, programs for adult women —both voluntary
and mandatory — worked rather than wages.
increase earnings by increasing hours
The successful programs emphasize classroom teaching and on-the-job training,
and the largest benefits go to women with little prior work experience. Second, most
programs for youth do not lead to significant increases in earnings. One exception
is the Job Corps program, which in some circumstances increases earnings. Third,
most training programs have relatively small effects on the earnings of adult men.
One exception is the current national voluntary training program (Job Training and
Partnership Act, or JTPA), which generates encouraging results for adult men.
Cash Payments and Payments in Kind

Table 13-2 shows federal spending on various antipoverty programs in 1994, before
the welfare reform program enacted in 1996. Of the total budget of $219 billion,
less than a quarter was spent on direct income transfers, which includes Aid to Fam-
ilies with Dependent Children (AFDC), Supplemental Security Income (SSI), and
Table 13-2 Spending on Transfer Programs in 1994
Number of Average
Recipients Monthly Annual Cost
Program {millions) Benefit ('$) ($ billion)
Income Transfers
Supplemental Security Income (SSI) 5.3 364 21
Aid to Families with Dependent
Children (AFDC) 14.1 137 23
General Assistance (GA) 1.2 150 2
In-Kind Transfers
Food stamps 24 66 22
Medicaid 27 200 127
Housing assistance 6 115 20
School Lunch Program 24 20 4
Total 219
SOURCE: Bradley Schiller. The Economics of Poverty and Discrimination. 6th ed. (Englewood Cliffs, N.J.:
Prentice-Hall. 1995).
General Assistance. The bulk of AFDC money went to female-headed households,

and the average payment per person was $ 1 37 per month. SSI is a special program for
the aged, the blind, and the permanently disabled. General Assistance is a program
for poor households that do not qualify for AFDC or SSI.
As shown in Table 1 3-2, federal spending on in-kind programs was $ 73 billion 1
in 1994. about three times the total spending on direct income transfers. The three
largest in-kind programs are Medicaid ($127 billion), food stamps ($22 billion), and
housing ($20 billion). Medicaid provides free medical care to the recipients of AFDC
and SSI. It is essentially a free medical insurance policy: the government agrees to
cover all the recipient's medical expenses, just as a private insurer would. Under the
food stamp program, poor households are given coupons that can be exchanged for
food, with the number of coupons determined by household income.
Welfare Reform
In August of 1996. President Clinton signed into law "The Personal Responsibility
and Work Opportunity Reconciliation Act of 996," a comprehensive welfare-reform
1
plan. The plan abolished AFDC and eliminated the entitlement of poor families to
receive cash assistance. The federal government now provides block grants to states
to provide Temporary Aid to Needy Families (TANF). with restrictions on what
recipients must do to qualify for assistance and how long they can receive it.
• After a maximum of 24 months of assistance (consecutive or

nonconsecutive), a recipient must participate in work activities, defined as
employment, on-the-job training, work experience, community sen ice.
vocational training, or child care services for individuals providing

community service. Single parents must participate at least 30 hours per
week, and two-parent families must participate at least 35 hours per week.
There are exceptions to this rule for the parents of very young children.
• After a total of 60 months of cash assistance (consecutive or
nonconsecutive). assistance is stopped. States can allow exceptions to this
rule, with the restriction that no more than 20 percent of the state's recipients
exceed the 60-month limit.
The federal government also provides incentives for states to reduce welfare
dependence. The federal government will cut the state's block grant if the state fails
to meet ambitious targets for ( 1 ) increasing the fraction of welfare recipients who
work or (2) reducing the number of people on welfare. Another incentive is $ 1 billion
available for performance bonuses to reward states for moving welfare recipients
into jobs. The general thrust of this welfare-reform change the welfare
package is to
system into one that requires work in exchange for time-limited assistance. There
are also block grants for child care subsidies to help working parents.
The states are responsible for implementing welfare reform. Florida's reform
—
law known as Work and Gain Economic Self-Sufficiency (WAGES) provides —
an example of one state's response to the new welfare law. The program gives
temporary assistance to poor families with children and provides parents with work
opportunities and support services. This is a "work-first" program: when a person
applies for assistance, he or she is immediately referred for employment. Unless
otherwise exempted, each recipient must participate in work activities to the full
extent of the federal guidelines. Individuals exempt from work activities include
( 1 ) custodial parents with children younger than three months. 2 ) recipients younger
(
than 16 years, and (3) recipients who are eligible for SSI because of age or disability.
For a family of three, the maximum cash assistance is $303 per month, which is less
than one-third of the poverty level for such a family. Cash assistance is limited to
24 consecutive months in any 60-month period, and no adult participant can receive
payments for more than 48 months.
What are the likely market effects of welfare reform? The switch to a program
requiring work in exchange for time-limited assistance will increase the number
of low-skilled people in the labor market. Bartik (1998) estimates that between
1996 and 2008, welfare reform will increase the supply of low-skilled workers by
between 1.0 and 1.7 million workers, with most of the new workers being women
with relatively low educational attainment. As shown in Figure 13-1, an increase in
the number of low-skilled workers shifts the labor-supply curve to the right: at each
wage, more hours of labor are supplied. In this example, the wage drops from $7.00
to $6.50 and the quantity of labor increases.
The actual effects of welfare reform on wages and employment will be deter-
mined by the elasticities of supply and demand in the market for low-skilled labor.
For the least educated and unskilled workers (high school dropouts and the least pro-
ductive high school graduates), wages could drop by about 6 percent (Holzer, 1996).
Women from female-headed households who worked before welfare reform could
Figure 13-1 Welfare Reform and the Market for

Low-Skilled Workers
Initial supply
Supply after
welfare reform
7.00
230 270
Quantity of labor ( 1 .000 hours)
Welfare reform will increase the number of low-skilled workers in the market, shifting the supply curve
to the ri«ht and decreasing the wages of low-skilled workers.
see their earnings drop by about 8 percent (Bartik. 1998). The largest losses in earn-
ings will be experienced by low-skilled women, because they will face additional
competition as many former AFDC recipients join the workforce.
The Homeless
The McKinney Homeless Assistance Act of 1987 defines a homeless person as
someone who sleeps (1) outside, (2) inside in places not intended for sleeping (e.g.,
the lobby of a public building), or (3 ) in housing shelters (places providing temporary
housing). A consensus estimate of the number of people who are homeless on a given
night between 250,000 and 350,000 (Honig and Filer, 1993).
is
The rate of homelessness varies considerably across metropolitan areas, with

the highest rates in the largest cities. For example, the number of homeless people
per 100,000 population is about 535 in San Francisco. 412 in Los Angeles. 349 in
Miami. 346 in New York, and 324 in Chicago. The population of New York is more
than 7.3 million, meaning that on any given night there are over 25.000 homeless
people on the streets or in shelters. In the middle range, the numbers for homeless
Chapter 13 Oilier Dimensions of Poverty 361
per 1 00,000 population are 87 for Boston. 76 for Philadelphia, 63 in Phoenix, and 60
in Louisville. In the lower range, the rate is 3 in Baltimore, 27 in Colorado Springs.
1
25 in Kansas City, 22 in Cleveland, and 14 in Rochester.

Studies of the homeless indicate that the typical homeless person is a single man.
One survey (Burt, 1992) computed the percentages of the adult homeless population
as follows:
• Single men: 73 percent.

• Single women: 9 percent.
• Women with children but no spouse: 8 percent.
• Married couples with children: 1 percent.
• Married couples without children: 2 percent.
The remainder of the homeless population consisted of adults traveling together (with
relatives or nonrelatives). Children comprise about 15 percent of the homeless popu-
lation.
Figure 13-2 shows the percentages of the homeless population that experience
various difficulties. About half are high school dropouts, and about one in five spent
some time in a mental hospital. About one in three homeless people spent some
time as a patient in a chemical-dependency program. Over half of the homeless
spent time in jail (a short period for misdemeanor violations), and about a quarter
spent time in prison (longer periods for felony violations). About one in five homeless
persons has attempted suicide.
Figure 13-2 Characteristics of Homeless Population

60 r
53
49
50 -
r- 40 -
33
i 30-
24
Z 20
igh school Spent time Treated for Spent time Spent time Attempted
dropouts in mental chemical in jail in prison suicide
hospital dependency
SOURCE: Author's calculations based on data in Burt. Martha R. Over the Edge: The Growth of
Homelessness in the 1980s (New York: Sage. 1992).
Figure 13-3 Elasticities for Homelessness

1.50
1.25
-
1.00
0.50
Size of
0.22
0.00
Rent for
Employment
growth
tzt
rate
welfare
payment
Institutionalization
of mentally ill
n
City size
low-quality 0.15
housing 0.32
-0.50 -
-
1.00
-1.50 -
1.54
-2.00
Source: Author's calculations based on data in Honig. Marjorie. and Randall K. Filer. "Causes of Intercity
Variation in Homelessness."American Economic Review 83 1993). pp. 248-55.
(
What causes homelessness? From an economic perspective, a person will be

homeless if his or her low enough relative to the price of housing that it is
income is
not sensible, or even possible, to purchase housing services. A study of the causes
of homelessness bears out this simple theory (Honig and Filer, 1993). Looking at
data for different metropolitan areas, we see that the homeless rate varies across
metropolitan areas, and so do several variables that affect a household's ability to
purchase housing services. The question is: do variations in these other variables
explain the observed variation in homeless rates?
Figure 13-3 shows the sensitivity of homeless rates to changes in several vari-
ables. The numbers shown are the elasticities of the homeless rate with respect to
selected variables. For example, the elasticity for rent on low-quality housing is
1.25, meaning that a 10 percent increase in rent on low-quality housing increases the
homeless rate by 12.5 percent. In the sample of cities, the average rent on low-quality
housing is $135, so a rent hike to $148 would increase the average city's homeless
rate from 186 per 100.000 population to 209.
One factor in the growth of the homeless population in the 1970s and 1980s was
a reduction in availability of low-quality housing in large cities (Burt. 1993). Histor-
ically, many of the poor lived in boarding houses, rooming houses, and residential
hotels (also known as single-room occupancy units, or SROs). This housing had small
living quarters, shared bathrooms, and limited kitchen facilities, but it was afford-
able for most pensioners and day laborers. During the 1970s and 1980s, many SROs
were eliminated, a result of urban renewal and conversion to other uses. Many of
the hotels and large houses used for boarding or rooming houses were "rejuvenated"
for higher-income households. The percentage loss in SRO units was 19 percent in
Chicago, 50 percent in Los Angeles, 38 percent in New York, 42 percent in Cincin-
nati, and 25 percent in San Diego. Nationwide, a total of 1.1 million SRO units
were lost during the 1970s and 1980s. The loss of SROs contributed to the homeless
problem by decreasing the supply of low-quality housing and increasing its price.
The numbers in Figure 13-3 indicate that changes in income affect the number
of homeless people. The elasticity for employment growth suggests that a 10 per-
cent increase in a city's employment growth rate will decrease the homeless rate
by 1.5 percent. This is sensible because employment growth results in more jobs
and higher wages for the working poor, so they are more likely to be able to pay for
housing. The elasticity for welfare payments suggests that cities with more generous
welfare programs have lower homeless rates: a 10 percent increase in welfare pay-
ments decreases the homeless rate by 15.4 percent. This is sensible because welfare
recipients can use the additional money to pay for housing.
The elasticity for institutionalization of the mentally ill indicates a negative re-
lationship between institutionalization and homelessness. A 10 percent increase in
the institutionalization rate (the number of patients in mental hospitals per 100,000
population) leads to a 3.2 percent decrease in the homeless rate. This result is con-
sistent with the idea that policies concerning the mentally ill affect the number of
homeless people. In the 1970s, many mentally ill people were released from state
mental hospitals, with the idea that they would live in the community and be treated
in community centers. A common perception is that many of the people released
from mental hospitals ended up homeless. Figure 13-3 suggests that this perception
is accurate.
Summary
1. Poverty rates are greatest among racial minorities, children, female-headed
households, high school dropouts, and central-city residents.
2. Discrimination in labor markets leads to lower wages for racial minorities and
some types of female workers.
3. The number of female-headed households increased significantly in the last few
decades.
4. Poverty rates are sensitive to the general health of the economy because the
poor are the last hired and the first fired.
5. Among the antipoverty policies used by the national government are

job-training programs, cash transfers, and in-kind transfers.
6. The general thrust of the recently adopted welfare-reform package is to change

the welfare system into one that requires work in exchange for time-limited
assistance.
364 Part HI Poverty and Housing
7. The factors that cause homelessness include high prices for low-quality
housing, low welfare payments, low rates of institutionalization for the mentally
ill, and slow growth in employment.

1 The section on job discrimination suggests that the market penalizes racist and
sexist firms because such firms are underpriced by color-blind and sex-blind
firms. As a result, there are limits on the extent of wage discrimination.
a. What if all firms (capitalists and managers) are color-blind, but white
laborers refuse to work with black laborers? Will wage discrimination (lower
wages for blacks) persist?
b. What if firms and employees are color-blind, but white consumers refuse to
purchase goods produced by blacks? Who bears the cost of bigotry? Will
wage discrimination persist?
2. Suppose comparable-worth law is enacted, requiring a 20 percent
that a
increase in the wages of female-dominated occupations. Who will bear the costs
of this policy? How will the policy affect nonworking women and women who
work in male-dominated occupations?
3. In the state of Florida, the implementation of a welfare-reform package that
forced many welfare recipients to work occurred at the same time as an increase
in the minimum wage from $4.25 to $4.75. Depict graphically the market effects
of these two changes on the market for low-skilled labor. To simplify matters,
assume that the old minimum wage was actually the market-equilibrium wage.

Poverty Facts and General References
Sawhill. Isabel V. "Poverty in the U.S.: Why Is It So Persistent?'" Journal of Economic
Literature 26 ( 1988). pp. 1073-1 19. Explores the reasons for the persistence of poverty
between the middle 1960s and 1988 and concludes that the most important factors
were demographic changes and rising unemployment.
Schiller. Bradley. The Economics of Poverty and Discrimination. 6th ed. Englewood Cliffs.
N.J.: Prentice-Hall. 1995.
U.S. Bureau of the Census. Statistical Abstract of the i 'nited States. 1997. Washington,
D.C.: U.S. Government Printing Office. 1997.
Poverty and Economic Growth

Bartik. Timothy. Who Benefits from State and Local Economic Development Policies?
Kalamazoo, Mich.: Upjohn Institute forEmployment Research. 1991.
Danziger, Sheldon, and Peter Gottschalk. "Do Rising Tides Lift All Boats? The Impact of
Secular and Cyclical Changes on Poverty." American Economic Review 76 (1986),
p. 410.
Gottschalk, Peter. "Inequality, Income Growth, and Mobility: The Basic Facts." Journal of
Economic Perspectives (1997), pp. 21-40.
1 1
Gramlich, Edward, and Deborah Laren. "How Widespread Are Income Losses in a
Recession?" In The Social Contract Revisited, ed. D. Lee Bauden. Washington, D.C.:
Urban Institute, 1984. Estimates the distributional effects of recessions.
Labor-Market Discrimination
Arrow, Kenneth. "The Theory of Discrimination," In Discrimination in Labor Markets,
and Albert Rees. Princeton: Princeton University Press, 1973.
ed. Orley Ashenfelter
Describes the phenomenon of statistical discrimination, which explains the persistence
of wage differentials.
Becker, Gary S. The Economics of Discrimination. Chicago: University of Chicago Press,
1957. Describes the market theory of discrimination, which suggests that competition
limits the extent of wage discrimination.
Bianchi, Suzanne M., and Daphne Spain. "Women, Work, and Family in America."
Population Bulletin 51, no. 3 (1996), pp. 2-48.
Darity, William A., and Patrick L. Mason. "Evidence on Discrimination in Employment:
Codes of Color, Codes of Gender." Journal of Economic Perspectives 1 2, no. 2 ( 1 998).
pp. 63-90.
Heckman, James J. "Detecting Discrimination." Journal of Economic Perspectives 12, no. 2
(1998), pp. 101-16.
Phelps, Edmund. "The Statistical Theory of Racism and Sexism." American Economic-
Review 62 (1972), pp. 659-61. Explains the theory of statistical discrimination.
Smith, James P., and Finis Welch. "Race Differences in Earnings: A Survey and New
Evidence." In Current Issues in Urban Economics, ed. Peter Meiszkowski and Mahlon
Straszheim. Baltimore: Johns Hopkins, 1979. Summarizes the evidence on wage
discrimination.
Waldfogel, Jane. "Understanding the 'Family Gap' in Pay for Women with Children."
Journal of Economic Perspectives 12, no. 1 (1998), pp. 137-56.
Female-Headed Households
Acs, Gregory. "Do Welfare Benefits Promote Out-of- Wedlock Childbearing?" In Welfare
Reform: An Analysis of the Issues, ed. Isabel Sawhill. Washington DC: The Urban
Institute, 1995.
Ellwood, David T, and Mary Jo Bane. "The Impact of AFDC on Family Structure and
Living Arrangements." Research in Labor Economics 1 ( 1985). Explores the
relationship between welfare benefit levels and family structure.
Ellwood, David and Jonathan Crane. "Family Change among Black Americans: What
J.,
Do We Know?" Journal of Economic Perspectives 4 (Fall 1990). pp. 65-84.

Hoynes, Hilary W. "Does Welfare Play Any Role in Female Headship Decisions?" NBER
Working Paper 5 149. Cambridge, Mass.. 1995.
Hoynes, Hilary W. "Work, Welfare, and Family Structure: What Have We Learned?" NBER
Working Paper 5644. Cambridge, Mass., 1996.
Moffitt, Robert. "Welfare Effects on Female Headship with Area Effects." Journal of
HumanResources 29 (1994), pp. 621-36.
Murray. Charles. Losing Ground: American Social Policy. 1950-1980. New York: Basic
Books, 1984. Argues that welfare policy was a large factor in the growth of female-
headed households.
Public Policy: Training Programs
Bassi, Laurie, and Orley Ashenfelter."The Effect of Direct Job Creation and Training
Programs on Low-Skilled Workers." In Fighting Poverty: What Works and What
Doesn't, ed. Sheldon H. Danziger and Daniel H. Weinberg. Cambridge. Mass.:
Harvard University Press. 1986. Discusses the history of employment and job-training
programs and summarizes the conclusions from several studies of these programs.
Friedlander. Daniel; David H. Greenberg: and Philip K. Robins. "Evaluating Government
Training Programs for the Economically Disadvantaged." Journal of Economic
Literature 25 (1997), pp. 1809-855.
Welfare Reform
Bartik, Timothy. Displacement and Wage Effects of Welfare Reform. (Kalamazoo. Mich.:
Upjohn Institute, 1998).
Holzer, Harry "Employer Demand, AFDC Recipients, and Labor Market Policy." In
J.
Research on Poverty Discussion Paper No. 1 15-96, 1996.

Institute for 1
Witte, Ann D.; Magaly Queralt; Tasneem Chipty: and Harriet Griesinger. "Unintended
Consequences? Welfare Reform and the Working Poor." National Bureau of Economic
Research, Working Paper 6798 (1998).
Homelessness
Burt. Martha R. Over the Edge: The Growth of Homelessness in the 1980s. New York:
Sage, 1992.
Honig, Marjorie, and Randall K. Filer. "Causes of Intercity Variation in Homelessness."
American Economic Review 83 (1993), pp. 248-55.
Chapter 14
w HY IS HOUSING DIFFERENT?
(^JT~ his is the first of two chapters on urban housing. It explains why housing is a
K-S unique commodity, focusing on six features that make housing different from
other goods. First, the stock of housing is heterogeneous: dwellings differ in size,
location, age, floor plan, interior features, and utilities. Second, housing is immobile:
it is impractical to move dwellings from one location to another. Third, housing is
durable: if properly maintained, a dwelling can be used for many decades. Fourth,
housing is expensive: to purchase a dwelling, the typical household must borrow
a large sum
of money. Fifth, moving costs are relatively high: in addition to the
substantial monetary cost of moving, there are also costs associated with leaving
the old neighborhood — —
with its schools, stores, and friends behind. Finally, some
people care about the racial and ethnic background of their neighbors, leading to
racial discrimination and segregation.
The remainder of the chapter explains how these six characteristics affect the
housing market. The first section shows that because housing is heterogeneous and
immobile, the housing market is split into a set of distinct but related housing submar-
kets. The second and third sections explore some of the implications of durability.
The second section discusses the landlord's decisions concerning maintenance and
repair and explains some of the market forces behind the abandonment of older
housing in central cities.The third section describes the filtering model of hous-
ing, a model that explains the process through which used housing is passed from
one type of household to another. The fourth section discusses the household's de-
cision about whether to rent or own and explores the effects of public policy on
the choice between renting and owning. The fifth section discusses the effects of
moving costs on housing consumption. The sixth section discusses the income and
price elasticities of demand for housing. The final section explores the effects of
racial prejudice and discrimination on housing prices and the racial composition of
neighborhoods.
367
Heterogeneity and Immobility

The housing stock is heterogeneous in the sense that each dwelling offers a different
set of features or a different bundle of housing services. There are two types of
housing features: dwelling characteristics and site characteristics.
Consider first the features of the dwelling itself. Dwellings differ in size (square
footage of living space) and layout (the arrangement of rooms within the dwelling).
They also differ in the quality and efficiency of kitchen equipment and utility sys-
tems (heating, air conditioning, plumbing,Other differences occur in the

electrical).
interior design (type of flooring, windows, cabinets) and structural integrity (the
durability of the foundation and the roof). To summarize, each dwelling offers a dif-
ferent combination of and structural integrity.
size, layout, utilities, interior design,
Because housing is immobile, one component of the housing bundle is the

residential location. A house buyer purchases both a dwelling and a set of site char-
acteristics. One site characteristic is accessibility: sites differ in their access to jobs,
shopping, and entertainment. Another characteristic is the provision of local pub-
lic services: metropolitan areas have dozens of local governments, each of which
provides a different combination of taxes and public services (schools, fire protec-
tion, police services). A third characteristic is environmental quality: sites differ
in air quality and noise levels (from cars, trucks, airplanes, and factories). A final
site characteristic is the appearance of the neighborhood (the exterior features of
neighboring houses and lots). To summarize, housing is consumed along with a
residential site, so the housing bundle includes several site attributes, including ac-
cess to different facilities, tax liabilities, public services, environmental quality, and
neighborhood characteristics.
The Price of Housing: The Hedonic Approach

What determines the equilibrium price of a dwelling? The hedonic approach is based
on the notion that a dwelling is composed of a bundle of individual components,
each of which has an implicit price. The market price of a dwelling is the sum of the
prices of the individual components.
To explain the hedonic approach, consider a housing market in which dwellings
differ in only five ways. The dwellings differ in access to jobs in the city center,
the number of bedrooms, the age of the roof, air pollution, and the quality of local
schools. A hedonic study of the market might generate the following information:
1 Base price. The average house has three bedrooms, is five miles from the
city center, and has a roof six years old. The price of the average house is
$70,000.
2. Access price. The price of housing drops by $ .000 for every additional
1
mile from the city center: more accessible dwellings have higher prices.
3. Bedroom price.The price of housing increases by S 10.000 for every
additional bedroom: larger dwellings have higher prices.
Chapter 14 Why Is Housing Different? 369
4. Roof price. The price of housing decreases by $ 100 for every additional
year of roof age; an older roof means that the roof must be replaced sooner,
so the market price of the dwelling is lower.
5. Air quality price. The price of housing decreases by $500 for every
additional unit of air pollution: dwellings in areas with relatively clean air
have higher prices.
6. School price. The price of housing increases by $600 for every one-unit
increase in the quality of the local elementary school (measured by the
average test score); dwellings in areas with better schools have higher
prices.
To predict the price of a particular dwelling, one needs information on location,

the number of bedrooms, the age of the roof, air quality, and the quality of the
elementary school. For example, consider a four-bedroom house located three miles
from the city center that has a two-year-old roof. The pollution level is three units
below the average and the average test score is two points above the average. Such a
house sells for $85, 1 00, the base price of $70,000 plus $ 1 0,000 for the extra bedroom,
$2,000 for the shorter commute, $400 for the four extra years of roof life, $ 1 ,500 for
the relatively low pollution level, and $ ,200 for the relatively high average test score.
1
Kain and Quigley (1975) use data from the St. Louis housing market in the
1960s to estimate the dollar values of different housing attributes. Table 14-1 shows
Table 14-1 Change in Housing Prices from One-Unit Increases in Housing

Attributes: A Hedonic Study of the St. Louis Housing Market
Rental Market: Owner-Occupied Market:

Attribute Increase in Monthly Rent <$) in Market Value <$)
Increase
Dwelling Quality
Interior 1.31 818
Central heating 4.44
Age -0.29 -100
Size of Dwelling
Number of rooms 22.63 1,453
Number of baths 9.07 769
Site Characteristics
Exterior quality of adjacent dwellings 1.86 777
Exterior quality of dwell ings on block 3.71 419
Miles from CBD -0.30 -354
Average monthly rent = $61 .34

Average market value = $14,596
SOURCE: John F. Kain and John M. Quigley. Housing Markets and Racial Discrimination: A Microeconomic
Analysis (New York: National Bureau of Economic Research. 1975). Table 8.3.
some of their There are three types of housing characteristics: the quality of
results.
the dwelling itself, the size of the dwelling, and site characteristics. The numbers
in the table show the increases in monthly rent (for rental housing) and market
value (for owner-occupied housing) resulting from one-unit increases in the various
housing attributes. For example, a one-unit increase in interior quality (a measure of
the quality of floors, windows, walls, ceilings, stairways) increases monthly rent by
$1.31 (about 2.1 percent of the average monthly rent of $61 .34). and increases market
value by $818 (about 5.6 percent of the average market value of $14,596). Rental
units with central heating rent for $4.44 more per month, and older dwellings have
lower rents and market values. The price of housing is also affected by dwelling size:
a second bathroom adds $9.07 to monthly rent and $769 to market value. The exterior
quality of nearby dwellings was measured on a scale of 1 (bad) to 5 (excellent). A
one-unit increase in the quality of adjacent dwellings increased rent by $1.86 and
market value by $777. while a one-unit increase in the quality of dwellings on the
block increased rent by $3.71 and market value by $419. The final variable is distance
to the central business district: a one-mile increase in distance to the employment
center decreased rent by $0.30 and value by $354.
Choosing a Dwelling
How does a household choose among alternative dwellings, each of which provides
a different bundle of characteristics? The household must find the dwelling with
the best combination of features at the best price. Most consumers do not have
access to a hedonic study of their housing market. As they shop, they gather their
own information about the implicit prices of location, size, and design features.
Eventually, the household chooses a bundle that maximizes the household's utility
subject to its budget constraints.

The household chooses the dwelling that provides the best affordable combi-
nation of features. Consider a housing market in which dwellings differ in only
two respects: size (square feet of living space) and quality (of floors, walls, roof).
Figure 14-1 shows the housing choices for three different households.
The budget lines in Figure 14-1 show the combinations of size and quality
that exhaust a household's housing budget. AB is the budget line for middle-income
households S and L. and CD is the line for H. a high-income household. The position
of the budget line is determined by the household's housing budget: the larger the
budget, the larger the set of affordable combinations of size and quality. The slope
of the budget line is determined by the implicit prices of size and quality: the larger
the price of quality relative to the price of size, the flatter the budget line.
The indifference curves in Figure 14-1 show the different combinations of
quality and size that generate the same utility level for a given household. Three
indifference curves are shown, one each for households S. H, and L. The indifference
curves are negatively sloped, reflecting the household's subjective trade-off between
quality and size: as quality decreases, the household needs a larger dwelling to
achieve the same utility level.
Housing choices reflect the households' preferences and income. Given the
shape of its indifference curves, household L (a middle-income household) chooses a
Figure 14-1 Households with Different Income and Preferences

Choose Dwellings of Different Size and Quality
Size (square feet of living space)
The household's choice of size and quality reflects its subjective trade-off (as represented in its
indifference curves), the market trade-off (as shown in the slope of the budget line), and household
income (which determines the position of Households S and L have different tastes for
the budget line).
housing: L chooses a large, low-quality dwelling, while S chooses a small, high-quality dwelling. The
high-income household (//) chooses a large, high-quality dwelling.
large, low-quality dwelling. Although household S has the same income and housing
budget as L, S chooses a smaller, higher-quality dwelling. Its choice is rational
because S has relatively strong preferences for quality, as shown by the shape of its
indifference curve. Household H has a higher income and a larger housing budget,
so it chooses a large, high-quality dwelling. The high-income household faces the
same monetary trade-off between size and quality (the same opportunity cost) but it
has more money to spend on both size and quality.
Segmented but Related Markets

The heterogeneity of the housing stock means that a city's housing market is com-
posed of a number of submarkets. The housing market is segmented with respect to
size, location, and quality. For example, some households look for small dwellings,
while others look for large ones. Therefore, there is one market for small dwellings
and another market for large ones. Similarly, some households look for houses acces-
sible to central-city jobs, while others look for houses in high-quality school districts.
Finally, some households look for new houses, full of modern conveniences, while
others look for older houses with old-world charm.
Although the housing market is segmented into submarkets, the submarkets

are related because consumers have some flexibility in their housing choices. As
the relative price of three-bedroom houses increases, some households switch from
the three-bedroom market to the two-bedroom market. Similarly, as the price of
older houses increases relative to the price of new ones, some households switch
to the new-housing market. In other words, dwellings from different submarkets
are imperfect substitutes, so households move between the submarkets in response
to changes in the relative prices of different types of housing. The implications of
housing submarkets are discussed later in the chapter.
Neighborhood Effects
The quantity of housing services produced by a particular dwelling depends not
only on the characteristics of the dwelling, but also on the characteristics of its
neighborhood. When one homeowner improves the appearance of his house by

painting it or repairing its broken windows, the neighborhood becomes a more
desirable place to live, so the market values of surrounding houses increase. This is
the neighborhood effect: positive changes in the exterior appearance of one house
cause spillover benefits (increases in market value) for surrounding houses. Although
thereis no doubt that neighborhood effects exist, there is some question as to their
magnitude and geographical extent.

Economists use the hedonic approach to estimate the magnitude of neighbor-
hood effects (see Crecine, Davis, and Jackson, 1967; and Kain and Quigley, 1975).
In Table 14-1, a one-unit increase in the quality of adjacent dwellings increased
monthly rent by $1.86 and increased market value by $777. Similarly, improve-
ments to other dwellings on the block increased rent and market value. The evidence
from hedonic studies suggests that neighborhood effects are highly localized: the
spillover benefits from positive changes in the exterior appearance of one house are
confined to dwellings within a few hundred feet of the house.
Durability of Housing
Housing is more durable than most goods. If a dwelling is maintained properly, it
can 100 years or more. Although dwellings deteriorate over time, they do so at a
last
relatively slow rate. The durability of housing has three implications for the housing
market. First, the landlord can control the rate of physical deterioration by spending
money on repair and maintenance. Second, there is a large supply of used housing on
the market every year. The general rule of thumb is that new construction in a given
year between 2 percent and 3 percent of the total housing stock. Over the course
is
of a decade, new dwellings provide about 20 percent to 30 percent of the housing

stock, so between 70 percent and 80 percent of households live in dwellings that are
at least 10 years old. The final implication is that the supply of housing is relatively
inelastic: the market dominated by the stock of used housing, so changes
is in price
cause relatively small changes in the quantity supplied.

Chapter 14 Why Is Housing Different': 373
Deterioration and Maintenance

A property owner can control the rate of physical deterioration by spending time and
money on repair and maintenance. The landlord has the incentive to maintain and
repair her property because these activities increase the quantity of housing services
(Q) generated by the dwelling. Consumers are willing to pay more for a dwelling
that generates a larger quantity of housing services, so an increase in maintenance
increases rent and market value. The question for the landlord is: how much should
Ispend on maintenance?
Figure 14-2 shows the benefits and costs of maintenance. The horizontal axis
measures the quantity of housing services generated by the dwelling (Q), which the
landlord can affect by repairing broken windows and leaky roofs, maintaining the
heating and plumbing systems, and painting the house. The total-cost curve shows
the sum of fixed costs (independent of Q) and variable costs. The fixed costs include
the cost of managing the property (collecting rent, advertising for tenants) and taxes.
Figure 14-2 Optimum Quantity of Housing Services
Total revenue
Total cost
in year 20
1
30 20
Quantity of housing services (Q)
The optimum quantity of housing services is the Q at which profit (total revenue less total cost)
ismaximized, in year 20. profit is maximized at 2:o- The aging of the dwelling increases
maintenance cost, shifting the cost curve upward and decreasing the optimum Q.
The variable costs include repair and maintenance costs. The cost curve is positively
sloped because variable costs increase with Q: it costs more to maintain a dwelling
at a higher Q. The cost curve is convex from below, a result of diminishing returns
to maintenance: as Q becomes progressively more costly to maintain
increases, it
the property at the given Q. The straight line in Figure 14-2 shows the relationship
between total revenue (rent per month) and Q. The implicit assumption is that the
price per unit of Q is constant: a dwelling with twice the quantity of housing services
commands twice the rent.
Profit equals the difference between total revenue and total cost. In year 20
(20 years after the dwelling was built), the landlord has the lower of the two cost
curves. Profit is positive above Q\, and reaches its maximum level at Q20. At this Q.
the slope of the revenue curve equals the slope of the cost curve: the marginal benefit
(change in total revenue per unit change in Q) equals marginal cost (the change in
total cost per unit change in Q). Therefore, the profit-maximizing landlord chooses
an output of Qiq units of housing services.
The optimum quantity changes over time, a result of changes in variable costs
or changes in demand.
1 Changes in variable costs. As the dwelling ages, it becomes progressively

more costly to maintain at a given quality level. In Figure 14-2, between
years 20 and 30, the fixed cost does not change, but maintenance cost
increases, causing an upward tilt of the total-cost curve. By year 30. the
optimum quantity drops from Qiq to (?30. and the landlord's profit falls.
2. Changes in demand. Suppose that the demand for housing in a particular

area drops, decreasing the equilibrium price of rental housing. In Fig-
ure 14-2, the total-revenue function would tilt downward (pivoting from
the origin), decreasing the profit associated with every Q. As the revenue
function tilts downward, the optimum Q decreases.
The Retirement Decision

What happens when the profit from rental housing is negative? Suppose that because
of changes in costs or demand, the total-cost curve lies above the total-revenue curve
for all values of Q. Rental housing is clearly unprofitable, leaving the landlord with
three options: conversion, boarding up. and abandonment.
1. Conversion. A dwelling can be converted to nonresidential use. such as an

office, a store, or a parking lot. Conversion is profitable if the alternative
activity generates enough profit to offset the cost of conversion to the
nonresidential use. If conversion would require the demolition of the
dwelling and the construction of a new building, conversion cost is
relatively high, and conversion is unlikely.
2. Boarding up. A dwelling can be boarded up and taken off the market
temporarily. The board-up option is profitable if two conditions are met.
First, if the landlord expects the market rent to increase sometime in the
future, the expected future profit is positive. Second, if the landlord's

carrying cost (the opportunity cost of keeping his money in housing instead
of a bank account) is relatively low, the cost associated with waiting for the
market rent to increase is relatively low. This type of temporary retirement
was common during the Great Depression.
3. Abandonment. The owner can disown the dwelling, walking away from
the property. Abandonment is profitable if the alternative uses (retail,
commercial, industrial) do not generate enough profit to cover the cost of
converting the property from residential to nonresidential use. If conversion
cost is high, the market value of the property is zero, so there is no reason to
keep title to the land.
Retirement results from three types of changes that decrease the profitability
of rental housing. On the demand side of the market, a decrease in average income
or a decrease in population lowers the demand for housing in certain areas. As
the market rent falls, the total-revenue curve tilts downward, decreasing profit. On
the supply side, an increase in the supply of rental housing decreases market rents
and profits. Finally, as the dwelling ages, increases in maintenance cost decrease
profits for a given total-revenue curve, increasing the likelihood of conversion or
abandonment.
Abandonment and Public Policy
How does tax policy affect abandonment? White (1986) shows that the property
tax increases the frequency of abandonment. Because old, low-quality dwellings
have relatively high tax liabilities, landlords have less incentive to keep title to their
properties. By abandoning them, they can avoid taxes.
Figure 14-3 shows the effects of the property tax on the landlord's revenue and
cost. Given the initial total-revenue curve, the optimum quantity of housing services
is <2o; total revenue is $340, total cost is is $150 per month. Suppose
$190, and profit
that the assessed value of the rental property $24,000 and the tax rate is 3 percent
is
of assessed value per year, or $720 per year ($60 per month). If so, the post-tax profit
is$90 ($150 -$60).
Suppose that the demand for housing decreases, tilting the total-revenue curve
downward. The profit-maximizing Q drops to Qi, and pre-tax profit drops to $48
($210 — $162). The decrease in profit decreases the market value of the property:
an investor is willing to pay less for the property because it yields less profit. In a
perfect tax world, the government would decrease the assessed value of the property
to reflect its lower market value, thus lowering the landlord's tax liability. If the tax
stays at $60, however, the post-tax profit is negative ($48 — $60 = —$12), so the
landlord is likely to abandon the property. To prevent abandonment, the government

would have to cut the tax liability to less than $48.
White (1986) explores the effects of various factors on abandonment in New
York City and concludes that the property tax is by far the most important factor.
She estimates that the elasticity of abandonment with respect to the property tax is
Figure 14-3 A Fixed Property Tax of $60 Encourages Abandonment

Original total
revenue
Total cost
Total revenue
after decrease in
demand
Quantity of housing services (Q)
If the property tax is fixed at profit is $90 instead of $ 50. A decrease in demand tilts the
$60 per month, post-tax 1
revenue curve downward, decreasing the optimum service level to Q : Pre-tax profit is $48. but post-tax prolit i>-
.
—$12. A fixed property tax causes abandonment despite a positive pre-tax profit. A reassessment of the properrj
would decrease the tax liability and make abandonment less likely.
1.65: a 10 percent increase in the property tax increases the frequency of abandon-
ment by 16.5 percent. For example, if the average assessed value of properties in
the Brownsville section of Brooklyn were cut by $1,000 (a 6 percent reduction) the
resulting decreases in property taxes would lower the abandonment rate from 7 per- 1
cent per year to 14.8 percent. Given this large elasticity, a tax cut would generate a
fiscal surplus for the city: although the tax liability per property would decrease, the
direct revenue loss would be offset by ( ) an increase in the number of properties on

1
the tax rolls and (2) a decrease in the number of properties that the city must either
take over or demolish.
Because abandonment is the ultimate neighborhood externality, it contributes
to theproblems of central-city neighborhoods. Abandoned buildings provide targets
for vandals and graffiti artists, and quickly become eyesores. Even worse, they often
become the temporary homes and retail outlets for transients and drug dealers, so they
contribute to crime. For these reasons, abandonment decreases the relative attrac-
tiveness of the neighborhood, decreasing the rent that other landlords can charge for
their properties. Abandonment feeds on itself, transforming livable neighborhoods
into unlivable ones.
Durability and Supply Elasticity
The durability of housing also has important implications for the market supply
curve. Consider the response to an increase in the demand for housing. In the short
run, the supply of housing is fixed, so the increase in demand increases the equi-
librium price. In the long run, suppliers respond to the increase in market price by
increasing the quantity of housing supplied. The questions are: by how much will
supply increase, and how soon will it increase?
There are three types of supply responses to an increase in price:
1 Build new dwellings. As the price rises, new housing becomes more
profitable, so more dwellings are built. Most of the new dwellings are built
on vacant land in the suburbs.
2. Slow deterioration of used dwellings. As the price of housing increases,

the benefit of maintaining dwellings increases. Landlords spend more on
maintenance and repair, slowing the deterioration rate and decreasing the
number of dwellings retired from the housing stock.
3. Remodel used dwellings. Some landlords upgrade their dwellings,
increasing the quantity of housing services generated.
Because the bulk of housing is used housing, the supply response will be relatively
large only if the second and third responses (the used-housing responses) are rela-
tively large.
How elastic is the supply of used housing? Because dwellings deteriorate slowly
over time, a decrease in the deterioration rate has only a small effect on the housing
market. Although an increase in price slows down the deterioration process, the
process is slow in the first place. In addition, remodeling is extremely expensive,
so it takes a very large price hike to make modifications worthwhile. For these two
reasons, housing supply is inelastic for relatively long periods of time. In other
words, the supply side of the housing market is sluggish: it takes suppliers a long
time to respond to an increase in the demand for housing. In the meantime, the price
of housing remains relatively high.
The same argument applies to a decrease in housing demand and the result-
ing decrease in the market price. A decrease in price decreases the incentives for
maintenance, so dwellings deteriorate at a faster rate and more dwellings are retired.
Even the fastest deterioration rate is relatively slow, so the decrease in the quantity
supplied is relatively small for relatively long periods of time. Although dwellings
can be converted to other uses, the high cost of conversion inhibits this response.
Because a decrease in price causes relatively small changes in a large part of the
market, supply is relatively inelastic for a long period of time. Therefore, the price
of housing will be relatively low for a long period of time.
What is the price elasticity of the supply of housing? Unfortunately, economists

have been unable to answer this question. It is difficult to estimate the supply elasticity
because it measure the quantity of housing services. The existing studies
is difficult to
of housing supply suffer from a number of statistical problems (see Olsen, 1987;
and Quigley, 1979), so their results must be interpreted with caution. Ozanne and
Struyk (1978) estimate that the supply elasticity of used housing is between 0.20 and
0.30. In other words, a 10 percent increase in the market price increases the quantity
of used housing on the market by between 2 percent and 3 percent. Over a 10-year
period, new construction provides only about 30 percent of the housing stock, so
their estimate applies to 70 percent of the housing stock for a 10-year period. De
Leeuw and Ekanem (1971) estimate that the long-run supply elasticity for rental
housing is between 0.30 to 0.70. In other words, the available evidence suggests that
the supply of housing is relatively inelastic over relatively long periods of time.
The Filtering Model of the Housing Market

The filtering model captures some of the essential features of the market for used
housing. It describes the interactions between different housing submarkets and the
process through which a dwelling passes from one use to another. The filtering
process has two basic features:
1 Decrease in housing services. The quantity of housing services produced

by a particular dwelling decreases over time. The decrease in Q results
from physical deterioration, technological obsolescence, and changes in
housing fashion.
2. Decrease in occupant income. The dwelling is occupied by households
with progressively lower incomes. As Q decreases, the dwelling is occupied
by households that demand progressively lower quantities of housing
services, typically households with lower incomes.
The filtering model can be used to address two questions. First, why do the poor
occupy used housing instead of new housing? Second, do the poor benefit from
subsidy policies that encourage the building of new housing for the wealthy?
Why Do the Poor Occupy Used Housing?

Most poor households occupy used housing. To explain why the poor occupy leftover
housing, consider a city with the following characteristics:
1 Three income groups. There are three income groups: wealthy,

middle-income, and poor.
2. Rising income. Real income increases over time, increasing the desired
quantities of housing services for all three income groups.
Chapter 14 Why Is Housing Different'.' 379
Figure 14-4 Filtering: The Allocation of Housing in 1980 and 1990
S= 180 (built in 1990)
o
S= 160 (built in 1980)
S =155 (built in 1980)
= 135 (built in 1970)
= 130 (built in 1970)
built in 1960)
Size (square feet of living space)
In 1980, the poor are at point b, the middle-income households are at point d, and the wealthy are at
point /. Rising income increases the housing consumption of the rich, and they move into new housing
in 1990 (point g). The middle-income households move into the 1980 housing vacated by the wealthy
(from point d to e), and the poor move into the 1970 housing vacated by the middle-income households
(point b to c).
3. Costly upgrading. It is relatively costly to upgrade dwellings: increases in

quality and size are relatively expensive.
4. Deterioration. Houses are built every decade, and deteriorate by five units
of housing services per decade.
Figure 14^4 shows the city's housing market in 1980. The wealthy demand
160 units of housing service and occupy new houses (point/). The middle-income
households demand 135 units of service (point d), occupying houses built in 1970
(which produced 140 units of service when they were new and 135 units one decade
later). The middle-income households occupy smaller and lower-quality dwellings
than the wealthy. The poor households demand 1 1 units of service and live in houses
built in1960 (point b).
Suppose that the real incomes of all three groups increase between 1980 and
1990. The desired quantities of housing service increase to 180 for the rich, 155
for the middle-income households, and 130 for the poor. Figure 14-4 shows one
possible arrangement of the 1990 housing market. New dwellings with 180 units of
service are built for the wealthy, so they move from point/ to point g. The middle-
income households occupy the dwellings vacated by the rich, moving from point d
380 Pari III Penary and Housing
to e (the dwellings built in 1980 generate 155 units of housing service one decade
later).The poor occupy the dwellings vacated by the middle-income households,
moving from point b to c.
Why should the wealthy live in new houses? If the wealthy did not occupy
new housing, they could be accommodated by upgrading the 1980 houses. Since
a 1980 house would normally produce only 155 units of housing service in 1990.
substantial modifications would be required to increase its service level to 180 units.
Rooms would be added, the roof and plumbing would be upgraded, and the house
would be remodeled to accommodate changes in housing tastes. Since these mod-
ifications would be costly, it is usually more efficient to build new housing for the
wealthy.
If new housing is built for the wealthy, there is a plentiful supply of used hous-
ing for middle-income households. When the wealthy vacate the 1980 dwellings,
the market values of the leftover houses decrease. Because the market value of a
used 155-unit dwelling is less than the cost of building a new 155-unit dwelling, the
middle-income households live in leftover houses. As the middle-income households
vacate the 1970 dwellings, their prices fall, making the used 130-unit dwellings a
bargain for the poor. Since there is a plentiful supply of used housing (with unfash-
ionable design and old pipes), a used, low-quality dwelling is less expensive than a
new one.
Construction Subsidies
Consider next the issue of who benefits from subsidies for new housing. Suppose
that the government subsidizes the construction of new housing for middle-income
(asopposed to low-income) households. How does the subsidy program affect low-
income households'?
To explain the effects of construction subsidies on low-income households,
consider a city with the following characteristics:
1 . Fixed number of households. There are 1 20 households in the city.
2. Housing submarkets. There are two submarkets: the market for

medium-quality housing and the market for low-quality housing.
3. Quality level. to changes in relative prices by moving
Landlords respond
their dwellingsbetween the two submarkets: dwellings can be upgraded to
the medium-quality submarket or downgraded to the low-quality submarket.
4. Imperfect substitutes. The two types of housing are imperfect substitutes:

consumers move between the two markets in response to changes in market
prices.
Figure 14-5 shows the equilibrium in the housing market. In the medium-
initial
quality submarket. there are 40 dwellings at a price of $30 per square foot. In the
low-quality submarket. there are 80 dwellings at a price of S20 per square foot.
Suppose that the city subsidizes the construction of new medium-quality hous-
ing. In Figure 14-5. the subsidy shifts the supply curve to the right from 5i to
Figure 14-5 Medium-Quality Housing on

Effects of Subsidy for
Medium-Quality and Low-Quality Submarkets
A: Medium-quality submarket B: Low-quality submarket
Initial equilibrium
A / ^ \
,s 2
Initial equilibrium
30
24
^&C S,
V\~
20 40 60 80 100 20 40 60 80 100
Quantity of medium-quality dwellings Quantity of low-quality dwellings

Construction subsidies shift the supply curve for The decrease in the price of medium-quality
medium-quality (new) dwellings outward. In equilibrium, dwellings increases the supply of low-quality
the price drops from $30 to $24 and the quantity increases dwellings: dwellings filter downward and decrease
from 40 to 60. thedemand for low-quality dwellings as consumers
switch from the low-quality market to the
medium-quality market. The price decreases from
$20 to $13 and the quantity decreases from 80 to 60.
52, increasing the quantity supplied at every price: at the original price of $30,
there would be 80 dwellings instead of 40. Since the quantity supplied exceeds the
quantity demanded, the price of medium-quality housing decreases. As the price of
medium-quality housing decreases, two things happen.
1. Demand effect. The decrease in the relative price of medium-quality

housing increases the quantity demanded: some households move from
low-quality dwellings into medium-quality dwellings, causing movement
downward along the demand curve D\.
2. Supply effect. As the price of medium-quality dwellings decreases, the
quantity supplied decreases as some landlords downgrade their dwellings to
the more lucrative low-quality submarket: the market moves downward
along the supply curve S2, decreasing the quantity of medium-quality
housing supplied.
In equilibrium, the subsidy increases the quantity of medium-quality housing

and decreases its price. In Figure 14-5, the number of medium-quality dwellings
increases from 40 to 60, and the price decreases from $30 to $24.
The changes in the medium-quality submarket cause two types of changes in
the low-quality submarket.
1 . Substitution effect. As consumers of low-quality dwellings move into

medium-quality dwellings, the demand curve for low-quality dwellings
shifts to the left. This is shown in Figure 14-5 as a shift from D3 to D4.
382 Pan III Poverty and Housing
Because the two types of housing are substitutes, a decrease in the price of
medium-quality housing decreases the demand for low-quality housing.
2. Filtering effect. The downgrading of dwellings from the medium-quality
submarket to the low-quality submarket shifts the supply curve of low-
quality housing to the right. This is shown in Figure 14-5 as a shift from
S3 to S4 .
The new equilibrium occurs where the new demand curve (D 4 intersects the new )
supply curve (54 The subsidy program for medium-quality housing decreases the
).
price of low-quality housing from S20 to S13 and decreases the quantity of low-
quality dwellings from 80 to 60.
The construction subsidies also cause the retirement of 40 dwellings. Since the
population is fixed at 120 households, even new house causes one retirement. The
number of households in the medium-quality submarket increases by only 20 (from
40 to 60). leaving 20 of the 40 new houses to filter down to the low-quality market.
The number of households in the low-quality submarket decreases by 20 (from 80
to 60). so there are 40 extra houses in the low-quality submarket. Therefore, the 40
lowest-quality houses are retired from the housing stock.
Summary: Filtering and Public Policy
The benefits of housing subsidies are not confined to the households who occupy the
subsidized dwellings. The households who
occupy low-quality dwellings
initially
benefit for two reasons. First, the subsidies decrease the price of medium-quality
houses, allowing some of the households to move into medium-quality dwellings.
Second, houses filter down from the medium-quality submarket to the low-quality
submarket. decreasing the price of low-quality housing.
What about the submarket for high-quality housing? Consider a model with
three submarkets: high-quality, medium-quality, and low-quality. The subsidy in-
creases the supply and decreases the price of medium-quality housing, so it slows
the downward filtering of dwellings from the high-quality submarket to the medium-
quality submarket. The increase in the supply of high-quality dwellings decreases
the price of high-quality housing. Therefore, the occupants of high-quality housing
also benefit from subsidies for medium-quality housing.
The same analysis applies to policies that decrease the supply oi medium-quality
dwellings. Suppose that growth controls or zoning policies decrease the supply of
new medium-quality houses. The decrease in supply increases the price of medium-
quality housing, increasing the benefits of maintaining dwellings for the medium-
quality market. Landlords spend more on maintenance and repair, slowing the
filtering process and decreasing the supply of low -quality housing. At the same time,
some consumers are unwilling to pay the higher prices in the medium-quality sub-
market, so they flee to the low-quality submarket. Growth controls on new medium-
quality housing increase the price of low-quality housing because they decrease the
supply an d increase the demand for these dwellings. Consequently, some of the costs
of growth control and zonins are borne b\ low -income households.
The High Cost of Housing

When a household buys a home, it typically spends between two and four times
its annual income on the house. The large cost of housing has two implications for
the housing market. First, given the large investment required for homeownership,
a large fraction of households rent instead of owning. In the United States, about
one-third of households are renters. Second, most middle-income households use
homeownership to accumulate wealth. In other words, a large fraction of middle-
income savings is tied up in dwellings. This section explores the household's choice
between renting and owning and the effects of public policy on this choice.
The Cost of Renting

The cost of renting a dwelling is simply the rent paid to the landlord. In a competitive
market, landlords make zero economic profit (normal accounting profit), so the
annual rent equals the landlord's annual cost. In the absence of taxes, the landlord
has three types of costs:
1 Capital cost. If the landlord borrows money to purchase the dwelling, the
capital cost is the annual interest cost, equal to the market interest rate (/)
times the purchase price, or value, of the dwelling ( V). If she uses her own
money to purchase the dwelling, the capital cost is the opportunity cost of
investing her money in housing instead of a bank account earning an
interest rate of i percent.
2. Depreciation. If the market value of the house decreases by d percent per

year, the annual depreciation cost is d times V.
3. Maintenance cost. If the annual cost of routine maintenance is in percent of

the market value of the dwelling, the annual maintenance cost is in times V.
The maintenance cost includes the opportunity cost of the landlord's time.
The expression for the annual cost of a rental dwelling is
C, = V-(i r +d +m r r ) (14-1)
where ir is the interest rate paid by the landlord, dr is the depreciation rate for
rental housing, and m r is the maintenance cost. Suppose that V = $50, 000, ir =
10 percent, dr = 1 percent, and m =
r 4 percent. If so, the annual cost is $7,500, the
sum of $5,000 in capital cost, $500 in depreciation cost, and $2,000 in maintenance
cost. In a taxless world, the landlord charges an annual rent of $7,500, thus covering
her costs and making zero economic profit.
What happens if the market value of the dwelling increases over time? If the
dwelling appreciates over time instead of depreciating, dr is a negative number. For
example, if the market value increases by 8 percent per year, dr is —0.08 instead
of +0.01, and the landlord earns $4,000 per year in appreciation. What is the new
equilibrium rent? Economic profit is zero if total cost (for capital and maintenance)
equals the sum of rent and appreciation. Since total cost is $7,000, the equilibrium
rent is $3,000 ($7,000 = $4,000 in appreciation plus $3,000 in rent). The equilibrium
rent is lower because the landlord earns money by owning the property, and therefore
needs less rental income to generate zero economic profit.
The Cost of Homeownership

Consider next the cost of owning a home. In the absence of taxes, the homeowner
incurs the same types of costs as the landlord (capital, depreciation, and mainte-
nance). If the homeowner borrows money to buy his home, the capital cost is the
annual mortgage-interest payment. If he pays cash for the house, the capital cost is
the opportunity cost of investing his money in a house rather than a bank account. In
either case, the capital cost is the market interest rate (/ ) times the value of the house
(V). The homeowner also incurs depreciation cost (d times V and maintenance cost )
(m times V The maintenance cost includes the opportunity cost of time spent fixing
).
faucets, painting walls, and keeping the books. The annual cost of ownership is
C = V-(i +d +m ) (14-2)
where i is the interest rate paid by the homeowner. d is the depreciation rate
for owner-occupied housing, and m is the maintenance cost for owner-occupied
housing. If the dwelling appreciates over time. d c
, is a negative number, and the net
cost of ownership is lower.
Indifference between Renting and Owning?

Under what circumstances is a household indifferent between owning and renting?
The household is indifferent between owning and renting if C r = C„. If landlords
face the same interest rate and depreciation rate and incur the same maintenance
cost, the annual cost of renting equals the annual cost of homeownership. The cost
of renting is often higher for three reasons.
The renter externality causes the depreciation rate on rental property to be
relatively high. Because the renter has no long-term stake in the dwelling, he has
little incentive to participate in everyday maintenance. As long as the tenant is not
inconvenienced by small problems (e.g.. a leaky pipe), he is unlikely to report the

problem to the landlord. Although the landlord would benefit from finding out about
the problem (she could prevent the small problem from becoming a big one), the
tenant does not have the proper incentive to inform the landowner. The externality
occurs because the tenant does not directly benefit from better care of the property. In
contrast, the homeowner receives all the benefits from everyday maintenance, so he
is more likely to fix the leaky pipe immediately to prevent long-term damage to the
dwelling. If tenants do not keep their landlords fully informed about the condition of
the house, rental property has a higher maintenance cost and a higher depreciation
rate. In other words, the renter externality increases the market rent on residential
property.
The cost of renting is relatively high for two other reasons. First, rental prop-
erty is a riskier investment (it has a higher default rate), so lenders charge higher
interest rates on loans for Second, some homeowners use work on

rental property.
the house (maintenance, repair, and yard work) as a hobby, so they have a lower
maintenance cost than a landlord. Homeowners often use work on the house as a
means of accumulating wealth in their spare time. For these reasons, the annual cost
of homeownership is often less than the annual rent on an equivalent rental dwelling,
even without considering the tax advantages of homeownership.
What type of household rents instead of owning? Three factors tend to push
some households toward renting. The first is low income: to qualify for a home
loan, the household must have enough income to pay the mortgage. The second is
mobility: given the substantial transactions cost of executing a home loan, frequent
movers tend to rent rather than own. The third is a distaste for work on the home and
yard: if a person is unwilling to do everyday maintenance and small repairs, renting
is relatively attractive.
Taxes and Housing
The federal government provides a number of tax breaks for both rental and owner-
occupied property. The tax breaks for rental property decrease the landlord's costs,
and these savings are passed on to consumers in the form of lower rent. The tax
breaks for homeowners are more explicit: homeowners deduct interest costs from
their gross income, so they pay lower taxes. What are the net effects of these tax
breaks? Do they encourage or discourage homeownership?
Rental Tax Break: Rapid Depreciation. The tax code allows the landlord to
deduct the full value of his property before the property actually wears out. In
other words, the tax code allows the rapid depreciation of property (also called
accelerated depreciation). Suppose that the landlord builds a dwelling at a cost of
$60,000. If the dwelling has a useful life of 60 years and deteriorates at a constant
rate, the economic cost of depreciation is $1,000 per year. Under a system of rapid
depreciation, the dwelling may have a "tax life" of only 15 years, meaning that the
deduction for depreciation is $4,000 per year for the first quarter of the dwelling's
life, and zero for the rest of its life.
Rapid depreciation does not change the total deductions for depreciation, but
simply allows landlords to deduct depreciation cost sooner, and thus pay taxes later.
If rapid depreciation defers a $100 tax bill for eight years, the landlord can place the
$ 1 00 in a bank and earn interest for eight years. When the tax comes due, the landlord
pays the tax bill with the $100 principal and keeps the interest. The deferral of taxes
decreases the net tax liability because the landlord can earn interest on the deferred
taxes. In other words, rapid depreciation decreases the present value of the landlord's
tax liability. In a competitive environment, all landlords make zero economic profits,
so the decrease in taxes is passed on to renters in the form of lower rent.
Ownership Tax Break: Mortgage Deduction. The federal tax code also de-
creases the cost of homeownership. Taxpayers can deduct mortgage-interest pay-
ments from their gross income, so every dollar of mortgage interest decreases the
Table 14-2 Tax Benefits of Homeownership and Income
Marginal Mortgage Tax Benefit as Percent

Household Income Tax Rate Interest Benefit of Income
A $20,000 Nonitemizer $ 5.000

B 30.000 1595 7.000 $1,050 3.50
C 50.000 28% ! 1 .000 3.080 6.16
federal tax liability by the taxpayer's marginal tax rate. For example, if the taxpayer's
marginal tax rate 28 percent, every dollar spent on mortgage interest decreases
is
the tax liability by 28 cents. Table 14-2 shows the tax benefits of homeownership
for three households. Household A, the poorest, does not have enough deductions
to justify itemizingits deductions, so it receives no benefit from the mortgage de-
duction. Household B. an itemizer with a marginal tax rate of 15 percent, receives a

15-cent subsidy, and household C. with a marginal tax rate of 28 percent, receives a
subsidy of 28 cents.
The tax benefit increases with household income for two reasons. First, under
a progressive tax system, the marginal tax rate increases with income, so the tax
benefit per dollar of mortgage payment increases with income. Second, because
the demand for housing increases with income, wealthier households have larger
mortgage payments and thus larger deductions. In 1998. the mortgage deduction
decreased federal tax revenue by about $60 billion.
The mortgage deduction increases housing consumption because it decreases
the net cost of housing. Figure14-6 shows the effect of the subsidy on the housing
consumption of a household with a 28 percent marginal tax rate. The demand curve
shows the marginal benefit of housing consumption (the willingness to pay for
housing). The horizontal line at $1 shows the marginal social cost or opportunity
cost of housing: by spending $1 on housing, society forgoes $1 worth of investment
in other capital goods (e.g.. factories, machines, schools). The optimum housing
consumption is H*. the amount such that the marginal social benefit equals the
opportunity cost. For consumption above H*, the marginal benefit is less than the
opportunity cost: the money would be more efficiently spent on factories, machines,
or schools.
The mortgage deduction causes inefficiency because it creates a gap between
the marginal private cost and the marginal social cost of housing. In the absence of the
subsidy, the household pays $1 per unit of housing, and chooses the point where the
demand curve intersects the marginal social-cost curve (point e). When the private
cost drops from $ 1 to 72 cents, the household increases its housing consumption from
H* to H' This is inefficient because the household's benefit from increased housing
.
consumption (from the demand curve) is less than the opportunity cost of spending
on housing (from the marginal social-cost curve). By spending more on housing,
society has less to spend on factories, machines, and schools. The household makes
an inefficient choice because the government pays 28 percent of its housing bill.
Figure 14-6 Mortgage Subsidy Increases Housing Consumption
Demand = marginal social

benefit
Marginal social cost

1.00
0.72 Marginal private cost

with 28% tax benefit
Housing consumption
The optimum housing consumption occurs at point e. where the marginal social benefit (shown by the
demand curve) equals the marginal social cost. If the marginal tax rate is 28 percent, the mortgage subsidy
decreases the marginal private cost of housing from $1 to 72 cents, increasing housing consumption from
H* to H' Consumers pay less than the marginal social cost of housing, so they consume too much housing
.
and divert investment from other uses.
Struyk, Mayer, and Tuccillo (1983) suggest that the mortgage deduction has
contributed to recent declines in U.S. productivity. The idea is that the deduction
encourages the investment in expense of industrial investment, so
housing at the
there is less capital equipment (factories and machines) per worker. Rosen (1979)
estimated the effect of the mortgage deduction on housing consumption in the 1970s,
before the 1981 and 1986 tax changes. He estimated that the deduction increased
housing consumption by about 14 percent, suggesting that a substantial amount of
capital is diverted to residential uses.
Mortgage Deduction and Imputed Rental Income. Under the federal tax code,
both landowners and homeowners can deduct their interest cost in computing their
taxable incomes. Does this mean that the mortgage deduction for homeowners is
neutral with respect to the consumer's choice between owning and renting?
To explain the bias toward ownership created by the mortgage deduction, con-
sider Fred the homeowner and Barney the landlord. All dwellings in city B are
two housing options. First, he can
identical rock hovels that sell for $2,000. Fred has
borrow $2,000 from the bank to purchase a home. If the interest rate is 10 percent,
the mortgage-interest payment for a perpetual loan is $200 per year, so the cost of
388 Pari Ill Poverty and Housing
ownership in the absence of tax benefits is $200. Second, he can rent an identical
dwelling from Barney. Since the houses are made of stone, there are no mainte-
nance or repair costs, so Barney's only expense is the capital cost of $200 per year.
The market rent ($200) is the same as the cost of ownership, so Fred is indifferent
between owning and renting.
The mortgage deduction creates a bias toward homeownership because it allows
Fred the homeowner to deduct expenses that do not generate taxable income. Con-
trast the mortgage deduction for Fred (homeowner) with the mortgage deduction for
Barney (landlord). Barney's deduction is sensible because Barney declares his rental
income as taxable income. He declares $200 of rental income per house, and deducts
$200 in mortgage cost. His taxable income from his property is zero because his de-
duction equals his rental income. In contrast, Fred does not declare any rental income
from his property, but still deducts his mortgage cost. Fred's mortgage deduction
does not offset rental income, but simply subsidizes the cost of homeownership.
The government could eliminate the tax bias toward homeownership in one of
two ways. The simple and obvious one is to eliminate the interest deduction for
homeowners. An alternative is to force Fred to declare his imputed rental income
as taxable income. The imputed rental income is the income earned from owning
a dwelling and renting it to yourself. Alternatively, it is the money you could earn
if you rented your dwelling to someone else. Fred's imputed rental income is $200
per year. If he were to declare $200 of imputed rental income and then deduct his
$200 mortgage cost, the two items would cancel one another, and the ownership bias
would disappear. His taxable income would be the same whether he rents or owns,
so he would be indifferent between renting and owning.
Rationale for the Favorable Treatment of Homeownership. Because the mort-

gage deduction favors ownership over renting, it increases homeownership rates.
In 1980, 66 percent of U.S. households were owner-occupants, up from 48 per-
cent in 1945. According to Hendershott and Shilling (1982), about one-quarter of
the increase in ownership rate was caused by the favorable tax treatment of home-
ownership.
The principal rationale for the tax breaks forhomeownership is the renter ex-
ternality. If renters do not have the proper incentive to participate in everyday main-
tenance, a subsidy for homeownership may be justified on efficiency grounds: the
subsidy discourages renting, decreasing the inefficiencies resulting from the renter
externality. This rationale raisestwo questions. First, how large is the renter exter-
nality,and how large must the ownership subsidy be to counteract the externality.'
Second, what is the appropriate form for the ownership subsidy? Does the present
system, under which the wealthiest households receive the largest subsidies, inter-
nalize the externality in the most efficient manner?
There have been a number of proposals to modify the mortgage deduction. One
option is to replace the deduction with a tax credit equal to a fixed percentage of
the household's mortgage cost. For example, if the tax credit is 15 percent, every
household would receive a tax cut equal to 1 5 percent of its mortgage cost. Under a tax
credit, all taxpayers would receive the same percentage subsidy for homeownership.
A second option is to place a ceiling on the amount of mortgage interest that can be
deducted. This would decrease the subsidy to the wealthiest households.
The Net Tax Code: Renting versus Owning. What is the net
Effects of the
code on the choice between renting and owning? For renters,
effect of the federal tax
rapid depreciation decreases the cost of supplying rental housing, decreasing market
rents. For owner-occupants, the mortgage deduction decreases the cost of ownership.
Does the tax code decrease the cost of renting by more than it decreases the cost of
ownership?
The net effect of the tax code depends on a number of factors. One of the factors
is the generosity of depreciation allowances: the more rapidly rental property can
be depreciated, the larger the tax benefits and the lower the market rent on rental
property. Two studies in the early 1980s suggested that the tax code decreased the
relative cost of renting, causing a bias toward rental housing (King and Fullerton,
1984; Gordon, Hines, and Summers, 1986). Since 1986, however, depreciation rules
have become less generous. For example, the tax life of rental property increased
from 1 5 years to 27.5 years; the allowable depreciation rate dropped from an average
of 6.7 percent per year to about 3.6 percent. As a result, the tax liabilities of rental
property have increased, increasing market rents and decreasing the bias toward
rental housing.
Inflation and Homeownership

How does inflation affect the costs and benefits of homeownership on a household
that has a fixed-rate mortgage? Consider first the effect of unanticipated inflation.
Suppose from 4 percent to 10 percent. The increase in
that the inflation rate increases
inflation increases the appreciation rate of the property from 4 percent to 10 percent,
so the benefits of homeownership increase. For a household holding a fixed-rate
mortgage, the increase in the inflation rate does not affect the monthly mortgage
payment, so it does not affect the costs of ownership. Consider next the effects of
anticipated inflation. If inflation is anticipated, mortgage lenders build the expected
inflation rate into the mortgage interest rate. For example, if the expected inflation
rate increases from 4 percent to 10 percent, the interest rate might increase from
8 percent to 14 percent. In this case, the home purchaser pays a higher mortgage
interest rate to get the higher appreciation rate.
An alternative to a fixed-rate mortgage is one where the mortgage interest rate is
adjusted periodically to reflect changes in market interest rates. Under an adjustable-
rate mortgage an increase in the inflation rate increases the monthly mortgage pay-
ment because the increase in inflation increases the mortgage interest rate. As a
result, unanticipated inflation increases both the benefits and costs of ownership.
Large Moving Cost

For most households, a change in housing consumption requires a move to a different
dwelling. Moving cost is substantial: in addition to the large cost of moving furniture,
clothes, and kitchen equipment, there is also a large cost associated with leaving
—
behind the old neighborhood with its people, schools, and stores. In other words,
a move to a new neighborhood disrupts social and consumption patterns, imposing
a substantial cost on the household. A household changes its housing consumption
only if the benefits of moving (the opportunity to live in a dwelling with a better
combination of housing services and cost) exceeds the cost of moving.

Figure 4-7 shows the choices faced by a household that experiences an increase
1
in income. The household starts with income 1\ and chooses point b (the point at
which its indifference curve is tangent to its initial budget line An increase in income ).
to h shifts the budget line upward, giving the household two choices. One option
is to stay in its original house and spend all the extra income on nonhousing goods
(point c). Alternatively, the household could move to a larger house (S2 square
feet instead of Si square feet), choosing point d. In the absence of moving cost.
Figure 14-7 Moving Cost and Responses to Increases in Income
Budget line with income /-,
Budget line with income l

:
Budeet line with income /,
Housing consumption (square feet)
\n increase in income siutts the budget line upward. In the absence of moving cost, the household would
move from point b to point </. moving to a largei house i\- square feet ol living space). If moving is
household m.i\v in its pa-sent house, choosing
costly, the instead ot 1/. It income increases again and
(
the household does not move, H chooses point The household moves to a larger duelling (point
i it I I
the resulting increase in utility is large enough to offset mo\mg eost

Chapter 14 Why Is Housing Different:' 391
utility would be higher at point d (point d

on a higher indifference curve), but
lies
the question is whether the increase enough to justify the cost of

in utility is large
moving. If the moving cost is relatively high, the household stays in its original house
and chooses point c. If income increases to /3, the budget line shifts upward again,
and the household faces the same choice: stay put (point e), or pay the moving cost
and move to a larger house (point/). Because the gap between the initial housing
consumption (S\) and the desired housing consumption (53) is relatively large, the
household is more likely to move than it was when the gap was relatively small
(S2-S1).
There are two lessons from Figure 14-7. First, a small change in income (or
price) is unlikely to change housing consumption. Most households tolerate some
dissatisfaction with their housing circumstances because moving is costly. A house-

hold moves to a different dwelling only if the change in income or price is large
relative to moving cost. Second, when a household changes houses, it is likely to
make a large change in its housing consumption. A household moves when its dis-
satisfaction with its current dwelling is large enough to justify the cost of moving.
Income and Price Elasticities

The demand for housing depends on household income and the price of housing. A
number of studies have estimated the responsiveness of housing demand to changes
in income and price.
Permanent Income and Income Elasticity
Because of large moving cost, households base their consumption choices on per-
manent, or long-run, income. Given the cost of moving, the household makes a
long-term commitment to living in a particular dwelling. The commitment is based
on the household's expected income over several years. Therefore, to estimate the
income elasticity of demand for housing (the percentage change in housing con-
sumption divided by the percentage change in income), one must use a measure of
the permanent (long-run) income of the household, not its income in any particular
year.
To explain the proper estimation of income elasticity, suppose that Lucky has a
steady wage income of $20,000. In 1991, she wins $10,000 at a bingo game. As a
result of her windfall, she increases her housing consumption by 5 percent, moving
to a slightly larger rental apartment. If her income elasticity is computed using her
income in 1990 and 1991, the elasticity would be computed as 0.10 (a 5 percent
increase in housing consumption divided by a 50 percent increase in income). To
correctly estimate the income elasticity, one must use a measure of her permanent
income, defined as her average income over some period of time. One approach
is to translate the windfall into an equivalent annual income. If Lucky invests her
$ 1 0,000 prize in a bank account yielding percent interest, her annual income from
1
392 Part Ill Poverty and Housing
the prize would be $1,000, meaning that the bingo prize increases her permanent
income by 5 percent. The income elasticity would then be computed as 1.0 (the
5 percent increase in housing consumption divided by the 5 percent increase in
permanent income).
What is the income elasticity of demand for housing? There have been dozens
of studies of housing demand, and there is a consensus on three points. First, the
overall income elasticity is about 0.75 (Ellwood and Polinski, 1979): a 10 percent
increase in income increases housing consumption by about 7.5 percent. Second, the
income elasticity for renters is less than the income elasticity for owner-occupants.
Third, the income elasticity increases with income. According to Ihlanfeldt 1982), (
the elasticity for low-income households is between 0. 14 and 0.62, and the elasticity
for high-income households is between 0.72 and 1.10.
The results of hedonic studies can be used to estimate the income elasticities
of demand for individual components of the housing bundle. Follain and Jiminez
(1985) summarize the results from several hedonic studies and come to three con-
clusions:
1. The demand for living space is inelastic with respect to income: in seven of
nine cases, the elasticity is below 0.46.
2. The demand for structural quality is highly elastic with respect to income:
in four of five cases, the elasticity exceeds 1 .64.
3. The demand for neighborhood amenities such as public safety is highly

elastic with respect to income.
The hedonic studies suggest that, compared to the demand for housing in general,
the demand for living space is less income-elastic and the demands for structural
quality and neighborhood amenities are more income-elastic.
Price Elasticities
What is the price elasticity of demand for housing? Most estimates of the price elas-
ticity fall between —0.75 and — 1 .20 (Ellwood and Polinski, 1979). The consensus is
that demand is slightly price-inelastic (an elasticity slightly less than 1 .0 in absolute
value). This means that an increase in price increases total expenditures on housing
by a small amount: an increase in price decreases the quantity demanded by a slightly
smaller percentage amount, so total expenditure (price times quantity) increases by
a small amount.
Racial Discrimination and Segregation

This section explores the issue of racial prejudice and discrimination in the urban
housing market. We discussed the facts o\' residential segregation in Chapter 12. As
shown by the index of dissimilarity, it is clear that U.S. cities are highly segregated
with respect to race. The reasons for segregation include racial attitudes, disci imi-
nation in the real estate market, exclusionary zoning, and past housing policies. In
this chapter, we explore the effects of discrimination on (1) housing prices paid by
blacks and whites and (2) the transformation of white neighborhoods into mixed or
black neighborhoods.
Yinger (1979) provides a useful set of definitions of prejudice, discrimination,
and segregation. Prejudice is defined as a negative attitude toward members of a
particular racial group. For the purposes of this chapter, prejudice is reflected as an
aversion to living near members of another race. Racial discrimination is defined as
behavior that results in differential treatment based on race: members of a group sub-
ject to racial discrimination are denied rights or opportunities because of their race,
not because they lack the formal qualifications for those rights or opportunities. The
degree of residential segregation indicates the racial mixture of a city's neighbor-
hoods. A city is completely segregated when racial groups are divided into homoge-
neous neighborhoods, so that no neighborhood contains members of more than one
racial group. A city is completely integrated when each neighborhood is heteroge-
neous, with proportional representation of each racial group in each neighborhood.
Housing Prices with Fixed Populations

This section explores the effects of racial prejudice and discrimination on housing
prices. The principal question is: how does the price of housing vary within a segre-
gated city? The conclusion is that in a city with a fixed black population, blacks pay
less than whites for identical dwellings.
To explain the effects of prejudice and discrimination on the price of housing,
consider a mixed-race city with the following characteristics:
1 There is a fixed number of households of each race.
2. All households have the same income.

3. All dwellings in the city are identical.
4. All dwellings are equally accessible to employment opportunities.
5. Blacks have a moderate preference for living near white households.

6. Whites have a strong aversion to living near black households.
Suppose that the city starts with a segregated housing market: blacks live close to the
city center and whites live in a suburban ring. The question what housing-price
is,
function is consistent with racial segregation? In other words, what set of housing
prices will sustain the segregated housing pattern?
Figure 14-8 shows a housing-price function that is consistent with racial seg-
regation. The border between and the white area is u*. The solid line
the black area
shows the price function under which households have no incentive to relocate. In
the black area (w < u*), the price function is positively sloped, reflecting blacks'
preferences for proximity to whites. Black households are willing to pay $48,000
for a dwelling in the city center, compared to $60,000 for a dwelling at the border
between the two areas. In the white area (w > w*), the positively sloped price func-
tion reflects whites' aversion to living close to blacks. White households are willing
Figure 14-8 Equilibrium Housing-Price Function in

Segregated City
In a segregated citj w here both blacks and whites

prefer to live near whites, the
housing-price function is positively sloped. Blacks living at u < u' are willing to
pa\ more for duellings closer to the white area, and whites living beyond u' are
willing to paj les^ for dwellings close to the black area.
to pay $60,000 for a dwelling compared to $84,000 for a dwelling at

at the border,
the location most distant The black price function is flatter than
from the black area.
the white price function because blacks have relatively weak preferences for living
near whites.
Figure 14-8 suggests that blacks pay less than whites for identical dwellings.
All dwellings are assumed to be identical in terms of their physical characteristics
and accessibility, yet the equilibrium price increases with the distance from the city
center.The average black household (located in the center of the black area) pays
$54,000. compared to $72,000 for the average white household ( located in the center
of the white area >. In this case, whites pay for their aversion to living near blacks in
the form of higher housing prices.

As pointed out by Yinger 1979). there are two key assumptions underlying the
(
result that blacks and whites are completely segregated. First, the preference of blacks
for white neighbors is weaker than the preference of whites for white neighbors.
This means that whites outbid blacks for all dwellings beyond u*. In contrast, if
some blacks had relatively strong preferences for white neighbors, they could outbid
\\ bites for dwellings beyond it
'
and thus disrupt the segregated pattern. In this case.
complete segregation is not a market equilibrium. The second key assumption is
Chapter 14 Why Is Housing Different.' 395
that all households have the same income. In combination with the assumption that
blacks have relatively weak preferences for living close to whites, the assumption of
equal incomes means that whites outbid blacks for all dwellings beyond u* . If this
assumption were dropped, wealthy black households could outbid whites for some
dwellings beyond u*, so complete segregation would not be a market equilibrium.
Population Growth and Broker Discrimination

Figure 14-8 predicts that blacks pay less than whites for identical dwellings. This
prediction is inconsistent with empirical studies from the 1970s, which suggest
that blacks paid more — not less —than whites. The apparent inconsistency can be
explained by the rapid growth of urban black populations during the 1960s and
1970s. In other words, the first assumption of the model underlying Figure 14-8
(a fixed number of households), was not satisfied during this period.
Consider a city that experiences rapid growth in its black population. In the
absence of restrictions on the size of the black area, we would expect the black area to
expand outward to accommodate the larger black population. As the black population
increases, competition for housing increases the amount blacks pay
are willing to
for housing near the old border («*). If black households are allowed to compete
with whites for housing near u*, blacks will outbid white households, causing the
black area to expand outward.
If whites are averse to living near blacks, they have an incentive to prevent the
expansion of the black area. White households can turn their prejudicial attitudes
into discriminatory behavior by refusing to deal with real estate brokers who sell
dwellings to blacks. A broker who sells to blacks will have fewer white customers
and may make less money. If the penalty for selling to blacks (fewer white customers)
is large enough, brokers will engage in discriminatory behavior that promotes seg-
regation and prevents the expansion of the black area.

Brokers use a number of techniques to discourage black buyers. The first tech-
nique is called racial steering: the broker directs individual buyers to neighborhoods
full of households of the same race. In other words, black buyers are not given in-
formation about dwellings in predominantly white neighborhoods. Second, brokers

provide inferior service to black households by (1) misrepresenting the housing
market, (2) delaying transactions, and (3) giving less help in arranging financing.
A fair-housing audit provides information on the extent of racial discrimi-
nation. The auditor sends two individuals who are identical except for race (they
have the same income, education, household size) to real estate brokers to inquire
about housing. According to Yinger (1998), there is conclusive evidence that blacks
are treated differently from otherwise identical whites: blacks are shown fewer
dwellings, steered into certain neighborhoods, and given less advice and assistance
on financing options.
Figure 14-9 shows the effects of black population growth when the border
between the black area and the white area is fixed. Population growth shifts the
black portion of the housing-price function upward. If racial discrimination on the
part of brokers prevents the expansion of the black area, the typical black household
Figure 14-9 Black and White Housing Prices with

Black Population Growth
White price
In a city with a growing black population and a fixed black area, the typical
black (at u pays more than the typical white (at u
) tor housing.
I
( at the midpoint u pays $78,000 for a dwelling, compared to $72,000 for the typical
' )
white (at it"). These figures are roughly consistent with empirical results from the
1970s that showed blacks paying higher housing prices. Unfortunately, there have
been no careful studies of the black-white price differential since the late 1970s, so
it's unclear whether blacks continue to pay more for housing.
Neighborhood Transition
Consider a neighborhood that is changing from a segregated white neighborhood to
a mixed-race neighborhood. This section discusses the factors that determine how
rapidly the neighborhood changes.
An important factor in the speed of neighborhood transition is the attitude to-
ward integration. A white household that prefers a segregated neighborhood is likely
to leave the neighborhood as its minority population grows. According to household
surveys, the number of departing whites increases with the minority's share of the
neighborhood population. According to one survey, the percentage of whites who
say that they will leave increases from 7 percent in a neighborhood with 7 percent
minorities, to 24 percent in a neighborhood with 20 percent minorities, to 41 percent
in a neighborhood with 33 percent minorities, to 64 percent in a neighborhood with
60 percent minorities. These numbers explain why many neighborhoods change

rapidly from highly segregated white neighborhoods to highly segregated minority
neighborhoods. An increase in the share of minority households decreases the at-
tractiveness of the neighborhood to many white households, causing white flight and
further increases in the minority share.
Galster (1990) studied the racial transformation of neighborhoods in the Cleve-
land metropolitan area during the 1970s. He explored the factors that determine the
rate at which the black share of a neighborhood's population increases over time.
The most important factors are (1) the initial black share of population, (2) white
attitudes toward segregation, and (3) the proximity of the neighborhood to other
neighborhoods where blacks are in the majority.
Figure 14-10 shows Galster's results for three types of neighborhoods. The
curves show the gains in the black share of population (in percentage points) over the
1970-1980 decade for different initial black shares. Curve A shows this relationship
for a neighborhood that ( 1 ) is not adjacent to a black-majority neighborhood and (2)
has "average" white preferences for segregation. For example, a neighborhood that
starts the decade as 10 percent black gains 5.5 percentage points over the decade, so
the neighborhood is 15.5 percent black at the end of the decade. Along the positively
Figure 14-10 Changes in the Racial Composition of Cleveland

Neighborhoods: 1970-1980
40
Strong segregation preferences and

Affirmative marketing
adjacent to a black-majority area
e 30
20 -
Id
30 37
Percent black in 1970
Source: George C. Galster, "Neighborhood Racial Change, Segregationist Sentiments, and Affirmative
Marketing Policies." Journal of Urban Economics 27 (1990), pp. 344-61
sloped portion of the curve, the gain in the black share increases with the initial
black share. For example, 20 percent, the gain is 9.5 percentage

if the initial share is
points, so the 1980 share is 29.5 percent. For the type of neighborhood shown by
curve A, there is a black majority by the end of the decade if the initial black share
exceeds about 37 percent (point X).
Curve C in Figure 14-10 shows the changes in the black share of population
for a neighborhood that is adjacent to a black-majority neighborhood and (2)
( 1 )
has relatively strong white preferences for segregation. For a given initial share,
this neighborhood experiences more white flight than the average neighborhood,
so the gain in the black share is relatively large. For example, for an initial share
of 10 percent, the gain is 20 percentage points, so the black share at the end of
the decade is 30 percent. This type of neighborhood is more likely to have a black
majority by the end of the decade than the average neighborhood: for any initial
share greater than 22 percent (point Z ). the black gain is large enough to generate a
black majority.
During the 1970s, two municipalities in the Cleveland area (Shaker Heights
and Cleveland Heights) used public policy to encourage racially balanced neighbor-
hoods. The idea was to encourage blacks to move into white neighborhoods, but to
prevent the development of black-majority neighborhoods. There were a number of
features of this "affirmative marketing" policy, including ( 1 ) information and per-
suasion campaigns to attract blacks. (2) policies to maintain the quality of housing
and public services (especially education) to discourage the departure of households
who place a high value on the quality of housing and public services, and (3) spe-
cial broker services for members of the "underrepresented" racial group, typically
whites.To prevent the white flight that might drop the number of white households
below 50 percent, the municipalities catered to white households.
The third curve in Figure 14-10 shows the gain in the black share for a neigh-
borhood that ( 1 ) engaged in affirmative marketing and (2) is not adjacent to a black-
majority neighborhood. For a neighborhood with a small initial black share, the
black gain in this type of neighborhood exceeds the gain experienced by the average
neighborhood (shown by curve A). For example, for an initial share of 10 percent,
the black gain in the affirmative marketing neighborhood is 24.5 percentage points,
compared to 5.5 points in the average neighborhood. In contrast, for a large initial
black share, the affirmative marketing neighborhood experiences a relatively small

gain. For example, for an initial share of 28 percent, the black gain is 3 percentage
points, compared to 12 points in the average neighborhood. In the neighborhood
with affirmative marketing, the black share of the population is stable if the initial
share is about 30 percent.
The success of affirmative marketing depends on whether or not the neighbor-
hood is adjacent to a black-majority neighborhood. If it is not, the black share o\'
population by the end of the decade never exceeds 30 percent, so the policy achieves
its objective of a racially balanced neighborhood. In contrast, the polic) is ineffec-
neighborhoods that are adjacent to black-majority neighborhoods:
tive in if such a
neighborhood starts the decade with a black share of at least 12 percent, it will have
a black majority by the end of the decade.
Figure 14-11 Threshold Black Shares for Different Types

of Neighborhoods
JU
46
40 37
33
30
22
20
10
Mild Average Average Strong

preferences preferences preferences preferences
Not adjacent to Adjacent to black

black majority majority
SOURCE: George C. Galster,"Neighborhood Racial Change, Segregationist Sentiments, and

Affirmative Marketing Policies," Journal of Urban Economics 27 (1990), pp. 344-61.
Figure 14-1 1 summarizes Galster's results for neighborhoods that did not adopt
affirmative marketing policies. For each type of neighborhood, the figure shows the
minimum initial black share that generates a black majority by the end of the decade.
The threshold share
is larger for neighborhoods that are distant from black-majority
neighborhoods and have relatively weak white preferences for segregation. The
smallest threshold (22 percent) is for a neighborhood that is adjacent to a black-
majority area and has strong white preferences for segregation.
Summary
1 Housing is from other commodities because it is heterogeneous,
different
immobile, and durable. Housing is costly in two senses: it is a large part of the
household budget, and changes in housing consumption are costly.

2. Housing is heterogeneous in the sense that every dwelling provides a different
bundle of features. Dwellings differ in size, layout, the quality of utility
systems, and interior design. Sites differ in access, tax liabilities, local public
goods, environmental quality, and neighborhood appearance.
a. The hedonic approach is based on the notion that a dwelling is composed of
a bundle of individual components, each of which has an implicit price.
b. Housing consumers choose the dwelling that provides the best combination
of features at the best price, trading off quality, size, location, and other
attributes.
c. The heterogeneity of the housing stock means that a city's housing market
is segmented with respect to size, location, and quality: the market is
composed of a number of submarkets.
d. Positive changes in the exterior appearance of one house cause spillover
benefits (increases in market value) for surrounding houses (the
neighborhood effect).
3. Housing is durable, and the landlord can control the rate of physical
deterioration by spending money on repair and maintenance.
a. Spending on maintenance and repair determines the quantity of housing
services generated by the dwelling. At the optimum quantity, the marginal
benefit (the change in total revenue) equals marginal cost (the change in
total cost).
b. When the profits from housing are negative, dwellings are retired from the
housing stock. There are three types of retirement: conversion to another
use. boarding up. and abandonment.
c. Retirement results from decreases in market price (from a decrease in
demand or an increase in supply) or an increase in maintenance cost.

d. A recent study suggests that the property tax encourages abandonment.
4. The supply of housing is relatively inelastic for relatively long periods of time
because the bulk of the housing stock is used housing. There are no reliable
estimates of the price elasticity of supply.
5. According to the filtering model, the quantity of housing services generated by
a dwelling decreases over time, so the dwelling is occupied by households
with progressively lower incomes. The filtering model explainswhy the poor
occupy used housing and shows that the poor benefit from the subsidization of
new housing for the middle-income households.
6. The costs of rental housing and owner-occupied housing are determined by the
cost of capital (the opportunity cost of money invested in housing),
depreciation, and maintenance cost. Three factors push many households
toward homeownership: the renter externality (tenants have little incentive to
inform landlords of maintenance problems), higher interest rates for rental
property loans, and the hobby aspect of repair and maintenance.
7. The federal tax code has special tax breaks for rental and owner-occupied
housing.
a. Rapid depreciation of rental property defers taxes, increasing the
profitability of rental property and decreasing monthly rent.
b. The deductibility of mortgage interest decreases the cost of homeownership,
increasing ownership rates and increasing housing consumption.
c. The code on homeownership depends on marginal tax
net effect of the tax
rates and the generosity of depreciation allowances, both of which have
changed over time.
8. The cost of moving is relatively large, so households change their bousing
consumption infrequently and make large changes when they move.
9. The income elasticity of demand for housing is about 0.75, and the price
elasticity is between —0.75 and —1.20.
10. Racial prejudice and discrimination affect the housing prices paid by blacks
and whites.
a. In a city with a fixed black population, blacks pay less than whites for
identical dwellings.
b. If discrimination prevents the expansion of the black area when the black
population grows, blacks pay more — not less —than whites for identical
dwellings. A number of empirical results from the 1970s showed that
blacks paid higher housing prices.
1 1 A study of the racial transformation of neighborhoods in the Cleveland
metropolitan area during the 1970s suggests that the most rapid transformation
occurs in neighborhoods that (a) start with a relatively large black share of
population, (b) have white households with strong preferences for living with
other whites, and (c) are close to neighborhoods where blacks are in the
majority.
12. An affirmative marketing program is more likely to result in a racially

balanced neighborhood if the neighborhood is not close to a black-majority
neighborhood.

1 Consider an apartment building that does not generate enough rental revenue
to cover the landlord's costs.
a. Use a graph to show this situation (see Figure 14-2).
b. Suppose government provides a 50 percent subsidy for operating,
that the
maintenance, and repair costs. Show the effects of the subsidy program on
your graph. What happens to the optimum Q and the likelihood of
retirement?
2. Suppose that you own an apartment building that does not generate enough
rental revenue to cover your costs. You have two options: first, you could
demolish the building and set up a parking lot; second, you could abandon the
building.Suppose that the cost of demolition is $20,000 and the annual profit
from the parking lot would be $1,500. Should you demolish or abandon? If
you don't have enough information to answer this question, list the
information you need and explain how you would use it.
3. Dwellings are eventually retired from the housing stock as a result of either
decreases in market rent (adownward tilt of the total-revenue curve in
Figure 14-2) or rising maintenance cost (an upward tilt of the total-cost curve).
a. Use a graph like Figure 14-2 to prove or disprove the following statement:
"A dwelling that leaves the market as a result of decreases in market rent
leaves the market at a higher quantity of housing services (a larger Q) than

one that leaves as a result of increases in maintenance cost."
b. Explain why this statement is true or false.
4. In city E, all households have the same income and real income is constant
over time. Will there be filtering in the city?
5. Consider city B, which has a fixed population. In an attempt to decrease total

residential energy consumption, the city has adopted policy E: all new housing
in the citymust be equipped with an energy-saving device that decreases the
dwelling's energy consumption by 10 percent.
a. Evaluate the effects of policy E on the city's new and used housing markets.
What happens to the prices of new and used housing?

b. What happens to the quantities of new and used housing?
c. Under what circumstances will total energy consumption in the city
increase?
d. Describe and analyze a more effective conservation policy.
6. The F are baffled. Although dwellings are subject to
real estate brokers in city
the normal physical and fashion deterioration over time, there is little filtering.
In contrast to most cities, in which the typical house changes hands every 7
years (usually going to a household with a lower income than the previous
owner), the typical house in the city changes hands every 50 years. Explain
this phenomenon.
7. Fill in the blanks to make the following statement correct and then explain
your choice of words. The benefits of a subsidy for medium-quality housing
are transmitted up the housing-quality hierarchy because it the
filtering process and down the housing-quality hierarchy because it
the filtering process.
8. Compute the annual cost of renting and homeownership under the following
set of assumptions:
/'.
The market interest rate is 12 percent for rental property and 10 percent
for owner-occupied property.
ii. The market value of the typical house is $100,000.
Hi. Maintenance cost is 5 percent per year on rental property and 2 percent
per year on owner-occupied property.
iv. The marginal tax rate of the typical household is 28 percent.
v. Both types of properties are appreciating by 4 percent per year.
vi. Rapid depreciation decreases the annual cost of rental property by
25 percent.
9. Comment on the following statement: "Property values in this city are
increasing rapidly, rising at an annual rate of about 20 percent. This will cause
many households to switch from renting to owning."
10. The owners of rental property can deduct routine maintenance cost from their
gross income: homeowners cannot. Does the deductibility of maintenance cost
decrease the relative cost of renting, thus creatine a bias toward renting?
Chapter 14 Why Is Housing Different.' 403
1 1. Consider a city where the demand for housing for the typical household is
Q = 1.000- 1.000- P
where Q is the square footage of housing demanded and P is the price per
square foot. The original price of housing is 50 cents per square foot.
a. Draw thedemand curve for housing.

b. How many square feet of housing does the typical household demand?
c. What is the consumer surplus associated with a price of 50 cents?
Suppose that the price of housing decreases to 40 cents.
d. What is the consumer surplus at the new price, assuming that the household
does not move to a new residence?
e. What is the consumer surplus at the new price, assuming that the household
moves to a new residence?
/ If the cost of moving is $25, will the typical household move to a new
residence?
12. In city D, landlords charge a security deposit of $900 and do not refund the
money when the tenant leaves. Suppose that a new deposit law is passed. The
law requires that all landlords refund 100 percent of the security deposit unless
the landlord can prove that the tenant damaged or dirtied the dwelling.
Comment on the following statement: "The new law will save the typical
tenant $900."
13. Comment on the following statement: "I live in an apartment complex built in
a year when interest rates were relatively low. My rent is lower than the rent
on other apartments because the cost of financing the complex is relatively
low."
14. In Figure 14-8, the housing-price function for blacks is flatter than the
housing-price function for whites. Why?
15. One of the assumptions behind Figure 14-8 is that all dwellings are equally
accessible to employment opportunities. Draw a new housing-price function
under the assumption that all workers commute to the city center, so
commuting cost increases with u (distance to the center).

Overview
Arnott, Richard. "Economic Theory and Housing." In Handbook of Regional and Urban
Economics, vol. 2, ed. Edwin S. Mills. New York: Elsevier-North Holland Publishing,
1987. Reviews various models of the urban housing market and suggests directions for
future research.
Olsen, Edgar O. "A Competitive Theory of the Housing Market." American Economic
Review 59 (1969), pp. 612-22.
"The Demand and Supply of Housing Service: A Critical Survey of the Empirical
.
Handbook of Regional and Urban Economics, vol. 2. ed. Edwin S.

Literature." In
Mills. New York: Elsevier-North Holland Publishing. 1987. Reviews empirical studies
of housing supply and demand.
Quigley. John M. "What Have We Learned about Housing Markets?" In Current Issues in
Urban Economics, ed. Peter Mieszkowski and Mahlon Straszheim. Baltimore: Johns
Hopkins University Press. 1979. A review of theoretical and empirical research in
housing.
U.S. Bureau of the Census. Statistical Abstract of the United States: 1986. Washington.
DC: U.S. Government Printing Office. 1985.
Supply of New Housing
Follain.James R. "The Price Elasticity of the Long-Run Supply of New Housing

Land Economics 55 1979). pp. 190-99.
Construction." (
Muth. Richard. "The Demand for Non-Farm Housing." In The Demand for Durable Goods.
ed. Arnold C. Harberger. University of Chicago Press. 1960. Estimates the elasticity of
supply of new housing.
Supply of Used Housing
Farb, Warren E. "An Estimate of the Relative Supply and Demand for Substandard Rental
Housing in Major U.S. Cities." Ph.D. dissertation. Washington University. St. Louis.
1971. Suggests that the supply of low-quality housing is highly elastic.
Ingram. Gregory K.. and Yitzhak Oron. "The Behavior of Housing Producers." In
Residential Location ana" Urban Housing Markets, ed. Gregory K. Ingram. New York:
National Bureau of Economic Research. 1977. Discusses a landlord's choice of quality
level, based on costs and benefits (rent) of different quality levels.
Kain, John, and William Apgar. "Simulation of Housing Market Dynamics." Journal of the
American Real Estate and Urban Economics Association 7 1979). pp. 505-38. A
(
simulation model in which landlords choose a quality level consistent with profit
maximization.
Abandonment
Muth. Richard F. "The Demand for Non-Farm Housing." In The Demand for Durable
Goods, ed. Arnold Harberger. University of Chicago Press. I960, pp. 29-96.
Peterson. George. "The Effect of Federal Taxes on Urban Form." In The Prospective City,
ed. Arthur Solomon. Cambridge. Mass.: MIT Press. 1979. Discusses the effects of
federal taxes on abandonment.
Sterlieb, George, and Robert Burchell. Residential Abandonment: The Tenement Landlord
Revisited. New Brunswick. N.J.: Rutgers University Press. I
l
)73.
White. Michelle. "Property Taxes and Urban Housing Abandonment." Journal of Urban
Economics 20 1986). pp. 312-30. Explores the effects of inflexible property taxation
(
on the frequency of abandonment and concludes that the property tax encourages
abandonment.
Supply Elasticities
De Leeuw, Frank, and Nkanta Ekanem. "The Supply of Rental Housing." American
Economic Review 61 197 ), pp. 806-17.
( 1
Grieson, Ronald E. "The Supply of Rental Housing: Comment." American Economic

Review 63 (1973). pp. 433-36.
Mason, John M. "The Supply Curve for Housing." Journal of the American Real Estate and
Urban Economics Association 7 (1979), pp. 362-77.
Ozanne, L., and Raymond Struyk. "The Price Elasticity of Supply of Housing Services." In
Urban Housing Markets: Recent Directions in Research and Policy, ed. L. S. Bourne
and J. R. Hitchcock. University of Toronto Press, 1978. Estimates the supply elasticity
of used housing.
Smith, Barton A. "The Supply of Urban Housing." Quarterly Journal of Economics 90
(1976), pp. 389^06.
The Hedonic Approach

Brown, H. James. "Micro Neighborhood Externalities and Hedonic Housing Prices." Land
Economics 56(1 980), pp. 25^4- 1 1
Crecine, John; Otto Davis; and John Jackson. "Urban Property Markets: Some Empirical
Results and Their Implications for Municipal Zoning." Journal of Law and Economics
10(1967), pp. 79-100.
De Salvo, Joseph S. "Benefits and Costs of New York City's Middle-Income Housing
Program." Journal of Political Economy 83 (1975), pp. 791-805. Found that housing
projects for middle-income housing increased the values of nearby properties by about
5 percent. Also estimated the welfare losses from using public housing instead of cash
transfers as 45 percent of the program cost.
Follain, James R., and Emmanuel Jiminez. "Estimating the Demand for Housing
Characteristics." Regional Science and Urban Economics 15 (1985), pp. 77-107 A '.
survey and critique of hedonic studies of the housing market.

Freeman, A. Myrick III. "The Hedonic Approach to Measuring Demand for Neighborhood
Characteristics." In The Economics of Neighborhood, ed. David Segal. New York:
Academic Press, 1979.
Kain, John F, and John M. Quigley. Housing Markets and Racial Discrimination: A
Microeconomic Analysis. New York: National Bureau of Economic Research, 1975. A
hedonic study of the St. Louis housing market.
. "Measuring the Value of Housing Quality." Journal of the American Statistical
Association 65 (1970), pp. 532-38.
King, Thomas. "The Demand for Housing: Integrating the Roles of Journey to Work.
Neighborhood Quality, and Prices." In Household Production and Consumption, ed.
Nester Terlecky. New York: National Bureau of Economic Research, 1975. A hedonic
study of housing.
. "The Demand for Housing: A Lancastrian Approach." Southern Economics Journal
43 (1976), pp. 1077-87. A hedonic study of housing prices.
Mills, Edwin S. "Housing Policy as a Means to Achieve National Growth Policy." In
Housing in the Seventies: Working Papers, vol. 1. Washington, D.C.: U.S. Government
Printing Office, 1976. Suggests that neighborhood effects are real but are relatively
small, between 5 and 15 percent of the amount invested in a dwelling.
Straszheim, Mahlon. "Estimation of the Demand for Urban Housing Services from
Household Interview Data." Review of Economics and Statistics 55 (1973), pp. 1-8.
A hedonic study of housing prices.
. An Economic Analysis of the Urban Housing Market. New York: National Bureau of
Economic Research, 1975. A hedonic study of housing prices.
Demand Elasticities
Ellwood, David, and Mitchell Polinski. "An Empirical Reconciliation of Micro and
Grouped Estimates of the Demand for Housing." Review of Economics and Statistics
61 (1979), pp. 199-205.
Tax Incidence on Owner-Occupied Housing: Evidence from
Ihlanfeldt. Keith R. "Property
Annual Housing Survey." National Tax Journal 35 (1982).
the
Mayo, Stephen K. "Theory and Estimation in the Economics of Housing Demand." Journal
of Urban Economics 10 ( 198 ), pp. 95-1 16.
1
Polinski, Mitchell. "The Demand for Housing: A Study in Specification and Grouping."
Econometrica 45 (1977), pp. 447-62.
Filtering
Brueggeman, William B. "An Analysis of the Filtering Process with Special Reference to
Housing Subsidies." In Housing in the Seventies: Working Papers, vol. 2. Washington,
D.C.: U.S. Government Printing Office, 1976, pp. 842-56. Discusses the process
through which subsidies for high-quality housing generate benefits for the occupants of
low-quality housing.
Grisby, William. Housing Markets and Public Policy. Philadelphia: University of
Pennsylvania Press, 1963. Discusses the filtering process.
Kain, John E, and William C. Apgar. "Simulation of Housing Market Dynamics." Journal
of the American Real Estate and Urban Economics Association 3 (1975), pp. 508-38.
Lowry, Ira S. "Filtering and Housing Standards: A Conceptual Analysis." Land Economics
36(1960), pp. 362-70.
Ohls, James C. "Public Policy toward Low Income Housing and Filtering in Housing
Markets." Journal of Urban Economics 2 1975). pp. 144-71.
(
Sweeney, James L. "A Commodity Hierarchy Model of the Rental Housing Market."
Journal of Urban Economics 1975), pp. 288-323.
1 (
Neighborhood Dynamics
Downs, Anthony. "Key Relationships between Urban Development and Neighborhood
Change." Journal of the American Planning Association 45 1979). pp. 462-72.(
Grisby, William; Morton Baratz; and Duncan Maclennan. The Dynamics of Neighborhood
Change and Decline. Philadelphia: University of Pennsylvania. 1984. Discusses the
theories of neighborhood decline.
Heilbrun, James. "On the Theory and Policy of Neighborhood Consolidation." Journal of
theAmerican Planning Association 45 (1979), pp. 417-27. Advocates a resettlement
program for central cities, under which households would be moved to certain
neighborhoods. The remaining neighborhoods would be cleared for new development.
Stegman. Michael A. "The Neighborhood Effects of Filtering." Journal of the American

Real Estate and Urban Economics Association 5 ( 1977). pp. 227-4 Discusses the
1 .
effects of new construction on thedemand for central-city housing. Advocates the

reorientation of federal programs away from policies that encourage the production of
new suburban housing.
Stegman, Michael, and David Rasmussen. "Neighborhood Stability in Changing Cities."
American Economic Review 70 (1980). pp. 415-19.
Taxes and Homeownership
Gordon. Roger H.; James R. Hines; and Lawrence Summers. "Notes on the Tax Treatment
of Structures." Working Paper no. 1896, National Bureau of Economic Research.
Cambridge. Mass., 1986.
Hendershott. Patric H., and James Shilling. "The Economics of Tenure Choice. 1955-79."
In Research in Real Estate, ed. C. Simians. Greenwich. Conn.: JAI Press, 1982.
pp. 105-33.
King,Mervyn A., and Don Fullerton, eds. The Taxation of Income from Capital: A
Comparative Study of the United States, the United Kingdom, Sweden, and West
Germany. University of Chicago Press, 1984. Estimates that rental housing is
subsidized more heavily than owner-occupied housing.
Rosen, Harvey S. "Housing Decisions and the U.S. Income Tax: An Econometric
Analysis." Journal of Public Economics 67 (1979), pp. 1-23.
Struyk. Raymond. Urban Homeownership. Lexington. Mass.: D. C. Heath, 1976. Explores
the effects of mortgage subsidy on homeownership. Suggests that households are
responsive to changes in the relative cost of ownership.
Struyk, Raymond J.; Neil Mayer: and John A. Tuccillo. Federal Housing Policy at
President Reagan's Midterm. Washington, D.C.: Urban Institute, 1983.
White, Larry J., and Michelle J. White. "The Tax Subsidy to Owner-Occupied Housing:
Who Benefits?" Journal of Public Economics 7 (1977), pp. 1 1 1-26.
Discrimination and Segregation
J. "A Note on the Economics of

Bailey, Martin Residential Zoning and Urban Renewal."
Land Economics 35 (1959). pp. 288-92.
Berry, Brian. "Ghetto Expansion and Racial Residential Segregation in an
Urban Model."
Journal of Urban Economics 3 (1976). pp. 397^123.
Clark, Thomas A. "The Suburbanization Process and Residential Segregation." In Divided
Neighborhoods, ed. Gary A. Tobin. Beverly Hills, Calif.: Sage Publications, 1987.
Courant, Paul. "Racial Prejudice in a Search Model of the Urban Housing Market." Journal
of Urban Economics 5 (1978). pp. 329-45.
Courant, Paul, and John Yinger. "On Models of Racial Prejudice and Urban Residential
Structure." Journal of Urban Economics 4 (1977), pp. 272-9 1
Darden, Joe T "Choosing Neighbors and Neighborhoods: The Role of Race in Housing
Preference." In Divided Neighborhoods, ed. Gary A. Tobin. Beverly Hills. Calif.: Sage
Publications. 1987. Shows that racial segregation is caused largely by racial
discrimination, not by income segregation or voluntary segregation.
Farley, John E. Segregated City, Segregated Suburbs: Are They Products of Black-White
Socioeconomic Differentials? Edwardsville: Southern Illinois University, 1983.
Concludes that differences in income and other socioeconomic characteristics play a

small role in racial segregation.
Galster. George C. "Neighborhood Racial Change. Segregationist Sentiments, and Affir-
mative Marketing Policies." Journal of Urban Economics 27 1990). pp. 344-61.(
Explores the factors that determine the rate at which the racial composition of a
neighborhood changes.
King. Thomas, and Peter Mieszkowski. "'Racial Discrimination. Segregation, and the Price
of Housing." Journal of Political Economy 8 1973). pp. 590-601. Provides evidence
(
that blacks pay more than whites for housing.

Smith. Barton. "Racial Composition as a Neighborhood Amenity." In The Economics of
Urban Amenities, ed. Douglas B. Diamond and George S. Tolley. New York: Academic
Press. 1982. Reviews theoretical models of racial discrimination and segregation and
provides evidence that blacks paid less than whites for housing in Houston.
U.S. Department of Housing and Urban Development. The 1978 HUD
Survey of the Quality
of Community Life: A Data Book. Washington. D.C.: U.S. Government Printing Office.
1979.
Woolbright. Louie Albert, and David J. Hartmann. "The New Segregation: Asians and
Hispanics." In Divided Neighborhoods, ed. Gary A. Tobin. Beverly Hills. Calif.: Sage
Publications. 1987.
Yinger. John. "Caught in the Act: Measuring Racial Discrimination with Fair Housing
Audits." American Economic Review 76 1986). ( pp. 881-93. Relates the results of
housing audits to various theories of discrimination and prejudice.
. "Evidence on Discrimination in Consumer Markets." Journal of Economic
Perspectives 12. no. 2 (Spring 1998). pp. 23—40.
. "Prejudice and Discrimination in the Urban Housing Market." In Current Issues in
Urban Economics, Mieszkowski and M. Straszheim. Baltimore: Johns Hopkins.

ed. P.
1979. pp. 430-68. Discusses the causes and consequences of racial prejudice and
discrimination. Summarizes the empirical evidence that blacks paid more than whites
for housing.
. "The Racial Dimension of Urban Housing Markets in the 1980s." In Divided
Neighborhoods, ed. Gary A. Tobin. Beverly Hills. Calif.: Sage Publications. 1987.
Discusses recent evidence of discriminatory treatment of black house buyers and
policy implications.
Chapter 15
&c OUSING POLICIES
(~~7~ his is the second of two chapters on the urban housing market. It discusses
^y three types of housing policies. The first is housing assistance: the federal
government uses various policies to improve the housing conditions and decrease
the housing cost of poor households. Second, the federal government uses a number
of community development programs to support local efforts to improve housing
conditions and revitalize neighborhoods. Third, under a rent-control policy, the local
government sets a maximum price for rental housing.
The federal government provides housing assistance through a number of pro-
grams administered by the Department of Housing and Urban Development (HUD).
In 1997, HUD's programs for rental housing provided subsidies for about 4.8 mil-
lion low-income households. About two-thirds of the households were served by
policies that operate on the supply side of the market: the government either builds
new housing for the poor (public housing) or subsidizes private developers to build
and manage low-income housing. About 1.4 million households occupied dwellings
in conventional public housing projects, and another .8 million households lived in
1
privately owned housing projects that were subsidized by the federal government. In
contrast, under a demand-side policy, the government hands out housing vouchers
(coupons) that low-income households can use to help pay for privately owned rental
housing. In 1997, about 1.6 million households received housing vouchers. In recent
years, the emphasis of housing policy has shifted away from supplying housing
either directly or indirectly —
in favor of housing vouchers. Most vouchers are now
portable: they can be used anywhere in a particular metropolitan area.
Conditions of Housing and Neighborhoods

The original goal of federal housing policy was to provide "a decent home and a
suitable environment for every U.S. family." Have these objectives been met? Are
most citizens living in decent houses in suitable neighborhoods?
409
Housing Conditions and Rent Burdens

Table 15-1 shows the percentage of households that experienced housing problems
in 1990. The definitions of housing problems come from the U.S. Department of
Housing and Urban Development (HUD). A dwelling is considered "inadequate" if
it has incomplete plumbing or kitchen facilities, structural problems (e.g., cracked
walls, leaking roof, broken plaster), deficiencies in common areas (stairwells, hall-
ways), or unsafe heating or electrical systems. A dwelling is "crowded" if there is
more than one person per room. A household is "cost-burdened" if it spends more
than 30 percent of its gross income on housing.
There are three conclusions we can draw from Table 15-1 First, overcrowding
.
is a relatively small problem: only 2.7 percent of all households live in overcrowded
housing, and only 7.5 percent of poor households (income less than half the median
area income) are overcrowded. Second, the most prevalent problem is affordabil-
ity. The number of cost-burdened households is almost four times the number of
households in inadequate housing. The affordability problem is most extensive
among black and poor households: 4 in 10 black households and 6 in 10 poor

households are cost-burdened. Third, the most severe problems are experienced by
Table 15-1 The Condition of

the Housing Stock
and Housing Cost
Burdens, 1991
Percent of
Households
All Households
Inadequate 7.9
Crowded 2.7
Cost-burdened 28.0
Total with problems 38.7
Poor Households
Inadequate 17.4
Crowded 7.5
Cost-burdened 62.1
Metropolitan Residents
Inadequate 7.3
Crowded 2.9
Cost-burdened 29.6
Black Households
Inadequate 17.4
Crowded 5.2
Cost-burdened 38.0
Soi k< iU.S Department of Commerce, Annual

Housing Survey forthe United States, 1991.
Chapter 15 Housing Policies 411
poor households and black households. Most (87 percent) poor households have
some sort of housing problem, and 6 in 10 are cost-burdened. Almost two-thirds
of black households experience problems: about 1 in 4 black households lives in
inadequate or overcrowded housing, and an additional 4 in 10 are cost-burdened.
In the last several decades, there has been considerable progress inimproving
the condition of the housing stock. The percentage of households in overcrowded
housing dropped from 13.5 percent in 1950 to 2.9 percent in 1991. The percentage
of urban households living in "substandard" housing (defined differently from inad-
equate housing) was 21.9 percent in 1950, but the percentage in inadequate housing
was only 7.3 percent in 1991.
Neighborhood Conditions
Table 15-2 provides evidence concerning the level of neighborhood dissatisfaction
for various types of households. The first column of numbers shows the percentage
Table 15-2 Opinion of Neighborhood by Race,

Income, Structure Type, and Location
Neighborhood Want to
Quality Is Poor (%) Move (%)
Household Type
Black, low. multi. city 16.8 36.1
Black, high, multi. city 9.7 32.2
Black, high, multi, suburbs 6.3 25.7
Black, low. multi, suburbs 10.5 25.6
White, low. multi, city 11.2 22.3
White, high, multi. city 5.7 21.8
Black, low. single, city 10.3 19.7
Black, high, single, city 5.4 17.4
White, high, multi, suburbs 3.9 16.3
White, low. single, city 6.5 15.9
White, high, single, city 3.3 15.4
Black, high, single, city 3.3 14.9
White, low, multi, suburbs 5.2 14.1
Black, low, single, suburbs 3.3 13.6
White, high, single, suburbs 1.8 10.4
White, low. single, suburbs 3.4 10.0
Definitions
Poor quality: score 3 or less on alO-point quality scale.
Low: income in bottom third of income distribution.
High: income in top two-thirds of income distribution.
Multi: live in multifamily housing.
Single: live in single-family dwelling.
City: live in central city.
Suburbs: live in suburban area.
SOURCE: Thomas P. Boehm and Keith R. Ihlanfeldt, "The Revelation of

Neighborhood Preferences: An N-Chotomous Multivariate Probit Approach,"
Journal of Housing Economics 1 ( 1991 ), pp. 33-59.
of households of each type that gave their neighborhood a score of 3 or less on

a 10-point scale. The is among households living in multi-
greatest dissatisfaction
family dwellings in central For example, 16.8 percent of low-income black
cities.
households living in multifamily housing in the central city said that their neigh-
borhood w as of "poor" quality and 36. 1 percent of these households said that the
neighborhood was so bad that they wanted to move. Among similar households
living in single-family housing in the suburbs, only 3.3 percent considered their
neighborhood to be of poor quality and only 13.6 percent said that they wanted to
move. The figures for low -income white households are 11.2 percent (poor qual-
ity) and 22.3 percent (want to move) for multifamily housing in the central city and
3.4 percent (poor quality) and 10.0 percent (want to move) for single-family housing
in the suburbs. Neighborhood dissatisfaction varies with race and income: blacks
are generally more dissatisfied than whites, and low -income households are more
dissatisfied than high-income households.
Boehm and Ihlanfeldt
99 explore the effects of various neighborhood charac-
( 1 1 )
teristicson the perceived quality of the neighborhood. Their results suggest that dis-
satisfaction is most prevalent in neighborhoods with high crime rates, high noise lev-
els, run-down and abandoned buildings, and large quantities of trash, litter, and junk.
One of the implications from Table 15-2 is that the incidence of neighborhood
dissatisfaction is higher among black households. Boehm and Ihlanfeldt show that
the most important factors in explaining the difference in satisfaction are differences
in ( crime rates. (2) the amount of litter, trash, andjunk. (3) the quality of buildings,
1 )
and (4) the number of abandoned buildings.

Figure 15-1 shows the results of the following thought experiment. Suppose
that the crime rate in the typical black neighborhood changes to the crime rate that
Figure 15-1 Reasons for the Gap between White and Black
Neighborhood Dissatisfaction
20 r-
M 15
o 10 -
SOI k< Thomas P. Bivhm and Keith R Ihlanfeldt. "The Revelation of Neighborhood PrefereiKw
i
\n \ Chotomous Multivariate Probil Approach," Journal of Housing Economics iiwii. pp 33 5°

1
prevails in the typical white neighborhood. By how much would

the gap between
black and white dissatisfaction decrease? The answer gap would decrease
is that the
by 12 percent. Figure 15-1 shows the results of repeating this thought experiment for
the other factors that contribute to neighborhood dissatisfaction: abandoned buildings
are responsible for 14 percent of the gap, and litter, trash, and junk are responsible
for another 17 percent.
Supply-Side Policy: Public Housing

In 1998, about 1.4 million households lived in public housing. In terms of budgetary
costs, the federal government's outlays for public housing included $3.1 billion for
operating subsidies (to cover the gap between the rent collected from tenants and the
costs of operating the project) and $3.8 billion for capital expenditures (for repair,
upgrading, and demolition). In recent years, there has been no investment in new
public housing; the emphasis in housing policy has shifted toward demand-side
policies (housing vouchers).
Public housing is managed by local housing authorities, subject to various rules
adopted by the federal government. The rent charged to a particular tenant can be
no more than 30 percent of the household's income. In most cities, there are long
waiting lists for public housing units, and the federal government has directed local
housing authorities to give priority to households with "very low" income (less than
half the median income of the area).
Public Housing and Housing Consumption

Figure 15-2 shows the effects of public housing on the housing consumption of
recipients. The horizontal axis measures housing consumption (measured as the
quantity of housing services generated by the dwelling), and the vertical axis mea-
sures the monthly consumption of all The original budget
other goods (in dollars).
line is drawn under the assumptions that (1) monthly income is $500 and (2) the
price of housing is 25 cents per unit of housing service per month. Given the house-
hold's preferences for housing (as represented by the negatively sloped indifference
curves), the household chooses point B, spending $150 on housing (occupying a
600-unit dwelling) and $350 on other goods.
Suppose that the typical public dwelling is an apartment that generates 800 units
of housing service and is provided to the poor for $100. Public housing adds point C
to the household's budget set: the household can consume an 800-unit apartment and
have $400 left to spend on other goods. Given the household's indifference curves,
utility ismaximized at point C, so the household moves into public housing. The
public housing program increases housing consumption (from 600 to 800 units of
housing service) and decreases housing costs (from $150 to $100).
Will public housing always increase housing consumption? Figure 15-3 shows
the choices of a household with the same income as the household in Figure 1 5-2, but
different tastes for housing. In the absence of public housing, the household lives in a
1,000-unit apartment (point D). The household spends $250 on housing (1,000 units
Figure 15-2 Public Housing Increases Housing Consumption

s
500
Additional budget point
400 — with SI 00 public housing
350
600 800 2.000
Housing consumption (units of housing sen ice |
Income is S500 andthe price of housing is 25 cents per unit of housing service. With the original
budget household chooses point B. Public housing (apartments generating 800 units of
line, the
service at a cost of S 100) adds point Cto the budget set. Public housing allows the household to
increase its consumption of both housing and other goods.
times 25 cents per unit), leaving S250 for other goods. If the household moves into
public housing, it occupies a dwelling with fewer units of housing service (800 units)
and has an additional SI 50 to spend on other goods. Public housing increases utility
if the additional $150 worth of nonhousing goods is worth the sacrifice of 200 units
of housing service. Given the indifference curves in Figure 15-3. utility is higher at
point C than at point D. so the household accepts the offer of public housing and
decreases its housing consumption.
Public Housing and Recipient Welfare
Would the household be better off with cash instead of subsidized public housing?
Consider first the possibility that the cost o\~ public housing is the same as the cost
of private housing (25 cents per government S200 to build and
unit). It costs the
maintain the 800-unit apartment, so the subsidy per dwelling is SI 00 (S200 less the
Figure 15-3 Public Housing Decreases Housing Consumption
Additional budget point

with $100 public housing
Original choice
Original budget line
800 1,000 2,000
Housing consumption (units of housing service)
The household starts at point D ( ,000 square feet of housing and $250 of other goods). Public
1
housing adds point C to the budget set. If the household occupies public housing, it decreases its
housing consumption from ,000 to 800 and increases its consumption of other goods from $250
1
to $400.
$100 rent charged to tenants), and the alternative to public housing is a $100 cash
payment to the household. In Figure 1 5 —4, a $100 cash payment shifts the recipient's
budget line upward by $ 100. The household chooses point E instead of point C and
achieves a higher utility level.
This conclusion is not a result of an arbitrary placement of the indifference

curves, but of basic microeconomic principles. The household would prefer the cash
because it provides more consumption options. The cash allows the household to
choose point E (the public housing point) or any other point along the new budget
line. The household would be better off with the cash if it chooses any point except
C(i.e., if it does not spend exactly $200 on housing). If the objective of public policy
is to increase the household's utility, a cash payment is superior to public housing.
How much is The value of public housing
public housing worth to the tenant?
can be measured by answering the following question: what cash payment would
make the household indifferent between the cash and a $100 housing subsidy? In
Figure 15-4 Public Housing versus Cash Payment

S600
500 Choice with SI 00 cash
Additional budget point with SI 00

^^^~^ public housing
400
\\ C -u 2
—
350
\fl
Z
Uj
^s Budget line with S 1 00 cash
^s ^^^^~^ payment
<
^s ^s ^ Original budget line
600 800 2.000 .400
The alternative to public housing (800 units of housing sen ice at a price of SI 00 instead of S200) is a
$100 cash payment. Under public housing, the household chooses point C. The $100 cash payment
shifts ihe budget line up by S100. and the household chooses point E. Utilit\ is higher under ihe cash
payment because the household can decide how to spend the money.
other words, what cash payment would move the household to indifference curve
U-{1 In Figure 1 5-5. a $75 payment shifts the budget line outward, and the household
maximizes point F. Since the household reaches the indifference curve U-i,
utility at
it is between $75 in cash and the $100 housing subsidy. Studies by Kraft
indifferent
and Olsen 977 ). Barton and Olsen 976), and Murray ( 1 975 suggest that tenants
( 1 ( 1 )
receive a benefit of about 75 cents per dollar spent on public housing. More recently.
Smeeding 1982) estimated that the benefit per dollar is 80 cents.
(
How does the cost of new public housing compare to the cost of private housing?
Public housing is more expensive for two reasons. First, the private sector can build
new low-income housing more efficiently than the public sector: Weicher (1979)
cites a number of studies showing that new public housing costs more than new
private housing. Second, there is a plentiful supply of used low-quality housing, so

even the least costly new housing costs more than used housing. According to the
National Housing Review (1974). the cost of new public housing is about 33 percent
Figurk 15-5 Monetary Value of Public Housing

$
575
500 Choice with $75 cash
Choice with $100 public housing
400
S 350
Budget line with
$75 cash payment
Original budget line
600 800 2,000 2.300
Public housing ($100 for an 800-unit apartment) is equivalent to a $75 cash payment. A $75 payment
allows the household to choose point F. which is on the same indifference curve as point C (the point
chosen with public housing). The household is indifferent between $100 public housing and a cash
payment of $75.
higher than the cost of used private housing. If new public housing costs one-third
more than used private housing, the government spends $266 (not $200) to provide
a new 800-unit apartment. Therefore, the government's cost of getting to point C in
Figure 15-4 (the public housing option) is $166 ($266 less the $100 rent paid by
tenants). In contrast, the cost of getting to point E (the cash option) is only $100,
so the cash option generates a larger increase in utility for about three-fifths of the
cost.
What is the bang per buck of public housing? In other words, what is the recipient
benefit per dollar spent on public housing? As shown in Figure 15-5, recipients
receive a benefit of $75 from public housing, so the bang from public housing is
$75. If the government spends $266 to build the apartment and collects $100 in
rent from the tenant, the government's cost is $166. Therefore, the bang per buck is
45 cents ($75 divided by $166).

Figure 15-6 Crime Rates in Chicago

Housing Projects
90
84
SI)
80
70
c
u
7 60
4>
O
S 50
o
c
40
I
'J
| 30
20
10
SOLRCE: Newsweek, Januan 2. 1989. p. 25.
Living Conditions in Public Housing
Public housing projects suffer from a number of problems, including high rates of
drug abuse and crime. Figure 15-6 shows the incidence of violent crime in three
public housing projects in the city of Chicago. The crime rates in the projects were
between 1.86 and 3.81 times the crime rate for the central city as a whole.
The crime rates in housing projects are relatively high for three reasons. First,
the residents of the projects are relatively poor (in Chicago, the median household
income of project residents is $5,000. compared to $22,000 for the central city as a
whole), and crime rates are higher among the poor. Second, many of the units in the
projects have been abandoned by the city and have been taken over by drug abusers
—
and gangs. Third, the physical layout of the typical project a cluster of high-rise
buildings — may contribute to crime. If crime rates are higher in high-density clusters
of poor households, one option for decreasing crime is to replace the high-rise clusters
with low-rise buildings dispersed throughout the metropolitan area.
Figure 15-7 Short-Run and Long-Run Effects

of Public Housing
Short-run
private supply
Private supply
after 5 years
250
208
Original
demand
300 380 500

Number of private low-income dwellings
Public housing decreases the demand for private low-income housing by 200 dwellings. In the
short run, the supply of private housing is of private dwellings drops from
fixed, so the price
$250 to $150. Over a five-year period, the decrease in the price of private housing decreases the
quantity of housing supplied from 500 to 380. so the price rises to $208. The longer the period of
time, the greater the displacement of private housing and the smaller the net decrease in price.
The Market Effects of Public Housing

Figure 15-7 shows the short-run and long-run market effects of public housing. In
the short run, the supply of housing from the private sector is perfectly inelastic:
500 dwellings are provided, regardless of price. At the initial equilibrium (point B),
the price is $250 per dwelling per month. If the local housing authority builds 200
new public housing dwellings, the demand curve for private housing shifts to the left
by 200 dwellings: the 200 households in public housing leave the private housing
market. As the poor move into public housing, they vacate private dwellings, causing
excess supply that decreases the price of housing from $250 to $150 (point C). In
the short run, all housing consumers gain from the public housing program: some
households occupy subsidized public housing, and others pay lower prices for private
housing.
What happens to the private market over time? In Figure 1 5-7, the intermediate-
run supply curve (the supply curve over a five-year period) is positively sloped. The
decrease in the price of housing caused by public housing decreases the profitabil-
ity of private dwellings, so fewer low-quality dwellings are supplied. The supply
response takes two forms:
1. More retirement. As explained in Chapter 14 (Why Housing Differ-

Is
ent?), a decrease in the price of low-quality dwellings makes economic

profits negative for some dwellings, so more dwellings are retired from the
housing market. The retired dwellings are either converted to another use or
abandoned.
2. Slower downward As the price of low-quality housing decreases
filtering.
medium-quality housing, landlords hold their dwell-
relative to the price of
ings in the medium-quality submarket for a longer period of time.
In other words, more dwellings drop out of the low-quality submarket and fewer
dwellings filter down into the low-quality submarket. As the number of low-quality
dwellings decreases, the market price rises. The exit process continues until the
price is high enough to restore zero economic profits (normal accounting profits).
In Figure 15-7, a total of 120 private dwellings are withdrawn from the market, and
the price rises from $150 to $208. Because public housing displaces private housing,
there is a net increase in supply of only 80 dwellings.
Ohls (1975) used a computer model of the housing market to explore the effects
of public housing on the filtering process. He divided the housing stock into 60
quality levels, and allocated housing in each level to the city's households: the
highest-quality dwellings were occupied by the wealthy: the lowest by the poor.
Under a public housing policy, the bottom fifth of the population is removed from
the private housing market, causing the following changes:
1. The amount of new private construction decreased: the decrease in the

demand for used low-quality housing lowered the incentive to build new
housing.
2. Fewer dwellings filtered down to the low-quality submarket: the low-
quality marketbecame less lucrative, so the filtering rate decreased.
3. The retirement rate increased: the decrease in the demand for low-quality
dwellings lowered the incentive to keep low-quality dwellings on the
market, so the retirement rate increased.
In general, the private market responded by building fewer dwellings, slowing the
filtering process, and increasing the retirement rate.
Supply-Side Policy: Subsidies for Private Housing
One alternative to public housing is a system of subsidies to encourage the private

sector to build and manage low-income housing. There
are tw o subsidy programs for
rental housing: Section —
236 and Section 8 -Project Based. Under both programs,
the government pays the property owner the difference between the household's
contribution to rent and the "fair market rent." In most cases, the household's contri-
bution is 30 percent of its income. The fair market rent is determined by the cost of
building and managing the property (Section 236) or the prevailing rent in the area
(Section 8). For example, suppose the fair market rent of an apartment is $300, and
an eligible household's income is $600. In this case, the household would pay $180
(30 percent of $600) and the government would pay $120, for a total of $300.
Under these two subsidy programs, the federal government signed long-term
contracts to provide annual payments to property owners. Under the contracts, the
owner was guaranteed the fair market rent on all units occupied by eligible house-
holds. Together these programs accommodate about 1.8 million households. In re-
cent years, no new contracts have been signed, but the government continues to pay
property owners under the terms of old contracts. In addition, many contracts have
been renewed.
Demand-Side Policy: Consumer Subsidies

Supply-side housing policies help the poor by building more low-income housing.
An alternative approach is to give subsidies directly to the poor and let them choose
their own housing. Under a demand-side policy, low-income households are given
housing coupons that can be redeemed for housing. Like the food-stamp program, the
housing-coupon program allows the poor to make their own consumption choices.
Coupons have two advantages over public housing. First, under some circumstances,
the coupons are equivalent to cash. As shown earlier in the chapter, a cash payment
generates a larger increase in utility than an equal expenditure on public housing.
Second, the coupons can be spent on used dwellings, which are typically less ex-
pensive than new dwellings. There are two types of housing coupon programs: rent
certificates and housing vouchers.
Rent Certificates: Section 8
Under the Section 8 program, poor households are issued rent certificates (coupons)
to help pay the rent on privately owned dwellings. The bulk of rent certificates are
issued to households with "very low" income (less than half the median income of
the area). In 1997, about 920,000 households received rent certificates.
There are two restrictions on the housing choices of recipients. First, the rental
dwelling must meet minimum physical standards for size and quality. Second, the
household cannot spend more than the fair market rent, defined by HUD as a reason-
able rent for a standard low-income dwelling. The eligible household pays 30 percent
of its income toward housing and receives a rent certificate to cover the difference
between its contribution and the actual rent:
Subsidy = A.ctual rent — 0.30 • Income (15-1)
For example, suppose that the fair market rent is $400 and the household's income
is $600. The household's contribution is $180, and the subsidy equals the difference
between the actual rent (up to a maximum of $400) and $180. If the actual rent is
only $350, the subsidy is $170.

Figure 15-8 Effects of Section 8 Rent Certificate

on Housing Consumption
600
180 300 400

Spending on housing (monthly rent)
With the original budget AH. the household spends S300 on rent and S300 on other goods
line
Ipoint .' The rent-certificate program « ith a lair market rent of S400 adds points between C and
)
I) to the budget set: the government pays the difference between actual rent and 30 percent of
income s SO), so the opportunity cost of spending he\ond S180 is zero. In the absence of
i 1
nun ing cost, the household switches to point /) because it lies on a higher indifference curve. If
the mm ing cost is relatively large, the household chooses point F.
Figure 15-8 shows the household's choice between housing and other goods
under the rent-certificate program. Household income $600 per month, and the is
fair market rent is $400. The original budget line

shows the initial trade-off AB
between spending on housing and other goods: every dollar of spending on housing
decreases spending on other goods by one dollar. The certificate program adds the
points between C and D to the budget set: the household receives a subsidy equal
to the gap between its actual spending on housing (its monthly rent) and 30 percent
of its income ($180). Therefore, it can consume $420 worth of nonhousing goods
and between $180 and $400 worth of housing. If the household rents an apartment
for $181. it receives a subsidy of $1 : if it rents an apartment for $400. it receives a
subsidy of $220. Because the government pays the difference between actual rent
and $ 80. an increase
1 in housing consumption does not decrease the consumption of
other goods as long as rent is less than the fair market rent. If the household spends
more than the fair market rent ($400), it is ineligible for the certificate program, so
it is stuck with the original budget line.
How does the certificate program affect the household's budget choice? In the
absence of moving cost, a move from E to D increases utility. It makes sense to go all
the way to D because between points C and D the opportunity cost of more housing
is zero: the tenant contribution to housing is fixed at $180 and the government pays
any increase in rent. If moving cost is relatively large, however, the household moves
straight up from E to F. In this case, the increase in utility is not large enough to
offset the moving cost, so the household stays in its original residence and spends
the entire subsidy on other goods.
One of
the surprising results of the rent-certificate program was that rela-
tively few households moved to dwellings that rented for the fair market rent. In
Figure 15-8, most households moved from point E to F, not E to D. This occurred
despite the fact that the entire increase in rent would have been covered by the sub-
sidy. Most households stayed in their existing dwellings, and used the subsidy to cut
their housing costs. This result suggests that the cost of moving is relatively large.
Housing Vouchers
The housing-voucher program was started on an experimental basis in 1983. The
eligibility requirements for housing vouchers are similar to those for rent certificates:
the bulk of payments go to the very poor. To qualify for a voucher, the household must
occupy a dwelling meets the minimum quality standards. A voucher differs from
that
a rent certificate in one important respect: the recipient can use the voucher to rent any
dwelling that meets minimum physical standards. In 1997, about 400,000 households
received housing vouchers. The federal budget for fiscal year 1999 includes 50,000
new vouchers (at a cost of $283 million) for households affected by the welfare-
reform package passed in 1996. These vouchers are "portable" in the sense that they
can be used anywhere in a particular metropolitan area. The idea behind making them
portable is to allow the recipients to move to residential areas accessible to jobs.
The face value of the voucher is based on household income and the fair market
rent. The formula is
Face value = Fair market rent — 0.30 •

Income (15-2)
Figure 15-9 shows the household's choice between housing and other goods under
the voucher program. Household income is $600 and the fair market rent is $400. so
the face value of the voucher is $220 ($400 — $180). The voucher program shifts the
budget line from AB to AFG. Point F is in the new budget set because the household
could use the voucher to get $220 worth of housing and spend all of its own income
($600) on other goods. As spending on housing rises above $220, there is a dollar-
for-dollar trade-off between housing and other goods. The voucher program gives
the household more options than the rent-certificate program because the voucher is
equivalent to a cash transfer. In the absence of moving cost, the household with the
choices shown in Figure 15-9 increases spending on housing from $300 (point E)
Figure 15-9 Effects of Housing- Voucher Program on Housing

Consumption
600
420 -
300
120 300 400 600 820

Spending on housing (monthly rent I
With the original budget line AR. the household spends $300 on rent and $300 on other goods
(point £*). The voucher program with a fair market rent of $400 shifts the budget line outward to
AFG. The household receives a voucher worth $220 (the difference between the rent standard
and 30 percent of income). In the absence of moving cost, the household chooses point H, and is
belter off than it would be under a rent-certificate program (point /)). If the cost of moving is
relatively large, the household chooses point J.
to about $345 (point H), leaving $175 of the subsidy to spend on other goods. If
the cost of moving is relatively large, the household chooses point J and spends the
entire subsidy on other goods.
The voucher household also has the option of spending more than the fair market
rent. In other words, the household has access to points along the budget segment
DG. In contrast to the rent-certificate program, the voucher program allows the
household to spend as much as it wants on housing.
Market Effects of Consumer Subsidies

What are the market effects of rent certificates and housing vouchers? To explain
the market effects, consider a city with the following characteristics:
1. The city has poor families and middle-income families.

2. To qualify for a housing voucher, a poor household must occupy a
moderate-quality dwelling (one that generates at least 800 units of housing
service).
3. The poor households initially live in low-quality dwellings (which generate

only 600 units of housing service).
4. Middle-income households live in moderate-quality dwellings (800 units of
housing service) and medium-quality dwellings (1,000 units of housing
service).
Figure 15-10 shows the short-run effects of the voucher program on the moderate-
quality submarket. The voucher program shifts the demand curve outward, increas-
ing the market price from $250 (point B) to $350 (point E). Because the supply
of moderate-quality housing is fixed in the short run, the vouchers simply bid up
the price of housing. Who who loses in the short run? Although voucher
gains and
recipients pay more for housing, they also have more money to spend on housing.
In contrast, nonrecipients (middle-income households) pay higher housing prices
without the benefit of housing vouchers.
Figure 15-10 Short-Run and Long-Run Effects of Housing Vouchers

Short-run supply
Supply after 5
years
Demand with
housing vouchers
500 620 700

Number of moderate-quality dwellings
The housing-voucher program increases the demand for moderate-quality dwellings. In ihe short run, the
supply of moderate-quality housing is fixed, so the price increases from $250 to $350. Over a five-year
period, the increase in the price of housing increases the quantity supplied from 500 to 620. and the price
drops to $292. The longer the period of time, the greater the increase in the quantity supplied and the smaller
the net increase in price.
What happens to the moderate-quality market after five years? Figure 15-10
shows the intermediate-run supply curve (the supply curve over a five-year period).
The increase in price increases the profitability of moderate-quality dwellings, in-
creasing the quantity supplied, for three reasons:
1 . Downward filtering. Dwellings filter down more rapidly from the

medium-quality submarket to the more lucrative moderate-quality
submarket.
2. Increased maintenance. The owners of moderate-quality dwellings spend
more on maintenance and repair to keep their dwellings in the market for a
longer period of time: downward filtering is slowed.
3. Upward filtering. Dwellings from the low-quality submarket are upgraded
to serve the more lucrative moderate-quality submarket.
For these reasons, the quantity of moderate-quality dwellings supplied increases as

the price increases. In equilibrium (point F), the price of moderate-quality housing
is $292 and the number of dwellings is 620.
How does the voucher program affect the medium-quality submarket? The in-
crease in the price of moderate-quality dwellings increases the price of medium-
quality dwellings for two reasons. First is the demand effect: the increase in the
relative price of moderate-quality dwellings causes some consumers to switch to the
medium-quality submarket. The demand for medium-quality dwellings increases,
increasing the market price. Second is the supply effect: the increase in the rel-
ative price of moderate-quality dwellings hastens the filtering of medium-quality
dwellings downward to the moderate-quality submarket. In combination, the in-
crease in demand and decrease in supply increase the market price of medium-quality
dwellings. Because of the interactions between the housing submarkets. the voucher
program increases the prices of all types of housing.
Ohls (1975) used a computer model of the housing market to explore the effects
of consumer subsidies on the filtering process. As explained earlier, his model divides
the housing stock into 60 quality levels, with the wealthy occupying the highest-
quality dwellings and the poor occupying the lowest-quality dwellings. Under his
consumer subsidy program, the government provides a 50 percent housing subsidy
for the poorest fifth of the population. The subsidy program causes the following
changes in the market:
1 The poorest households upgrade to higher-quality housing, moving from

quality level 5 to 8. The typical recipient moves from quality level 8 to 11.
2. The increase in the demand for moderate-quality housing stimulates new

construction (at the high-quality level) because the market for used housing
becomes more lucrative.
3. Filtering down to the moderate-quality range speeds up. and filtering down
from the moderate-quality range slows down.
The Experimental Housing Allowance Program

An alternative to a housing voucher is a housing allowance. A housing allowance is a
cash payment, the size of which equals the gap between a given fraction of household
income 30 percent) and the fair market rent. To qualify for a housing allowance,
(e.g.,
the household must occupy a dwelling that meets minimum quality standards. A
housing allowance is equivalent to a housing voucher if (1) the two programs have
the same quality standards, (2) the face value of the voucher equals the allowance
payment, and (3) the household's rent exceeds the face value of the voucher. If
the household spends more than the voucher value on housing, the household can
substitute the voucher for its own spending on housing, freeing up an amount of cash
equal to the face value of the voucher.
Under the Experimental Housing Allowance Program (EHAP), the federal gov-
ernment conducted a number of experiments with housing allowances. The demand
experiment offered housing allowances to a small sample of poor renter house-
holds in two cities (Phoenix and Pittsburgh). The purpose of the experiment was to
examine the effects of the program on eligible households. The most important pol-
icy issues concerned participation rates (how many would participate), the mobility
of participants (how many would move to higher-quality dwellings), and housing
consumption (how much housing consumption would increase). The supply exper-
iments offered allowances to all poor households in two counties (Brown County,
Wisconsin, and St. Joseph County, Indiana). The purpose of the supply experiment
was to examine the effects of housing allowances on the market price of housing.
The principal question was whether a citywide assistance program would increase
housing prices.
The experiments generated three conclusions about the responses of eligible
households:
1. Low participation rates. Between 40 percent and 45 percent of the

eligible households participated in the allowance programs. Participation
rates were sensitive to the quality standards: the stricter the standards, the
lower the participation rate.
2. Small changes housing consumption. Between one-quarter and

in
one-third of the allowance payment was used to increase housing
consumption, and the remainder was spent on other goods. In other
words, most of the allowance money was used to decrease rent burdens.
The measured income elasticity of demand of allowance recipients was
relatively low: between 0.20 and 0.44 for renters, and about 0.45 for
owner-occupants.
3. Mobility. Relatively few households changed dwellings. In Pittsburgh,
40 percent of allowance recipients moved over a two-year period, compared
to 35 percent of households in a control group. In Phoenix, 63 percent of
recipients moved, compared to 53 percent of the control households. The
allowance program did not encourage integration with respect to race or
income.
The general conclusion is that the housing allowances caused relatively small
changes in housing choices. Recipients spent most of the money on nonhousing
goods, and few moved from their original dwellings.
The allowance program did not cause any measurable price increases in the two
cities involved in the supply experiment. Because participation rates were relatively
low and recipients spent a relatively small fraction of their allowances on housing,
the rightward shift of the demand curve was relatively small. To the extent that
thedemand for housing increased, the supply of housing was sufficiently elastic to
accommodate the increased demand without any measurable price effects.
What are the lessons from the supply experiment? As pointed out by Kain 198 ( 1 ).
the supply experiment tested the effects of one type of allowance program, with
minimal housing quality standards and relatively low participation rates. Because
the allowanceprogram increased the housing consumption of a small number of
households by relatively small amounts, it is not surprising that housing prices did
not rise. If a national housing allowance program had higher participation rates,
larger allowance payments, and more stringent housing standards, it would shift the
housing demand curve by a larger amount and would probably increase housing
Kain cites studies by the National Bureau of Economic Research and the
prices.
Urban Institute that suggest such an allowance program would increase prices in
some housing submarkets.
Supply versus Demand Policies

What are the trade-offs associated with supply-side policies (public housing) and
demand-side policies (housing vouchers)? From the perspective of the recipient,
the voucher program has two advantages over subsidized public housing. First, the
voucher is equivalent to an income transfer, so it generates a larger increase in utility
per dollar spent on housing programs. Second, the voucher allows tenants to occupy
relatively inexpensive used housing, so a fixed budget provides more housing per
budget dollar.
The disadvantage of the voucher program is that it increases the price of housing.
The recipient would be better off under the voucher program if the program caused a
relatively small increase in the price of housing. This occurs if the elasticity of supply
isrelatively large. As explained in Chapter 14 (Why Is Housing Different?), little
isknown about the intermediate-run and long-run elasticities of supply of housing.
Unless one knows something about the individual housing submarkets in a city, it is
of a voucher program on the price of housing.
difficult to predict the effects
What and nonrecipients* perspectives toward public housing and

are taxpayers'
housing vouchers? Two factors contribute to public support of public housing over
housing vouchers. First, taxpayers seem to care about the housing consumption of
recipients, not necessarily their utility levels. If public housing generates a larger
increase in housing consumption, it is considered a superior policy. Second, nonre-
cipients pay a higher price of housing under a demand-side policy, and a lower price
under a supply-side policy. For these two reasons, taxpayers and nonrecipients may
prefer public housing over housing vouchers even if recipients are better off under
the voucher program.
Community Development and Urban Renewal

There are dozens of programs and policies that fit under the term community de-
velopment. The legislated mandate for federally supported community development
calls for "systematic and sustained action by the federal, state, and local govern-
ments to eliminate blight, to conserve and renew older areas, to improve the living
environment of low- and moderate-income families, and to develop new centers
of population growth and economic activity." The two principal purposes of com-
munity development policies are to (1) revitalize declining areas of the city and
(2) improve the housing of poor households. Community development policies are
neighborhood-oriented in the sense that they concentrate spending in relatively small
geographical areas.
Urban Renewal
Urban Renewal, the first community development program in the United States, was
established under the Housing Act of 1949 and eventually dropped in 1973. The
national government provided local governments with the power and the money to
demolish and rebuild parts of their cities. Local agencies acquired property under
the right of eminent domain, cleared the site of "undesirable" uses (like low-income
housing and small businesses), and then either built a public facility or sold the site
to a private developer. The local agencies charged the developer less than the cost
of acquiring and clearing the site, and the federal government covered two-thirds
of the local government losses. The private developer built housing (usually for
middle-income and high-income households), government buildings, or commercial
establishments.
According to the Congressional Research Service, Urban Renewal had the fol-
lowing effects:
1. Demolition. A total of 600,000 dwellings were demolished, displacing

about 2 million people, most of whom were poor.
2. New housing. A total of 250,000 new dwellings were built, most of which
were occupied by middle-income and high-income households.
3. Public facilities. The total floor space of new public facilities was 120
million square feet.
4. Commercial facilities. The total floor space of new commercial facilities
was 224 million square feet.
5. Property assessments. The assessed value of property on renewed sites
increased by 360 percent.
430 Part 111 Poverty and Housing
In sum, Urban Renewal generated both costs and benefits. It displaced poor house-
holds, but also provided housing for middle-income and rich households, allowed
the construction of commercial and public facilities, and increased tax revenue.
Was Urban Renewal worthwhile? The critics of the program focus on its demo-
lition aspects, pointing out that 2 million poor people were displaced. The defenders
of the program focus on its rebuilding aspects, pointing out that the new commercial
developments provided jobs for the poor residents of the central city.
Recent Community Development Programs
More recent federal community development programs have avoided many of the
problems of the Urban Renewal program. The newer programs are executed on a
smaller scale, so they displace fewer households, and they place a greater emphasis
on providing housing for the poor.
The bulk of funding for community development is for a program called Com-
munity Development Block Grants (CDBG). In 1997, the CDBG budget was $4.7
billion, with about 70 percent of the funds going to central cities and urban coun-
ties, and the remainder going to smaller localities. The allocation formula used to
distribute CDBG funds gives large grants to cities with relatively old and over-
crowded housing, high poverty rates, and slow economic growth. The funds are
used to improve housing, support public services, promote economic development,
and clear land for new development. Under the Urban Development Action Grants
(UDAG), funds are used to leverage private investment in community development.
The government provides small subsidies to transform slightly unprofitable private
development projects into profitable ones.
Cities spend their community development grants in a wide variety of projects.
The projects can be divided into four types:
1 . Improved housing. Some programs support local projects that renovate

buildings, enforce building codes, and build new low-income housing.
• The Boston Housing Partnership used $8.5 million in CDBG funds to
acquire and rehabilitate 700 rental units.
• The city of Cleveland joined with a nonprofit organization and a private
developer to build a 183-unit rental housing project, using $4.4 million in
federal funds ($1.7 million from CDBG and $2.7 million from UDAG) to
leverage $9.9 million from private and other sources.

2. Renewed infrastructure. Some programs support projects that improve
streets, roads, and water and sewage facilities.
• In Fort Worth, Texas, the Main Street Project used $2.6 million in UDAG
funds to improve the street's infrastructure (new curbs, gutters, sidewalks,
and street lighting). As part of the renovation program, private developers
agreed to build a $48 million hotel.
• The city of Oshkosh. Wisconsin, used $600,000 of CDBG and UDAG
funds to expand its airport to accommodate increased traffic.
3. Job development. Some programs support projects that encourage

economic development and job creation.
• In St. Louis,Union Station was renovated with $10.2 million in CDBG
and UDAG money and $126 million in other public and private funds.
The station was developed into a hotel-retail entertainment complex that
produced 2,000 new jobs, over 50 percent of which were filled by
minorities. The project rejuvenated a part of the city that had been
dormant for 15 years.
• Salt Lake City is using a $1 .1 million UDAG to support the construction
of a downtown auto mall, using the grant to leverage $4.7 million in
private financing.
• Baltimore used $10 million in UDAG funds to support the revitalization
of its Inner Harbor. Private investment in office space, retail space, and a
hotel totaled $73.4 million and produced 1,600 jobs.
4. Responsive public services. Some programs support public services for

and children.
the elderly, the homeless,
• Orange County, California, used $892,000 in CDBG funds to support the
construction of a $7.5 million facility to provide emergency services for
abused, abandoned, or neglected children.
• Phoenix used $361,000 in CDBG funds to build a recreation center for
the handicapped.
• Fort Worth, Texas, used $106,000 in CDBG funds to start a neighborhood
youth project designed to curtail gang activity.
These examples show that community development funds are spent on a wide variety
of projects.
What are the effects of community development programs? The programs that
support housing projects supplement federal housing programs (public housing, sub-
sidized new construction, housing vouchers). The difference is that the city has more
control over the spending of community development funds. Other community de-
velopment programs support economic development projects and can be evaluated as
alternative means of promoting local economic growth. The costs and benefits of eco-
nomic development programs are discussed in Chapter 6 (Urban Economic Growth).
Rent Control
During World War II, government instituted a national system of rent
the federal
controls. Following the war, New
York City was the only city to retain rent con-
trols. During the 1970s, rent control returned to dozens of cities, including Boston,
Cambridge, Los Angeles, Washington, D.C., Albany, Berkeley, and Santa Monica.
According to the U.S. Department of Housing and Urban Development (1991), more
than 200 localities had rent control in 199 and over 1 percent of the nation's private
1 ,
rental housing was subject to these controls.

Market Effects of Rent Control

To explain the market effects of rent control, consider a city with the following
characteristics:
1. All households live in identical apartments.
2. The equilibrium rent is $600 per month.

3. The equilibrium number of apartments is 90.
4. The apartments are made of stone and do not deteriorate over time.
5. The city sets a maximum rent of $400.
Figure 15-1 1 shows the short-run and long-run effects of rent control. The short
run is defined as the period of time over which the supply of housing is fixed at 90
dwellings. In the short run, rent control does not affect the supply of housing, but
simply transfers $200 per month from the landlord to the tenant.
In the long run, the supply of apartments depends on the price of rental housing.
As rent decreases, some landlords convert their property to more lucrative uses:
some apartment buildings are converted to condominiums, and other buildings are
Figure 15-11 Short-Run and Long-Run Effects of Rent Control
Long-run
supply
600
400
Demand
50 <)() 120
Numbei of apartments
In the short run. the suppl) of apartments rem control (maximum rem $400) simplj
is fixed, so
transfers income from landlords market moves from poinl li to (' In the long run.
to tenants: the
market moves from poinl (' to point /).
ihe decrease in price de< reases the quantity supplied: the
The shortage (the gap between demand ami suppl) is 7ii apartments
>
demolished to make way for commercial land uses. In Figure 15-1 1, rent control
decreases the number of apartments from 90 to 50. As the price of rental housing
decreases, the market moves downward along the long-run supply curve. The supply
response to rent control depends on the long-run elasticity of supply. The more elastic
the supply, the larger the decrease in the quantity of housing supplied.
Rent control causes a shortage of housing. At the controlled price of $400, the
quantity of apartmentsdemanded is 120, compared to a supply of only 50 apartments.
There are too many tenants chasing too few apartments. In such a market, tenants
spend relatively large amounts of time and money searching for apartments. For
tenants, rent control brings good news and bad news: although rent control decreases
the rent, it also increases search costs.
Rent control produces winners and losers. The city's population can be divided
into four groups:
1 Nonmovers. Some households live in the city when rent control is

implemented and occupy the same apartment for life. These households
are winners in the rent-control game: they pay a lower price for the same
apartment.
2. Successful searchers. Some people move to the city after rent control is
implemented. They spend several months searching for an apartment and
eventually find one. The full cost of a rent-controlled apartment is the sum
of the rent ($400 per month) and the opportunity cost of search time. The
full cost of a rent-controlled apartment may exceed the full cost of an
uncontrolled apartment ($600 per month and small search cost). In general,
if the search cost is relatively large, the benefit of rent control (decreased
rent) is smaller than the cost (increased search cost), so successful searchers
are losers in the rent-control game.
3. Displaced households. Some tenants are displaced when apartments are

converted or demolished, and are unsuccessful in their search for a new
apartment. In Figure 15-1 1, the number of dwellings decreases by 40 (from
90 to 50), so 40 households are displaced. The displaced households are
obviously losers.
4. Landlords. Rent control decreases the annual rental income per apartment
from $7,200 ($600 times 12 months) to $4,800. Because the market value
of an apartment building reflects its income-earning potential, the
announcement of the rent-control program decreases the market value of
the city's apartment buildings. If Barney owns an apartment building when
rent control is announced, he pays for rent control in the form of a
decreased market value. He cannot escape the effects of rent control by
selling the building to someone else: once prospective owners hear about
rent control, they want to pay less for the building. According to Smith and
Tomlinson (1981), Toronto's rent-control program decreased the market
value of apartment buildings by about 40 percent over a five-year period
(1975-1980).
Olsen (1972) estimated the effects of New York City's rent-control program on
tenants and landlords. Ignoring search cost, he estimated that the occupants of con-
trolled units experienced a 3.4 percent increase in real income. On the other hand,
landlords experienced losses in real income about twice as large as the total gain of
tenants.
Distributional Effects of Rent Control
Gyourko and Linneman (1989) use data from 1968 to estimate the distributional
effects of New York City's rent-control program. They estimate the decrease in rents
experienced by different types of households. Their results provide insights into the
from rent control across income groups. The principal policy
distribution of benefits
question do most of the benefits from rent control go to poor households'?
is:
Figure 15-12 shows the relationship between the benefit from rent control (the
decrease in annual rental payments) and household income. All the figures are in
1984 dollars. The top line shows the benefit-income relationship for households
that occupy controlled units. The average benefit among occupants of controlled
dwellings was S2.440 per year, and the average savings as a percentage of income
Figure 15-12 Benefits from Rent Control for Different Income Levels
in New York Citv
2.600
2,400 _
2.200
2.000
-
-
\ Households in
controlled units
1.800 — >v^
1.600 All renter
1.400 - V households
1.200
1.000 -
^^^^ \~^
All households
800
600
400
200
1 1 1 1
1
10 20 30 40 50
Vnnual income ($1 .000)
Soi RCE: Joseph Gyourko and Petei inneman, "Equit) and Efficienc) Aspects of
I Rem Control: \n Empirical
S(ud\ of New YorkCity" Journal of Urban Economics 26(1989), pp 54 "^
was 27 percent. The At each

benefit per household increases slowly with income.
income wide variation in the benefits experienced by individual
level, there is a
households. For example, the average gain for households with incomes of $20,000
was $2,440, but many households had much larger benefits and many had much
smaller benefits.
The middle line in Figure15-12 shows the benefit-income relationship for all
renting households. The average benefit among renting households was about $ 1 ,3 1
per year. The average benefit among renting households is smaller than the average
benefit among the occupants of controlled units because the benefits are spread over
more households: only 57 percent of renting households occupy controlled units.
The benefit per household decreases as income increases, reflecting the fact that the
average income of households in controlled units is lower than the average income
of renting households: poor households are overrepresented in controlled dwellings,
so poor households, on average, experience larger benefits from rent control.
The lower line in Figure 15-12 shows the benefit-income relationship for all
households, including both renters and homeowners. The average benefit among all
households was about $856 per year. The average benefit among all households is
relatively small because less than a third of the households occupy controlled units.
The benefit decreases as income increases, reflecting the relatively low income of
households in controlled units.
Figure 15-12 suggests that the benefits of rent control are mildly progressive
in the sense that the average benefit per household decreases as income increases.
It's important to note, however, that the benefits experienced by households within
an income group vary widely across households. The variation in benefits across
households within an income group occurs for two reasons. First, only some of
the households of a given income
occupy controlled dwellings (and receive
level
benefits from rent control). The other households (renters of uncontrolled units and
homeowners) receive no benefits from rent control. Second, among households that
occupy controlled dwellings, there is a large variation in the benefits from rent
control. Given the large variation in the benefits within each income group, rent
control is a very blunt tool for income redistribution.
Deterioration and Maintenance

One of the assumptions of the simple model of the rent-control city is that apartments
are made of stone, so dwellings do not deteriorate over time. This is clearly unrealis-
tic. Suppose that housing is subject to physical deterioration: to maintain a dwelling
at a given service level, the landlord must spend money on routine maintenance and
repair. Suppose that in the initial equilibrium, the typical $600 apartment generates
1,000 units of housing service (Q = 1,000).
Rent control discourages routine maintenance, so apartments deteriorate more
rapidly. In an uncontrolled city, landlords spend money to keep their apartments at
a service level high enough to justify a $600 rent. If the apartment deteriorates to Q
= 900, tenants are unwilling to pay $600 rent for the dwelling. In the rent-control
city, there is no incentive to keep the apartment at Q = 1,000 because the landlord
cannot charge S600 for the dwelling. Landlords decrease their spending on repair
and maintenance, thus allowing apartments to deteriorate to a service level consistent
with the $400 rent. Under a rent-control program, tenants may eventually get what
they pay for.
Studies of rent-control cities show that landlords do indeed cut back on main-
tenance and repair. The Rand Corporation studied the effects of rent controls in
New York City (1970). and concluded that rent control decreased the quality and
quantity of rental housing. Between 1960 and 1967. the inventory of "sound" hous-
ing increased by 2.4 percent, the inventory of "dilapidated" housing increased by
44 percent, and the inventory of "deteriorating" housing increased by 37 percent.
Between 1965 and 1967. 114.000 dwellings were retired from the housing stock.
A study of Cambridge. Massachusetts, found that maintenance cost per apartment
decreased by $50 per year (Navarro. 1987).
Gyourko and Linneman (1990) examined the effects of rent controls on the
quality of rental housing in New York City in 1968. They show that rent-controlled
buildings are less likely than uncontrolled buildings to be in "sound" condition,
meaning that rent-controlled buildings are more likely to be deteriorating or dilapi-
dated. One measure of the effect of rent control is the decrease in the probability that
a building is sound, given that the building is subject to rent control. For example,
if the probability of being sound is 0.90 for uncontrolled buildings (90 percent of
uncontrolled buildings are sound) but only 0.85 for controlled buildings, rent control
decreases the probability of being sound by 0.05.
Figure 15-13 shows the decrease in the probability that a building is sound for
several types of low-rise buildings (less than seven stories) in Manhattan. Brooklyn,
and Queens. For buildings in Manhattan built before 1947. rent control decreases
the probability by about 0.09 (from 0.63 to 0.54). Newer buildings, which are more
likely to be sound, experience smaller decreases in the probability of being sound.
For buildings Manhattan built between 1947 and 1959. rent control decreases the
in
probability of being sound by about 0.04 (from 0.89 to 0.85): for buildings built
between 1960 and 1968. rent control decreases the probability of being sound by
0.01 (from 0.98 to 0.97). The effects of rent control are smaller in the two other
boroughs of New York.
Housing Submarkets
One of the assumptions of the rent-control model is that all households live in
identical apartments. If there are housing submarkets. the analysis changes in two
ways.
Consider first two controlled submarkets. There are
the possibility that there are
large apartments (equilibrium rent of $600 and controlled rent of $400) and small
apartments (equilibrium rent of $300 and controlled rent of $200). Rent control
inhibits movement between the two markets because the search cost is large. Suppose
that a household would like to move from a large apartment lo a small one. In the
rent-control city, there are two options. First, the household can staj in their large
Figure 15-13 Rent Control and the Probability that a

Building Is Sound
0.10 r-
0.08
0.06
0.04
0.02
0.00
Pre- 1947 1947-59
n 1960-68
Year of construction
J Manhattan
1 Brooklyn
I Queens
SOURCE: Joseph Gyourko and Peter Linneman. "Rent Controls and Rental Housing
Quality: A Note on the Effects of New York City's Old Controls," Journal of Urban
Economics 27 (1990), pp. 398^109.
apartment, paying $400 per month. Second, they can spend time and money searching
for a small apartment (which rents for $200). If the search cost is relatively high,
they are better off staying in their original apartment.
Consider next the possibility that some of the dwellings in the city are not subject
Rent control decreases the supply of housing in the controlled sector.
to rent control.
The displaced households move into the uncontrolled submarket, bidding up the
price of housing. Therefore, part of the cost of rent control is borne by households
living in uncontrolled dwellings.
Controlling the Supply Response to Rent Control
Rent control decreases the quantity and quality of rental housing. Cities have devel-
oped a number of policies to diminish this supply response.
1 Rent adjustments. Most rent-control laws allow periodic increases in rent,

with the allowable increase tied to some price index, for example, the
consumer price index (CPI). Rent adjustments weaken rent controls and
diminish the negative supply effects of the controls. In most rent-control
cities that use rent adjustments, controlled rents rise more slowly than the
cost of building and maintaining rental housing. In New York City,
controlled rents rose by 1 10 percent between 1943 and 1968. while the
operating costs of rental housing increased by 285 percent (de Salvo, 1971).
In Toronto, controlled rents increased by 39 percent between 1975 and
1980, while the operating cost of rental housing increased by 75 percent
(Smith and Tomlinson. 1981 ).
2. Exemption of new housing. Most rent-control laws exempt new con-

struction from controls. If new housing is exempt from controls, it may be
unaffected by rent control. In fact, tenants displaced from the controlled
market increase the demand for new rental housing, stimulating the supply
of new rental housing. It is possible that new housing offsets the loss in
housing stock caused by rent control. But there are two reasons to suspect
that new housing will not offset the effects of rent control. First, new
housing is typically a small part of the housing market, so it would take a
relatively large increase in the supply of new housing to counteract the
effects of rent control. Second, builders and owners are likely to be
skeptical about the city's promise to never impose rent control on this
year'snew rental housing. If builders suspect that today's new housing will
become tomorrow's rent-control housing, they may be reluctant to supply
new housing.
3. Restrictions on conversions. Many cities restrict the conversion of rental
housing to condominiums. If successful, these restrictions limit the
conversion of rental housing to other uses, diminishing the supply effects of
rent control.
4. Subsidies for new construction. Public policies encourage the building of

new dwellings to replace the dwellings lost as a result of rent control. The
city of Toronto offered low-interest construction loans to housing firms. In
the United States, one of the purposes of public housing is to replace the
dwellings lost because of rent control.
5. Vacancy decontrol. In some cities, rent is fixed during a tenancy, but can
be changed when a new tenant moves into the dwelling. This provision
increases the landlord's rental income, weakening rent control. It also
increases the incentive for long-term occupancy.
Another way to decrease the supply effect of rent control is to cheat. Tenants and
landlords often find ways to violate the spirit of rent-control laws. Landlords charge
tenants "key money" (e.g.. $5,000 for a key to the apartment) or nonrefundable
security deposits. To the extent that landlords use indirect payments to increase the
net price of housing, the negative supply response to rent control is reduced. In
the city of Cairo, which has had rent control since 1944, key money is an accepted
practice. On average, payments to property owners in the form of key money increase
the effective rent from about 38 percent of the market price to about 71 percent
(Malpezzi, 1998).
Tenants also collect bribes for rent-controlled apartments: when a tenant vacates
a controlled apartment, he can charge the new tenant a finder's fee. The fee is often
implicit: in California, old tenants reportedly sold bean-bag chairs to new tenants,
charging hundreds of dollars for chairs that sold for just a few dollars in thrift stores.
Tenant bribery does not increase the supply of housing, but simply transfers income
from the new tenant to the old one.
Alternatives to Rent Control
In many cities, rent control is form of income redistribution.

viewed as an indirect
The idea is to take money from wealthy it to poor tenants. At
landlords and give
best, rent control is a blunt instrument for redistribution. Rent control helps all of
the occupants of controlled dwellings, rich and poor alike. Moreover, a poor person
who occupies a controlled dwelling is worse off under rent control if the search cost
is larger than the savings in rent.
Suppose that a city wants to redistribute income from the wealthy to the poor.
What are the alternatives to rent control? The first option is to finance income
redistribution with a land tax. Because the supply of land is perfectly inelastic, the
land tax will not affect the supply of land or the supply of housing. Therefore, the land
tax will not cause the same supply problems as rent control. The second approach is
to leave income income

redistribution to the federal government. Local attempts at
redistribution are likely to be frustrated by the movement of households into and out
of the city. If the city gives the proceeds of a land tax to its poor residents, poor people
from other cities may migrate to the city, diluting the effects of the redistribution
program. Because international mobility is less than intercity mobility, a national
redistribution program does not suffer from this problem.
Summary
1. The condition of the U.S. housing stock has improved considerably over the last
several decades. Nonetheless, about two-thirds of poor households experience
some sort of housing problem, usually inadequate housing or high housing
costs. Over half of black households experience housing problems.
2. About 1.4 million households live in public housing. Public housing is built
and managed by local housing authorities. The federal government provides
subsidies for capital cost, operating cost, and modernization, and specifies who
is eligible for public housing.
a. Public housing may increase or decrease housing consumption, depending
on the recipient's willingness to trade off consumption of housing and other
goods.
b. Public housing produces a relatively small increase in recipient welfare per

dollar because (i)monetary value of public housing is about 75 percent of
the
its market value and public housing costs more than used private housing.
(if)
c. In the short run. the supply of private housing is fixed, so public housing
decreases the market price of housing by a large amount.
d. Over time, the decrease in the price of housing decreases the quantity of
private dwellings supplied as more dwellings are retired and fewer dwellings
filter down to the low-quality submarket. The more elastic the supply, the
greater the displacement of private housing and the smaller the net decrease
in market price.
3. There are two types of consumer subsidies: Section 8 rent certificates and
housing vouchers.
a. The rent certificates cover the difference between the actual rent and
30 percent of household income, and can be used on dwellings that rent for
less than the fair market rent.
b. The face value of a housing voucher equals the gap between fair market rent
and 30 percent of household income, and can be used on any dwelling
satisfying minimum physical standards.
c. Both subsidy programs shift the demand curve outward, increasing the
market price in the short run. Over time, the quantity of housing increases as
(i) fewer dwellings are retired and (if) more dwellings filter down from
higher-quality submarkets. The larger the supply elasticity, the smaller the
net increase in price.
4. The Experimental Housing Allowance Program (EHAP) tested a program of

housing allowances (cash payments to poor households). The demand for
housing increased by a relatively small amount because (i) participation rates
were relatively low, (ii) few households changed dwellings, and {Hi) only a
small fraction of the allowance payment was spent on housing. The small shift
of the demand curve caused no measurable increases in housing prices.
5. There are dozens of policies and programs that fit under the term community
development. The two principal purposes of such policies are to (i) revitalize
declining areas of the city and (if) improve the housing of poor households.
6. Under the Urban Renewal program, the federal government subsidized the
replacement of low-income housing with middle-income and high-income
housing, public facilities, and commercial establishments.
7. More community development programs support local projects that
recent
improve low-income housing and local infrastructure, stimulate economic
development, and develop new public sen ices for handicapped, elderly, and
disadvantaged citizens.
8. Rent control has the following effects on the housing market.
a. The supply of dwellings decreases, causing a housing shortage that increases
search cost.
b. Spending on maintenance and repair decreases, so individual dwellings
produce lower quantities of housing sen ices.
c. Consumer mobility decreases, and consumers make inefficient housing

choices.
9. Rent-control cities use a number of policies to control the decrease in supply

caused by rent control, including rent adjustments, exemption of new housing,
restrictions on conversion, subsidies for new construction, and vacancy
decontrol.

1. In Figure 15-3, the offer of $100 public housing decreased housing
consumption. Suppose that the government increases the rent on an 800-unit
dwelling from $100 to $150.
a. Using a graph similar show the budget point with (/) $100
to Figure 15-3,
public housing and $150 public housing.
(ii)
b. Assume that the household starts with housing consumption of 1,000 units.
Draw a set of indifference curves such that the household accepts the offer
of $100 public housing, but is indifferent between its original choice and
$150 public housing.
2. Suppose that the household in Figure 15-3 starts with housing consumption of
1,200 units of housing service instead of 1,000. Is this household more or less
likely to decrease its housing consumption to get public housing? Explain.
3. Figure \5-A shows that the consumer prefers cash to free public housing. The
figure was drawn under the assumption that a system of cash transfers would
not increase the price of housing.
a. Redraw the budget line under the assumption that the cash transfers
increase the price of housing from 25 cents to 50 cents per unit.
b. Draw a set of indifference curves such that the consumer prefers free public
housing (with no change in price) to the cash transfer (with the associated
increase in price).
c. Under what conditions (what values of the relevant elasticities) will
low-income households prefer new housing to the cash?
d. Where are these conditions likely to be met? That is, for what group of
individuals and what locations are these conditions likely to be met?
4. Comment on the following statement of Mr. Taxpayer: "I support public

housing rather than cash transfers because I want to be sure that the recipients
of public assistance increase their housing consumption. If we give them cash,
they will spend the money on something else."
5. According to Mr. Wizard, a realtor in Modville, "If my assumptions are

correct, the long-run market effects of public housing are nil. The building of
200 public housing dwellings will affect neither the equilibrium price of
housing nor the quantity of housing."
a. What are Mr. Wizard's assumptions?
b. Use Mr. Wizard's assumptions to predict the long-run market effects of
housing vouchers.
6. In the early days of Section 8 rent certificates, the recipient received 25 percent
of the savings associated with renting a dwelling for less than the fair market
rent. Draw Figure 15-8 and add the budget line under the early version of the
rent-certificate program.
7. Suppose that you are interested in maximizing the welfare of the poor. You
must choose between cash payments and free housing. What data do you
need? Explain how you would use the data.
8. Suppose that the poor households of city H are enrolled in a federal
housing-allowance program. You are a middle-income renter in the city.
a. Will you be helped or harmed by the allowance program?

b. If you will be harmed, what variables determine how much you will be
harmed? In other words, what are the relevant elasticities, and how does
the cost you incur vary with them?
c. How would your answer to (a) change if you owned your home?
9. Contrast the effects of rent control in two cities. In city R. the elasticity of
supply of housing is 5.0 (i.e.. a 10 percent increase in the price of housing
increases the quantity of housing by 50 percent). In city S. the supply of
housing is less elastic: the supply elasticity is 0.40.
10. Cheating on rent control is common. If you were a prospective tenant in a
rent-controlled city, would you encourage your prospective landlord to cheat

1 1 Figure 1 5- 1 3 suggests that the effects of rent control increase with the age of
the building. Provide an explanation for this result.
12. Discuss the trade-offs associated with demolishing the high-rise clusters of
public housing and replacing them with dispersed low-rise buildings.

Description of Federal Housing Policies
Aaron. Henry J. "Rationale for a Housing Policy." In Federal Housing Policies and
Programs, ed. J. Paul Mitchell. New Brunswick. N.J.: Center for Urban Policj
Research. 1985. Discusses frequenth cited justifications for federal housing
policies.
Jacobs. Barry G.: Kenneth R. Harney: Charles L. Edson: and Bruce S. Lane. Guide to
Federal Housing Programs. 2nd ed. Washington. D.C.: Bureau of National Affairs.
1986. A detailed description of federal housing policies (public housing, housing
subsidies, consumer subsidies, and coinmunit\ development).
Mariano. Ann. "The Voucher Controversj Drags On." Washington Post. I
:
ebruar\ IS. 19S C).
Discusses the budgetary costs of vouchers and rent certificates.
Mitchell. J. Paul. "The Historical Context tor Housing Policy." In Federal Housing Policies
and Programs, ed. J. Paul Mitchell. New Brunswick. N.J.: Center tor Urban Policj
Research. 1985. A historical sketch of federal housing policy.
Stmyk, Raymond J.: Neil Mayer; and John Tuccillo. Federal Housing Policy at President
Reagan's Midterm. Washington. D.C.: Urban Institute. ls>S.v Explains the changes in
federal housing policies (assisted housing, community development, and housing

finance) during the early 1980s.
Struyk, Raymond J.; Margery A. Turner; and Makiko Ueno. Future U.S. Housing Policy:
Meeting the Demographic Challenge. Washington, D.C.: Urban Institute, 1983.
Discusses the implications of demographic changes (changes in family structure and
the aging of society) for federal housing policy.
U.S. Department of Commerce, Annual Housing Survey for the United States, 1991.
Weicher, John. "Urban Housing Policy." In Current Issues Urban Economics, ed. Peter
in
Mieszkowski and Mahlon Straszheim. Baltimore: Johns Hopkins University Press.

1979.
Condition of the Housing Stock and Neighborhood Dissatisfaction
Boehm, Thomas P., and Keith R. Ihlanfeldt. "The Revelation of Neighborhood Preferences:
An N-Chotomous Multivariate Probit Approach." Journal of Housing Economics 1
(1991), pp. 33-59.
Irby, Iredia. "Attaining the Housing Goal." Washington, D.C.: U.S. Department of Housing
and Urban Development, Division of Housing and Demographic Analysis, July 1986.
Describes the condition of the housing stock in 1983.
Supply versus Demand Policies

Maisel, Sherman. "The Agenda for Metropolitan Housing Policies." In American Domestic
Priorities —An Economic Appraisal, ed. John M. Quigley and Daniel L. Rubinfeld.
Berkeley: University of California Press, 1985. Discusses housing policies in the
middle 1980s.
Olsen, Edgar. "Housing Programs and the Forgotten Taxpayer." The Public Interest 66
(1982), pp. 97-109. Discusses alternative housing policies from the perspective of
taxpayers.
Welfare Effects of Public Housing
Aaron, Henry, and George M. von Furstenberg. "The Inefficiency of Transfers in Kind: The
Case of Housing Assistance." Western Economic Journal 9 (1971), pp. 184-91.
Estimates a welfare loss from public housing of between 10 percent and 18 percent.
Barton, David M., and Edgar O. Olsen. "The Benefits and Costs of Public Housing in New
York City." Institute for Research of Poverty Discussion Paper 373-76, University of
Wisconsin, Madison, 1976. Estimates the welfare loss from public housing (relative to
cash transfers) as 25 percent of program cost.
Kraft, John, and Edgar O. Olsen. "The Distribution of Benefits from Public Housing." In
The Distribution of Benefits from Public Housing. Vol. 41 of Studies in Income and
Wealth, ed. F. Thomas Juster. New York: National Bureau of Economic Research,
1977. Estimates the welfare losses from using public housing instead of cash transfers
as 27 percent of the program cost.
Murray, Michael P. "The Distribution of Tenant Benefits in Public Housing." Econometrica
43 (1975), pp. 771-88. Estimates the welfare losses from using public housing instead
of cash transfers as 16 percent to 20 percent of the program cost.
Muth. Richard F. Public Housing: An Economic Evaluation. Washington, D.C.: American

Enterprise Institute, 1973. Estimates the welfare losses from using public housing
instead of cash transfers as 45 percent of the program cost. Argues for the replacement
of traditional housing programs with rent certificates, which would operate in the same
way as food stamps.
National Housing Review. Housing in the Seventies. Washington. D.C.: U.S. Government
Printing Office. 1974. Discusses the equity and efficiency effects of housing
programs.
Smeeding, Timothy M. "Alternative Methods for Evaluating Selected In-Kind Transfer
Benefits and Measuring Their Effect on Poverty." Technical Paper no. 50. U.S. Bureau
of the Census, Washington, D.C., 1982. Estimates the value of in-kind transfers to the
recipients. Estimates that $1 of public housing is worth 80 cents to the typical recipient.
Smolensky, Eugene. "Public Housing or Income Supplements: The Economics of Housing
for the Poor.'* Journal of the American Institute of Planners 34 1968), pp. 94-101. (
Market Effects of Public Housing

Ohls. James C. "Public Policy toward Low-Income Housing and Filtering in Housing
Markets." Journal of Urban Economics 2 (1975), pp. 144-71. Uses a computer model
to simulate the effects of public housing on the filtering process.
Experimental Housing Allowance Program (EHAP)

Aaron, Henry J. "Policy Implications: A Policy Report." In Do Housing Allowances Work?
ed. KatherineBradbury and Anthony Downs. Washington. D.C.: Brookings Institution.
1981. Compares the effect of allowances to cash transfers and construction subsidies.
Discusses the value of social experiments such as EHAP.
Allen. Garland E.: Jerry J. Fitts: and Evelyn "The Experimental Housing
S. Glatt.
Allowance Program." In Do Housing Allowances Work? ed. Katherine Bradbury and
Anthony Downs. Washington. D.C.: Brookings Institution, 1981. A description of the
housing allowance experiments.
Downs, Anthony, and Katherine Bradbury. "Conference Discussion." In Do Housing
Allowances Work? ed. Katherine Bradbury and Anthony Downs. Washington. D.C.:
Brookings Institution. 1981. A summary of the discussion of a conference on EHAP.
Frieden. Bernard J. "Housing Allowances: An Experiment That Worked." The Public
Interest 59 ( 1980). pp. 15-35. A summary of the results of the allowance experiment.
Hanushek. Eric A., and John M. Quigley. "Consumption Aspects." In Do Housing
Allowances Work? ed. Katherine Bradbury and Anthony Downs. Washington. D.C.:
Brookings Institution. 1981. Discusses the responses of EHAP participants to housing
allowances, focusing on the effects on housing consumption and the quality of housing.
Kam. John F. "A Universal Housing Allowance Program." In Do Housing Allowances
Work.' cd. Katherine Bradbury and Anthony Downs. Washington. D.C.: Brookings
Institution, 1 98 1 . Discusses the implications of EHAP for a national housing-
allowance program. Suggests that EHAP was not a true test of a national program
because participation rates and housing-quality standards were relatively low.
Mills. Edw in S.. and Arthur M. Sullivan. "Market Effects." In Do Housing Allowances
Work? ed. Katherine Bradbury and Anthony Downs. Washington, D.C.: Brookings
Institution. 1981. Discusses the market effects of the EHAP suppl) experiments.
Rossi, Peter H. "Residential Mobility." In Do Housing Allowances Work? ed. Katherine

Bradbury and Anthony Downs. Washington. D.C.: Brookings Institution, 1981.
Discusses the effects of EHAP on household mobility.
Straszheim, Mahlon R. "Participation." In Do Housing Allowances Work? ed. Katherine
Bradbury and Anthony Downs. Washington, D.C.: Brookings Institution, 1981.
Discusses the factors that explain the relatively low participation rates in EHAP.
Struyk, Raymond J., and Mac Bendick. Housing Vouchers for the Poor. Washington, D.C.:
Urban Institute, 1981. A detailed description of the Experimental Housing Allowance
Program.
Rent Controls
De Salvo, J. S. "Reforming Rent Controls in New York City." Regional Science Association
Papers 2 (1971), pp. 195-227.
Gyourko, Joseph, and Peter Linneman. "Equity and Efficiency Aspects of Rent Control: An
Empirical Study of New York City." Journal of Urban Economics 26 (1989), pp. 54-74.
. "Rent Controls and Rental Housing Quality: A Note on the Effects of New York
City's Old Controls." Journal of Urban Economics 27 ( 1990), pp. 398^09.
Lett, Monica. Rent Control: Concepts, Realities and Mechanisms. New Brunswick, N.J.:
Rutgers University Press, 1976. A general description of rent-control policies.
Malpezzi. Stephen. "Welfare Analysis of Rent Control with Side Payments: A Natural
Experiment in Cairo, Egypt." Regional Science and Urban Economics 28 (1998),
pp. 773-95.
Navarro, Peter. "Rent Control in Cambridge. Massachusetts." Public Interest 91 (1987),
pp. 83-100.
Olsen, Edgar. "An Econometric Analysis of Rent Control." Journal of Political Economy
80(1972), pp. 1081-100.
."Economics of Rent Control." Regional Science and Urban Economics
28(1998), pp. 673-78.
Rand Corporation. "The Effects of Rent Control on Housing in New York City." In Rental
Housing in New York City: Confronting the Crisis. Publication no. RM-6190-NYC.
New York, 1970.
Smith, Lawrence B., and Peter Tomlinson. "Rent Control in Ontario, Roofs or Ceilings."
American Reed Estate and Urban Economics Journal 9(1981), pp. 93- 114.
U.S. Department of Housing and Urban Development. Office of Policy Development and
Research. Report to Congress on Rent Control ( 1991 ).
Urban Renewal
Anderson, Martin. The Federal Bulldozer. Cambridge, Mass.: MIT Press, 1964. Argues that
urban renewal destroyed low-income housing, displacing poor households.
Davis, Otto, and Andrew B. Whinston. "The Economics of Urban Renewal." In Urban
Renewal: The Record and the Controversy, ed. James Q. Wilson. Cambridge, Mass.:
MIT Press, 1966, pp. 50-67.
Rothenberg, Jerome. Economic Evaluation of Urban Renewal. Washington, D.C.:
Brookings Institution, 1967.
Weicher, John. Urban Renewal: National Program for Local Problems. Washington, D.C.:
American Enterprise Institute, 1973.
Weiss, Marc A. "The Origins and Legacy of Urban Renewal." In Federal Housing Policies
and Programs, ed. J. Paul Mitchell. New Brunswick. N.J.: Center for Urban Policy
Research. 1985. Discusses the politics of urban renewal and housing policies.
Community Development
U.S. Department of Housing and Urban Development. 1988 Community Development
Programs: State Reports. Washington. D.C.: Government Printing Office. 1989. Shows
HUD"s community development budget and describes how cities and states have used
federal community development funds.
Part IV
OCAL GOVERNMENT
(J7~ n contrast with the federal government, whose urban policies are primarily
t_y redistributional in nature, local governments are involved in resource alloca-
tion. Local governments build the streets, provide mass transit, run the public school
system, and manage the criminal justice system. Local spending programs are sup-
ported by taxes on property, income, and retail sales. This section discusses the
efficiency and distributional effects of local spending and tax policies. Chapter 16
provides an overview of local government, explaining the basic reasons for local
government. Chapter 17 discusses how local budget decisions are made, focusing
on how citizens influence budget decisions through their voting and location choices.
Chapter 18 deals with local revenue sources, examining the incidence of local taxes
and the local responses to intergovernmental grants.
Chapter 16
& V £ R V I E W of LOCAL GOVERNMENT
A community is like a ship: everyone ought to be prepared to take the helm.

Henrik Ibsen
his chapter provides an overview of the taxing and spending policies of the
(~J7~
t_x local public sector. The first part of the chapter presents the facts on local
government, including information on spending programs and revenue sources. The
second and third parts discuss the economic rationale for local government, explain-
ingwhy goods such as public schooling, public-safety services, parks, and transit
systems are produced by local governments rather than private firms or higher levels
of government. The fourth part explores the trade-offs associated with metropolitan
consolidation. The fifth part discusses the reasons for —and responses — to the rising
cost of local public services. The final part explores the reasons for the fiscal distress
experienced by many central cities in the last few decades.
Local-Government Facts
Local governments are creatures of state governments in the sense that localities
derive their taxing and spending powers from the states. There is considerable vari-
ation across states in the amount of power delegated to local governments. One
measure of the relative importance of local government in a particular state is the
ratio of local spending to state spending. For the United States as a whole in 1992,
the ratio of local to state spending was 1 .38 (U.S. Census Bureau, 1997). The ratio of
local to state spending varies considerably from state to state, from a low of 0.28 in
Hawaii to a high of 2.27 in Nevada. Some of the states with relatively low spending
ratios were Delaware (0.53), Alaska (0.54), Rhode Island (0.62), Vermont (0.65),
and Massachusetts (0.79). Among the states with the highest local spending ratios
are California (2.14), Florida (2.10), New York (1.78), and Texas (1.74).
449
450 Part IV Local Government
Table 16-1 Types of Local Governments
Number of Percent of Local

Type of Local Government Units in 1992 Expenditures
County 3,043 23
Municipal 19.279 33
Township and town 16.656 3
School district 14,422 30
Special district 3 1 .555 11
Total S4.995 100
SOURCES: U.S. Bureau of the Census. Statistical Abstract of the United States
1997, Table 474 (Washington. D.C.: U.S. Government Printing Office. 1998); U.S.
Bureau of Census. Census of Governments, Government Finances 1991-92
(Washington, D.C.: U.S. Government Printing Office. 1994).
As shown in Table 16-1 , there are almost 85,000 local governments in the United
States. In terms of total expenditures, the most important types of local governments
are municipalities (33 percent of local-government expenditures) and school districts
(30 percent of local expenditures).
Table 16-2 shows the spending of local governments and municipalities on
various goods and services. The largest spending category for local government is
education, with $961 per capita, about 42 percent of total spending. The spending
of municipalities is more evenly divided, with the largest expenditures on police
protection, education, highways, sewerage, and fire protection.
Table 1 6-3 shows the distribution of revenue for different types of local govern-
ments. For local governments as a whole, 38 percent of total revenue comes from
intergovernmental grants, 39 percent comes from local taxes, and the remaining
23 percent comes from charges and general revenue. School districts are heavily
dependent on intergovernmental grants, receiving about 54 percent of their revenue
from the federal and state governments. For municipal governments, 28 percent of
revenue comes from grants, 44 percent comes from local taxes, and 28 percent comes
from user charges and general revenue. In contrast, special districts are heavily de-
pendent on user charges, receiving 55 percent of their revenue from charges and
general revenue.
Figure 1 6- shows
1 the distribution of tax revenue for municipalities and all local
governments. In 1992, the property tax generated about three-fourths of all local tax
revenue. Municipalities depend less on the property tax: for municipalities as a
whole, the property tax generated about half of tax revenue; for large municipalities
(population greater than 300,000). the property tax generated only about 46 percent
of tax revenue. The second largest revenue source is the sales tax: it generated
15 percent of total tax revenue and 27 percent of municipal tax revenue. The third
largest source is the local income tax, which generated 5 percent of local taxes and
13 percent of municipal taxes.
Intergovernmental grants support dozens of local programs. The programs re-
ceiving the most grant money are education (about 57 percent of all intergovern-
mental grants), public welfare (13 percent), housing and community development
Chapter 16 Overview of Local Government 451
Table 16-2 Per Capita Expenditures of Local Government
Annual Spending per Capita, 1991-1992
All Local Municipalities and

Expenditure Category Governments Townships
Education Services
Education $961 $81
Libraries 18 10
Social Services
Public welfare 133 33
Hospitals 18 30
Health 56 12
Transportation
Highways 108 64
Other 40 19
Public Safety
Police protection 123 90
Fire protection 59 47
Other 51 16
Environment and Housing

Natural resources 12 1
Parks and recreation 54 36

Housing and community development 64 35
Sewage services 81 51
Solid waste management 45 31
Administration and Finance 125 58
General Expenditure 120 73
SOURCE: U.S. Bureau of the Census. Compendium of Government Fincmaes. no. 5 (Washington.
D.C.: U.S. Government Printing Office. 1997).
Table 16-3 Revenue Sources of Local Governments, 1991- -1992
Percent of General Revenue from
Charges and
Intergovernmental Local General
Type of Government Grants Taxes Revenue Total
County 37% 38% 25% 100%

Municipality 28 44 28 100
Township 24 60 16 100
School district 54 38 8 100
Special district 29 16 55 100
All local governments 38 39 23 100
SOURCE: U.S. Bureau of the Census. Compendium of Government Finances, no. 5 (Washington,
452 Pari IV Local Government
Figure 16-1 Sources of Tax Revenue for Local and Municipal

Governments
80 76
70
60
'? 5()
> 40
£ 30 27
S 20
-
10
Property Sales Income

&i
Other taxes
All local Municipal

government government
SOURCE: U.S. Bureau of the Census. Compendium qj Government Finances, no. 5 (Washington,
DC: U.S. Government Printing Office. 1997).
(4 percent), highways (4 percent), and health and hospitals (4 percent). For munic-
ipalities, the programs receiving the most grant money are education (20 percent
of total grants), public welfare (14 percent), housing and community development
(8 percent), and highways (7 percent). About 20 percent of grant money to munici-
palities is used for the general support of municipal operations (U.S. Bureau of the
Census, 1997).
The Role of Local Government

What is the role of local government in the market economy? This question can be
answered by asking two separate questions. First, what is the role of government
in the market economy? Second, how does local government tit into the general
scheme of government involvement?

Musgrave and Musgrave 1980) distinguish between three types of government
(
policies. The first is stabilization policy: the government uses monetary and fiscal pol-
icy to control unemployment and inflation. Under the second type of policy, income
redistribution, the government uses taxes and transfers to alter the distributions of
income and wealth. The third policy is resource allocation: the government makes de-
cisions about what to produce and how to produce it. When the government actually
produces a particular good or service, it makes these resource-allocation decisions

directly. When the government subsidizes or taxes private activities, it influences the
resource-allocation decisions of the private sector.
How do local governments fit scheme of governmental ac-
into this three-part
tivity? The assumed by the national
responsibility for stabilization policy has been
government for two reasons. First, although each local government could print its
own money and execute its own monetary policy, such a system would be chaotic.
Instead, the national government prints the money and manages a national monetary
policy. Second, because a large fraction of local income is spent on goods produced
outside the local area, local monetary and fiscal policies would be relatively weak
and ineffective. Consider a city that wants to stimulate the local economy, and tries
to do so by cutting its taxes without changing its spending on local public goods.
This deficit-spending policy will be ineffective because a large fraction of the tax cut
will be spent on imports, so the tax cut will have a relatively small local multiplier
effect. Fiscal policy is more effective at the national level because a relatively small
fraction of national income is spent on imports, so a national tax cut has a relatively
large multiplier effect.
Consider next the distribution role of government. Local attempts to redistribute
income will be frustrated by the mobility of taxpayers and transfer recipients. Sup-
pose that a city imposes a tax on its wealthy citizens and uses the tax revenue to
finance transfer payments to the poor. To escape the tax, some wealthy households
will leave the city, causing a decrease in total tax revenue. At the same time, some
poor households will migrate to the relatively generous city, causing a decrease in
the transfer payment per recipient. In combination, the fleeing of the wealthy and
the migration of the poor weakens the city's redistribution program: there is less
money to transfer to more poor households. A program is
national redistribution
more effective because there is less mobility between nations than between cities.
The third role of government is resource allocation. As was shown in
Table 16-2, local governments are responsible for the provision of several goods and
services, including education, highways, police and fire protection, parks, and sew-
ers. Under what conditions can government provision of these goods be justified on
efficiency grounds? There are three possibilities. First, the good may be produced un-
der conditions that generate a natural monopoly. Second, the good may generate posi-
tive externalities. Finally, the good may have the characteristics of a local public good.
Natural Monopoly
A natural monopoly occurs if the production of a particular good is subject to rela-

economies. In Figure 16-2, the long-run average total-cost curve
tively large scale
for sewage services is negatively sloped over a wide range of output, reflecting the
fact that ( 1 ) there is a substantial fixed cost associated with setting up a sewage sys-
tem and (2) the marginal cost of adding another customer is relatively small. Because
the average-cost curve is negatively sloped, the marginal cost is less than average
cost (the marginal cost pulls down the average cost as quantity increases). Sewage
service is a natural monopoly in the sense that a single sewage authority (providing
454 Part PV Local Government
Figure 16-2 Natural Monopoly in Sewage Services
Demand = marginal
social benefit
Long-run average cost
Lone-run marginal cost
Quantity of sewage services
Sewage service is a natural monopoly: scale economies generate a negatively sloped long-run average-cost curve.
The optimum output is where the demand curve intersects the marginal-cost curve. Since the optimum price is less
than the average total cost, the sewage system generates a deficit. The government could provide sewage sen ice
itself, covering the deficit with general tax revenue. Alternatively, the government could regulate a private
sewage authority.
5* units of service) could serve the city at a lower average cost than several sewage
authorities (for example, five firms, each supplying 5' units).
What is the optimum level of sewage service? The basic efficiency rule is that
an activity should be increased to the point at which the marginal social benefit of
the activity equals the marginal social cost. In Figure 16-2. the demand curve, which
shows consumers' willingness pay for different levels of sewage service, is also
to
the marginal-social-benefit curve. The optimum output S* is the output at which the( )
demand curve intersects the marginal-cost curve. One problem with an output of 5*
is that a firm producing this output would lose money: to get consumers to purchase
S* units of sewage service, the firm would charge a price of P*, which is less than
the average total cost of producing 5*. The deficit is shown by the shaded area.
The government has several options. First, it can provide the service itself, charge
a price of P* and cover the deficit with general tax revenue. Second, it can get a
private firm to provide sewage service. If the government forces the firm to charge a
price of P*, subsidies are required to cover the deficit. A final option is to allow the
private firm to charge P" instead of P*. At P". average total cost equals the price,
so the firm would cover all of its cost. Under this scheme, the output (S") is less than
the optimum output.

Local governments have the same options for other natural monopolies. Chap-
ter 20 (Mass Transit) discusses the pricing and regulation of mass transit. For
other natural monopolies (water systems, cable television networks, telephone net-
works), the government can either produce the service itself or regulate a private
producer.
What level of government should operate or regulate a natural monopoly? Should
sewage and transit services be run by local, metropolitan, regional, state, or national
governments? The appropriate level of government is determined by the extent of
scale economies: the greater the scale economies, the larger the political jurisdiction
required to realize the scale economies associated with providing the service. Local
governments are responsible for the provision and regulation of services for which
scale economies are exhausted at the local level.
Externalities
One of the roles of government is to internalize externalities. An externality occurs

ifone person's consumption of a good generates benefits or costs for other people.
The basic efficiency rule is that an activity should be increased to the point at which
the marginal social benefit equals the marginal social cost. An externality causes a
divergence between private and social benefits or between private and social costs.
Individual consumers, who base their decisions on private benefits and costs, make
choices that are inefficient from the social perspective. When this occurs, government
intervention may promote efficiency.
Education Externalities. Figure 16-3 shows the private and social benefits of
education. Education generates private benefits (for the student himself) and external
benefits (for the student's fellow workers and citizens). As a person becomes more
educated, he becomes a better team worker: he understands instructions more readily
and is more likely to suggest ways to improve production. In other words, education
generates workplace externalities. In addition, a more educated citizen makes better
choices on election day, generating external benefits for his fellow citizens. Because
education generates workplace and civic externalities, the marginal social benefit
of education exceeds the marginal private benefit. In the absence of government
intervention, the worker-citizen ignores these external benefits, choosing £" units of
education instead of the optimum (£*).
The government has two options to solve the externality problem. The first is to
take responsibility for providing the good. A system of free compulsory education
could encourage citizens to consume more education. Alternatively, the government
could subsidize education, encouraging citizens to consume more education while
allowing them to choose their own schools. In Figure 16-3, the subsidy would be
equal to the gap between the private and social benefits of education. The subsidy
would increase consumption to E* units of education. As discussed in Chapter 21
(Education), there is heated debate over whether the government should supplement
the public school system with subsidies for private education (tuition tax credits and
education vouchers).
456 Part TV Local Government
Figure 16-3 Externalities in the Consumption of Education

Marginal private benefit
Marginal social benefit
Marginal cost
£' £*
Quantity of education
There are externalities in the consumption of education, so the marginal social benefit
exceeds the marginal private benefit. In the absence of government involvement, the student
chooses E (where the marginal private benefit equals the marginal cost). The optimum level
of education is E*, which the student will choose if the government provides a subsidy equal
to the gap between the marginal private and marginal social benefit.
What level of government should deal with education externalities'? Should the
public school system (or education subsidies) be managed by local, metropolitan,
regional, state, or national governments? One factor in determining the optimum level
of government is the geographical extent of the externalities. If people throughout
the country benefit from the education of children in a particular school, national
participation in education decisions (national schools or a national subsidy) may
promote efficiency. If the benefits are confined to a particular city, local decision
making may be efficient. As explained later in the chapter, there are other factors
optimum level of government.
that affect the
Public-Safety Externalities. There are two types of externalities from spending

on police services, positive and negative. Both externalities result from the fact that
criminals are mobile: they can move freely from one jurisdiction to another. The
positive externality occurs when one municipality uses its police budget to capture
a criminal: by doing so. the municipality generates benefits for surrounding mu-
nicipalities, so the marginal social benefit of police spending exceeds the marginal
local (municipal) benefit. The negative externality occurs when one municipality's
crime-fighting activities cause criminals to victimize people in surrounding munici-
palities. In other words, the police sometimes chase criminals from one municipality
to another. In this case, police spending just moves crime around, so the marginal
local benefit of police spending exceeds the marginal social benefit.
What is the optimum level of government for police protection? One factor is
the geographical extent of the positive and negative externalities, which depends on
the mobility of criminals. The greater the mobility of criminals, the larger the op-
timum jurisdiction.As explained later in the chapter, there are other factors that
affect the optimum level of government. The typical response to these positive
and negative externalities is to provide police services through municipal govern-
ments.
The other public-safety service, fire protection, also generates externalities. Fires
can spread from one house to another, so the marginal social benefit of fire protection
exceeds the marginal private benefit. In most metropolitan areas, fire protection is
provided by local governments. Some municipalities contract with private firms to

provide fire protection. For the trade-offs between public and private production, see
Savas (1982) and Stevens (1984).
Characteristics of a Local Public Good

A local public good has three characteristics. First, it is nonrivalrous in consumption:
several people consume the same good at the same time. Second, it is nonexcludable:
it is impossible or impractical to exclude some people (e.g., those who do not pay
for the good) from consuming the good. Third, the benefits of the good are confined
to a relatively small geographical area.
Nonrivalrous Consumption (Joint Consumption), Consider first the issue of

rivalry in consumption. A private good (a hot dog) can be consumed by only one
person. In contrast, a public good can be consumed by many people at the same
time. For example, hundreds of people can watch a parade without interfering with
each other's enjoyment of the parade. In other words, a parade is subject to joint
consumption: everyone can consume the same good. The parade is a pure lo-
cal public good if the marginal cost of an additional parade viewer is zero. This
occurs if an additional viewer does not decrease the enjoyment of other parade
viewers.
Many of the goods provided by local governments are semirivalrous. If the num-
ber of consumers becomes large enough, an individual consumer's benefit decreases.
For example, if the city does not set up bleachers for parade viewers, tall people block
the views of short people. This is a problem only if the number of parade viewers is
large enough do not get a front-row view. In this case, the parade
that short citizens
is an impure or congestable public good. Other examples of impure public goods
are city streets and highways: during the peak travel periods, travel speeds decrease
as traffic volume increases, so the benefit of using the highway decreases as the
number of consumers increases.
What about police and fire services? Although two citizens cannot use a police
officer or a fire truck at the same time, public-safety systems are designed to respond
quickly to emergencies. Most cities have enough police officers and fire trucks to
simultaneously handle a reasonable number of calls for help. If everyone called
at once, consumption would of course be rivalrous, but not everyone needs safety
services at the same time.
Nonexcludability. The second characteristic of a local public good is that it would

be impractical to exclude some people from consuming the good. In other words, the
good is nonexcludable. Consider the parade example. Although it may be possible to
charge everyone for watching the parade and exclude those who don't pay, it would
be very costly to do so. The government would have to first empty the streets and
buildings along the parade route, and then admit people who paid a parade-viewing
fee. If some people promised watch the parade, they might be permitted to
to not
enter a building along the route, but the governmentwould have to find a way to
prevent them from sneaking a look. Such a scheme would obviously be impractical.
What about streets and public parks? One way to charge people for using city
streets would be to set up a system of toll booths, one on each corner. Such a system
would be costly: the city would have to hire a large number of toll collectors, and
road users would incur a substantial time cost in stopping at every corner to pay the
tolls. Chapter 19 (Autos and Highways) describes a low-cost system for charging
drivers for using streets and highways. There is little political support for such a
system, so streets are likely to remain nonexcludable. The same argument applies
to parks. Although it might be possible to install fences and turnstiles in city parks,
such a system would be relatively costly.
What about public-safety services? For some safety activities, it would be easy
and send them a bill for the safety services rendered. For
to identify the beneficiaries
example, the fire department could charge a fee for putting out fires, and the police
could charge a fee for responding to burglaries and robberies. Under such a system,
the fire and police departments would not provide services to people who could not
afford the service fees. This system of user charges is politically infeasible because
it would most people's notions of equity and fairness. For other types of
violate
safety activities, it would be more difficult to identify the beneficiaries and charge
for the safety services. For example, when a city captures, prosecutes, and imprisons
a criminal, every potential crime victim benefits. Other examples are crowd and
traffic control. For these services, it would be impractical to exclude people who do
not pay.
Localized Benefits. The third characteristic of a local public good is that its ben-
efits are confined to a relatively small geographical area. Unlike national defense,
which generates benefits for the entire nation, most of the benefits of the local police
force and local fire department go to local citizens. Similarly, local citizens get most
of the benefits from local streets and highways. The appropriate size of the jurisdic-
tion is determined by the "localness" of the public good (the geographical extent of
the benefits from the public good): the more extensive the benefits, the larger the
jurisdiction required to contain all the beneficiaries.
The Optimum Quantity of a Local Public Good

What is theoptimum level of a local public good? According to the basic efficiency
rule, the optimum level of an activity is the amount at which the marginal social
Chapter Hi Overview of Local Government 459
Figure 16-4 Optimum versus Equilibrium Provision of a Local Public Good

$
103
\^ Marginal social benefit = MB S

+ MB m + MB h
60 Marginal social cost
53
[\WA/,
30 : ^r\Mfi„,
20
__ ^^\^ Marginal cost with
equal cost shares
MB,
-»«^_ 1
^^
~~~~~~~~~--4~---J
1 ir*"---
1 1 1
1 1 1
Park acreage
The marginal social benefit of park acreage is the sum of the private benefits of three citizens. The optimum park
acreage is A' (the amount such that the marginal social benefit equals the marginal social cost).
benefit equals the marginal social cost. To determine the optimum level, we must
determine the marginal social benefit of the local public good.
Consider a three-person city that must decide how many acres of public parks to
provide. In Figure 16-4, the marginal-benefit curves show how much each citizen is
willing to pay for different amounts of parks: Sam has a low demand for parks and
prefers relatively small parks (MB S ); Marian has a medium demand (MB,,,): Bertha
has a high demand and prefers big parks (MBb). Suppose that city parks have the
following characteristics:
1 . No congestion. In the relevant range of park use, park acreage is
nonrivalrous: park usage is low enough that there is no congestion.

2. No intercity spillovers. The city's park does not generate benefits or costs
for people outside the city.
If these two conditions are satisfied, the park is a pure local public good. The social
benefit of a given park size is the sum of the marginal benefits of Sam, Marian,
and Bertha: if there is no congestion, the three citizens do not affect their fellow
citizens' enjoyment of the park: if there are no spillovers, there are no other potential
beneficiaries to worry about. For A s acres of land, the marginal social benefit is
$103 (Sam's marginal benefit is $20. Marian's is $30. and Bertha's is $53). The
marginal-social-benefit curve is the vertical sum of the individual marginal-benefit
curves.
At the optimum, the marginal social benefit of park acreage equals its marginal
social cost. In Figure 1 $60 per acre, so the optimum acreage
6-4, the marginal cost is
is A*. For any amount less than A*, citizens in the city would be willing to pay more
than $60 per additional acre, so an increase in size would increase social welfare.
For example, suppose the city starts at A s At this point, the willingness to pay for
.
parks (from the social-benefit curve) is $103 and the cost is only $60, so an increase
in park acreage would generate a large net gain. Similarly, for any amount exceeding
A*, the willingness to pay would be less than the social cost, so a smaller park would
be more efficient.
How do local governments decide how much of local public goods to provide?
Chapter 17 discusses two methods for determining how much of a local public good
to provide. One approach is to hold an election and let voters choose how much
to spend on local public goods. Another approach is to allow citizens to vote with
their feet by moving to the jurisdiction that provides the best combination of public
services and taxes. One of the issues examined in Chapter 17 is whether either of
these approaches generates the efficient amount of local public goods.
Federalism and the Optimum Level of Government

Under the federal system of government, the responsibility for providing public
goods is divided between the federal, state, and local governments. Some goods,
such as defense and space exploration, are provided at the national level. Others,
such as education and police protection, are provided at the local level. Why are some
public goods provided by local governments rather than state or federal governments?
Oates ( 1972) discusses the advantages and disadvantages of the local provision
of public goods.
1. Diversity in demand. A system of small local governments allows com-

munities to choose different levels of spending on public goods (parks,
public safety, education, libraries). Because local governments can
accommodate the diverse demands for public goods, they promote
efficiency.
2. Externalities. If the local jurisdiction is not large enough to contain all the
people affected by its spending programs, there are externalities (benefit
spillovers). Under a system of small local governments, local voters ignore
the benefits that accrue to people outside the jurisdiction and therefore
make inefficient choices.
3. Scale economies. If there are scale economies in the provision of public

goods, a system of small local governments has a relatively high production
cost.
The local provision of a public good is efficient if the advantages outweigh the
disadvantages. In other words, local provision is efficient if ( 1 ) diversity in demand
is relatively large, (2) externalities are relatively small, and (3) scale economies are
relatively small.
Externalities versus Demand Diversity

Consider first the trade-off between internalizing externalities and accommodating
diverse demands for local public goods. Suppose that because of interjurisdictional
mobility, public education generates benefits for citizens outside the jurisdiction:
once a citizen leaves her local school district, her co-workers and fellow workers in
another jurisdiction benefit from her education. Because the municipality considers
all the costs of the local public good but ignores some of the benefits, the local
government provides the good at an inefficiently low level. The inefficiency occurs
because the municipality is too small to contain all the beneficiaries of its spending
program.
The same analysis applies to other local public goods that generate externalities.
When the police capture a criminal, potential victims in surrounding municipalities
benefit, so the social benefit of police spending exceeds the local benefit. If house-
holds cross municipal borders to use the services of other local governments, the
social benefit of these services exceeds the local benefit. These interjurisdictional
spillovers cause inefficient choices by municipalities and school districts.
One solution to the spillover problem is a system of subsidies from a higher
level of government. If the municipality receives a subsidy equal to the marginal
external benefit of the public good, it bases its spending decisions on the marginal
social benefit of the good. The subsidy increases the municipality's benefit to the
level of the marginal social benefit, so the municipality increases its spending to
the optimum level. Chapters 18 (Local Taxes and Intergovernmental Grants) and 21
(Education) explore the effects of various types of intergovernmental grants on local
spending.
An alternative response to the externality problem is metropolitan consolidation.
The idea is that the jurisdiction could be expanded to include all the people affected
by local-government policies. For example, a metropolitan area could consolidate all
of its municipalities, school districts, and special districts into a single metropolitan
government that would make spending decisions for the entire metropolitan area.
What are the trade-offs associated with metropolitan consolidation? The prob-
lem with a metropolitan government is that it chooses a single spending level for
all households in the metropolitan area. Under the metropolitan government, exter-
nalities are internalized, but since a single spending level must be chosen, citizens
are forced to compromise on their ideal spending levels. Metropolitan governments
will be more efficient than municipal governments if interjurisdictional spillovers
462 Part [V Local Government
are large relative to diversity in demand. In this case, the advantages of small local
governments (the ability to accommodate diverse demands for local public goods)
are relatively small, and the disadvantages (the inefficiencies associated with exter-
nalities) are relatively large. Therefore, a metropolitan system of government will
be more efficient.
Scale Economies versus Demand Diversity

There is also a trade-off between demand diversity and scale economies. Consider a
metropolitan area with a single public good (fire protection) and two municipalities.
The metropolitan area has the following characteristics:
1 . Diversity in demand. The demand for fire protection is relatively high in

municipality H and relatively low in municipality L.
2. No externalities. The benefits of fire protection do not cross municipal

borders.
3. Scale economies. The average cost of fire protection decreases as the

number of fire trucks increases. Therefore, a metropolitan fire department
can produce fire protection at a lower average cost than a municipal fire
department.
4. Uniformity. Under a metropolitan government, the two municipalities have
identical fire protection programs.
Under a system of municipal governments, each municipality makes its own deci-
sions on fire protection. The decisions are based on the local benefits and costs of
fire protection.
Suppose that municipality L has four fire trucks (low demand), while munici-
pality H has eight The two municipalities merge to form a metropolitan
fire trucks.
government. Each community gets six fire trucks (the average number of fire trucks
that would occur under municipal government) at a lower average cost. The switch to
metropolitan government brings good news and bad news. The good news is that the
average cost of fire protection decreases. The bad news is that both municipalities
must compromise on their ideal level of fire protection, with H getting too few trucks
(six instead of eight), and L getting too many (six instead of four). Metropolitan con-
solidation will be efficient if the savings in production cost are large relative to the
losses resulting from the uniform provision of fire protection. In other words, con-
solidation will be efficient if scale economies are large relative to the diversity in
demand for fire protection.
What are the facts on scale economies in the provision of local public goods?
There have been dozens of studies of the relationship between production costs and
jurisdiction sizes. The evidence suggests that there are moderate scale economies
in the provision of water and sewage services. Because these sen ices are capital-
intensive, average cost decreases as population increases. In contrast, most studies
of other local public goods (police protection, fire protection, schools) suggest that
scale economies occur up to a population of about 100.000.
Metropolitan Consolidation
Should local governments be consolidated into metropolitan governments? Four
arguments have been advanced in support of metropolitan consolidation.
Prevention of Suburban Exploitation
According to the suburban exploitation hypothesis, suburban commuters do not pay

their "fair share" of the cost of central-city public services. Consider a worker who
commutes from her suburban home Although she uses city services
to the city.
(streets, water, fire protection, police protection), she pays her property taxes to a
suburban government. Does the commuter "exploit" central-city residents by using

city services without paying for them?
The exploitation hypothesis ignores the fact that commuters pay taxes to the city.
As shown in Chapter 8, over two-fifths of municipal tax revenue comes from sales
1
and income taxes, meaning that suburbanites who work and shop in the city center pay
taxes to city governments. For large cities (population greater than 300,000), sales
and income tax generate over half of municipal tax revenue. In addition, most cities
have user fees for public services (tolls, admission prices for zoos and museums),
so suburbanites pay directly for some city services. Most studies of the exploitation
hypothesis suggest that suburbanites pay close to their share of the cost of city
services (see Bradford and Oates, 1974, for a review of these studies).
Exploitation of Scale Economies
Another argument advanced in favor of consolidation is that a metropolitan govern-

ment can exploit scale economies to provide public services at a lower cost. The
scale-economy argument is incomplete. The switch to metropolitan government re-
quires communities to compromise on their ideal levels of public goods, so the cost
savings from scale economies must be compared to the welfare losses resulting from
the uniform provision of public goods. Metropolitan consolidation will be efficient
if scale economies are large relative to the diversity in demand.
Equalization of Tax Rates
Under a system of fragmented local government, municipalities with large tax bases
often have relatively low tax rates. Under a metropolitan government, there would
be a single property-tax base and thus a single-tax rate. Therefore, the switch from a
municipal system to a metropolitan government would increase tax rates in relatively
wealthy communities and decrease tax rates in relatively poor communities.
To what extent does the equalization of tax rates redistribute income from
the wealthy to the poor? Consider a metropolitan area with two municipalities,
one with middle-income households and one with poor households. Everyone in the
poor municipality lives in rental housing, and every household in the middle-income
municipality owns the home it occupies. The poor community has a low tax base
and a high tax rate, and the middle-income community has a large tax base and a
low tax rate. Therefore, metropolitan consolidation increases property taxes in the
middle-income community and decreases taxes in the poor community.
The changes in tax liabilities affect housing prices and property values in the
two communities. In the poor community, the tax cut increases the profitability of
housing, so the supply of housing increases and the price of housing (monthly rent)
falls. Therefore, tenants benefit from the tax cut. Property owners also benefit: the
decrease in the tax on rental housing increases the net income from rental property,
increasing market values. This is the capitalization process: changes in tax liabilities
are capitalized into property values. In the middle-income community, homeowners
face higher tax liabilities, so their property values fall.
Who gains and who loses as a result of the equalization of taxes? Tenants and
landlords in the poor community gain: the tenants pay lower rent, and the landlords
have higher property values. Homeowners in the middle-income community lose
because their property values fall. To predict the distributional effect of the equaliza-
tion of tax rates,one needs information on the income levels of tenants, landlords,
and homeowners. If landlords in the poor community are relatively wealthy, the
equalization of tax rates could actually increase income inequality.
Equalization of Spending
Under the fragmented system of local government, spending varies within the
in the demands for public services. Con-
metropolitan area, reflecting variation
solidating localgovernment would equalize spending on local public goods. Be-
cause spending on public services typically increases with income, metropolitan
consolidation would narrow the spending gap between wealthy and poor comm-
unities.
Bradford and Oates ( 1974) explore the effects of metropolitan consolidation in
northeastern New Jersey. They predict the effects of consolidating 58 local juris-
dictions (5 central cities and 53 suburban governments) into a single metropolitan
government.
Bradford and Oates evaluate the trade-offs associated with the equalization
of educational spending. To explain their results, suppose that there are only two
school districts, a wealthy one and a poor one. Before consolidation, the wealthy
community spends $3,000 per pupil on education, and the poor one spends $2,600.
Suppose that under the consolidated government, spending is equalized at S2.800
per pupil: spending increases by $200 in the poor district and decreases by the same
amount in the wealthy According to Bradford and Oates. the extra $200
district.
of education spending worth only about $130 to the typical poor household: the
is
household would be indifferent between receiving $200 in educational spending

and $130 in cash. In contrast, the typical wealthy household would be indifferent
between losing $200 in educational spending and giving up $230 in cash. Therefore,
the equalization program produces a SI 30 benefit (to poor households) at a cost of
$230 (to wealth) households).
Using the Bradford-Oates numbers, the equalization program seems to be fool-

ish. Both the poor and the wealthy would be better off if the government used cash
transfers —not education spending —
to redistribute income. For example, if the gov-
ernment took just $130 from the wealthy and gave it to the poor, the poor would be
just as well off, but the wealthy would be better off by $100.
This simplistic analysis is deceptive. As pointed out by Bradford and Oates,
these calculations ignore the external benefits of education. As explained earlier
in this chapter, education generates externalities, so individual households spend
less than the optimum amount on education. The question is whether the external
benefits of the equalization program are large enough to offset the private costs.
Suppose that the equalization program increases the educational achievement of the
poor by a relatively large amount. The net external benefit is defined as the external
benefit of extra education for the poor less the external cost of less education for the
wealthy. If the net external benefit is relatively large, the apparent inefficiency of the
equalization program disappears: the equalization program would not be foolish, but
would promote efficiency.
Another possible justification for the equalization program is that education may
be a merit good, defined as a good that is assumed to be systematically undervalued
by consumers. If individualsdo not make rational choices concerning education,
their willingness to pay for education understates the private value of education. In
the example above, perhaps the poor don't realize that they would actually receive
a benefit of $190 from the increase in education spending. If so, it wouldn't take
much of an external benefit to make the equalization program socially efficient. The
notion of a merit good can also be stated in terms of the willingness to donate: if
the donors (the people in the wealthy school district) are allowed to make choices
for the recipients (in the poor district), they are willing to donate more. They may
be willing to give only $50 in cash, but $200 in extra education. If so, everyone is
better off under the equalization program than they would be under the cash-transfer
program.
Alternatives to Metropolitan Consolidation
An alternative to metropolitan consolidation is Under

a metropolitan federation.
a federated government, a metropolitan government would be responsible for the
provision of goods with large scale economies (e.g., water and sewage systems,
sanitation, highways, transit). Municipalities would continue to provide goods with
relatively small scale economies (e.g., schools, police and fire protection, parks).
The Toronto area established a federated government in 1954.
Another alternative to metropolitan consolidation is to contract with another
level of government to provide public services. Many U.S. cities provide some
public services under intergovernmental contracts (under which one unit of govern-
ment arranges with another to provide a public service), and many use joint-service
contracts (under which two or more governmental units join to produce a public
service). Both types of arrangements allow a relatively small jurisdiction to realize
scale economies in the provision of public services.
466 Parr IV Local Government
Interjurisdictional Spillovers and Metropolitan Government

Another option is to use metropolitan government to deal with the urban problems
that cross municipal boundaries. Many economists and geographers believe that
metropolitan areas —not municipalities or states —
are the most important spatial
units in today's economy. In the words of Anthony Downs (1998), metropolitan
government is sensible
because the various spatial sections of each metropolitan area are linked together in a series of
densely interlocking networks. These networks transcend the boundaries of most individual
communities but are not as intensive at the larger state level.
Among these networks are streets and highways for commuting and nonwork
trips, water systems, sewage-disposal systems, school systems, airsheds, and water-
sheds. Some of the problems that cross jurisdictional boundaries are highway con-
gestion, air pollution, crime, low educational achievement in central-city schools,
and many of the problems caused by poverty. In the current political system, the
power to deal with these problems is divided among many small jurisdictions, most
of which are too small to contain but a fraction of the people affected by the problems.
One solution to the problem of interjurisdictional spillovers is to establish gov-
ernment bodies that have the power to deal with problems that cross jurisdictional
boundaries. Metropolitan government is working in Portland, Oregon, and in the
Twin Cities metropolitan area of Minnesota. As explained by Downs (1998). the
federal government could boost the prospects for metropolitan government:
Congress could improve this situation almost overnight by requiring each metropolitan region
to develop an overall plan for the use of federal funds of all types, before any government in
that region could receive any federal money at all. In 1991, Congress adopted this approach
concerning federal surface transportation funds. Now it should extend the same idea to all
types of federal funding of local governments and services.
The Growth in Local Spending

This section explores some of the reasons for growth in local public spending.
Because total spending equals the quantity of output times unit cost, an increase in
spending could be caused by an increase in unit cost, an increase in the quantity

produced, or increases in both cost and quantity.
Increases in Labor Cost: The Baumol Model

Baumol (1967) argues that rising labor cost has contributed to increases in local
spending. To explain the Baumol model, consider an economy that produces man-
ufactured goods (watches) and personal services (haircuts, public education). The
labor market has the following characteristics:
1 . Manufacturing productivity. The productivity of manufacturing workers

increases over time, a result of innovations that allow the substitution of
capital for labor.
2. Service productivity. The productivity of service workers (barbers and

teachers) is A haircut is a nonstandardized good: every head has a
constant.
different shape,and every customer wants a unique haircut. Since a barber
cannot be replaced by a machine, productivity is constant. Like the barber,
the teacher provides a nonstandardized output: every child enters the
classroom with a different set of skills and educational needs, and requires
personal attention. In addition, every child requires a different educational
output. Given the difficulty of capital substitution, teacher productivity is
constant.
3. Worker mobility and wages. Workers move costlessly between

manufacturing and service jobs. In equilibrium, all workers earn the same
daily wage.
4. Fixed labor supply. The total supply of workers is fixed.
Figure 16-5 shows the market for labor in the two-sector economy. The demand
curves for both manufacturing and services are negatively sloped. In the initial
equilibrium (points B and C), the wage is $13.

Suppose that the productivity of manufacturing workers increases. An increase in
productivity shifts the demand curve for manufacturing workers to the right: at every
wage, more workers are demanded. In the short run, the supply of manufacturing
workers is fixed, so the manufacturing wage rises to $17 (point F).
In the long run, the increase in the manufacturing wage causes some service
workers to switch to manufacturing jobs. As the quantity of manufacturing workers
increases, the manufacturing wage drops: the market moves downward along the
new demand curve from point F toward point D. In addition, the decrease in the
number of service workers causes a shortage of service workers, so the service wage
Figure 16-5 Effects of Increased Manufacturing Productivity on Wages

in the Service Sector
Demand after
productivity increase
M° M' S' S°
Manufacturing employment Service employment
In the initial B and C), the wage is $13. An increase in manufacturing productivity shifts the
equilibrium (points
demand curve manufacturing workers upward, causing excess demand for labor. As the manufacturing wage
for
rises, service workers move to the manufacturing sector, causing a shortage in the service sector that increases
the wage. In the new equilibrium (points D and E). the wage is $15.
increases: the market moves up the service demand curve, from point C toward point
E. Equilibrium is when two conditions are satisfied. First, total employment
restored
in the two sectors equals the fixed supply of labor {M + S ). Second, workers are
indifferent between the two types of jobs. Workers will be indifferent if the two
sectors pay the same wage. These two conditions are satisfied at points D and E.
What are the implications for the prices of manufacturing goods and services?
Because the productivity of manufacturing workers increases over time, the increase
in the wage is offset by increases in labor productivity, so production cost and market
prices do not change. In contrast, the wages of service workers increase while their
productivity is constant, so production cost and market prices increase over time.
In other words, the price of haircuts and the unit cost of education increase over
time.
Other local public goods share the personal-service features of haircuts and
education. Consider the cost of police protection. Police officers interview witnesses
and look for clues at the scenes of crimes. Neither of these tasks can be performed by
a machine, so the growth in police productivity is relatively slow. The same is true for
fire fighting: every fire presents a different set of circumstances, preventing the sort
of capital substitution that occurs in the manufacturing sector. If the productivities
of police and fire fighters are constant, an increase in wages (caused by an increase
in the wage of manufacturers) increases the cost of police and fire protection.
Other Reasons for Rising Labor Cost
The Baumol model suggests that the labor cost of government increases because the
government provides personal-service goods, not because public servants are lazy
or greedy. There are two other possible reasons for increases in labor cost:
1. Competition and innovation. The nature of the public sector discourages

innovation. Unlike a firm in a competitive market, the local government is
shielded from competition, so there is less incentive to develop new
labor-saving techniques. If this view is correct, the monopoly status of local
government prevents innovations that would increase labor productivity.
2. Union power. Most government workers are represented by powerful labor
unions. If unions use their power to increase the wages of government
workers, public servants will be overpaid relative to their private-sector
counterparts, increasing the cost of government. Freeman (1985) examined
the wages of private and government employees. He compared the wages of
public and private workers with similar qualifications (education, work
experience, personal characteristics). He found that state and local admini-
about 5 percent more than equally qualified private-sector
strators earn
workers, and public school teachers earn about 6 percent less than equally
qualified private workers. On balance, it appears that the differences
between private and government wages are relatively small. Freeman
did not include pension plans in his analysis. Since public pension plans
are generally more generous than private plans, the inclusion of pension
benefits would increase public-sector wages relative to private-sector

wages.
Increases in Quantity
An increase in local spending may be caused, in part, by an increase in the quantity

of local services demanded. The Baumol model provides a useful framework for
analyzing changes in the quantity of local public goods.
The quantity of education demanded changes over time for two reasons. The
the income effect. The increase in manufacturing productivity increases the
first is
real income of the economy: the total output of the economy increases, increasing
the purchasing power of its citizens. As real income increases, the citizens spend
more on all normal goods, including education. In Figure 16-6, the demand curve
for education shifts to the right: at every price, more education is demanded. At P',
the quantity of education increases from E to E'
The second reason the demand for education changes is the substitution effect.
The substitution effect causes movement upward along the new demand curve.
Figure 16-6 Changes in the Quantity of Education Demanded
Original demand curve
New demand curve
cd p
Quantity of education
An increase in real income shifts thedemand curve for education outward and increases the
quantity demanded (the income effect: B to C). If the unit cost (price) of education increases
from P' to P". the quantity demanded decreases (the substitution effect: C to D). In this case,
the income effect is large relative to the substitution effect, so the quantity demanded increases
(B to D).
Because the productivity of manufacturing workers increases while the productivity

of teachers is constant, the price of education increases relative to the price of other
goods. As a result, consumers substitute manufactured goods for education (and
other personal-service goods), decreasing the quantity of education demanded. In
Figure 16-6. the price of education increases from P' to P", and the substitution
effect causes movement upward along the new demand curve from point C to D.
Figure 16-6 shows a net increase in the quantity of education demanded from
E° to £"'. The income effect (the outward shift of the demand curve) dominates
the substitution effect (the upward movement along the new demand curve), so the
equilibrium quantity of education increases. This occurs under the assumption that
the income elasticity of demand for education is large relative to the price elasticity
of demand. Total spending on education increases: the community consumes more
education (£"' instead of E ) at a higher unit cost (P" instead of P').
Another possibility is that the quantity of education demanded decreases over
time. If the income effect is small relative to the substitution effect (if the income
elasticity is small relative to the price elasticity), the outward shift of the demand
curve would be small relative to the movement up the new demand curve, and the
quantity of education demanded would decrease. If this happens, total expenditures
on education may either increase or decrease. If the decrease in the quantity de-
manded is large enough to offset the increase in unit cost, total expenditures would
decrease. But if the decrease in quantity demanded is relatively small, total expen-
ditures would increase.
Taxpayer Revolt: Expenditure Limits
During the late 1970s and early 1980s, dozens of states imposed new constraints on
local-government spending and taxes. According to opinion surveys, the supporters
of tax limits believed that there was substantial inefficiency in local governments,
so taxes could be cut without affecting public services (Courant, Gramlich, and
Rubinfeld, 1987). For example, before the passage of Proposition 13 in Califor-
nia, 38 percent of California voters believed that local governments could absorb a
40 percent reduction in tax revenue without any reductions in local public services
(Citrin, 1979).
Although voters believed that there was no trade-off between public and private
goods, there is indeed a trade-off. In states that have tax limits, local schools have
larger class sizesand lower-paid teachers. In addition, tax-limit states have lower
educational achievement, as measured by scores on standardized tests on reading,
math, science, and social studies (Figlio, 1996). Municipalities subject to tax limits
provide lower-quality fire protection services (Doyle, 1994).
The tax limits have been effective in reducing local-government spending (Pre-
ston and Ichniowski. 1991 ). But local governments have cushioned the effects of tax
limits in a number of ways (Aronson and Hilley, 1986):
1. Local governments have shifted to user fees and taxes that are not subject to
restrictions (income, sales, and business taxes).
2. Some spending programs have been transferred to special districts to evade

the spending limits.
3. Intergovernmental grants from state governments have increased.

4. In cases where limits apply to specific spending categories, programs have
been moved from restricted categories to unrestricted ones.
The debate over tax and spending limitations has been spirited. The supporters
of limitations view local governments as inefficient monopolies. According to this
view, the limitations force local governments to "cut out the fat" by increasing
productivity and decreasing wages. According to Courant, Gramlich, and Rubinfeld
(1987), many of the people who vote for tax and spending limitations hold this view.
The critics of tax and spending limitations believe that the Baumol model is
applicable to the local public sector. They believe that the cost of local services
has increased because of rising productivity in the private sector, not because pub-
lic workers are lazy and wasteful. According to this view, there is little "fat" in
the local public budget, so the spending limitations decrease the quantity of local
public services. Another concern is that local tax limitations may increase the state
involvement in local budget choices, decreasing local control of tax money.
Fiscal Problems of Central Cities

Since the early 1970s, many cities have experienced serious fiscal problems. The
most sensational episode was the near default of New
York City in 1975. In many
other cities, tax revenues have fallen short of the amount required to provide the
average level of public services, requiring either tax hikes or service cuts to prevent
budget deficits. This part of the chapter explores some of reasons for the fiscal
problems experienced by central cities.
The most severe problems have been experienced by old, large cities,
fiscal
especially those that have experienced either sloweconomic growth or decline. As

explained by Peterson (1974) and Ladd and Yinger (1992), there are several reasons
for the fiscal problems of large central cities:
1. Suburbanization. In most cities, the suburbanization of employment and

population has decreased the city's tax bases (income, property, sales),
decreasing tax revenue.
2. Decrease in population. In many cities, population losses have not led to
lower costs for providing public services. If a city invests in infrastructure
(e.g., streets, transit systems, schools, and fire stations) suitable for a city of
5 million, it will pay the same capital costs even if its population drops to 4
million. A decrease in population increases per capita costs because there
are fewer "capitas" to pay the costs. In some cities, labor unions are
powerful enough to prevent layoffs when population decreases, leading to
higher labor costs per capita.
3. Old infrastructure. The maintenance costs of infrastructure increase as it
ages, so old cities with aging infrastructure have relatively high costs.
4. Diseconomies of scale. According to Ladd and Yinger (1992), there are

diseconomies of scale in providing urban services. The elasticity of general
public-service cost with respect to population is 0.14, meaning that a city
that is 10 percent larger than the average city has per capita service
costs that are 1.4 percent higher. The elasticity of police costs is 0.21,
indicating that per capita police costs increase by 2.1 percent for every
10 percent increase in population.
5. Low per capita income. Cities with low income per capita generate less
tax revenuefrom a given tax rate, and so must either collect less revenue or
impose higher tax rates.
6. Small share of metropolitan population. A central city that has a
relatively small share of metropolitan population has higher costs because it
provides services for a relatively large number of commuters.
Poverty and Fiscal Problems

Urban poverty is an important factor in the fiscal problems of central cities. As
explained in Chapter 12, poverty rates in central cities are about twice as high as
poverty rates in suburban areas. Although most of the money
programs to assist for
the poor come from the federal government, local governments
United States in the
devote about 12 percent of their own revenue to public welfare, health, and hospitals
(Gyourko and Summers, 1998). In 1992, large cities (population exceeding 300,000)
spent $364 per capita on these programs, or 30 percent of their own revenue. In
contrast, smaller cities and suburban municipalities spent only $40 per capita on
these programs (9 percent of revenue).
Poverty also has indirect effects on municipal spending. Ladd and Yinger 992) ( 1
estimate that the elasticity of police costs with respect to the poverty rate is 1.1.
meaning that a city with a poverty rate 1 percent higher than the average city has
police costs that are about 1 1 percent higher. Gyourko and Summers 1998) estimate(
that Philadelphia, with a poverty rate of 2 1 more per year

percent, spent $23 million
on police services than it would have, had its poverty rate been equal to the national
average of 14 percent. Cities with high poverty rates also have higher costs for
education and fire protection.
Index of Fiscal Health

Ladd and Yinger 1989) estimate an index of fiscal health for dozens of central cities.
(
The index measures the ability of a city to deliver a standard set of public services,
assuming that the city's residents pay 3 percent of their income as municipal taxes.
The ability of a city to deliver the standard tax/service package depends on the
economic circumstances of the city. For the average city, the value of the index is
Figure 16-7 Standardized Fiscal Health of Selected

Central Cities in 1988
New York
Los Angeles
Chicago
Philadelphia
Detroit
San Diego
New Orleans
Nashville
Pittsburgh
Minneapolis
Oakland
Sacramento :
Rochester
Santa Ana. Calif.
Greensboro
Garden Grove, Calif.
Ontario, Calif.
High Point. N.C.
I I
-100 -80 -60 -40 -20 20 40 60

Standardized fiscal health: percent deficit or surplus
Definitions:
1 Negative value: percentage increase in revenue that a city would have to receive
from an outside source to be able to provide the baseline quality at the standard
tax burden.
2. Positive value: excess tax revenue (as a percent of total) after providing the baseline
service quality at the standard tax burden.
SOURCE: Helen Ladd and John Yinger. America's Ailing Cities: Fiscal Health and the
Design of Urban Policy (Baltimore: Johns Hopkins University Press, 1989). Table E.l.
zero. A negative value indicates that the cost of delivering the standard set of services
would exceed the revenue generated by the standard tax package.
Figure 16-7 shows the index of fiscal health for several cities. A negative value
for the index indicates the fiscal deficit associated with the standard tax/service
package —
or the size of the transfer required to fill the gap between local tax revenue
and the cost of providing the standard service level. New York's value is —83,
meaning that to provide the standard level of public services with the standard
tax system, the city would need a transfer equal to 83 percent of its tax revenue.
A positive value for the index indicates the percentage surplus resulting from the
standard tax-spending package.
A city receives a negative score if its revenue-raising capacity is less than the
cost of providing the standard service package. The negative scores result from one
or more of the following conditions: low per capita income (resulting in a small tax
base), large population (resulting in higher service costs because of diseconomies
of scale in public services), high poverty rate (resulting in higher costs of public
services), old infrastructure (resulting in high maintenance costs), and a small share
of the metropolitan population (resulting in higher costs to provide services for
commuters).
The index of
fiscal health gives us a list of cities where fiscal crises are most
likely. In the real world of municipal budgeting, a city with a negative score has two
options. First, if the city wants to provide the standard service package, it must have
higher tax rates. Second, if the city wants to keep its tax rates at the national average,
it must provide a lower-than-average level of public services.
Summary
1. There are almost 85,000 local governments in the United States, including
municipalities, school districts, counties, and special districts. Municipalities
are responsible for about 33 percent of local-government spending, and school
districts are responsible for about 30 percent. The largest spending program is
education (42 percent of local-government spending).

2. 39 percent of local revenue came from taxes on property, income, and
In 1992.
and 38 percent came from intergovernmental grants. The rest of local
sales,
money comes from user fees and charges.
3. The national government is responsible for stabilization and redistribution
policies.
4. Local governments are involved in resource allocation for three types of goods.
a. Natural monopoly. Local governments either produce or regulate the private
production of goods generated by natural monopolies (water, sewage
services, transit).
b. Externalities. One option is for the government to provide goods that
generate externalities. Another option is to internalize the externality w ith a
subsidy.
c. Local public goods. A local public good is nonrivalrous and nonexcludable,
and its benefits are confined to a relatively small geographical area. For the
optimum quantity, the marginal social benefit (the sum of the marginal
private benefits) equals the marginal cost.
5. There are advantages and disadvantages associated with the local provision of
public goods.
a. Local governments promote efficiency if there is diversity in demand for

public goods.
b. Local governments inhibit efficiency if the benefits or costs of public goods
spill over into other jurisdictions (externalities).
c. Local governments inhibit efficiency if there are scale economies in the
production of public goods.

6. Local provision of a public good is efficient if the diversity in demand for local
public goods is large relative to externalities and scale economies.
7. Metropolitan consolidation has been advocated as a way to prevent suburban
exploitation, exploit scale economies, equalize tax rates, and equalize
spending.
8. The Baumol model suggests that increases in the labor cost of personal
services (haircuts and education) result from rising productivity in
manufacturing, which pulls up the wages of all workers.
a. Other theories of rising labor cost focus on the monopoly power of local
governments and labor unions.
b. The quantity of public services demanded will increase if the income effect
(from rising productivity) is large relative to the substitution effect (from
rising unit cost).
9. In the 1970s and 1980s, many states tightened restrictions on local tax rates,
tax revenue, and spending.
10. Since the early 1970s, many cities have experienced serious fiscal problems.
The most severe problems have been experienced by
fiscal large, older cities
with declining populations and employment.
1 1 The index of standardized fiscal health shows the fiscal health of cities in the
absence of intercity differences in taxes and spending.

12. The cost of providing city services depends on the characteristics of the city.
a.The elasticity of general costs with respect to population is 0. 14, meaning
there are diseconomies of scale in providing general services.
b. The elasticity of police costs with respect to the poverty rate is 1.1: a
10 percent difference in poverty rates between cities causes an 1 1 percent
difference in the cost of police services.
13. Larger cities and cities with relatively large numbers of poor people have
relatively small revenue-raising capacities and relatively large expenditure
needs, so they have low scores for standardized fiscal health.

1 . Why do we have public libraries? Why don't people rent books from firms, just
as they rent trucks, Rototillers, and videos? If we have U-Haul, why not
U-Read; if we have Captain Video, why not Captain Book?
476 Pan IV Local Government
2. Figure 16-4 showing the marginal social benefit of local parks assumes that
there is no congestion in the use of the parks; that is. the good is literally
nonrivalrous.
a. Draw a new MSB curve under the assumption that congestion occurs.
b. What are the implications for the optimum park acreage? Does it increase or
decrease?
3. Consider Metro, a metropolitan area with three school districts. East. West, and
North. There are externalities in the provision of education, but there are no
scale economies in the provision of schools. Suppose that the three districts are
consolidated into a single metropolitan school district. Will it be more or less
efficient than the original arrangement' 1
4. Consider Metro, a metropolitan area with two municipalities. Sportville (high
demand for parks) and Slothburg (low demand for parks). The provision of
parks is subject to scale economies (the cost per acre of parks is lower under
metropolitan management) and generates externalities. Suppose that the two

municipalities are consolidated into a single metropolitan government. Under
what circumstances will total expenditures on parks decrease?
5. Consider Metro, a metropolitan area with two municipalities, each of which has
a small police force. The only activity of the police is chasing criminals: the
police never capture criminals but simply cause them to move to another
municipality.
a. Suppose that the provision of police services is subject to scale economies. If
the two municipal police forces are consolidated into a single metropolitan
force, will total expenditures on police services increase, decrease, or stay
the same?
b. If you don't have enough information to answer the question, describe the
information you need and explain how you would use it.
6. In the city of Crewville. everyone gets crew cuts: the output of the barber is
standardized. The output of the teacher is also standardized: the only objective
is to teach children how to add and subtract. Will the costs of haircuts and
education increase relative to the cost of manufactured goods over time?
7. Consider the cost of admission to a performance by a popular rock band. Would
you expect the relative cost of admission to increase over time (relative to the
price of manufactured goods)

Overview
Advisory Commission on Intergovernmental Relations. Significant Features of Fiscal

Federalism. Washington. D.C.: U.S. Government Printing Office. 1987.
Inman. Robert P. "The Fiscal Performance of Local Governments: An interpretive Review."
In Current Issues in I 'rban Economics, ed. Peter Mieszkowski and Mahlon
Straszheim. Baltimore: Johns Hopkins University Press, 1979, pp. 270-321.

Summarizes studies of the price and income elasticities of demand for local public
goods and the elasticity of local expenditures with respect to intergovernmental aid.
Musgrave, Richard A., and Peggy B. Musgrave. Public Finance in Theory and Practice.
New York: McGraw-Hill, 1980.
U.S. Bureau of the Census. Census of Governments. Washington, D.C.: Government
Printing Office, 1987.
.
City Government Finances. Washington, D.C.: Department of Commerce, 1987.
. Compendium of Government Finances, no. 5. Washington, D.C.: U.S. Government
. Statistical Abstract of the United States. Washington, D.C.: U.S. Government Printing
Office, 1990.
The Role of Local Government in the Federal System
Break, George. Financing Government in a Federal System. Washington, D.C.: Brookings

Institution, 1980.
Oates, Wallace E. Fiscal Federalism. New York: Harcourt Brace Jovanovich, 1972.
Analyzes the division of responsibilities between different levels of government,
focusing on spillovers, scale economies, and diversity in demand.
Olson, Mancur. "The Principle of Fiscal Equivalence: The Division of Responsibilities
among Different Levels of Government." American Economic Review 59 ( 1 969),
pp. 479-87. Argues that the federal system is required to control benefit spillovers.
Stigler, George J. "The Tenable Range of Functions of Local Government." In Private
Wants and Public Needs, ed. Edmund S. Phelps. New York: Norton, 1965.
Municipal versus Metropolitan Government

Bahl, Roy, and David Puryear. "Regional Tax Base Sharing: Possibilities and Implications."
National Tax Journal 29 (1976), pp. 328-35.
Bollens, John C, and Henry J. Schmandt. The Metropolis. New York: Harper & Row, 1982.
Describes the metropolitan federation form of government, under which some public
services are provided by a metropolitan government and others by small local
governments.
Bradford, David, and Wallace E. Oates. "Suburban Exploitation of Central Cities and
Government Structure." In Redistribution through Public Choice, ed. Harold Hochman
and George Peterson. New York: Columbia University Press, 1974. Discusses the
suburban exploitation hypothesis and predicts the effects of metropolitan consolidation
on tax rates, expenditure levels, and income segregation.
Downs, Anthony. "The Devolution Revolution: Why Congress Is Shifting a Lot of Power to
the Wrong Levels." Brookings Policy Brief no. 18. Washington, D.C.: Brookings
Institution, 1998.
Greene, Kenneth V., and Thomas J. Parliament. "Political Externalities, Efficiency, and the
Welfare Losses from Consolidation." National Tax Journal 33 (1980), pp. 209-17. A
study of the efficiency effects of metropolitan consolidation.
Reschovsky, Andrew. "An Evaluation of Metropolitan Area Tax Base Sharing." National
Tax Journal 33 (1980), pp. 55-66. Describes the experience with tax-base sharing in
Minneapolis-St. Paul.
Scale Economies in the Provision of Local Public Goods
Advisory Commission on Intergovernmental Relations. Intergovernmental Service

Arrangements for Delivering Local Public Services: Update 1983. Washington. D.C.:
U.S. Government Printing Office. 1985. A survey of the use of intergovernmental and
joint-service contracts in the provision of local public services.
Gardner. John L. "City Size and Municipal Sen ice Costs." In Urban Growth Policy in a
Market Economy, ed. George S. Tolley. Philip E. Graves, and John L. Gardner. New
York: Academic Press. 1979, pp. 51-61. Examines the relationship between
public-service costs and city size, controlling for the fact that the demand for public
services depends on income, which varies across cities. Concludes that cost per family
increases with city size and decreases as population density increases.
Gyimahbrempong. K. "Economies of Scale in Municipal Police Departments The Case of —
Florida.'" Review of Economics and Statistics 69 (1987). pp. 352-56.
Hines. Lawrence G. "The Long-Run Cost Function of Water Production for Selected
Wisconsin Communities." Land Economics 45 (1969). pp. 133—40. Detects scale
economies in the provision of water services.
Hirsch. Werner. "Expenditure Implications of Metropolitan Growth and Consolidation."
Review of Economics and Statistics 41 1959). pp. 234-35. Estimates the degree of
(
scale economies for police, fire protection, and public schools. Did not detect scale
economies for police services or public schools. Found scale economies for fire
protection up to a population of about 100.000.
Neutze. G. M. Economic Policy and the Size of Cities. Canberra: Australian National
University. 1965. Discusses scale economies in the provision of urban services.
Walzer. Norman. "Economies of Scale and Municipal Police Sen ices: The Illinois
Experience." Review of Economics and Statistics 54 1972). pp. 431-38. Suggests that
(
there are small scale economies in the provision of police senices for small cities (up
to 140.000 population).
Rising Cost of Government and the Tax Revolt

Aronson. J. Richard, and John L. Hilley. Financing State and Local Governments.
Washington. D.C.: Brookings Institution. 1986. Chapter 12 discusses revenue and
expenditure limits and the responses of local governments to the limits.
Baumol. William. "The Macroeconomics of Unbalanced Growth: The Anatomy of Urban
Crisis." American Economic Review 65 1967). pp. 414-26.
(
Bradford. David F.; R. A. Malt; and Wallace E. Oates. "The Rising Cost of Local Public
Services: Some Evidence and Reflections." National Tax Journal 22 1969). (
pp. 415-26.
Brennan. Geoffrey. "Tax Limits and the Logic of Constitutional Restrictions." Chapter 6 in
Tax and Expenditure Limitations, ed. Helen Ladd and Nicholas Tideman. Washington.
D.C.: Urban Institute. 1987. pp. 121-38.
Citrin. Jack. "Do People Want Something for Nothing? Public Opinion on Taxes and
Government Spending." National Tax Journal 31 1979). pp. 13-29. ( 1
Courant, Paul: Edward Gramlich: and Daniel Rubinfeld. "Win Voters Support Tax
Limitation Amendments: The Michigan Case." Chapter 3 in Tax and Expenditure
Limitations, ed. Helen Ladd and Nicholas Tideman. Washington. D.C.: Urban Institute.
1987. pp. 37-72.
Chapter 16 Oven'iew of Local Government 479
Doyle, Maura. "Property Tax Limitations and the Delivery of Fire Protection Services."
Federal Reserve Board of Governors (1994).
Figlio, David."Did the Tax Revolt Reduce School Performance? Journal of Public
Economics 65 (1997), pp. 245-69.
Freeman, Richard B. "How Do Public Sector Waste and Employment Respond to
Economic Conditions?" Working Paper no. 1653, National Bureau of Economic
Research, Cambridge, Mass., 1985.
Ladd, Helen F. "An Economic Evaluation of State Limitations on Local Tax and Spending
Powers." National Tax Journal 31 (1978), pp. 1-18. Explores the reasons behind tax
revolts.
Ladd, Helen F, and Julie Boatwright Wilson. "Why Voters Support Tax Limitations:
Evidence from Massachusetts' Proposition 2 1/2." National Tax Journal 35 (1982), pp.
137^10.
Musgrave, Richard. "Leviathan Cometh-Or Does He?" Chapter 5 in Tax and Expenditure
Limitations, ed. Helen Ladd and Nicholas Tideman. Washington. D.C.: Urban Institute,
1987, pp. 77-120.
Oakland, William. "Proposition 13: Genesis and Consequences." National Tax Journal 32
(1979), pp. 387^09. Discusses the reasons for the tax revolt in the state of California.
Preston, Anne and Casey Ichniowski. "A National Perspective on the Nature and Effects of
E.,
the Local Property Tax Revolt, 976-1 986." National Tax Journal 44(1 99 ). pp. 123^5.
1 1
Rabushka, A., and P. Ryan.

The Tax Revolt. Stanford, Calif.: Hoover Institution, 1982. Discusses the attempts
by citizens to control government spending by tax and expenditure limitations.
Fiscal Health and Distress

Bradbury, Katharine L.; Anthony Downs; and Kenneth
A. Small. Urban Decline and the Future of American Cities. Washington, D.C.:
Brookings Institution, 1982. Discusses the reasons for the economic problems of cities.
Gyourko, Joseph, and Anita A. Summers. "A New Strategy for Helping Cities Pay for the
Poor." Brookings Policy Brief no. 18. Washington, D.C.: Brookings Institution, 1998.
Humphrey,
Nance; George Peterson; and Peter Wilson. The Future of Cleveland's Capital Plant.
Washington, D.C.: Urban Institute, 1979. Discusses the fiscal problems of Cleveland.
Ladd, Helen, and John Yinger. America 's Ailing Cities:
Fiscal Health and the Design of Urban Policy. Baltimore: Johns Hopkins University
Press, 1989. A comprehensive study of the factors that determine a city's fiscal health.
Oakland, William H. "Central Cities: Fiscal
Urban Economics, ed. Peter
Plight and Prospects for Reform." In Current Issues in
Mieszkowski and Mahlon Straszheim. Baltimore: Johns Hopkins University Press,
1979, pp. 322-58. Discusses the sources of central-city fiscal distress, suggesting
is caused in large part by the attempt of local governments
that fiscal distress
to redistribute income. Suggests that intergovernmental grants should be designed to
compensate local governments for the money spent on high-cost, low-revenue citizens.
Peterson, George E. "Finance." In The Urban Predicament, ed. William Gorham and Nathan
Glazer. Washington, D.C.: Urban Institute, 1976. Provides a comprehensive analysis of
the reasons for the fiscal crises of large cities and suggests ways to avoid future problems.
Reischauer. Robert; Peter Clark: and Peggy Cuciti.

New York City's Fiscal Problem. Washington. D.C.: Congressional Budget Office.
1975. A detailed discussion of the fiscal problems experienced by New York City.
Privatization
Savas. E. S. Privatizing the Public Sector. Chatham, N.J.: Chatham House. 1982. Argues
for shifting the responsibility for providing some public services to the private sector.
Stevens. Barbara J. Delivering Municipal Services Efficiently: A Comparison of Municipal
and Private Sector Delivery: Summary. Report no. 3744. U.S. Department of
Housing and Urban Development, Washington. D.C.. June 1984. Compares the costs
of providing municipal services by local governments and by private firms (through
the contracting process). Concludes that government cost was substantially higher.
Chapter 17
v T I N G with
BALLOTS and WW
his chapter discusses two ways that citizens express their preferences for
(~J7~
t_y local government. One option is to vote in budget elections and in other
elections that pick mayors, city council members, and other local officials. The
policy question is whether voting in local elections promotes efficiency in local
government. Specifically, under majority rule, do local governments provide the
optimum level of local public goods? A second option for citizens is to express their
preferences by voting with their feet. The typical large metropolitan area, with its
large number of local governments, provides citizens with a wide range of choices
for local tax and spending programs. One way to get a better combination of local
services and taxes is to move to another jurisdiction. The policy question is whether
this foot voting promotes efficiency in local government. Specifically, would the
consolidation of several municipalities into a metropolitan government improve the
efficiency of the local public sector?
Voting with Ballots: The Median- Voter Rule

Local governments must decide how much of a local public good to provide. One
approach to local decision making is to hold an election and let voters decide how
much to spend on a local public good. Will voters choose the optimum amount of a
local public good?
Recall the example (from the previous chapter) of the three-person city that must
decide how many acres of public parks to provide. In Figure 17-1, the marginal-
benefit curvesshow the willingness to pay for different amounts of parks: Sam
has a low demand for parks and prefers them relatively small (MB S )\ Marian has a
medium demand and prefers medium-sized parks (MB m ); Bertha has a high demand
and prefers big parks (MBj,). If there are no interjurisdictional spillovers, the marginal
social benefit is the sum of the marginal benefits of the three citizens. At the optimum
481
Figure 17-1 Provision of Parks under Majority Rule and Benefit Taxation
v Marginal social benefit = MB S

+ MB m + MB^
60 Marginal social cost
\ 6/>
' \v
35
^« m T\. N.
Marginal cost with

h=T=,
equal cost shares
1
|^~~-=~^_ ^^-~~L
1
i ^-t
A A A* A,
S in b
Park acreage
The optimum park acreage is (the amount where the marginal social benefit equals the marginal social cost). II
\
each household pays S20 per acre, the preferred amounts are A s A,„. and .4,,. Under majority voting, the preferred
,
si/e of the median votei (A„ Us chosen. Under a benefit tax, each household pays a tax (per aerei equal to its
marginal benefit at the optimum park si/e so there will be unanimous support for the optimum park si/e.
i . \
'
I,
acreage, the marginal social benefit of park acreage equals its marginal social cost.
In Figure 17-1. the marginal cost is $60 per acre, so the optimum acreage is A*.
The city must pick a tax system to finance the provision of parks. Suppose that
the city divides park costs equally among its three citizens. If the marginal cost is $60
per acre, every citizen pays $20 per acre of park land. What are the preferred park
si/cs for the three citizens? The preferred size of a particular voter is the acreage such
that the marginal private cost equals the marginal private benefit. In Figure 17-1, the
marginal private cost is the horizontal line at $20. For Sam. the cost curve intersects
the benefit curve at a park size of As acres, so his preferred size is A s acres. For the
other two citizens, the preferred sizes are A„, and A/,.
Suppose that the city holds a series of elections in which citizens vote for one
of two proposed park sizes. In an election between A and A m
s , there is one vote for
As (Sam), but two votes for Am (Marian and Bertha), so Am wins the election. A m
also wins an election against Ai,: Sam and Marian vote for Thus A„, wins both
-4,,,.
Chapter 17 Voting with Ballots and Feet 483
elections because it meets the preferred budget of the median voter (the voter that
splits the rest of the voting public into two equal halves). The preferred budget of
the median voter wins an election against a smaller size (as Bertha joins the median
voter to make the vote 2 to 1) and any larger size (as Sam joins the median voter to
make the vote 2 to ). This is the median-voter result: if the spending level for a
1
public good is determined by a majority vote, the winning budget is the preferred
budget of the median voter.
In Figure 17—1, the size chosen by voters (A m ) is less than the optimum size
(A*). Is this an artifact of the election sequence, or would A m win a direct election
against the optimum? In an election between A m and A*, Sam would join Marian in
voting for A,„ because A m is closer to Sam's preferred size. Therefore, A,„ defeats
the optimum size in a direct election.
— —
The power and inefficiency of the median-voter result can be shown by con-
sidering the effects of changes in Bertha's preferences. Suppose that her marginal-
benefit curve shifts upward by 100 percent: for every size, her willingness to pay for
parks doubles. Because the marginal-social-benefit curve is the vertical sum of the
individual marginal-private-benefit curves, the social-benefit curve would also shift
upward, and the optimum park size would increase by a large amount. Nonetheless,
voters still choose A,„ Sam joins with Marian to defeat any size above A,„ The same
: .
is true of changes in the other nonmedian benefit curve MB S The voting outcome .
depends only on the position of the median voter's marginal-benefit curve.
Benefit Taxation
Majority voting is inefficient because the costs of the public good are divided equally
among the three voters. An alternative approach
is to allocate park cost according
to the willingness to pay for parks: the greater the citizen's willingness to pay, the
higher the tax liability. This is the Lindahl approach to local decision making
(named after Erik Lindahl): according to the benefit principle, tax liabilities are
based on the benefits received from public goods.

Figure 17-1 shows how benefit taxes work. Suppose that the government knows
the marginal-benefit curves of its citizens. It announces that Sam's tax liability per
acre is $9, Marian's is $16, and Bertha's is $35. Given these private costs, all three
citizens have the same preferred park size, A*, which also happens to be the optimum
size. In this case, the government is clever enough to figure out how much individual
citizens are willing to pay for different levels of the local public good, and it sets
each citizen's tax equal to his or her marginal willingness to pay. Sam (who prefers
small parks) agrees to a larger park because he faces a lower tax liability: he pays $9
per acre instead of $20. Bertha (who prefers big parks) gets a larger park at a cost:
she pays $35 per acre instead of $20. Under benefit taxation, citizens with larger
benefits pay higher taxes.
Is the benefit principle practical?One problem with this approach is that the
government doesn't know the shapes of the marginal-benefit curves of individual
citizens, so it can't estimate the appropriate taxes for them. Moreover, there is little
incentive for taxpayers with relatively high benefits to reveal their willingness to pay
for local public goods: in doing so. they would be volunteering to pay higher taxes.
It may be rational for high demanders to hide their true preferences in the hope that
someone else will pay for the public good. This is the free-rider problem: citizens
may understate their willingness to pay in an attempt to get a free ride.
Budget Elections and the Median Voter

The median-voter result is applicable to a wide variety of election formats. For
example, if a school district holds a series of elections to determine its budget, the
winning budget is the preferred budget of the median voter. To show this, consider
a school district with the following characteristics:
1 Citizens'' preferences. Citizens have different preferences for school

spending. Table 17-1 shows the preferred budgets for seven voters in the
school district. Voter A's preferred budget is $49, voter B*s is $56. and so on.
2. Series of elections. The school board proposes a budget and holds an
election in which citizens vote yes or no. If a given budget proposal fails to
receive a majority of votes, the school board will decrease its proposed
budget by $10 and hold another election. This process continues until a
majority of citizens vote in favor of the budget.
Under this election scheme, the school district chooses the largest budget receiving
majority support.
Suppose that the school board starts with a budget of $90. The voters know that
if $90 budget fails, the board will propose an $80 budget in the second election.
the
A citizen votes yes if the $90 budget is better than an $80 budget (if the citizen's
preferred budget is greater than $85). For example, voter F votes against the $90
budget because an $80 budget is closer to his preferred budget ($84). G is the only
voter who votes for the $90 budget, so the first budget proposal loses by a vote of
6 to 1 The second proposal ($80) fails by a vote of 4 to 3: only citizens with preferred
.
budgets greater than $75 vote yes.
Table 17-1 The Median Voter in a Series

of School Buc get Elections
Preferred Vote with Vote with Vote with

Voter Budge $90 Budget $80 Budget $70 Budget
A $49 N N N
B $56 N N N
C $63 N N N
D $70 N N Y
E $77 N Y Y
F $84 N Y Y
G $91 Y Y Y
Aproposed budget of $70 wins an election by a vote of 4 to 3. All citizens

F, and G) vote for the $70
with a preferred budget greater than $65 (citizens D, E,
budget. The $70 budget proposal is successful because it's the preferred budget of
the median voter. A $70 budget splits the voters into two equal halves, with half
preferring a larger budget (E, F, G) and half preferring a smaller budget (A, B, C).
By threatening to propose a smaller budget in the next election, the school board
convinces citizens with preferred budgets above the median budget (E, F, G) to vote
for the $70 budget; combined with the vote of the median voter (D), this generates
a majority in favor of the proposed budget.
In this example, the school district decreases its proposed budget in $10 in-
crements, from $90 down to the preferred budget of the median voter. The same
resultwould occur if the board decreased the proposed budget in increments of $1
Similarly, thesame result occurs if the school district reverses the direction of the
budget sequence: if it starts out with a low budget and works its way upward, the
median voter is still decisive.
The Median Voter in a Representative Democracy
In a representative democracy, decisions about budgets are made by elected offi-

cials. make budgetary decisions indirectly, by electing people whose budget
Citizens
philosophies are consistent with their own preferences. For example. Bill votes for
the mayoral candidate whose expressed position on the city budget is closest to Bill's
preferred budget.
To explain the budget choices in a representative democracy, consider a city
that provides a single local public good (police services). The city has the following
characteristics:
1. Two candidates. There are two candidates for mayor, Penny (a low
spender) and Buck (a big spender).
2. Single issue. The only issue in the election is the police budget, which is set
by the mayor.
3. Citizen choice. Every citizen votes for the candidate whose proposed
police budget is closest to the citizen's preferred budget.
4. Distribution of preferences. Figure 1 7-2 shows the distribution of budget

preferences. The horizontal axis measures the police budget, and the
vertical axis measures the number of votes for a given budget. For example,
10 citizens have a preferred budget of $2, 15 have a preferred budget of $3,
and so on. The distribution of budget preferences is symmetrical, and the
median budget (the budget that splits the rest of the voters into two equal
halves) is $6.
Suppose that the candidates start out with different proposed police budgets:
Penny proposes a budget of $4, and Buck proposes $8. Citizens with preferred
budgets less than or equal to $5 vote for Penny (70 votes), and those with preferred
budgets greater than or equal to $7 vote for Buck (70 votes). The two candidates
Figire 17-2 The Median Voter in a Representative

Democracy
30
25
20
10 Id
4 5 6 7! 10
Preferred police budget (S)
Citizens have different preferences for the police budget. If Penny proposes a budget of
S4 and Buck proposes a budget of S8. the election will result in a tie vote: the two
candidates split the votes of citizens with S6 preferred budgets, while Penn\ gets the
votes of those w ith lower preferred budgets, and Buck gets the votes of those w ith
higher budgets. By moving toward the median budget. Penin can increase her chances
of being elected. In equilibrium, both candidates propose the budget of the median
voter ($61
split the 30 voters with a preferred budget of S6 (halfwaj between the two proposed
budgets), so each candidate gets a total of 85 votes.The election results in a tie vote.
Penny could increase her chance of being elected by increasing her proposed
budget. Suppose that she proposes S5 instead of S4. She picks up all 30 votes of
the citizens with a S6 preferred budget (Penny's S5 proposal is closer than Buck's
$8). and she wins the election by a vote of 100 (10 + 15 + 20 + 25 + 30) to 70
(25 + 20 4- 15 + 10). If Buck responds by decreasing his proposed budget to S7.
the election results in a tie vote again. Penny and Buck will continue to revise their
proposed budgets until each candidate's budget is close to the budget of the median
voter ($6).
This example shows that the median-voter result occurs in a representative
democracy. In equilibrium, both candidates propose a budget equal to the preferred
budget of the median voter. Since both candidates have virtually the same proposed
budget, it doesn't matter which candidate actually wins the election. In either case.
the median voter determines the size of the police budget.
Implications of the Median-Voter Rule
There are four implications of the median-voter rule:
1. Inefficient voting outcome. There is nothing that guarantees that the

preferred budget of the median voter is the optimum budget (the budget at
Chapter 17 Voting with Hal lots and Feet 487
which the marginal social cost of spending equals the marginal social
benefit).
2. Predicting election outcomes. To predict the outcome of an election, one

needs to (a) identify the median voter and (b) estimate his or her preferred
budget. As a practical matter, it may be difficult to identify the median
voter. One approach is to assume that the desired spending depends only on
income. Under this assumption, the median voter is the voter with the
median income. To the extent that desired spending depends on other
variables (household size, age, political philosophy), this approach
generates inaccurate predictions.
3. Estimating the income elasticity. Differences in income cause differences
in spending on local public services. These differences can be used to
compute income elasticity of demand for local public services. To
the
explain, consider two cities: city S has a small police budget ($100 per
capita) and a low median income ($1,000), and city L has a large police
budget ($125 per capita) and high median income ($1,200). Assume that the
"price" of police services (the opportunity cost of money spent on police) is
the same in the two cities. The income elasticity of demand for police
services is defined as the percentage change in the police budget divided by
the percentage change in income. City L, with 20 percent higher income,
has a 25 percent larger police budget, so the income elasticity of demand is
1.25 (25 percent divided by 20 percent). Inman (1979) summarizes the
results of empirical studies based on the median-voter model. As shown in
Table 17-2, the income elasticities for most local goods are less than 1.0.
4. Estimating price communities in the
elasticity. If there is variation across
price of local public goods, the median-voter model can also be used to
estimate the price elasticity of demand for public goods. To plot the
demand curve for local spending, one needs information on price (the
opportunity cost of local spending) and quantity (the local spending level).
The price elasticity can then be computed from the demand curve. As
Table 17-2 Income and Price Elasticities of

Demand for Local Public Goods
Public Good or Senice Price Elasticity Income Elasticity
Total expenditures -0.23 to -

-0.56 0.34 to 0.89
Education -0.07 to -0.51 0.24 to 0.85
Parks and recreation -0.19 to -0.92 0.99 to 1.32
Public safety (police and fire) -0.19 to -1.00 0.52 to 0.71
Public works -0.92 to -1.00 0.79
Source: Robert Inman, "The Fiscal Performance of Local Governments," in

Current Issues in Urban Economics, ed. Peter Mieszkowski and Mahlon
Straszheim (Baltimore: Johns Hopkins University Press, 1979).
shown in Table 17-2, the demands for local public goods are price-
inelastic: the price elasticities are all less than or equal to 1.0 (in absolute
value). Some of the implications of these elasticities are discussed in
Chapter 18 (Local Taxes and Intergovernmental Grants).
Qualifications of the Median-Voter Model

The median-voter model has a number of unrealistic assumptions. Although the
model provides a useful framework for thinking about local-government decisions,
it should be used with caution. There are three assumptions that limit the model's
applicability:
1 Ideology. The politicians in the model care only about winning elections,
so they slavishly adhere to voter preferences. An alternative assumption is
that candidates base their positions on ideology. An ideological candidate

uses the election campaign to persuade voters that her position is the correct
one. playing the role of a leader, not a follower. In a model with ideological
candidates, there is no reason to expect that the choice of the median voter
will prevail.
2. Single issue. There is a single issue in the simple median-voter model. If

there are several election issues (e.g.. the police budget, the park budget,
policies for the homeless), candidates offer package deals to voters. Unless
the same citizen is the median voter on every issue, the winning budget is
not the preferred budget of the median voter.

3. All citizens vote. In the median-voter model, all citizens vote. In the typical
budget election, only a fraction of eligible voters actually cast ballots. A
rational citizen abstainsfrom voting if the cost of voting (the time and
monetary costs of becoming informed and going to the polls) exceeds the
benefit. The benefit of voting is relatively small if (a) the two proposed
budgets are so close to one another that it makes little difference which
person is elected (indifference), or (b) the proposed budget of the best
candidate is so far from the citizen's preferred budget that the citizen is
alienated from the election process (alienation). If some citizens abstain
from voting, the median-voter result will not necessarily occur.
Voting with Feet: The Tiebout Model

Most metropolitan areas in the United States have dozens of municipalities, school
districts,and other local governments. A model developed by Tiebout 1956) shows (
the role of consumer choice in the development of our fragmented system of local
governments. The idea is that people vote with their feet, choosing the local govern-
ment that provides the best combination of taxes and local public goods. One of the
implications of the Tiebout model is that interjurisdictional mobility (voting with
feet) may prevent the inefficiencies associated with majority voting.

Chapter 17 Voting with Ballots and Feel 489
The Simple Tiebout Model

There are five assumptions in the simple version of the Tiebout model:
1 Jurisdictional choice. A household "shops" for a local government and

chooses the jurisdiction that provides the ideal level of local public goods.
There are enough jurisdictions to ensure that every household finds the
perfect jurisdiction.
2. Information and mobility. All citizens have access to all relevant

information about the alternative jurisdictions, and moving is costless.
3. No interjurisdictional spillovers. There are no spillovers (externalities)

associated with local public goods: all the benefits from local public goods
accrue to citizens within the local jurisdiction.
4. No scale economies. The average cost of production is independent of
output.
5. Head taxes. Local governments impose head taxes to pay for public goods:
if you have a head, you pay the head tax.
Citizens in the Tiebout world sort themselves into municipalities and school
districts according to their demands for local public goods. People with relatively
low demand for parks choose a municipality with a relatively small park budget (and
a relatively low head tax). Citizens with relatively high demand for parks choose a
municipality with a relatively large park budget. Citizens vote with their feet: if they
are dissatisfied with their current municipality, they move to another one.
Efficiency Effects of the Tiebout Process
The simple model suggests that shopping and sorting increase the efficiency of
local governments. As explained earlier in the chapter, majority voting generates an
inefficient level of local spending if the desired budget of the median voter differs
from the optimum budget. One of the reasons for the inefficient outcome is that local
citizens have different preferences for local public goods, so they disagree about the
appropriate spending level. Under the Tiebout process, households sort themselves
into homogeneous communities, so citizens agree on how much to spend.
To explain the efficiency effects of the sorting process, suppose that a region
starts out with three identical municipalities. Each community has three citizens, one
Sam-type citizen (who wants a small park), one Marian-type (medium-sized park),
and one Bertha-type (big park). In other words, each municipality is heterogeneous
with respect to the demand for parks. As shown in Figure 17-1, each heterogeneous
municipality chooses Am (the choice of the median voter), which is less than the
optimum park size.
Suppose that the Tiebout process causes households to sort themselves into ho-
mogeneous municipalities. One municipality has three Sams (municipality S); one
has three Marians (municipality M); one has three Berthas (municipality B). Fig-
ure 17-3 shows the spending choices in municipality S. The marginal-social-benefit
Figure 17-3 Citizens in a Homogeneous Municipality Choose

the Optimum Spending Level
Marginal
cost
Marginal cost w ith
equal cost shares
Park si/e (acres i
Municipality S has three Sam-type households (who want a small park), so the marginal social
benetii is three times the marginal private benefit. If costs are divided equally between the three
citizens, they all agree to a park size of A, acres. The inefficiencies of majority voting are
eliminated by the sorting of households into homogeneous municipalities.
curve is three times the marginal-benefit curve of Sam. and intersects the marginal-
cost curve at a park acreage of .4,. If costs are divided equally between the three
households, the marginal benefit equals the marginal cost at the optimum budget
(A 5 ). All three households have the same preferences for parks, so there is unanimous
support for the optimum park size.
The same analysis applies to decision making in the two other homogeneous
communities. In each municipality, every household has the same marginal-benefit
curve, so an equal sharing of park costs causes unanimous support for the opti-
mum park size. In municipality M. medium-sized park.
the three Marians vote for a
In municipality B. the three Berthas vote for a big park. The potential inefficien-
cies o\ majority voting are avoided by the sorting of households into homogeneous
communities.
Tiebout Model with Property Taxes
Hamilton (1975) made the Tiebout model more realistic by introducing property
taxes into the model. Hismodel assumes that local governments finance their spend-
ing programs with a property tax, not a head tax. To explain the effects of the property
tax. consider a municipality with the following characteristics:
1 Ever) household demands $500 worth oi park spending.

2. There are two income groups: the wealth) live m S75.000 houses, and the
poor live in $25,000 houses.
3. Under a head tax, every household pays $500 in local taxes to support the
parks.
Suppose the municipality switches from the head tax to a 1 percent property tax.
Under the property tax, each wealthy household pays $750 in taxes (up from $500),
and each poor household pays $250 (down from $500).
How does the property tax affect location choices? The wealthy, who now pay
$750 for $500 worth of parks, will try to decrease their tax liabilities. One option is
to set up a new municipality for wealthy households. If every household in the new
wealthy municipality owns a house worth $75,000, a property tax of 0.667 percent
generates a tax bill of $500 per household. In other words, if the wealthy households
form an exclusive municipality, they again pay $500 in taxes for $500 worth of parks.
How do the wealthy establish their exclusive suburb? If it were legal, they
might simply outlaw houses worth less than $75,000, thus guaranteeing that every
household would pay at least $500 in taxes. Because this is illegal, they use large-lot
zoning: the municipality establishes a minimum lot size for every dwelling. Since
land and housing are complementary goods, the municipality can control housing
consumption by controlling land consumption. One rule of thumb is that the market
value of land is about one-fifth of the total property value (the value of the structure
and the land). If the price of land is $30,000 per acre, a half-acre lot ($15,000 worth
of land) generates a property value of $75,000 (five times $15,000). Therefore, a
minimum lot size of half an acre meets the target property value.
What about the poor people who remain in the original municipality? The
wealthy households are eventually replaced by poor households in less expensive
housing. The average property value in the original municipality decreases, requir-
ing an increase in the tax rate to generate $500 of taxes per household. If the average
property value falls to $25,000, the property tax rate rises to 2 percent.
The switch from the head tax to the property tax causes households to sort them-
selves with respect to housing consumption. There is one community for households
with a large demand and one for households with a small demand for
for housing
housing. This sorting occurs because tax liabilities are based on housing consump-
tion. If the municipalities were to tax Spam instead of housing, citizens would sort
—
themselves into two communities one where people consume large quantities of
Spam and one where people consume small quantities.
The switch from a head tax to a property tax increases the equilibrium number
of municipalities. Citizens sort themselves with respect to two characteristics: the
demand for local public goods and the demand for housing. The equilibrium number
of municipalities depends on the number of housing types and the number of public-
good types. Suppose that there are three income groups (rich, middle-income, and
poor) and three types of houses, large (occupied by the wealthy), medium (middle-
income), and small (poor). Suppose further that for each income group there are
two different levels of demand for public goods: some households in each income
group have relatively high demands for local public goods, and others have relatively
low demands. Given these assumptions, there will be six communities: three high-
spending municipalities (one with large houses, one with medium houses, and one
with small houses), and three low-spending municipalities. If local governments
492 PartIV Local Government
used a head tax instead of a property tax. there would be only two municipalities, a
high-spending one and a low-spending one.
Capitalization and the Tiebout Model

Most municipalities are not homogeneous with respect to property values. The sort-
ing of households with respect to property values is imperfect: there are some small
houses in municipalities where most of the houses are large. A small house sur-
rounded by large houses has a lower tax liability than a small house surrounded
by small houses because the large-house community has a lower property tax rate.
In the example above, a $25,000 house surrounded by $75,000 houses might have
a tax rate of 0.667 percent, compared to a tax rate of 2 percent for a small house
suiTounded by small houses.
As explained in the chapter on housing, the market price of housing depends
on the characteristics of the dwelling and the site. Among the site characteristics
are the local tax liability and the quality of public services. Everything else being
equal, a dwelling with a lower tax liability commands a higher price: the small house
in the large-house (low-tax) municipality costs more than an equivalent house in a
small-house (high-tax) municipality. In equilibrium, the consumers of small houses
must be indifferent between the small houses in the two communities, and they bid
up the price of the small house in the low-tax community until the price difference is
large enough to offset the tax difference. In other words, lower taxes are capitalized
into the market value of the small house.
The capitalization of taxes diminishes the incentive for free riding. If the small-
house buyer moves to a community with a relatively large tax base and a low tax
rate, he pays a premium for the small house. In other words, the benefits of low taxes
are offset by higher housing costs.
Evidence of Shopping and Sorting
There is good deal of evidence that citizens shop for municipalities and school
a
districts.Using data for the New York metropolitan area. Oates 1969) estimated (
the relationship between house value, taxes, and school expenditures. He found that
communities with relatively low tax rates had higher property values. This result
suggests that people shop for school districts, bidding up the price of housing in
areas with favorable combinations of taxes and local public goods. In other words,
differences in taxes and expenditures are capitalized into house values.
There is also evidence that citi/ens sort themselves with respect to their de-
mands tor local public goods. Gramlich and Rubinfeld 1982) examined the tastes
(
for local public goods in different metropolitan areas. In a metropolitan area with
a single large municipality, households have no opportunity for shopping, so the
single municipality contains households with a wide variety of demands for lo-
cal public services. In contrast, if the metropolitan area has a large number of
jurisdictions, each household can pick a municipality that pun ides the preferred
level of public services, so households sort themselves into homogeneous commu-
Chapter 17 Voting with Ballots and Feel 493
Using data from questionnaires, Gramlich and Rubinfeld concluded that the
nities.
greater thenumber of municipalities to choose from, the more homogeneous the

demand for local public goods within each municipality. Heikkila (1996) shows that
in the Los Angeles metropolitan area, with over 100 local governments, people of
similar preferences for local public goods cluster, forming relatively homogeneous
communities.
There is substantial variation across metropolitan areas in the number and size
of local governments within the metropolitan area (Fisher and Wassmer, 1998).
Among the 168 metropolitan areas with a 1980 population of at least 200,000, the
average municipal population varied from 2,300 to 316,000. Fisher and Wassmer
show that municipalities are more numerous — and smaller— in metropolitan areas
where there is more variation among citizens demand for local public goods.
in the
In a metropolitan area with very little demand for local-government

variation in the
services (a result of little variation in income, age, and race), the expected number
of general-purpose local governments is five and the expected number of school
districts is one. In a metropolitan area with more variation in income, age, and
race —and thus more variation in demands for local public goods — the expected
number of general-purpose local governments is 101 and the expected number
of school districts is 54. These results are consistent with the idea that the num-
ber of local governments increases with variation in the demand for local public
goods.
The Tiebout process is primarily a suburban phenomenon. In the suburbs, house-
holds can choose from dozens of municipalities and school districts. In the central
city, there is typically a single large jurisdiction. One reason for the suburbanization
of population is that the only option for central-city foot voters is to move to a sub-
urban municipality. If large central-city governments were broken up into smaller
municipalities, central-city residents would be able to vote with their feet without
moving to the suburbs.
Critique of the Tiebout Model

Like any simple model, the Tiebout model can be criticized on a number of grounds.
Some of its assumptions are unrealistic. For example, the model assumes that house-
holds are perfectly informed and perfectly mobile, so they can easily change juris-
dictions if they become dissatisfied with their local government. This is clearly
unrealistic: information about alternative locations is costly to acquire, and the cost
of moving is high. As explained in Chapter 14 (Why Is Housing Different?), most
households tolerate some dissatisfaction with their dwellings because moving is
costly. In context of the unhappy with its local

Tiebout model, a household that is
government will switch jurisdictions only if the benefits from moving to a differ-
ent jurisdiction are large enough to offset the associated moving cost. Although the
shopping and sorting predicted by the Tiebout model certainly occur, the matching
of households and jurisdictions is by no means perfect.
Another assumption of the Tiebout model is that there are enough local govern-
ments to allow every household to pick the ideal jurisdiction. In the Tiebout world,
494 Part [V Local Government
every household finds a local government that spends exactly the right amount on
every public good. This is clearly unrealistic: there are simply not enough local
jurisdictions to provide every household with the ideal level of even public good.
This problem is complicated by the use of the property tax: if local governments use
the property tax instead of a head tax. households must also sort themselves with re-
enough jurisdictions to allow every household to
spect to house value. If there are not
choose the ideal jurisdiction, citizens compromise on their ideal tax-spending pack-
age, choosing a jurisdiction that comes closest to their ideal package. Households
express their preferences for local public goods in local budget elections: after the
household votes with its feet, it starts voting with ballots.
A final criticism of the Tiebout model concerns its efficiency implications. The
model suggests that the fragmented system of local government is efficient because
the sorting of households allows each household to consume its ideal level of local
public goods. A key assumption of the model is that local public goods do not gener-
ate interjurisdictional externalities: all the benefits and costs of local public goods are
confined within the local jurisdiction, regardless of the size of the jurisdiction. In the
Tiebout world, even a one-person jurisdiction is efficient: the model assumes that all
the benefits of spending on police, fire protection, sewerage, and education would
be confined to the single person in the jurisdiction. Under this assumption, there
is no real difference between local publicgoods and private goods. So the house-
hold's choice of a jurisdiction, however small, is as efficient as its choice of private
goods.
The goods provided by local governments are indeed different from private
goods. Some of the goods are nonrivalrous in consumption. For example, when a
small municipality uses its police force to capture a criminal, people outside the
municipality benefit. The municipality ignores these external benefits, and therefore
spends too little on police services. Some goods provided by governments
local
generate externalities for people in surrounding areas. If a child in one city
educated
moves workers and citizens in
to another city as an adult, the second cit\ benefit from
the education program of the first. An individual citj ignores the benefits of its edu-
cation program on outsiders, and therefore spends too little on education. In general,
the goods provided by local governments are nonrivalrous and generate externalities.
so Tiebout's fragmented system of local governments may be inefficient.
Metropolitan Consolidation and the Tiebout Model

The fragmented system o( local government predicted by the Tiebout model has
good points and bad points. On the positive side, citizens can choose from a wide
variet) of jurisdictions, so the fragmented system accommodates diversity in de-
mand for local public goods. On the negative side, there are likely to be spillover
benefits and costs that cause inefficiencies. Under what circumstances do the costs
o( the fragmented system dominate the benefits? Alternatively, under what circum-
stances would metropolitan consolidation improve the efficiency o( local
governments?
Chapter 1 7 Voting with Ballots and Feel 495
Figure 17-4 Interjurisdictional Externalities and Local Decision Making
Marginal social cost
Marginal social benefit = Marginal local

benefit plus marginal external benefit
Marginal local benefit

3 times MBr
Marginal cost with

equal cost shares
Park size (acres)
If municipal spending generates benefits for people outside the municipality, the marginal social benefit
exceeds the marginal local benefit. The optimum spending level is A*. Local voters, basing their decisions on
local benefits and costs, choose less than the optimum spending level (A, ). A subsidy equal to the gap between
the local benefit and the social benefit would cause the municipality to choose the optimum spending level.
Inefficiency from Spillovers
Figure 17-4 shows the effects of benefit spillovers on the efficiency of municipal
government. The marginal social benefit of parks equals the marginal local benefit
(the benefit of citizens within the municipality) plus the marginal external benefit
(the benefit of people outside the municipality). The municipality chooses the level
at which the marginal social benefit equals the marginal cost (A s ). The optimum
level is where the marginal local benefit equals the marginal cost (A*). Because
the municipality ignores the spillover benefits of its parks, it spends less than the
optimum amount on parks. The inefficiency occurs because the municipality is too
small to contain all the beneficiaries of its spending programs.
Figure 17-5 shows the inefficiency of municipal government in the metropolitan
area with three municipalities, S (small parks), M (medium-sized parks), and B (big
parks). Under the municipal system, there are spillovers: in each community, the
marginal local benefit of spending is less than the marginal social benefit, so each
municipality spends less than the optimum amount. For example, in community S,
the marginal local benefit equals the marginal cost at a spending level 5", which is
less than the optimum level S*. Similarly, M and B spend less than the optimum
amounts (Af ' < M* and B' < B*).
Effects of Consolidation
Suppose that the three municipalities are consolidated into a single metropolitan
government. The metropolitan government is large enough to contain all the citizens
496 Part l\' Local Government
Figure 17-5 Municipal versus Metropolitan System of Government
Community S Community M Community B

Small parks Medium parks Big parks
MLB_ MSB MLB,,, MSB, MLB, MSB,,
Marginal
cost
There are two options For local decision making. Under the municipal system, each community chooses Us own level
of spending, ignoring interjurisdictional spillover benefits. Municipality S chooses 5'. chooses M and B chooses M ,
/>' In each oi the municipalities, spending falls short of the optimum spending level (V. \l Under a .
/>' i
metropolitan government, the median voter Marian in community M) picks the metropolitanwide spending level
<
i\/ The metropolitan spending level exceeds the optimum in community S A/'
i. i ,s" and falls short of the
• i
optimum in community B i U' - A' |
affected by local spending, so citizens base their votes in the metropolitan election
on the social benefits of parks. For example, citizens in community S base their
votes on the marginal social benefits (MSB, not the marginal local benefits (MLB,
) ).
Similarly, citizens in communities M

and B base their votes on MSB,,, and MSB/,.
Given the marginal-cost curve, the preferred budgets for the three types of household
are S*,M*, and B*
Suppose that the metropolitan government intends to build three parks, one in
each community. The three parks will be the same size. If the metropolitan govern-
ment holds an election to determine the size of the parks, the winning size is the size
preferred by the median voter. M*. Therefore, each community will get a park with
M* acres.
The spending under the metropolitan government is inefficient because it
chooses a single park size for all three communities. The actual size (
M* ) exceeds the
optimum size for low-demand households (S*) and falls short of the optimum size
for the high-spending household (B*). Under the metropolitan government, exter-
nalities are internalized, but two of the three communities are forced to compromise
on their ideal park sizes.
Is the metropolitan system superior to the municipal system? The municipal

system accommodates diversity in demand, but suffers from the externality problem.
The metropolitan system solves the externality problem, but requires two of the
communities to compromise on their ideal spending packages. It's not clear from
figure 17-5 which of the two systems of government is more efficient. Under both
systems, the actual spending differs from the optimum combination (.V. M* B*) by .
Chapter 1 7 Voting with Ballots and Feet 497
Figure 17-6 Superiority of the Municipal System if Diversity in Demand Is

Large Relative to Externalities
Community B
Big parks
Marginal
cost
S' S* M' M* B' B*

If externalities are relatively small,municipal decision making will be relatively efficient: the gaps between the
equilibrium and optimum spending amounts (5* — 5'; M' — M'; B' - B ') will be relatively small. If the diversity in
demand is relatively large, metropolitan decision making will be relatively inefficient: the gaps between the
equilibrium (A-/* and optimum spending amounts (A/* — S': B" — M") will be relatively large. If externalities
)
(spillovers) are small relative to diversity in demand, the municipal system will be more efficient than the
metropolitan system.
large amounts: the municipal system generates a spending package of S', M\ and B'
and the metropolitan system generates a spending of M* in all three communities.
If the diversity in demand is large relative to externalities, the municipal system
will be more efficient than the metropolitan system. In Figure 17-6, interjurisdic-
tional externalities are relatively small, so the gaps between the local- and social-
benefit curves are relatively small. Therefore, under municipal decision making, the
gaps between the equilibrium and optimum amounts of park acreage are relatively
small. For example, the small-park municipality spends S\ which falls short of the
optimum spending S* by a relatively small amount. Similarly, the gaps between
M* and M' and between B* and B' are relatively small. In contrast, because di-
versity in demand is relatively large, the compromises associated with metropolitan
government are relatively large. The gaps between M* (the budget chosen under ma-
jority rule) and the optimum budget of community S (M* — S*) is relatively large.
Similarly, the gap between M* and B* is relatively large. Therefore, the municipal
system is more efficient than the metropolitan system.
In general, metropolitan government will be more efficient than municipal gov-
ernment if interjurisdictional spillovers are large relative to diversity in demand. In
this case, the advantages of small local government (the ability to accommodate
diverse demands for local public goods) are relatively small, and the disadvantages
(the inefficiencies associated with externalities) are relatively large. Therefore, a
metropolitan system of government will be more efficient.
Summary
1 The median-voter model predicts that the government will adopt the preferred
budget of the median voter (the voter that splits voters into equal halves, with
one-half preferring smaller budgets and the other half preferring larger
budgets).
a. Majority rule does not necessarily produce the optimum level of spending.
b. The median voter is decisive in both a direct democracy and a representative
democracy.
c. Economists have used the median-voter model to estimate the elasticities of
demand for local public goods.
2. The inefficiencies resulting from majority rule could be eliminated by the use of
Lindahl taxes (tax equal to the marginal benefit of the local public good).
3. According to the Tiebout model, citizens shop for jurisdictions and sort
themselves with respect to demands for local public goods. Because the Tiebout
model assumes that there are no externalities, the Tiebout process is just as
efficient as shopping for private goods.
4. The property tax increases the equilibrium number of communities: households
sort themselves with respect to both the demand for local public goods and
housing consumption.
a. The wealthy (consumers of large houses) use large-lot zoning to exclude the
poor (consumers of small houses) from wealthy communities.
b. Differences in taxes are capitalized into property values, so a small house in
a wealthy community (with a low tax rate) costs more than a small house in a
poor community (with a high tax rate).
5. There is a good deal of evidence for the Tiebout process.
a. Citizens shop for local governments and sort themselves into relatively
homogeneous communities.
b. Most of the shopping is in the suburbs: large central-city governments limit
the shopping options in the central city.

1 In the example shown in Table 17-1. the school board started with a budget of
$90 and worked its way downward until a majority of voters voted yes. Show
that the same outcome would occur if the school board stalled with a budget of
$10 and worked upward.
2. In the example shown inTable 17-1. the winning budget is $70. Explain the
effects of the following changes on the winning budget. In each case, use as
the starting point the distribution of voters shown in Table 17-1.
a. Four elderly voters move into the school district. The elderly voters have
preferred budgets of $20. $25. $30. and $35.
b. In response to "A Nation at Risk," a report critical of educational programs,

the desired spending of every household increases by $5.
3. Consider the effect of private schools on the local school election. Suppose
that households F and G (in Table 17-1) enroll their children in private
schools. Predict the effects of private schooling on the public school budget
under the following two cases.
a. F and G abstain from voting in the school election.
b. F and G vote in the election. Their preferred budgets drop by $40
($44 instead of $84 and $5 1 instead of $9 1 ).
4. Comment on the following: "If the median-voter model is correct, it's silly to
mess around with elections. We should hire Mr. Gallup to identify the median
voter and find out his or her preferred budget. Such a system would be cheaper
than a series of budget elections."
5. You are the campaign manager for Dick Tracy, who is running for mayor. His
opponent has committed herself to spending $500 per capita on police
services. You must decide how much Dick will propose for the police budget.
What information would you collect, and how would you use it?
6. The conventional head tax is amount per head: if you have a head, you
a fixed
pay a $500 tax. Suppose that the head tax is based on head weight: the heavier
your head, the larger your tax. Would you expect the switch from the
conventional property tax to the head-weight tax to increase or decrease the
equilibrium number of municipalities?
7. In Metro, there are three types of houses: E (expensive), M (medium), and C
(cheap). A recent survey suggests that there are three types of preferences for
fire protection: H (high), I (intermediate), and L (low). Mr. Wizard, an
economic consultant to the city, recently made the following statement: "If my
assumptions are correct, the shopping and sorting from the Tiebout process
will generate three municipalities."
a. Assume that is correct. What are his assumptions?
Mr. Wizard's reasoning
b. Suppose number of municipalities turns out to be nine,
that the equilibrium
not three. Where did Mr. Wizard go wrong? What set of assumptions would
have given him the correct prediction?
8. Metro is a metropolitan area in which all households demand the same
spending per pupil on public education. Education is produced with constant
returns to scale. Because schools are financed with the property tax, there are
dozens of school districts. According to Peabody, a fiscal analyst in Metro, "If
my assumptions are correct, if we replace the property tax (a tax on market
value) with a tax based on square footage (e.g., $2 per square foot of living
space per year), the equilibrium number of school districts will decrease to
one."
What are Peabody's assumptions?
9. Under the traditional public school system, parents must send their children to
the neighborhood school. In recent years, many people have suggested that
parents be able to send their children to the public school of their choice, not
necessarily the neighborhood school. Suppose that this change is
implemented. Discuss the effects on the Tiebout shopping and sorting process.
Would there be more or less income segregation?
10. In the example of the Tiebout model with a property tax. wealthy households
established an exclusive suburb with a property tax rate of 0.667 percent,
spending of $500. and a minimum lot size of 0.50 acres. Discuss the effects of
the following changes on the property tax rate and the minimum lot size.
a. The market value of houses is SI 00.000 instead of $75,000.

b. The market value of land is $50,000 instead of $30,000 (and the market
value of houses is $100,000).
1 1 Consider Metro, a metropolitan area with three school districts. East. West,
and North. The optimum spending is $500 in East. $900 in West, and $700 in
North. If the districts make independent decisions, spending will be $450 in
East. $850 in West, and $650 in North. In other words, there are externalities
in the provision of education, causing each district to spend too little. Suppose
that the three districts are consolidated into a single metropolitan school
district.
a. Assume that there are no scale economies in the provision of schools. Will
the metropolitan school district be more or less efficient than the original
arrangement?
b. How would your answer to (a) change if the expenditures under
independent decision making were $200 (East) and $600 (West)?
c. How would your answer to (a) change if there are scale economies in the
provision of education?
12. The example shown in Figure 17-2 assumes that all citizens vote in the
mayoral election. Suppose that some citizens refuse to vote if the proposed
budgets of the two candidates are not very far apart (the) abstain because of
indifference). Specifically, suppose that if the difference between the
candidates is less than $1. only 50 percent of citizens (of all budget
preferences) vote.
a. Will Buck increase his chances of being elected if he increases his proposed
budget from $6 (the median budget) to $7?
b. Will the median-voter result persist if there is indifference?
13. The example shown in Figure 17-2 assumes that all citizens vote in the
mayoral election. Suppose that some citizens refuse to vote if the proposed
budget of the closest candidate is too far from their preferred budget (they
abstain because of alienation). Specifically, a voter abstains if the difference
between the closest candidate and the citizen's preferred budget is greater
than $3.
a. Suppose that Penny decreases her proposed budget to $5. Does she increase
or decrease her chances of being elected?
b. Will the median-voter result persist if there is alienation?
Chapter 1 7 Voting with Hal lots and Feci 50

General
Inman, Robert. "The Fiscal Performance of Local Governments: An Interpretive Review."
In Current Issues Urban Economics, ed. Peter Mieszkowski and Mahlon
in
Straszheim. Baltimore: Johns Hopkins University Press, 1979. pp. 270-321. Discusses
various models of decision making in the local public sector, including the median-
voter model and the dominant-party model. Summarizes studies of the price and
income elasticities of demand for local public goods and the elasticity of local
expenditures with respect to intergovernmental aid.
Median-Voter Model
Barr, J., and O. Davis. "An Elementary Political and Economic Theory of Local
Governments." Southern Economic Journal 33 (1966), pp. 149-65.

Bergstrom, T, and R. Goodman. "Private Demand for Public Goods." American Economic
Review 63 (1973), pp. 286-96. Derives the conditions under which the chosen budget
is equal to the preferred budget of the median-income family.
Black, D. "On the Rationale of Group Decision Making." Journal of Political Economy 56
(1948), pp. 88-95.
Downs, Anthony. An Economic Theory of Democracy. New York: Harper & Row, 1956.
Chapters 3, 8, and 14 discuss the rational voter and the median-voter model.
Edelson, N. "Budgetary Outcomes in a Referendum Setting." In Property Taxation and the
Finance of Education, ed. R. Lindholm. Madison: University of Wisconsin Press, 1974.
Hotelling, Harold. "Stability in Competition." Economic Journal 39 (1929), pp. 41-57. The
basis for the median- voter model. Shows why politicians are like ice-cream vendors.
Inman, Robert. "Testing Political Economy's 'As If Proposition: Is the Median Voter
Really Decisive?" Public Choice 33 (Winter 1978), pp. 45-65. Tested the
median-voter model for local public schools and found that the decisive voter was in
median-income homeowner whose children attended public schools. The

fact the
chosen budget moved away from the preferred budget of the median voter as the
number of renters, elderly, and private-school families increased.
Rubinfeld, D. "Voting in a Local School Election: A Micro Analysis." Review of Economics
and Statistics 59 (1977), pp. 30-42.
The Tiebout Model of Interjurisdictional Mobility

Bewley, T. "A Critique of Tiebout's Theory of Local Public Expenditures." Econometrica
49 (1981), pp. 713^40. Discusses the conditions under which the Tiebout process is
efficient.
Eberts, Randall W., and Timothy J. Gronberg. "Jurisdictional Homogeneity and the Tiebout
Hypothesis." Journal of Urban Economics 10 (1981), pp. 227-39. Discusses the
problems associated with a limited supply of local jurisdictions.
Hamilton, Bruce W. "Capitalization of Interjurisdictional Differences in Local Tax Prices."
American Economic Review 66 (1976), pp. 743-53. Explores the roles of zoning and
capitalization in attaining a Tiebout equilibrium in a community whose citizens have

different demands for housing.
"Zoning and Property Taxation in a System of Local Governments." Urban Studies 12
(1975). pp. 205-1 1. Discusses the role of zoning in maintaining the Tiebout
equilibrium under property tax finance.
Hamilton. Bruce W.; Edwin S. Mills; and David Puryear. "The Tiebout Hypothesis and
Residential Income Segregation." In Fiscal Zoning and Land Use Controls, ed. Edwin
S. Mills and Wallace Oates. Lexington. Mass.: D. C. Heath, 1975. Discusses the effects
of fragmented government on income segregation.
Henderson, J. V. "Theories of Group, Jurisdiction, and Size." In Current Issues in Urban
Economics, ed. Peter Mieszkowski and Mahlon Straszheim. Baltimore: Johns Hopkins
University Press, 1979, pp. 235-69. Discusses the allocation of population among local
communities.
Oates. Wallace E. "On the Use of Local Zoning Ordinances to Regulate Population Flows
and the Quality of Local Services." In Essays in Modem Labor Market Analysis, ed.
Orley Ashenfelter and Wallace E. Oates. New York: John Wiley & Sons. 1977, pp.
201-19.
Pack, H., and J. Pack. "Metropolitan Fragmentation and Local Public Expenditure."
National Tax Journal 31 ( 1978), pp. 349-62. Discusses the problems associated with a
limited supply of local jurisdictions.
Tiebout. Charles. "A Pure Theory of Local Expenditures." Journal of Political Economy 64
(1956), pp. 416-24. The original statement of the Tiebout model.
Tests of the Tiebout Model

Fisher.Ronald C., and Robert W. Wassmer. "Economic Influences on the Structure of Local
Government in U.S. Metropolitan Areas." Journal of Urban Economics 43 1998), (
pp. 444-70.
Gramlich, Edward M., and Daniel L. Rubinfeld. "Microestimates of Public Spending
Demand Functions and Test of the Tiebout and Median- Voter Hypothesis." Journal of
Political Economy 90 (1982), pp. 536-60. Found relatively small differences in the
demands for local public goods in metropolitan areas where there are many local
jurisdictions.
Heikkila, EricJ. "Are Municipalities Tieboutian Clubs'?" Regional Science and Urban
Economics 26 (1996), pp. 203-26.

Oates, Wallace E. "The Effects of Property Taxes and Local Public Spending on Property
Values: An Empirical Study of Tax Capitalization and the Tiebout Hypothesis."
Journal of Political Economy 11 (1969). pp. 957-70. Estimates the effects of local
taxes and expenditures on property values. Provides evidence that households shop for
local jurisdictions.
Zodrow, George. The Tiebout Model after Twenty-Five Years. New York: Academic Press,
1983. A collection of papers on the Tiebout model.
Chapter 18
C A L TAXES and
INTERGOVERNMENTAL GRANTS
The thing generally raised on city land is taxes.

Charles Dudley Warner
(^~7~ his chapter explores the revenue side of local government, discussing the
l/ economics of local taxes and intergovernmental grants. When a government
imposes a tax, the people who pay the tax in a legal sense have the incentive to change
their behavior to avoid the tax. By doing so, they shift it to someone else. The policy
question is: after taxpayers have fully adjusted to the new tax, who actually pays
the tax? The first two parts of the chapter answer this incidence question for the
residential property tax, and the third part answers the incidence question for the tax
on commercial and industrial property. The fourth part discusses the effects of the
Tiebout shopping/sorting process on the distribution of the burden of the property tax.
The fifth part discusses the administration of the property tax. The last four parts of
the chapter discuss the economics of intergovernmental grants, which provide about
two-fifths of the revenue for local governments. We'll explore the responses of local
governments to various types of grants, focusing on how local governments allocate
grant money between additional spending on local public goods and tax cuts. One
issue concerns states' responses to the welfare-reform plan enacted in 1996. The
new plan replaces matching grants from the federal government with block (lump-
sum) grants. The question is: by how much will spending on welfare programs
decrease?
Who Pays the Residential Property Tax?

The property tax is a tax on residential, commercial, and industrial property. This
section considers the effect of the residential property tax, using a model of a
503
residential city to explore the market effects of the tax. The analysis focuses on
two principal questions about the residential property tax. First, how do property
owners respond to the tax? Second, to what extent do property owners pass on the
tax to consumers, landowners, and other property owners?
It is useful to divide rental housing into two components: the dwelling itself
(the structure) and the land that comes with the dwelling. The annual rent on the
residential property is the sum of dwelling rent and land rent. For example, an annual
rent of $5,000 can be broken down into $4,000 of dwelling rent and $ 1 .000 of land
rent. Similarly, the market value of residential property is the sum of the market
values of the dwelling and the land. As explained in Chapter 7 (Introduction to
Land Rent and Land Use), the market value of property equals the present value
of the rental income from the property. If the market interest rate is 10 percent,
the market value of the dwelling is $40,000 ($4,000/0. 10), the market value of the
land $10,000 ($1,000/0.10), and the market value of the residential property (both
is
dwelling and land) is $50,000.

The residential property tax is an annual tax based on the market value of the
property. For example, under a 2 percent property tax, the owner of a $50,000
property would pay $1,000 per year in taxes. This can be broken down into an $800
tax on the dwelling itself (2 percent of a market value of $40,000) and a $200 tax on
the land (2 percent of a market value of $10,000).
To explain the market effects of the property tax, consider the city of Taxville,
where all land is used for rental housing. The city has the following characteristics:
1 . Identical dwellings. The city has 900 identical rental housing units.
2. Housing inputs. There are two inputs to housing: land and improvements
(the dwelling). Every dwelling occupies a quarter-acre lot.
3. Capital mobility. Dwellings are immobile in the short run, but can be
moved from one city to another in the long run.
4. Unit property tax. The annual property tax is paid in legal terms by the
property owner. The tax is $200 per quarter-acre lot and $800 per dwelling,
regardless of the market values of land and improvements.
The fourth assumption means that the property tax is not based on market value, but
is a fixed unit tax. This assumption simplifies the exposition without changing the
analysis in any substantive way.
The Land Portion of the Property Tax

Consider first the effect of the land portion of the property tax. Figure 18-1 shows
the city's land market. The supply of land is perfectly inelastic, with a fixed supply
of 900 quarter-acre lots. The demand curve intersects the supply curve at point B.
generating an annual rent of $1,000 per lot. If the market interest rate is 10 percent,
market value of land is $10,000 per lot.
Can landowners shift the land tax onto land consumers? The property tax
increases the property owner's tax liability by $200 per year. Suppose that the
Chapter 18 Local Taxes and Intergovernmental Grants 505
Figure 18-1 Market Effects of the Land Tax

Supply
1.200
^^c
i
B
8 l.ooo
1
S 800 , n
i
i
^^ Demand
i
V i
i 1
450 900 1,350

Quantity of land number of quarter-acre
( lots)
Land rent is SI. 000 in the pre-tax equilibrium. Since the supply of land is fixed, the land tax decreases rent
by the amount of the tax ($200). If the landowner tried to pass on the $200 annual tax to consumers, the net
cost of land would rise to $1,200 (point C ). and the quantity supplied (900) would exceed the quantity
demanded (450). The surplus of land would cause rent to fall to $800.
property owner increased land rent from $1,000 per year (point B) to $1,200
(point C). At point C, the quantity of land demanded (450 lots) is less than the
fixed supply (900 lots), so some land is The excess supply of land causes
vacant.
landowners to cut prices to attract consumers. The price cutting continues until
thegap between supply and demand is eliminated. In equilibrium, land rent falls to
$ ,000 and the net return to property owners drops to $800 ($1 ,000 less the $200 tax).
1
Because the supply of land is fixed, the land tax is ultimately borne by landowners.
The decrease in net rental income decreases the market value of land from $10,000
to $8,000 ($800/0.10).
The Improvement Tax: The Traditional View

The traditional analysis of the property tax examines the effects of the property tax
on the city that imposes the tax. The traditional view is based on partial-equilibrium
analysis. The analysis is "partial" in the sense that it ignores the effect of the tax on
other cities.
Figure 18-2 shows the market tor improvements (dwellings) in the taxing city.
In the long run, the supply of dwellings is assumed to be perfectly elastic: the city
is assumed market for dwellings, so the market price

to be a small part of a national
is unaffected by events in theGiven the supply and demand curves, the initial
city.
equilibrium (point B) generates a dwelling rent of $4,000 per year and 900 dwellings.
Figure 18-2 Partial-Equilibrium Effects of Improvement Tax
4.800 Supply after tax
C 4.400
*£ 4.000 Initial supply
J> 3.600 -
Demand
3.200
420 660 900 1.140
Quantity of land number of dwellings

(
Before the dwelling tax.B is the equilibrium point (rent = $4,000 per year). The property tax is $800 per
dwelling, so the supply tune shifts up by S800. Dwelling rent increases to S4.800 per year, and the
number of dwellings decreases to 420.
If the market interest rate is 10 percent, the market value of the rental dwelling is
$40,000.
Why is the long-run supply curve perfectly elastic? One way to picture the long
run is to imagine that propertyowners can put wheels on their dwellings and roll
them to other cities. Although property owners do not literally move their dwellings
from place to place, they can decrease the supply of housing in other ways. As
explained in the chapter on housing, property owners can decrease their invest- ( 1 )
ment in repair and maintenance, (2) abandon dwellings. (3) convert property to other
uses, and (4) not respond to increases in the demand for housing. Given these four
responses, property owners can decrease the effective supply of housing without
actually moving their dwellings to another location. The house-on-wheels metaphor
captures the idea that property owners can decrease the supply of housing in the
long run.
The property tax generates an annual tax of $800 per dwelling. Because the
property owner pays the improvement tax in a legal sense, the tax shifts the long-run
supply curve up by the amount of the tax: the supply curve is a marginal-cost curve,
and the tax increases the marginal cost of rental property by $800 per dwelling per
year. The market moves from point R to F: the equilibrium rent rises by the full
amount of the tax from $4,000 to $4,800). and the quantity of dwellings drops from
(
Chapter 1H Local Taxes and Intergovernmental Grants 507
900 to 420. Net rental income remains at $4,000: the property owner collects $4,800
in rent and pays $800 in taxes, leaving $4,000 in net income. The tenant pays the
entire tax: the annual cost of rental housing rises by $800.
The improvement tax is borne by consumers because the supply of dwellings is
perfectly elastic. The city is market for dwellings, so it

a small part of the national
takes the national rent of $4,000 as given: if the city's property owners were to receive
anything less than $4,000 in net rental income, they would roll their dwellings out
of town. In fact, the decrease in the supply of dwellings (from 900 to 420) increases
dwelling rent by an amount large enough to restore the original net rental income of
$4,000.
The New View of the Property Tax

The partial-equilibrium analysis explores the effects of the property tax on the mar-
kets for land and improvements in the taxing city. General-equilibrium analysis
explores the effects of the property tax on markets outside the city, providing a new
view of the property tax.
Suppose that Taxville shares a region with Notax, a residential city without a
property tax. The region has the following characteristics:
1 . Fixed supply of dwellings. The total number of dwellings in the region is
fixed at 1,800.
2. Capital mobility. In the short run, dwellings cannot be moved between

cities. In the long run, dwellings can be moved between cities at zero cost.
3. Household mobility. Households are immobile in the sense that they
cannot move from one city to another.
4. Equilibrium price. Property owners move their dwellings to the city with
the highest return (the highest annual rent). In equilibrium, the market rent
on dwellings is the same in the two cities.
The regional dwelling market is shown in Figure 18-3. Each city has a negatively
sloped demand curve, and the initial equilibrium occurs with points B (in Taxville)
and K (in Notax). The equilibrium price of dwellings is $4,000, and each city has
900 dwellings.
In the short run, the property owner pays the improvement tax. In the short
run, the supply of dwellings in each city is fixed. The property tax ($800 per year)
decreases the net rental income in Taxville from $4,000 (point B) to $3,200 (point T).
The property tax is borne by property owners because the supply of dwellings is
perfectly inelastic. In the nontaxing city, rent stays at $4,000.

In the long run, property owners move their dwellings to the more lucrative
market in Notax. Suppose that property owners move 120 dwellings from Taxville
to Notax. In Figure 18-3, the Taxville market moves from point B (900 dwellings) to
point C (780 dwellings), causing the rent on Taxville dwellings to rise from $4,000 to
$4,200. The net income of Taxville's property owners increases from $3,200 ($4,000
rent less $800 tax) to $3,400 ($4,200 less $800). Taxville's loss is Notax's gain: the
o Mi _;
.c
= u c
- a
li
§/ ^r
—
.5
o
^"3
2_
-=
yf _. (N
—~
c
~
'->
c r -
-a S
>>
u
CO
CO —s B
^ c !!
-
_| 5 s g
"3
c /
/
l '
1
i -- s;
s
- >> 5
>S S
a/ i
]
i
^2 =Z
S 5 =
u 3 =
I
s
-
L^^J ^_
E -
= -a HH
i
> and so
2 = long
$3,600.
a the
urvc)
return,
E is
ft
a s 1 s
-^: a = 7 .-
H j; -3 £. n
U: ~ -j — c
—
c?
l! ~ 9 •>
.£
EX)
*£ Jc c-.
_4J I I .c c- a II
o E
= > £ a i
X a 1 T3 E
03
-
H H -3 « 3q
>
« U | = | | Q
- T = >
^ 6 5
5 -f
1 f- 3 N* 5 B
B
u
U
w a .a J .g a
—
ai
—— »oc —
—
'= >
x
a
C
5 ^v E- i>
3 s o o c o
c 2 5 a o
o
5
U 30 \C "* <N O *C
508
Notax market moves from point K to point L (an increase of 120 dwellings, from
900 to 1,020), and rent drops from $4,000 to $3,800.
Is the dwelling market in equilibrium with points C and L? Equilibrium occurs
when two conditions are satisfied. First, the total number of dwellings in the two
cities must equal ,800 (the fixed supply). Second, property owners have no incentive
1
to move their dwellings from one city to another: the net rental income must be the
same in the two cities. For points C and L, the first condition is satisfied (780 + 1 ,020
= 1,800), but the second condition is violated: the net income in Taxville ($3,400)
is still less than net income in Notax ($3,800). Therefore, the market will continue
to move up the Taxville demand curve and down the Notax demand curve. The
movements along the demand curves will continue until both equilibrium conditions
are satisfied. Equilibrium occurs with points D and M: there are 660 dwellings in
Taxville and 1,140 in Notax, and the net income is $3,600 per rental dwelling.
There are two lessons from the general-equilibrium analysis:
1 Property owners pay the tax. The tax decreases net rental income in both
cities to $3,600. The movement of dwellings to the untaxed city decreases
rent in the untaxed city (to $3,600) and increases rent in the taxing city (to
$4,400). The tax is paid by property owners throughout the region, not just
property owners in the taxing city.
2. Consumer effects. Consumers in Taxville lose (they pay $4,400 instead of

$4,000), and consumers in Notax gain (they pay $3,600). The changes in
housing costs offset one another, so the tax does not change the average rent.
The general-equilibrium analysis suggests that the improvement tax is paid by

property owners (the owners of dwellings). Because the total (regional) supply of
dwellings is perfectly inelastic, property owners are unable to escape the tax. On the
other side of the market, housing consumers as a class escape the tax: the losses of
Taxville households are offset by the gains of Notax households. The improvement
tax causes zero-sum changes in rent, so housing consumers are no worse off as a
result of the tax.
Changing the Assumptions

The simple general-equilibrium model has a number of unrealistic assumptions.
How does the analysis change when these assumptions are dropped?
More than Two Municipalities. One of the key assumptions of the simple model
is that onlytwo municipalities share a fixed supply of dwellings. An alternative
assumption is that Taxville is one of 100 cities in a region, so that Taxville shares a
fixed supply of dwellings with 99 other cities.
When Taxville imposes a property tax, dwellings flee to 99 other cities, not just
to Notax. Suppose that 240 dwellings leave Taxville (the market moves from point B
to point D in Figure 1 8-3), so that each of the other 99 cities gets about 2.4 additional
dwellings. Each city experiences a relatively small increase in dwellings, and thus
a relatively small decrease in dwelling rent (e.g., from $4,000 to $3,996). Given the
small decrease in rent, there is still a gap between the net income on dwellings in
Taxville ($4,400 - $800 = $3,600) and net income elsewhere ($3,996). Therefore,
point D in Figure 18-3 is not an equilibrium point in the 100-city case.
Dwellings continue to leave Taxville until the income gap is closed. Given the
small price effect of fleeing dwellings, a large number of dwellings must leave the
taxing city. The market moves a relatively long distance up Taxville's demand curve.
In equilibrium, the market is near point F (dwelling rent is close to $4,800 and a
total of 480 dwellings have left the taxing city). Suppose that 475 dwellings leave
Taxville, meaning that the other 99 cities each gain about 4.8 dwellings. Dwelling
rent is $4,792 in Taxville and $3,992 in the other 99 cities. At this rent, the net income
of Taxville property owners ($3,992 = $4,792 — $800) equals the net income of
property owners in other cities.
The 100-city case has the same aggregate effects as the 2-city case. The tax is
paid by property owners: every property owner in the 100-city region receives $8 less
per dwelling. The only difference is that there are 50 times as many dwellings in the
100-city case (100 cities full of dwellings instead of 2 cities), so the decrease in
equilibrium rent in the nontaxing cities is one-fiftieth as large ($8 instead of $400).
In the taxing city, the tax increases dwelling rent by a relatively large amount because
a large number of dwellings leave the city.
Tax in Both The simple model assumes that the property tax is imposed in
Cities.
only one Suppose that both cities impose a $400 improvement tax. In the short
city.
run, the improvement taxes decrease the net income of property owners to $3,600.
Because both cities impose the tax, there is no incentive for property owners to move
their dwellings to the other city. Therefore, the short-run effect is the same as the
long-run effect: the tax decreases net income, but does not affect the market rent.
Property owners pay the improvement tax, just as they did in the single-tax case.
Because the tax does not change the equilibrium rent, it does not affect housing
consumers. Housing consumers escape the tax because dwellings do not move from
one city to another.
Variable Supply of Dwellings. The simple model assumes that the total supply of
dwellings is fixed. Although the tax decreases income from rental property, it does
not decrease the total supply of dwellings. If the total supply of dwellings decreases
as net rental income decreases, the analysis changes in three ways:
1 . Dwellings are withdrawn. Only some of the dwellings that leave Taxville
arrive in the untaxed city. Some of the dwellings are withdrawn from the
regional housing market. Suppose that one-third of the dwellings that leave
Taxville are transferred to the untaxed city, and two-thirds are withdrawn
from the regional market. In Figure 18-3, equilibrium occurs with points E
and /. (compared to D and M
if the total supply of dwellings is fixed). The
Chapter IX Local Taxes and Intergovernmental Grants 511
quantity of dwellings in Taxville decreases to 540 (a reduction of 360),

while the quantity of dwellings in Notax increases to 1,020 (an increase
of only 120), and 240 dwellings are retired from the market.
2. Property owners lose less. The withdrawal of dwellings increases net
rentalincome. At points L and E, dwelling rent is $3,800 in Notax and
$4,600 in Taxville. The net income for property owners in both cities is
$3,800 (compared to $3,600 with a fixed supply of dwellings). The tax
causes a smaller decrease in rental income because property owners
respond to the tax by decreasing the supply of dwellings.
3. Tenants pay part of the tax.The withdrawal of dwellings shifts part of the
tax to housing consumers. The tax increases dwelling rent to $4,600 in
Taxville and decreases rent in Notax to $3,800. In other words, the loss of
Taxville tenants (rent increase of $600) exceeds the gain of Notax tenants
(rent reduction of $200). The
harms Taxville tenants more than it helps
tax
Notax is worse off. Because property owners
tenants, so the tenant class
decrease the supply of dwellings, they can shift part of the improvement tax
to housing consumers.
Mobile Consumers. One of the assumptions of the simple model is that tenants
are immobile. When dwelling rent increases in the taxing city, none of the households
leave the city. What happens if households can move to the nontaxing city?
If households can move costlessly between the two cities, they will leave the
taxing city as long as it has a higher housing rent. If the market is at points D and
M in Figure 18-3, housing rent (the sum of dwelling rent and land rent) is $5,400
in the taxing city($4,400 for the dwelling plus $1,000 for the land) and $4,600
in theuntaxed city ($3,600 plus $1,000). When households leave the taxing city,
the demand for Taxville land decreases, decreasing land rent and housing rent. For
example, if land rent drops to $900, housing rent decreases to $5,300. At the same
time, the migration of households to the untaxed city bids up land rent, increasing
housing rent: if land rent increases to $1,100, housing rent increases to $4,700.
In other words, migration from Taxville
Notax narrows the difference between
to
housing rents in the two cities. two cities have the same
In locational equilibrium, the
housing rent. For example, if land rent drops to $600 in Taxville and rises to $1 ,400
in Notax, housing rent is $5,000 in both cities ($4,400 plus $600 in the taxing city
and $3,600 plus $1,400 in the untaxed city).
The migration of households causes capital gains and losses for landowners. If
the annual rent on Taxville land drops from $ 1 ,000 to $600, the market value of the
land drops from $10,000 to $6,000. In contrast, if rent on Notax land rises to $1,400,
its market value increases to $14,000. The migration of households eliminates the
intercity differences in housing cost. Instead of differences in housing cost, there are
differences in land rent and differences in the market values of land. In other words,
the migration of households allows consumers to shift the benefits and the costs of
the improvement tax to landowners.
Summary: Who Pays the Residential Property Tax?

The different views and models of the property tax are cleverly designed to baf-
fle students. Who actually pays the property tax? The incidence of the land tax
is straightforward: given the perfectly inelastic supply of land, the tax is paid by
landowners. The incidence of the improvement tax is not so easily summarized, as
it depends on one's perspective.
The Local Effects of a Local Tax

Tom is the mayor of a small city in a nation of 100 cities. To predict the effects of his
city's improvement he should use the partial-equilibrium model. Because his
tax.
city is small, it does not affect the national market for dwellings, so the assumption
of a perfectly elastic supply of dwellings is a good approximation of reality. Using
the partial-equilibrium model. Tom would predict that an $800 improvement tax will
increase dwelling rent by the full amount of the tax ($800): the tax will be borne
entirely by housing consumers.
Tom's analysis is not literally correct, but it is a good approximation. In one
of the examples in the previous section, an improvement tax in a 100-city region
increased dwelling rent in the taxing city by $792 and decreased dwelling rent in
the rest of the region by $8. Because Tom doesn't care about people outside his city,
he can ignore the effects of the tax on outsiders. From Tom's perspective, all that
matters is pay about $800 more for dwelling rent.
that his citizens
Table 18-1 shows the distribution of the property tax burden from the local per-
spective (the traditional view). In computing the numbers. Pechman ( 1985) assumes
Table 18-1 Property Tax Incidence, Local

and National Perspectives
Property Tax as Percent of Income
Local Perspective National Perspective

Income ($) 1 partial equilibrium l (general equilibrium)
0-5.000 7.9% 1.0<7r
5.000-10.000 3.0 0.6

10.000-15.000 2.4 0.9
15.000-20.000 2.1 0.9
20.000-25.000 2.1 1.0
25.000-30.000 2.1 1.2
30,000-50,000 2.2 1.4
50.000-100.000 2.3 2.2

100,000-500,000 2.2 3.9
500.000-1 million 5.2
()\cr 1 million 2.3 5.8
loseph A Pechman, Who Paul tin- Taxis. 1966-1985

'
Shi R( i
(Washington, D.C.: Brookings Institution, 1985), Tabic 3-1.

that ( 1 ) the land tax is borne entirely by landowners and (2) the improvement tax
is borne entirely by housing consumers. In other words, the computational results
are based on partial-equilibrium analysis. Since the income elasticity of demand
for housing (or improvements) is slightly less than 1 .0, the improvement tax is re-
gressive: the ratio of tax to income decreases as income increases. In contrast, land
ownership rises rapidly with income, so the land portion of the property tax is pro-
gressive. In combination, the two components of the tax cause the property tax to
be highly regressive for incomes less than $20,000 and roughly proportional for
incomes above $20,000.
The National Effects of a Local Tax

Bertha, the president of a nation of 100 cities, wants to estimate the national effects
of a property tax She should use the general-equilibrium model.
in a single city.
Under improvement tax
the assumption of a fixed national supply of dwellings, the
is paid by capitalists in the form of a lower return on capital (lower net income from
rental property). The tax causes zero-sum changes in the cost of housing: consumers
outside the taxing city gain at the expense of consumers in the taxing city.
Table 18-1 shows the distribution of the property tax burden from the national
perspective (the new view). In computing the numbers. Pechman 1 985 assumes that ( )
( 1 ) the land tax is borne entirely by landowners and (2) the improvement tax is borne
entirely by capitalists (property owners). In other words, the computational results are
based on general-equilibrium analysis. Because capital and land ownership increase
rapidly with income, the property tax is progressive: the tax burden as a fraction of
income increases with income.
What if Bertha assumes that the supply of capital is not fixed? If the supply is
not perfectly inelastic, part of the tax is shifted to housing consumers. The price of
housing increases by a relatively large amount in the taxing city and decreases by a
relatively small amount in the other 99 cities. As a result, the average cost of housing
increases.
The Effects of a National Tax

Sam is the president of a nation in which all 100 cities have the same property tax.
He wants to estimate the effect of the improvement tax. Because all cities impose the
tax, it does not cause any movement of capital between cities. Under the assumption
of a fixed national supply of capital, the tax is borne entirely by capitalists. The tax
does not affect the cost of housing because there is nowhere for the dwellings to flee.
Under the assumption that the supply of capital is not perfectly inelastic, the tax is
split between housing consumers and capitalists, with the consumer share increasing
as the elasticity of supply of capital increases.
Summary of Incidence Results
Table 18-2 summarizes the results of the analysis of the property tax. It shows who
pays the tax under different assumptions about the type of tax (a tax in a small
Table 18-2 Who Pays the Property Tax?
Tax Imposed by a Small City

Welfare effects in the taxing city
1. Immobile households: consumers pay higher dwelling rent.
2. Mobile households: landowners receive lower land rent.
Welfare effects in a nontaxing city

1. Immobile households: consumers pay lower dwelling rent.
2. Mobile households: landowners receive higher land rent.
National welfare effects

1. Property owners (capitalists) receive lower net rental income.
2. Immobile households
a. Fixed supply of dwellings: zero-sum changes in dwelling rent.
b. Variable supply of dwellings: positive-sum changes in dwelling rent.

3. Mobile households
a. Fixed supply of dwellings: zero-sum changes in land rent.
b. Variable supply of dwellings: negative-sum changes in land rent.
Tax Imposed by All Cities (A National Property Tax)

Welfare effects in the taxing cities
1. Fixed supply of dwellings: housing consumers and landowners unaffected by the tax.
2. Variable supply of dwellings: dwelling rent increases and land rent decreases.
National welfare effects

1 Property owners receive lower net rental income.
2. Fixed supply of dwellings: property owners pay the entire tax.
3. Variable supply ot dwellings: tax shared by property owners, consumers, and landowners.
city versus a national tax), household mobility, and the response of property owners
to the tax (fixed versus variable supply of dwellings). It also discusses the welfare
effects for three types of jurisdictions: the taxing city, the nontaxing cities, and the
nation.
The Tax on Commercial and Industrial Property

Because Taxville has only residential property, the property tax is effectively a tax
on residential property. This section discusses the effects of the general property tax
(a tax on residential, commercial, and industrial property) in cities with all three
types of property.
The General Property Tax

The analysis of the land portion of the general property tax is straightforward.
Because the supply of land is perfectly inelastic, the tax is paid entirely by
landowners. The land tax does not affect land rent, but simply decreases the
annual rental income by the annual tax. The decrease in net rental income decreases
the market value of the land.
What about the tax on improvements? Suppose that a city imposes a 2 per-
cent tax on residential, commercial, and industrial structures. In the short run, the
supply of structures is fixed, so the tax decreases the net income of the city's prop-
erty owners. In the long run, property owners move their structures to Notax, the
other city in the region. The movement of capital increases the rent on structures
in the taxing city and decreases it elsewhere. For example, rent might rise from
$4,000 to $4,400 in the taxing city and decrease to $3,600 elsewhere. The changes
in structure rents increase the price of consumer goods in the taxing city and decrease
prices elsewhere. The changes in structure rents also equalize net property income at
$3,600.
Differential Tax Rates
Most jurisdictions impose the same nominal tax rate on residential and business
property. Because assessment ratios differ across property types, however, effective
tax rates often differ. Consider a city with the following characteristics:
1 There are two types of structures: houses and pizza parlors.
2. The city starts with the same effective tax rates on pizza parlors and houses.
3. The total capital in the city (the total number of structures) is fixed.
4. Structures can be used for either houses or pizza parlors.
5. In the initial equilibrium, the structure rent is $4,000.
Suppose that the city cuts the property tax rate on pizza parlors and increases
the tax rate on housing. The change in tax policy is revenue-neutral: the loss of
revenue from pizza parlors equals the gain in tax revenue from residences. The new
tax policy causes the following changes in the city.
1. Transfer of capital. Property owners move capital from the residential

sector to the more lucrative pizza sector.
2. Increased dwelling rent. The movement of capital out of the housing

sector increases dwelling rent.
3. Decreased pizza prices. The movement of capital into the pizza sector
decreases the price of pizza.
Who gains and who loses? The change in tax policy is revenue-neutral, so property
owners are not affected by the change in tax policy. Housing consumers lose because
the price of housing increases, and pizza consumers gain because the price of pizza
decreases. This general-equilibrium analysis suggests that differences in effective
tax rates are reflected in differences in consumer prices. If a city taxes residential
property at a higher rate than business property, its citizens pay more for housing,
but less for other goods.
What if nonresidential goods are consumed by people outside the city? The
taxation of property used in the production of exports increases the export price, so
part of the city's tax burden is borne by consumers outside the city. This is known as
tax exporting: the city uses tax revenue from outsiders to pay for local government.
Suppose that a large fraction of the city's pizza is consumed by outsiders. If the city
wants to decrease the local tax burden, it could cut the residential tax rate and increase
the tax rate on commercial property (including pizza parlors). Although pizza prices
increase, partly offsetting the decrease in the cost of housing, local consumers are
better off because a larger fraction of local taxes is shifted to outsiders.
What are the limits to tax exporting? The taxation of exports decreases the city's
export volume: consumers respond to the increase in price by decreasing the quantity
of exports demanded, so the city's tax base decreases. In addition, export firms have
an incentive to move to cities with lower taxes. Because taxation decreases the city's
tax base, there are limits to tax exporting. Tax exporting is more likely in a city that
has a unique production advantage that makes it the dominant locational choice for
export firms.
The Tiebout Model and the Property Tax

The Tiebout model, introduced in the previous chapter, suggests that citizens shop
for municipalities and school districts. In equilibrium, households sort themselves
with respect to ( 1 ) the demand for public goods and (2) housing consumption. This
section discusses the implications of the Tiebout process for the incidence of the
property tax.
To explain the effects of the property tax in the Tiebout world, consider Metro,
a metropolitan area with the following characteristics:
1. Local public good. Municipalities provide a single public good (parks).
2. Income and house value. Half the households are wealthy and have
$100,000 houses. Half are poor and have $50,000 houses.
3. Park demands. Half the households in each income group want a large
park budget ($2,000), and half want a small park budget ($200).
4. Municipalities and zoning. Households can set up new municipalities,
using large-lot zoning to establish minimum house values.
As shown in Table 18-3, households sort themselves with respect to park demands
and property values, forming four municipalities. There are two wealthy communi-
ties, one with a large park budget (tax rate = 2.0%) and one with a small park budget
(tax rate = 0.20%). Similarly, there are two poor municipalities.

Because the households in Metro sort themselves into homogeneous commu-
nities, the property tax is a user fee, not a conventional tax. The tax liability of a
particular household depends on its park consumption, not on its housing consump-
tion (property value). Every household pays a tax equal to its desired park budget,
regardless of its property value: the poor pay a higher tax rate to generate the same
park budget as their counterparts in the wealthy community; the low spenders pay
a lower tax rate than their counterparts in the high-spending municipalities. In the
Table 18-3 Tiebout Sorting and the Property Tax
Municipality Property Value ($> Park Budget 1$) Tax Rate (%)
Wealthy —Large budget 100,000 2.000 2.0

Wealthy — Small budget 100.000 200 0.20
Poor — Large budget 50.000 2.000 4.0
Poor —Small budget 50.000 200 0.40
Tiebout world, there no link between property value and property tax liability.
is
depends strictly on park consumption.

Instead, the tax liability
In the Tiebout world, an increase in housing consumption does not affect prop-
erty taxes. Suppose that a few of the wealthy, high-spending households decide to
buy $200,000 homes. They will form a new community of high-spending house-
holds with $200,000 houses, and finance their $2,000 park program with a 1 percent
property tax. The doubling of housing consumption cuts the property tax rate in half,
so property taxes remain at $2,000. In general, a change in housing consumption
causes the household to move to a community where other households have ( 1 ) the
same demand for public goods and (2) the same housing consumption.
In the Tiebout world, the discussion of the incidence of the property tax is
meaningless. If households are sorted into homogeneous communities, the property

tax is a user fee, and every household gets exactly what it pays for. The "incidence"
of a $200 tax liability is the same as the "incidence" of $1 paid for a hot dog: you
pay the $1 to get the hot dog; you pay the $200 tax to get $200 worth of parks. Since
the consumer's tax liability is independent of his housing choice, the property tax is
a user fee, not a conventional tax.

How realistic is the Tiebout model? The conditions required for Tiebout equilib-
rium are unlikely to be satisfied. There are simply not enough local governments to
allow households to sort themselves into perfectly homogeneous communities. The
Tiebout model is clearly inapplicable to central cities, where a single municipality
serves a large and diverse population. Therefore, the central-city property tax is not
a user fee, but a conventional tax. In the suburbs, where there is more sorting of
households with respect to demands for public services and housing consumption,
the property tax is closer to a user fee.
Administering the Property Tax

Because the property tax is based on the value of the property, the local government
must periodically assess the value of the property. The annual tax liability of a piece
of property is
Tax = Nominal tax rate •

Assessed value (18-1)
Table 18-4 Residential Property Tax Rates in

Selected Cities, 1995
Effective ta.x Assessment Nominal ta.x

c
City rate(%) ratio (
'(
rate (%)
Bridgeport, CT 4.75 70 6.78

Newark. NJ 3.68 17 21.64
Milwaukee, WI 3.58 98 3.65
Des Moines. IA 2.98 50 5.77
Houston. TX 2.67 100 2.67
Philadelphia. PA 2.64 32 8.26
Jacksonville. FL 2.20 100 2.20
Phoenix. AZ 1.82 10 18.19
Portland. OR 1.80 100 1.80
Chicago. EL 1.79 19 9.44
Memphis, TN 1.53 24 6.34
Boston. MA 1.39 100 1.39
Minneapolis, MN 1.34 92 1.46
New York NY
City. 0.86 8 10.69
Los Angeles, CA 0.74 70 1.06
SOURCE: Statistical Abstract of the United States 1997. Table 497.
The assessed value is
Assessed value = Assessment ratio •

Market value (18-2)
If the assessment ratio is 25 percent, a house with a market value of $120,000 has
an assessed value of $30,000. If the nominal tax rate is 10 percent, the tax liability
is $3,000 (10 percent of the assessed value). The effective tax rate, defined as
the tax liability divided by the market value, is only 2.5 percent ($3,000 divided
by $120,000). The effective tax rate is less than the nominal tax rate because the
assessed value is less than the market value.
Table 8-4 shows assessment ratios, nominal tax
1 rates, and effective tax rates
for several cities. There is substantial variation among all three variables: the range
of effective tax rates is 0.74 percent to 4.75 percent: the range for assessment ratios is
8 percent to 1 00 percent: the range for nominal tax rates is 1 .06 to 2 1 .64 percent. If all
property in a jurisdiction is subject to the same assessment ratio, underassessment
does not affect tax liabilities or effective tax rates, but simply requires a higher
nominal tax rate to generate the same tax revenue. In other words, a local government
can offset a low assessment ratio with a higher nominal tax rate.
most jurisdictions, assessment ratios also vary from one piece of property to
In
another. Some of the variation is caused by random assessment errors, a reflection
o( the imperfections of the assessment process. Another reason for variation in
assessment ratios is systematic assessment errors: that is, the assessment ratio varies
sj stematically with property value. In many cities, the assessment ratio is inversely
related to property value: as property value increases, the assessment ratio decreases.
This practice of regressive assessment has been documented by Engle (1975), Black
(1977), Ihlanfeldt (1982), and Netzer (1980). Under a regressive assessment policy,
the poor, who occupy low-value housing, face higher assessment ratios and thus
higher effective tax rates.
There are two reasons for regressive assessment practices. First, assessors do
not always keep up with changes in property value: unless property is reassessed
as it increases in value, the assessment ratio decreases over time. Since market
values typically rise more rapidly in wealthy neighborhoods, the assessment ratio of
high-income housing decreases relatively fast over time (see Engle, 1975). Second,
assessors sometimes discriminate against multiple-family properties, assessing them
at higher rates than single-family properties. Since a large fraction of the poor live in
multiple-family buildings, the poor face higher effective tax rates. This phenomenon
is discussed by Aaron (1975) and Netzer (1980).
Taxes on Income and Retail Sales

In addition to property taxes, most municipalities have taxes on income and retail
sales. For municipalities as a whole, 27 percent of tax revenue comes from income
taxes, and 13 percent comes from sales taxes. In this part of the chapter, we discuss
the incidence of these two taxes.
Municipal Sales Tax

Who pays the municipal sales tax? According to the theory of tax incidence, the
distribution of the tax burden depends on the elasticities of supply and demand: the
relatively inelastic side of the market pays a relatively large share of the tax.
Figure 1 8 —4 shows the effects of a 25 percent general sales tax. The market starts
at point 5, with an equilibrium price of $8 and an equilibrium quantity of 30 units.
If the tax is paid, in legal terms,by the city's merchants, their costs increase by
25 percent, shifting the supply curve upward. The equilibrium net price (the price
charged by the merchant plus the tax) increases to $9.00, and total output decreases
to 25 units (point C). Merchants charge consumers $9.00 and pay the government
$1.80 in taxes, leaving them with $7.20. In this example, consumers pay about
55 percent of the tax (the $1 increase in price divided by the $1.80 tax).
How does a change in the elasticity of demand affect the distribution of the tax
burden? Suppose that the demand curve in Figure 18-4 becomes more elastic: its
slope decreases and it pivots around point B. The flatter demand curve (shown by the
dashed line) intersects the post-tax supply curve at a lower net price, so consumers
pay a smaller fraction of the tax. A relatively elastic demand means that consumers
can more easily change their behavior, so they are better able to escape the tax.
How does a change in supply elasticity affect the distribution of the tax burden?
For example, suppose the initial supply curve in Figure 8^1 is horizontal (perfectly
1
elastic). The sales tax shifts the supply curve up by the amount of the tax, and the
520 Part PV Local Government
Figure 18-4 Market Effects of a Municipal Sales Tax

Supply after tax
Initial supply
- Relatively elastic
demand
Before the sales tax. B is the equilibrium point. A 25 percent sales tax shifts the supply curve upward
by 25 percent, generating a new equilibrium at point C. The consumer price rises from $8.00 to
$9.00. and the revenue of the merchant (price less tax) drops from $8.00 to $7.20.
post-tax supply curve intersects the demand curve at a price equal to the original
price plus the tax. In other words, if supply is perfectly elastic, the consumer bears
the entire tax. In general, the more elastic the supply curve (the flatter the curve),
the smaller the supplier burden and the larger the consumer burden.
The sales tax also affects the urban land market. As the city's sales volume
decreases, the demands for all inputs (labor, land, capital) decrease: merchants need
fewer laborers, machines, buildings, and land. In the short run. the prices of these
inputs decrease. In the long run. inputs that are mobile (labor and capital) leave the
taxing city, and the prices of the mobile inputs recover. is immobile, In contrast, land
so the tax-induced decrease in the demand
permanent decrease in
for land causes a
the price of land. Therefore, part of the sales tax is borne by landowners. Although
merchants receive a lower net price ($7.20 instead of $8.00). they can still make
zero economic profit (normal accounting profit) because they pay less for immobile
inputs (e.g., land).
This discussion of the city's input markets shows the link between tax policy and
economic development. The sales tax increases the net price of the city's products, so
Chapter IS Local Taxes and Intergovernmental Grants 521
the city produces less output and uses less capital, labor, and land. In other words, the
tax decreases output and employment. The local economy is particularly sensitive
to changes in taxes because firms and households can easily move between jurisdic-
tions. Because of the high degree of intercity mobility, relatively small differences
in tax rates can cause relatively large changes in the level of economic activity.
Municipal Income Tax

As shown in the first chapter on the local public sector, the income tax generates about
one-sixth of municipal tax revenue. It will be useful to compare the municipal income
tax to a national income tax. Figure 18-5 shows the effect of a national income tax.
In the initial equilibrium (point B), thewage is $8 and the quantity of labor is N„.
Suppose government imposes an income tax of $2 per hour, which
that the federal
is paid in legal terms by the employer. The tax shifts the demand curve down by
$2, decreasing the market wage (paid to employees) from $8.00 to $6.10. The net
cost of labor (the wage plus the employer's tax liability) rises from $8.00 to $8.10.
Workers pay almost the entire tax because the supply of labor is relatively inelastic.
Figure 18-6 shows the effect of a municipal income tax. Since laborers can
move easily from one municipality to another, the municipal supply curve is more
elastic than the national supply curve. When the municipality imposes a $2 income
Figure 18-5 Market Effects of a National Income Tax

Supply
Units of labor
The initial equilibrium is point B. A $2 labor tax (paid by the employer) shifts the demand curve down by the
amount of the tax, generating a new equilibrium at point C. The equilibrium wage (paid to laborers) drops
from $8.00 to $6.10. meaning that workers pay about 95 percent of the $2 tax.
Figi re 18-6 Market Effects of Municipal Income Tax

s
8.00
7.00
6.10
Initial demand
Demand after S2 tax
M \1
Units of labor
The initial equilibrium is point D. A S2 labor tax (paid by the employer) shifts the demand cur\e down b\ the
amount of the tax. generating a new equilibrium at point E. The equilibrium wage (paid to laborers) drops
from S8 to S7. meaning that suppliers pay only about half of the S2 tax. Worker-- pa\ a --mailer portion of the
municipal tax because the municipal supply of labor is more elastic than the national >uppl\
tax. the demand curve shifts down by S2. decreasing the market wage from $8 to S7.
Compared to the national income tax. the municipal tax causes a smaller decrease in
the market wage S .00 instead of S .90) because it is easier to escape the municipal
( 1 1
tax. Because intercity migration is less costly than international migration, laborers
bear a smaller fraction of the municipal tax.
The municipal income tax increases the net cost of labor from $8 to S9. In
other words, the demander"s share of the tax is $1. If all firms make zero profits,
who actually pays this part of the tax? The tax is paid by the owners of immobile
resources: it is paid by owners of inputs that cannot be moved to other cities with
the fleeing laborers. As
economic activity in the taxing city decreases
the level of
(the number of laborers decreases from to M M
"). the demand for land decreases,
lowering the price of land. Although firms pay a higher net wage ($9 instead of S8).
the) also pay less for land, so they can still make zero economic profits.
Intergovernmental Grants
As was shown in Table 16-3. intergovernmental grants provide about two-fifths of
local-government revenue and about a quarter of municipal revenue. Fig-
ure 18-7 shows how the intergovernmental revenue for municipalities is divided
Figure 18-7 Shares of Intergovernmental Grants for Different

Programs
General support
22%
Health and hospitals
Housing and community Highways
3%
development 8%
9%
SOURCE: U.S. Bureau of the Census. Compendium of Government Finances, no. 5 (Washington.
among various spending programs. About a fifth of grant money supports the general
operations of local government, and another fifth supports education. The two redis-
tributional programs —public welfare and housing and community development
together receive just over a quarter of municipal grant money.
Why Intergovernmental Grants?
Why don't local governments pay their own way, supporting their spending programs
with local taxes? There are two principal reasons. First, intergovernmental grants
can be used to internalize interjurisdictional spillovers, as discussed in Chapter 16.
Second, local tax instruments are relatively inflexible, causing mismatches between
local tax revenue and local expenditures.
The mismatch problem occurs ifthe desired spending on local public goods
rises faster than the tax base. Figure 18-8 shows the mismatch problem for a city
with the following characteristics:
1 Local public goods are financed with the residential property tax.
2. Income is rising twice as fast as property values.
3. The desired spending on local public goods is proportional to income: the

income elasticity of demand for local public goods is .0. 1
At the start of the decade, the desired spending equals the average property tax
liability. Because income increases at twice the rate of property values, the desired
spending increases twice as fast as che average tax liability. By the end of the decade,
there is$50 gap between desired spending and taxes. The mismatch occurs because
a
(1 ) property values are relatively sluggish and (2) the tax rate is fixed.
There are two possible responses to the mismatch between taxes and desired
spending. The obvious response is to increase the tax rate. The alternative approach
is to transfer surplus federal tax revenue to the local government. The federal income
Figure 18-8 Mismatch between Desired Spending and Tax Revenue
Desired spending
(proportional to income)
Property tax revenue
Years
Ifincome rises twice as fast as property values, the desired spending rises twice as fast as property tax
revenue. The gap between tax revenue and desired spending grows to $50 by the end of the decade. An
increase in the tax rate would close the gap.
tax will generate a surplus if ( 1 ) the income tax has a progressive rate structure and (2
federal spending is a constant fraction of income. In Figure 18-8. a federal transfer
of $50 per capita would solve the local mismatch problem.
Unconditional Grants
An unconditional grant is a lump-sum grant with no strings attached. The local

response to an unconditional grant can be predicted by examining the effects of the
grant on the preferences of the median voter.
Figure 18-9 shows the budget lines and indifference curves of Marian, the
median voter in city G. The city provides a number of local public goods and pays
for them with on public goods there is one dollar of
local taxes: for every dollar spent
taxes, so there is one spend on private goods. The indifference curves,
less dollar to
which show Marian's subjective trade-off between public and private goods, are
convex, reflecting the diminishing marginal rate of substitution between public and
private goods. Given the initial budget line (AB) and her indifference curves. Marian's
utility is maximized at point E. According to the median-voter model described in
the previous chapter, the city chooses the preferred budget of the median voter, so the
city spends $25 per household on public goods, leaving Marian $45 for private goods.
Figure 18-9 Effects of a $20 Unconditional Grant

$
90
C
70 A \.
60
F
45
E
\, Initial budget line
30
/ n. Budget line with $20 grant
15
\B \D
25 30 70 90
Spending on local public goods (3
An unconditional grant of $20 per capita shifts the budget line of the median voter from AB to CD.
The utility-maximizing combination of public and goods moves from point E to point F.
private
Spending on local public goods increases by $5, and spending on private goods increases by $15.
The local government responds to the grant by cutting local taxes, allowing citizens to spend
three-fourths of the grant on private goods.
Suppose that the state gives the city an unconditional grant of $20 per household.
The city makes the following announcement to its citizens: "The state has just given
us $20 per household. Tell us how much to spend on public services, and we'll cut
taxes to distribute the rest of the grant to you." The grant shifts Marian's budget
line upward, from AB to CD, increasing her desired spending on public goods to
$30 (point F). The city, adhering to the preferences of the median voter, increases
its budget by $5 and cuts taxes, thus distributing the rest of the grant ($15) to its
citizens. The tax cut allows Marian to increase private consumption to $60.
One of the predictions of the median-voter model is that an unconditional grant
has the same effect on local spending as an equal increase in per capita income. In
fact, there is conclusive evidence that cities spend a relatively large fraction of an
unconditional grant on public goods, meaning that a grant is not equivalent to an
increase in income. This phenomenon is known as the flypaper effect: the grant
money sticks where it first hits (the local government).
Some of the theories of the flypaper effect assume that city officials either
confuse or deceive their citizens. In a model developed by Filimon, Romer, and
Rosenthal ( 1982). the objective of city bureaucrats is to maximize the city budget.
If bureaucrats do not publicize the grant, citizens are unlikely to demand tax cuts,
so a large fraction of the grant can be used to increase government spending.
Lump-Sum Conditional Grants
The money from a conditional grant, or categorical grant, must be spent on a spe-
cific program. Conditional grants are provided for education, public welfare, health
and hospitals, highways, housing, and community development. There are program-
specific grants within each expenditure group. For example, the $1.94 billion in
education grants that went to local governments in 1992 included specific grants for
remedial reading, school libraries, special education, and other programs.
Suppose that the state gives the city a lump-sum grant of $20 per capita for
special-education programs. Figure 18-10 shows Marian's preferences for special
Figure 18-10 Effects of a $20 Conditional Grant

$
C
90 \
\
\
\
\
\
\
\
70 A "\G
60
iV\~
1
\
\\
1
E
45 1
==U 1
^s Initial budget line
\\ /^^s Budget line with S2(> grant

\
1
t .
1 \^
N
1
1
\ xf
1
20 25 30 70 90
Spending on special education ($)
\ conditional grant ol $20 per capita shifts the budget line of the median voter from H to \C \ />.
The utility-maximizing combination moves from point £ to point / one-fourth of the yrant is :
spent on special education. Hie citj uses the grant to decrease its contribution to special education,
pro> iding more monej to spend on othei goods. Because the desired spending on special education
exceeds the grant, the conditional grant is equivalent to an unconditional grant.
education and all other goods. "Other goods" includes private goods and other local
public goods. The grant shifts Marian's budget line from AB to AGD. Point G is
in the new budget set because Marian could spend all of her own money on other
goods and use the $20 grant to support special education. As spending on special
education increases above $20, there is a dollar-for-dollar trade-off between spending
on special education and other goods, so the slope of the budget line is 1.0.
The grant increases Marian's desired spending on both special education and
other goods. If the city adheres to the preferences of the median voter, spending on
special education increases from $25 to $30 and spending on other goods increases
from $45 to $60. One-quarter of the grant is spent on special education, and three-
fourths is spent on other goods.
The city can spend part of the grant on other goods because it decreases its own
contribution to special education. Before the grant, $25 of local tax money was spent
on special education. After the grant, spending on special education is $30, and the
city can pay for its special-education program with the $20 grant and $10 of local
tax money. The grant frees up $15 worth of local tax money, which can be spent on
other local public goods and private goods.
In Figure 18-10, the conditional grant is equivalent to an unconditional grant.
Because the desired spending exceeds the grant, the community can replace its own
spending on special education with the grant: in effect, the grant has no strings
attached. In graphical terms, the budget lines of the two grants are identical as long
as the desired spending exceeds the grant (as long as the city chooses a point to the
right of G).
The conditional grant is not equivalent to the unconditional grant if the desired
spending on special education is less than $20. Suppose that the city starts with
spending of $5 on special education, and the desired spending increases to $10
under the unconditional grant. In other words, the city moves to a point midway
along the segment CG. In contrast, the city is likely to choose point G under the
conditional grant. Since the city wishes to spend less than $20 on special education,
it cannot substitute the conditional grant for its own spending on special education,
so a conditional grant is not equivalent to an unconditional grant. The unconditional
grant generates a higher utility level for voters in the city. The conditional grant
causes a larger increase in spending on special education.
Matching Grants: Open-Ended

Under a matching grant, the government contributes $m for every $ of local spending 1
on a program. One possibility is a one-for-one matching grant: every $1 of local

spending generates $1 in grant money, so $1 of local spending buys $2 worth of local
public goods. There are two types of matching grants, open-ended and closed-ended.
Figure 18-1 1 shows the effect of a one-for-one matching grant for special ed-
ucation. The grant decreases the slope of Marian's budget line from 1.0 to 0.50:
instead of a dollar-for-dollar trade-off, Marian gets $2 worth of special education for
every $1 she diverts from other goods. Given thenew budget line, Marian's optimum
point moves from point E to point H. If the city adheres to the preferences of the
median voter, spending on special education increases from $25 to $40 and spending
on other goods increases from $45 to $50. Half of the city's $40 special-education
528 Part l\' Local Government
Figure 18-11 Effects of an Open-Ended Matching Grant

$
A
70
50
H
45 \e f*
20 \ /
Budget line with matching
grant (S for $)
\ f Initial budget line
\fi \f
25 i 10 70 140
An open-ended matching grant shifts the budget line of the median voter from AH to AC. The utility-maximizing
combination moves from point /-.
to point H . three-fourths of the grant is spent on special education. The matching grant
decreases the opportunity cost of spending on special education. pro\ iding a greater stimulus to special education than
an equivalent lump-sum grant.
budget comes from the state government, so the local contribution to special educa-
tion is $20. leaving $50 for spending on other goods. In other words, the city spends
one-quarter of the matching grant on other goods.
The matching grant provides a greater stimulus to special education than an
equivalent lump-sum conditional grant. Although the state spends the same amount
on each type of grant ($20). the matching grant increases spending on special
education to $40, while the lump-sum grant increases spending to only $30 (see
Figure 18-10). The matching grant has a relatively large stimulative effect because
it has both an income effect and a substitution effect. Both grants increase real
income by $20 per capita, increasing the demand for all normal goods, including
special education (the income effect). The matching grant also has a substitution
effect: the matching grant cuts the relative price oi special education in half (from
1.0 to 0.50), causing the substitution of special education for other goods. Since
thematching grant has a substitution effect as well as an income effect, it increases
spending by a larger amount.
Chapter IS Local Taxes and Intergovernmental Grants 529
The information in Figure 18-1 could be used to plot two points on the city's
1
demand curve for special education. Before the matching grant, the "price" of special
education is $1: there is a dollar- for-dollar trade-off between special education and
other goods. At a price of $1, the quantity of special education demanded is $25
The matching grant decreases the price to $0.50, and increases the quantity
(point E).
demanded to $40 (point H). Since the grant does not affect the other determinants
of demand (income and other prices), points E and H identify two points on the
demand curve: ($1, $25) and ($0.50, $40).
Matching Grants: Closed-Ended

Under a closed-ended matching grant, the government specifies a maximum grant.
In Figure 18-12, the maximum grant is $30: the government provides matching
Figure 18-12 Effects of a Closed-Ended Matching Grant
Budget line with closed-ended

matching grant ($ for $ up to $30)
40 60 70 90 100 140
A closed-ended matching grant shifts the budget line of the median voter from AB to ADI The
.
utility-maximizing combination moves from point E to point H Since the desired spending level ($40) is less
.
than the limit ($60), the grant limit is irrelevant, and the closed-ended grant is equivalent to the open-ended grant.
If desired spending exceeded the limit, the city would spend less under the closed-ended grant than under the
open-ended grant.
funds up to a total of $30. The budget line has a kink at the point at which the grant
limit sets in. Given a one-for-one matching when spending
grant, the kink occurs
on special education (the sum of city spending and grant money) reaches $60. For
spending above $60 (below point D). every additional dollar spent on special educa-
tion decreases local spending on other goods by $1, so the slope of the budget line
is 1.0.
In Figure 18-12, the city's response to the closed-ended grant program is the
same as its response to the open-ended program. Since the desired spending level
($40), is less than the limit, the grant limit is irrelevant. Therefore, the closed-ended
grant is equivalent to the open-ended grant.
The two grants are not equivalent if the desired spending under the open-ended
grant exceeds $60. In Figure 18-12, the open-ended budget line is the line ADC.
and the closed-ended budget line is ADI. If the desired spending under the open-
ended grant exceeds $60, the city will choose a point along the segment DC. In this
region, the open-ended budget line (segment DC) lies above the closed-ended grant
(segment DI). Spending is higher under the open-ended grant for two reasons. First,
real income is higher under the open-ended grant, producing an income effect that
increases spending on all normal goods, including special education. Second, the
opportunity cost of special education is lower under the open-ended grant, producing
a substitution effect that increases spending on special education.
Block Grants
Under a block grant program, dozens of conditional grants are consolidated into a
single general grant. For example, all grants for education (special education, reme-
dial reading, kindergartens, libraries, and so on) could be consolidated into a single
block grant for education. The block grant is a compromise between unconditional
and conditional grants. Although the city is forced to spend the entire education
block grant on education, it can choose how to allocate themoney between special
education, remedial reading, kindergartens, and libraries. This greater flexibility is
welcomed by local officials, but is often opposed by state and federal officials, who
lose some control over grant money.
Summary: The Stimulative Effects of Grants
As explained governments use intergovernmental grants to increase

earlier, local
local expenditures and to cut local taxes. The tax cuts allow local citizens to spend
part of the grant on private goods. The empirical question is: what fraction of grant
money is used for additional local spending, and how much is used to cut taxes?
Inman 1979) summarizes the results of several studies of the local response to
(
giants. He concludes that local-government spending increases by about 30 cents

per dollar of grant money, leaving 70 cents per dollar for tax relief.
Welfare Reform: Matching Grants to Block Grants

A key component of the welfare-reform plan adopted in 1 996 is the replacement of
federal matching grants with block grants. Under the old system, each state picked a
level of welfare spending (subject to a minimum level set by the federal government)
and the federal government used matching grants to issue "rebates" to states, the
size of which increased with the state's welfare spending. For poor states, the federal
rebate per dollar spent on welfare was $0.78, so from the state's perspective, each
dollar spent on welfare cost the state only $0.22. In other words, the price per dollar
of welfare spending was only $0.22. For rich states, the rebate per dollar was $0.50,
so the price of welfare spending was $0.50. Under the new block-grant system, the
federal grant no longer depends on how much the state spends on welfare. There
are no matching funds or rebates for state welfare spending, so the price of a dollar
spent on welfare is $ 1 .00.
Figure 18-13 shows the effects of welfare reform on the budget choices of a
poor state. The original budget line (shown by points A, m, and E) is relatively
Figure 18-13 A Switch from a Matching Grant to a Block Grant

Decreases Welfare Spending
Initial budget line:
matching grant
140 170 210
Spending on welfare programs ($ million)
A switch from a matching grant to a block grant that makes the initial point (m) affordable decreases
spending on welfare programs. Because the initial point is affordable, there is no income effect, and the
substitution effect (substitution away from the good whose price increases) decreases welfare spending
from $210 million to $170 million (point b).
flat, low price of welfare spending under a matching grant. The state's
reflecting the
original choice shown as point m (for "matching"): the state spends $210 million
is
on welfare and $260 million on other goods. The new block grant is $140 million,
so the new budget line is shown by points A, G, and D: there is a kink in the new
budget line at point G. Given the new dollar-for-dollar trade-off between welfare
spending and other spending, the block grant is large enough that the state has the
option of picking point m. If the original point is possible, does that mean that the
state will choose it?
Under the block grant, the state will actually spend less on welfare programs. To
maximize utility of the median voter, the slope of the indifference curve will be equal
to the slope of the budget line (the marginal rate of substitution will equal the price
ratio). For the original choice (point //; ). the slope of the indifference curve equals
the slope of the original budget line (—0.22). Under the block grant, the slope of the
budget line becomes —1.0, so the indifference curve is now flatter than the budget
line, and point m

no longer maximizes utility. The new utility-maximizing point.
b (for "block") is the point at which an indifference curve is tangent to the new.
steeper budget line. In this case, when the price of welfare spending increases, the
state's spending on welfare programs decreases by $40 million, from $210 million
to $170 million. Consistent with the new dollar-for-dollar trade-off between the two
goods, spending on other goods increases by $40, from $260 to $300.
We can explain the cut in welfare spending using the concepts of income and
substitution effects. Normally, when the price of a particular good increases, there
are two effects:
1 . Substitution effect. The relative price of the good increases, and the
consumer substitutes other goods for the good that has become relatively
more expensive.
2. Income effect. The price hike means that the consumer has less purchasing
power or real income, reflected in the fact that the consumer can no longer
afford the original bundle. A decrease in real income could either increase
or decrease consumption.
from a matching grant to a block grant that is large enough to

In the case of a switch
make is no income effect. Because there is only
the original point affordable, there
a substitution effect —
which always decreases the consumption of a good that has
—
become more expensive we can be sure that spending on welfare programs will
drop.
By how much will the state's welfare spending decrease? For a poor state, the
price hike from $0.22 to $1 .00 will decrease welfare spending by between 40 percent
and 66 percent (Inman and Rubinfeld. 997 ). For a rich state, the price hike is smaller
1
($0.50 to $1.00). and the switch to block grants will decrease welfare spending by
between percent and 18 percent. Congress was apparently aware that welfare
1
reform would cause states to cut their welfare spending: the law requires states to
continue to spend at least SO percent of the amount spent under the old matching-grant
policy.
Capitalization and the Distributional Effects of Grants

Intergovernmental grants are often used to promote income redistribution. Grants
are targeted to two types of communities: those with a large number of low-income
households and those with relatively high tax rates.
Grants to Low-Income Communities
Consider a state with two cities, a wealthy one and a poor one. In the poor city, all
households rent housing for $400 per month. In an attempt to redistribute income
from the rich to the poor, the state imposes a tax on the wealthy city and uses the tax
revenue to support a grant to the poor city of $100 per household. Suppose that the
poor city responds to the grant by cutting its income tax by $100 per household.
Figure 18-14 shows the effect of the grant program on the housing market in
the poor city. The decrease in local taxes increases the relative attractiveness of the
city, causing migration to the city that shifts thedemand curve for housing upward.
The equilibrium from $400 to $435. Because the grant
price of housing increases
increases the price of housing, the net benefit per household is only $65 ($100 less
the $35 increase in rent).
The intergovernmental grant is a blunt instrument for income redistribution for
two reasons. First, the grant is tied to a location, not to a poor household. The grant
Figure 18-14 Grant to Poor City and the Price of Housing
Supply
Demand with grant
Initial demand
60 70
Quant y of dwellings
A grant to a poor community increases the relative attractiveness of the community, increasing the
demand for housing and housing prices. The grant is capitalized into housing prices, so some of
the benefits of the grant accrue to property owners.
increases the market value of land and housing in the poor city, generating benefits
for property owners. In other words, the grant is capitalized into housing and land
prices. Second, although part of the grant goes to households in the recipient city,
all households in the city benefit, not just poor households.
Grants to High-Tax Communities

Under some grant programs, the size of the grant depends on "local tax effort." A
city with relatively high tax rates receives a relatively large grant. Is this equitable?
Suppose that there are two cities in a region. North and South. North has harsh
winters, so the city has high taxes to support large expenditures on highway mainte-
nance. In contrast. South has mild winters and a relatively low tax rate. In equilibrium,
people must be indifferent between living in the two cities, so the differences in tax
rates are capitalized into property values. The residents of North pay high taxes, but
are able to buy inexpensive land and housing. Similarly, the low taxes in South are
offset by relatively high land and housing prices.
Suppose that highway grants are based on local tax effort, so North receives
a larger highway grant. Is the grant policy equitable? Before the highway grant,
households were indifferent between the two cities, with differences in tax rates
capitalized into housing prices. Because citizens in the two cities achieve the same
utility level before the grant, a larger grant to North does not necessarily promote
"fairness." On the contrary, the relatively large grant to North increases property
values in the city, generating capital gains for property owners.
Tax Expenditures: Implicit Grants

The federal government uses its tax code to influence local government spending.
Two provisions of the federal income tax code affect local spending: the deductibility
of local taxes and the exemption of interest from municipal bonds.
Deductibility of Local Taxes
Federal taxpayers can deduct local and state taxes on income and property from their
taxable income. Consider a city with the following characteristics:
1 Marian, the median voter in the city, itemizes her deductions on her federal
tax form.
2. Marian has a marginal tax rate of 28 percent.

3. All local taxes are deductible.
Figure 18-15 shows Marian's spending choices. In the absence of the deduction
for local taxes, the opportunity cost of $ 1 of public spending is $ 1 of private spending,
so the budget line AB. The deductibility of local taxes decreases the opportunity
is
cost of local spending to 72 cents on the dollar: every dollar of local taxes decreases
Figure 18-15 Deductibility of Local Taxes and Local

Spending
Initial budget line
Spending on local public goods ($)
The deduction for local taxes decreases the opportunity cost of local spending, increasing
the desired spending of the median voter. The deduction is equivalent to an open-ended
matching grant.
Marian's federal tax liability by 28 cents, so she gives up only 72 cents worth of
private goods to consume $1 worth of local public goods. The deduction shifts the
budget line from AB to AC, shifting Marian's preferred point from E to F. If the city
adheres to the preferences of its median voter, it chooses point F.
The deduction for local taxes is equivalent to an open-ended matching grant. This
can be seen by comparing Figures 18-15 and 18-11. The deduction is an example
of a tax expenditure: the federal government stimulates local spending indirectly,
providing tax cuts with the expectation that part of the tax cut will be spent on local
public goods. Instead of giving the locality an intergovernmental grant of $20 per
capita, the federal government gives local citizens a $20 tax cut. In other words,
the federal government uses local taxpayers as intermediaries in its subsidization of
local spending.
The stimulative effect of the federal tax deduction varies across cities. If the
median voter in a middle-income city has a marginal tax rate of 15 percent, the city's
federal subsidy is 15 cents per dollar of local spending. If the median-income voter in
a higher-income city has a marginal tax rate of 28 percent, the city's federal subsidy
is 28 cents per dollar. If the median voter in a low-income city does not itemize his
deductions, the city's federal subsidy is zero.
Exemption of Interest on Local-Government Bonds

Because the interest earned on local-government bonds is not subject to the federal
income tax, local governments can pay bondholders a relatively low interest rate.
Consider Bill, whose marginal tax rate is 28 percent. Suppose that the market interest
rate for private bonds is 12 percent. Since Bill declares his interest earnings on
private bonds as taxable income, every $ 1 of interest income increases his federal
tax liability by 28 cents. Therefore, the net return on private bonds is about 9 percent,
not 12 percent. If the interest on local-government bonds is not taxable, the net
return on a government bond is equal to its gross return. The local government
can compete with private bonds by setting the interest rate on government bonds at
about 9 percent. The tax exemption allows local governments to finance their capital
projects at a fraction of the cost of private projects, so it stimulates local spending.
The exemption is another example of a tax expenditure: the federal government
stimulates local spending by giving up tax revenue, not by providing money directly
to local governments.
Summary
The property tax is a tax on land and improvements (capital). Since the supply
of land is fixed, the land tax is borne by landowners. There are two
perspectives on the incidence of the improvement tax.
a. The view takes a partial-equilibrium approach, focusing on the
traditional
effect of theimprovement tax on the taxing jurisdiction. The supply of
dwellings is perfectly elastic, so the tax is paid by housing consumers.
b. The new view takes a general-equilibrium approach, focusing on the effect
of the tax on people inside and outside the taxing jurisdiction. Property
owners throughout the region bear the tax, and consumers outside the
taxing city gain at the expense of consumers in the taxing city.
The incidence of the improvement tax depends on one's perspective.
a.A mayor should use the partial-equilibrium model (the traditional view) to
estimate the effects of the city's property tax. From the local perspective,
the tax is paid by local housing consumers.
b. The president of a nation should use the general-equilibrium model to
estimate the national effect of one city's tax. Because the tax causes
zero-sum changes in housing rent, the president can ignore the effects on
housing consumers. The tax is paid by the nation's capitalists.
c. If all cities in the nation have the same property tax rate, the tax is paid by
capitalists. There are no housing-price effects because there is nowhere for
capitalists to move their capital.
The simple model of the property tax assumes that the national supply of
capital is fixed. If the supply of capital is responsive to changes in net returns,
capitalists shift part of the tax onto housing consumers in the form of higher
housing prices.
Chapter IH Local Taxes and Intergovernmental Grants 537
4. The simple model assumes that households are immobile. If they can leave the
taxing jurisdiction, they pass the tax onto landowners.
5. The analysis of the general property tax (on residential, commercial, and
industrial property) is the same as the analysis of the residential property tax.
a. According to the general-equilibrium model, the tax decreases the return on
capital; consumer prices increase in the taxing city and decrease elsewhere.
b. If a city imposes a higher effective tax rate on residential property,
consumers pay more for housing, but less for other goods (e.g., pizza,
clothes).
6. In the Tiebout world, a household's property tax bill is independent of its
housing consumption: households sort themselves into homogeneous

communities, so a household's tax bill depends only on its consumption of
public goods.
7. Assessment practices cause differences between nominal and effective
property tax rates.
a. Because of underassessment, the nominal property tax rate exceeds the
effective tax rate.
b. Some cities systematically overassess low-value property and underassess
high-value property, so the poor face a relatively high effective tax rate.
8. Taxes on sales and income are paid in part by landowners.

a. The municipal sales tax is paid by consumers, who pay higher prices, and
landowners, who receive less rent.

b. The municipal income tax is paid by laborers, who receive lower wages,
and landowners. Because labor is more mobile between municipalities than
between nations, labor's share of a municipal income tax is smaller than its
share of a national income tax.
9. The states provide the bulk of grants to local governments (86 percent of all
grants). Grants are used to internalize externalities and solve the mismatch
problem (desired spending rising faster than tax revenue).
10. The local response to an intergovernmental grantcan be predicted by

examining the effects of the grant on the budget choices of the median voter.
a. Local governments respond to grants by cutting local taxes, so part of the
grant money is spent on private goods.

b. If citizens don't know about grants, government spending may increase by
a relatively large amount (the flypaper effect).

c. A conditional grant is equivalent to an unconditional grant if the desired
spending exceeds the grant: the local government can replace its own
spending with the grant money, so the conditional grant is the same as a
cash payment.
d. A matching grant has a relatively large stimulative effect because it has a
substitution effect and an income effect.
1 1 Under a block grant program, dozens of conditional grants are consolidated
into a single general grant.
538 PartlY Local Government
12. The welfare-reform plan adopted in 1996 replaces matching grants with block
grants. Thischange in policy increases the net cost or price of state welfare
spending, and will decrease state spending on welfare programs.
13. Intergovernmental grants are capitalized into property values, so part of a grant
to a low-income community ends up in the pockets of property owners.
Similarly, grants to high-taxcommunities generate benefits for property
owners.
14. The federal government uses tax expenditures to stimulate local spending.
Taxpayers act as intermediaries in the stimulation of local spending.
a. The deduction for local taxes decreases the opportunity cost of government
spending.
b. The exemption of interest on local bonds decreases the cost of government
borrowing.

1 Consider the new view of the property tax. Based on the Taxville example,
compute on the following
the long-run effects of Taxville's property tax
individuals. How much
do they gain or lose?
a. Ms. Partial rents a house in Taxville.
b. Mr. Wash is a traveling salesman who rents one dwelling in Taxville and
one in Notax.
c. Ms. Capitalist owns five dwellings in Taxville.
cl. Mr. Capitalist owns five dwellings in Notax.
2. In the Taxville example, a 2 percent property tax decreased net rental income
by $400 in each city. Suppose that the region has four cities instead of just two.
a. By how much will the property tax increase dwelling rent in the taxing city?
b. By how much will the tax decrease net rental income?
3. Choose the word in parentheses that makes the following statement correct,
and then explain your choice: As the moving cost of households decreases, the
tax-induced increase in Taxville's dwelling rent (increases, decreases).
4. Mr. Javelin is the economic forecaster tor the city of Drip. Oregon. You are his
research assistant and do all of his computations. The cit) is considering a tax
of $2 per raincoat. In estimating the revenue from the raincoat tax. you make
the following assumptions:
I. The supph of raincoats to the city is perfectly elastic.

//'. The current price of raincoats is $20. and the current quantity o\
raincoats demanded is 1.000 per week.

Hi. The price elasticity of demand for raincoats is -2.0.
a. Estimate the tax revenue from the raincoat tax.
b. When Mr. Javelin sees your computations, lie says: "Those revenue
projections look pretty low. You know, the state imposed a S2 raincoat tax
several years ago and it raised S3 ). 000 per week. Since about
1
5 percent of
Chapter IH Local Taxes and Intergovernmental Grants 539
the state's raincoats are bought in Drip, I would expect that we would raise
about 5 percent of the revenue raised by the state tax. If your revenue
projection is less than $1,950, it's wrong." Is Mr. Javelin correct? If so,
what is your mistake? If not, what is wrong with his analysis?
5. The city of Tinytown is one of 10 municipalities in a large metropolitan area.

Compare the incidence of a Tinytown income tax to the incidence of a
metropolitan income tax.
6. Suppose Collegetown eliminates its tax on pizza parlors and increases its tax
on residential property. Under what conditions will the change in tax policy
decrease the tax liability of Joe College?
7. Consider a school district with a total budget of $500,000. The district is
currently spending $250,000 on special-education programs for disadvantaged

students (SE programs). There are three types of federal grants for SE
programs:
Grant A: Unconditional grant of $100,000 to the school district.
Grant B: Conditional grant (money to be spent on SE programs) of $100,000.
To ensure that the categorical grant is spent on SE, the federal
government gives 100,000 marked dollar bills to the school district.
On each dollar bill is written: "This dollar must be spent on special

education."
Grant C: Conditional matching grant (dollar for dollar) with a maximum
subsidy of $100,000. For each $1 spent by the school district on SE,
the federal government gives it $1.
a. Use diagrams to illustrate the district's responses to the three types of grants.
b. How would the district's responses differ if the pre-grant expenditures on
SE were only $50,000?
School District Type of Grant Educational Spending
E None $50
F $1 grant for $1 local 70
G $0.33 grant for $1 local 60
8. The state of Grantland has experimented with different types of matching

grants for education. The state picked three identical school districts and used
three different matching grants. The results are shown in the table above.
a. Map the demand curve for education — different combinations of price
(opportunity cost) and quantity (educational spending).
b. Using the data for school districts E and F, compute the price elasticity of
demand for education.
9. If the flypaper effect occurs, will the deductibility of local taxes still be
equivalent to a matching grant? Which policy will generate a larger increase in
local-government spending?
10. The discussion of the mismatch between local tax revenue and desired
spending assumes that the desired spending on local public goods is
proportional to income: the income elasticity of demand for public spending is
1.0. Another implicit assumption is that the unit cost of local public goods is
constant. Draw the spending and tax lines for the following two cases:
a. The income elasticity of demand for local spending is 2.0. Does the larger
income elasticity generate a larger or a smaller gap between desired

spending and tax revenue?
b. The income elasticity of demand for local spending is 1 .0, but the unit cost
of local public goods increases by 25 percent over the course of the decade.
Does rising cost generate a larger or a smaller gap between desired
spending and tax revenue?
1 1. Consider the distributional effects of intergovernmental grants. In the example
shown in Figure 18-14, the $100 grant to the poor city decreased the city's
income taxes, generating benefits for both tenants (the poor) and landowners.
Suppose that instead of cutting taxes, the city increases spending on schools,
police, fire protection, and parks. Will the benefits of the grants still be shared
by tenants and landowners?

Revenue Facts
U.S. Bureau of the Census. Census of Governments. Washington. D.C: Department of
Commerce. 1987.
— — Government Finances
. in 1984-85. Washington, D.C: Department of Commerce,
1987.
Property Tax Incidence
Aaron, Henry J. Who Pays the Property Tax'.' A New View. Washington. D.C: Brookings
Institution. 1975. Discusses the traditional and new views of the property lax.
Describes the administrative problems of the property tax. including assessment errors.
Aronson, J. Richard, and John L. Hilley. Financing State and Local Governments.
Washington, D.C: Brookings Institution, 1986.
Buskin. Michael. "Taxation. Saving, and the Rate of Interest." Journal of Political Economy
86 (1978). pp. S3-S28. Estimates that the supply of savings is responsive to the rate of
between 0.20 and 0.60.
return, with an elasticity
Freeman. Richard B. "How Do
Public Sector Waste and Employment Respond to
Economic Conditions?" Working Paper no. 1653. National Bureau of Economic
Research. Cambridge. Mass.. 1985.
Mieszkowski. Peter. "The Property Tax: An Excise Tax or a Profits Tax?" Journal of Public
Economics 1972). pp. 73-96. Discusses the new view of the property tax.
1 (
Pechman. Joseph A. Who Paid the Taxes. 1 966 - 985 Washington. D.C: Brookings
.''
Institution, 1985. Estimates the incidence of the property tax under the traditional view
and the new view.
Property Tax: Administration and Reform

Aaron, Henry J. Who Pays the Property Tax? A New View. Washington, D.C.: Brookings
Institution, 1975. Chapter 4 discusses the administrative problems of the property tax,
including assessment errors. Chapters 5 and 6 discuss reform proposals.

Washington, D.C.: Brookings Institution, 1986. Chapter 7 discusses some of the
assessment problems associated with the property tax.
Black, David E. "'Property Tax Incidence: The Excise-Tax Effect and Assessment
Practices." National Tax Journal 30 (1977), pp. 429-34. Discusses the effects of
regressive assessment practices on the incidence of the property tax.
Engle, Robert F. "De Facto Discrimination in Residential Assessments: Boston." National
Tax Journal 28 (1975), pp. 445-51. Shows that assessment ratios are inversely related
to market values.
Ihlanfeldt, Keith. "PropertyTax Incidence on Owner-Occupied Housing: Evidence from the
Annual Housing Survey." National Tax Journal 35 (1982), pp. 89-97. A study of
regressive assessment in Philadelphia and Atlanta.
Netzer, Richard. Real Property Tax Policy for New York City. New York: New York
University, 1980. Shows that assessment ratios are relatively low for one-family and
two-family houses.
Peterson, George E. Property Tax Reform. Washington, D.C.: Urban Institute, 1973.
Income and Sales Taxes

Washington. D.C.: Brookings Institution, 1986. Chapter 8 discusses the local sales tax,
the income tax, and user fees.
Intergovernmental Grants: Reasons, Types, and Effects
Washington, D.C.: Brookings Institution, 1986. Chapter 3 discusses federal
intergovernmental grants, and Chapter 4 discusses state grants.
Courant, Paul N.; Edward M. Gramlich: and Daniel Rubinfeld. "The Stimulative Effect of
Intergovernmental Grants, or Why Money Sticks Where It Hits." In Fiscal Federalism
and Grants -in -Aid, ed. Peter Mieszkowski and William Oakland. Washington, D.C.:
Urban Institute, 1979. Discusses the flypaper effect.
Filimon, R.; T Romer; and H. Rosenthal. "Asymmetric Information and Agenda Control:
The Bases of Monopoly Power and Public Spending." Journal of Public Economics
17(1982), pp. 51-70.
Gramlich, E. M. "Intergovernmental Grants: A Review of the Empirical Literature." In The
Political Economy of Fiscal Federalism, ed. Wallace E. Oates. Lexington, Mass.:
Lexington Books, 1977. Reviews the literature on the effects of grants on local
spending. Concludes that matching grants are more stimulative than lump-sum grants
and that the categorical grant has a larger stimulative effect than an equivalent increase
in private income.
Inman, Robert P. "The Fiscal Performance of Local Government: An Interpretive Review."
In Current Issues inUrban Economics, ed. Peter Mieszkowski and Mahlon
Straszheim. Baltimore: The Johns Hopkins University Press, 1979, pp. 270-321.
Inman. Robert P.. and Daniel L. Rubinfeld. "Rethinking Federalism." Journal of Economic
Perspectives 1 1:4 (Fall 1997), pp. 43-65.
Ornstein, Norman J. "Chipping Away at the Old Blocks." The Brookings Bulletin 18 (1982),
pp. 1 1-15. Discusses the reasons for categorical and block grants.
Tax Expenditures
Ladd, Helen F. "Federal Aid to State and Local Governments." In Federal Budget Policy in
the 1980s, ed. Gregory V. Mills and John L. Palmer. Washington, D.C.: Urban Institute.
1984. Discusses the role of tax expenditures in intergovernmental finance.

PartV
u BAN TRANSPORTATION
a ne of the advantages of an urban location

ties
is its proximity to the many
within a metropolitan area. This part of the book examines the two main
components of the urban transportation system. Chapter 19 describes the automo-
activi-
bile/highway system, focusing on three problems caused by automobiles: conges-

tion, pollution, and highway accidents. Chapter 20 deals with urban mass transit,
focusing on the choice of a commuting mode by a commuter (e.g., automobile, bus,
or light rail) and the choice of a transportation system by a city planner. The chapter
explains why so few commuters in U.S. cities use mass transit and why light-rail
and heavy-rail transit systems are usually less efficient than bus systems.
Chapter 19
HIGHWAYS
^€ UTO S a n d
The home is where part of the family waits until the others are through with
the car.
Herbert Prochnow
C^7~ his is the first of two chapters on urban transportation. It discusses the most
«_x popular travel mode, the automobile, which is used by almost 90 percent of
U.S. commuters. It examines three transportation problems caused by the automo-
bile: congestion, air pollution, and highway accidents. Congestion during rush hours
is inevitable, and a certain level of congestion is actually efficient. Just as it would
be inefficient to eliminate all air pollution, it would be inefficient to eliminate all
congestion. The question is whether congestion is at the optimum level. If not, there
are a number of policies that could decrease congestion, including various taxes on
auto travel, subsidies for mass transit, and highway construction. The second prob-
lem, air pollution, is controlled by the federal government through its auto emissions
policies. The question is whether other pollution-control policies such as pollution
taxes or gas taxes might be more efficient in controlling pollution. The third trans-
portation problem is auto safety. Traffic accidents injure and kill people, and also
The policy question is:
disrupt traffic flow, contributing to the congestion problem.
how do government highway death rates and accident rates?
policies affect
Figure 19-1 shows the modal choices of commuters in U.S. metropolitan areas.
About 77 percent of central-city residents commute by automobile (car, truck, or
van) with about four of five of these commuters driving alone, and the rest riding
in carpools. Suburban residents are more numerous, and a larger fraction of them
(93 percent) commute by automobile. As shown in Table 10-3 (in Chapter 10),
about 40 percent of metropolitan workers commute from one suburb to another. Over
97 percent of these commuters use automobiles. In contrast, among the 33 percent
of metropolitan workers who commute within the central city, only 72 percent use
automobiles.
545
546 Part V Urban Transportation
Figure 19-1 Modal Choice of Central-City and Suburban Commuters
50
c
1 40 -
—
E 30
E
o
to
Number
Central-eity residents Suburban residents
D Walk 1.96 1.45
D Bus or rail 3.93 1.24

D Carpool 4.72 7.07
D Drive alone 23.11 43.73
Source: U.S. Bureau of the Census. 7990 Census of Population and Housing, SSTF20. Journey to Work in the
United States, (Washington. D.C.: U.S. Government Printing Office. 1994).
Congestion: Equilibrium versus Optimum Traffic Volume

Most cities suffer from traffic congestion during the morning and evening rush
hours. In recent years, congestion has become worse in most metropolitan areas in
the United States, Western Europe, and Japan (Giuliano and Small, 1998). During
the 1970s and 1980s, vehicle traffic per capita increased at a rate of about 2.5 percent
per year in the United States. Highway capacity hasn't kept pace with rising traffic
volume: the ratio of traffic volume to road capacity increased in 47 of the 50 largest
metropolitan areas, with an average increase of 16 percent. The result is more con-
gestion. For example, on theSan Francisco-Oakland Bay Bridge, the aggregate time
loss resultingfrom congestion delays more than doubled between 1984 and 1991.
The annual cost of congestion (for extra time and fuel consumption) in the 50 largest
metropolitan areas is about $35 billion (Small. 1997).
We'll use a simple model to explain the congestion phenomenon and evaluate
some alternative public policies to deal with it. Consider a city with the following
characteristics:
1 Radial highway. There is a two-lane highway from the suburbs to the city
center (a distance of 10 miles).
2. Monetary travel cost. The monetary cost of auto travel is 20 cents per mile.
3. Time cost. The opportunity cost o\~ travel time is 10 cents per minute.
Chapter 19 Amos and Highways 547
The total cost of a commuting trip is the sum of the monetary and time costs of the
10-mile trip. The monetary cost is $2 10 miles times 20 cents per mile). The time
(
cost depends on travel time: a trip that takes 30 minutes has a time cost of $3 (30
minutes times 10 cents per minute); a 20-minute trip has a time cost of only $2.
Therefore, the total cost of a 30-minute trip is $5 ($2 plus $3), and the total cost of
a 20-minute trip is $4.
The Demand for Urban Travel

Consider first the demand side of the urban travel market. Figure 19-2 shows the de-
mand curve for travel along the radial highway. The horizontal axis measures the
number of vehicles per lane per hour. The vertical axis measures the cost of the com-
muting trip, the sum of the monetary and time costs of the 10-mile trip. The demand
curve shows, for each trip cost, how many travelers use the highway. For example,
if the trip cost is $12.80, there are 1,200 people for whom the benefit of the trip
exceeds the cost, so the traffic volume is 1 ,200 vehicles per lane per hour. As the
cost of the trip decreases, there are more people for whom the benefit exceeds the
Figure 19-2 Congestion Externalities and the Congestion Tax
Social trip
cost
*- 7.13
marginal benefit
1.000 1,200 ,600 1,800
Vehicles per lane per hour
Drivers use the highway as long as the marginal benefit (shown by the demand curve (exceeds the private trip cost, so the
equilibrium traffic volume is .600 vehicles per lane per hour. At the optimum, the marginal benefit equals the marginal
1
social cost (the social trip cost). The equilibrium volume (1,600) exceeds the optimum volume ( 1.400) because drivers
ignore the external costs of their trips. A congestion tax of $4.34 internalizes the congestion externality, generating the
optimum traffic volume.
cost, so the city moves downward along the travel demand curve: there are 1.400
vehicles at a cost of $9. 14, and 1 .600 vehicles at a cost of $5.48.
The demand curve is a marginal-benefit curve. For each traffic volume, it shows
how much the marginal traveler is willing to pay for the highway trip. Suppose that
the city starts with a trip cost of $9.15 and a traffic volume of 1.399. If the trip cost
decreases to $9. 14. traffic volume increases to 1 meaning that the .400th driver
.400. 1
is make the trip; at

willing to pay $9. 14 to any cost above $9.14. the trip would not
be worthwhile. The demand curve shows that the marginal benefit of the 10-mile
trip decreases from $12.80 for the 1.200th driver to $1.82 for the 1.800th driver.
The Private and Social Costs of Travel

Table 19-1 shows the relationships between traffic volume and travel time. Column
B lists the trip time (the travel time per driver) for different traffic volumes. For traffic
volumes up to 400 vehicles per lane per hour, there is no congestion: everyone travels
at the legal speed limit of 50 miles per hour, and the 10-mile trip takes 12 minutes.
For traffic volume above 400 vehicles, the computations are based on the following
formula:
Trip time = 12.0 + 0.001 •(Volume-400) + 0.000015-(Volume-400)

2
(19-1)
When the 40 1 st driver enters the highway, the congestion threshold is crossed. As the
highway becomes crowded, the space between vehicles decreases and drivers slow
down to maintain safe distances between cars. As more and more drivers enter the
highway, travel speeds decrease and travel times increase: the trip takes 1 2.8 minutes
if there are 600 vehicles, 22.4 minutes for 1,200 vehicles, and 52.0 minutes for
2.000 vehicles.
Table 19- -1 Traffic Volume, Travel Time, Travel Cost, and Congestion Externalities
A B C D E F G H
Increase in Increase m
Traffic Travel Time Total Travel External Private Social Marginal
Volume Trip Tune per Driver Time Trip Cost Trip Cost Trip Cost Benefit
{vehicles) (minutes 1 1 minutes i I minutes i
($) ($) <$) (demand)
200 12.0 3.20 3.20 31.10

400 12.0 3.20 3.20 27.44
600 12.8 0.007 4.2 0.42 3.28 3.70 23.78
800 14.8 0.013 10.4 1.04 3.48 4.52 20.12
1,000 18.0 0.014 19.0 1.90 3.80 5.70 16.46
1,200 22.4 0.025 30.0 3.00 4.24 7.24 12.80
1,400 28.0 0.031 43.4 4.34 4. SO 9.14 9.14
1,600 J4.8 0.037 59.2 5.92 5.48 11.40 5.48
1,800 42.8 0.043 77.4 7.74 6.2X 14.02 1.S2
:.ooo 52.0 0.049 98.0 9.80 7.20 17.00
Chapter 19 Autos and Highways 549
Columns C and D show on the travel times

the effects of the marginal driver
of other drivers. For low volumes (below 400 vehicles), an additional driver
traffic
does not affect speeds or travel times. For volumes above 400 vehicles, however,
an additional driver slows traffic and increases travel times. For example, the 600th
driver increases the travel time per driver by 0.007 minutes (column C): when the
driver enters the highway, the trip time increases from 12.793 to 12.80. The increase
in total travel time is simply the extra time per driver (0.007) times the number of
other drivers (599), or 4.2 minutes (column D). This is the congestion externality:
the marginal driver slows traffic and increases travel time, forcing other drivers
to spend more time on the road. The congestion externality increases with traffic
volume: the 1 ,200th driver increases travel time by 30 minutes (0.025 minutes times
1,199), and the 1,600th driver increases travel time by 59.2 minutes (0.037 times
1,599).
The external trip cost equals the monetary value of the congestion externality.
The figures in column E are based on the assumption that the opportunity cost of
travel time is 10 cents per minute. For the 600th driver, the external trip cost equals
the increase in total travel time (4.2 minutes) times 10 cents per minute, or 42 cents.
The external trip cost increases with traffic volume, rising to $3.00 for the 1,200th
driver, $5.92 for the 1,600th driver, and $9.80 for the 2,000th driver.
Columns F and G show the private and social costs of travel. The private trip
cost is commuter, defined as the sum of
the travel cost incurred by the individual
the monetary cost ($2.00) and the private time cost. To compute the private time
cost, multiply the trip time by the opportunity cost of travel time: the opportunity
cost is 10 cents per minute, so the private time cost is $1.20 for a volume of 200
vehicles, $1.28 for 600 vehicles, and so on. Therefore, the private trip cost is $3.20
for 200 vehicles, $3.28 for 600 vehicles, and so on. The social trip cost is the sum
of the private trip cost (column F) and the external trip cost (column E). Figure 19-2
shows the cost curves for private trip cost and social trip cost. The social-cost curve
lies above the private-cost curve, with the gap between the two curves equal to the
external trip cost.

There are some alternative labels for private and social trip costs. An alternative
is average travel cost, defined as total travel cost divided
label for private trip cost
by the number of drivers. Since each driver travels at the same speed and thus has the
same travel cost, the average travel cost equals the private trip cost. An alternative
label for social trip cost marginal travel cost, defined as the increase in the total
is
cost of travel resulting from adding one more traveler. Since the social trip cost
equals the trip cost incurred by the marginal driver plus the external cost he imposes
on other drivers, the social trip cost is the same as the marginal travel cost.
Equilibrium versus Optimum Traffic Volume
What is the equilibrium number of drivers? A driver uses the highway if the marginal
benefit of a trip (from the demand curve) exceeds the private trip cost. In Figure 1 9-2,
the demand curve intersects the private-trip-cost curve at 1,600, so the equilibrium
number of vehicles is ,600 and the equilibrium private trip cost is $5.48. The ,60 st
1 1 1
driver does not use the highway because the marginal benefit of using the highway
is less than the trip cost.
What is the optimum number of drivers? The basic efficiency rule is that an ac-
tivity should be increased as long as the marginal social benefit exceeds the marginal
social cost; at the optimum level, the marginal benefit equals the marginal cost. In
Figure 19-2. the marginal social benefit is shown by the demand curve, and the
marginal social cost is shown by the social-trip-cost curve. The demand curve inter-
sects the social-trip-cost curve at a volume of 1,400. so the optimum traffic volume
is 1,400 vehicles. For the first 1.400 drivers, the social benefit of travel exceeds the
social cost, so their use of the highway is efficient. For the 1,401st driver, the social
cost exceeds the benefit, so her use of the highway is inefficient.
The equilibrium volume exceeds the optimum volume because drivers ignore
the costs they impose on other drivers. An additional driver slows traffic, forcing
other drivers to spend more time on the road. Suppose that Carla. the 1.599th driver,
has a private benefit of $5.49. From column Fin Table 19-1. the private trip cost for
1,599 vehicles is about $5.48, so Carla uses the highway. Her use of the highway is
inefficient because the benefit of the trip ($5.49) is less than the social cost of the trip
($1 1.40, equal to the sum of Carla's private cost of $5.48 and the external trip cost
of $5.92). Because Carla ignores the $5.92 external cost, she makes an inefficient
choice.
The Policy Response: Congestion Tax

The government could use a congestion tax to generate the optimum traffic volume.
A tax equal to the external trip cost would internalize the congestion externality,
generating the optimum number of drivers. In Figure 19-2. a congestion tax of
$4.34 per would shift the private-trip-cost curve upward by $4.34. decreasing
trip
the equilibrium number of drivers from 1.600 to 1.400. The 1.401st driver would
not use the road because the benefit of the trip ($9.13) would be less than the full
cost of the trip ($9.14, the sum of the $4.80 private cost and the $4.34 congestion
tax).Because the congestion tax closes the gap between private and social costs, the
individual driver bases his travel decision on the social cost of travel. Therefore, the
highway is used efficiently.
Benefits and Costs of Congestion Taxes

From the perspective of the individual traveler, the imposition of congestion taxes
generates good news and bad news. People who continue to use the highway after the
tax imposed pay the tax. but also have a lower time cost: the tax decreases traffic
is
volume, which decreases travel times. People who stop using the highway avoid the
tax. but forgo the benefits associatedwith using the highway. In other words, there
are costsand benefits for both types of people. Do the benefits outweigh the costs?
A key consideration in the evaluation of the congestion tax is the disposition of
the revenue it generates. The government does not throw the tax revenue away, but
presumably uses it to finance public services or to decrease other local taxes. In fact,
as explained later in the chapter, the total revenue from the congestion tax is just
enough pay for the optimum highway. Therefore, the government could substitute
to
congestion taxes for the gasoline tax, which is currently used to finance highways.
To simplify government redistributes the tax revenue, in
matters, suppose that the
equal shares, to all of the households who use
the highway before the congestion tax
is imposed. In Figure 19-2, the pre-tax traffic volume is 1,600, so there are 1,600
highway users. The total tax revenue is $6,076 (1,400 vehicles at $4.34 per vehicle),
so the transfer per household is $3.79 ($6,076/1,600).
Table 19-2 shows the costs and benefits for two travelers. Helen has a relatively
large willingness to pay for highway use, so she uses the highway even after the
congestion tax is imposed. In Figure 19-2, her position on the demand curve is
shown by point H. Louis has a relatively small willingness to pay, so he stops using
the highway after the tax is imposed (point L in Figure 19-2). What are the benefits
and costs for the two travelers?
1 Helen pays a tax of $4.34 and receives a refund of $3.79, for a 55-cent net
increase in taxes. The congestion tax decreases the private trip cost from
$5.48 (at a volume of 1,600) to $4.80 (at a volume of 1,400). for a savings
of 68 cents. The net benefit is 13 cents: a benefit of 68 cents less the tax
increase of 55 cents.
2. Louis doesn't use the highway after the congestion tax is imposed. Before
the tax, his benefit highway equaled the gap between his
from using the
willingness to pay (from the demand curve) and the private trip cost.
This gap is his consumer surplus from using the highway. From
Figure 19-2, Louis was willing to pay $7.13 for a trip with a private trip
cost of $5.48, so his consumer surplus was $1.65 ($7.13 — $5.48). After the
congestion tax, Louis does not use the highway, so he loses this surplus.
Because the tax also provides a transfer payment of $3.79, Louis is better
off after the congestion tax by $2.14: the transfer payment more than offsets
his loss of consumer surplus.
Table 19-2 Benefits and Costs of Congestion Taxes
Congestion Tax Net Time Consumer Net

Tax Refund Transfer Savings Surplus Lost Benefit
Helen (H) $4.34 $3.79 -$0.55 $0.68 +$0.13

Louis (L) 3.79 + 3.79 $1.65 + 2.14
Definitions
1. Total tax revenue = Tax per driver x Volume: 1,400 x $4.34 = $6,076.
2. Transfer payment = Total tax revenue H- Number of citizens: $6,076/1.600 = $3.79.
3. Net transfer = Refund — Congestion tax (if any).
4. Consumer surplus lost = Willingness to pay (from demand curve) — Private trip cost
(before congestion tax).
In this example, both citizens benefit from the imposition of congestion taxes.
A more rigorous analysis of the benefits and costs would show that some people
in the city would be harmed by the congestion tax policy: for some travelers, the
savings in travel costs and the transfer payment would not be large enough to offset
the congestion tax or the loss in consumer surplus.
A more rigorous analysis would also show that the winners' benefits exceed the
losers' costs. In other words, the move from the equilibrium to the optimum generates
a net gain for society. In Figure 19-2, the net gain is shown by the shaded area, the
area between the demand curve (the marginal-benefit curve) and the social-trip-cost
curve. To explain this, suppose the city starts with the equilibrium volume (1,600
vehicles) and somehow persuades the 1,600th driver to not use the road. What are
the benefits and costs of diverting the driver? On the benefit side, total travel cost
decreases by the social trip cost for the 1.600th driver (about $11.40 at point J).
On the cost side, the driver loses the benefits of the highway trip: the willingness
to pay for the trip is shown by the demand curve (about $5.48 at point B). The net
benefit from diverting the driver is the difference between the social trip cost and
the willingness to pay, or $5.92. The social gain from diverting the 1,599th driver
is slightly lower than this because the social trip cost is lower (the city starts lower
on the cost curve) and the willingness to pay is higher (the city starts further up the
demand curve). The net gain to society from moving from the equilibrium (point
B) to the optimum (point C) is the sum of the differences between the social trip
cost and the willingness to pay (from the demand curve) for the 1,600th through
the 1,401st drivers. In other words, it is the area between the two curves, shown by
the area of triangle CJB. Because there is a net gain from the move to the optimum
point, the government could, in principle, redistribute income from the winners to
the losers to ensure that everyone is made better off by the congestion tax.
Peak versus Off-Peak Travel

To be efficient, the congestion tax must vary across time and space. The tax should be
higher on congested highways. The most congested highways are the ones leading
to and from employment centers. Most congestion occurs during the morning and
evening rush hours. According to Straszheim (1979). about a quarter of all trips are
made during the rush hours. McConnell-Fay ( 1 986) reports that 64 percent of the trips
from homes to workplaces San Francisco Bay area occur between 6:30 and
in the
8:30 A.M.. and 57 percent of the workplace-home trips occur between 4:30 and 6:30
P.M. Figure 19-3 shows the demand curves and congestion tolls for the peak period
and the off-peak period. Given the high volume of traffic during the peak period (V^,
compared to V„ during the off-peak period), the peak-period congestion toll is higher.
Estimates of Congestion Taxes
What is the optimum congestion tax? In a study of the San Francisco Bay area,
Keeler and Small (1977) estimate congestion taxes for different locations and times.
Pozdena ( 1 988) updated the Keeler and Small estimates, and he estimates that during
(
'hapter 19 Autos and Highways 553
Figure 19-3 Congestion Tax during Peak and Off-Peak Periods
Peak demand
Traffic volume
During the off-peak period, (he demand for (ravel is relatively low. generating a low traffic volume
(l'„, compared to V during (he peak period), so the optimum congestion tax is relatively low.
,
f
the peak travel periods, the congestion tax would be 65 cents per mile on central
urban highways, 21 cents per mile on suburban highways, and 17 cents per mile on
fringe highways. During the off-peak periods, the taxes would be between 3 and 5
cents per mile atall locations. Because the Bay area is more congested than most
metropolitan areas, the optimum congestion taxes would be lower in most other
metropolitan areas.
Suppose the Bay area implemented such a system of congestion tolls. How much
would different types of commuters pay for travel during the peak periods? Consider
first a suburban commuter who travels seven miles on suburban highways and three
miles on central urban highways to a job in the central city. The commuter's one-way
congestion bill would be $3.42 (seven miles times 21 cents plus three miles times
65 cents), so the daily congestion tax would be $6.84. Consider next a suburban
commuter who travels eight miles to a job in another suburb. The commuter's one-
way congestion tax would be $1.68 (eight miles times 21 cents per mile).
Implementing the Congestion Tax

How would the government collect the congestion tax? One possibility is to install
toll booths on every road and collect the tax directly from drivers. This option is
impractical because the collection process slows traffic, causing more congestion.
The high-technology version of toll booths is a vehicle-identification system (VIS).
Under a VIS. every car is equipped with a transponder an electronic device that —
allows sensors along the road to identify the car as it passes. The system records the
number of times a vehicle uses the congested highway and sends a congestion bill
to the driver at the end of the month. For example, a driver who travels 10 miles
along a congested highway 20 times per month would pay a monthly congestion bill
of $86.80 (20 times $4.34 as shown in Figure 19-2). The alternative approach is to
install a device in each car that decreases the value of a cash card or a debit card:
when the car passes a checkpoint, the value of the card (inserted into the device in
the car) is reduced by an amount equal to the toll.
Experiences with Congestion Pricing
Singapore was the first city to use prices to control the volume of traffic. Under the
Area Licensing System (ALS) implemented in 1975. drivers were charged about
$2 per day to travel in a restricted zone in the central area of the city. Initially. ALS
applied only to the morning rush hour, but later was extended for the entire day. The
system decreased traffic volume by about 44 percent and increased travel speeds.
In 1998. Singapore's ALS was replaced with Electronic Road Pricing (ERP).
Under ERP. drivers pay fees for passing different points en route to the central
business district. The fees vary by location and time of day. with the highest fees
for the most congested areas during rush hours. The collection system uses an in-
vehicle electronic device that accepts a stored- value cash card (purchased at local
banks). The in-vehicle device deducts the appropriate fee from the cash card every
time the vehicle passes through an ERP gantry (a structure spanning the roads at
each collection point). Vehicles that do not have an in-vehicle unit or sufficient
value on their cash cards are photographed by gantry cameras for what Singapore
officials call "subsequent enforcement action." The fee for the private cars entering
the central district between 8:30 A.M. and 9:00 A.M. is $2. compared to $1 for the
period 9:30 A.M. to 3:00 P.M. Motorcycles pay half as much as private cars, and taxis
pay one-third as much.
In Toronto, the users of the 407 Express Toll Road pay tolls that depend on
the distance traveled and time of day. The per-kilometer toll is 10 cents during rush
hours ( 10 cents Canadian, equal to 7 cents U.S. ). 7 cents during other weekday times,
and 4 cents on the weekend. The system uses a trip-toll method that, with the help
of a transponder in each car. determines when and where a vehicle enters and exits
the tollway; a central system matches entry and exit registrations and computes the
vehicle's toll for the trip. The toll road is open to occasional users without in-vehicle
transponders: the system photographs license plates and sends bills to the registered
owners.
One approach that is gaining popularity involves changing the rules for high-
—
occupancy vehicle (HOV) lanes. An HOV lane sometimes known as a "diamond"
lane or an "express" lane —
is a lane designated for use by high-occupancy vehicles,
typically defined as a vehicle with at least three passengers. A HOT lane is a lane
that can be used both by high-occupancy vehicles and other vehicles that pay a toll
(HOT stands for "high occupancy and toll").

The first HOT project was along Riverside Freeway (California State Route
91), which connects employment centers in Los Angeles and Orange Counties with
rapidly growing areas to the east. Two HOV lanes that had been added in the median
strip of the freeway were switched to HOT lanes. The toll varies by time of day, with
the highest toll ($2.75 per trip) between 5 A.M. and 9 A.M. on weekdays. Each car
has a transponder in its windshield for identification purposes and a corresponding
account that is billed each time the vehicle uses the HOT lane. Changing to a HOT
lane increased traffic volume on the old HOV lanes, with about 80 percent of users
paying the toll. The change also decreased traffic volume and increased speeds along
the regular lanes on Route 9 generating benefits for commuters who did not pay
1 ,
the toll.
Another HOT project is along Interstate 15 in San Diego. The reversible facility
consists of an 8-mile stretch of two lanes in the freeway median, which are accessible
only at the endpoints of the facility. A fee is charged for each trip, with the toll varying
in "real time" from $0.50 to $4.00, depending on the level of congestion. Each vehicle
has a transponder and a pre-paid account that is billed each time the vehicle uses the
HOT lane. The toll is highest between 7 A.M. and 8 A.M., and 4 P.M. and 5 P.M.
These recent experiences with congestion pricing are promising. Travelers re-
spond to higher prices by changing their travel behavior in ways that decrease traffic
volume and improve the efficiency of travel. The most frequent responses: ( form- 1 )
ing carpools, (2) switching to mass transit, (3) switching to off-peak travel, (4) pick-
ing alternative routes, and (5) combining two or more trips into a single trip (Small
and Gomez-Ibanez, 1998). In many other cities in the United States, Europe, and
Asia, policymakers are evaluating the merits of congestion pricing.
Alternatives to a Congestion Tax: Taxes on Auto Use

A number of alternative congestion policies have been proposed. One set of policies
discourages auto use by increasing the cost of auto travel. Three of the pricing options
are gasoline taxes, parking taxes, and congestion-zone taxes. How effective are these
policies compared to a congestion tax?
To set the stage for a discussion of the efficacy of the alternative policies, it will
be useful to list the four ways that the congestion tax decreases traffic volume:
1 Modal substitution. The tax increases the cost of auto travel relative to
carpooling and mass transit (buses, subways, light rail), causing some
travelers to switch to these other travel modes.
2. Time of travel. The tax is highest during the peak travel periods, causing
some travelers to travel at different times. Because work and study
schedules are relatively inflexible, commuters and students would be less
likely to change their travel times than other travelers (e.g., shoppers).
Nonetheless, firms would have a greater incentive to change work
schedules to allow their workers to avoid costly travel during the peak
period.The institution of "flex-time** and the rearrangement of shift times
would cause some workers to change their travel times.
3. Travel route. The congestion tax is highest on the most congested routes,
causing some travelers to switch to alternative routes.
4. Location choices. The congestion tax increases the unit cost of travel
(travel cost per mile), causing some commuters to decrease their
commuting distances. Some workers may move closer to their jobs, and
others may switch to jobs closer to their residences.
These four changes cause the city to move up the travel-demand curve as the cost
of travel increases. In Figure 19-2. the congestion tax decreases traffic volume from
1.600 to 1.400 because it changes travel modes, times, routes, and distances.
Gasoline Tax
One alternative to the congestion tax is a gasoline tax. The simple idea is that if
travel is more expensive, traffic volume decreases. The problem is that the gas tax
increases the costs of all automobile travel, not just travel along congested routes
during peak periods. In contrast with the congestion tax. which changes travel times
and routes, the gas tax does not encourage drivers to switch to other travel times or
routes.
Suppose the government wants to use the gasoline tax to decrease the peak-
period traffic volume to its optimum level. What is the required gasoline tax? To
have the same effect as the congestion tax in Figure 19-2. the gas tax must increase
the cost of travel by 43.4 cents per mile ($4.34 per 10-mile trip). If the typical car
gets 22 miles per gallon of gasoline (the average mileage in 1992). the required gas
tax is $9.55 per gallon (43.4 cents times 22): if gas mileage is only 10 miles per
gallon, the tax is $4.34 per gallon. Even if a gas tax at this level were feasible, it
would be inefficient because it also increases the cost of off-peak travel by 43.4 cents
per mile.
Parking Tax
A number of cities use parking taxes to discourage driving to central business district
jobs. In an experiment in Madison. Wisconsin, a tax surcharge of $1 was imposed
on drivers who arrived at parking garages during the peak travel period (7 A.M. to
c
) A.M. and left their cars for more than three hours. As explained by Parody
) 984 ( 1 ).
the surcharge decreased traffic volume because ) some commuters sw itched to car-
( 1
some changed their travel times. In Washington. D.C..

pools and mass transit and (2)
the government increased parking costs for its employees, causing some workers to
switch to carpools and mass transit (Miller and Everett. 1982). In Ottawa. Canada.
the government increased parking rates for government employees from zero to 70
percent of the commercial rate, causing a 23 percent decrease in the number
( 1 )
of workers dining to work, (2) a 6 percent increase in the auto occupancy rate
(from 1.33 to 1.41), and (3) a 16 percent increase in bus ridership (DiRenzo, Cima,
and Barber, 1981 ). In a series of experiments in Los Angeles, the elimination of free
parking decreased the number of solo drivers by between 18 percent and 83 per-
cent. When one Los Angeles firm increased its parking fee from zero to $28.75 per
month, the number of solo drivers dropped by 44 percent; when the firm increased
the monthly fee to $57.50, solo driving decreased 81 percent (Small, 1992).
There are three potential problems with using parking taxes to decrease con-
gestion. The first is that the taxes must be imposed only on peak-period commuters:
drivers who travel during the off-peak period should be exempt from the tax. As
shown by the Madison experiment, this problem can be solved by imposing a sur-
charge for drivers who arrive at parking garages during the peak travel period. Sec-
ond, in contrast with the congestion tax, which increases the unit cost of travel and
decreases travel distances, the parking tax does not depend on the distance traveled.
Therefore, there is less incentive for commuters to economize on travel cost by living
closer to their workplaces. Third, because much of the congestion problem iscaused
by cars do not park in congested areas, the tax does not force all peak-period
that
travelers to pay for the congestion they cause.
Shoup (1994) suggests that the subsidization of parking by employers con-
tributes to the congestion problem. In 1990, about 95 percent of American com-
muters who drove to work benefited from free parking at their place of work. In Los
Angeles, the average subsidy from free parking provided by employers is $3.87 per
day. Shoup estimatesthat employer-paid parking shifts 25 percent of all commuters
into solo driving and increases the number of cars driven to work by 19 percent. One
possible response to this problem is to "cash out" the parking subsidy. An employer
that provides free parking worth, say, $150 per month, would give its workers the
option of free parking or $150 in cash. When faced with the cash option, many
workers would take the cash and make other arrangements for getting to work, thus
decreasing traffic volume.
Alternatives to Auto Taxes

The congestion policies discussed so far decrease congestion by increasing the cost
of driving. There are several alternative approaches. One option is to increase the
carrying capacity of the highway, and another is to subsidize transit ridership.
Capacity Expansion and Traffic Design
One response to the congestion problem is to widen the highway to increase its
carrying capacity. Figure 9 —4 shows the effects of such a policy on travel times and
1
traffic volume. The wider road reaches the congestion threshold at a higher traffic
volume and has a lower private trip cost at all volumes above the original congestion
threshold(V = 400). The decrease in trip costs increases traffic volume as the city
moves down the demand curve from point D to E.
Figure 19—4 Effects of Widening the Highway

Private trip cost on
narrow highway
Private trip cost on

wide highway
Traffic volume
The widening of the highway shifts the private trip cost to the right: the congestion threshold increases,
and trip cost is lower at every traffic volume above the original threshold volume. The decrease in trip cost
increases traffic volume from V„ to V'„
. The benefit of the widening is the increase in consumer surplus
(the shaded area).
Under what conditions wouldthe widening of the highway be efficient? One

way measure the total benefit is with consumer surplus. As explained earlier,
to
consumer surplus equals the excess of the willingness to pay (from the demand
curve) over the actual cost of travel. Widening the highway decreases the trip costs
of all drivers, so it increases consumer surplus. The increase in consumer surplus is
shown as the shaded area in Figure 1 9 —4. For the original drivers (Volume = V„), the
consumer surplus per driver increases by the savings in private trip cost (C„ — C„ .).
The increase in consumer surplus is the rectangular area with a width of (C„ — C w )
and a length of V The decrease in trip cost also increases traffic volume from V„ to
.
Vw The consumer surplus of the new drivers equals the area of the triangle DEF.
.
The benefit of widening the highway exceeds the cost if the increase in consumer
surplus (the sum of the rectangle and the triangle) exceeds the cost of widening the
highway. This analysis ignores pollution cost, which is discussed later in the chapter.
There are many anecdotes about increases in highway capacity that did not
result in less congestion and faster travel during the rush hour. The reason is that
the demand for rush-hour travel is highly elastic, with many commuters initially
deterred from using the congested highway because of slow travel speeds. This is
the phenomenon of "latent demand." In the language of Small (1993). there is a
"reserve army of the unfulfilled" that will switch to a previously congested highway
as soon as an increase in capacity increases travel speeds. This latent demand will
fill most or all of the new capacity during peak periods. Many of the new users have
switched from other routes or other times, so we must look beyond the rush hour
along the widened highway to see the effects of the increase in capacity.
The city could also improve the flow of traffic on the existing highway. Street
lights can be synchronized to keep traffic flowing at a steady speed. Designating
one-way and restricting on-street parking increase the carrying capacity of
streets
streets. Some cities have placed stoplights on expressway on-ramps, thus smoothing
the flow of vehicles onto expressways. In the language of transportation engineers.

the ramp lights decrease the "turbulence" caused by entering vehicles, thus increasing
travel speeds. In Los Angeles, the installation of such a system on the Harbor Freeway
increased average travel speeds from 5 miles per hour to 40 miles per hour. A similar
1
system in Dallas increased travel speeds from 4 miles per hour to 30 miles per hour.
1
Subsidies for Transit
Another alternative to the taxation of auto travel is the subsidization of mass transit
(buses, subways, commuter trains, light rail). Transit and autos are substitute travel
modes, so a decrease in the cost of transit causes some consumers to switch from
autos to transit. In other words, a transit subsidy decreases the auto volume, narrowing
the gap between the equilibrium and optimum traffic volumes.
Figure 19-5 shows the interactions between the auto and transit markets. Under
a system of optimum congestion taxes, the auto market reaches point 5, with traffic
volume of A* (where the demand curve intersects the social-trip-cost curve), and
the transit market reaches point Q, with ridership of T* (where the demand curve
intersects the marginal-social-cost curve). In the presence of congestion taxes, both
markets are at their optimum points, defined as the points at which the marginal
social benefit (from the demand curve) equals the marginal social cost.
What if there are no congestion taxes? The auto market reaches point C, where
the demand curve intersects the private-trip-cost curve, generating an equilibrium
traffic volume of A'. The underpricing of auto demand curve for
travel shifts the
transit (a substitute shown by point R, with
good) downward. The equilibrium is
ridership of T If auto travel is underpriced, the auto volume exceeds the optimum
'.
volume and transit ridership is less than the optimum ridership.

A small transit subsidy moves both markets closer to their optimum points.
In the transit market, the subsidy decreases the cost of transit, causing movement
down the demand curve
to point 5: ridership increases to T", closing about two-
thirds of the gap between the equilibrium and the optimum volume. In the auto
market, the decrease in the price of transit shifts the auto-demand curve downward,
moving the market from point C to point D. The equilibrium volume decreases from
A' to A", closing about one-quarter of the gap between the equilibrium and the
—
optimum volume. The subsidy narrows but does not eliminate the gaps between —
the optimum and equilibrium volumes: A* is still less than A", and T" is still less
than T*.
- = >> - - ^
£ * 12 — J :
C y:
•J .C
3 6 ff
"a
=
—- us
j=
5, >
S X 5 — s
_ = —
u 3
r: 2.
>c 2
3 c
1
•Sis
Eh
/ S g 8
/
1
o | . g>SS
c ^ S<S.
a £ 5 A / =
—~ —
ft _S i Oj —s
Is 3 * ^ / i / s -
S 3 •^j/ "3
u —
— > z
5 ill
5
p " ac / i
a 2^
*- r t *-|£
o
a ° N i
E
u i
Q i
S
—
B
r: i
1 o"?
jsoo dux
Z —
c_
>»
u u
3 O > Si
> Eh
3 =
—O
s.
c_ 2- y - -
— _ =
~ - B.2-
'u *-•
u P a' —u -
-
= c Q -=
<Sj N 00 3. B
S
c
C 5 %5 ~
U A--
B..2 3
^ * 1 r -;
aiV
S
i)
T3
u
t
3
03
= CJ /
H CJ>
3 fl" ®
ir,
c
I
lso3 dpx
560
Would it be possible to eliminate both the auto gap and the transit gap? In
Figure 19-5, the transit gap is closing faster than the auto gap: the subsidy closes
two-thirds of the transit gap, but only one-quarter of the auto gap. In other words,
the subsidy increases transit ridership by more than it decreases auto volume. This
occurs because in addition to diverting some drivers from cars to transit, the subsidy
also attracts new travelers to transit. For example, suppose that the subsidy causes
250 drivers to switch from autos to transit (A' — A" — 250) and also generates 280
new travelers. The total increase in transit ridership is 530 (T" — T= 530), the
sum of the diverted auto drivers and the new transit riders. Because the transit gap
is closing faster than the auto gap, if the city decides to eliminate the transit gap,
the city will be left with an auto gap. Alternatively, if it decides to eliminate the
auto gap, the city will "overshoot" the transit gap: transit ridership will exceed the
optimum ridership. The city can close one but not both gaps. — —
The fundamental problem of the transit subsidy is that it underprices transit,
increasing ridership above its optimum level. In the absence of congestion taxes,
a transit subsidy may improve the efficiency of the transportation system, but the
subsidy will never be as efficient as a system of congestion taxes. The policy question
is whether the benefit of the transit subsidy (the diversion of drivers from underpriced
and congested roads) is larger or smaller than the cost (excessive transit ridership).
Highway Pricing and Traffic Volume in the Long Run

This section shows how the government could pick both the optimum traffic vol-
ume and the optimum road width. The long-run analysis of highway pricing and
investment has four steps:
1 Derive the average total-cost curves for different highway widths.

2. Derive the long-run cost curves (average and marginal).
3. Pick the optimum traffic volume and road width.
4. Pick the congestion tax that generates the optimum traffic volume on the
optimum road.
Average Total Cost for Two-Lane and Four-Lane Highways
The average total cost of travel is the sum of the trip cost and the average road
cost. Figure 19-6 shows the trip-cost curves for both the two-lane and the four-lane
highways: the cost curves are horizontal up to the point at which congestion sets in;
the congestion threshold is V2 vehicles per hour for the two-lane highway and V4
vehicles for the four-lane highway. The average road cost equals the total cost of
building the highway divided by the number of trips.
The average total-cost curves are U-shaped. An increase in traffic volume has
two effects on average total cost. First, an increase in volume spreads construction
cost over more drivers, so the average road cost decreases (the road-cost effect).
Figure 19-6 Average Cost for Two- and Four-Lane Highways
Long-run average
cost (LRAC)
Total traffic volume (vehicles per hour)
The average of travel is the sum of private trip cost and average road cost. The average total-cost curves
total cost
are U-shaped. If there are constant returns to scale in road building, the two average total-cost curves reach the
same minimum cost C", but the four-lane highway reaches this cost with twice the traffic volume ( V
instead of
I"). The long-run average cost is composed of the minimum points of the average total-cosl curves.
Second, an increase volume may increase congestion, increasing travel times and
in
At traffic volumes below the congestion threshold

the trip cost (the trip-cost effect).
(VS for the two-lane highway and V4 for the four-lane highway), there is no trip-cost
effect, so the average total-cost curve is negatively sloped. At high traffic volumes,
the average road cost is relatively low and congestion is severe, so the trip-cost effect
dominates the road-cost effect. Therefore, the average total-cost curve is positively
sloped. In Figure 1 9-6, the average total-cost curve for the two-lane highway reaches
its minimum point at a volume of V" drivers, indicating that the road-cost effect
dominates for volumes up to V

drivers and the trip-cost effect dominates for higher
volumes. For the four-lane highway, the trip-cost effect dominates for volumes above
V* drivers.
The cost curves in Figure 19-6 are drawn under the assumption of constant
returns to scale in highway building. If there are constant returns to scale, a doubling
of construction cost (to build four lanes instead of two) doubles the carrying capacity
of the highway. The congestion threshold of the four-lane highway ( V4) is double
that of the two-lane highway (VS). Similarly, it takes twice as many drivers on the
four-lane road to reach a given private trip cost: to reach the private trip cost C",
the four-lane highway requires twice the traffic volume. The two total-cost curves
reach the same minimum cost (C*), but the four-lane highway reaches this point
with twice the traffic volume (V* instead of V').
Long-Run Average Cost and Marginal-Cost Curves

The long-run average cost curve (LRAC) shows the minimum average total cost for
each traffic volume. The curve is drawn by answering the following question: for a
given traffic volume, what is the average total cost of the most efficient highway?
Figure 19-6 shows two points on LRAC: for a volume of V" drivers, the two-
lane highway minimizes average cost at an average cost of C*; for a volume of
V* drivers, the four-lane highway minimizes average cost at the same cost. The
LRAC curve is made up of the minimum points of the average total-cost curves for
different highway widths, so it indicates the average total cost under the assumption
that the government builds the optimum highway for each traffic volume. Given
the assumption of constant returns to scale, LRAC is horizontal, with an average
total cost (the sum of average road cost and average trip cost) of C*. The long-
run marginal-cost curve (LRMC) indicates the marginal social cost of an additional
driver, including the cost of widening the road to accommodate the new driver. In
Figure 19-6, LRAC is horizontal (a result of constant returns to scale), so LRMC is

the same as LRAC (if the average is constant, the marginal equals the average).
Optimum Volume and Road Width

For the optimum traffic volume, the marginal benefit of travel equals the long-run
marginal cost of travel. In Figure 19-7, the demand curve shows the marginal benefit
of travel, and the LRMC shows the long-run marginal cost of travel. The two curves
intersect at V*, so the optimum traffic volume is V* drivers. From Figure 19-6, the
four-lane highway generates the minimum average total cost for a volume of V*
drivers, so the optimum road width is four lanes.
How does the city persuade the optimum number of drivers to use the highway?
From the demand curve, a total of V* drivers will use the road if the cost of the trip
is C*. Drivers will face a cost of C* only if the city imposes a congestion tax. If
the city builds a four-lane road but does notimpose a congestion tax, there will be
V drivers (volumewhich the demand curve intersects the private-trip-cost curve)
at
instead of V*. If the government adds a congestion tax (C* — C), the traffic volume
will be V* (volume at which the demand curve intersects the social-trip-cost curve).
Congestion Tolls Pay for the Optimum Road

If there are constant returns to road building, the revenue from the congestion tax
covers the cost of building the road. In Figure 19-7, the congestion tax equals the
gap between the social trip cost and the private trip cost(C* — C). Because the
long-run average cost is the sum of the private trip cost and the average road cost,
the gap between the LRAC and the private trip cost is the average road cost (C* — C).
Figure 19-7 Optimum Road Width and Congestion Tax

Social trip
v cost . . ,
,. , , Average total cost

(4 lanes
(4 lanes I
Private trip cost

i 4 lanes
8 c LRAC = LRMC
H C
Total traffic volume (vehicles per hour)
The optimum traffic volume (V*) is the volume such that the marginal benefit of travel ( show n b\ the demand
curve) equals the long-run marginal cost of travel (LRMC). The four-lane highway minimizes the average total
V" drivers, so the optimum road width
cost of travel for is four lanes. The appropriate price of travel is C*, so a
congestion tax of T is added to the private tnp cost of C . The congestion tax equals the average road tax. so
congestion taxes cover the cost of the optimum road.
Because the congestion tax equals the average road cost, a system of congestion tolls
pays for the optimum road.

Is this a fluke, or is there something systematic about congestion tolls and road
costs? The expressions for average road cost and the congestion tax are as follows:
Average road cost = Average total cost — Private trip cost (19-2)
Congestion tax = Social trip cost — Private trip cost (19-3)
Therefore, the average road cost equals the congestion tax if the average total cost
(ATC) equals the social trip cost (STC). STC is actually the short-run marginal
cost of driving; it shows the increase in social cost from one additional driver. The
marginal-cost curve (STC) intersects the ATC curve minimum point. If the
at its
government builds the optimum highway, the market reaches the minimum point of
some ATC curve, so STC equals ATC. Therefore, the average road cost equals the
congestion tax. meaning that the revenue from congestion taxes covers the cost of
the road.
The revenue from a congestion tax provides an important signal to road builders.
[f tax revenue exceeds the total cost of building the road, the road should be widened.
In other words, any surplus tax revenue should be spent on widening the road. To
explain, suppose that the city builds a two-lane highway and imposes a congestion
tax of $4.34 (see Figure 19-2). If the total revenue from the tax ($4.34 times 1,400
drivers) exceeds the total cost of building the road, the road should be widened.
Who Pays for Highways?
Highways Auto drivers pay federal and state

are financed through various user fees.
taxes on gas, and auto parts. In addition to these normal user taxes, truck drivers
oil,
also pay user fees based on mileage and weight.

Since the early 1970s, user taxes have failed to cover the cost of building and
maintaining roads (Meyer and Gomez-Ibanez, 1981). During the 1960s, revenue
from user taxes exceeded the highway cost by about 25 percent. During the 1970s,
revenue from the gas tax did not keep pace with inflation because the gas tax is a
fixed amount per gallon, not a fixed percentage of the price. By 1975, there was a
deficit of about 25 percent. Urban road users actually paid more than their share of
the highway cost: the tax revenue collected from urban road users exceeded the cost
of urban roads by about 14 percent. In contrast, rural travelers paid only about 56
percent of the cost of rural roads.
Congestion and Land-Use Patterns

This section explores the effects of congestion externalities and congestion taxes
on urban land-use patterns. It uses the concepts developed in Part II (Land Rent
and Urban Land Use Patterns) to show how a system of congestion taxes would
change residential land use and city size. The land-use effects can be divided into
partial-equilibrium effects and general-equilibrium effects.
Partial-Equilibrium Effects
The congestion tax increases the slope of the residential bid-rent function. The con-
gestion tax increases the unit cost of commuting: in Figure 19-2, the congestion
tax adds 43.4 cents per mile to commuting cost. As explained in Chapter 8 (Land
Use Monocentric City), an increase in the unit cost of commuting increases
in the
the slopes of the housing-price function and the residential bid-rent function. Fig-
ure 19-8 shows three residential bid-rent functions. The relatively flat curve (R )
is the initial bid-rent function (before the imposition of congestion tax). After
congestion tax is imposed, the bid-rent function shifts to Rc if (1) the government
throws the tax money away and (2) the congestion tax does not change traffic volume
or travel times. Under these two assumptions, the congestion tax decreases the bid
rent at all locations, with the gap between R and R c increasing with distance to the
city center: the longer the commuting distance, the larger the decrease in the bid-rent
for land.
566 Pan V Urban Transportation
Figi re 19-8 Partial-Equilibrium Effects of Congestion Taxes

s
Bid rent with redistribution of congestion

tax revenue and decreased time cost
Distance to cit\ center (miles)
The imposition of congestion taxes shifts the bid-rent function from R„ to R c if the tax revenue is not redistributed
to the city's households. The redistribution of tax revenue combined with the decreases in tra\el times caused by
i
decreases in traffic % olume shifts the bid-rent function upward from R c to R e The bid rent rises above the original
> .
bid rent for distances up to u' miles and drops below the original bid rent for locations beyond u'
The third bid-rent function. /?,.. shows the actual effect of congestion taxes.
It lies above Rc for two reasons. First, if the government uses the congestion tax
revenue to cut other taxes, residents are willing to pay more for housing and land at
all locations. Second, the congestion tax decreases traffic volume, increasing travel
speeds and decreasing travel times. The decrease in the time cost decreases the unit
cost of travel, partly offsetting the effects of the congestion tax on the slope of the
bid-rent function. By comparing R t
, to the original bid-rent function. R„. we can
see that the congestion tax increases bid rents within /<* miles of the city center
and decreases bid rents beyond »*. Locations near the city center become more
attractive because the congestion tax bill is relatively low, so the bad news (increase
in congestion taxes)dominated by the good news (decrease in other taxes and
is
lower time from the city center become less attractive

cost). In contrast, locations far
because the congestion tax bill is relatively large. To summarize, the congestion tax
increases the relative attractiveness o\' locations close to the city center, increasing
the slope of the bid-rent function.
Figure 19-9 shows the effect of congestion taxes on the residential density
function. The density function shows population density (people per acre) at different
locations. As explained in Chapter 8. density increases as one approaches the city
center: land is more expensive, and people economize on the relatively expensive
land b\ consuming less of it (they live in smaller dwellings on less land), so there are
more residents pei acre. The congestion tax shifts the density function from D„ to D,
Figure 19-9 Congestion Taxes and Residential Density
Density function with

congestion tax
The congestion tax increases residential density up to u* miles from the city center and decreases density
beyond u*
in locations where land rent rises (distance less than i/*), households consume less
land, so density increases; beyond u* where land

, rent falls, households consume
more land, so density decreases.
The change in the density function shows that the congestion tax generates a
more compact city. After the tax, the city has a larger fraction of its population living
within a given distance of the city center, so it has shorter commuting distances
and travel times. As explained earlier in the chapter, the congestion tax causes the
city to move up its travel-demand curve: the increase in travel cost decreases traffic
volume. One reason for the decrease in traffic volume is that workers have shorter
commuting distances. This is shown by the change in the population density function:
the congestion tax increases the fraction of households living within a given distance
of the city center, so the average commuting distance decreases.
General-Equilibrium Effects
Figures 19-8 and 19-9 show the partial-equilibrium effects of the congestion tax.
In other words, they show the effects of the congestion tax on the residential land
market, ignoring the effects on other markets in the city. As explained in Chapter 9
(General-Equilibrium Land Use), general-equilibrium analysis examines the effects
of changes in the land market on the city's labor market.
Does the congestion tax increase or decrease total labor supply? The tax in-
creases total labor supply because it increases the efficiency of the city's transporta-
tion system and increases the welfare of city residents. The switch to the optimum
pricing scheme makes highway travel more efficient, generating a net increase in the
welfare of city residents. As shown in the example in Table 19-2, the congestion tax
makes both Helen (the typical highway user) and Louis (the typical nonuser) better
off. In Figure 19-2, the net gain from the congestion tax is the shaded area CJB.
Because the congestion tax generates a net gain for city residents, it increases the
relative attractiveness of the city, increasing total labor supply.
How does the congestion tax affect the labor market and the city's wage? The
initial increase in total labor supply causes excess supply of labor, so the city's
wage decreases. Firms respond to the wage cut by increasing the quantity of labor
demanded, narrowing the gap between labor supply and demand. The wage continues
to drop until labor supply equals labor demand. Equilibrium is restored with a lower
wage and a larger workforce.
How will the city look after general equilibrium is restored? One way to answer
this question is to use a computer to generate two pictures of the urban economy,
one before the congestion tax and one after the economy has adjusted to the new
tax. An urban general-equilibrium computer model allocates different plots of land
in a hypothetical city to firms and households. If a particular allocation does not
satisfy the conditions for general equilibrium, the computer tries a different one.
This groping process continues until an equilibrium allocation of land is found.
The computer model can be designed to compute a pre-tax equilibrium and a post-
tax equilibrium. To see the general-equilibrium effects of the congestion tax. one
compares the "before" and "after" snapshots of the urban economy.
Sullivan (1983) used a computer model to simulate the effects of congestion
taxes in a monocentric city. The results of the study are listed in Table 19-3. The
hypothetical city has an initial population of 5 1 8.000 and a wage of $6.00 per hour.
The average commuter travels 3.72 miles to work and spends 3.73 hours per week
commuting. As shown in the second column of the table, the congestion tax increases
the city's population and decreases its wage. The changes in population and wages
Table 19-3 General-Equilibrium Effects

of Congestion Taxes
Variable Initial City Congestion Tax City
Population 5 8.000
1 561,000
Wage $ per hour)
i 6.00 5.31
Commute distance (miles) 3.72 3.42
Commuting time (hours per week) 3.73 3.41
Percent of land used for transportation 17.9 16.2
Source: Arthur M. Sullivan, "The General Equilibrium Effects ol Congestion

Externalities," Journal oj I rban I onomu s 141 1983), pp. SO-KU.
<
are relatively large because the hypothetical cityis highly congested, much more
so than the typical U.S. The congestion tax causes households to move closer
city.
to their workplaces: the average commuting distance drops from 3.72 miles to 3.42
miles, and the average weekly commuting time drops by 0.32 hours. Another change
occurs in the transportation network: the decrease in traffic volume allows the city
to build narrower roads, so it uses less of its land for highways (16.2 percent instead
of 17.9 percent).
The analysis of congestion taxes provides an important lesson about urban prob-
lems and policies. Congestion is truly an urban problem in the sense that the largest
cities usuallyhave the most severe congestion. Nonetheless, a city that imposes a
congestion tax is likely to grow, not shrink. Because the congestion tax internal-
izes the congestion externality, it improves the efficiency of the city's transportation
network and thus increases the relative attractiveness of the city. The lesson is that
solving an urban problem is likely to cause economic growth, not decline.
Autos and Air Pollution

Another urban transportation problem is air pollution resulting from the emissions
of motor vehicles. Motor vehicles emit several pollutants that contribute to air pol-
lution in a significant way: volatile organic compounds (VOC), carbon monoxide

(CO), nitrogen oxides (NO x ), and sulfur dioxide (SO2). VOCs react with NO x in the
atmosphere to form ozone (O3) and also generate particulate matter. In the United
States, transport activities are responsible for about two-thirds of CO emissions,
about half ofVOC emissions, and about two-fifths of NO x emissions (Small and
Kazami, 1995). Motor vehicles powered by petroleum products emit carbon diox-
ide (CO2), one of the "greenhouse" gases that are accumulating in the atmosphere.
Most scientists believe the buildup of these gases will cause global temperatures to
increase, resulting in significant changes in weather patterns.
In the United States, most of the regulations concerning vehicle emissions are
included in the Clean Air Act of 1963, amended in 1970, 1977, and 1990. The act
specifies emission standards for various pollutants, including CO, VOC, and NO x .
Over the few decades, emissions have decreased dramatically, so a new car today
last
emits a small fraction of the amount emitted by cars made two decades ago (Small
and Kazami, 1995).
The Clean Air Act also defines standards for ambient air quality, specifying
maximum concentrations of CO, O3, NO x SO2, and particulate matter. The indi-
,
vidual states are responsible for meeting the standards. The states have used various
policies to decrease auto pollution, including vehicle inspection and repair programs,
restrictions on parking, and "cash for clunkers" (paying people for old cars and then
crushing the cars).
Amendments to the Clean Air Act (in 1977 and 1990) included one innovative
feature for the control of stationary pollution sources. In cities where the EPA stan-
dards were exceeded, new pollution sources were allowed if two conditions were
satisfied. First, the new sources had to adopt the most effective abatement technol-
ogy. Second, existing sources had to decrease their emissions to ensure that there was
a net improvement in air quality. This policy set the stage for the development of a
market for pollution rights, in which new firms purchase pollution rights from exist-
ing firms. Because this "offset" policy mimics a system of effluent fees, it generates
efficient abatement activity.
The Economic Approach: Effluent Fees

The economic approach to air pollution starts with a description of the externality
problem. Drivers base their travel decisions on the marginal private cost of driv-
ing, which is less than the marginal social cost because autos and trucks cause air
pollution. The pollution externality causes people (1) to drive cars that generate a
relatively large amount of pollution per mile driven and (2) to drive too many miles.
To an economist, the obvious response is to force people to pay for pollution, that
is, to internalize the externality.
One approach would be to install a in every car. Such a device
monitoring device
would measure government to tax the car owner for
the car's emissions, allowing the
the air pollution caused by the car. For example, if the estimated pollution cost per
unit of carbon monoxide is one cent, the monthly bill would be $2 for a vehicle that
emits 200 units per month, and $3 for a vehicle that emits 300 units. Such a scheme
is obviously impractical, but it illustrates the basic objective of policy: drivers should
pay for the pollution their cars generate. The trick is to estimate how much pollution
each car generates.
An alternative approach is to impose a one-time pollution tax on automobiles.
The government could estimate ) the lifetime emissions of a particular car model
( 1
and (2) the cost per unit of pollution. The pollution tax for a particular model would
be set equal to the product of total emissions and the unit cost. For example, if model
B is expected to emit 5,000 units of CO over its lifetime and the unit cost of CO
is one cent, its CO tax would be $50. If the same car is expected to emit 400 units
of lead and the unit cost of lead pollution is 50 cents, its lead tax would be $200.
The car's total pollution tax would equal the sum of the taxes for CO, lead, nitrogen
oxides, hydrocarbons, and particulates. Under this scheme, car buyers would pay
more for cars that generate more pollution, so consumers would have the incentive
to buy cleaner cars.
Another approach is to use a gasoline tax to increase the private cost of auto
travel. The tax would increase the cost per mile driven, so it would decrease the
total miles driven, thus decreasing air pollution. The problem with the gas tax is that
every driver would pay the same tax, regardless of how much pollution his or her
car generates. In other words, there would be no practical way to impose a higher
tax on cars that generate more pollution. Such a tax would decrease pollution by
decreasing miles driven, not by persuading people to drive cleaner cars.
What is the marginal external cost of auto driving? Small ( 1997) estimates an air
pollution cost of between $0,007 and $0,037 per vehicle-mile, or roughly between
$900 and $4,400 over 22 miles per

the life of the typical car. If the typical car gets
gallon, the associated gasoline tax would be between $0.15 and $0.81. Concerning
CO2 emissions and global warming, one proposal is to impose a tax on the carbon
content of fuels such as gasoline. Under a carbon tax that would stabilize CO2
emissions at 80 percent of the 1990 level, the tax per gallon of gasoline would
be between $0.45 and $0.67. Adding the taxes for conventional air pollution and
greenhouse gasses, the associated hike in the gasoline tax is between $0.60 and $ 1 .48,
which would generate gasoline taxes well below most European levels.
Transit Subsidies
An alternative approach for controlling auto pollution is to subsidize mass transit.

Although both autos and mass transit generate air pollution, the pollution per passen-
ger is lower on mass transit. If there are no pollution taxes, auto travel is underpriced
relative to transit. As explained earlier in the case of congestion externalities, the
underpricing of autos generates an excessive traffic volume (the equilibrium auto
volume exceeds the optimum volume) and too little transit ridership (the equilibrium
transit ridership is less than the optimum ridership). One response to the underpricing
of autos is to subsidize transit, the substitute good.
As shown in Figure 19-5, transit subsidies can be used to move the auto and
transitmarkets closer to their optimum points (A* and 7"*). As explained for the
case of congestion externalities, the basic problem with transit subsidies is that
they decrease the cost of transit below the marginal social cost, causing too much
transit travel. The excessive amount of travel is costly because society's productive
resources (labor, capital, fuel) are used inefficiently: the transit system will use
an excessive number of drivers, mechanics, and buses, and consume an excessive
amount of fuel. If the pollution cost per auto passenger exceeds the pollution cost
per transit rider by a wide margin, there are large air-quality benefits from transit
subsidies. The real question is whether the air-quality benefits dominate the resource
cost of the policy.
Auto Safety
The third transportation problem is auto safety. Traffic accidents generate two types
of costs. First, people involved in accidents are often injured or killed. Motor- vehicle
accidents are the leading cause of death for people ages 1 to 24; in 1993, 42,000
people died in motor vehicle accidents in the United States, and another 2 million
were injured (Small, 1997). About a third of the deaths and about three-fifths of
the injuries occur in urban areas. Second, traffic accidents during rush hours disrupt
traffic flows and contribute to congestion.
This section explores the effects of various public policies on auto safety and
traffic accidents. Federal safety legislation requires automakers to install a number of
safety features (e.g.. seatbelts. shatterproof windshields). One policy issue concerns
the effects of these regulations on highway death rates and accident rates. Another
issue concerns the effects of congestion and pollution policies on traffic safety. For
example, policies that encourage the switch to smaller cars decrease the congestion
problem (smaller cars take up less space on the highway) and the pollution problem
(small cars use less fuel and generate less pollution). Unfortunately, smaller cars are
less safe (injuries are more severe in collisions involving small cars), so there are
trade-offs between safety, congestion, and pollution.
The Effects of Safety Regulations

The Vehicle Safety Act of 1966 established safety standards for new cars, and leg-
islation since then has made the standards more stringent. Among the features man-
dated by the regulations are head restraints, padded dashboards, seatbelts. shatter-
proof windshields, dual braking systems, collapsible steering columns, and air bags.
These safety features add about $1,000 to the price of a car (Small. 1997). In the
three decades since the safety regulations have been in force, the number of highway
deaths per vehicle mile has decreased significantly.
The data on highway death two questions. First, to what extent did
rates raise
the safety regulations contribute to the decrease in highway death rates? Crandall
et al. 1986) have studied the effects of various factors on highway death rates. They
(
estimate that safety regulations decreased the number of highway fatalities by about
30 percent, saving about 20.000 lives per year.
The second question concerns the response of drivers to safety regulations.
Pelt/.man ( 1975) argues that because safety regulations make people feel safer, they
drive more recklessly and are therefore involved in more collisions. This is the
theory of risk compensation. This theory suggests that safety regulations gener-
ate benefits (greater safety in the event of a collision) and costs (more collisions).
If the regulations increase the number of collisions, the death rates of nonoccu-
pants (pedestrians and bicyclists) are likely to increase. In addition, an increase
in the number of accidents increases the congestion problem on crowded urban
highways.
The Optimum Speed. To explain a driver's response to safety equipment, consider

Duke, who lives in Hazard City. Every Saturday night. Duke must decide how fast
to drive from his suburban home to the dance hall in the city center. The benefit of
fast driving is that he spends less time on the road, giving him more time to dance
with Betty Lou. Figure 19-10 shows the marginal-benefit curve for speed. The curve
indicates how much Duke benefits from a one-unit increase in speed. When his speed
increases from 45 m.p.h. to 46 m.p.h.. he can spend more time dancing, and he is
willing to pay $16 for the additional dance time. The benefit curve is negatively
sloped, reflecting the diminishing marginal utility of dancing.
Table 1
9 —4 shows the expected cost of crashing at different speeds. As the speed
increases, the probability of crashing increases: the probability is only 0.001 for a
speed of 45 m.p.h.. but rises to 0.01 1 for a speed of 55 m.p.h. Faster driving also
causes more damage to Duke and his car. If he crashes at a speed of 45 m.p.h.. his
total crash cost (medical bills plus auto-repair costs) will be $2,000. If he crashes at
Chapter IV Autos and Highways 573
Figure 19-10 Optimum Speed and Safety Regulations
Original marginal cost
Marginal cost with

safety regulations
Speed (miles per hour)

The optimum speed is the speed at which the marginal benefit of speed (value of decreased
travel time) equals its
marginal cost (increased expected accident costs). In the absence of safety regulations. Duke chooses point A (50 m.p.h.).
Safety regulations decrease the expected cost of collisions, shifting the marginal-cost curve downward, increasing
the optimum speed to 52 m.p.h. (point B).
Table 19-4 Crash Costs and Optimum Speed
Expected Marginal Marginal

Speed Probability Crash Crash Cost of Benefit of
(m.p.h.) of Crash Cost ($) Cost ($) Speed ($) Speed ($)
45 0.001 2,000 2 17
46 0.002 3,000 6 4 16
47 0.003 4,000 12 6 15
48 0.004 5,000 20 8 14
49 0.005 6,000 30 10 13
50 0.006 7,000 42 12 12
51 0.007 8,000 56 14 11
52 0.008 9,000 72 16 10
53 0.009 10,000 90 18 9
54 0.01 11,000 110 20 8
55 0.011 12,000 132 22 7
a speed of 55 m.p.h., his total crash cost will be $12,000. The expected crash cost is
the probability of crashing times total crash cost. The expected crash cost rises from
$2 for a speed of 45 m.p.h. (0.001 times $2,000) to $132 for a speed of 55 m.p.h.
(0.01 1 times $12,000). The marginal cost of speed is the change in the expected
crash cost for a one-unit increase in speed. An increase in speed from 45 m.p.h. to
46 m.p.h. increases the expected crash cost from $2 to $6, so the marginal cost of
speed is $4 for a speed of 46 m.p.h. The marginal cost of speed increases with the
speed, reaching $22 for a speed of 55 m.p.h.
What is the optimum travel speed? Duke increases his speed as long as the
marginal benefit of speed (the value of the increase in dance time) exceeds the
marginal cost (the increase in the expected crash cost). In Figure 19-10. the optimum
speed (where the marginal-benefit curve intersects the marginal-cost curve) is 50
m.p.h. Table 19 —4 shows that a speed of 50 m.p.h. generates a marginal cost of $12
and an expected crash cost of $42.
What are the effects of safety regulations on the expected crash cost? Suppose
that Duke is forced to buy a car with seatbelts and a shatterproof windshield. If he
crashes in his new car, he is less likely to suffer serious injuries, so the expected cost of
crashing at each speed decreases. In Figure 19-10. the decrease in the expected crash
cost shifts the marginal-cost curve downward, increasing
optimum speed fromhis
50 m.p.h. to 52 m.p.h. Because the safety regulations protect Duke in an accident,
they cause him to take more risks, so the probability of crashing increases. The net
effects of the safety regulations are smaller than they would be if his speed did not
change: he has a lower injury cost if he crashes, but he is also more likely to crash.
Evidence of Risk Compensation. Peltzman (1975) uses two observations to sup-

port his theory of increased risk taking. First, collision rates were higher than ex-
pected in the years following the implementation of safety regulations. Second,
pedestrian death rates were also higher than expected. If people drive faster and
more recklessly in their safer cars, they take more risks for themselves and others
and may kill more pedestrians.
The results of Crandall et al. (1986) provide support for the risk compensation
theory. The death rates for pedestrians and bicyclists are inversely related to an index
of safety features, suggesting that drivers in safe cars take more risks and endanger
others. Because the decrease in driver deaths exceeds the increase in nondriver
deaths, safety regulations produce a net decrease in highway deaths.
Small Cars and Safety

Another safety issue concerns the costs and benefits of shifting to a fleet of smaller
cars.A greater reliance on small cars would provide some relief from transportation
problems because small cars use less fuel and generate less air pollution. In addition,
small cars take up less space on streets and highways, so a switch to smaller cars
would increase the carrying capacity of the road system, reducing congestion. On the
other hand, travel in small cars is less safe. The policy issue is whether the decreases
in energy consumption, pollution, and congestion are worth the cost (the increase in
traffic injuries and deaths).
The occupants of small cars are more likely to suffer serious injuries in a col-
lision. The state of New York has estimated the injury rates for different car types
(Meyer and G6mez-Ibanez, 1981). In a collision between two large cars (weight
greater than 3,500 pounds), the rate of serious injury is 5.2 per hundred million
vehicle miles. In a collision between a large car and a small car (weight less than
2,000 pounds), the injury rate is 3.7 for the occupants of the large car and 13.8 for
the occupants of the small car. The combined injury rate for such a collision is 8.8.
In a collision between two small cars, the injury rate is 9.7 for the occupants of each
car. In other words, a collision between two small cars is more lethal than a collision
involving a small and a large car.

Are small cars more or less likely to be involved in collisions? Studies of the
accident rates of different car sizes have generated mixed results: some studies found
higher collision rates for small cars, while others found no significant differences
in accident rates. Overall, it appears that small-car occupants are just as likely to be
involved in collisions and more likely to suffer serious injuries. Therefore, travel in
small cars is more dangerous than travel in large cars.
A shift toward a small-car fleet is likely to increase injury and death rates. As
the fraction of small cars increases, the fraction of collisions involving small cars in-
creases, increasing the rate of serious injury. According to Meyer and Gomez-Ibanez
( 1 98 1 ), if the percent of small cars in the fleet increases from 20 percent to 70 percent,
the percent of dangerous accidents (between a large and a small car, or between two
small cars) will increase by 50 percent to 95 percent. The increase in the fraction of
dangerous accidents increases traffic death rates. These numbers suggest that there
is a considerable cost associated with the switch to smaller cars. The policy question
is whether the benefits (less congestion and pollution) are greater than the cost.
Meyer and Gomez-Ibanez point out two additional factors that must be taken
into account in predicting the safety effects of small cars. First, about 37 percent of
urban traffic deaths involve pedestrians and bicyclists. If small cars are less lethal
for pedestrians and bicyclists, there would be fewer pedestrian and bicycle deaths.
Second, about 1 1 percent of urban traffic deaths result from single-car crashes (a car
hitting a fixed object). Since small cars are about as safe as large cars in this type of
accident, a shift to small cars is unlikely to increase the number of single-car fatalities.
The switch to small cars is also likely to change driving habits. A driver who
switches to a smaller, more dangerous car is likely to drive more cautiously to mitigate
the risk of serious injury. If Duke switches to a small car, his expected injury cost
would increase, shifting his marginal-cost curve upward. As a result, he decreases
his speed, partly offsetting the increase in the severity of accidents by decreasing
the probability of crashing. In addition, Duke would be more likely to fasten his
seatbelt, thus decreasing the severity of injuries.
Summary
The three most popular commuting trips are between suburbs (40 percent of
commuters), within the central city (33 percent), and from a suburb to the
central city (20 percent). About 4 of 5 workers (and 9 of 10 suburbanites)
commute by auto, truck, or van.
2. The demand for travel depends on the private cost of travel, the sum of the
monetary and time costs of the trip. The larger the trip cost, the fewer the
drivers.
3. Auto travel is subject to congestion externalities.

a.The external cost of travel is the time cost imposed on other drivers: the
marginal driver slows down traffic, forcing other drivers to spend more time
on the highway.
h. The social trip cost is the sum of the private and external trip cost.
c. Drivers base their travel decisions on private — not social — costs, so the
equilibrium volume exceeds the optimum volume.
d. A congestion tax internalizes the congestion externality, generating the
optimum traffic volume.
e The congestion tax would be higher during peak travel periods and along
more congested routes.
/ Estimates for the San Francisco Bay area suggest that the congestion tax for
travel during the peak period would be 65 cents per mile on central
highways. 21 cents on suburban highways, and 17 cents on fringe highways.
4. One way to implement congestion taxes is with a vehicle identification system,
under which every car is equipped with an electronic device that allows
sensors along the road to identify the car as it passes.
5. There are several alternatives to the congestion tax.

a. The gas tax is inappropriate because it would be the same on all routes at all
times.
b. Experiences with parking taxes suggest that they decrease traffic volume by
encouraging auto drivers to carpool and take mass transit. One problem is
that the tax does not depend on distance traveled.

c. The supply response to congestion is to increase highway capacity. The
benefit of capacity expansion is the increase in consumer surplus.
d. A transit subsidy would cause some auto drivers to switch to transit,
decreasing the congestion problem. The problem is that the subsidies may
increase transit ridership above the optimum level.
6. In the long run. the optimum road width and traffic volume are determined
simultaneously.
a. The optimum volume is where the demand (marginal-benefit) curve
intersects the long-run marginal-cost curve.
h. A congestion tax will generate the optimum traffic volume.
c. The congestion pay the full cost of the optimum highway. If the
tax will
revenue from congestion taxes exceeds the cost of the road, the road should
be widened.
7. A congestion tax increases the efficiency of the city's transportation system,
decreases commuting distances, and increases city size.
8. Autos and trucks generate several types of air pollution.
a. One response to the auto pollution problem is to impose pollution taxes on
new cars.
b. Another option is to impose a gasoline tax equal to the average external cost.
c. A third option is to subsidize transit, a policy that decreases pollution but
also increases transit ridership above its optimum level.
9. There are trade-offs associated with safety regulations: the regulations

decrease death rates of accident victims in autos, but increase death rates of
nonoccupants (pedestrians and bicyclists) and increase collision rates.
10. A shift toward small cars is and death rates: small-car

likely to increase injury
collisions are more lethal than large-car collisions, and small cars seem to be
just as likely to be involved in accidents. The policy question is whether the
benefits of small cars (less pollution and congestion) are greater than the safety
cost.

1. Suppose that the relationship between traffic volume and travel time is
Trip time = 12.0 + 0.001 • (Volume - 400) + 0.00001 • (Volume - 400)

2
The marginal benefit (demand) is $22.12 for a volume of 200 and drops by
$2.48 for every additional 200 drivers.
a. Use this equation to derive a table like Table 19-1.
b. What is the equilibrium traffic volume? What is the optimum volume?
c. What is the appropriate congestion tax?
2. You have been assigned the task of estimating the external trip cost at a traffic
volume of 1,600 vehicles per lane per hour for a one-mile section of an urban
freeway between two exits. You have traffic data for the relevant stretch of
highway for the last year. Outline your strategy for computing the external trip
cost. What information would you collect and how would you use it?
3. In estimating the congestion tax for peak-period travel, Keeler and Small
(1977) assumed that the imposition of a congestion tax would not affect traffic
volume.
a. Depict graphically the model of auto travel underlying this assumption.
b. Did Keeler and Small overestimate or underestimate the congestion tax? In
other words, if the assumption of a fixed traffic volume were dropped,
would the estimated congestion tax increase or decrease?
4. Consider the following quote from Peabody and Associates, a transportation
consulting firm: "According to the Peabody principle, an increase in highway
capacity increases traffic volume by an amount sufficient to leave the private
trip cost unchanged."
a. Use a graph to depict a situation in which the Peabody principle is correct.
b. Choose the word in parentheses that makes the following statement correct
and then explain your choice of words: "As the price elasticity of demand
for highway travel (increases, decreases) in absolute value, the Peabody
principle becomes a more accurate prediction of reality."
578 Tart V Urban Transportation
5. Sinead commutes from her suburban residence to the city center. When asked
her opinion of a proposed congestion tax of $5.00 per trip, she says, "Of
course I oppose the congestion tax. It would make me worse off by $5.00 per
trip. What do you think I am, stupid?" Critically appraise Sinead's statement.
6. A city has announced plans to widen a radial highway. The demand curve for
the typical traveler is T = 40 — P, where T is the number of trips per month
and P is the cost per trip (in cents). For example, if the cost is 10 cents, the
typical traveler will make 30 trips per month. If the new highway is built, the
cost per trip would decrease from 30 cents to 20 cents and the taxes of the
typical traveler would increase by $1.25 per month. Will the typical traveler
support the construction of the new highway? Explain.
7. In Figure 19-4, the widening of the road decreased trip cost and increased
traffic volume.
a. Compute the increase in consumer surplus if C = $5, Cw = $4,
V = 1.600, and Vw = 1,800. If the widening project costs $1,400, is it
worthwhile?
b. How would your answer change if every auto using the stretch of highway
generates $2 worth of air pollution?
c. What is the lesson from this exercise?
8. Consider a city that has both a congestion problem (it does not have a
congestion tax) and an auto pollution problem (drivers do not pay for the
pollution they generate).
a. Draw the private-trip-cost curve and the social-trip-cost curve. How do your
cost curves differ from those in Figure 19-2?
b. Label the optimum volume V* and the equilibrium volume V".
c. Suppose that the city builds a light-rail transit system and subsidizes
light-rail fares. Depict graphically a situation in which the subsidized
light-rail system generates an auto volume of V*.
d. Does the fact that the light-rail system generates the optimum auto volume
mean that it is an efficient policy?
9. Using the data in Table 19^L compute the effects of the following events on
Duke's optimum speed. Illustrate your answers with a graph (like
Figure 19-10).
a. Betty Lou is grounded for make-up violations, leaving Duke without his
favorite dance partner. The marginal benefit of speed decreases by $9 for
every speed.
/;. The normal country western band is replaced by Adam Smith and the
Invisible Hands, a punk rock band. Although Duke and Betty Lou enjoy
slam-dancing, it generates one-quarter of the utility of the two-step (a
country western dance).
c. The legal speed limit decreases to 46 m.p.h. The expected cost from
speeding tickets increases by $3 for every 1 m.p.h. over the speed limit: $3
for47 m.p.h.. $6 for 48 m.p.h., $9 for 49 m.p.h., and so on.
d. Highway improvements cut the probabilities of crashing in half.
10. Duchess, a resident of Hazard City, has the same marginal-benefit curve as
Duke (Table 19-4), but her marginal-cost curve is horizontal.
a. Suppose that the marginal cost of speed without air bags is $14. What is her
optimum speed?
b. Suppose that an air-bag system would decrease the marginal cost of speed
to $8. If the air-bag system costs $300, should Duchess buy the system?
1 1. Chopperville is evaluating the merits of using helicopters to clear up highway
accidents. Suppose that an accident simply stops traffic from the time the
accident occurs to the time the disabled vehicles are removed from the
highway. During rush hours, the typical accident stops traffic for 7,000 cars.
Under the current tow-truck system, the typical rush-hour accident stops traffic
for 20 minutes. Under a helicopter system, the typical rush-hour accident
would stop traffic for only eight minutes. Suppose that the cost of operating the
tow-truck system (including the costs of labor, fuel, and equipment) is $200
per accident and the cost of operating the helicopter system (including the
costs of labor, fuel, and equipment) would be $3,000 per accident. Your task is
to determine whether the helicopter system is superior to the tow-truck system.

a. Is there enough information to determine which system would be superior?
If not, what additional information do you need, and how would you use it?
b. If possible, specify "reasonable" values for the missing bits of information
and compute the benefits and costs of switching to the helicopter system.
c. Suppose that it is possible, for each accident, to instantaneously determine
which driver is at fault and that the responsible driver must pay for either a
helicopter or a tow truck to clear up the accident (at a cost of $200 for the
truck or $3,000 for the helicopter). Describe a simple public policy that
would provide the incentive for drivers to freely choose the socially
method of clearing up the accident (either the truck or helicopter).
efficient

General
Kraft, Gerald, and Thomas A. Domencich. "Free Transit." In Readings in Urban
Economics, ed. Matthew Edel and Jerome Rothenberg. New York: Macmillan, 1972,
pp. 459-80. Reports estimates of the elasticities of demand for auto travel.
Meyer, John R., and Jose A. Gomez-Ibafiez. Autos, Transit and Cities. Cambridge, Mass.:
Harvard University Press, 1981.
Small, Kenneth A. Urban Transportation Economics. Philadelphia, Penn.: Harwood
Academic Publishers, 1992. A comprehensive survey of the economics of urban
transportation.
"Economics and Urban Transportation Policy in the United States." Regional Science
.
and Urban Economics 27 997), pp. 67 1-9

( 1 1
U.S. Bureau of the Census. 7950 Census of Population: Journey to Work Metropolitan —
Commuting Flows. Washington, D.C: U.S. Government Printing Office, 1984.
Congestion
Alan A. The Urban Transportation System. Cambridge. Mass.: Joint Center for
Altshuler,
Urban Studies of MIT and Harvard. 1979. Chapter 9. pp. 317-73.
Downs, Anthony. Stuck in Traffic: Coping with Peak-Hour Traffic Congestion. Washington.
D.C.: The Brookings Institution. 1992. Explores the effects of adopting several
anticongestion remedies, including gasoline taxes, highway expansion.
high-occupancy lanes, and increased employment and population density.
Giuliano. Genevieve, and Kenneth A. Small. "Alternative Strategies for Coping with Traffic
Congestion." University of California Transportation Center. Working Paper 188.
Berkeley: University of California Transportation Center. 1994.
Keeler. Theodore E., and Kenneth A. Small. "Optimal Peak-Load Pricing. Investment and
Service Levels on Urban Expressways." Journal of Political Economy 85 1977). (
pp. 1-25. Develops a long-run model of highway pricing and investment and uses the
model to estimate optimum congestion taxes.
Keeler,Theodore E.; Kenneth A. Small; George S. Cluff; and Jeffrey K. Finke. "Optimal
Peak-Load Pricing. Investment and Service Levels on Urban Expressways." Working
Paper no. 253, Institute of Urban and Regional Development. University of California,
Berkeley. 1975. Develops a long-run model of highway pricing and investment and
uses the model to estimate optimum congestion taxes.
McConnell-Fay, Natalie. "Tackling Traffic Congestion in the San Francisco Bay Area."
Transportation Quarterly 40 1986), pp. 159-70. A planner's view of the congestion
(
problem and alternative policy responses.

Meyer, John R., and Jose A. Gomez-Ibanez. Autos. Transit and Cities. Cambridge. Mass.:
Harvard University Press. 1981. Chapter 1. pp. 185-229. 1
Moore, Terry, and Paul Thorsnes. The Transportation/Land Use Connection. Chicago:
American Planning Association, 1994. Explores the effects of public policx on
transportation systems and land-use patterns.
Pozdena, Randall J. "Unlocking Gridlock." Federal Reserve Bank of Sun Francisco Weekly
Letter. December 1988, pp. 1-5. Updates the Keeler and Small (1977) estimates of
congestion tolls.
Small, Kenneth A. "Urban Traffic Congestion: A New Approach to the Gordian Knot."
Brookings Review 1 1 ( 1993). pp. 6-11.
Small. Kenneth A., and Jose A. Gomez-Ibanez. "Road Pricing for Congestion
Management: The Transition from Theory to Policy." In Road Pricing. Traffic
Congestion, and the Environment, ed. Kenneth J. Button and Erik T. Verhoef.
Cheltenham. U.K.: Edward Elfar. 1998.
Sullivan, Arthur M. "The General Equilibrium Effects of Congestion Externalities." Journal
of Urban Economics 14 1983). pp. 80-104. Uses a general-equilibrium computer
(
model to simulate the effects of optimum congestion externalities on land rent, land
use, and city size.
Vickrey, William S. "Pricing in Urban and Suburban Transport." American Economic
Review 53 1963), pp. 452-65. Reprinted in Urban Economics: Readings and Analysis,
(
cd. Ronald L. Grieson. Boston: Little, Brown. 1973, pp. 106-18.

Walters. A. A. The Economics of Road User Charges. Baltimore: Johns Hopkins University
Press, 1968. Chapters 2 and 3.
Watson. Peter I... and Edward P. Holland. "Congestion Pricing — The Example of
Singapore." In Urban Transportation Economics, Special Report ISI. Washington.
D.C.: Transportation Research Board, 1978, pp. 27-30. Reprinted from Finance and
Development 13 (March 1976), pp. 20-24.
Williams, Stephen R. "Getting Downtown, Relief of Highway Congestion through Pricing."
In Urban Economic Issues, ed. Stephan Mehay and Geoffrey Nunn. Glenview, 111.:
Scott, Foresman, 1984. Discusses the efficiency and equity effects of congestion tolls
and describes Singapore's congestion policy.

Wohl, Martin. "Congestion Toll Pricing for Public Transport Facilities." In Public Prices for
Public Products, ed. Selma Mushkin. Washington, D.C.: Urban Institute, 1972,
pp. 245-46.
Parking Taxes and Other Alternatives to Congestion Taxes
DiRenzo, J.; B. Cima; and E. Barber. "Parking Management Tactics." Vol. 3, Reference
Guide. Washington, D.C.: U.S. Department of Transportation, 1981. Reports the
results of the increase in parking cost in Ottawa, Canada.
Kulash, Damian. Parking Taxes as Roadway Prices: A Case Study of the San Francisco
Experience. Washington, D.C.: Urban Institute, 1973, pp. 25-28.
McElhiney, Paul T. "Evaluating Freeway Performance in Los Angeles." Traffic Quarterly
14, no. 3 (July 1960), pp. 296-312.
Miller. Gerald K., and Carol T. Everett. "Raising Commuter Parking Prices — An Empirical
Study." Transportation 1 1 (1982), pp. 105-29. Examines the result of an increase in
parking cost for government employees in Washington, D.C.
Mohring, Herbert. "The Benefits of Reserved Bus Lanes, Mass Transit Subsidies, and
Marginal Cost Pricing in Alleviating Traffic Congestion." Chapter 6 in Current Issues
inUrban Economics, ed. Peter Mieszkowski and Mahlon Straszheim. Baltimore: Johns
Hopkins University Press, 1979, pp. 165-95.
Parody, Thomas E. "Implementation of a Peak-Period Pricing Strategy for CBD Parking."
Transportation Quarterly 38 (1984), pp. 153-69. Describes the results of an
experiment with peak-period parking taxes in Madison, Wisconsin.
Shoup, Donald C. "Cashing Out Free Parking." Transportation Quarterly 36 (1982),
pp. 351-64.
. "Cashing Out Employer-Paid Parking." Access (1993). pp. 3-9.
. "Cashing Out Employer-Paid Parking: An Opportunity to Reduce Minimum Parking
Requirements." University of California Transportation Center Working Paper 204,
1994.
Willson, Richard W. "Parking Subsidies and the Drive-Alone Commuter: New Evidence
and Implications." Paper presented at the Transportation Research Board Annual
Meeting, 1988. Reports the results of a survey of parking subsidies in Los Angeles.
Highway Investment
Downs, Anthony. "The Law of Peak Hour Expressway Congestion." Traffic Quarterly, July
1962, pp. 393-409. Reprinted in Anthony Downs, Urban Problems and Prospects, 2nd
ed. Skokie, 111.: Rand McNally, 1976, pp. 185-99.
Keeler, Theodore E., and Kenneth A. Small. "Optimal Peak-Load Pricing, Investment and
Service Levels on Urban Expressways." Journal of Political Economy 85 (1977),
pp. 1-25. Develops a long-run model of highway pricing and investment and uses the
model to estimate optimum congestion taxes.
Keeler.Theodore E.: Kenneth A. Small: George S. Cluff; and Jeffrey K. Finke. "Optimal
Peak-Load Pricing. Investment and Sen ice Levels on Urban Expressways." Working
Paper no. 253. Institute of Urban and Regional Development. University of California.
Berkeley. 1975. Develops a long-run model of highway pricing.
Kraus. Marvin. "Scale Economies Analysis for Urban Highway Networks." Journal of
Urban Economics 9 1981). pp. 1-22.
(
Mohring. Herbert. "Urban Highway Investments." In Measuring Benefits of Government

Investments, ed. Robert Dorfman. Washington. D.C.: Brookings Institution. 1965.
Auto Safety
Altshuler, Alan A. The Urban Transportation System. Cambridge. Mass.: Joint Center for
Urban Studies of MIT and Harvard. 1979. Chapter 7. pp. 210-51.
Case. H. Vs.: A. Burg: and J. D. Baird. "Vehicle Size and Accident Involvement: A
Preliminary Study." Journal of Safety Research 5 1973). pp. 26-35.
(
Crandall. Robert W.; Howard K. Gruenspecht: Theodore E. Keeler: and Lester B. Lave.
Regulating the Automobile. Washington. D.C.: Brookings Institution. 1986. Chapter 4
discusses the effects of safety regulations on highway death rates. Chapter 3 explores
the cost of safety regulations. Chapter 7 discusses the conflicting goals of regulations
on auto and fuel economy.
safety, auto emissions,
Meyer, John and Jose A. Gomez-Ibanez. Autos, Transit and
R.. Cities. Cambridge. Mass.:
Harvard University Press. 1981. Chapter 13. pp. 254-76.
Peltzman. Sam. Regulation of Automobile Safety. Washington. D.C.: American Enterprise
Institute. 1975. Explores the effect of auto safety regulation on driver behavior.
Auto Pollution
Ackerman. Bruce. "Clean (Cough) Air." New York Times. August 20. 1977. p. 21.
Altshuler. Alan A. The Urban Transportation System. Cambridge. Mass.: Joint Center for
Urban Studies of MIT and Harvard. 1979. Chapter 6. pp. U2-2(R
Crandall. Robert W; Howard K. Gruenspecht: Theodore E. Keeler: and Lester B. Lave.
Regulating the Automobile. Washington. D.C.: Brookings Institution. 1986. Chapter 5
discusses the effects of emissions policy on air pollution. Chapter 3 explores the cost ot
emissions controls. Chapter 7 discusses the conflicting goals of regulations on auto
safety, auto emissions, and fuel economy.
Freeman. A. Myrick. Air and Water Pollution Control: A Benefit-Cost Assessment New
York: John Wile} & Sons. 1982. Fstimates the benefits and costs of air pollution
programs.
Gomez-Ibanez. Jose A., and Gary R. Fauth. "Downtown Auto Restraint Policies: The Costs
and Benetits for Boston." Journal of Transport Economics and Policy 14 1^80). (
pp. 133-53.
Meyer, John R.. and Jose A. Gomez-Ibanez. Autos, Transit and Cities. Cambridge. Mass.:
Harvard University Press. 1981. Chapter 9. pp. 149-70.
Mills. Fdwin S.. and Lawrence
J. White. "Government Policies toward Automotive
Emissions Control." Approaches to Controlling Air Pollution, ed. Ann F.

In
Friedlaender. Cambridge. Mass.: MIT Press. 1978. pp. 348—409.

National Academy o\ Sciences and National Acadenn oi Engineering. Air Quality and
Automotive Emission Control. Vol. 4. The Costs and Benetits ofAutomobile Emission
Control. Prepared for the Senate Committee on Public Works. Washington, D.C.: U.S.
Government Printing Office, 1974.
Small, Kenneth A., and Camilla Kazimi. "On the Costs of Air Pollution from Motor
Vehicles." Journal of Transport Economics and Policy (1995).
Straszheim, Mahlon R. "Assessing the Social Costs of Urban Transportation Technologies."
In Current Issues in Urban Economics, ed. Peter Mieszkowski and Mahlon
Straszheim. Baltimore: Johns Hopkins University Press, 1979. Reviews the literature
on the externalities associated with automobiles and highways.
Chapter 20
M ASS TRANSIT
While real trolleys in Newark, Philadelphia, Pittsburgh, and Boston languish for
lack of patronage and government support, millions of people flock to
Disneyland to ride fake trains that don 't go anywhere.
Kenneth T. Jackson
f~J7~ his chapter explores the problems and prospects of urban mass transit. In the
^y last 40 years, the percentage of urban travelers using mass transit decreased
and the mass-transit sector went from a marginally profitable private-sector enter-
prise to a deficit-ridden public-sector operation. This chapter explores some of the
reasons for this transformation and identifies some changes that could improve the
urban mass-transit system. Here are some of the questions we'll consider:
1 How do commuters pick a travel mode, and why do so few choose mass
transit?
2. What are the relative costs of different transit systems (heavy rail, light rail,
and buses) and how do they compare to the costs of an auto-based system?
3. Is there a dark side to light rail?
4. Are high-occupancy vehicle lanes effective in increasing vehicle occupancy

and decreasing congestion?
5. How would the privatization of urban mass transit affect the features of the
transit system, ridership, and costs?
6. Do heavy-rail systems like San Francisco's BART have large effects on
land-use patterns?
Mass Transit Facts

As shown in Figure 19-1 (in Chapter 19), a small percentage of metropolitan com-
muters use mass transit. The share of commuters using transit is 6 percent for all
585
Figure 20-1 Transit Modal Choice for Residents of Central Cities

and Suburbs
4.500.000
4.000.000
~ 3,500,000
si
C
-
o 3.000.000
s
w> 2.500.000
B
g 2.000.000
^ 1.500.000
1,000,1
500.000
Central-city residents Suburban residents
Heav) rail 1 .492.634 260.648

Light rail 52.810 24.197
Bus 2.381.213 950.807
Si n R( i : U.S. Bureau of the Census. I WO ('< 'nsus oj Population and Housing, SSTF20. Journey to Work in
the United States. (Washington, D.C.: U.S. Government Printing Office, 1994).
metropolitan commuters, 1 1 percent for central-city residents, and 2 percent for

suburban residents. Figure 20-1 shows the distribution of transit ridership among
three transit modes and suburbs. About 6 percent of
for residents of the central-city 1
central-city transit commuters use the bus, and 38 percent use heavy rail (subways
and elevated trains). Suburban transit riders are less numerous; about 77 percent of
them use the bus and 21 percent use heavy rail. Although light rail systems have
received a lot of attention in the popular press, ridership on these newer systems is
relatively low.
Transit ridership varies with the path of the commuter trip. As shown in
Table 10-3 (in Chapter 10), about 40 percent of metropolitan workers commute
from one suburb to another. Less than 2 percent of these commuters use mass tran-
sit. Transit usage is highest among workers who commute within the central city:
about one in six of these commuters use mass transit.
Table 20-1 shows trends in transit ridership. Total ridership in 1995 was about
45 percent ol ridership in 1 950. After steady increases in ridership between 1970 and
1990. the number of transit rides has decreased recently, with the largest reductions in
bus ridership. The trolley coach (a bus powered by overhead electric wires) reached
its peak in 1 950 and has declined since then. The use of light-rail systems | streetcars)
decreased steadih between 1940 and 1980. but has recently staged a recovery:
Chapter 20 Mass Transit 587
Table 20-1 Public Transit Ridership, 1940-1995 (Numbers in millions)
Year Total Rides Heavy Rail Light Rail Trolley Coach Motor Bus
1940 13,098 2,382 5,943 534 4.239

1950 17,246 2,264 3,904 1.658 9,420
1960 9,395 1,850 463 657 6,425
1970 7,332 1.881 235 182 5,034
1980 8.567 2,388 133 142 5,837
1990 8,873 2,346 176 126 5.784
1995 7,763 2,033 251 119 4.848
SOURCE: American Public Transit Association. Transit Fact Book (Washington. DC. 1991 ).
systems have recently been built or restored in Portland, San Jose, Sacramento,
Buffalo, San Diego, and Pittsburgh. In Canada, there are new light-rail systems in
Edmonton and Calgary.

There is substantial variation in transit ridership across metropolitan areas.
Figure 20-2 shows the percent of workers who use public transportation for
Figure 20-2 Percentage of Workers Using Public Transportation,

Selected Cities
30.0 r
24.8
25.0
-
•e 2o.o
15.0
-
3.4
111
0.2
10.0
- 9.3
ta 8.6
7.4
6.1
5.1
4.6 4.4
5.0 3.7 4.0
2.9
2.3
n
2 S 2
U .;o
—
u
— U 43
u
R" 3>
a u. -
Source U.S. Bureau of the Census. 1990 Census of Population and Housing, Labor Force Characteristics and Journey
to Work. (Washington, D.C.: U.S. Government Printing Office, 1994).
metropolitan areas (CMSAs or MSAs) with at least 1 million workers. The cities
are arranged, left to right, in descending order of total employment. In the New
York CMSA, about 25 percent of metropolitan workers use public transit. No other
metropolitan area has a transit share greater than 14 percent. In fact, there are only
three other metropolitan areas where transit ridership is at least 10 percent: Chicago,
Washington D.C., and Philadelphia. Transit ridership is much higher among central-
city residents: 47 percent of workers who live in New York City use transit, as do
26 percent of workers in Chicago and 25 percent of workers in Philadelphia.
What are the elasticities of demand for mass transit? By how much would transit
ridership increase when transit fares decrease or transit service improves? There are
four general conclusions from empirical studies of transit ridership:
1. Price elasticity. The demand for transit is price-inelastic, with a price

elasticity between —0.20 and —0.50 (Beesley and Kemp, 1987). A common
rule of thumb is that a 10 percent increase in fares decreases ridership by
about 3.3 percent, meaning that the price elasticity is —0.33. According to
Small ( 1992), the price elasticity is relatively large for off-peak trips and
trips by high-income commuters.
2. Time elasticities. The demand for transit is more responsive to changes in
travel time. For the line-haul portion of the trip (time spent in the vehicle),
Domencich, Kraft, and Valette (1972) estimate an elasticity of —0.39: a
10 percent increase in line-haul time decreases ridership by about
3.9 percent. For access time (time spent getting to the bus stop or transit
station), they estimate an elasticity of —0.71.
3. Value of travel time. According to Small ( 1992), the average commuter
values the time spent in transit vehicles at about half the wage: the typical
commuter would be willing to pay half of his hourly wage to avoid an hour
on the bus or train. The value of time spent walking and waiting is two to
three times larger: the typical commuter would be willing to pay between
1.0 and .5 times his hourly wage to avoid an hour of walking or waiting
1
time. The value of travel time increases less than proportionately with
income: a 50 percent difference in income generates less than a 50 percent
difference in the value of travel time.
4. Noncommuting trips. The elasticities of demand for noncommuting travel
are higher than the elasticities for commuting trips.
There are three principal implications from these empirical results. First, an
increase in transit fares increases total fare revenue. A fare increase decreases rider-
ship by a relatively small amount, so total revenue (fare times ridership) increases.
Kraft and Domencich (1972) suggest that dropping the price of mass transit to
zero — making transit free— would increase ridership by only about a third. Second,
a simultaneous improvement in service and fares may increase ridership. Suppose
that a transit authority increases the frequency and speed of buses and finances the im-
proved service with increased fares. Because people are more sensitive to changes in
time cost than changes in fares, ridership may increase. Third, service improvements
that decreasewalking and waiting time (more frequent service, shorter distances be-
tween stops) generate larger increases in ridership than improvements that decrease
line-haul time.
Choosing a Travel Mode: The Commuter

This section discusses modal choice from the perspective of the commuter. The
commuter chooses the mode that minimizes the total cost of travel (the sum of
time and monetary costs). Suppose that a commuter has three travel options: the
automobile, the bus, and a fixed-rail transit system such as San Francisco's Bay
Area Rapid Transit (BART) or Washington, D.C.'s, Metro.
The commuting trip can be divided into three parts. The collection phase in-
volves travel from the home to the main travel vehicle. The auto mode has no col-
lection cost because the driver uses his own vehicle. The bus mode has a moderate
collection cost because the rider must walk from his home to the bus stop. The fixed-
rail system has the highest collection cost because there is a relatively long average
distance between transit stations, so riders must either walk a long distance or take
another mode (e.g., auto or bus) from the home to the transit station. The line-haul
phase is on the main travel vehicle. The heavy-rail mode
the part of the trip spent
has the shortest line-haul time because it has an exclusive right-of-way, so it avoids
rush-hour congestion. Although the bus and the auto both travel on congested streets
and highways, the auto is faster because the bus must stop to pick up passengers
along the way. The distribution phase involves travel from the end of the vehicular
trip (at the downtown transit station, parking garage, bus stop) to the workplace. If
parking is available near the workplace, the auto has the shortest distribution time,
followed by the bus and the fixed-rail system.
An Example of Modal Choice

Table 20-2 lists the monetary and time costs of the three travel modes for Carla, a
commuter who travels 10 miles from her suburban home to a job in the central city.
The computations are based on the assumption that Carla values the time spent on
the transit vehicle at half her wage rate and values time spent walking and waiting at
1.5 times her wage rate. If her wage is $12 per hour, she is willing to pay $6 to avoid
one hour of in-vehicle time and $ 1 8 to avoid one hour of walking and waiting time.
Therefore, the cost of walking and waiting time is 30 cents per minute, and the cost
of in-vehicle time is 10 cents per minute.
1 Collection time cost. Carla walks to the bus stop or the BART station.
Since the bus stop is closer to her home, the bus has a lower collection cost.
2. Line-haul time cost. BART is the fastest mode (it operates on an exclusive
right-of-way), followed by autos (which travel on congested streets) and
buses (which travel on congested streets and stop to pick up passengers).
Table 20-2 Costs of Travel Modes and Modal

Choice
Auto Bus BART
Collection Time Cost

Collection time (minutes) 10 15
Cost per minute ($) 0.30 0.30 0.30
Collection time cost ($) 0.00 3.00 4.50
Line-Haul Time Cost

Line-haul time (minutes) 40 50 30
Cost per minute ($) 0.10 0.10 0.10
Line-haul time cost ($) 4.00 5.00 3.00
Distribution Time Cost

Distribution time (minutes) 5 9
Cost per minute ($) 0.30 0.30 0.30
Distribution time cost ($) 0.00 1.50 2.70
Monetary Cost
Operating cost or fare ($) 2.00 1.00 1.50
Parking cost ($) 3.00 0.00 0.00
Total Monetary Cost ($) 5.00 1.00 1.50

Total Time Cost ($) 4.00 9.50 10.20
Total Cost ($) 9.00 10.50 11.70
3. Distribution cost. Carla parks her auto in a company parking lot under her
office building, so the distribution cost of the auto trip is zero. The bus stop
is relatively close to the office, so the bus has a lower distribution cost than
BART.
4. Monetary cost. The monetary cost of the auto trip is $0.20 per mile, or
$2.00. The bus fare is $1, and the BART fare is $1.50. Half of the $6.00
parking cost is allocated to the morning commute.
The cost of driving is less than the cost of the bus and BART. Although the
monetary cost of driving exceeds the monetary cost of the bus by $4. the lower time
cost of the auto more than offsets its higher monetary cost. Similarly, the auto is more
expensive but faster than BART. The largest difference in time cost is for collection
and distribution costs, where the auto has a cost advantage of $4.50 over the bus
($3.00 for collection + $1.50 for distribution), and $7.20 over BART ($4.50 +
$2.70). Since the auto has a lower total cost than both the bus and BART, Carla
drives to work.
What would it take to persuade Carla to switch from her auto to mass transit?
There are several possibilities:
1 . Subsidized transit. If the bus and BART were free, Carla would still drive.
To get her to switch to the bus. she would have to be paid a bribe of 5 cents 1
per bus ride. A bribe of $1.21 would cause her to switch to BART. Transit
fares must decrease by relatively large amounts to offset the time-cost
advantages of the automobile.
2. Line-haul time. If the line-haul time of the bus decreased from 50 minutes
to less than 35 minutes, Carla would ride the bus. She would take BART if
its line-haul time decreased to less than 3 minutes.
3. Collection and distribution time. Carla would ride the bus if its collection
and distribution time decreased from 15 minutes to less than 10 minutes
and would switch to BART if its collection and distribution time decreased
from 24 minutes to less than 15 minutes.
4. Auto monetary cost. Carla would ride the bus if the car's unit cost
increased from 20 cents to 36 cents per mile. As reported in Chapter 19,
estimates of congestion tax are in the range of 2 1 to 65 cents per mile for
rush-hour traffic, so the imposition of a congestion tax would cause her to
switch to the bus. Similarly, if the city imposed a pollution tax of at least
16 cents per mile, Carla would stop driving.
5. Parking cost. Carla would ride the bus if the parking cost rose to at least
$4.50 ($9 per day).
6. Wage. wage dropped to $8, the bus would be less costly than the
If Carla's
auto, and wage dropped to $6, BART would be less costly than the
if her
auto. As the wage decreases, the opportunity cost of travel time decreases,
increasing the relative attractiveness of the modes with relatively low
monetary cost and high time cost.
To summarize, to get Carla to switch to transit, the changes in either the monetary
would have to be relatively large. On the other hand, she
cost or the line-haul time cost
would switch to transit with relatively small changes in collection and distribution
time costs. These conclusions are consistent with the transit elasticities discussed
earlier in the chapter.
What type of commuters take mass transit instead of driving? There are five
possibilities:
1 Proximity to stops and stations. A person who lives near a bus stop or a
has a relatively low collection cost for transit and is more likely
rail station
to take the bus or BART.

2. Low opportunity cost. For a person who has a low opportunity cost of
travel time (a worker with a low wage, a student, a retired person), the time
cost of travel is relatively low. As a result, the advantages of transit (lower
monetary cost) dominate the disadvantages (longer commuting time). Such
a person is more likely to choose transit.
3. Low walking cost. A person who enjoys walking has relatively low
collection and distribution costs and is more likely to choose transit.
4. Disutility of driving. A person who dislikes the hassle and anxiety of

driving is more likely to choose transit. In terms of the numerical example.
such a person has a relatively high cost for in-vehicle auto time and is more
likely to take the bus or BART. In contrast, a person who considers driving
a challenging form of athletics will buy a pair of gloves and drive to work.
5. No automobile. Many of the poor do not have access to an auto, so their
only option is to use public transit.
Transit Service and Modal Choice

How do changes in transit design and scheduling affect ridership? Transit ridership
increases as a result of service improvements that decrease the time cost of transit.
As explained earlier, commuters are most responsive to changes in collection and
distribution time (walking and waiting time).
Designing the Bus System. Consider first the bus system. The bus company
affects time cost in two ways. First, it chooses the bus headway, the period of time
between buses on the bus route. As the headway decreases, riders spend less time
waiting at the bus stop, so their time cost decreases. Second, the bus company
chooses the frequency of stops in the residential collection area. An increase in the
frequency o\' stops decreases walking distances and collection cost. Similarly, the
more frequent the stops in the downtown distribution area, the lower the distribution
cost. An increase in the frequency of stops also increases line-haul (in-vehicle) time:
more time is spent picking up and dropping off passengers, so the trip takes longer.
Designing the Fixed-Rail System. Consider next the design of a fixed-rail sj stem.
San Francisco's BART provides a nice illustration of the trade-offs associated with
the design of a rail transit system. There are two basic design trade-offs:
1 . Mainline versus integrated system. BART is considered a mainline

system because it relies on other transit systems to collect its riders from
residential neighborhoods. Riders must either walk, drive, or ride a bus to
the BART station. The alternative to the mainline system is an integrated
system, under which commuters make the entire commute trip on a single
transit mode.
2. Spacing between stations. BART has widely spaced stations (about

2.5 miles apart), so there are few stops on the way to the city center. As
a result, travel time from the suburban station to the downtown station
(line-haul time) is relatively short, and in-vehicle time cost is relatively low.
On the other hand, the average commuter must travel a relatively long
distance from home to one of the widely spaced transit stations, so
collection cost (by foot, bus. or can is relatively high.
BART was designed to compete with the line-haul portion of the automobile trip.
It achieves this objective, providing comfortable, speedy service from the suburban
stations to the city center. There is a trade-off, however: collection cost is relatively
high because BART is a mainline system with widely spaced stations. Because
walking and waiting time is more costly than in-vehicle time, the negative attribute
(high collection cost) dominates the positive one (comfortable, speedy line-haul
travel), so the full cost of a BART trip is relatively high and BART has diverted a
relatively small number of auto commuters.
High-Occupancy Vehicle (HOV) Facilities
Many cities have established exclusive rights of way for buses, vans, and carpools,
allowing these high-occupancy vehicles (HOV) to bypass congested roadways. At
one extreme of the range of possible HOV facilities is a separate roadway, sometimes
called a "busway"; at the other extreme is simply designating a single lane on an
existing highway for use by HOVs (diamond lanes). There are HOV facilities in at
least 17 cities, and several cities are using HOV facilities as a core element in their
transportation plans (Giuliano and Small, 1994).
As explained by Giuliano and Small ( 994), there is evidence that HOV facilities
1
generate large benefits to commuters. HOV facilities improve bus service in two
ways. First, line-haul times decrease because buses are able to bypass congestion.
Second, any increases in bus ridership allow the bus authority to run buses more
frequently, and service means lower collection and distribution costs.
more frequent
Mohring 1979) and Small (1983) show that the combined effect of these two service
(
improvements leads to significant increases in bus patronage.

Richmond (1998) describes recent experiences with HOV facilities that use
exclusive right-of-way for buses, van pools, and car pools:
• Pittsburgh's East Busway has higher travel speeds than similar light-rail
facilities, and it increased bus ridership.

• Miami's South Dade Busway increased bus ridership by 50 percent, in large
part because it offered more frequent service and thus economized on riders'
collection and distribution costs.
• Ottawa's bus transitway system allows fast and frequent express service,
which has helped the transit system achieve the highest market share of
North American metropolitan areas of its size.
• Houston's HOV system is extensive, with almost 70 miles of HOV lanes
(generally in freeway medians) in five transportation corridors. The ridership
figures for the HOV system include bus riders, van poolers, and car poolers,
reflecting Houston's emphasis on mobility rather than transit ridership.
What are the effects of diamond lanes on commuters who continue to drive?
There good news and bad news. The good news is that some auto drivers switch
is
to buses and car pools, so there is less auto traffic. In Figure 20-3, the diamond lanes
shift the demand curve to the left, decreasing trip cost for the original trip-cost curve.
The bad news is that there are fewer lanes for the remaining autos: the diamond lanes
decrease the amount of road space, so the congestion threshold decreases and the trip-
cost curve shifts to the left. In this example, the shift of the demand curve is small rel-
ative to the shift of the cost curve, so the trip cost increases from $5 to $6. This was the
result of the ill-fated diamond lane on the Santa Monica Freeway in Los Angeles. The
diamond lane shifted a relatively small number of commuters to car pools, so 25 per-
cent of the freeway's capacity was used by only 6 percent of the vehicles (Dahlgren,
1995). After the resulting public outcry, the lane was returned to general use.
594 Pan V Urban Transportation
Figure 20-3 Diamond Lanes and Auto Trip Costs

Private trip cost
with diamond lane-.
Initial private
trip cost
Initial demand
Traffic volume
The addition of diamond lanes (reserved for buves and carpools) affects travel innondiamond lai^ in two ways:
there fewer vehicles in nondiamond lanes, so the demand curve shifts to the left: there are feu or lanes, so the
.ire
trip-cost curve shifts to the left. If a relatively small number of vehicles are diverted to the diamond lanes, the net
ci^t o( travel in nondiamond lanes increases from S3 to S6.
Some HOV lanes are more favorable for commuters who continue to use general-
purpose lanes. If the shift of the demand curve in Figure 20-3 were large enough, the
demand shift would more than offset the cost-curve shift, leading to less congestion
and a lower private trip cost. Houston's HOY system is an example of a system that is
favorable to nonusers. The system caused large increases in carpooling and bus rider-
ship, so the HOY lanes diverted large volumes of traffic from general-purpose lanes.
The average vehicle occupancy increased by 20 percent and congestion dropped by
4 percent (Richmond. 1998).
Choosing a System: The Planner's Problem

This section discusses modal choice from the perspective of a transportation planner.
The planner must decide u hat t\ pe of transportation S) stem to build. There are three
options: an auto-based highway system, an integrated bus system, and a fixed-rail

system like BART or Metro. Estimates of the cost of these alternative systems are
provided b\ of transportation options
a stud) in the San Francisco Bay area (Keeler.
Merewitz, Fisher, and Small 1975).
Figure 20-4 Costs of Alternative Transit Systems

$
BART and feeder

8
bus
Peak passengers per hour (thousands)
The average cost of auto trafficindependent of auto volume, but the average costs of an integrated bus system
is
and BART decrease as volume increases. The bus system is more efficient than BART for all volumes studied and
more efficient than an auto-based system for volumes exceeding 1.100 passengers per hour.
Keeler et al. estimated the cost of three different commuting systems: an auto-
based system, an integrated bus system, and BART. The principal conclusions of
their study are shown in Figure 20-4. The horizontal axis measures the number
of commuters traveling through a transportation corridor during the one-hour peak
period. The vertical axis measures the long-run average cost of a "typical" commuting
trip (a six-mile line haul and additional time spent in residential collection and
downtown distribution). The cost curves show that the bus system is more efficient
than the auto system for volumes above 1,100 passengers per hour and is more
efficient than BART for all traffic volumes. The auto system is more efficient than
BART for volumes up to about 22,000 passengers per hour.
Cost of the Auto System
The cost of the auto system is the sum of the driver's time and operating cost and the
public cost of auto traffic. The public cost includes the cost of building the optimum
road system. As explained in the previous chapter, the revenue from the optimum
congestion tax equals the cost of building the optimum road, so congestion taxes
can be used to both internalize congestion externalities and pay for the roads. The
public cost also includes the cost of air and noise pollution.
The average cost curve is two reasons. First. Keeler et al. assume
horizontal for
that the average operating cost and the average pollution cost (cost per mile) do
not depend on traffic volume. Second, the trip time (and private trip cost) does not
depend on traffic volume: as volume increases, the road is widened to accommodate
the increased traffic without any reduction of travel speeds. Given constant returns
to scale in highways, a doubling of highway capacity doubles traffic volume without
changing the trip time. Using data for 1972. Keeler et al. estimate that the full cost
of an auto trip with a six-mile line haul is $4. 15.
Cost of the Bus System
The cost of the integrated bus system is the sum of time cost, agency cost, and the
public cost of the bus system. The agency cost includes both the operating cost and
the capital cost. Included in the public cost are ( 1 ) the cost of modifying the roadway
to accommodate buses and (2) pollution cost.
The average cost curve is negatively sloped for two reasons. First, as ridership
increases, the fixed cost of administering the bus system is spread over more riders.
Second, as ridership increases, collection and distribution times decrease. An in-
crease in ridership decreases headways (the time between buses) and decreases
the space between bus stops, so riders spend less time walking to and waiting
for the bus. In addition, an increase in ridership allows a bus to fill up with a
shorter collection route. Less time is spent picking up passengers, so in-vehicle
time decreases. According to Keeler et al.. if the traffic volume along the corridor
exceeds 1,100 passengers, the integrated bus system is less costly than the auto
system.
Cost of BART
The BART option involves the mainline heavy-rail system and feeder buses to bring
commuters to the BART stations. The time cost of the BART option is the sum
of walking and waiting time costs and in-vehicle time cost. The agency cost is the
sum (if the operating and capital costs of the feeder buses and BART. Given the
substantial investment in tracks and rights-of-way. the capital cost of the BART
option is large.
The cost curve is negatively sloped for three reasons. First, the system has
a substantial fixed cost, which is spread over more riders as ridership increases.
Second, as ridership increases. BART headways decrease, decreasing waiting time.
Third, an increase in bus ridership decreases the time and monetary cost of feeder-bus
service.
BART is more costly than the bus system for three reasons. First, given its design
as a mainline system with widely spaced stations, the collection and distribution
costs are relatively high. Second, its capital cost is very large: the fixed cost of the
system was $1.6 billion. According to Webber (1976). BART could buy enough
buses to carry its passengers for only $40 million (2.5 percent of BART's capital
cost). Third. BART has a surprisingly high operating cost: in 1975. the operating
cost was 15.7 cents per passenger mile, compared to 13.6 cents for a bus. Most
planners were surprised to discover that the large investment in capital equipment
did not generate large savings in operating cost.
System Choice
The Keeler et al. study provides important information for transit planners. For
all corridor volumes studied (up to 30,000 passengers per hour), BART was more
costly than an integrated bus system. At a peak volume of 30,000 passengers per
hour, BART is 50 percent more costly than the bus system. BART's peak ridership
through the transbay tube is about half this amount. The lesson for planners is clear:
with the possible exceptions of New York City and Chicago, which have corridor
volumes exceeding 30,000 passengers per hour, an integrated bus system is likely
to be more efficient than a modern fixed-rail system like BART.
Experiences with new heavy-rail systems in other metropolitan areas are similar
to the BART experience. The Metro system in Washington, D.C., would have to
charge a one-way fare of $8 (Gordon and Richardson, 1989) to cover its operating
and capital costs. The Red Line
Los Angeles would have to charge a fare of at
in
least $1 1 to cover all its costs. Ridership on the new systems in Atlanta, Miami, and
Baltimore have fallen well short of levels required to make heavy rail less costly
than a bus system. The problems with heavy-rail systems should not be surprising
in light of Keeler et al.'s conclusion that corridor volumes must exceed 30,000 to
make a heavy-rail system competitive with a bus system.
Light Rail
In recent years, many medium-sized cities have built light-rail transit systems. Light
rail is modern version of the trolley and streetcar systems that were built in the
the
late 800s and early 1900s. The first modern light-rail system opened in Edmonton in
1
1978. Since then, new systems have opened in San Diego, Calgary, Buffalo, Portland,
Vancouver, Sacramento, and San Jose. Many other cities are planning to build new
systems or extend old streetcar and trolley systems.
How does light rail compare to buses and heavy rail? Gomez-Ibanez (1985) has
studied the cost and performance of new light-rail systems in San Diego, Edmonton,
and Calgary, and has come to the following conclusions:
1 . Capital cost. The capital cost of light rail includes the costs of the vehicles,
the track, and the power These costs make light rail
distribution system.
more costly than a bus system,which uses less expensive vehicles and
travels along regular streets and highways. Light rail is much less costly
than heavy rail: the cost per mile is $7 million to $20 million, compared to
$50 million to $200 million per mile for heavy rail. Light rail is less
expensive because it typically uses surface streets (instead of requiring
subways or elevated guideways) and is designed for lower speeds.
Table 20-3 Comparison of San Diego's Light-Rail

and Bus Systems
Bus System Light-Rail System
Costs per passenger

Operating cost $0.82 $1.02
Capital cost $0.34 $2.09
Total cost $1.16 $3.11
Ridership (thousands per year) 3.398 4.845
Cost per new light-rail passenger $5.47
1 "'
SOURCE: Jose A. Gomez-Ibanez. "A Dark Side to Light Rail' Journal of the American
Planning Association 51 (Summer 1985). pp. 337-51.
2. Operating cost. Compared to a bus system, light rail has a lower labor cost
because it needs fewer operators: a three-car train can be operated by one
person and can carry about 10 times as many riders as a standard bus.
Because light rail has a higher maintenance cost on its vehicles and its
right-of-way, however, its total operating cost is equal to or greater than the
operating cost of buses.
3. Performance. Most light-rail systems consist of one or two radial rail lines,
with feeder buses for the collection and distribution phases of the transit
trip. Like the heavy-rail systems, the light-rail systems have higher
collection and distribution costs than a bus system. Light rail also has
shorter line-haul times and provides a more comfortable ride than a bus.
Table 20-3 compares the cost of the San Diego light-rail system to the bus
routes it replaced. The total cost per light-rail passenger was $3.11, compared to
$1.16 for the bus system. Light rail has a moderately higher operating cost and
a much higher capital cost. The
system seems to have increased transit
light-rail
ridership: ridership in the area served by the system increased from 3.4 million to
4.8 million per year. According to Gomez-Ibanez, much of the increased ridership
was caused by population growth in the corridor served by light rail. If one assumes
that the entire increase in ridership was caused by the introduction of the light-rail
system, the cost per additional rider was $5.47. The San Diego experience suggests
that although light rail may increase ridership. it does so at a relatively high cost.
More recent studies of light-rail systems have provided additional evidence
that light rail is more expensive than a bus system. Richmond (1998) examines
light-rail systems in 1 1 Denver, Los Angeles.
cities (Baltimore. Buffalo. Dallas.
Pittsburgh, Portland, Sacramento. San Diego. San Jose, and St. Louis) and compares
their performance to bus systems. Here are some of his conclusions.
1 . Light rail has higher capital costs. For example, the capital cost of the
Long Beach light-rail system was $88 million, compared to the
1
$168 million in capital costs that would have been required for an
equivalent bus system (Richmond, 1996).
2. Light rail has higher operating costs. Most data reported on the operating
costs of light rail omits the costs of the feeder buses that bring riders to
light-rail stations. Ignoring these costs, the average operating cost for light
rail is higher or perhaps slightly lower than that for equivalent bus lines. For
example, the cost per passenger mile for Portland's MAX is $0.38,
compared to $0.32 for one equivalent bus line, and $0.39 for another. Once
the cost of feeder buses is included, light rail is more expensive than
equivalent bus lines.
3. Light rail takes passengers out of buses. For the Blue Line in Los
Angeles, 63 percent of riders were previously bus riders. In Portland, about
55 percent of MAX's riders switched from buses to light rail.
Light rail from many of the same problems experienced by heavy rail.
suffers
The system some other system to collect passengers, and this
requires feeder buses or
is expensive for transit authorities and bothersome to potential riders. In the modern
metropolitan area with dispersed employment and retail activities, it is difficult to

attract enough riders to make light rail less costly than a well-designed bus system. In
many cases, busways and other HOV systems would be less expensive and more —
effective in increasing transit ridership (Richmond, 1998). In other cases, simple
and inexpensive changes in regular bus service (adding buses, changing routes or
schedules, decreasing fares) may be more efficient than big projects.
Subsidies for Mass Transit

Taxpayers provide huge subsidies for mass transit service. In 1995, the total value
of the subsidies from federal, and local governments was $18.6 billion, with
state,
$11.4 billion in subsidies for operating costs and another $7.2 billion in capital
subsidies. These subsidies are necessary because the revenue from transit fares falls
short of operating and capital costs. Figure 20-5 shows the fare-box ratios (the
percentage of operating costs covered by fares) for three transit modes. Overall,
passenger fares covered only 38 percent of the operating costs (and none of the
capital costs). The remainder of operating costs were covered by local government
(22 percent), state governments (21 percent), the federal government (4 percent),
and other sources (15 percent). What is the rationale for these enormous subsidies?
Justification for Transit Subsidies
There are two justifications for the subsidization of mass transit. First, there are
substantial scale economies in the provision of transit, so transit systems have the
characteristics of a natural monopoly. As explained in Chapter 16 (Overview of
Local Government), the subsidization of a natural monopoly promotes efficiency.
Figure 20-6 shows the long-run average cost and marginal-cost curves for a transit
Figi re 20-5 Fare-Box Ratios for Public Transportation, 1995
7
3
:
a 60% 57
u
ft
— 49%
5096
-
-
u 38%
40°;
34%
s
?2%
u
r 307r
c
u
>
u
- 20%
-
10 r r
X
-
ncz.
Bus Heavy rail Light rail Commute rail .All public transit
SOURCE: American Public Transit Association. Transit Fact Book. (Washington D.C.. 1997).
Figure 20-6 Transit Subsidies for a Natural Monopoly

$
Demand = Marginal
social benefit
Long-run average cost
Lona-run marginal cost
Ridership
Mass transit is a natural monopoly: scale economies generate a negatively sloped long-run average cost curve. The
optimum output is where the demand curve (the marginal-benefit curve) intersects the marginal-cost curse. At
the optimum ndership R'i
|,the pnce P" in le^s than the average total cost, so the system generates a deficit. The
I
i
-•> stem would cover its cost at a price of P but ndership R would be below the optimum level.
. i i
600
system. The curves are negatively sloped because ( 1

) there is a substantial fixed cost
associated with setting up the system and (2) the marginal cost of adding another
rider is The optimum ridership (/?*) is where the marginal social
relatively small.
benefit (from the demand curve) equals the marginal cost. To generate this level of
ridership, the price must be P* which is less than the average total cost. The deficit
,
associated with marginal-cost pricing is shown by the shaded area. If the transit
system were forced to cover all its costs, it would charge a price of P' and generate
a ridership of /?', which is less than the optimum level.
The second justification of transit subsidies is explained in Chapter 19 (Autos
and Highways). Auto travel is underpriced because drivers do not pay the external
costs associated with highway congestion and air pollution. Therefore, transit is
overpriced relative to auto travel, and transit subsidies can be used to offset the
artificially low price of autos. As explained in Chapter 19, there are trade-offs with
transit subsidies: although the subsidies discourage auto use, decreasing congestion
and pollution, they also cause the underpricing of urban travel in general, leading
to an excessive amount of travel. The policy question is whether the benefit of
the subsidy (decreased congestion and air pollution) exceeds the cost (too many
resources used in transportation).
Reasons for Transit Deficits

The fare-box one measure of the fiscal health of mass transit. The lower
ratio is
between fare revenue and operating costs. Between
the ratio, the larger the shortfall
1960 and 1995, the fare-box ratio decreased from 102 percent to 38 percent. There
were several factors behind this dramatic decline in the fiscal health of mass transit
systems (Pickerell, 1983; Small, 1997).
1 . Low fares. Transit fares have been kept relatively low, in part to attract
riders. Because demand for transit services is price-inelastic, a decrease in
fares decreases total revenue.
2. Higher wages. Wages for transit workers have increased relatively rapidly
over the last few decades.
3. Increase in transit mileage. Transit service was extended into low-density
suburban areas, where ridership is relatively low.
4. Decrease in labor productivity. Between 1970 and 1995, total transit
employment
ridership increased slightly (by 6 percent) while total transit
more than doubled. Between 1980 and 1995, ridership dropped by 9 per-
cent but employment increased by 66 percent. The result, as shown in
—
Figure 20-7, is rapidly declining or more accurately, diving produc- —
tivity. Between 1970 and 1995, the average productivity (the number of
transit rides per employee) was cut in half.
Why number of passengers per employee decrease? According to

did the
Fielding (1987), an important factorwas an increase in the fraction of transit trips
taken during peak periods. To explain the peaking problem, consider a city where
Figure 20-7 Average Productivity of Transit Workers, 1970-1995

60.000
50.000 -
40.000
30.000
-
20,000
10.000
1970 1975 1980 1985 1990 1995
SOURCE: American Public Transit Association. Transit Fact Book. ( Washington DC 1997.)
transit ridership is uniform throughout an eight-hour workday: the city can serve its
transit riders with a single bus running all day. Suppose that all the riders suddenly
decide to travel to work during two morning rush hours: traffic volume during the
two rush hours quadruples. The city will no longer be able to serve its transit riders
with a single bus and a single bus driver, but will be forced to buy three more buses
and hire more drivers.
What are the effects of the shift to peak ridership on the city's labor productivity
and labor cost? Consider first the possibility that the city hires part-time drivers.
If the city hires drivers by the hour, productivity does not change: the city hires
four part-time drivers instead of one full-time driver. On the other hand, if the city
must hire four full-time drivers (who will be idle for six hours per day), its labor
productivity decreases and its labor cost rises. In other words, if the shift to peak-
period travel is combined with restrictive work rules that prevent part-time workers
and split shifts, labor productivity decreases and labor cost increases.
Policy Reform: Contracting and Paratransit

In most cities in the United States, urban transportation is heavily regulated. Firms
are prohibited from providing services that compete with the local transit authority,
so the transit authority has a monopoly. In addition, a taxi cannot serve as a common
carrier: the taxi can carry a small group of people traveling together, but cannot give
rides to strangers in transit. In other words, taxis cannot pick up additional passengers
en route to one passenger's destination and cannot pick up customers along routes
chosen by the driver.
What is the rationale for restricting entry in the market for urban transportation
services?
1 . Cross subsidization. The city regulates transit fares and routes, forcing the
transit authority to cover all routes, even ones that don't generate enough
revenue to cover operating costs. The fares on profitable routes are kept
relatively high to cover the losses on unprofitable routes. If entry were
allowed, a new firm could provide service along the profitable routes,
underpricing public transit and capturing most of its riders. The transit
authority would make less money on the lucrative routes, so there would be
less money to cover its losses on unprofitable routes.
2. Scale economies. As explained earlier in the chapter, there are substantial

fixed costs in providing masssome types of mass transit are
transit, so
natural monopolies and subsidies are required for efficiency. Although
fixed-rail systems are certainly natural monopolies, the case for bus systems
is not so clear-cut. Walters (1982) and Morlok and Viton (1985) suggest
that scale economies in bus transit are exhausted at ridership levels well
below the volume of the typical medium-sized city.
The problems experienced by public transit (declining ridership, decreasing

productivity, and growing deficits) suggest that alternative approaches should be
considered. We'll explore two reform proposals: private provision of transit services
through service contracts, and deregulation of urban transit to allow free entry into
the market.
Contracting for Transit Services
One option is for a city to sign contracts with private firms to provide specific
transit services. Under a contracting arrangement, the local government specifies the
service characteristics of the transit system (e.g., headways, travel times, location
of bus stops, fares) and then accepts bids from private firms for the transit service.
The transit monopoly is granted to the firm that provides the service at the lowest
cost.
How much money can be saved by contracting for transit services? The Fed-
eral Transit Administration estimates savings in operating costs between 25 and
30 percent, and other studies suggest savings between 15 and 35 percent (McCul-
lough, Taylor, and Wachs, 1997; Giuliano and Teal, 1987; Peskin, Mundle, and
Varma, 1992).
Lave (1985), Echols (1985), and Rosenbloom (1985) discuss various experi-
ences with contracting arrangements. The city of Tidewater, Virginia, contracted for
bus service to low-density areas. The private transit company provided the same
transit service at a fraction of the cost of public bus service. Along one route, the
deficit per passenger decreased from $4.75 to $1.63. Other cities using transit con-
tracts have had similar experiences: the typical private bus company provides transit
service for about half the cost of the public bus system. Contracts are also used
for subsidized dial-a-ride services: in 1983. the city of Phoenix paid a local taxi
company $104,425 to provide dial-a-ride services on $undays. about one-sixth of
the cost of using city buses for Sunday service.
Private firms provide transit services for a lower cost for three reasons. First,
they pay lower wages. In San Francisco, when BART accepted bids for feeder-bus
service to serve transit stations, the wage rates on all the private bids were below
$9.10, compared to a wage rate of $1 1.01 for the sole public bidder. AC Transit.
Second, they have more flexible work rules. They have split shifts and use part-
time workers, so they don"t pay idle workers during the off-peak periods. Third,
they use minibuses on low-density routes. In Phoenix, the private bus company uses
minibuses with an operating cost of $1.22 per mile, compared to $2.86 per mile for
the standard bus.
Deregulation and Paratransit
A second option for reform is to deregulate the urban transportation market by allow-
ing free entry and competition. Deregulation would change the mix of transportation
services available in most cities. The current system has two extremes solo-rider —
taxisand large transit vehicles (large buses and rail cars) —
and would be replaced
by a system that provides travelers with a wider variety of services in a variety of
vehicles.
is a term adopted in the 1970s to describe a w ide variety of services
Paratransit
that fallbetween the private automobile and the conventional bus. Some examples
are shared-ride taxis (3 to 4 passengers), dial-a-ride vans (6 to 10 passengers), jitneys
(6 to 15 passengers), and subscription commuter vans and buses 10 to 60 passen- (
gers). These paratransit services fill the gap between solo-rider taxis and large buses.
The experiences with paratransit suggest that these services could play a major role
in a deregulated transit market (Cervero. 1996).
Shared-Ride Taxis and Dial-a-Ride Vans. During World War II. shared-ride
taxis thrived in Washington. D.C.. Cab drivers displayed destination signs, allowing
people along the route to hail cabs going their way. Currently, shared-ride taxis are
legal for downtown trips in Los Angeles and New York City, and for airport trips in
Chicago. Denver, Seattle, and other cities. In Phoenix and Norfolk, low -volume bus
routes were replaced w ith shared-ride taxis, leading to substantial savings.
Dial-a-ride vans are similar to shared-ride taxis in the sense that both are on-
demand services. There are two main differences: dial-a-ride vehicles are larger, and
arrangements are made by phone instead of by curbside hailing. The most popular
and profitable dial-a-ride service is the airport shuttle. Airport shuttles serve people
who don't want to drive to the airport but want something between the two extremes
of taxi service and conventional transit (bus or rail).
Jitneys. A jitney is a vehicle with a capacity of between 6 and 15 passengers.

Although jitneys are similar to conventional buses operating along fixed routes, there
are several important differences. First, jitney service involves smaller vehicles, so
the cost per passenger is lower. Second, jitneys are more anywhere
flexible: they stop
along a route, don't follow a set schedule, and may deviatefrom their normal
slightly
routes. Third, jitneys have shorter headways (the interval of time between vehicles
plying a route) and usually stop less often. The shorter headways mean that jitneys
have lower collection costs.
Jitneys were popular in the United States in the early part of the twentieth century,
but were banned when trolley operators complained about unfair competition and
"cream skimming" (serving only high-volume, profitable routes and leaving low-
volume routes for public transit). The name comes from the fact that the fare for a
jitney ride was 5 cents, and the slang for nickel was "jitney." Most jitneys operated
only during rush hour, allowing the driver to work at another job during the off-peak
hours. Some jitneys operated as casual car pools: the driver posted his workplace on
the windshield and picked up commuters on his way to work. Other jitney drivers
started driving without a specific destination and then modified their routes as they
picked up passengers.
More recently, jitneys have been used in Miami, Atlantic City, New York City,
and along the California-Mexico border. Jitneys have lower costs than public buses
operating along similar routes, and in contrast with subsidized public buses, earn a
small profit (Cervero, 1996). For example, jitneys in San Ysidro, California, earn a
profit of $0.18 per passenger, while a public bus serving Route 932 had a deficit of
$0.18 per passenger. (Route 932's fare-box ratio of 82 percent makes it one of the
highest performing bus routes in the country). In Miami, jitneys generate a profit of
$0.39 per passenger, compared to a loss of $1.03 per passenger on public buses.
The two cities with the most extensive legal jitney services are Atlantic City
and Miami (Cervero, 1996). In Atlantic City, jitneys operate along Atlantic Avenue,
the city's main commercial strip, and serve 1 1 million passengers per year. Entry
is restricted, with the number of jitneys fixed at 190. Reflecting the substantial
profits that can be earned, the market value of a jitney medallion (license) and the
accompanying vehicle was $160,000 in 1995. In Miami, jitneys operate along 21
designated routes that are drawn to limit competition between jitneys and public
buses: no jitney route duplicates more than 30 percent of a bus route. The city of
Miami number of jitneys, a response to losses in revenue and
recently decreased the
ridership experienced by the public bus system. The unsubsidized jitneys took riders
away from the subsidized buses, so the city reduced the number of jitneys.
Subscription Commuter Services. In contrast with jitneys and shared-ride taxis,

travelers who use subscription services make regular trips to common destinations
(Cervero, 1996). This regularity allows the development of subscription bus ser-
vice: riders pay in advance to ride buses that travel between their neighborhoods
and workplaces. Most subscription bus services started out in the 1970s as private
operations, but have gradually been replaced by government programs. The appeal
of government involvement is that government grants cover a large fraction (up to

80 percent) of vehicle costs. Many transit agencies that operate subscription buses
contract the services out to private operators, which keeps costs down because the
firms pay lower wages to their workers.
Among the metropolitan areas with extensive subscription bus service are
Chicago, Houston, San Francisco, and New York. In San Francisco, Golden Gate
Transit established 22 commuter routes between suburban communities and the
downtown financial district, and contracts with firms to provide the services. In New
York City, private bus lines carry about 60,000 workers each day to jobs Manhat-
in
tan. The private operators have lower costs than public buses, (e.g., about 30 percent
lower in New York).
Summary of Paratransit. Paratransit has the potential to fill the gap in the urban
transportation system between solo-rider taxis and large public buses. In contrast
with subsidized public buses, paratransit operations actually earn profits. In addition,
paratransit does not simply siphon riders from other mass transit modes, but also
causes some travelers to switch from single-occupancy vehicles. In summarizing the
prospects for paratransit, Cervero (1996) notes:
Given the fiscal cutbacks facing America's public transit industry today, the expansion of more
entrepreneurial, commercial transportation services seems unavoidable. While critics charge
that the poor will suffer as a result, other remedies — like user-side subsidies — are available
for redressing inequities. Moreover the history of commercial paratransit is certainly not one
of ignoring poor neighborhoods. For jitneys and neighborhood car services, low-income areas
have traditionally been their market base.
British Experience with Deregulation
In Great Britain, the transit industry was deregulated under the British Transport Act
of 1985. As explained by Gomez-Ibanez and Meyer (1990), the act relaxed controls
on entry most public transit authorities as for-
into the transit industry, reorganized
profit organizations,and introduced competitive bidding for certain transit services.
Local governments can continue to subsidize transit services as long as the subsidized
service is provided by the low bidder in a competitive auction. Although the London
area was not covered under the Transport Act, contracting out is used extensively in
the metropolitan area, and full deregulation is expected sometime in the 1990s.
Britain's new combined deregulation with substantial cuts in tran-
transit policy
sit were cut by about 23 percent in the
subsidies. For example, transit subsidies
metropolitan counties and about 6 percent in other counties. The decrease in transit
subsidies generated pressure to increase fares and cut services. Gomez-Ibanez and
Meyer (1990) and Small (1992) summarize some of the results of the new transit
policy:
1 . Service. In the first full year following deregulation, total mileage was up
by 3.3 percent in the metropolitan counties and 16.5 percent in other
counties. The use of minibuses (with 12 to 25 seats) increased dramatically.
Some unprofitable services were curtailed: for example, some off-peak and
low-density routes were abandoned.
2. Fares. Transit fares increased by about 35 percent in real terms over a
two-year period.
3. Ridership. Ridership decreased by about 14 percent.
4. Production cost. Production cost fell as a result of labor concessions
(relaxed work rules and lower compensation), the introduction of
minibuses, and the elimination of excess capacity. According to Heseltine
and Silcock (1990), unit cost fell by between 15 percent and 30 percent.
Gomez-Ibanez and Meyer discuss three lessons from the British experience
with the deregulation of transit. First, it is possible to have both competition in
the local bus industry and subsidies for unprofitable services. By using competitive
bidding to pick a firm to provide the subsidized services, the public sector can
provide the services at the lowest possible cost. Second, in addition to cutting cost
and increasing productivity, deregulation generates service innovations: the private

bus companies introduced minibuses, and local governments improved the services
along unprofitable routes. Third, because most of the benefits from deregulation
come from competition among transit firms, the public sector must develop policies
to ensure competition.
Transportation and Land-Use Patterns

Part II of this book (Land Rent and Urban Land-Use Patterns) discusses the effects of
changes in transportation technology on urban land-use patterns. The development
of the internal-combustion engine allowed the development of new travel modes (the
car and the truck) that were more flexible and less costly than streetcars and horse-
drawn wagons. The new modes decreased the reliance of firms and households on the
central area of the city, allowing the decentralization of employment and residents.
This section discusses two types of changes in transportation: an increase in gasoline
cost and the construction of fixed-rail transit systems.
Gasoline Prices
Suppose that the real cost of gasoline increases. How will the increased cost of auto
travel affect housing and land prices? As explained in Part II, increase in trans-
portation cost increases the slopes of the housing-price function and the residential
bid-rent function. Studies by Small (1986) and Coulson and Engle (1987) confirm
these theoretical predictions: an increase in the price of gasoline increases the rel-
ative attractiveness of locations near employment centers, increasing the prices of
housing and land.
How will an increase in the cost of gasoline affect the distributions of employ-
ment and population? The changes are likely to be relatively small for three reasons.
First, the cost of gasoline is a relatively small part of the cost of travel. Second,
other responses to an increase in gasoline prices (buying more fuel-efficient cars,

carpooling, switching to public transit) allow drivers to decrease travel cost without
changing residences or workplaces. For these two reasons, the changes in the prices
of housing and land are likely to be relatively small. Even if the changes in price
were large, itwould take a long time before there would be changes in population
density near employment centers because structures last a long time and are costly
to demolish. Small (1985) estimated that a $1 per gallon increase in the price of
gasoline would cause a relatively small number of suburban households to move to
the central city.
Fixed-Rail Systems and Land Use

One of the objectives of San Francisco's BART was to increase employment op-
portunities in the areas near transit stations. An early study of BART (Dyett and
Escudero, 1977) suggests that BART had a moderate positive effect on employment
in downtown San Francisco, but not much of an effect elsewhere. Similarly, Web-
ber (1976) concludes that BART increased downtown employment but didn*t affect
employment near suburban stations. Webber suggests that office development in the
downtown area would have been about same if transit authorities had expanded bus
service instead of building BART. More recently, Cervero and Landis (1995) show
that BART increased employment in areas near transit stations, with most of the
growth in the downtown area.
A recent study of Atlanta's MARTA suggests that the heavy-rail system did not
increase employment in areas near transit stations (Bollinger and Ihlanfeldt. 1997).
Although public employment in these areas grew, the gains in public employment
were offset by losses in private-sector employment. This result reflects the policy of
shifting government employment to areas along the rail line.
There are two main reasons why fixed-rail systems have such small effects on
employment. First, the typical new system does not significantly increase the ac-
cessibility of sites near station areas because those sites were already accessible by
private automobile. In contrast, streetcars in the late 1800s made areas along the
spokes of the hub-and-spoke system accessible to urban workers, causing dramatic
changes in land use. Similarly, the building of suburban highways in the 1950s and
1960s made suburban areas much more accessible, resulting in fundamental changes
in land-use patterns. The second reason for the small land-use effects of fixed-rail
systems is that buildings in urban areas are durable, lasting at least 50 years. Even
if a new transit line increased the accessibility of a particular area enough to make
high-density employment sensible, the transition from low-density to high-density
would probably occur slowly.
In summarizing the results from several studies of the effects of rail systems on
land-use patterns. Altshuler (1979) makes two observations:
1 . In a growing economy, rail transit contributes to the clustering of activities

near downtown stations. These clustering effects are usually negligible
outside the central business district.
2. Investment in rail transit is sensible only if it is used in concert with more

powerful land-use instruments such as zoning and property taxation. If the
government uses its zoning and tax policies to generate high-density

development, rail transit provides an efficient means of delivering a large
number of workers to the dense central area.
In general, there is evidence that fixed-rail systems cause small changes in the spatial
distribution of employment. The primary effect of transit is an increase in downtown
employment.
Summary
1. Mass transit is most popular among central-city residents.
2. Transit ridership is relatively unresponsive to changes in monetary cost (fares),
but responsive to changes in time cost, especially waiting and walking costs.
On average, commuters value in-vehicle time at about half their wage rate and
collection and distribution time at between 1.0 and 1.5 times the wage rate.
3. The time is the sum of collection cost, in-vehicle cost, and
cost of a trip
Commuters choose the mode that minimizes the sum of
distribution cost.
monetary and time costs. Mass transit has a lower monetary cost than autos,
but higher collection and distribution time costs.
4. There are trade-offs associated with bus service: a decrease in the headway or
an increase in the frequency of stops decreases collection and distribution
costs, but increases line-haul cost.
5. There are two basic trade-offs in the design of a fixed-rail system.

a. A mainline system forces commuters to use other travel modes in the
collection and distribution phases of the trip. An alternative is an
integrated system, which allows commuters to take the entire trip on
a single mode.
b. A system with widely spaced stations has shorter line-haul times but higher
collection and distribution costs.
c. BART is a mainline system with widely spaced stations. Because its
collection and distribution costs are relatively high, it caused a relatively

small number of commuters to switch from autos to transit.
6. Diamond lanes cause a moderate number of commuters to switch from autos

to buses. The decrease in the number of auto lanes increases congestion,
increasing the time cost of auto commuters.
7. A study of alternative transport systems for the San Francisco Bay area
generated the following conclusions.
a. For all corridor volumes studied, BART was more costly than an integrated
bus system.
b. For corridor volumes exceeding 1,100 vehicles per hour, an integrated bus
system was less costly than an auto-based system.
8. In 1995, passenger fares covered only about 38 percent of the operating cost of
transit. Transit subsidies can be justified on efficiency grounds because transit
is a natural monopoly and automobile travel is underpriced. Transit deficits
have increased over the last few decades, a result of decreases in fares and
productivity and increases in wages, energy cost, and transit mileage.
9. Deregulation of transit could take two forms: the private provision of transit
through service contracts and free entry into the transit market. Contracting
arrangements have decreased the cost of providing transit services in a number
of cities. Free entry could involve private buses, jitneys, and taxis serving as
common carriers.
10. Increases in gasoline prices change housing and land prices, but do not cause
large changes in location patterns.
11. Fixed-rail systems cause relatively small changes in location patterns. If the
government uses its zoning and tax policies to generate high-density

development, rail transit can provide an efficient means of delivering workers
to the high-density areas.

1. Fill in the blanks in the following table.
Auto Bus BART
Collection Time Cost

Collection lime (minutes) 12 16
Cost per minute ($) 0.20 0.20 0.20
Collection time cost ($)
Line-Haul Time Cost

Line-haul time (minutes) 30 40 20
Cost per minute ($) 0.15 0.15 0.15
Line-haul time cost ($)
Distribution Time Cost

Distribution time (minutes) 4 11
Cost per minute $ 1

i 0.20 0.20 0.20
Distribution time cost ($)
Monetary Cost
Operating cost or fare ($) 3 1.5 2
Parking cost ($) 4
Total monetary cost ($)

Total time cost ($)
Total cost ($)
2. Buster has proposed that the distance between bus stops along a certain route
be shortened to decrease the walking distance for the typical rider from 10
Chapter 20 Maw Transit 611
blocks to 8 blocks. The decrease in bus-stop distance will increase line-haul

time from 50 to 55 minutes and increase the total operating cost of the route by
10 percent.
a. Given the available evidence concerning the size of the relevant elasticities,
is ridership likely to increase or decrease?
b. Suppose that the transit authority passes on the higher operating cost in the
form of an increased fare. Given the available evidence concerning the
relevant elasticities, is ridership likely to increase or decrease?
3. Marta the transportation planner recently proposed that bus headways (the
time between buses) along a high-density route be shortened from 20 minutes
to 10 minutes. Your mission is to estimate the effects of such a policy on bus
ridership. Assume that any increase in operating costs will be passed on to
riders in the form of a higher fare.
a. What information would you collect, and how would you use it?
b. Provide a numerical example of a situation in which ridership will increase.
4. Figure 20-4 shows average cost curves for three types of travel systems: an
auto system, an integrated bus system, and a heavy-rail system.
a. Contrast the auto curve in Figure 20—4 to the cost curves in Figure 19-2.
Why is the cost curve in 20-4 flat while the cost curves in 19-2 are
positively sloped?
b. Suppose that public opposition prevents the expansion of the highway
network. Draw a new pair of cost curves for Figure 20^4, one for the auto
system and one for the BART system. Explain any differences between the
new curves and the old ones.
5. The cost curves in Figure 20-4 were derived by Keeler et al. (1975) under a
number of assumptions about the disutilities of walking, waiting, and
in-vehicle time. Draw a new set of curves for the following changes:
a. Decrease in the disutility of walking and waiting time.
b. Increase in the disutility of line-haul time.
6. In drawing the cost curves in Figure 20-4, Keeler et al. assumed that the
disutility of time spent riding a bus is the same as the disutility of time
driving a car.
a. Based on your own preferences, is this a realistic assumption?
b. Draw a new set of cost curves consistent with your preferences.
7. In drawing the cost curves in Figure 20-4, Keeler et al. assumed that the
interest rate was 6 percent. Draw a new set of cost curves under the assumption
that the interest rate is 12 percent.
8. Use a graph like Figure 20-4 to show the long-run average cost of a light-rail
system. Draw the light-rail curve along with the curves for BART and a bus
system. Explain your placement of the light-rail curve.
9. Zirconium City just converted one of the four lanes on its freeways to a
diamond lane. The conversion shortened the line-haul time of buses by 10
minutes and increased the line-haul time of autos by 3 minutes. Consider the
responses of two commuters: Maeve switched from driving to taking the bus,
but Cormac continues to drive. Explain these different responses to the

changes in travel times caused by the diamond lane. How does Maeve differ
from Cormac?
10. In Weather City, the winters are cold and the summers are hot and humid. In
Mild City, the winters and summers are mild. Except for the differences in
weather, the two cities are similar. Would you expect the distance between bus
stops to be longer or shorter in Mild City?
11. The city of Congestville is examining alternative policies to deal with its
congestion problem. One of the options is free rush-hour bus service. Depict
graphically the effects of free bus service on auto congestion and auto trip
costs.
12. Consider the effects of dropping a planner's bomb on the San Francisco Bay
area. A planner's bomb doesn't hurt any people, but destroys everything except
the BART infrastructure (tracks, vehicles, and station). Most important, it
destroys all buildings, so the metropolitan area must be completely rebuilt.
Design a set of public policies that will ensure that in the rebuilt Bay area,
BART ridership will be high enough that BART is as efficient as an integrated
bus system.
13. In the city of Phoenix, private companies provide bus services on low-volume
routes at a fraction of the cost of the public bus agency. In New York City and
Chicago, private companies provide bus service on high-volume (peak-period)
routes at a fraction of the cost of the public bus agency. If private firms can
underprice the public sector on high-volume and low-volume routes, what, if
anything, should be the role of the public sector in the provision of transit
services?
14. Suppose that the purpose of cross-subsidization is to increase the welfare of

people along low-volume routes. Discuss some alternative means of improving
their welfare.

General
Altshuler,Alan A. The Urban Transportation System. Cambridge. Mass.: Joint Center for
Urban Studies of MIT and Harvard, 1979.
American Public Transit Association. Transit Fact Book. Washington, D.C., 1991.
Cervero, R. "Efficiency and Equity Impacts of Current Transit Fare Policies."
Transportation Research Record 790 (1981). pp. 7-15.
Cherwony, W.. and S. R. Mundle. "Peak-Base Cost Allocation Models." Transportation
Research Record 663 (1978), pp. 52-56.
Dahlgren, Joy. "Are HOV Lanes Really Better?" Access 6 (Spring 1995). pp. 25-9.
Dewees, Donald. "Travel Cost. Transit, and Urban Motoring." Public Policy 24 (Winter
1976), pp. 59-79.
Chapter 20 Muss Transit 613
Giuliano, Genevieve, and Kenneth A. Small. "Alternative Strategies for Coping with Traffic
Congestion." University of California Transportation Center, Working Paper 188.
Berkeley: University of California Transportation Center. 1994.
Kemp, Michael A., and Melvyn D. Cheslow. "Transportation." In The Urban Predicament,
ed. William Gorham and Nathan Glazer. Washington, D.C.: Urban Institute, 1976,
pp. 321-24.
Lave, Charles A. "The Demand for Urban Mass Transportation." Review of Economics and
Statistics 52 (1970), pp. 320-23.
. "Transportation and Energy: Some Current Myths." Policy Analysis 4 (Summer
1978), pp. 297-315.
Meyer, John R., and Jose A. Gomez-Ibanez. Autos, Transit and Cities. Cambridge, Mass.:
Harvard University Press, 1981.
Meyer, John; John Kain; and Martin Wohl. The Urban Transportation Problem. Cambridge,
Mass.: Harvard University Press, 1965.
Mohring, Herbert. "The Benefits of Reserved Bus Lanes, Mass Transit Subsidies, and
Marginal Cost Pricing in Alleviating Traffic Congestion." In Current Issues in Urban
Economics, ed. P. Mieszkowski and M. Straszheim. Baltimore: Johns Hopkins
University Press, 1979.
. Transportation Economics. Cambridge, Mass.: Ballinger, 1976.
—
Richmond, Jonathan E. "New Rail Transit Investments A Review." Taubman Center for
State and Local Government, 1998.
Small, Kenneth A. "Economics and Urban Transportation Policy in the United States."
Regional Science and Urban Economics 27 (1997), pp. 671-91.
. "Urban Traffic Congestion: A New Approach to the Gordian Knot." Brookings
Review 11 (1993), pp. 6-1 1.
. Urban Transportation Economics. Philadelphia: Harwood, 1992. A comprehensive
summary and review of the economics of urban transportation.
U.S. Bureau of the Census. 1980 Census of Population: Journey to Work Characteristics—
of Workers. Washington, D.C.: U.S. Government Printing Office, 1984.
Buses and Scale Economies

Mohring, Herbert. "Minibuses in Urban Transportation." Journal of Urban Economics 14
(1983), pp. 293-317.
."Optimization and Scale Economies in Urban Bus Transportation." American
Economic Review 62 (1972), pp. 591-604.
Morlok, Edward K., and Philip A. Viton. "The Comparative Costs of Public and Private
Providers of Mass Transit." Chapter 10 in Urban Transit: The Private Challenge to
Public Transportation, ed. Charles Lave. Cambridge, Mass.: Ballinger, 1985;
pp. 233-50.
Walters, A. A. "Externalities in Urban Buses." Journal of Urban Economics 1 1 (1982).
Discusses the advantages of small buses.
Demand Elasticities
Beesley, Michael E. "The Value of Time Spent Traveling: Some New Evidence." In The
Demand for Travel: Theory and Measurement, ed. Richard Quandt. Lexington, Mass.
D. C. Heath, 1970.
614 Pun V Urban Transportation
Kemp. "Urban Transportation." Chapter 26 in

Beesley. Michael E.. and Michael A.
Handbook of Regional and Urban Economics. Vol. 2 of Urban Economics, ed. Edwin
S. Mills. Amsterdam: North Holland. 1987.
Domencich. T.: G. Kraft: and J. Valette. "Estimation of Urban Passenger Travel Behavior."
In Readings in Urban Economics, ed. M. Edel and J. Rothenberg. New York:
Macmillan. 1972.
Domencich. Thomas, and Daniel McFadden. Urban Travel Demand. Amsterdam: North
Holland. 1975.
Kemp. Michael A. "Some Evidence of Transit Demand Elasticities." Transportation 2
(1973).
Kraft. Gerald, and Thomas Domencich. "Free Transit." In Readings in Urban Economics.
ed.Matthew Edel and Jerome Rothenberg. New York: Macmillan. 1972. pp. 459-80.
Lago. Armando M.: Patrick Mayworm; and J. Matthew McEnroe. "Transit Sen ice
Elasticities." Journal of Transport Economics and Policy 15 1981 ). pp. 99-1 19.
(
Meyer, John, and Mahlon R. Straszheim. Pricing and Project Evaluation. Vol. of 1
Techniques of Transport Planning. Washington. D.C.: Brookings Institution. 1971.

Chapter 7.
Improving Mass Transit

Fielding. Gordon J. Managing Public Transit Strategically: A Comprehensive Approach to
Strengthening Service and Monitoring Performance. San Francisco: Jossey-Bass. 1987.
Hamer. Andrew Marshall. The Selling of Rail Rapid Transit. Lexington. Mass.: D. C. Heath.
1976. pp. 61-67. 75-77.
Kain. John F. "How to Improve Urban Transportation at Practically No Cost." Public Policy
20 (Summer 1972). pp. 335-58.

.
"Toward Better Urban Transport Pricing: Using What Is Known and Exploring What
Is Not Known." In Urban Transportation Pricing Alternatives. Washington. D.C.:
Transportation Research Board. 1976.

Meyer, John R.. and Jose A. Gomez-Ibanez. Improving Urban Mass Transportation
Productivity. Washington. D.C.: U.S. Department of Transportation. Urban Mass
Transportation Administration. 1977.
Webber. Melvin W. "The BART Experience — What Have We Learned?" The Public
Interest. Fall 1976. pp. 79-108.
Transit Deficits
Hilton. George. Federal Transit Subsidies. Washington. D.C.: American Enterprise Institute
tor Public Policy Research. 1974.
Pickerell, Don H. "The Causes of Rising Transit Operating Deficits." Transportation
Research Record 915 (1983), pp. 18-24.
Pucher. John; Anders Markstedt; and Ira Hirshman. "Impacts of Subsidies on the Costs of
Urban Public Transport." Journal of Transport Economics and Policy 17 (1983).
pp. 155-76.
Comparative Costs: Auto versus Bus versus Rail

Keeler. Theodore E.: L. Merewitz: P. Fisher: and K. Small. The Full Costs of Urban
Transport. Monograph 21, part 3. Institute of Urban and Regional Development.
University of California, Berkeley, 1975. Estimates the costs of alternative travel

modes.
Straszheim, M. "Assessing the Social Costs of Urban Transportation Technologies."
Chapter 7 in Current Issues in Urban Economics, ed. Peter Mieszkowski and Mahlon
Straszheim. Baltimore: Johns Hopkins University Press, 1979, pp. 196-232.
Public Transit
Glaister, Stephen. "Competition on an Urban Bus Route." Journal of Transport Economics

and Policy 9 1 ( 1 985), pp. 65-8 1
Gomez-Ibanez, Jose A. "A Dark Side to Light Rail?" Journal of the American Planning
Association 5 1 (Summer 1985), pp. 337-5 1 . Compares the cost of new light-rail
systems to the cost of conventional bus systems.
Gordon, Peter, and Harry Richardson. "Notes from Underground: The Failure of Urban
Mass Transit." The Public Interest 94 (1989), pp. 77-86.
George W. "The Rise and Fall of Monopolized Transit." Chapter 2 in Urban Transit:
Hilton,
The Private Challenge to Public Transportation, ed. Charles Lave. Cambridge, Mass.:
Ballinger, 1985.
Kemp, Michael A., "Government Policies Affecting Competition in
and Ronald F. Kirby.
Public Transportation." Chapter 12 in Urban Transit: The Private Challenge to Public-
Transportation, ed. Charles Lave. Cambridge, Mass.: Ballinger, 1985.
Savas, E. S. "Municipal Monopolies vs. Competition in Delivering Public Services." Urban
Analysis 2 (1974), pp. 93-1 16.
Wilson, James Q. "The Dead Hand of Regulation." The Public Interest 49 (Fall 1977),
pp. 39-58.
Public versus Private Transit
Cervero, Robert. "Commercial Paratransit in the United States: Service Options, Markets,
and Performance." Working Paper no. 299, University of California Transportation
Center, 1996.
Echols, James C. "Use of Private Companies to Provide Public Transportation Services in
Tidewater, Virginia." Chapter 4 in Urban Transit: The Private Challenge to Public
Giuliano, Genevieve, and Roger F. Teal. "Estimating the Potential Cost Savings of Transit
Service Contracting." Transportation Research Record 1 108 (1987), pp. 1-11.
. "Privately Provided Commuter Bus Services: Experiences, Problems, and Prospects."
Chapter 7 in Urban Transit: The Private Challenge to Public Transportation, ed.
Charles Lave. Cambridge, Mass.: Ballinger, 1985.
Gomez-Ibanez, Jose A., and John R. Meyer. "Privatizing and Deregulating Local Public
Services: Lessons from Britain's Buses." Journal of the American Planning
Association 56 (Winter 1990), pp. 9-21. Discusses Britain's experience with the
deregulation of the bus industry, focusing on the effects of deregulation on service,
fares, ridership, and production cost.
Heseltine, M., and D. T. Silcock. "The Effects of Bus Deregulation on Costs." Journal of
P.
Transportation Economics and Policy 24 (1990), pp. 239-54. Discusses the effects of
the deregulation of the British bus industry on production cost.
Johnson. Christine M.. and Milton Pilarsky. "Toward Fragmentation: Transportation in

Chicago." Chapter 3 in Urban Transit: The Private Challenge to Public
Lave, Charles A. "The Private Challenge to Public Transportation —An Overview." Chapter
1 in Urban Transit: The Private Challenge to Public Transportation, ed. Charles Lave.
Cambridge, Mass.: Ballinger, 1985.
McCullough. William S.; Brian D. Taylor: and Martin Wachs. "Does Contracting Transit
Service Save Money?" Access 1 (Fall. 1997), pp. 22-6.
1
Morlok. Edward K.. and Philip A. Viton. "The Comparative Costs of Public and Private
Providers of Mass Transit." Chapter 10 in Urban Transit: The Private Challenge to
Public Transportation, ed. Charles Lave. Cambridge. Mass.: Ballinger. 1985.
. in Large U.S.
"Recent Experience with Successful Private Transit Cities." Chapter 6
in Urban The Private Challenge to Public Transportation,
Transit: ed. Charles Lave.
Cambridge. Mass.: Ballinger, 1985, pp. 121-48.
Pagano, Anthony M. "Private Sector Alternatives for Public Transportation."
Transportation Quarterly 38 (1984), pp. 433^47.
Peskin. R.: S. Mundle: and P. K. Varma. "Transit Privatization in Denver: Experience in the
Second Year." Transportation Research Record 1402 (1992), pp. 17-24.

Rosenbloom. Sandra. "The Taxi in the Urban Transport System." Chapter 8 in Urban
Transit: The Private Challenge to Public Transportation, ed. Charles Lave.
Cambridge, Mass.: Ballinger, 1985.
Simpson. Anthony U. "Implications of Efficiency Incentives on Use of Private Sector
Contracting by the Public Transit Industry ." Chapter 13 in Urban Transit: The Private
Challenge to Public Transportation, ed. Charles Lave. Cambridge. Mass.: Ballinger.
1985.
Walder, Jay H. "Private Commuter Vans in New York." Chapter 5 in Urban Transit: The
Private Challenge to Public Transportation, ed. Charles Lave. Cambridge. Mass.:
Ballinger. 1985, pp. 101-18.
Taxis and Jitneys

Eckert. Ross, and George Hilton. "The Jitneys." Journal of Law and Economics 15 (1972).
pp. 293-325.
Kirby. Ronald F. Paratransit: Neglected Options for Future Mobility. Washington. D.C.:
Urban Institute. 1974.
Rosenbloom. Sandra. "Taxis. Jitneys, and Poverty." Transaction 7 (1970), pp. 47-54.
Wohl, Martin. "The Taxi's Role in Urban America: Today and Tomorrow." Transportation 4
(1975).
Transportation Systems and Land Use

Anas. Alex, and Leon N. Moses. "Transportation and Land Use in the Mature Metropolis."
In The Mature Metropolis, ed. C. L. Levin. Lexington, Mass.: D. C. Heath. 1978,
p. 161.
"The Impact of Rapid Rail Transit on
Bollinger. Christopher, and Keith Ihlanfeldt.
Economic Development: The Case of Atlanta's MARTA." Journal of Urban
Economics 42 (1997), pp. 179-204.
Cervero, Robert, and John Landis. "The Transportation-Land Use Connection Still
Matters." Access 7 (Fall, 1995). pp. 2-10.
Coulson, N. Ed. and Robert F. Engle. "Transportation Costs and the Rent Gradient."
Journal of Urban Economics 21 1987), pp. 287-97.
(
De Leuw, Cather, and Company. "Land Use Impacts of Rapid Transit: Implications of
Recent Experience." Washington. D.C.: U.S. Department of Transportation. August
1977.
Dewees, D. N. "The Effect of a Subway on Residential Property Values in Toronto."
Journal of Urban Economics 3 (1976), pp. 357-69. Estimates the effects of Toronto's
subway on property values along the transit line and around the transit stations.
Downs, Anthony. Stuck in Traffic: Coping with Peak-Hour Traffic Congestion. Washington,
D.C.: The Brookings Institution, 1992.
Dyett, Michael V., and Emilio Escudero. "Effects of BART on Urban Development." 1977.
Mimeo.
Giuliano. Genevieve. "The Weakening Transportation-Land Use Connection." Access 6
(Spring, 1995), pp. 3-11.
Goldstein, Gerald, and Leon Moses. "Transport Controls, Travel Costs, and Urban Spatial
Structure." Public Policy 23 (Summer 1975). pp. 355-80.
Heenan, Warren G. "The Economic Effect of Rapid Transit on Real Estate Development."
Appraisal Journal 36 (April 1968), pp. 213-24.
Hilton, George W. "Rail Transit and the Pattern of Modern Cities: The California Case."
TrafficQuarterly 2] (July 1967), pp. 379-93.
Moore, Terry, and Paul Thorsnes. The Transportation/Land Use Connection. Chicago:
American Planning Association, 1994. Explores the effects of public policy on
transportation systems and land-use patterns.
Small, Kenneth. "The Effect of the 1979 Gasoline Shortages on Philadelphia Housing
Prices." Journal of Urban Economics 19(1 986). pp. 371-81.
. "Transportation and Urban Change." In The New Urban Reality, ed. P. Peterson.
Washington, D.C.: Brookings Institution. 1985. pp. 197-223.
Smerk, George M. "The Streetcar: Shaper of American Cities." Traffic Quarterly 21
(October 1967). pp. 569-84.
Warner, Samuel B., Jr. Streetcar Suburbs: The Process of Growth in Boston, 1870-1900.
Cambridge, Mass.: Harvard University Press, 1932.
Webber. Melvin W. "The BART Experience —What Have We Learned?" The Public
Interest, Fall 1976. pp. 79-108.
Future Transit Systems

Anderson, Edward J. Personal Rapid Transit. Minneapolis: University of Minnesota,
Department of Audio Visual Extension, 1972.

. Personal Rapid Transit II. Minneapolis: University of Minnesota. Department of
Audio Visual Extension, 1974.
Fisher, Ronald J. "'Megatrends' in Urban Transport." Transportation Quarterly 38 (1984).
Kain, John F. "The Unexplored Potential of Freeway Rapid Transit in Regional
Transportation Planning: An Atlanta Case Study." In Unorthodox Approaches to
Urban Transportation: The Emerging Challenge to Conventional Planning, ed.
Andrew Hamer. Atlanta: Georgia State University, Bureau of Business and Economic
Research, 1972, pp. 38-51.
Skinner. Robert E., Jr., and Thomas B. Deen. "Second Generation U.S. Rail Transit
Systems: Prospects and Perils." Transportation 8 (1980).
Benefit-Cost Analysis
Beesley, M. E. Urban Transport: Studies in Economic Policy. London: Butterworths. 1973.

Foster. C. D.. and M. E. Beesley. "Estimating the Social Benefit of Constructing an
Underground Railway in London." Journal of the Royal Statistical Society 126 (1963).
pp. 46^92.
Meyer, John, and Mahlon R. Straszheim. Pricing and Project Evaluation. Volume 1 of
Techniques of Transport Planning. Washington. D.C.: Brookings Institution, 1971.
Chapters 11 and 12.
Peterson. Thomas C. "Cost-Benefit Analysis for Evaluating Transportation Proposals: Los

Angeles Case Study." In Urban Economic Issues, ed. Stephan Mehay and Geoffrey
Nunn. Glenview. 111.: Scott. Foresman. 1984. Critique of a cost-benefit study that was
used to justify the construction of a fixed-rail transit system in Los Angeles.
Part VI
/EDUCATION and CRIME
(~~7~ he last part of the book explores two urban policy challenges. Chapter 21
K-S considers some of the spatial aspects of education, showing how a frag-
mented system of local government causes inequalities in educational spending and
achievement. It also explores some of the policy responses to these inequalities,
including intergovernmental grants and desegregation. Chapter 22 presents the eco-
nomic view of crime and crime fighting. The model of the rational criminal provides
some important insights into the reasons for crime and also provides a framework
for evaluating the merits of alternative crime-fighting policies.
(lifter 21
8 D U C AT 1 N
Human history becomes more and more a race between education and
catastrophe.
H. G. Wells
his chapter explores several issues concerning elementary and secondary

(^J7~
t_y education. One issue concerns the connection between educational spending
and educational outcomes. In the last few decades, real spending on K-12 education
has increased, but student performance on standard tests has stayed at the same level
or dropped. Although the quantity of educational inputs has increased, educational
output, at least as measured by test scores, has not. The production function for
education differs from the production function for other goods and services, and
this difference helps explain why test scores haven't increased with spending on
education.
A second set of issues concerns the spatial aspects of education. Given the large
number of school districts in most metropolitan areas, households can choose from a
wide variety of schools, each of which provides a different combination of teaching
philosophies, educational quality, and tax costs. As explained in the chapter on
voting, the Tiebout process causes households to sort themselves geographically with
respect to desired spending on local public goods, property values, and income. This
chapter explains how this sorting process contributes to inequalities in educational
spending and achievement, and discusses the effects of various public policies on
spending and achievement inequalities.
What's special about education? Spending inequalities occur for other local
public goods: poor municipalities spend less on parks, recreation, and fire protection.
Education is special because a good education is necessary for economic survival. A
person with an inferior education is unlikely to find a well-paying job and is therefore
more likely to be poor and more likely to turn to crime. One approach to solving the
problems of poverty and crime is to improve the educational opportunities for poor
children in the central city.
621
622 Part VI Education anil Crime
The chapter is organized as follows. The first section discusses the education
production function, which is used to estimate the relative importance of various
inputs to the education process. The most important factors in educational achieve-
ment are the home environment of the student and the intelligence and motivation
of his fellow students. The second section explains the effects of the Tiebout sorting
process on the educational system, showing how the sorting process causes inequal-
ities in educational spending, racial and income segregation, and inequalities in
educational achievement. The third section explores the issue of inequalities in ed-
ucation spending across school districts, documenting the extent of inequalities and
evaluating various public policies designed to diminish the inequalities. The fourth
section discusses racial segregation in the schools and the choice between public
and private schools. The final section explores the effects of education vouchers on
the educational system.
The Education Production Function

Figure 21-1 shows the time trends in education spending and class sizes. Between
1960 and 1990, real spending per pupil nearly tripled, reaching $5,582 (in 1992
dollars) in 1990. Teacher salaries comprise a large fraction of the education budget,
so an important reason for the dramatic increase in real spending was smaller class
sizes. Over this period, the average pupihteacher ratio dropped from 25.6 to 17.3.
Other changes over this period included increases in the percentage of teachers with
master's degrees (from 23 percent to 53 percent) and increases in the average amount
of teaching experience (from 1 1 years to 15 years).
During this period of rising expenditures and shrinking class sizes, educational
performance was at best constant and may have fallen (Hanushek. 1996). Between
1970 and 1990, scores on standardized science tests given to 17-year-old students
dropped, while scores on reading and math increased slightly. On standardized tests
taken by students in 39 countries, American students are below average in most
subjects. To explain the puzzling combination of rising expenditures (inputs) and
stagnant performance (output), we must take a careful look at the education produc-
tion function.
The production function summarizes the relationship between the inputs to the
educational process and the output (achievement ). The production-function approach
can be used to estimate the relative importance of the various inputs to the education
process.
Inputs and Outputs of the School

What is the output of a school? The purpose of education is to develop cognitive,
social, and physical skills. The basic cognitive skills (reading, writing, mathematics,
logic) are necessary for employment and participation in a democracy. Cognitive
Chapter 21 Education 623
Figure 21-1 Trends in Class Size and Spending Per Student
$5,582
$4,116
$3,269
I 30.0
U 25.6
22.3
$1,903
20.0 17.3
10.0
Class Spending Class Spending Class Spending Class Spending

size per pupil size per pupil size per pupil size per pupil
1960 1970 1980 1990

SOURCE: Hanushek, Eric A. "Measuring Investment in Education." Journal of Economic Perspectives 10, no. 4 (Fall 1996),
pp. 9-30.
skills also increase the enjoyment of leisure activities: they allow people to read
books, understand jokes, and compute bowling scores. Schools also develop so-
cial skills: they teach children how to exchange ideas and make group decisions.
Finally, schools develop physical skills: they teach children how to exercise and
play games.
Education isfrom other production activities because the output (ed-
different
ucational achievement) to measure. While tests have been developed to
is difficult
measure basic cognitive skills, it is virtually impossible to accurately measure so-

cial and physical skills. Consequently, any empirical study of schools can, at best,
measure only one component of the output of schools. Because empirical studies
ignore social and physical skills, they provide an incomplete picture of the education
process and must be interpreted with caution.
The production-function approach explores the contributions of different inputs
to the cognitive achievement of students. Cognitive achievement is measured by
scores on standardized achievement tests. Suppose that achievement is defined as
the change in the test score of a particular child over a one-year period. If the output
of the school is defined as the change in the test score, the production function can
624 Part VI Education and Crime
be written as
Achievement = f(C, E, T, H, P) (21-1)
where
C= School curriculum
E= Quantity of instructional equipment of the school
T= Quantity of labor input (teacher) per student
H = Home environment of the child
P= Average achievement level of other students in the class
Achievement depends on five inputs: the school's curriculum and educational equip-
ment, the classroom teacher, the home environment, and the achievement level of
the child's classmates.
These five inputs to the production function can be divided into three groups:
1. School resources. The school has control over three inputs: curriculum,
instructional equipment, and classroom teachers. Under the supervision
of the local school district, the school designs a curriculum, purchases
instructional equipment (building, books, science labs, computers) and
hires teachers. An increase in the number of teachers decreases class size,
increasing the teacher input per student.
2. Home environment. Educational achievement is influenced by the home
environment of the child in three ways. First, parents set the rules of the
household, establishing an environment that is either favorable or
unfavorable to education. For example, an unfavorable environment is one
in which children watch television instead of reading books or doing their

homework. Second, parents can motivate their children by encouraging
reading, helping with homework, and rewarding success. Third, parents
can provide instructional materials such as books and home computers,
encouraging independent learning. The quality of the home environment
depends in part on the income and education level of the parents. The
children of wealthy and educated parents learn more because they receive
more encouragement and assistance at home, and also pick up verbal and
quantitative skills in everyday interactions with their parents. In contrast,
children from poor and less educated families learn less because there is
often less encouragement to achieve and less learning from parents. In
addition, children living in poverty are often malnourished, inhibiting their
ability to learn.
3. Peer group effects. The final input to the production process is the peer
group of the child. A child learns more if he or she is surrounded by smart
and motivated children. Smart peers promote achievement because of
cooperation (children learn from one another) and competition (children
compete with one another). Motivated peers promote achievement because
the teacher can spend less time disciplining and motivating students, and
more time teaching them. In addition, an unmotivated student provides an
undesirable role model for other students.
The Coleman Report

Which inputs to the production process are the most important? Using the production-
function approach, educators and economists have estimated the contributions of the
various inputs to the education process. The Coleman report (Coleman et al., 1966),
was the first attempt to estimate the education production function. The conclusions
of the report are as follows:
1 The most important inputs are the home environment of the student and the
characteristics of the peer group.
2. High-income families have more favorable home environments, so children

from wealthy families learn more than children from poor families.
3. Children from wealthy families provide a more favorable peer group, so
children with wealthy classmates learn more than children with poor
classmates.
4. School inputs (curriculum, instructional equipment, teachers, and class
size) do not affect achievement.
The Coleman report shocked the education community. The report was disturb-
ing and perplexing for three reasons. First, it implies that schools don't matter, that
educational achievement is determined by factors beyond the control of educators.
This was a surprise to teachers and educators, who assumed that they contributed
to the education process. Second, the report suggests that education cannot be used
to lift children out of poverty. If achievement depends exclusively on the child's
home environment and peer group, children in poor areas are trapped: they receive
inferior education precisely because they come from poor families (with inferior
home environments) and poor neighborhoods (with inferior peer groups). If
live in
school resources do not affect achievement, the government cannot compensate for
inferior home and peer environments by spending more money on the education of
poor students. Third, the report suggests that teachers do not matter, that all teachers
are equally effective, regardless of education, experience, innate intelligence, and
motivation. In other words, a teacher is a standardized input like a blackboard: put
one in the classroom and it works just as well as any other teacher.
To put the conclusions of the Coleman report in perspective, suppose that you
are a parent choosing a school for your child. What questions would you ask about
a prospective school? According to the Coleman report, you need to ask only one
question: what is the average income level of students in the school? It would be
senseless to ask questions about curriculum, instructional equipment, and teachers
because, according to the Coleman report, these inputs do not affect achievement.
The Coleman report precipitated a flood of criticism. As several researchers
pointed out, the statistical methods of the report were flawed by two basic problems:
1. Unit of observation. Coleman et al. (1966) chose the school, not the
individual student, as the unit of observation. The report explores the
relationship between the average achievement level of a school and
the average school inputs (e.g., average class size, average level of teacher
experience, average income) and found that schools with different inputs
did not have different achievement levels. The alternative approach is to
examine the relationship between the achievement level of an individual
child and the quantities of inputs experienced by that child (e.g.. the size of
the child's class, the teacher's experience and education level, the family
income). Because Coleman et al. use schoolwide data, their approach
obscures important differences between students in a given school and may
generate inaccurate assessments of the relative importance of education
inputs.
Robustness of results. The second problem is that the Coleman results are
not robust: they change with small changes in the statistical methodology.
Their results are sensitive to the order in which the various inputs (home
environment, peer environment, and school resources) are incorporated into
the statistical analysis. If the home variables enter the analysis before the
school variables, it appears that school inputs don't matter. If, however, the
school variables are entered first, it appears that school inputs affect
achievement. Since the Coleman results are sensitive to the order in which
inputs are included in the study, the results are of questionable validity.
Other Studies of the Production Function
Researchers did more than criticize the statistical methods of the Coleman report.
Using superior data and more sophisticated statistical methods, they generated new
estimates of the education production function. The new results cast doubts on some
of the conclusions of the Coleman report. The results of these new studies are by no
means definitive, so many issues remain unresolved. There is a consensus emerging,
however, on the influences of the home environment, the peer group, and teachers.
Home Environment and Peer Group Effects. Most researchers agree that the
home environment has an extraordinary effect on achievement. Wealthy and well-
educated parents provide a more favorable home environment. Researchers also
agree about the effects of the peer group on achievement: a student learns more
if she is surrounded by smart and motivated students. There is some evidence that
the largest peer group effects are experienced by low achievers: the students at the
bottom of the class have the most to gain from adding smart and motivated students
to the class. There is also evidence that these peer effects are most important in the
middle and upper grades (grades 5 through 12).
The Effectiveness of Teachers. There is also agreement that teachers matter.

Contrary to the conclusions of the Coleman report, teachers are not like blackboards,
but in fact vary in effectiveness. The most effective teachers are the ones that are
smart, motivated, innovative, and flexible.
A study of inner-city schools found that during a single academic year, a student
with a high-quality teacher outperforms a child with a low-quality teacher by up to
Chapter 21 Education 62/
one full grade level (Hanushek, 1992). For example, consider a child who starts third
grade at grade level 2.0, right on schedule. With a high-quality third-grade teacher,
the child could move up to grade level 3.2 by the end of the year and be ahead of
schedule; with a low-quality teacher, the child could move up slightly, to grade level
2.2, and not be ready for fourth grade.
While researchers agree that some teachers are more effective than others, they
do not agree on what makes a good teacher. In looking for teacher characteristics
that explain differences in effectiveness, researchers have focused on education level
(years of postgraduate study), communication skills (verbal ability), and experience
(years of teaching). The results of the studies are as follows:
1 . Education level. There is no evidence that teachers increase their teaching

skills by going to graduate school. It appears that students do not learn
more from teachers with advanced degrees.

2. Verbal skills. The most effective teachers have superior communication
skills. Students learn more from teachers who score high on standard tests
of verbal ability.
3. Experience. There is some disagreement about the effects of experience
on teaching skills. Murnane (1975, 1983, 1985) suggests that teaching

effectiveness increases for the first few years of teaching and then levels
off. In contrast, Hanushek (1971, 1981) suggests that teaching skills are
independent of teaching experience. Summers and Wolfe (1977) found that
low achievers learn more from inexperienced teachers, while high achievers
learn more from experienced teachers. Given these conflicting results, the
link between teacher experience and effectiveness is still an open question.
Curriculum and Instructional Equipment. There is evidence that curriculum

matters. One of the most important factors in the highly publicized decline in SAT
test scores during the 1970s and 1980s was the relaxation of graduation requirements.
Under the "smorgasbord" approach to education, students were allowed to choose
from a wide variety of courses, many of which were not academic in nature. The
typical student took fewer rigorous academic courses, and academic achievement
declined. This is the "Twinkie" effect: students skipped the meat and vegetable
courses (academic courses) and filled up on the Twinkie courses.
Researchers agree that differences in achievement are not caused by differences
in instructional equipment. In other words, students do not learn more in schools
that spend a relatively large amount on books, projectors, and buildings. This does
not mean that achievement would be unaffected if society returned to the days of log
schoolhouses and shared books. It means that, given the current levels of spending on
books and buildings, small changes in spending are unlikely to affect achievement.
Class Size. There is some disagreement about the effects of class size on achieve-
ment. As Hanushek (1981) points out, there have been 1 1 2 studies of the relationship
between class size and achievement, with 9 studies finding a negative relation-
ship, 14 studies finding a positive relationship, and 89 studies finding no significant
628 Parr VI Education and Crime
relationship. Summers and Wolfe 1977) ( suggest that low achievers learn more in
small classes, and high achievers learn more
in large classes. Given these conflicting
results, the link between class size and achievement is still an open question. The
issue is not whether class size matters at all, but whether small changes from the
current class size affect achievement. No one doubts that achievement would decline
if the average class size increased from 20 to 60 students.
Alternative Approach: School Resources and Market Earnings

The traditional studies of the education production function use test scores as a
measure of the output of education. An alternative approach looks at the relationship
between schooling and market earnings, using the wages earned by adults as a
measure of the output of their earlier education. The production question is: does
the wage earned by an adult depend on the quantity of resources used in her K-12
education? For example, consider two adults who differ only in their class sizes
during grade school. If school resources matter, we would expect the adult who had
smaller classes to have a higher skill level and thus earn a higher wage.
Card and Krueger (1996) summarize the results of several studies that exam-
ine the relationship between school resources and earnings. There is evidence that
school resources matter: students who were educated in schools with relatively high
spending levels earn higher wages. The estimated elasticity of wages with respect to
per pupil spending is about 0.15, meaning that a 10 percent difference in spending
generates a 1 .5 percent difference in wages. The elasticity of wages with respect to
class size is about 0.07. meaning that a 10 percent difference in class size generates a
0.7 percent difference in wages. As Card and Krueger note, some studies suggest that
the relationship between educational spending and wages is weak or nonexistent, so
the issue has not been resolved.
The Tiebout Process and Inequalities in Achievement

How does the Tiebout sorting process affect educational spending and educational
achievement? Under the Tiebout process, households sort themselves with respect to
three characteristics: desired spending on schools, property values, and peer group
characteristics.
Consider first the sorting with respect to educational spending. The desired
spending on education varies across households, reflecting differences in income
and preferences for education. Households sort themselves into high-spending and
low-spending school districts. In this case, spending inequalities result from the
simple fact that the wealth) spend more on education, just as they spend more on
clothes, cars, and housing. The spending inequalities cause inequalities in educa-
tional opportunity: the low-spending school districts have less money for teachers,
so they may have lower achievement levels.
Because local schools are financed with the property tax. households also sort
themselves with respect to property values. Because housing is a normal good
(positive income elasticity), the sorting with respect to house value causes
sorting with respect to income. Suppose that all the households in a city, wealthy
and poor, have the same desired spending on education. Under the property tax,
the city has two school districts, one for the wealthy (who live in large houses) and
one for the poor (small houses). Although the two school districts have the same
per pupil spending, achievement is likely to be higher in the wealthy school district
for two reasons. First, achievement is positively related to household income, so
the average achievement is higher in the wealthier school district. Second, because
of the more favorable peer environment in the wealthy district, a given child will
reach a higher achievement level in the wealthy Income segregation deprives
district.
poor students of the possible benefits of being surrounded by smart and motivated
students.
Consider finally the effects of student peer groups on location choices. Par-
ents care about educational achievement, not simply educational spending. Given
the importance of the peer group in achievement, the household looks for a school
with smart and motivated students. In practical terms, this means that the household
looks for a neighborhood full of high-income, well-educated parents who provide
a favorable home environment. The shopping for a peer group reinforces income
segregation. Suppose that a poor household has the same desired spending on educa-
tion asmiddle-income households and occupies a house with the same market value
(same tax liability). In the absence of peer group effects, middle-income households
would welcome the poor household into their school district. However, because the
poor student is likely to have a below-average achievement level, middle-income
households have an incentive to exclude the child from their school. In general, the
peer group effect adds another item to the household's shopping list, increasing the
pressures for segregation with respect to income, race, and social status. As a result,
it widens the spatial inequalities in educational achievement. Poor households are
denied access to schools that provide a favorable peer environment.
Spending Inequalities and Public Policy

By how much does spending per pupil vary across school districts in a particular
state? In 1987, spending per pupil in the Whitley School District in Kentucky was
$2,574, equal to the sum of $247 of local property revenue and $2,287 from the state
government (Evans, Murray, and Schwab, 1997). In contrast, the Walton Vernona
School District spent $3,508 per pupil, with $1,010 from local property taxes and
$2,498 from the state. In this case, the low-income district had lower local revenue
(a result of relatively low property values) and lower state aid.
Figure 21-2 shows the spending inequalities for four states whose education-
finance systems were eventually declared unconstitutional and reformed. For each
state, the taller bar shows the average spending per pupil in the 10 highest-spending
districts, and the shorter bar shows spending in the 10 lowest-spending districts. In
three of the four states, the high-spending districts spent more than twice as much
as the low-spending districts. In the 1980s, spending ratios above 1.5 were common
in other states (Katz, 1991).
Figure 21-2 Inequalities in Spending per Pupil in Selected States,

1986-1987
7.000 -
6,000 -
g- 5,000
gP 4,000
3.000
< 2,000 -
1,000 -
Kentucky Montana New Jersey

i\ Texas
| Lowest-spending districts J Highest-spending districts
Source: Jeffrey Katz. "The Search for Equity in School Funding," Governing. August 1991, pp. 20-22.
The Legal Mandate to Equalize Expenditures
State courts became involved in the financing of education in the 1970s. In response
to persistent inequalities in education spending, parents from poor school districts
sued state governments, claiming that the spending inequalities violated state con-
stitutions. In a number of states, the courts ruled in favor of the parents, finding that
inequalities in educational spending violated the equal-protection clauses of state
constitutions. Education is considered a fundamental right of citizens of the state,
meaning that all citizens are to receive the same quality of education. The courts
developed the principle of fiscal neutrality:
The quality of education may not be a function of the wealth of the local community.
In other words, children from poor school districts must receive the same quality
of education as children from wealthy school districts. The courts directed state
governments to equalize spending on primary and secondary education.
Serrano Priest.
v. The most frequently cited court case is Serrano v. Priest, de-
cided 1974 by the California Supreme Court. When John Serrano complained
in
about the low quality of the local high school's education program, the school's
principal suggested that Serrano move his family to a school district where per pupil
Table 21-1 Taxes and Spending in Baldwin Park and

Beverly Hills
Baldwin Park Beverly Hills
Property value per student $4,169 $49,501

Property tax rate 3.34% 2.55%
Educational spending per student $272 $1,535
SOURCE: James Guthrie. "United States School Finance Policy 1955-1980" In

School Finance Policies and Practices, ed. James Guthrie (Cambridge. Mass.:
Ballinger. 1980). p. 14.
spending was higher. Instead of moving, Serrano sued the state of California, arguing
that spending inequalities were unconstitutional.
To illustrate the spending disparities, Serrano cited the differences between two
school districts in the Los Angeles area, Beverly Hills and Baldwin Park. As shown in
Table 21-1 Beverly Hills used
, its larger property tax base to spend more per student
while charging a lower tax rate. In ruling in favor of Serrano, the court ruled that the
property tax system violated the equal-protection clause of the state's constitution.
The court ordered the state legislature to develop a financing system under which
per pupil spending would vary by no more than $100.
There have been similar court cases in dozens of other states. Between 1970
and 1983, the school finance systems of more than half the states were challenged,
but only seven finance systems were declared unconstitutional during that period. In
1989 and 1990, the courts struck down the the school finance systems of four states:
Kentucky, Montana, New Jersey, and Texas.
Limits on Reform. Two court cases established the limits on education reform.
In Mclnnis v. Ogilvie, the plaintiffs argued that the Constitution guarantees the right
to equal educational outcomes rather than equal educational spending. They argued
thatgovernment should eliminate inequalities in educational achievement, spending
more on low-achieving students to bring them up to the level of high achievers. The
court ruled against the plaintiffs, arguing that there was no practical way to enforce
an equal-achievement standard. Given this decision, the objective of reform efforts
is to equalize spending, not achievement.
Another case, San Antonio Independent School District v. Rodriguez, estab-
lished the limits for federal involvement in school finance. The court ruled that
education is not a fundamental right guaranteed to U.S. citizens. This ruling is based
on the fact that education is not mentioned in the U.S. Constitution. Consequently, the
equal-protection clause of the U.S. Constitution does not apply to education, and vari-
ation in per pupil spending is not proscribed by the U.S. Constitution. There are two
implications from the San Antonio decision. First, any reform of the school finance
system must come from state governments, not the federal government. Second,
although the court system may promote equalization of spending within individual
states, the courts will do nothing to promote equalization of spending across states.
Foundation Grants
About 80 percent of states use foundation grants to help narrow the gaps in education
spending across school districts. The foundation grant per pupil is determined by
the following formula:
_, Foundation Foundation Local property value ,„, _.

Grant = . . - . . r, (21-2)
level tax rate per pupil
Suppose a state picks a foundation level of $5,000 and a foundation tax rate of
1.5 percent (0.015). For a school district with $200,000 of local property value per
pupil, the foundation grant is $2,000:
Grant = $5,000 - 0.015 •

$200,000 = $2,000 (21-3)
In effect, the foundation grant equals the difference between the foundation level and
the amount of money the school district could raise locally if it were to impose the
foundation tax rate. A district with a lower tax base will receive a larger grant (e.g..
a district with only $100,000 of property value per pupil would receive a grant of
$3,500). Note that the foundation grant is independent of the school district's actual
tax rate; depends only on the local tax base and the state's foundation variables.
it
We can use a simple example to illustrate how a foundation plan affects the
tax and spending options of a local school district. Table 21-2 shows spending
options with different tax rates. The district has $200,000 of property per stu-
dent, so per pupil spending increases by $2,000 for every percentage point of
the property tax. One option is to pick a tax rate of 2.0 percent, which gener-
ates $4,000 of local revenue per pupil. If the foundation grant is $2,000, the dis-
trict has a total of $6,000 per pupil. If the district picks a tax rate of
2.5 percent, would collect an additional $1,000 in local revenue; because the
it
grant doesn't depend on the district's tax rate, education spending increases by only
$1,000.
How does a foundation grant affect spending in the school district? As ex-
plained in Chapter 17 (Voting with Ballots and Feet), a local school district is likely
to pick the preferred budget of the median voter. Figure 21-3 shows the budget
lines and indifference curves of Marian the median voter. The horizontal axis mea-
sures per pupil education spending and the vertical axis measures spending on other
goods. The budget line BC shows the initial trade-off between education and other
goods: Marian has a total of $ 1 0.000 per pupil to spend on education and other goods.
Table 21- -2 Tax and Spending Options under a

Foundation Grant
Tax Rate Local Tax Revenue Grant Education Spending
2.0% $4,000 $2,000 $6,000

2.5$ $5,000 $2,000 $7,000
Figure 21-3 Foundation Grant and Spending Per Student

B D
10.000
fe 7.600
°° 6,000
Budget line with $2,000 grant
Initial budget line
2,000 4,000 4,400
Spending on education ($ per pupil)
A $2,000 foundation grant shifts the budget line of the median voter from BC to BDF and the utility-maximizing
point moves from i ($400) of the grant is spent on education, and the remaining four-fifths is
to /. One-fifth
spent on the other goods. The school district cuts the tax rate from 2.0 percent (raising $4,000 in local revenue)
to 1.2 percent (raising $2,400), allowing citizens to spend part of the grant on other goods.
and every dollar spent on education decreases spending on other goods by one dol-
lar. Before the foundation grant, Marian maximizes utility at point i (education
spending = $4,000), and this is the point chosen by the local school district.
The foundation grant shifts Marian's budget line outward from BC to BDF.
Point D is in the new budget set because Marian can use the $2,000 grant to get
$2,000 worth of education while spending all of her own money ($10,000) on other
goods. The points along the line connecting D and F are in the new budget set:
starting from point D, Marian faces a dollar-for-dollar trade-off between education
and other goods. The shift of Marian's budget line moves the utility-maximizing
point from i to /. In other words, the grant increases education spending by $400
(to $4,400) and spending on other goods by $1,600 (to $7,600).
How can the foundation grant increase spending on other goods? The district
can fund its $4,400 education program with the $2,000 grant and only $2,400 of local
tax revenue. The grant decreases the local contribution to education from $4,000 to
$2,400, so the school district cuts local taxes by $1,600 and allow its citizens to
spend that much more on other goods. From the perspective of local citizens, the
foundation grant is equivalent to an increase in consumer income, which increases

the demand "normal" goods, including education and other goods. By cutting
for all
taxes, the school district is simply responding to the demands of its citizens.
In principle, foundation grants can be negative for some wealthy school districts,
meaning that a school district would transfer money to the state government. In our
example, if a wealthy district has a tax base per pupil of $400,000. the foundation
grant would be negative:
Grant = $5,000 - 0.015 • $400,000 = -$1,000 (21-4)
Under a pure foundation system, the district would transfer $ 1 ,000 to the state. The
transfer would shift the budget line of the median voter to the left, decreasing the
desired spending on education and other goods.
How does a foundation plan affect spending inequalities? The largest grants are
given to the school districts with the lowest property values, so the greatest stimulus
for education will occur in low-wealth districts. If the foundation plan is pure in the
sense that high-wealth districts are taxed, education spending in these districts will
actually decrease, and the spending gaps will narrow even more.
Guaranteed Tax Base or Power Equalization

A more aggressive approach to equalization involves an attempt to equalize the
revenue-generating capacities of school districts. Under a guaranteed tax base
(GTB) plan (a.k.a. district power equalizing), a state specifies a "guaranteed" tax
base per pupil, meaning that each school district has access to the same effective tax
base. The grant per pupil is determined by the following formula:
GTB Grant = Local tax rate • (Guaranteed tax base per pupil
— Local tax base per pupil) (21-5)
The tax base per pupil is the same as the property value per pupil. Once the state
picks the guaranteed tax base, it's up to each school district to pick a tax rate, which
then determines how much tax is collected locally and how big the grant is.
We can use a simple example to illustrate how a GTB plan works. Suppose
the guaranteed tax base is $300,000 and the local tax base is $200,000, so the gap
between the two tax bases is $100,000. Table 21-3 shows the implications of two
Table 21- -3 Options under a Guaranteed Tax

Base Plan
Tax Rate Local Tax Revenue Grant Total
2.0% $4,000 $2,000 $6,000

$5,000 $2,500 $7,500
different tax rates. With the lower tax rate (2.0 percent), the district raises $4,000
locally and gets a grant of $2,000, for a total of $6,000. With the higher tax rate
(2.5 percent), local tax revenue is $5,000 and the grant is $2,500, for a total of
$7,500. Because a tax-rate hike increases the grant as well as local tax revenue, the
local cost of getting an additional $1,500 in school spending is only $1,000.
The GTB plan decreases the opportunity cost of spending on education. Before
the grant program, there is a dollar-for-dollar trade-off between spending on edu-
cation and other goods: every dollar spent on education requires one dollar of local
taxes, so every dollar of education decreases spending on other goods by one dollar.
Under the GTB plan, the trade-off is lower. In Table 21-3, increasing the tax rate
from 2.0 percent to 2.5 percent increases local tax revenue by $ ,000 (and decreases
1
local spending on other goods by the same amount), but increases spending on edu-
cation by $1,500. On average, the school district gets a dollar's worth of education
by sacrificing about 67 cents' worth of other goods. In this example, the GTB plan
is equivalent to a matching grant with a match rate of one-half: an additional $1,000
of local taxes increases the grant by $500.

How does the GTB plan affect the choices of the median voter? In Figure 21-4,
the plan pivots Marian's budget line from BC to BD. The slope decreases because
the grant decreases the opportunity cost of education spending from $1 to 67 cents.
Marian chooses point g instead of point i, so school spending increases from $4,000
to $4,800. As in the case of the foundation plan, part of the GTB grant is spent on
other goods, with spending on these goods increasing by $800. In this example, the
GTB grant is split equally between education and other goods.
A GTB grant has a larger stimulative effect than a foundation grant because the
GTB grant has both an income effect and a substitution effect. Like the foundation
grant, the GTB grant increases Marian's real income, increasing her desired spending
on all "normal" goods, including education (the income effect). The GTB grant
also decreases the opportunity cost of education spending from $1 to 67 cents,
causing Marian to substitute education for other goods (the substitution effect). In
Figure 21-3, one-fifth of the foundation grant is spent on education. In Figure 21 —
one-half of the GTB grant is spent on education.
Under a pure GTB plan, wealthy districts would transfer money to the state
government. If a district's tax base exceeds the guaranteed tax base, the local revenue
will exceed guaranteed revenue, and the surplus revenue would go to the state. Since
the district must transfer to the state a fraction of each dollar collected, the GTB
plan increases the opportunity cost or price of education. In a wealthy district, the
income and substitution effects of the GTB plan combine to decrease spending on
education.
Would a pure GTB plan equalize spending on education? The GTB plan provides
each school district with the same effective tax base. If every community chooses
the same tax rate, education spending would be the same in all districts. If, however,
communities chose different tax rates, spending inequalities would persist. Under a
pure GTB plan, any remaining differences in education spending are the result of
differences in chosen tax rates, not differences in tax bases.
Figure 21-4 Guaranteed Tax Base Grant and Spending Per Student
g. 6.800
6.000
Budget line with

GTB grant
4.000 4.800 $
Spending on education ($ per pupil)
A GTB grant tilts the budget line of the median voter, reflecting the decrease in the opportunity cost of
education spending The utility-maximizing point moves from i to g. Half ($800) of the grant is spent on
education, and the other half is spent on other goods. The school district cuts the tax rate from 2.0 percent
(raising$4,000 in local revenue) to 1.6 percent (raising $3,200). allowing citizens to spend part of the giant
on other goods
Effects of Equalization Plans on Spending Inequalities

A recent study explores the changes in spending inequalities between 1972 and 1992
(Evans. Murray. Schwab, 1997). During this period, a total of 37 states reformed
their school finance systems, with 16 of these states under court orders to reform
their systems. In the states adopting court-ordered reform, state spending on local
education increased by 22 percent ($437 per pupil). Local support of education didn*t
change much over the period, so the states' share of education funding increased
from 44 percent to 52 percent.
In the states implementing court-ordered reform, spending inequalities dec-
reased significantly. Spending per pupil increased in low-spending school districts,
but didn't change much in high-spending districts, so reductions in ine-

qualities resulted from "leveling up." For example, for a school district in the
Chapter 2 1 Education 637
25th percentile (25 percent of districts initially spent less per pupil), education spend-
ing increased by 27 percent. In contrast, spending in the median district increased
by only 15 percent, and spending in the district in the 75th percentile did not change.
Another study of equalization programs suggests that the degree of spending inequal-
ity decreased between 16 percent and 38 percent (depending on how one measures
inequality).
What about the 2 1 states that reformed their education finance system without
the pressure of court orders? The reform plans in most of these states were ineffec-
tive: they did not have significant effects on either spending per pupil or spending
inequalities across school districts (Evans, Murray, Schwab, 1997). It appears that
court mandates are necessary for real reform.
Michigan is the exception to the rule that real reform requires court orders. In
1993, the state eliminated the local property tax as a source of education funding.
The elimination of the local property tax was combined with a moderate reduction
in income taxes and a large increase in the sales tax (Courant and Loeb, 1997).
The state now determines educational spending in all but the 28 wealthiest school
districts. Spending per pupil increased in small rural districts, and decreased in poor
urban areas and rich suburban areas. The other states with complete state control of
school finance are California, Hawaii, Florida, and Wisconsin.
Education Finance Reform and Central City Schools

How did central-city schools fare under the reform of the education finance system?
As explained in Chapter 1 2 (Segregation and Poverty in the Central City), central-city
schools have relatively high costs because a large fraction of their students come from
poor families. These schools devote more time and resources to security measures,
dealing with family and health crises, trying to teach children who do not speak
standard English, and trying to teach children with weak educational preparation.
Because these schools have relatively high costs, they have above-average spending
levels. As a result, they did not benefit from equalization programs, and in some
cases actually did worse (Courant and Loeb, 1997).
The formulas used in many equalization plans are biased against central cities. As
explained by Duncombe and Yinger (1997), one defect of the equalization formulas
is that they do not account for differences in the costs of producing education. If
the formulas were modified to incorporate cost differences, some central-city school
districts would receive two or three times as much grant money.
Desegregation Policy
Segregation and Public Policy
How segregated is the U.S. school system? In 1995, about two-thirds of black stu-
dents attended schools that had a majority (at least 50 percent) of minority students,
and about one-third attended schools in which at least 90 percent of students were
minorities. The figures for Hispanics are a bit higher: 74 percent in minority-majority
schools and 35 percent in schools where minorities made up 90 percent of students.
Segregation most pronounced in urban schools. In the 10 largest school districts,
is
white students comprised more than 22 percent of the student population in only
one district (Institute on Race and Poverty. 1998). In 1995. the student body of the
Chicago school district was only 1 percent white; the student body of Detroit school
1
district was only 6 percent white.

There is also segregation inside schools. According to a report from the De-
partment of Education ( 1987), classrooms are even more segregated than schools.
Under the ability-tracking system, students are segregated with respect to ability and
achievement: the high achievers are put one class and the low achievers are put in
in
another class. Because minority students often are behind white students in the early
grades, ability tracking in the later grades contributes to racial segregation within
the schools.
What causes racial segregation in the schools? One reason for segregation is
racial prejudice. Some white parents are unwilling to let their children be educated
with racial minorities, regardless of the achievement level of the minority children
(and their peer group effects). Others use race as a signal about the likely achieve-
ment level of students: if minority children have, on average, a lower achievement
level, a simple way to produce a more favorable peer group is to exclude all racial
minorities.
number of ways. The first approach
Racial segregation can be accomplished in a
is method used in many southern cities until the early
to use explicit segregation, a
1970s. A more subtle approach, used by many northern cities, is to manipulate atten-
dance boundaries to send most black children to predominantly black schools. The
third approach is to set up neighborhood schools and let households sort themselves
with respect to race: households interested in sending their children to white schools
can move to wealthy suburban school districts that exclude blacks. Sending children
to private schools may increase segregation because many private schools are too
expensive for minority households.
How does the educational achievement of blacks and Hispanics compare to that
of whites? Here are some facts on educational achievement (Institute on Race and
Poverty, 1998):
1. In grades K-6, 31 percent of urban black students had above-average

reading scores, compared to 61 percent for urban white students.
2. In grades K-6. 38 percent of urban black students had above-average math
scores, compared to 67 percent for urban white students.
3. In grades 9-12, 31 percent of urban black students had above-average
scores on standardized math tests, compared to 65 percent for urban white
students.
4. In grades 9-12, 29 percent of urban black students had above-average

scores on reading achievement tests, compared to 67 percent for urban
white students.
Court-Ordered Desegregation
The courts have outlawed the first two methods of segregation. In Brown v. Board
of Education ofTopeka (1954), the Supreme Court declared that explicit racial seg-
regation is unconstitutional. This Brown case reversed Plessy v. Ferguson (1896),
which established the constitutionality of "separate but equal" schools. The Brown
decision required the desegregation of schools only in areas where local govern-
ments pursued a policy of explicit segregation. Most of the cities forced to de-
segregate their schools under the Brown decision were in the South. As pointed
out by Clotfelter (1979), the mandate for desegregation was extended in the 1970s
to northern cities that used attendance boundaries and other subtle techniques to
promote segregation. The courts ordered several large northern and western cities
to desegregate their schools. The objective was to integrate the schools to the
extent that would have occurred in the absence ofearlier government policies
that had encouraged segregation. In other words, the court did not require the
cities to develop a truly integrated school system, only one in which government
policy had a neutral effect. The segregation that results from the individual
location choices of households (the Tiebout shopping process) is considered
constitutional.
How did white households respond to forced desegregation? One of the lessons
from the history of desegregation policy is that it often increases rather than decreases
segregation. There are powerful forces behind segregation that are not neutralized
by court-ordered desegregation, and households can take a number of actions to
counteract desegregation plans. There are two responses that weaken or reverse
the effects of desegregation: households can move from the integrating school dis-
trict to an all-white school district, or parents can enroll their children in private
schools.
Moving to Another School District

Consider first the possibility of moving to another school district. In the South,
many central cities had marbled spatial distributions of households: black areas were
interspersed within white areas. The cities used explicit segregation to send whites
and blacks living in the same area to different schools. When the courts outlawed
explicit segregation, many central-city white families moved to the predominantly
white suburbs to keep their children in all-white schools.
Clotfelter (1979) examined the various responses to desegregation in Baltimore
and Atlanta and came to the following conclusions:
1 The fleeing of whites to the suburbs decreased the demand for central-city
housing and decreased housing prices. In stable neighborhoods, the price of
central-city housing decreased by about 4.7 percent for a 10 percent
increase in the proportion of blacks in local public schools.
2. Families with school-age children were the most likely to move to the
suburbs when central-city schools were integrated.
The same phenomenon occurred in other cities that used busing to integrate their
schools: some white households moved from desegregating school districts to pre-
dominantly white school districts.
Switching to Private Schools
Another option for parents in a desegregating school district is to enroll their chil-
dren in private schools. To evaluate the effects of desegregation on private-school
enrollment, it will be useful to develop a model of the choice between private and
public schools.
The Choice between Public and Private Schools. In choosing between public
and private schools, parents face a trade-off between the consumption of education
and other goods. Private schools charge tuition, but also provide a better education,
on average, than a public school. Parents send a child to a private school only if the
increase in educational achievement is worth the sacrifice of other goods.
Figure 21-5 shows the options faced by the Smith family. The horizontal axis
measures the consumption of education (measured by the achievement level of the
Figure 21-5 Desegregation Causes Switch from Public to Private

Education
Public education budget point
after desegregation
int for public education

before desegregation
Budget point for private

education
Achievement level
The budget points are H (public education) and V (private). Given its indifference curves, the
initial
household is not willing to sacrifice $400 worth of other goods to increase achievement b> E, - Eb so it ,
chooses public education Desegregation decreases achievement in public schools (the budget point shiiis
from li to D), causing a switch to private schools
child), and the vertical axis measures consumption of other goods (in dollars). The
family faces a binary choice: it chooses either a public-school education or a private-
school education, not a mixture of the two. Therefore, the budget set consists of two
points, V (private education) and B (public education). If the family has $2,000
to spend on education and other goods, the public-school option in the absence of
desegregation is point B: if the family chooses public education, it receives Ef, units
of free public education, leaving $2,000 for other goods. If private-school tuition is
$400, the family could get Ev units of education at a cost of $400, leaving $1,600
to spend on other goods (point V).
The Smiths choose public education over private education. The household's
negatively sloped indifference curves (U\ and Ui) reflect its subjective trade-off be-
tween education and other goods. The indifference curves show that the family is
willing to sacrifice only a small amount of noneducation goods to increase educa-
tional achievement. Given the indifference curves and the budget points, the utility
level from public education {II 2) exceeds the utility level from private education
{U\), so the family chooses public education. The benefit of increased achievement
(from Ef, to E v ) is outweighed by the loss of $400 worth of other goods.
Differences in Achievement? Why do students in private schools learn more

than their counterparts in public schools? According to Murnane (1986), private
schools are superior in part because they have more favorable peer group effects. As
discussed earlier in this chapter, a particular student learns more if he is surrounded
by smart and motivated classmates, and the smartest and most motivated students
come from high-income families. Since a large fraction of private-school children
come from high-income families, the peer groups in private schools are superior to
the peer groups in public schools. According to Murnane, the favorable peer group
effects explain over half of the difference between the achievement levels of public
and private students.
Murnane quantifies the peer group effect by predicting student test scores under
different circumstances. The first question he poses is: how would the test score of the
typical public-school student change if the student transferred from the public school
to a private school? The answer is that the student's score would increase from 24.4
to 26.1. By transferring to a private school, the typical student would experience a
7 percent increase in achievement. The second question posed by Murnane is: how
would the test score of the typical public-school student change if the student, along
with his public-school classmates, transferred from the public school to a private
school? The answer is that the student's score would increase from 24.4 to 25.1,
an increase of about 2.9 percent. If the student brings his public-school peers along
with him to the private school, he experiences less than half the achievement gain
that occurs if he leaves his public-school peers behind.
Murnane's results suggest that a large part of the difference between pub-
lic and private schools is caused by peer group effects. A student in a private
school is surrounded by smarter and more motivated students and therefore learns
more. The other half of the achievement difference is presumably caused by dif-
ferences in curriculum, teachers, educational methods, and discipline. An important
advantage of private schools is that they have more flexibility in expelling

troublemakers.
More recent studies of public and private schools suggest that the differences in
achievement are very small, at least as measured by test scores. Goldhaber (1996)
finds no difference in achievement between comparable students in public and private
schools, and Gamoran (1996) finds a very small advantage for students in private
schools. Rouse ( 1996) found a small private-school advantage in mathematics, but
no difference in reading.
Although the differences in achievement between public and private schools
appear to be small, the differences in educational attainment may be large. Achieve-
ment is usually measured by test scores in a particular grade, while attainment is
measured by the years of schooling. The key measures of attainment are high school
graduation rates, college attendance rates, and college graduation rates. Neal ( 1997)
found that urban minority students attending private (Catholic) high schools have
much higher graduation rates than comparable students in public high schools. In
addition, private- school students are more likely to get a college degree. These higher
attainment rates translate into higher earnings after high school: on average, students
who attend private high schools earn 8 percent more than comparable students who
attend public schools (Neal, 1997).
Desegregation and Private Schools. How does desegregation affect the house-
hold's choice between private and public schools? Suppose that desegregation enrolls
more poor children in the public schools and decreases the average achievement level
of the Smiths' public school. In Figure 21-5, the budget point for public education
would shift to the left, from B to D. For the Smiths, who chose the public school
before desegregation, the utility generated by the public school is now less than the
from private schools: the achievement gap is larger, so the increase in achieve-
utility
ment generated by a switch from the public to the private school is now worth the
$400 sacrifice of other goods. In this case, desegregation causes a switch to a private
school.
There is conclusive evidence that desegregation causes "white flight" to private
schools. Clotfelter (1976) hypothesizes the following relationship:
E= f(S.I.B) (21-6)
where
E= Proportion of white students enrolled in private schools

5 = Family size
/ = Household income
B — Percent of public-school students who are black
Using data from Mississippi, he concludes that white enrollment in private schools
increased with the number of blacks in the neighborhood public school. The re-
lationship is especially strong when black enrollment in the neighborhood public
school exceeds 50 percent. For B less than 50 percent, a one-unit increase in B (one
percentage point increase) increased E by 0.40 percentage points; for B greater
than 50 percent, a one-unit increase in B increased E by 2.8 percentage points. In
other words, the tipping point (where the relationship between B and E suddenly
becomes much stronger) is about 50 percent.
Tiebout-Based Segregation
The segregation that results from the individual location choices of households (the
Tiebout shopping process) isconsidered constitutional. What sort of government
actions would diminish this type of segregation? In other words, what policies di-
minish the incentives for households to sort themselves with respect to income and
race?
One approach would be to consolidate local school districts into a metropolitan
school district or a statewide school authority. Such a policy would generate a uniform
spending level and a single tax base, so there would be less incentive for households
to sort themselves with respect to desired school spending and property values.
As explained in the chapters on local government, there are some trade-offs with
this policy: to the extent that there is diversity in demand for education, the large
jurisdiction — with uniform provision of education will be less efficient than
its —
the system of smaller jurisdictions. Moreover, as pointed out by Bradford and Oates
(1974), such a policy would not eliminate segregation because parents will still
seek a school that provides a favorable peer group for their children. Middle-income
households still have an incentive to exclude poor children from their neighborhoods
and schools because the poor have lower average achievement levels.
A less drastic approach would be to use intergovernmental grants to equalize
spending. This policy has an advantage over consolidation because it gives the
local school district greater control over how to spend its fixed budget. It eliminates
the incentive for fiscally based sorting (sorting with respect to desired spending
levelsand property values), but would not affect sorting motivated by peer group
governments have tried a number of grant programs, none of which
effects. State
has narrowed spending gaps significantly. The only equalization policy that worked
was the California program, under which the state assumed complete control of
education funding.
Is there any way to diminish the sorting motivated by peer group effects? Sup-
pose that state grants were tied to students (they traveled with the student to her
school) and were inversely related to the income level of the student's family. By
admitting a low-income student, a school would increase its grant revenue; it could
use the extra money to improve
its education program (e.g., decrease class sizes, hire
Such a scheme would decrease the incentive for middle-income and

better teachers).
wealthy households to exclude poor students, so it would promote integration. To
the extent it fails to integrate the schools, a school full of poor (highly subsidized)
students could at least use the extra money to provide a better education program.
644 Parr VI Education and Crime
Subsidies for Private Education: Education Vouchers

Should the government subsidize students who attend private as well as public
schools? The advocates of private-school subsidies view the public-school system
as an inflexible and inefficient monopoly. They argue that a subsidy for private
education would force public schools to compete with private schools for students,
so public schools would be more concerned about efficiency and more responsive
to parental concerns. Would the public support of private schools lead to greater
inequalities in educational achievement?
Under a voucher system, every child is issued a coupon that can be used to pay
for either public or private education. The typical voucher plan has the following
features:
1. The state gives each child an educational voucher with a face value equal to
the current per pupil spending in public schools (for example, $5,000).
2. The family uses the voucher to pay for tuition at a qualifying school, public
or private. The school redeems the voucher with the state government.
3. Public schools charge tuition equal to $5,000, making public education

effectively free.
4. To qualify for the voucher program, a school must teach basic cognitive
skills and civics, and must admit students without regard for race, sex, or
religion.
5. The family is free to supplement the voucher with its own funds, so the
child is free to attend a school charging more than $5,000 in tuition.
6. The government collects and disseminates information on the teaching

methods, curriculum, and test scores of qualifying schools.
The voucher program involves a mixture of subsidies and regulations. The subsidies
place private and public schools on equal footing with respect to governmental
support. The regulations prevent explicit racism and sexism and force private schools
to teach basic cognitive skills and civics. The government's information program
helps parents choose among the alternative schools.
The debate over education vouchers has been spirited. Like most policy pro-
posals, the voucher plan has both good and bad points. Because a true voucher plan
has never been tried, there is a great deal of uncertainty about its possible effects.
The obvious response is to experiment with voucher systems in one or more school
districts.
The advocates of education vouchers stress two points in their arguments for
vouchers:
1. Competition. The voucher plan would allow private schools to compete

more effectively for students. Competition could improve the efficiency of
the educational system, increasing educational achievement per dollar
spent.
2. Diversity. The voucher plan would encourage diversity in educational

philosophy and methods, providing parents with more options.
Voucher advocates argue that vouchers would break up the public-school monopoly,
leading to greater efficiency and wider consumer choice.
The opponents of vouchers suggest that wider school choice is likely to increase
segregation with respect to income, race, and academic ability. Although a parent
who receives a voucher can pick any school —public or private — not all parents
The experience with school-choice programs suggests that
will exercise this option.
parents with the most education and the highest income are more likely to switch
schools (Levin. 1997). The children in these families are above-average in terms of
achievement and parental involvement, so the movement of these children to other
schools makes the peer group at their original schools less favorable. As a result,
students of low socioeconomic status will have less favorable peer groups and thus
lower educational achievement and attainment.
A recent model of vouchers generates some predictions about the effects of
vouchers on different types of students (Epple and Romano, 1998). The model
assumes that private and public schools are equally effective, and that peer group
effects are important in educational achievement and parental choices. The simulated
effects of a universal voucher of $2,000 (which covers only part of the cost of private
schools) are as follows.
1 The share of students in public schools drops from 90 percent to 70 percent.

2. A large number of high-ability students switch from public schools to
private schools, and the average-ability level of public schools drops by
about 1 6 percent.
3. There is a large per capita benefit for high-ability, low-income students,
most of whom switch to private schools and benefit from a more favorable
peer group.
4. There is a small per capita loss for low-ability, low-income students, most
of whom stay in public schools and suffer from the less favorable peer
group.
5. The total gains experienced by high-ability students exceed the total losses
of low-ability students by a small margin, so there is a net gain in terms of
achievement.
There are clearly trade-offs associated with education vouchers. The efficiency
of the school system could improve because of increased competition and more
choice. On the other hand, vouchers could increase segregation with respect to
income, race, and ability. In addition, the achievement gains of some students could
come expense of others. Before voters can make an informed decision about
at the
vouchers, these trade-offs must be quantified.

One way to quantify the trade-offs is through policy experimentation. In 1990,
the city of Milwaukee launched a voucher experiment that by 1995 involved 1,500
students. Because the attrition rate of participants has been relatively high, there
646 Pan VI Education and Crime
are several problems in interpreting the data from this experiment. But in the next
few years researchers hope to use the ongoing experience of Milwaukee and new
experiments in other cities to gain some insights into the trade-offs associated with
vouchers.
Summary
1 The production function summarizes the relationship between education
inputs and achievement.
a. There are five inputs to education: curriculum, educational equipment,
teachers, the home environment, and the peer group.
b. The home environment has an extraordinary effect on achievement.
c. A child learns more when surrounded by smart and motivated peers.
d. Teachers vary in effectiveness. Teacher productivity increases with
communication skills, but is not affected by graduate education. Some
researchers suggest that productivity increases for the first few years of
teaching, while others suggest that productivity is independent of teaching
experience.
e. The link between class size and achievement is still an open question. If
there is a negative relationship, it seems to be a weak one.
2. Under the Tiebout process, households sort themselves with respect to three
characteristics: desired spending on schools, property values, and peer group
traits.
3. Education spending varies considerably across school districts, a result of

differences in desired spending and differences in property tax bases.
4. U.S. schools are highly segregated: in 1995. two-thirds of minority students

attended schools in which over 50 percent of the students were minorities.
Under the ability-tracking system, classrooms are even more segregated than
the schools.
5. Blacks and Hispanics have relatively low achievement levels.
6. Under a foundation plan, the state grant is inversely related to the district's tax
base per pupil. The grant increases spending on education and also decreases
taxes, so households can spend part of the grant on other goods.
7. Under the guaranteed tax-base (GTB) plan, every school district has access to
the same effective tax base. The pure version of GTB would decrease spending
inequalities by a relatively large amount because it (i) decreases the tax base of
wealthy districts and increases the tax base of poor districts and (ii) increases
the opportunity cost of education in the wealthy district and decreases the
opportunity cost in the poor district.
8. The Supreme Court outlawed segregation based on explicit and implicit

government actions and ordered many cities to desegregate their school
s\ stems.
a. Many white households moved from central-city school districts to

all-white school districts in the suburbs.
b. Other white households switched to private schools.
9. In choosing between public and private schools, parents face a trade-off

between the consumption of education and other goods.
a. Achievement is perhaps slighty higher in private schools, a result of more
favorable peer groups and other factors.
b. Desegregation causes some households to switch to private schools if the
integration of public schools decreases achievement.
10. Under is issued a coupon to pay for either public

a voucher system, every child
or private education. The voucher plan has advantages and disadvantages.
a. Competition among schools would increase, improving efficiency and
increasing consumer choice.
b. Segregation with respect to income, race, and ability may increase.

1. According to Summers and Wolfe (1977), low achievers learn more in small
classesand with inexperienced teachers, while high achievers learn more in
large classes with experienced teachers. Why?
2. Consider a school district that can hire teachers from a large pool of applicants
that differ in their verbal abilities and the number of credit hours of graduate
study in education. Under the current salary structure, a teacher's annual salary
increases by $100 for a one-unit increase in credit hours and increases by $200
for a one-unit increase in verbal ability. Suppose that the district's objective is to
minimize the cost of providing a given educational output. Use the standard
isoquant-isocost analysis (covered in intermediate microeconomics) to show the
district's choice (the cost-minimizing combination of verbal ability and hours of
graduate study):
a. Use the available empirical evidence to draw the education isoquant, with
verbal ability on the horizontal axis and graduate credit hours on the vertical
axis.
b. Show the cost-minimizing combination of verbal ability and graduate
education. Provide a brief explanation of your choice.
3. The objective of the state of Egalitaria is to equalize educational opportunity
(defined as the same average achievement level in every school). Comment on
the following:"To achieve its objective, the state must equalize educational
spending (spending the same amount per student in every school)."
4. Your task is to predict the effects of a state's foundation plan on the school
district and the state budget. The foundation tax rate (r) is 2 percent, and the
foundation level is $6,000. The income elasticity of demand for education is
+ 1.20.
a. Fill in the blanks in the following table.
Tax base per pupil $200,000

Initial spending per pupil $4,000
Tax rate —
Foundation grant —
Initial income of median voter $20,000
Percentage change of median- voter income —
New (postgrant) spending per pupil —
Local contribution to education (per pupil)
New tax rate —
b. If the school district has 1,000 students, how much does the foundation plan
cost the state?
c. What is the bang (increase in education spending) per buck of the foundation
grant?
5. Compute the price elasticity of demand for education given the following facts:
a. The local school district has a property value per student of $300,000.
b. The state institutes a guaranteed tax-base (GTB) plan = $400,000.
c. As a result of the GTB, the school district increases spending per student
from $4,000 to $4,200.
6. Suppose that you are the budget analyst for the state government. Your task is to
estimate the fiscal impact of a guaranteed tax-base plan. Assume the following:
;'.
The price elasticity of demand for education is —0.80.
ii. The guaranteed tax base is $400,000.
iii. The typical school district has 100 students and a property tax base of
$300,000 per student, and initially spends $3,000 per student.
Compute the effects of the GTB on the following:
a. Spending per pupil.
b. The local contribution to education (per pupil).
c. The state contribution to education (per pupil) and the total budgetary cost of
the GTB plan.
d. The bang (increase in education spending) per buck (budgetary cost).
7. Use the median-voter model (see Chapter 17) to predict the effects of
desegregation on spending (per pupil) in public schools. Will spending increase
or decrease?
8. One of the criticisms of a voucher program for education is that it would
increase spending inequalities and achievement inequalities. Design a new
—
voucher plan that would narrow rather than widen spending and —
achievement differences across schools. How would you change the
conventional voucher proposal to decrease spending and achievement
inequalities?
9. Consider Creditland, a state that recently implemented a system of income tax

credits for private schools. The characteristics of the state and the program are
as follows:
I. Each family has one child in school.

ii. The government spends $4,000 on each child in the public schools.
Hi. Tuition in private schools is $3,500.
iv. The tuition tax credit is 50 percent of private-school tuition.
a. What additional information do you need to compute the fiscal effects of the
tax-credit program?
b. Make up some numbers and compute the fiscal effect.
c. Choose the word in parentheses that makes the following statement correct
and then explain your choice: "The net budgetary cost of the tax-credit plan
increases as the price elasticity of demand for private education (increases,
decreases) in absolute value."

Overview
Garms, Walter I.; James W. Guthrie; and Lawrence C. Pierce. School Finance: The
Economics and Politics of Public Education. Englewood Cliffs, N.J.: Prentice Hall,
1978. Chapter 3 discusses the reasons for government intervention in education.
Chapter 4 discusses the roles of the federal, state, and local governments.
Murnane, Richard J. "An Economist's Look at Federal and State Education Policies." In
American Domestic Priorities: An Economic Appraisal, ed. John Quigley and Daniel
Rubinfeld. Berkeley and Los Angeles: University of California. 1985. Discusses a
number of issues in public education, including trends in achievement scores, merit
pay, the restrictive effects of state certification requirements, and curriculum
reforms.
Education Facts
Advisory Commission on Intergovernmental Relations. State Aid to Local Governments.
Washington, DC: U.S. Government Printing Office. 1969.
Brown. L.; J. Killalea: R. Rosthal; and E. Tron. "School Finance Reform in
A. Ginsberg:
Achievements and Failures." Selected Papers in School Finance, 1978.
the Seventies:
Washington, D.C: U.S. Department of Health. Education, and Welfare, 1978.
Congressional Budget Office. Trends in Educational Achievement. Washington, D.C: U.S.
Congress, 1986. Reports trends in test scores.
Katz, Jeffrey. Has Title I Improved Education for Disadvantaged Students? Evidence from
Three National Assessments of Reading. Denver: Education Commission of the States,
1981a. Evaluates the effects of federal support programs.
"The Search for Equity in School Funding." Governing, August 1991. pp. 20-22.
.
The Third National Mathematics Assessment: Results, Trends, and Issues. Denver:
.
Education Commission of the States, 1982. Reports trends in math scores for 9-, 13-,
and 17-year-olds.
Three Assessments of Reading: Changes in Performance, 1970-1980. Denver:
.
Education Commission of the States, 1981b. Reports trends in reading scores for 9-,
13-, and 17-year-olds.
.Three National Assessments of Science: Changes in Achievement, 1969-77. Denver:

Education Commission of the States. 1978. Reports trends in science scores for
9-. 13-. and 17-year-olds.
National Center for Education Statistics. Digest of Education Statistics, 1982. Washington.
D.C.: U.S. Government Printing Office, 1982. Documents the differences in
achievement between whites, blacks, and Hispanics.
Office of Civil Rights, Department of Education. 1986 Elementary and Secondary School
Civil Rights Survey: NationalSummaries. Washington, D.C.: U.S. Government
Printing Office, 1987. Shows that classrooms are more segregated than schools.
Schiller. Bradley. The Economics of Poverty and Discrimination. 5th ed. Englewood Cliffs.
N.J.: Prentice Hall, 1989.

U.S. Census Bureau. Statistical Abstract of the United States, 1987. Washington, D.C.: U.S.
Government Printing Office.
U.S. Department of Education. The Condition of Education: A Statistical Report, 1983,
1985, 1986. Washington. D.C.: U.S. Government Printing Office. 1987. Reports the
facts on education, including information on student performance, fiscal resources,
teacher characteristics, public opinion of the schools, and governance.
Production Function
Brown, B. "Achievement, Costs, and the Demand for Public Education." Western Economic
Journal 10 (1972). pp. 198-219. Suggests that an increase in the teacher-pupil ratio
does not affect achievement, but an increase in teacher quality (experience and
training) does increase achievement.
Brown, B., and D. Sachs. "The Production and Distribution of Cognitive Skills within
Schools." Journal of Political Economy 83 (1975). pp. 571-94.
Card, David, and Alan B. Krueger. "School Resources and Student Outcomes: An
Overview of the Literature and New Evidence from North and South Carolina."
Journal of Economic Perspectives 10. no. 4 (Fall 1996). pp. 31-50.
Coleman, James S.; E. Q. Campbell; C. J. Hobson; J. McPartland; A. M. Mood; F. D.
Weinfield; and R. L. York. Equality of Educational Opportunity. Washington. D.C.:
U.S. Government Printing Office. 1966. The controversial "Coleman report." which
was interpreted to mean that schools and teachers do not matter.
Hanushek, Eric A. "The Economics of Schooling: Production and Efficiency in Public
Schools." Journal of Economic Literature 24 1986), pp. 141-77. A survey of
( 1
research on education, focusing on production and efficiency. Discusses the results of

studies that have estimated the education production function.
. "Measuring Investment in Education." Journal of Economic Perspectives 10. no. 4
(Fall 1996). pp. 9-30.
."Throwing Money at Schools." Journal of Policy Analysis and Management 198 1 ), 1 (
pp. 19-41. Summarizes the results of estimates of the production function, focusing on
the effects of teachers on achievement. Concludes that teachers do in fact matter, with
teachers of higher verbal ability doing better in terms of student achievement. Suggests
that spending more money on schools does not increase achievement; argues for the
development of direct performance incentives.
.
"The Tradeoff Between Child Quantity and Quality." Journal of Political Economy
100.no. (1992). pp. 84-117.
1
Murnane. Richard J. The Impact of School Resources on the Learning of Inner City-
Children. Cambridge, Mass.: Ballinger, 1975. Surveys the results of previous studies
and provides new estimates of the education production function. Concludes that
teacher experience matters for the first few years of teaching.
Summers, Anita, and Barbara Wolfe. "Do Schools Make a Difference?" American
Economic Review 67, no. 4 (1977), pp. 639-52. Estimates the education production
function, showing that achievement depends on teacher characteristics and class size,
with different effects on high achievers and low achievers.
Teachers: Productivity and Class Size
Eberts, Randall W., and Joe Stone. "Wages, Fringe Benefits, and Working Conditions: An
Analysis of Compensating Differentials." Southern Economic Journal 52, no. 1 (1985).
pp. 274-79. Estimates the responsiveness of teacher salaries to changes in working
conditions, including changes in class size.
Hanushek, Eric A. "Teacher Characteristics and Gains in Student Achievement: Estimates
Using Micro Data." American Economic Review 61, no. 2 (1971), pp. 280-88.
Estimates the education production function, showing that achievement depends on the
general ability of teachers (as measured by scores on verbal tests) and how recently the
teacher has received formal training.
Murnane, Richard J. "Understanding the Sources of Teaching Competence: Choice, Skills,
and the Limits of Training." Teachers College Record 84 (1983), pp. 564-69. Discusses
the evidence about which teacher characteristics matter (verbal ability and experience)
and which do not (formal education) and the policy implications of these findings.
Decline of Test Scores
Advisory Panel on the Scholastic Test Score Decline. On Further Examination. New York:
College Entrance Examination Board, 1977. Discusses some of the reasons for the
decline in SAT scores.
Harnischfeger, Annegret, and David E. Wiley. Achievement Test Score Decline: Do We
Need Worry? Chicago: Cemrel, 1975.
to
. Time and Learning in California Schools. Sacramento: California Testing Program,
1984. Suggests that student achievement depends on the number of academic courses
completed.
Spending Inequalities
Berne, Robert, and Leanna Stiefel. The Measurement of Equity in School Finance.
Baltimore: Johns Hopkins University Press, 1984.
Coons, John E.; William H. Clune; and Stephen D. Sugarman. Private Wealth and Public
Education. Cambridge, Mass.: Harvard University Press, 1970. Developed the notion
of fiscal neutrality.
Guthrie, JamesW. "United States School Finance Policy 1955-1980." Chapter 1 in School
Finance Policies and Practices, ed. James W. Guthrie. Cambridge, Mass.: Ballinger,
1980. Relates the history of school finance, focusing on court cases that forced states to
reform their financing policies.
McDonald, M. Brian. "Educational Equity and the Fiscal Incidence of Public Education."
National Tax Journal 33 (1980), pp. 45-54. Shows that there is an inverse relationship
between income and the net benefits from education.
Equalization Policies
Bradford, David, and Wallace E. Oates. "Suburban Exploitation of Central Cities and
Government Structure.'* In Redistribution through Public Choice, ed. Harold Hochman
and George Peterson. New York: Columbia University Press. 1974. Discusses the
suburban exploitation hypothesis and predicts the effects of metropolitan consolidation
on tax rates, expenditure levels, and income segregation.
California Department of Education. Selected Financial and Related Data for California
Public Schools, K-12, 1984. Sacramento: State of California, 1985.
Carroll, Stephen J. The Search for Equity in School Finance: Summary and Conclusions.
Santa Monica: Rand Corporation, 1979. Discusses the reasons why GTB plans have
not had a significant effect on spending inequalities.
Courant, Paul N.; Edward M. Gramlich; and Daniel Rubinfeld. "The Stimulative Effect of
Intergovernmental Grants, or Why Money Sticks Where It Hits." In Fiscal Federalism
and Grants-in-Aid, ed. Peter Mieszkowski and William Oakland. Washington, D.C.:
Urban Institute. 1979. Discusses the flypaper effect of intergovernmental grants.
Courant, Paul N„ and Susanna Loeb. "Centralization of School Finance in Michigan."
Journal of Policy Analysis and Management 16, no. 1 (1997), pp. 114-36.
Duncombe, William, and John Yinger. "Why Is It So Hard to Help Central-City Schools?"
Journal of Policy Analysis and Management 16. no. 1 (1997). pp. 85-1 13.
Evans. William N.; Sheila E. Murray; and Robert M. Schwab. "Schoolhouses. Courthouses,
and Statehouses After Serrano." Journal of Policy Analysis and Management 16, no. 1
(1997), pp. 10-31.

Gramlich, E. M. "Intergovernmental Grants: A Review of the Empirical Literature." In The
Political Economy of Fiscal Federalism, ed. Wallace E. Oates. Lexington, Mass.:
Lexington Books. 1977. Reviews the literature on the effects of grants on local
spending. Concludes that matching grants are more stimulative than lump-sum grants
and the categorical grant has a larger stimulative effect than an equivalent increase in
private income.
Inman, Robert. "Editor's Introduction." Journal of Policy Analysis and Management 16.
no. 1 (1997), pp. 1-9.

Neenan, William B. "Fiscal Relations between State of Michigan and Southeastern
Michigan." In Financing the Metropolis, ed. Kent Mathewson and William B. Neenan.
New York: Praeger, 1980.
Public versus Private Schools
Barzel, Y "Private Schools and Public School Finance." Journal of Political Economy 81
(1973), pp. 174-86. Analyzes the exit of wealthy households to private schools when
the achievement levels of public schools decrease.
Coleman, James S.; Thomas Hoffer; and Sally Kilgore. Achievement in High School: Public
and Private Schools Compared. New York: Basic Books, 1981. Shows that students in
private schools have higher test scores than students in public schools. Also suggests
that achievement depends on the number of academic courses completed.
Gamoran, Adam. "Student Achievement in Public Magnet, Public Comprehensive, and

Private City High Schools." Educational Evaluation and Policy Analysis 8. no. 1 1
(Spring 1996), pp. 1-18.

Goldhaber, Dan D. "Public and Private High Schools: Is School Choice an Answer to the
Productivity Problem?" Economics of Education Review 15, no. 2 (1997). pp. 93-109.
Inman. Robert. "Optimal Fiscal Reform of Metropolitan Schools: Some Simulation
Results." American Economic Review 68 (1978), pp. 107-22. Discusses the choice
between private and public schools and the implications for education finance.
Murnane, Richard J. "Comparisons of Private and Public Schools: The Critical Role of
Regulations." Chapter 4 in Private Education: Studies in Choice and Public Policy, ed.
Daniel C. Levy. Oxford: Oxford University Press. 1986a.
."Comparisons of Private and Public Schools: What Can We Learn?" Chapter 5 in
Private Education: Studies in Choice and Public Policy, ed. Daniel C. Levy. Oxford:
Oxford University Press, 1986b.
Neal, Derek. "The Effect of Catholic Secondary Schooling on Educational Attainment."
Journal of Labor Economics 15 (1997). pp. 98-123.
Rouse, Cecilia. "Private School Vouchers and Student Achievement: An Evaluation of the
Milwaukee Parental Choice Program." Quarterly Journal of Economics 1 13. no. 2
(1998), pp. 553-602.
School Problems: Desegregation and Busing
"The Effect of School Desegregation on Housing Prices." Review of

Clotfelter, Charles T.
Economics and Statistics 57 (1975), pp. 446-5 1 Shows that desegregation in the South
.
decreased the demand for central-city housing, decreasing housing prices.

. "School Desegregation. Tipping, and Private School Enrollment." Journal of Human
Resources 1(1976), pp. 28-50. Shows that an increase in black enrollment in public
1
schools increases the proportion of whites enrolled in private schools.

."School Desegregation as Urban Public Policy." In Current Issues in Urban
Economics, ed. Peter Mieszkowski and Mahlon Straszheim. Baltimore: Johns Hopkins
University Press. 1979, pp. 359-88. Describes the evolution of desegregation policy,
explores the "white flight" phenomenon, and discusses the effects of desegregation on
achievement.
Coleman, James, and Sara Kelly. "Education." In The Urban Predicament, ed. William
Gorham and Nathan Glazer. Washington, D.C.: Urban Institute, 1976. Analyzes racial
segregation in schools between 1968 and 1973.
Giles, Michael W.; Douglas Gatlin; and Everett F. Cataldo. "White Flight and Percent
Black: The Tipping Point Re-Examined." Social Science Quarterly 56 (1975),
pp. 85-92. Using data for Florida, estimates that the tipping point for white enrollment
in private schools is about 30 percent black enrollment in public schools.
Education Vouchers
Bridge, Gary. "Citizen Choice in Public Services." In Alternatives for Delivering Public
Services, ed. E. S. Savas. Boulder. Colo.: Westview Press, 1977. Suggests that parents
are likely to pick education programs for them for the same
their children that prepare
type of occupation.
Coons, John E., and Stephen D. Sugarman. Education by Choice: The Case for Family
Control. Berkeley and Los Angeles: University of California, 1978. Discusses the
voucher program, arguing that a voucher system would provide better education
because parents and students would have greater opportunity for educational choice.
Epple, Dennis, and Richard E. Romano. "'Competition between Private and Public Schools,
Vouchers, and Peer-Group Effects." American Economic Review 88. no. 1 (1998),
pp. 33-62.
Levin, Henry. "Educational Vouchers: Effectiveness, Choice and Costs." Journal of Policy
Analysis and Management 17, no. 3 (1997). pp. 373-92.
Levin, Henry M. "Educational Vouchers and Social Policy." In School Finance Policies and
Practices, ed. James W. Guthrie. Cambridge, Mass.: Ballinger, 1980. Relates the
history of the voucher proposal and describes the features of the most recent voucher
proposals. Discusses the role of integrated classrooms for the promotion of
occupational mobility and the functioning of a pluralist democracy and suggests that a
voucher scheme is likely to cause a stratification of the education system.
Chapter 22
& RIME and PUNISHMENT
/ don 't worry- about crime in the streets; it 's the sidewalks I stay off of.
Johnson Letellier
C~~7~ his chapter presents the economic approach to crime and crime prevention.
iS The economic approach assumes that both criminals and victims are rational
in the sense that they base their choices on the expected benefits and costs of alter-
natives. A rational person commits a property crime if the expected benefit of the
crime exceeds the expected cost. Similarly, potential victims use their resources to
prevent crime if the expected benefit of prevention exceeds the expected cost. The
economic approach leads logically to a discussion of the optimum amount of crime:
because crime is costly to prevent, it is rational to allow some crime to occur.
Why study crime? The analysis of crime is an important part of urban economics
for three reasons. First, most crime occurs in metropolitan areas, and crime rates are
highest in central cities. Second, households are sensitive to crime rates, so their
location decisions are affected by local crime rates. In other words, crime affects
the spatial distribution of people within and between cities. Because households are
attracted to areas with low crime rates, an area with a relatively low crime rate has
relatively high housing prices. Third, the relatively high crime rates in central cities
have contributed to the suburbanization of population: many households have moved

to the suburbs to escape central-city crime.
Crime Facts
Most crime data come from the FBI's Uniform Crime Reports. The FBI collects data
from local police departments on seven index crimes. The index crimes are divided
into personal crimes and property crimes.
655
1 Personal crime. The victim of a personal crime is placed in physical

danger. For some personal crimes, the criminal's objective is to injure the
victim (homicide, rape, aggravated assault). For other personal crimes, the
objective is to steal property, but the criminal uses a show of force to coerce
the victim (robbery).
2. Property crime. Property crimes are crimes of stealth, not force. Examples
are burglary (illegal entry of a building), larceny (purse snatching, pocket
picking, and bicycle theft), and auto theft.
The FBI data provide a partial picture of the crime scene. Among the crimes omitted
in the Uniform Crime Reports are disorderly conduct, shoplifting, arson, employee
theft, and drug-related offenses (possession and sale of narcotics, public drunken-
ness, drunk driving).
Table 22-1 lists the crime rates for the seven index crimes for the last few
decades. The crime rates are expressed as the number of crimes per 100,000 pop-
ulation. Most of the reported index crimes are economically motivated. The four
economic crimes (robbery, burglary, larceny, and auto theft) make up about 90 per-
cent of total reported crimes. The total crime rate rose from 1,870 in 1960 to 5,950
in 1980, and then fell to 5,278 in 1995. Because the FBI data include only the
crimes that are reported to the police, they may be deceptive. About two-thirds of
all crimes —and over half of violent crimes — are not reported to police and do not
appear in the FBI data.
Since 1973, the Justice Department has collected crime data from its semi-
annual victimization surveys. The surveys suggest that the overall level of crime has
decreased since its peak in 1981.
Table 22-1 FBI Index Crimes, 1960- -1990
Number of Crimes per 100,000 People
I960 1970 1980 1990 1995
Personal Crime
Murder 5.0 7.8 10.2 9.4 8.2
Rape 9.5 18.6 36.8 41.2 37.1
Aggravated assault 85.2 176.9 298.5 424.1 418.3
Robbery 59.5 187.2 251.1 257.0 220.9
Property Crime
Auto theft 182 457 502 658 561
Larceny 1,024 2,124 3.167 3.184 3.045
Burglary 504 1.152 1.684 1.236 988
Total index crimes 1.870 3.949 5.950 5.820 5,278
Source: U.S. Federal Bureau of Investigation, Crime in the United States, 1995 (Washington.
DC: U.S. Government Printing Office, 1995)
Chapter 22 Crime and Punishment 657
The Victims of Crime

Who are the victims of crime? Victimization rates vary with income, place of resi-
dence, and race.
1 . Crime and income. The victimization rates for crimes of violence (rape,
robbery, and assault) decrease as income increases: the poor are more likely
to be victims of violent crime. For example, a person with income of
$7,500 is twice as likely to be a victim of violent crime as a person with an
income of $50,000. In contrast, victimization rates for property crime
increase with income. For example, compared to a person with annual
income less than $7,500, a person with income above $50,000 is 50 percent
more likely to be a victim of property crime.
2. Crime and place of residence. Table 22-2 shows victimization rates

for different residential locations. Crime rates are higher in metropolitan
areas than in nonmetropolitan areas, and higher in central cities than in
suburbs. Crime is concentrated in cities for three reasons. First, poverty is
concentrated in cities. Because a large fraction of criminals come from poor

households, the cities have a large supply of potential criminals. Second,
wealth is concentrated in cities, and criminals are attracted by urban loot.
Third, population density is higher in cities, providing criminals with a

large concentration of targets.
3. Crime and race. Table 22-2 shows the victimization rates for whites and
blacks. For all five crimes listed, blacks are victimized more frequently than
whites, with the biggest difference for robbery and auto theft.
Crime and the Price of Housing

Crime rates vary across space, causing variation in the price of housing. Gray and
Joelson (1979) studied the relationship between crime rates and property values in
Table 22—2 Victimization Rates in 1993
Household
Robbery Assault Theft burglary Auto theft
White 5 42 241 57 17
Black 13 51 250 86 34
Central cities 11 55 292 84 34
Suburban areas 5 39 241 47 17
Rural areas 3 37 190 54 7
Theft and burglary: Rate per 1 .000 households
Robbery and assault: Rate per 1.000 persons age 12 and over
SOURCE: U.S. Bureau of Justice. Criminal Victimization in the United States. 1993 (Washington. DC.
1995).
Minneapolis. In 1970, the burglary rate varied considerably across the city's census
tracts, from 1 percent of dwellings per year in the lowest-crime tract to 14 percent
in the highest. A one-unit increase in the crime rate (an additional percentage point)
decreased the average property value by $336. The vandalism rate varied from less
than 1 incident to 30 incidents per 1,000 population. A one-unit increase in the
vandalism rate (an additional incident per 1,000 population) decreased the average
property value by $1 17. In another study. Thaler ( 1978) estimated the elasticity of
property values with respect to crime rates as —0.067: a 10 percent increase in the
crime rate decreased the market value of housing by about 0.67 percent.
The Costs of Crime

Figure 22-1 shows the estimated costs of crime in 1992. The costs incurred by victims
include the value of lost property, medical expenses for injuries, the opportunity cost
of lost work time, the costs associated with the pain and suffering from victimization,
and the value of lives cut short. The costs of the criminal justice system include the
costs of police, the courts, and correction facilities. The cost of correction facilities
was $25 billion, or about $22,000 per prisoner. Citizens spent about $39 billion on
their own prevention measures, including locks and hired guards. The opportunity
cost of having 1.35 million people incarcerated (in jails and prisons) instead of
working is $46 billion. All together, the cost of crime is $250 billion per year, which
is 3.8 percent of GDP.
Figure 22-1 The Costs of Crime
Criminal justice
system
$74 billion
Opportunity cost of
incarcerated
$46 billion
Source: Freeman, Richard. "Why Do So Main Young American Men Commit

Crimes and What Might We Do About It'.'" Journal oj Economic Perspectives 10,
no I
(1996), pp. 25-42.
The Rational Criminal and the Supply of Crime

The economic model of crime assumes that the criminal commits a crime if the
expected benefit of the crime exceeds its expected cost. The model of the rational
criminal is relevant for economically motivated crimes (property crimes), but not
for crimes of passion and violence.
A Numerical Example of Burglary

Consider Boris, who must decide whether to spend a day planning and executing
a burglary. Boris will commit the burglary if the expected net benefit of burglary
exceeds the expected benefit of spending the day in some legal activity. The costs
and benefits of the two activities are listed in Table 22-3.
1 . The expected loot. The expected value of the burglary loot is the
probability that Boris succeeds in the burglary (P ) times the monetary 5
value of the loot:
EL = P S Loot (22-1)
If the probability of success is 80 percent and the resale value of the stolen
property is $600, the expected value of the loot is $480.
Table 22—3 Expected Benefits and Costs

for a Burglary
Benefits and Costs
Burglary Activity
Expected loot
Loot $ 600
Probability of success ( P s ) 0.80
Expected loot (EL — P, •
Loot) $ 480
Expected cost
Probability of punishment (P, = Pa Pp ) 0.02
Opportunity cost of prison time
Daily wage (W) $ 40
Expected workdays per year (D) 150
Forgone income per year (/ = W D) $6,000
Cost of lost freedom per year (F) $2,000
Opportunity cost per year (C = I + F) $8,000
Expected cost (EC = P, C) $ 160
Expected benefit of burglary (B = EL — EC) $ 320
Legal Activity
Daily income if employed (MO $ 40
Net return from burglary (R = B - W) $ 280

2. Expected cost. On the cost side, Boris faces the possibility of being sent to
prison for committing the burglary. The expected value of the penalty for
burglary is the probability of being imprisoned ( P, ) times the opportunity
cost of time spent in prison.
a. Probability of imprisonment. The probability of being imprisoned is
determined by the police and the court system. The probability of

imprisonment (P,) is the probability of being arrested (Pa ) times the
probability of being sentenced to prison once arrested (Pp ).
Pi = Pa Pp (22-2)
In this example, we will assume that the probability of imprisonment

is 0.02.
b. The opportunity cost of prison time. The opportunity cost of time
spent in prison equals the annual opportunity cost times the length of the
prison sentence. Suppose that Boris is a relatively low-skilled worker
who can work only about 50 days per year and is paid $40 per day.
find 1
For every year in prison, he gives up $6,000 in income (150 times $40).
Another cost of imprisonment is the loss of liberty and freedom. If he
values his freedom at $2,000 per year, the annual opportunity cost of
prison is $8,000 ($6,000 in forgone wages plus $2,000 in lost freedom).
If the prison sentence for burglary is one year, the opportunity cost of
imprisonment is $8,000.
c. Expected cost. The expected cost is $160 [the probability of being
imprisoned (0.02) times the opportunity cost of imprisonment for
burglary ($8,000)].
3. Expected benefit from burglary. The expected benefit from burglary is the
expected loot ($480) less the expected cost ($160), or $320.
4. Net return from burglary. The alternative to burglary is to spend the day
The net return from a burglary is the expected benefit of the
in a legal job.
burglary ($320) less the money Boris could earn in a legal job ($40), or
$280. The net return is positive, so if Boris bases his decision exclusively
on the expected benefits and costs of burglary, he will commit the
burglary.
Morality and Aversion to Crime? Most people have an underlying aversion to

committing crime. For the typical person, crime is unattractive at any price. For other
people, the net return on crime must be high enough to overcome their underlying
aversion to it. One way to incorporate the aversion to crime into the benefit-cost
analysis is to estimate a person's anguish cost, defined as the cost associated with
engaging in antisocial activity. If Gabriel has a strong aversion to crime, his anguish
cost may be $10,000, meaning that he will commit a crime only if the net return
exceeds $10,000. If Lucifer is less averse to crime, his anguish cost is only $10, so
he will commit a crime if the net return exceeds $10. Returning to the example in
Table 22-3, Boris will commit the burglary if his anguish cost is less than the net
return of the burglary ($280).
Who Commits Crimes? The model of the rational criminal provides three rea-
sons why some people commit crimes. some people are relatively skillful at
First,
committing crime and escaping punishment. For such people, the expected loot is
relatively large and the expected cost of crime is relatively small. Second, some peo-
ple have a relativelylow opportunity cost of time spent committing crime and time
is relatively low. For example, the poor
spent in prison, so the expected cost of crime
may have a lower opportunity cost because they are unemployed or earn low wages.
Third, some people have less respect for society and are less averse to committing
crimes, so they need a relatively small net return to make crime worthwhile. For
most people, the moral anguish associated with crime is large enough to make crime
unattractive at almost any price.
The Crime-Supply Curve

The model of the rational criminal can be used to derive the supply curve for burglary.
The supply curve shows the relationship between the net return or "price" of burglary
and the number of burglaries committed.
Figure 22-2 shows a city's supply curve for burglary. The horizontal axis mea-
sures the number of burglaries committed in the city per month, and the vertical axis
Figure 22-2 The Supply Curve for Burglary
Supply with
greater aversion
Supply with
moderate aversion
20 40 60 80 100
Number of burglaries per month
The supply curve shows the number of burglaries for different net returns ( R) on crime. It has a
positive intercept because of anguish cost and is positively sloped because an increase in the net
return increases the number of burglars and increases the number of burglaries per burglar. An
increase in the aversion to crime shifts the supply curve upward.
measures the net return (R) of burglary. The supply curve for moderate aversion has
a positive intercept of $100 because of anguish cost: for the first burglary, the net
return must be high enough to exceed the anguish cost of the first burglary ($100).
The curve is on burglary increases, more bur-
positively sloped: as the net return
glaries are committed because ( 1 the net return exceeds the anguish cost for more
)
potential criminals, so more people become burglars and (2) burglars commit more
crimes per month. When the net return increases from 400 to 600, the number of
burglaries increases from 60 per month to 100.
The supply curve shows the effects of crime-fighting policies on the crime rate.
Suppose that the city increases its police force and thus increases the probability of
a burglar being arrested. The increase in the probability of arrest decreases the net
return on burglary, so the city moves down its supply curve to a lower crime rate.
The crime rate decreases as (1) each criminal commits fewer offenses and (2) some
criminals exit from the crime business because the net return of crime falls short of
their anguish cost. Marginal criminals "go straight" if the return on crime is not high
enough to offset their underlying aversion to crime.
The aggregate supply curve shifts when the underlying aversion to crime
changes. In Figure 22-2, an increase in the aversion to committing crime shifts
the supply curve upward: there are fewer crimes at each net return. For example,
the crime rate for a net return of $500 drops from 80
to 40. Because an individ-
ual's aversion to on age, changes in the age distribution of
crime depends in part
the population shift the supply curve. Young males commit more than their share
of crimes. Therefore, a society with a relatively large number of young males will
have relatively high crime rates. Recent shifts in the age distribution of the U.S.
population have decreased the proportion of young males. As the proportion of the
population in the crime-prone years decreased, crime rates fell.
Optimum Amount of Crime

Because crime prevention is costly, the optimum amount of crime is positive. In other
words, it is rational to tolerate some crime. How much crime should be allowed?
Victim Cost and Prevention Cost

The social costs of crime can be divided into victim cost and prevention cost. The
\ ictims of property crime lose property and are sometimes injured in the process.
Figure 22-3 shows total victim cost as a function of the number of burglaries. The
victim-cost curve is positively sloped and linear, reflecting the assumption that the
victim cost per burglary is independent of the number of burglaries: if the number
of burglaries doubles, victim cost also doubles.
Although society can decrease crime by decreasing the net return on burglary,
crime prevention is costly. There are a number of options for crime prevention:
Figure 22-3 The Optimum Number of Burglaries
100
Number of burglaries
The total social cost of burglaries equals the sum of the cost incurred by victims and the cost of prevention.
Victim cost increases at a constant rate if the cost per crime is constant. The prevention cost increases rapidly as
the number of burglaries decreases, a result of diminishing marginal returns. Total cost is minimized with B*
burglaries (point G).
1. Hardening the target. Victims can decrease the expected loot by

decreasing the probability of successful burglary (e.g., installing locks and
using guard dogs).
2. Increased probability of arrest. The police can increase the expected cost
of crime by increasing the probability of arrest (e.g., hiring more police
officers).
3. Increased probability of imprisonment. The criminal justice system can

increase the expected cost of crime by increasing the probabilities of
conviction (e.g., hiring more investigators to gather evidence on crime).
4. Increased severity of punishment. The criminal justice system can
increase the expected cost of crime by lengthening prison sentences (e.g.,
building more prison cells).
5. Increased value of legal opportunities. Society can increase the

attractiveness of legal activity by increasing the job skills of potential
criminals, thus increasing their legal wages. As the opportunity cost of
crime increases, the net return on burglary decreases.
In Figure 22-3 the prevention-cost curve is negatively sloped. All of these

preventive measures are costly, so the larger the number of crimes prevented (the
fewer the burglaries committed), the higher the prevention cost. The prevention-
cost curve increases in slope as the number of burglaries decreases, reflecting the
assumption of diminishing marginal returns to crime prevention: as prevention efforts
increase and the number of crimes decreases, it becomes progressively more difficult
to prevent more crime. For example, it is relatively easy to decrease the number of
burglaries from 100 to 99, but more difficult to decrease the number from 1 1 to 10.
The total cost of crime is the sum of the victim cost and the prevention cost.
In Figure 22-3. the total-cost curve is U-shaped, reflecting the trade-offs between
victim and prevention cost. Suppose that a city starts with zero prevention cost, so
it has 100 burglaries per day and the total cost of crime is the victim cost (point
F). The city could decrease the number of burglaries and decrease victim cost by
spending a small amount on prevention (e.g.. hardening targets, lengthening prison
sentences, improving legal opportunities). By doing
would decrease the total
so. it
cost of crime: starting from point F, the savings in victim cost would exceed the
increase in prevention cost, so the city would move down the total-cost curve. The
total cost of crime is minimized at point G. so the optimum crime rate is B*.
How does the city persuade burglars to commit the optimum number of burglar-
ies (5*)? The city would use its crime prevention resources to decrease the net return
to burglary to the point at which burglars commit B* burglaries per day. In other
words, the city picks the point on the burglary-supply curve and invests resources to
bring the net return down to the necessary level.
Marginal Cost and Variation in Optimum Crime Levels
The optimum crime rate can also be derived from the marginal-cost curves underlying
the total-cost curve in Figure 22-3. In Figure 22^4, the lower horizontal line shows
the marginal victim cost for burglary, defined as the increase in victim cost per addi-
tional burglary (the slope of the total-victim-cost curve in Figure 22-3). This horizon-
tal cost curve reflects the assumption that the cost per burglary victim is constant. The
negatively sloped curve shows the marginal prevention cost, defined as the change in
prevention cost per unit change in the number of crimes (the slope of the prevention-
cost curve in Figure 22-3). It is negatively sloped because there are diminishing
returns to crime prevention: the first unit of crime prevention (decreasing the num-
ber of crimes from 100 to 99) has a relatively low marginal cost: the last unit of crime
prevention (decreasing thenumber of crimes from to 0) has a high marginal cost.
1
The optimum amount of crime is shown by the intersection of the marginal-

victim-cost curve and the marginal-prevention-cost curve. Starting with 100 burglar-
Figure 22-^4 Marginal Victim and Prevention Costs and the Optimum
Crime Level
Marginal
prevention cost
Marginal victim cost

for armed robbery
Marginal victim
cost for burglary
Number of crimes
The optimum crime level is where the marginal victim cost equals the marginal prevention cost B* ( for
burglaries and A* for armed robberies). The optimum number of robberies is smaller because robbery has a
higher victim cost.
ies, the marginal victim cost for burglary exceeds the marginal cost of prevention, so
it is sensible to deter the 100th crime. Prevention is efficient as long as the marginal
victim cost exceeds the marginal prevention cost, so the optimum burglary rate
is B*.
The marginal-cost curves explain why optimum crime rates vary across crimes.
Suppose burglary and armed robbery have the same prevention cost, but different
victim costs: the victims of armed robbery are often injured during the crime, so
armed robbery has a higher marginal victim cost. In Figure 22-4, the optimum
crime rate for robbery (A*) is less than the optimum crime rate for burglary: more
resources are invested in robbery prevention because the savings in victim cost
per robbery are higher. Optimum crime rates also differ because of differences in
prevention cost: crimes with higher marginal prevention costs have higher optimum
crime rates, everything else being equal.
The same arguments can be used to explain why crime rates are higher in some
central-city areas. These areas have large concentrations of poor and unemployed
people, so the opportunity cost of crime is relatively low and the number of potential
criminals is relatively large. Given the plentiful supply of criminals, the marginal
prevention cost is relatively high. If the marginal victim cost is the same throughout
the metropolitan area, the optimum crime rate is higher in the central city.
Crime Prevention Activities

This section discusses some of the details of crime prevention, exploring the effects
of various "carrot" and "stick" approaches to preventing crime. The first part of
the section discusses carrot policies: by increasing the value of legal opportunities,
the government increases the incentives to choose legal activities over crime. The
second part shows how potential victims can protect themselves from crime. The
final four parts examine stick policies: by increasing the likelihood and severity of
punishment, the government deters crime by increasing the expected cost of crime.
Increasing the Value of Legal Opportunities
The rational criminal chooses crime instead of legal activities because the expected
benefit from crime is higher. One way to discourage crime is to increase the payoff
from legal opportunities. Conversely, we expect relatively high crime rates when
wages are relatively low or the unemployment rate is relatively high.
There is a negative relationship between crime rates and wages paid for legal
activities (Freeman, 1996; Grogger. 1994). In the last two decades, the real earnings
of low-skilled workers has dropped by 20 to 30 percent (Freeman. 1996). As ex-
plained in Chapter 13, the drop in real wages was caused by changes in the economy
that decreased the demand for low-skilled workers. Grogger (1994) suggests that
the decrease in wages during this period was an important factor in the 18 percent
increase in the arrest rate for young men.
What is the relationship between unemployment and crime? Freeman (1983)
summarizes the evidence from dozens of studies of the relationship and comes to two
basic conclusions. First, although there is a positive relationship between crime and
unemployment, the relationship is relatively weak. Second, the crime rates of first-
time offenders are relatively sensitive to the unemployment rate. This conclusion
is consistent with the results of Fleisher (1966) and Phillips. Votey. and Maxwell
( 1972), both of whom found a strong relationship between youth unemployment and
crime rates. A similar conclusion was reached by the National Bureau of Economic
Research study of youth unemployment (Freeman and Holzer. 1986): teenage crime
rates were higher in cities with fewer legal opportunities.
There are other public policies that could improve the prospects for the young
males who comprise a large fraction of the criminal population. As explained earlier
in the book, the key to economic success is education, and investments in central-city
education could increase the payoff from legal activities and thus decrease crime
rates. In addition, racial segregation in housing increases poverty and encourages
criminal activity, so policies that combat segregation could reduce crime rates. The
general problem of decreasing real earnings of less skilled workers presents a con-
siderable policy challenge (Freeman, 1996):
How to improve the job market for less skilled young American men. and reverse the huge
decline in their earnings and employment opportunities, is the problem of our times, with
implications both for crime and many other social ills.

Table 22-4 Protective Measures for $100

Bicvcle
Protective Protection Probability Expected Total

Action Cost of Theft Theft Cost Cost
No protection $0 0.30 $30 $30

Flimsy lock 11 0.10 10 21
Sturdy lock 24 0.02 2 26
Armed guard 80 0.00 80
Victims as Crime Fighters
Potential victims can discourage crime by decreasing the thief's expected benefit.
There are two strategies for decreasing the expected loot: hardening the target and
decreasing the market value of the stolen property.
Hardening the Target. The potential victim can harden the target by investing in
protective devices. Alice's options for bike protection are listed in Table 22-4. If she
does not spend any money on protection, there is a 30 percent chance that her $100
bike will be stolen, so her expected loss from theft is $30 (0.30 times $100). If she
locks her bike with a flimsy lock, the probability of theft drops to 10 percent; if she
locks her bike with a sturdy lock, the probability drops to 2 percent. If she hires an
armed guard, the probability of theft is zero.
What is the optimum level of protection? Alice's objective is to minimize the
total cost of protecting her bike, equal to the sum of protection cost and the expected
loss from theft. She minimizes total cost with the flimsy lock: her protection cost
is $11 and the expected loss is $10, for a total cost of $21. Although a sturdy
lock would decrease the probability of theft, the marginal benefit of the additional
protection (the $8 decrease in the expected cost) is less than the marginal cost (the
$13 increase in protection cost). Alice tolerates a 10 percent chance of having her
bike stolen because the marginal cost of additional protection exceeds the marginal
benefit.
The optimum level of protection depends on the value of the item being pro-
tected. If Alice replaces her $100 bike with a $200 bike, she will upgrade her pro-
tection to a sturdy lock: the switch to the more expensive bike doubles the expected
loss from theft, so the marginal benefit of the sturdy lock ($16) exceeds its marginal
cost ($13). In general, people with more property to protect spend more money on
protection.
How do people and firms harden crime targets? Most people lock their cars
and houses, and some install burglar alarms. Purchasers of expensive car-stereo
systems can buy plastic covers that make the stereos look cheap and unattractive.
Most businesses keep their valuables in safes, and install alarms to deter robbers.
Convenience stores keep their money in timed safes so clerks (and thieves) have
access to a limited amount of cash.
668 Part VI Education ami Crime
Protection and Insurance. How does insurance affect Alice's spending on pro-
tective measures? Suppose that she insures her $100 bicycle against theft, and there
are no costs associated with filing a claim (no transactions cost and no deductible
payment). If so, the total cost of leaving her bike unprotected is zero: every time her
bike is stolen, she collects $100 from her insurance company and then buys a new
bike. Therefore, Alice will leave her bike unprotected.
This moral-hazard problem. If Alice buys insurance, she reduces her own
is the
protective measures and is more likely to be victimized. The moral-hazard problem
explains why many people do not lock their cars. According to the FBI, 80 percent
of stolen cars are left unlocked, and 40 percent have the keys in the ignition. The
moral-hazard problem also explains why most insurance policies have deductibles:
the deductible increases the property owner's expected loss from theft, encouraging
the owner to invest in protective measures. Recently, some insurance companies have
tried to control the moral-hazard problem by providing free home-security systems
to their policyholders.
Market Value of Loot. Potential victims can also decrease the resale value of
stolen property. As the resale value of the stolen property decreases, the net re-
turn from theft decreases, discouraging crime. A potential victim can decrease the
likelihood of being victimized by making his property less valuable to criminals.
The resale value depends on how easily the loot can be identified as stolen prop-
erty. A person who purchases stolen property is subject to prosecution for receiving
stolen goods. Given the risks associated with buying stolen goods, a consumer is
willing to pay less for a good that is easily identified as "hot." The more readily the
stolen goods can be identified, the lower the resale value. The typical criminal sells
his loot to a fencing operation and is paid between 20 percent and 50 percent of the
legalmarket value of the goods.
Suppose that Natasha can steal either a hundred-dollar bill or an original painting
with a legal market value of $100. What will she steal? The resale value of the
painting is less than $100 because the purchaser of the painting could be arrested for
the possession of stolen property. In contrast, the "resale" value of the cash is equal
to its face value: given the number of hundred-dollar bills in circulation, it would be
impossible to find and punish the person who "buys" the stolen hundred-dollar bill.
Therefore, Natasha will steal the cash, not the painting.

How do potential victims facilitate the identification of stolen goods? In the
Wild West, ranchers branded their cattle to discourage cattle rustling. More recently,
ranchers have experimented with implanting microchips into cow hides to iden-
tify and locate stolen cattle. Many communities have developed programs under
which residents mark their possessions with identification numbers (social security
numbers, driver's license numbers) to discourage burglary. Participants in Opera-
tion Identification place warning labels in conspicuous places to inform potential
burglars that "all items of value have been marked for ready identification."
The Police
One measure of the effectiveness of police is the arrest ratio, defined as the number
of arrests divided by the number of crimes committed. The arrest ratios for property
crimes such as burglary, larceny, and auto theft are in the range of 0.13 to 0.18,
and the arrest ratio for robbery is about 0.27 (Levitt, 1998). Arrest ratios for violent
crimes aremuch higher, (e.g., 0.42 for aggravated assault and 0.82 for murder).
One way to deter crime is to have a relatively high arrest ratio. From the criminal's
perspective, the arrest ratio is the probability of being arrested for a crime, and the
higher the probability, the lower the expected payoff from crime. According to a
recent study, the elasticity of crime rates with respect to the arrest ratio is around
—0.30 for robbery, burglary, and larceny (Levitt, 1998a). This means that a 1 percent
increase in the arrest ratio will decrease crime rates by about 3 percent. In earlier
studies, the estimated elasticities for property crimes were around —0.50 (Ehrlich,
1972; Wilson and Boland, 1976; Bartel, 1979; Witte, 1980).
Other studies explore the relationship between spending on police activities
and criminal activity. Tauchen (1994) shows that additional spending on the police
decreases the likelihood that youths will commit crime. A 10 percent increase in
police spending increased the average length of time a youth is a law-abiding citizen
by about 4.7 percent.
Police Patrol. The police spend a lot of time patrolling streets and sidewalks. The
idea behind patrol is that the police should be in a position to respond quickly when
crime occurs. Does patrol deter crime? A number of studies show that car patrol does
not have a significant effect on crime rates. In a famous experiment in Kansas City,
some neighborhoods and decreased patrolling
the city increased patrolling activity in
in others.There were no significant changes in either arrest ratios or crime rates in
any of the neighborhoods, suggesting that patrolling does not decrease crime. Other
experiments have come to the same conclusion (Sherman, 1983).
Why doesn't car patrol work? One reason is that crime victims are slow to
report crimes to the police. According to Sherman, the typical victim takes about
50 minutes to report the crime to police. Suppose that an increase in patrolling could
decrease the response time of police (defined as the period of time between receiving
a report of the crime and arriving on the scene) from 10 minutes to 2 minutes. If the
typical victim takes 50 minutes to report the crime, the police could arrive on the
crime scene 52 minutes after the crime is committed instead of 60 minutes. In either
case, the criminal's trail is already cold, so the improved response time is unlikely to
increase the arrest ratio. On the other hand, if the victim took only 1 minute to report
the crime, an increase in patrolling could make a difference: the police would arrive
3 minutes after the crime instead of 1 1 minutes and would have a better chance of
capturing the criminal.
What about foot patrol? A number of studies show that foot patrol can in-
crease arrest ratios and deter crime. Police on foot patrol are constantly gathering
670 Pari VI Education ami Crime
information about the people and businesses in their territories. The foot-patrol
knows who is arguing with whom, who is suddenly living beyond his means,
officer
and whose children are running amok. As a result, the foot-patrol officer is better
equipped to prevent crimes and to solve crimes once they occur.
Police Investigation. In addition to arresting criminals, police gather information

for public prosecutors and the courts. The success of investigations
is measured
by the clearance rate, defined as the fraction of arrests that result in convictions.
According to Forst (1983) about 32 percent of felony arrests result in conviction.
The most important investigative task is to find eyewitnesses to the crime. One rule
of thumb is that if the police provide two eyewitnesses to a crime, a conviction is
virtually guaranteed.
Under the traditional investigative system, crimes are investigated by police
detectives rather than patrol officers. The patrol officer usually makes the arrest,
prepares a report on the crime, and then hands over the investigation to a detective.
Some police forces have experimented with alternative investigative systems.
Under a community service system, each police officer is assigned a small territory
and is responsible for both making arrests and investigating crimes. Instead of simply
filing a report about the crime, the arresting officer finds eyewitnesses and gathers
other evidence for prosecutors. The idea behind this system is that the local patrol
officer has a great deal of information about his territory that would be useful in the
investigation. There is some evidence that the community service approach increases
clearance rates.
Other cities have experimented with different incentive systems. Under the tra-
ditional system, police rewards are based on the number of arrests, not on clearance
rates or local crime rates. If police were rewarded for increasing clearance rates and
decreasing crime rates, they might have a greater incentive to follow up arrests by
gathering information and finding eyewitnesses. The city of Orange, California, has
experimented with a system under which police salaries are tied to crime rates (see
Reynolds. 1986).
The Court System

Forst (1983) shows how the court system handles a typical set of 100 felony arrests.
His results are summarized in Figure 22-5. Only 9 of 100 felony arrests result in a
prison sentence longer than a year. The process by which the 100 arrests are whittled
down to nine imprisonments is as follows:
1 Juvenile cases. About one-third of the offenses are committed by juveniles

and are handled by the juvenile justice system. Little is known about the
disposition of these cases.
2. Dismissed by prosecutor. Of the 65 cases presented to the district attorney.

25 are dropped: most cases are dropped because of weak evidence: a small
number of cases are dropped because the offense is trivial. Only about 1
Figure 22-5 Typical Disposition of 100 Felony Arrests,

1974-1984
100
felony
arrests
35-*-^ ^-65
Juvenile justice Presented to
system district attorney
25 ~<T ^>-40
Rejected or Accepted for
dismissed by prosecution
prosecutor
6 -*-' ^-34
Dismissed by Plead guilty (27)
judge (4) or fail or go to trial (1)
to appear (2)
2~*T "~^32
Aquitted in Convicted (5 in
trial trial)
\2~<< "->~20
Probation Incarcerated
11'*-' "~*-9
Jailed (less Imprisoned
than 1 (greater than 1
year) year)
SOURCE: Brian Forst, "Prosecution and Sentencing," Chapter 10 in Crime and Public
James Q. Wilson (San Francisco: Institute for Contemporary Studies, 1983),
Policy, ed.
percent of cases are dropped because of violations of rules concerning

arrest procedures and evidence.
3. Dismissed by judge. Of the 40 cases accepted by the district attorney, 4 are
dismissed by the judge, usually because the evidence is insufficient or the
offense is trivial.
4. Disappearing defendant. Two cases are eventually dropped because the

defendant disappears after being released on bond, personal recognizance,
or third-party custody.
5. Plea bargaining. Of the 34 remaining cases, 27 of the defendants plead
guilty, usually to a lesser charge.
6. Trial acquittal. Of the seven cases that go to trial, two result in acquittal
and five result in a guilty verdict.
Table 22-5 Plea Bargaining
Armed Robberx Petit Larcem
Penaltv 3 years 1 year

Probability of conviction 0.50 1.0
Expected penaln 1 .5 years 1 .0 \ear

Court costs $2,000 $300
Defense costs SI. 000 S200
7. Probation. Of the 32 defendants found guilty (27 from plea bargains and 5
in trials). 12 receive probation (supervised release).
8. Jail versus prison. Twenty convicts are incarcerated. Eleven serve time in
jails (receiving sentences of less than one year), and nine are imprisoned
(receiving sentences of more than one year).
Plea Bargaining. As shown in Figure 22-5. of the 32 cases that result in convic-
tion, only 5 cases are actually decided by a trial. The other 27 cases are resolved
through the plea-bargaining process.
The district attorney has broad powers to negotiate with the accused oxer the
criminal charge and the resulting punishment. Under the plea-bargaining process,
the district attorney reduces the criminal charge and the accused pleads guilty to the
lesser charge. The plea bargain generates benefits for both sides: the district attorney
gets a certain conviction, and the defendant receives a relatively light sentence. In
addition, both parties avoid the cost of a criminal trial.
Table 22-5 provides a simple example of the plea-bargaining process. Suppose

that Bonnie is accused of armed robbers . a crime that carries a three-) ear prison
sentence. Given the evidence in the case, both Bonnie and the district attorney (DA)
agree that there is a 50 percent chance that a criminal trial will produce a guilty
verdict. Therefore, the expected penalty from the armed robbery charge is 1.5 years
(3 years times 0.50). The alternative to a trial on the robbery charge is a plea of guilty
to petit larceny, a crime that carries a one-year prison sentence.
Should Bonnie accept the plea bargain? Suppose that a criminal trial would cost
Bonnie $1,000 in legal fees, but the plea bargain would cost only $200. She would
be better off with a guilty plea for two reasons. First, she saves $800 in legal costs
($1,000 — $200). Second, the expected prison sentence is shorter ( 1 year versus
1.5 years). If the DA offers this plea bargain. Bonnie will accept it.
Should the district attorney offer the plea bargain or pursue a trial on the charge
of robbery Suppose that a criminal trial would cost taxpayers S2.000. but the plea
'
bargain would cost only $300. From the perspective of law-abiding citizens who
wish to see Bonnie punished for her crime, the plea bargain produces good news
and bad news. The good news is that the plea bargain saves taxpayers SI. 700. The
bad news is that the plea bargain decreases Bonnie's expected prison time by half
a year. If society is willing to give up half a year of punishment to save $ 1 ,700, the
district attorney should offer the plea bargain.
Under what circumstances will Bonnie and the DA not agree to a plea bargain,
but instead bring the case to trial? Suppose that both sides are optimistic about their
chances of winning the court battle. Bonnie assumes that the probability of a guilty
verdict is very low (10 percent), so the expected penalty from a robbery trial is 0.30
years, compared to a certain penalty of 1 year under the proposed plea bargain. She
will demand a trial if the decrease in expected punishment (0.70 years) is worth at
least $800 (the increase in legal cost if the case goes to trial). If the DA assumes
that the probability of a guilty verdict is very high (90 percent), the expected penalty
from a trial is 2.7 years, compared to a certain penalty of 1 year under the proposed
plea bargain. The DA will seek a trial if the extra 1 .7 years in expected punishment
is worth at least $1,700. In this example, Bonnie's case will go to trial because the
two parties disagree about the likely outcome of the trial.
Proof of Guilt. In any justice system, two types of errors are inevitable. First,
some innocent people are found guilty and punished for crimes they did not commit
(a false positive result). Second, some guilty people are found innocent and escape
punishment (a false negative result).
The U.S. courts have developed a set of rules that place the burden of proof on
the shoulders of the prosecutors. To convict a person of a crime, the state must prove
the person's guilt "beyond a reasonable doubt." Under the reasonable-doubt rule, the
accused is presumed innocent until proved guilty, and close cases are decided in favor
of defendants. The rule decreases the frequency of false positives (fewer innocents
are convicted), but increases the frequency of false negatives (more guilty people are
acquitted). In other words, the courts err on the side of innocence rather than guilt.
Other court systems err on the side of guilt rather than innocence. If the accused
is presumed guilty until proved innocent, the burden of proof is on the shoulders of
the accused. Under this type of rule, close cases are decided in favor of the state. The
adoption of this rule would increase the number of false positives (more innocents are
convicted) and decrease the number of false negatives (fewer guilty are acquitted).
The Principle of Marginal Deterrence

The penalties for various crimes are determined by legislators, prosecutors, and
judges. Legislatures pass laws that establish minimum and maximum penalties,
providing guidelines for prosecutors and judges. In the plea-bargaining process, the
prosecutors negotiate with suspects over criminal charges and penalties and then pass
on recommendations to judges, who usually accept the plea bargains. If a case
their
goes to and the defendant is found guilty, the presiding judge sets the penalty.
trial
In determining the penalties for different crimes, legislators and judges are
guided by the principle of marginal deterrence. According to this principle, the
penalty for a particular crime should depend on the seriousness of the crime, with
greater penalties for more serious crimes.
Table 22-6 Marginal Deterrence
Burglar) Burglary
Armed Robbery 1 1 year) 1 2 xcars i
Expected loot $2,000 $1,700 $1,700

Probability of punishment 0.10 0.10 0.10
Prison sentence 2 years 1 year 2 years
Annual opportunity cost of prison S8.000 $8,000 S8.000
Expected prison cost $1,600 $800 $ .600
1
Net benefit $400 $900 $100
A Numerical Example. The principle of marginal deterrence can be explained

with a simple example. Suppose that a state initially has a one-year penalty for
burglary and a two-year penalty for armed robbery. Robbery carries a greater penalty
because robbery has a higher social cost: the victims of robbery are often injured or
killed, while the victims of burglary simply lose their property.
Emil is a career criminal who must decide whether to commit a robbery or a
burglary. He will commit the crime that provides the larger net benefit. His options
are shown in Table 22-6. Robbery generates a greater expected loot ($2,000 versus
$1,700). but also carries a higher expected cost ($1,600 versus $800). The expected
net benefit of burglary is $900 compared to $400 for armed robbery, so Emil picks
burglary rather than robbery.
The state's penalties for robbery and burglary provide a marginal deterrent for
robbery, the more serious crime. In deciding on the level of crime. Emil weighs
the marginal benefit and marginal cost associated with upgrading from burglary
to robbery. The marginal benefit of upgrading is the increase in the expected loot
($300). The marginal cost is the increase in the expected prison cost ($800). Since
the marginal benefit is less than the marginal cost. Emil sticks with burglary. Society
deters Emil from the more serious crime by imposing a larger penalty on robbery.
Suppose that the state passes a law that equalizes the penalties for burglary and
robbery at two years. The increase in the burglary penalty decreases the expected
net benefit from burglary to $100 ($1,700 - $1,600). but does not affect the net
benefit from robbery. Since the net benefit of robbery now exceeds the net benefit
of burglary. Emil switches to robbery. The equalization of penalties eliminates the
marginal deterrent: the marginal benefit of moving from burglary to robbery is still
$300. but the marginal cost of upgrading (the increase in the expected penalty)
is zero. If Emil receives the same penalties for the two crimes, he will commit the
more lucrative crime. The equalization of penalties eliminates the marginal deterrent
effect, so Emil switches to the crime with a higher social cost.
The General Effects of Equalizing Penalties. Figure 22-6 shows the effects of
increasing the burglary penalty on the number of burglaries and robberies. Suppose
that initially the two crimes have the same net return (/?'). Given the two supply
curves, there are B' burglaries and A' armed robberies. When the state increases the
CIS
u o
B u
Xu 3
X o
u
c p
X;
0)
o
u u
X i
u = SP
e -
< B X3
u c
U £ OJ
C u «
x
xa
5 u
u u
X) B T3
XI •< 4J
s — B
5
a
o 1
•S
u u
i
s a —
< i
. Z X u
< a •c
6a
g
XX
X
E
o
1)
5 c
1 —3 u
o B a
E u
as
>
o X
E B p
=
u 03 t?
— g 0)
X
= X
'J p
o
>
cd
i
Xa 3
y o
X
X u uc
— X= a
</3 p
U
160 B a 06
B
OQ 3 X
XI c
E? u 3
Xa CO aB
o>
X) ^ y a
E XQ — > :
3 3 X_^00
; Z g
9
6s
U
>.'&
aa c
•c
X 3. §
u
—c
'3
so
«
QJ
S
Wj
u bfl 5 £
3d 3
-C
aO s
Qfi
"5
§ 3-°
a- o;
X
-5 ir s
675
676 Pan VI Education and Crime
penalty for burglary, the net return drops to R". decreasing the number of burglaries
to B". The decrease in the net return to burglary causes some criminals to upgrade
to robbery, so the robbery supply curve shifts to the right. The net return to robber)
is still R '. number of robberies increases to A". Because the increase

so the in the
burglar) penalty weakens the marginal deterrent, some criminals upgrade to the
more serious crime.
The increase in the burglary penalty decreases the total number of crimes: the
decrease in the number of burglaries (from B' to B") exceeds the increase in the
number of robberies (from .4 to A This occurs because the increased penalty
).
causes some burglars to stop committing crimes. In other words, only some of the
burglars who change their behavior shift to robbery.
Is society better or worse off under the higher burglar) penalty' 1 It depends on
the social cost of the two crimes. Because victims are often injured or killed in rob-
beries, the social cost of robbery exceeds the social cost of burglary. According to
Phillips and Votey 1979) the social cost of the typical armed robber) is about S5.400
(
and the social cost of the typical burglar) is only about S900. In Figure 22-6. the
increase in the number of robberies is about one-fourth the decrease in burglaries.
If the social cost of robbery is six times the social cost of burglar), the total social
cost of crime increases. In this case, the increase in the burglar) penalty increases
the social cost of crime.
There are two lessons for crime because stiffer penalties provide
fighters. First,
a marginal deterrent, an increase in the penalty for one crime causes some criminals
to upgrade to a higher-level crime. Second, policymakers could use this approach
to estimate the effects of changes in penalties on the social cost of crime. To do so.
they would need information on the social cost of different crimes and the criminals'
responses to the change in penalties. The principal questions are: how many criminals
would be deterred: how many would upgrade to a higher-level crime: and how many
would downgrade to a lower-level crime?
The Prison System

The prison system has four functions. The first is the rehabilitation function, which
is based on the idea that society can use education and counseling to persuade
convicts to "go straight." The second function is deterrence: by punishing convicts,
society makes the threat of punishment credible, discouraging other people from
committing crime. The third function is incapacitation: prisons isolate criminals
from their potential victims, decreasing crime by taking criminals out of circulation.
A final function is retribution or revenge: law-abiding citizens like to see criminals
punished for their actions.
Several studies explored the effects of the size of the prison system on crime
rates. The larger the capacity of prisons and the number of criminals in prisons, the
greater the deterrent effect and the incapacitation effect. The elasticity of crime rates
with respect to the number of prisoners is about —0.50: a 10 percent increase in the
number of prisoners decreases crime rates by about 5 percent (Levitt. 1996).

Chapter 22 Crime and Punishment Oil
Between 1975 and 1995, the number of people in prison roughly tripled (Free-
man, 1996). Given the large deterrent and incapacitation effects of prisons (an elas-
ticity of —0.50), we would expect a substantial reduction in crime rates. The fact
that crime rates didn't fall by very much suggests that something else was going on.
As explained earlier in the chapter, that "something else" was a dramatic drop in the
earnings of low-skilled workers. Although the higher imprisonment rate decreased
the expected payoff from criminal activity, the drop in earnings decreased the payoff
from legal activities, so crime rates decreased by relatively small amounts.
The Rehabilitation Function. According to the popular view, society can reha-
bilitate convicts.The purpose of education and counseling programs is to provide
convicts with the skills and the desires to become lawful citizens. The success of a
particular rehabilitation program is measured by its effect on the recidivism rate,
defined as the percentage of convicts that are arrested for another crime after they
leave prison. The overall recidivism rate is about 80 percent: four-fifths of convicts
are arrested after they leave prison.
Do rehabilitation programs work? According to Wilson (1975) there have been
over 200 studies of rehabilitation programs, none of which demonstrates that rehabil-
itation decreases recidivism rates. The studies evaluated the effects of several types of
rehabilitation schemes, including education and job-training programs, work-release
programs, and counseling programs. The rehabilitation programs did not affect the
recidivism rates of either juveniles or adults. Given the results of these studies, there
is good reason to be pessimistic about the prospects for using rehabilitation to reduce
crime.
Why doesn't rehabilitation work? First, it is difficult to change the antisocial
attitudes that make criminals relatively receptive to crime. Second, by the time a
criminal reaches the rehabilitation stage (i.e., by the time the criminal is imprisoned),
he is firmly committed to crime, having committed dozens of crimes. Third, it is
difficult to make legal opportunities more profitable than crime: crime pays, and it
is difficult to increase the job skills of an adult criminal.

There is evidence that anticrime programs targeted on youths pass the benefit-
cost test. The high cost of crime means that even a relatively expensive program
that reduces crime rates by a small amount will pass the test. The average juvenile-
delinquent program reduces crime rates by a small amount (Lipsey, 1992). Some
early-intervention programs help reduce crime among youths, again by modest
amounts (Mendel, 1995).
The parole system is based on the principle of rehabilitation. Under the parole
system, a convict stays in prison until the parole board declares that he is rehabil-
itated. The rehabilitated convict is released and serves the remainder of his prison
term under the supervision of a parole officer. A parolee cannot get married or own
a car, and must receive the approval of his parole officer before borrowing money
or changing residence.
Most parole boards have large caseloads and base most parole decisions on
simple rules of thumb. One popular rule is that a convict is released after serving either
three years or one-third of the prison sentence, whichever is shorter. Studies of the
678 Part \'l Education and Crime
parole system have generated mixed results. In some states, parolees have lower rates
of recidivism than convicts who serve out their entire sentence. In other states, there is
no measurable difference between the recidivism rates of parolees and other convicts.
The Deterrence Function. A number of studies show that the threat of imprison-
ment decreases crime rates. An increase in either the probability of imprisonment
or the length of the prison term increases the expected cost of crime, decreasing the
number of crimes.
There is evidence that an increase in the certainty of punishment provides
a greater deterrent than an increase in the severity of punishment (Barrel. 1979;
Grogger. 1991 ). In other words, an increase in the probability of imprisonment de-
ters more crime than an increase in the length of the prison term. For example.
Bartel ( 1979) estimates that the elasticity of the supply of crime with respect to the
probability of punishment is about —0.30: a 10 percent increase in the chance of
imprisonment decreases the crime rate by 3 percent. On the other hand, the elasticity
of supply with respect to the length of imprisonment is close to zero.
Why don't longer prison sentences deter crime? Although an increase in the
penalty may discourage crime by increasing the expected cost of crime, it may
simultaneously encourage crime for three reasons:
1 . Hardening the criminal. Perhaps a long prison sentence hardens the

making antisocial behavior more
convict, likely. In Figure 22-7. a longer
Figure 22-7 Effects of Increased Punishment on Crime
Supply with lower

aversion
B' B' B*
Number of burglaries
I he increase in the severity of punishment decreases the net return from R* to H causing .
a move from ( to /» It a longer prison term hardens the criminal, her aversion to crime
'
will decrease, shifting the supply curve to the right and causing a move from point D to
/ crime uses close to lis original level i B is close to B*). Ii the criminal also learns
more in prison, she will have a higher probability of success when she gets out. causing a
move from / to / increasing the crime rate to Us original level).
i
prison term decreases the net return to burglary from R* to /?', causing
movement down the initial supply curve (from point C to D) and
decreasing the number of crimes from B* to B'. If criminals are hardened
by longer prison terms, their anguish cost decreases, shifting the supply
curve to the right and causing a move from point D to E. The net effect is a
small decrease in the crime rate (from B* to B").
2. Prison schooling. If prison allows a criminal to learn the tricks of the trade
from other criminals, a longer prison term means a more educated and
skillful criminal. The postprison probability of success increases, increasing
from crime.
the net return In Figure 22-7, the prison-schooling effect
causes a move from point E to F, restoring the original crime level (B*).
3. Discount rates. If criminals are "present-oriented," they are more
concerned with instant gratification than the long-term consequences of
crime (imprisonment). The perceived cost of an additional year in prison
(four years instead of three) is small relative to the benefits of today's crime.
In formal terms, if criminals have large discount rates, the perceived cost of
an additional year of imprisonment is relatively small in present- value
terms, so the additional year has a relatively small deterrent effect.
The Incapacitation Function. The third function of the prison system is to take
criminals out of circulation. An obvious way to reduce crime up potential is to lock
criminals. The practical policy issue is whether the benefit of incapacitation exceeds
the cost. The benefit equals the number of crimes avoided by holding a criminal
times the social cost per crime. The consensus is that between 17 and 21 crimes are
avoided by keeping the typical criminal in prison for a year (Marvel and Moody,
1994; Dilulio, 1996). The cost of imprisonment is about $23,000 per prisoner per
year, so if the social cost per crime avoided by imprisonment is at least $1,353, the
incapacitation benefit exceeds the cost of prisons.
Deterrence versus Incapacitation and "Three Strikes" Laws. Which is more

important, deterrence or incapacitation? The answer to this question tells us whether
a recent innovation in criminal sentencing —
Three Strikes and You're Out is likely —
to be efficient. Under the three-strikes law, a person convicted for a third felony is
given a very long prison sentence, e.g., 25 years or perhaps life imprisonment. About
half the states have three-strikes laws, but just a handful have implemented the law
in a significant way. California has used the law to sentence 40,000 people — about a
quarter of its prison population — to lengthy sentences (Carelli, 1999). Georgia has
sentenced about 2,000 people to lengthy terms under its three-strikes law.
To think about the effects of three-strikes laws, consider two extremes. First, if
prisons have no deterrent effect but only an incapacitation effect, the three-strikes
law will be ineffective and inefficient. By the time a person is convicted of a third
felony, he will probably be late in his criminal career, and will be committing fewer
and fewer crimes each year. There will be few crimes avoided by having a relatively
old criminal in prison, so the cost of the three-strikes law ($23,000 per year for many
years) is likely to exceed the benefit (the decrease in the social cost of crime).
At the other extreme, suppose the only effect of prisons is deterrence. In this
case, the threat of longer prison sentences under the three-strikes law will decrease
the expected payoff from crime for all criminals, decreasing crime rates. If the three-
strikes law deters enough crime, there could actually be fewer people in prison. If the
law has a large deterrent effect, its benefit (the decrease in the social cost of crime)
is likely to exceed its cost (extra expenses for the prison system).
A recent study suggests that deterrence is more important than incapacitation

(Levitt. 1998). For property crimes, deterrence is responsible for about three-fourths
of the decreases in crime caused by prisons. For example, if the number of burglary
arrests increases by 10. the deterrent effect will be 23 fewer burglaries, but the inca-
pacitation effect is just 1 less burglary. If the number of auto-theft arrests increases
by 10, a total of 5 auto thefts are avoided through deterrence, but only 1 is avoided
through incapacitation.
Although it is too early to determine the actual effects of three-strikes laws, the
early experiences are consistent with the notion that deterrence is more important
than incapacitation (Levitt, 1998). In Washington state, which has had a three-strikes
law since 1993, violent crime has dropped but the prison population is expected to
grow by only 10 percent over the next 10 years. During the first year of California's
three-strikes law. violent crime dropped by 7.2 percent and the projections for the
prison population were revised downward by 25 percent.
Juvenile versus Adult Penalties. There is considerable variation across states in

the relative severity of punishments for adults and juveniles (less than 1 8 years of age
inmost states). In general, adults face higher penalties for crime, so the expected cost
of crime for an 1 8-year-old is higher than the cost for a 1 7-year-old. In states where the
difference in penalties is relatively large —
generating a harsh transition from juvenile
to adult status —
property crime among 18-year-olds is about 21 percent lower than
crime among 17-year-olds (Levitt, 1998). In states where there are relatively small
differences in the treatment of juveniles and adults, the difference in crime rates is
only 9 percent. The differences are much larger for crimes of violence: in states with
a harsh transition, the crime rate for 1 8-year-olds is 4 percent lower than the rate for
17-year-olds, while in states with a gentler transition, the crime rate for 1 8-year-olds
is 23 percent higher (a result of a generally positive relationship between age and
violent crime rates among young adults).
Why Are Crime Rates Higher in Big Cities?

Crime rates increase with city size. The rate of violent crime in large cities (population
at least 250.000) is about double the crime rate in small cities (population less than
10.000). The rate of property crime in big cities is about 30 percent higher than the
crime rate in small cities. Overall, the elasticity of general crime with respect to city
size is 0.15, meaning that a 10 percent increase in population increases the crime
rate by about 1 .5 percent (Glaeser and Sacerdote. 1996).
Figure 22-8 Arrest Rate and City Size

14 i-
12
12
-
10
10
~ 6 -
4 -
25.000 to 50,000 to 100,000 to 250,000 to 500,000 to 1.000.000

49,999 99,999 249,999 499,999 999,999 and more
Population
SOURCE: Glaeser, Edward L„ and B. Sacerdote. "Why is There More Crime in Cities?" NBER
Working Paper #5430, 1996.
Why are crime rates so much higher in big cities? Glaeser and Sacerdote provide
three reasons for higher crime rates in big cities:
1 A higher payoff for crime is responsible for 25 percent of the difference in

crime rates. Big cities have more lucrative crime targets: the average value
per crime is about $700 in a city of 500,000 and $900 in a city of 1 million,
compared to a value between $500 and $600 for smaller cities.
2. A lower probability of arrest is responsible for 15 percent of the difference

in crime rates. Figure 22-8 shows the arrest rate (the percentage of crimes
that result in an arrest) for cities of different sizes. Although the reasons
behind the negative relationship are unclear, Glaeser and Sacerdote suggest
some possibilities: (a) it may be more difficult for police to identify a
culprit when the pool of potential suspects is larger; (b) lawful citizens in
impersonal big cities may be less inclined to help their neighbors and the
police in crime-control efforts.
3. A higher concentration of female-headed households is responsible for 50

percent of difference in crime rates. It's not clear why higher rates of single
parenthood have such a powerful influence on crime rates. Glaeser and
Sacerdote speculate that children raised in single-parent families may have
fewer job skills and less powerful ethical restraints on criminal behavior.
Since 1970, the correlation between city size and crime has weakened, and in recent
years, there were dramatic reductions in crime rates in big cities, especially New
York City.
Illegal Commodities and Victimless Crime

The fact that a product is illegal does not prevent people from consuming it. Among
the illegal goods and services that are consumed in large quantities are illegal drugs
(heroin, cocaine, marijuana), illegal gambling, loansharking, and prostitution. Why
are some goods outlawed? First, lawmakers assume that consumers are irrational
about some products: a gambler is unable to make rational choices in the heat of a
gambling frenzy, and a drug user is incapable of rational thought in the middle of
a drug trip. Second, the consumption of certain goods is assumed to be contagious.
Society protects itself from a drug "epidemic" for the same reasons that it protects
itself from a polio epidemic. Third, drug addiction is often passed from parent to
child, through either physiological or psychological means. By outlawing certain

drugs, the government protects children from drug dependence. Fourth, lawmakers
assume that drug use leads to unemployment, poverty, and welfare dependence.
Product Illegality and Production Costs
A firm selling an illegal product differs from a lawful firm in two ways. First, there
are large penalties if the illegal activity is discovered by the police. The participants
may be fined or imprisoned, and the assets of the firm may be seized. Second,
contracts with illegal enterprises are not legally enforceable, so creditors cannot be
certain that they will be paid for services rendered. These two characteristics make
the production cost of the illegal firm relatively high.
The punishment increases wages. To compensate for the risk of illegal
threat of
must pay its workers a wage premium. Just as a coal miner is paid
activity, the firm
more than a person in a safe, clean job, a worker in an illegal enterprise is paid
more than a person in a legal job. In addition, the firm pays a relatively high wage to
prevent "squealing." If workers earn relatively high wages, they have more to lose
if the firm's illegal activities are discovered, so workers have greater incentive to
conceal those activities.
The threat of punishment also affects the way that a firm organizes its production
process. To maintain secrecy, the firm controls the flow of information about its
activities. One way to control information segment the workforce into small
is to
work groups and limit the information that passes from one group to another. For
example, heroin passes through many groups on the way from the poppy field to
the consumer (importer, processor, wholesaler, distributor), and each group has little
knowledge of the activities of the other groups in the chain. These evasion techniques
add to the production cost.
The illegal activity also has a relatively high capital cost. A loan to an illegal
enterprise is relatively risky for three reasons. First, contracts are not legally enforce-
able, so lenders can never be certain that they will be repaid. Second, given its desire
for secrecy, the illegal firm cannot provide the lender with financial information.
Since the lender cannot assess the profitability of the illegal enterprise, the loan is
relatively risky. Finally, given the risk of confiscation, the illegal firm does not have
any real collateral, so the lender may be unable to collect if the borrower defaults on
the loan. Because a loan to an illegal enterprise is relatively risky, the lender charges
a relatively high interest rate, so the production cost is relatively high.
Product Illegality and Price

Figure 22-9 shows the effect of product illegality on the market for drugs. Sup-
pose that a particular drug is suddenly declared illegal. The cost of producing and
distributing the drug increases, shifting the supply curve to the left: at every price,
less is supplied. The illegal status of the drug also affects its demand curve. The
total cost of consuming the illegal drug equals the sum of the market price and the
expected penalty for consuming the drug. For example, if there is a 2 percent chance
of a $1,000 penalty for drug consumption, the expected penalty is $20 per dose
(0.02 times $1,000). When the product is declared illegal, the demand curve shifts
downward by the expected penalty. The penalty for drug consumption is like a tax:
it shifts the demand curve by the amount of the implicit tax.
If enforcement efforts are concentrated on the supply side of the drug market, the
prohibition of drugs increases the market price. In Figure 22-9, the shift of the supply
Figure 22—9 Effects of Increased Punishment on Crime
Illegal
supply
Legal
supply
K"
Kilos of drugs
When a commodity is declared illegal, the cost of producing and selling it increases, shifting the supply
curve upward. Consumer penalties increase the net cost of consuming the good, shifting the demand curve
downward. The equilibrium moves from point B to C. The supplier penalty is large relative to the
consumer penalty, so the supply shift is larger than the demand shift, and the price increases from P' to
P". The equilibrium quantity decreases from K' to K"
curve is larger than the shift of the demand curve, reflecting the assumption that the
expected penalty for the production and sale of drugs is large relative to the expected
penalty for consumption. The price rises from P' to P", and the equilibrium quantity
decreases from K' to K" By
. declaring the drug illegal, the government increases
the price and decreases the quantity consumed.
What
are the implications for the consumers of illegal goods? Consumers pay
higher prices for the illegal goods and services. The typical heroin addict, for ex-
ample, pays about $60 per day for his habit, about 30 times the amount he would
pay if heroin were legal. The same analysis applies to other types of illegal goods.
The patrons of loansharks pay higher interest rates because the loanshark must be
compensated for the risk of imprisonment and the cost of concealing his activities.
Similarly, gamblers face lower odds (lower expected returns) because bookies face
higher risks and incur a large concealment cost.
Drugs and Property Crime

There is a positive correlation between drug use and criminal activity. According to
Wilson (1975). between 25 percent and 67 percent of property crime is committed by
drug addicts. Despite this high correlation, the causal link between drug consumption
and crime is not fully understood. Although there is evidence that some addicts
support their drug habits with property crime, it is unclear just how much addict
crime is committed solely to support drug habits. Boyum and Kleiman (1995) report
that 39 percent of cocaine and crack users claim to have committed crime to get
money to buy drugs.
Will the enforcement of drug laws increase total property crime? In Fig-
ure 22-9, the illegality of drugs increases price and decreases the quantity con-
sumed. The increase in price is relatively large because ( 1 ) the leftward shift of
the demand curve is relatively small and (2) the demand curve is relatively steep
(the price elasticity of demand for drugs is relatively low). Because the price rises
by a relatively large amount, total expenditures on drugs (price times quantity) in-
crease. If drug addicts support their habits with property crime, the increase in total
expenditures increases property crime.
Silverman and Sprull (1977) estimate the effects of changes in drug prices on
property crime. They estimate that the price elasticity of demand for heroin is about
—0.27. so a 50 percent increase in the price of heroin would decrease the quantity of
heroin demanded by about 13 percent and increase total drug expenditures by about
30 percent. Because only a portion of property crime is drug-related, the resulting
increase in property crime ( 14 percent) is less than the increase in drug expenditures.
There are two implications from this analysis. First, there is often a trade-off
between drug control and property crime. Because a supply-side policy increases
total expenditures on drugs, property crime is likely to increase. Although drug-
enforcement policies decrease drug consumption and thus decrease the social cost
associated with drug abuse, they also increase property crime. The difficult policy
question is whether the benefit (the decrease in the social cost of drug abuse) exceeds
the cost i increased property crime). The second implication is that demand-side
policies do not have the same trade-offs. Since a demand-side policy simply shifts
the demand curve downward, decreasing both the priceand quantity of drugs, it
decreases total expenditure on drugs and thus decreases drug-related property crime.
Summary
1. The facts on crime are collected by the FBI (from local police departments)
and the Justice Department (from victimization surveys).
2. Victimization rates vary with income, place of residence, and race.
Victimization rates are generally higher for racial minorities and the residents
of central cities.
3. The direct costs of crime include injury cost and property losses. The indirect
costs include protection cost and the cost of the criminal justice system.
According to one study, the total cost of crime in 1992 was $250 billion.
4. The optimum amount of crime where the marginal victim cost equals the
is
marginal protection cost. Crimes differ in victim and prevention costs, so

optimum crime rates vary.
5. A rational person commits a property crime if the net return on crime is large
enough to offset her underlying aversion to crime.
a. The expected benefit depends on the monetary value of the loot and the
probability of success.
b. The expected cost depends on the probability of imprisonment, the
opportunity cost of prison time, and the length of the prison sentence.
c. The opportunity cost of crime is the income forgone in a legal job.
d. The net return is the expected benefit (the expected loot less the expected
cost) minus the money that could be earned in a legal job.
e. Crime is rational if the net return exceeds the anguish cost.
6. The supply curve for crime shows the relationship between the net return and
the number of crimes. Society determines the net return to crime, thus picking
a point on the supply curve.
a. Investments in education and job training increase the value of legal
opportunities.
b. Potential victims can decrease the expected benefit of crime by hardening
the target or decreasing the resale value of the loot.
c. The police can increase the expected cost of crime by arresting more
criminals.
d. The courts can increase the expected cost of crime by increasing the
likelihood and severity of punishment.
7. The supply curve shifts as a result of demographic and social changes.

a.As the share of the population in the crime-prone years increases, the
supply curve shifts to the right.
b. Social changes that increase the underlying aversion to crime shift the
supply curve upward.
8. Police are responsible for arresting criminals and investigating crimes.

a. The arrest ratio is the fraction of crimes that lead to an arrest. The elasticity
of crime with respect to the probability of arrest is about —0.30.
b. The success of police investigation is measured by the clearance ratio, the
fraction of arrests that lead to conviction. The most important investigative
task is to find eyewitnesses to the crime.
c. Although an increase in car patrols decreases the police response time, it
does not affect arrest rates and crime rates because victims are slow to
report crimes. Foot patrol is more effective because the police gather
information about people and potential criminals on their beats.
9. The courts are responsible for determining guilt and setting penalties.
a.About 9 in 100 felony arrests result in prison sentences. Many cases are
dropped because the offense is trivial or the evidence is insufficient. About
one-third of arrests lead to convictions, but most convicts receive probation
or short jail terms.
b. Most criminal cases are resolved through plea bargaining: the district
attorney reduces the criminal charge and the accused pleads guilty to the
lesser charge. From the perspective of society, plea bargaining is efficient if
the benefit (the savings in court costs) exceeds the cost (the decrease in the
punishment of convicts).
c. Two types of errors are inevitable in the criminal justice system: a false
positive occurs when an innocent person is found guilty, and a false negative
occurs when found innocent. In the United States, the
a guilty person is
state must prove the case beyond a reasonable doubt, which decreases the
number of false positives and increases the number of false negatives.
10. According to the principle of marginal deterrence, penalties should be scaled

to the severity of the crime, thus providing a marginal deterrent for more
serious crimes. An increase in the penalty for a particular crime causes some
criminals to switch to other crimes. For example, an increase in the penalty for
burglary causes some criminals to upgrade to robbery, others to downgrade to
larceny, and others to abandon crime.
1 1 The prison system has four purposes: rehabilitation, deterrence, incapacitation,
and retribution.
a. Contrary to popular opinion, rehabilitation is relatively ineffective in
reducing crime.
b. Deterrence works. An increase in the certainty of punishment provides a
greater deterrent than an increase in the severity of punishment.
c. Incapacitation is an efficient policy if the prisoner would commit a large
number of crimes if he were free.
12. Illegal goods are relatively expensive because firms must compensate their
workers and lenders for extra risks and must spend money on concealment.
13. Policies that decrease the supply of illegal drugs increase their prices. If drug
addicts support their habits with property crime, an increase in total
expenditures increases total property crime, partly offsetting the benefits
associated with decreased drus abuse.

1 . Consider Waivah, a potential burglar who weighs the benefits and cost of
burglary. Fill in the blanks in the following table:
Loot $1,000
Probability of success 0.70
Expected loot
Probability of being imprisoned 0.03
Wage per day $20
Workdays per year 100
Forgone income per year
Loss of freedom per year $3,000
Opportunity cost per year
Length of prison term 2 years
Expected penalty cost
Expected benefit from burglary
Daily income if employed
Net return from burglary
2. Consider Boris, whose benefit-cost analysis of burglary is shown in

Table 22-3. Compute the values of the following variables such that Boris is
indifferentbetween burglary and a legal job. In computing the threshold value

for each variable, assume that the values of the other variables are the same as
in Table 22-3.
a. The probability of success (Ps ).
b. The value of the loot.
c. The probability of imprisonment (P,).
d. The opportunity cost per year (C).
e. The length of the prison sentence.
3. Consider a city with a wealthy area and a poor one. The two areas have the
same horizontal MVC (marginal-victim-cost) curve and the same negatively
sloped MPC (marginal-prevention-cost) curve. The two curves intersect at a
crime rate B*. The crime rate in the wealthy area is B w and the crime rate in
is B
the poor area p Given . the current allocation of police between the two
areas, B w < B* < B p .
a. Use a graph to depict the current situation.

b. of resources efficient? If not, how should resources
Is the current allocation
be reallocated? Put some numbers on your graph and use the graph to show
that the reallocation of resources will generate net gains for society.
c. Suppose that the two areas have the same MVC curve but different MPC
curves. Depict a situation in which a higher crime rate in the poor area
(B p > B w )is efficient.
4. The city of Evilville suffers from two types of crimes, marijuana consumption
and burglary. Now let us suppose that you must decide how to allocate the
city's crime-fighting resources between the two types of crime. The
marginal-prevention-cost (MPC) curves are linear, and for a given crime rate,
the marijuana MPC is twice the burglar)' MPC. The marginal-victim-cost

(MVC) curves are horizontal. Under what conditions will the optimum
number of marijuana crimes be less than the optimum number of burglaries?
Illustrate your answer with a graph.
5. Suppose that your city has a fixed budget to allocate among 10 police
precincts. You are responsible for allocating the budget.
a. Your allocation decisions should be guided by some crime-fighting
objective.What is your objective'1
b. What information would you collect, and how would you use it?
6. Suppose that Alice purchases an insurance policy for her bike. Complete the
following table, assuming that her policy has a $20 deductible.
Protective Protection Probability Expected

Action Cost of Theft Theft Cost Total Cost
No protection SO 0.30
Flimsy lock 11 0.10
Sturdv lock 24 0.02
Armed euard 80 0.00
7. Recall the Bonnie example, shown in Table 22-5. You are negotiating with the
DA over the length of the prison term for petit larceny. Suppose that Bonnie's
opportunity cost of prison time is $8,000 per year. What is the maximum
larceny term she will accept? In other words, what larceny term will make her
indifferent between facing a criminal trial on armed robbery and a plea bargain
for larceny?
8. Consider the following quote from Monty, a former district attorney who was
recently elected the governor of a state in which the only crimes are burglary
and robbery. "1*11 stop the let's-make-a-deal approach to criminal justice and
replace the plea-bargaining process with a set of required penalties, one for
burglary and one for robbery."
a. Outline a strategy for determining the set of penalties. What information
would you collect, and how would you use it?
b. Is this a wise policy? Is it efficient
9. Critically appraise this statement: "The social cost of robbery is about six
times the social cost of burglary. Given the principle of marginal deterrence,
the length of the prison term for robber}' should be six times the length of the
prison term for burglary."
10. Consider the analysis of marginal deterrence shown in Figure 22-6. Under
what circumstances (yvhat values for the social cost of burglary and robber}')
would the increased penalty for burglar}' decrease the total social cost of crime?
1 1 Consider a cityin which there are only two types of crimes, burglary and
armed robbery. Choose one of the words in each set of parentheses to make the
following statement correct and then explain your choices: an increase in the
penalty for burglary will decrease the total social cost of crime if
a. The supply curve for burglary is relatively (steep, flat).

b. The cross-elasticity of the supply of armed robbery with respect to the net
return on burglary is relatively (small, large).
c. The social cost of burglary is (small, large) relative to the social cost of
armed robbery.
12. In the city of Swiftsure, there is only one crime, armed robbery. The city
recently decided to double the probability of imprisonment from 10 percent
(10 percent of armed robberies result in a prison sentence) to 20 percent.
Because the city has a fixed prison capacity, it also cut the average prison
sentence in half, from 3 years to 1.5 years.
a. Would you expect these changes in the city's crime policy to increase,
decrease, or not affect the city's crime rate?
b. Would you expect these changes to increase, decrease, or not affect the total
cost of crime (the sum of victim and prevention costs)?
13. Suppose that Gotham City doubles the required prison sentence for burglary
(its only crime). Consider the following quote from Robin: "If my assumptions
are correct, the longer prison sentence will increase the crime rate."
a. What are Robin's assumptions?
b. Use a graph to illustrate his prediction.
14. Consider a city with the following characteristics:
i. The supply curve of heroin is vertical (perfectly inelastic) at 200 kilos.

ii. The initial price is $40. The price elasticity of demand is —0.25.
Hi. The city will spend $500 on a drug policy to shift the supply curve to
the left by 5 percent. The drug policy will not shift the demand curve.
iv. The heroin market generates two types of social cost. First, consumption
of heroin generates a social cost of $150 per kilo: the citizens of the city
are willing to pay $150 for every one-kilo reduction of heroin. Second,
heroin consumers support their habits with property crime.
a. Is the drug program efficient?
b. Choose the word in parentheses that makes the statement correct, and
explain your choice: "As the price elasticity of demand for heroin increases
in absolute value, the efficiency of the drug policy (increases, decreases)."
15. Consider the following statement: "Our drug policy is inefficient. We should
legalize drugsand tax drug producers. We can set the tax at a level such that
we have the same equilibrium drug consumption. By doing so, we will save
all the money we currently spend on the enforcement of drug laws and the
prosecution and punishment of drug dealers." What are the advantages and
disadvantages of this policy, which combines legalization and taxation?
16. A number of insurance companies recently started offering free burglar alarms
with their homeowner insurance policies. Evaluate the merits of such a policy.
Would you expect such firms to make more or less money than firms that do
not provide alarms?
17. Sympacity is considering a policy under which every victim of crime would be
compensated for victim cost. The city would cover the property losses from
crime and would also compensate victims for injuries suffered as a result of
crime. Evaluate the efficiency and equity effects of the city's policy.
18. The property losses from business fraud (white-collar crime) totaled S600 per
household in 1983. compared to losses of $90 for robbery. S93 for burglary,
and $56 for auto theft. Yet white-collar crime receives less attention than other
types of property crime. Why do you suppose white-collar crime receives such
little attention?
19. During the 1970s, the crime rate of women increased five times faster than the
crime rate of men. Why?

General References
Carelli. Richard. "Supreme Court Rejects Case Involving Three-Strikes Law." The
Oregonian, January 20. 1999. p. A7.
Dilulio. John J. Jr. "Help Wanted: Economists. Crime, and Public Policy" Journal of
Economic Perspectives 10. no. 1 ( 1996). pp. 3-24.
Freeman. Richard B.. and Harry J. Holzer. "The Black Youth Employment Crisis: Summary
of Findings," In The Black Youth Employment Crisis, ed. Richard B. Freeman and
Harry J. Holzer. Chicago: University of Chicago Press. 1986.
Glaeser. Edward L. "Are Cities Dying?" Journal of Economic Perspectives 12. no. 2 ( 1998).
pp. 139-60.
Glaeser. Edward L.. and B. Sacerdote. "Why Is There More Crime in Cities?" NBER
Working Paper no. 5430. 1996.
Grogger, J. "Criminal Opportunities. Youth Crime, and Young Men's Labor Supply."
Universitj of California. Santa Barbara ( 1994i.
Levitt. Steven D. "The Effect of Prison Population Size on Crime Rates: Evidence from
Prison Overcrowding Litigation." Quarterly Journal of Economics 1 1 1 ( 1996).
pp. 319-52.
Levitt. Steven D. "Why Do
Increased Arrest Rates Appear to Reduce Crime: Deterrence.
Incapacitation, orMeasurement Error?" Economic Inquiry 36 1998a). pp. 353-72. (
Levitt. Steven D. "Juvenile Crime and Punishment." Journal of Political Economy 106.
no. 1 (1998b). pp. 1156-185.
Marvel, Thomas B., and Carlisle E. Moody. "Prison Population Growth and Crime
Reduction." Journal of Quantitative Criminology 10. no. 4 < 1994). pp. 109—40.
Crime Facts
Chaiken. Jan M.. and Marcia Chaiken. "Crime Rates and the Active Criminal." In Crime
and Public Policy, ed. James Q. Wilson. San Francisco: Institute for Contemporary
Studies. 1983. Discusses recent trends in crime rates.
U.S. Department of Justice. Criminal Victimization in the United States. 1993. Washington.
D.C.: U.S. Government Printine Office, 1995.
. Sourcebook of Criminal Justice Statistics, 1986. Washington, D.C.: U.S. Government

U.S. Federal Bureau of Investigation. Crime in the United States, 1990. Washington, D.C.:
U.S. Government Printing Office, 1991 . Data from the Uniform Crime Reports.
The Cost of Crime

Gray, Charles M., and Mitchell Joelson. "Neighborhood Crime and the Demand for Central
City Housing." Chapter 3 in The Costs of Crime, ed. Charles M. Gray. Beverly Hills,
Calif.:Sage Publications, 1979. Explores the effects of crime on property values.

Reynolds, Morgan. Crime by Choice: An Economic Analysis. Dallas: Fisher Institute, 1986.
Estimates the cost of personal and property crimes.
Thaler, Richard. "Note on the Value of Crime Control —Evidence from the Property
Market." Journal of Urban Economics 5 (1978), pp. 1 37^45. Examines the effects of
crime rates on property values.
The Rational Criminal

Two Hypotheses." In The Economics of Crime and
Cobb, William. "Theft and the
Punishment, ed. Simon Rottenberg. Washington, D.C.: American Enterprise Institute.
1973.
Deutsch, Joseph; Simon Hakim; and J. Weinblatt. "A Micro Model of the Criminal's
Location Choice." Journal of Urban Economics 22 (1987), pp. 198-208.
Gunning, J. Patrick, Jr. "How Profitable Is Burglary?" In The Economics of Crime and
Punishment, ed. Simon Rottenberg. Washington, D.C.: American Enterprise Institute,
1973.
Herrnstein, Richard J. "Some Criminogenic Traits of Offenders." In Crime and Public
Policy, ed. James Q. Wilson. San Francisco: Institute for Contemporary Studies, 1983.
traits that predispose some people to crime, that is, the traits
Discusses the personal
that causesome people to have a relatively low aversion to crime.
Krohm, Gregory. "The Pecuniary Incentives of Property Crime." In The Economics of
Crime and Punishment, by Simon Rottenberg. Washington, D.C.: American Enterprise
Institute, 1973.
Sesnowitz, Michael. "Returns to Burglary." Western Economic Journal 10, no. 4 (1972),
pp. 477-81.
Crime, Demographics, and Culture
Blumstein, Alfred; Jacqueline Cohen; and Harold Miller. "Demographically Disaggregated

Projections of Prison Population." Journal of Criminal Justice 8 (1980), pp. 1-26.
Predicts that crime rates decline as the relative number of teenagers decreases.
Fox, James A. Forecasting Crime. Lexington, Mass: D. C. Heath, 1978. Predicts crime rates
on the basis of the age distribution and racial composition of the population.
Schapiro, Morton Owen, and Dennis A. Ahlburg. "Why Crime Is Down." American
Demographics 8 (1986).
Wilson, James Q. "Crime and American Culture." The Public Interest 70 (Winter 1983),
pp. 22^18.
Unemployment and Crime

Cook, Philip J., and Gary A. Zarkin. "Crime and the Business Cycle." Journal of Legal
Studies, 14 (1985). pp. 5-28. Found that robbery and burglary rates rise by a small
1 1
amount during recessions.

Fleisher. Belton. The Economics of Delinquency. Chicago: Quadrangle Books, 1966.
Suggests that an increase in the unemployment rate increases the juvenile crime rate.
Freeman, Richard B. "Crime and Unemployment." In Crime and Public Policy, ed. James
Q. Wilson. San Francisco: Institute for Contemporary Studies, 1983. Reviews dozens
of studies of the relationship between crime and unemployment. Concludes that there
is a negative, but weak, relationship between the crime rate and the unemployment rate
and that the link between the crime rate and sanctions (expected punishment) is
stronger. Speculates that an improvement in labor-market conditions may deter
first-time offenders.
.
"The Labor Market." In Crime, eds. J.Q. Wilson and Joan Petersilia. San Francisco:
Contemporary Policies. 1995.
Institute for
Orsagh, Thomas, and Ann D. Witte. "Economic Status and Crime: Implications for
Offender Rehabilitation." Journal of Criminal Law and Criminology 72 1981 ). no. 3. (
Found little evidence of a significant relationship between unemployment and

crime.
Phillips, Llad; Harold Votey; and Harold Maxwell. "Crime, Youth, and the Labor Market."
Journal of Political Economy 80 (1972), pp. 491-504. Suggests that the crime rates of
18- and 19-year-old males are sensitive to labor-market conditions: the lower the
labor-force participation rate, the higher the crime rate.
Wilson, James Q. "Crime and Unemployment." The Public Interest 79 (Spring 1985),
pp. 3-8. Discusses the relationship between unemployment rates and crime rates.
Victim as Crime Fighter
Balkin, S., and J. F. McDonald. "The Market for Street Crime — An Economic Analysis of
Victim-Offender Interactions." Journal of Urban Economics 1 ( 1 98 1 ). pp. 390-405.
A discussion of the subtle interactions between potential victims and criminals.
Clotfelter, Charles D. "Explaining Unselfish Behavior —Crime and the Helpful Bystander."
Journal of Urban Economics 8 ( 1980). pp. 196-212. Discusses altruistic behavior of
witnesses to crime.
. "Urban Crime and Household Protective Measures." Review of Economics and
Statistics59 1977), pp. 499-503. Discusses the effects of increased crime rates on the
(
protective measures of potential victims.

Clotfelter. Charles D.. and Robert D. Seeley. "The Private Costs of Crime." In The Costs
of Crime, ed. Charles M. Gray. Beverly Hills, Calif.: Sage Publications. 1979,
pp. 213-32. Discusses the private and social costs of private protection measures.
Wilson. James Q.. and Barbara Boland. "Crime." Chapter 4 in The Urban Predicament.
ed. William Gorham and Nathan Glazer. Washington, D.C.: Urban Institute, 1976,
pp. 179-230. Discusses how potential victims protect themselves from crime.
Police Issues
Carr-Hill, R., and N. Stern. "An Econometric Model of the Supply and Control of Recorded
Offenses in England and Wales." Journal of Public Economics 2 1973). pp. 289-318.
(
Concludes that an increase in police and enforcement expenditures per capita increased
reported (not actual) crime per capita.
. "Theory and Estimation in Models of Crime and Its Social Control and Their
Relations to the Concepts of Social Outputs." In The Economics of Public Services, ed.
Martin Feldstein and Robert Inman. London: Macmillan. 1977. Discusses the issue of
how to measure the output of crime enforcement.
Hakim, S.; A. Ovadia; E. Sagi; and J. Weinblatt. "Interjurisdictional Spillover of Crime and
Police Expenditure." Land Economics 55 (1979), pp. 200-212.
Johnson, T. G.; B. M. Swallow; and B. J. Deaton. "Does Crime Prevention Pay? —An
Analysis of Local Government Crime Prevention Expenditures." American Journal of
Agricultural Economics 66 (1984).
Kelling, G. L.; T. Pate; D. Dieckman; and C. E. Brown. The Kansas City Preventative
Patrol Experiment: A Technical Report. Washington, D.C.: The Police Foundation,
1974. Suggests that increases in patrol activity do not affect crime rates.
Levine, James. "The Ineffectiveness of Adding Police to Prevent Crime." Public Policy 23
(Fall 1975), pp. 523^5.
Mehay, Stephen. "Inter-Governmental Contracting for Municipal Police Services An —
Empirical Analysis." Land Economics 55 (1979), pp. 59-72.
Phillips, Llad, and Harold Votey. "Crime Control in California." Journal of Legal Studies
4 (1975), pp. 327-50. Estimates the effects of enforcement expenditures on crime. An
increase in the number of law enforcement personnel increases the probability of
conviction.
Pogue, Thomas F. "Effect of Police Expenditures on Crime Rates." Public Finance
Quarterly 3 ( 1975), pp. 14—44. Finds no relationship between police spending and
crime rates.
Reynolds, Morgan. Crime by Choice: An Economic Analysis. Dallas: Fisher Institute, 1986.
Sherman, Lawrence. "Patrol Strategies for Police." In Crime and Public Policy, ed. James
Q. Wilson. San Francisco: Institute for Contemporary Studies, 1983. Discusses the
efficacy of alternative police patrol strategies.
Thaler, Richard. "Econometric Analysis of Property —
Crime Interaction between Police
and Criminals." Journal of Public Economics 8 (1977), pp. 37-5 1
Wilson, James Q. Thinking about Crime. New York: Basic Books, 1975. Chapter 5 (The
Police and Crime) discusses the efficacy of patrol in deterring crime and contrasts the
traditional patrol approach to the community service approach.
Evidence of Deterrence
Bartel, Ann P. "Women and Crime: An Economic Analysis." Economic Inquiry 42 (1979).
pp. 29-51. The probabilities of arrest and conviction have significant deterrent effects
on female property crime. The decrease in the number of preschool children accounts
for over half of the increase in female property crime between 960 and 1970. 1
Craig, Steven G. "The Deterrent Impact of Police: An Examination of a Locally Provided

Public Service." Journal of Urban Economics 21 (1987) pp. 298-31 1. Discusses the
effects of changes in clearance rates on the crime rate and the reporting rate.
Ehrlich, Isaac. "The Deterrent Effect of Criminal Law Enforcement." Journal of Legal
Studies 1 (1972), pp. 259-76.
. "Participation in Illegitimate Activities." Journal of Political Economy 81 (1973),
pp. 521-65.
Goldberg, I., and F. C. Nold. '"Does Reporting Deter Criminals? —An Empirical Analysis of
Risk and Return Crime." Review of Economics and Statistics 62 (1980), pp. 424—31.
in
Discusses the effects of increased reporting rates on the return on crime and the
probability of being victimized.
Grogger, Jeffrey. "Certainty vs. Severity of Punishment." Economic Inquiry 29, no. 2
(1991), pp. 297-309.
Mathur, V. K. "Economics of Crime — Urban
Investigation of the Deterrent Hypothesis for
Areas." Review of Economics and
60 (1978), pp. 459-66. Discusses the
Statistics
deterrent effects of increases in the certainty and severity of punishment.
Myers, S. L. "Crime in Urban Areas, New Evidence and Results." Journal of Urban
Economics 1 1982), pp. 148-58. Examines the effects of increased penalties and
1 (
increased income on crime rates.

Phillips, Llad. "Crime Control: The Case for Deterrence." In The Economics of Crime and
1973.
Sjoquist,David L. "Property Crime and Economic Behavior: Some Empirical Results."
American Economic Review 63 1973), pp. 439—46. Found that an increase in the
(
probability of arrest decreased the crime rates for burglary, robbery, and larceny.
Tauchen, Helen; Ann Dryden Witte; and Harriet Griesinger. "Criminal Deterrence:
Revisiting the Issue with a Birth Cohort." Review of Economics and Statistics 76
(1994), pp. 399-412.
Tullock, Gordon. "Does Punishment Deter Crime?" The Public Interest, Summer 1974,
pp. 103-111.
Viscusi, W. Kip. "The Risks and Rewards of Criminal Activity: A Comprehensive Test of
Criminal Deterrence." Journal of Labor Economics 4 ( 1 986), pp. 3 1 7—10.
Wilson. James Q.. and Barbara Boland. "Crime." Chapter 4 in The Urban Predicament, ed.
William Gorham and Nathan Glazer. Washington, D.C.: Urban Institute. 1976,
pp. 179-230. Overview of crime and crime prevention, including a summary of

deterrent studies. Estimates that the elasticity of supply of burglaries with respect to the
probability of arrest is —0.50.
Witte, Ann D. "Estimating the Economic Model of Crime with Individual Data." Quarterly
Journal of Economics 94 (1980), pp. 57-84. Estimates deterrent effects and finds that
the increased certainty of punishment (larger probability of punishment) has a larger
deterrent effect than increased severity (larger penalty).
Wolpin, V. "A Time Series Cross-Sectional Analysis of International Variation in Crime and
Punishment." Review of Economics and Statistics 62 1980). pp. 1417-23. Compares (
the crime rates of the United States. Japan, and the United Kingdom.
Economics of the Courts

Forst, Brian. "Prosecution and Sentencing." In Crime and Public Policy, ed. James
Q. Wilson. San Francisco: Institute for Contemporary Studies, 1983. Discusses how
the criminal justice system processes a typical set of 100 criminal cases.
Landes, William. "The Bail System: An Economic Approach." Journal of Legal Studies 2,
no. 1 (1973). pp. 79-105.

Rhodes, W. M. "Economics of Criminal Courts —Theoretical and Empirical Investigation."
Journal of Legal Studies 5 ( 1976). pp. 3 1 1-430.
Schlesinger. Steven R. "Criminal Procedures in the Courtroom." In Crime and Public

Policy, ed.by James Q. Wilson. San Francisco: Institute for Contemporary Studies.
1983. Discusses the bail system and the exclusionary rule.
Stepan. Jan. "Possible Lessons from Continental Criminal Procedure." In The Economics of
Crime and Punishment, ed. Simon Rottenberg. Washington. D.C.: American
Enterprise Institute. 1973.
Optimum Penalties
Becker. Gary S. "The Case for Fines." Journal of Political Economy 76. no. 2 ( 1968).
pp. 168-217.
Dickens, William T. "Crime and Punishment Again: The Economic Approach with a
Psychological Twist." Journal of Public Economics 30 (1985), pp. 97-107. Discusses
the effects of psychological factors in the deterrent effect of increased penalties.
Stigler. George. "The Optimum Enforcement of Laws," Journal of Political Economy 78
(May 1970), pp. 526-36. Discusses the principle of marginal deterrence.
Wilson. James Q. Thinking about Crime. New York: Basic Books. 1975. Chapter 8
advocates several reforms of the sentencing procedure, including ( 1 ) centralized
sentencing management and uniform sentences, (2) some form of confinement for
every crime (certain punishment), (3) a shortening of prison and jail sentences. In other
words, he advocates short but certain punishment.
Rehabilitation
Evans, Robert. "The Labor Market and Parole Success." Journal of Human Resources 3.
no. 2(1968), pp. 201-11.

Crime and Public Policy, ed.
Glaser. Daniel. "Supervising Offenders outside of Prison." In
James Q. Wilson. San Francisco: Contemporary Studies, 1983. Discusses
Institute for
parole, halfway houses, counseling programs, and employment assistance.
Lipsey, Mark. "Juvenile Delinquency Treatment: A Meta-Analysis Inquiry into the
Variability of Effects." In Meta-Analysis for Explanation:A Casebook, ed. Thomas
D. Cook et al. Sage Foundation (1992).
Beverly Hills. Calif.:
—
Martinson. Robert. "What Works Questions and Answers about Prison Reform." The
Public Interest 35 (Spring 1974), pp. 22-54. Reviewed the results of 231 studies of the
effects of rehabilitation programs on recidivism rates. Concluded that rehabilitation has
no appreciable effect on recidivism.
Mendel, Richard A. "Prevention or Pork?A Hard-Headed Look at Youth-Oriented
Anti-Crime Programs." Washington, D.C.: American Youth Policy Forum, 1995.
Incapacitation
Greenberg, David F. "The Incapacitative Effects of Imprisonment: Some Estimates." Liiw

and Society Review 9 (Summer 1975), pp. 541-80.
Greenwood, Peter. "Controlling the Crime Rate through Imprisonment." In Crime and
Public Policy, ed. James Q. Wilson. San Francisco: Institute for Contemporary Studies,
1983. Discusses the trade-offs associated with selective incapacitation.
Shinnar. Reuel, and Shlomo Shinnar. "The Effects of the Criminal Justice System on the
Control of Crime." Law and Society Review 9 (Summer 1975), pp. 581-61 1.
Wilson. James Q. "Dealing with the High-Rate Offender." The Public Interest 72 (Summer
1983). pp. 52-71.
Wilson. James Q., and Barbara Boland. Chapter 4 in The Urban Predicament, ed. William
Gorham and Nathan Glazer. Washington, D.C.: Urban Institute, 1976, pp. 179-230.
Estimates the incapacitation effect for different values of crimes per offender.
Prison Reform
Blumstein, Alfred. "Prisons: Population, Capacity, and Alternatives." In Crime and Public
Policy, ed. James Q. Wilson. San Francisco: Institute for Contemporary Studies, 1983.
Discusses alternative strategies for dealing with prison congestion, including new
construction and selective incapacitation.
Conley, J. "Economics and the Social Reality of Prisons." Journal of Criminal Justice 10
(1981), pp. 25-35.
Dilulio, John J. "Prison Discipline and Prison Reform." The Public Interest 89 (Fall 1987).
pp. 71-90.
Engel, Kathleen, and Stanley Rothman. "Prison Violence and the Paradox of Reform." The
Public Interest 73 (Fall 1983). pp. 91-105.
Drugs and Property Crime

Becker, Gary S. "Should Drug Use Be Legalized?" Business Week, August 17, 1987, p. 22.
Boyum. David, and Mark A. R. Kleiman. "Alcohol and Other Drugs." In Crime, eds. J.Q.
Wilson and Joan Petersilia, San Francisco: Institute for Contemporary Policies, 1995.
Clague, Christopher. "Legal Strategies for Dealing with Heroin Addiction." American
Economic Review 63, no. 2 (1973), pp. 263-69.
Moore, Mark. "Controlling Criminogenic Commodities: Drugs, Guns, and Alcohol." In
Crime and Public Policy, ed. James Q. Wilson. San Francisco: Institute for
Contemporary Studies, 1983. Discusses the roles of drugs, guns, and alcohol in
criminal activity.
. "Policies to Achieve Discrimination of the Effective Price of Heroin." American
Economic Review 63, no. 2 (1973), pp. 270-77.
Silverman, L. P., and N. L. Sprull. "Urban Crime and the Price of Heroin." Journal of
Urban Economics 4 (1977), pp. 80-103. Discusses the interactions between drug
policies and property crime.
Wilson, James Q. Thinking about Crime. New York: Basic Books, 1975. Chapter 7
discusses the relationship between heroin addiction and property crime and examines
the British system of legalized heroin.
Organized Crime
Allsop, Kenneth. The Bootleggers: The Story of Chicago's Prohibition Era. New Rochelle.
NY.: Arlington House, 1970.
Buchanen, James. "A Defense of Organized Crime." In The Economics of Crime and
1973.
Reuter. Peter, and Jonathan Rubinstein. "Fact. Fancy and Organized Crime." The Public
Interest 53 (Fall 1978). pp. 45-67.
Rottenberg. Simon. "The Clandestine Distribution of Heroin. Its Discovery and
Suppression." Journal of Political Economy 76 (January 1968). pp. 78-90.
Rubin. Paul. "The Economic Theory of the Criminal Firm." In The Economics of Crime and
Punishment, ed. Simon Rottenberg. Washington. D.C.: American Enterprise Institute.
1973.
Schelling. Thomas C. "Economics and Criminal Enterprise." The Public Interest 1 (1967).
pp. 114-26.
Miscellaneous
Gordon. David. "Capitalism. Class, and Crime in America." Review of Radical Political
Economy 3, no. 3 1971 pp. 5 1-75.
( ),
Kates. Don B.. Jr. "The Battle over Gun Control." The Public Interest 84 (Summer 1986).
pp. 42-52.
Katzman, Martin. "The Economics of Defense against Crime in the Streets." Land
Economics 19 (1968), pp. 431-40.
Phillips, Llad. and Harold Votey. Crime and Public Policy. Beverly Hills, Calif.: Sage
Publications. 1979.
Shoup. Carl. "Standards for Distributing a Free Governmental Service: Crime Prevention."
Public Finance 19 (December 1964), pp. 383^401 Discusses the trade-off between
.
efficiency and equity in the distribution of crime-fighting resources.

Simon, Carl P., and Ann D. Witte. Beating the System: The Underground Economy. Boston:
Auburn House. 1982.
Thurow. Lester. "Equity vs. Efficiency in Law Enforcement." Public Policy 18. no. 4
(1970). pp. 451-59.
Toby. Jackson. "Crime in the Schools." In Crime and Public Policy, ed. James Q. Wilson.
San Francisco: Contemporary Studies, 1983.
Institute for
Wright, James D. "Second Thoughts about Gun Control." The Public Interest 91 (Spring
1988). pp. 23-39.
Appendix
0~ L S of MICROECONOMICS
Model of Supply and Demand, 701

The Market Demand Curve, 701
The Market Supply Curve, 703
Elasticities of Demand and Supply, 704
Market Equilibrium, 705
Shortage or Surplus, 706
Market Effects of Changes in Demand or Supply, 707
Theory of Consumer Choice, 708
Constructing an Indifference Curve, 709
An Indifference Map, 712
The Budget Set and Budget Line, 712
Maximizing Utility, 715

Changes in Price and Income, 717
Theory of Worker's Choice, 717
Indifference Curves for Worker, 7 1
Budget Line for Worker, 718

Maximizing Utility, 719
Theory of Producer Choice, 719
Marginal Decision Making, 721
Marginal Benefits and Costs, 721
Using the Marginal Principle, 721
699
0~ LS of MICROECONOMICS
(~~7~ his appendix provides a review of some of the microeconomic tools that
t_x are used throughout the text.The first part reviews the model of supply and
demand. The second and third parts show how to use indifference-curve analysis to
explore the decisions of consumers and workers. The fourth part explains the model
of producer input choice, using isoquants and budget lines. The fifth part reviews
the notion of marginal decision making.
Model of Supply and Demand

The most frequently used model in economics is the model of supply and demand.
Economists use supply and demand curves to show how prices are determined in
markets and to predict the effects of changes in the market on equilibrium prices.
The Market Demand Curve

The market demand curve shows the relationship between the price of a particular
good and the quantity of the good purchased by consumers. To draw the market
demand curve, we change and observe how consumers respond, assuming
the price
that everything else that affects theconsumption of the good remains fixed. In Fig-
ure A-l, the market demand for shirts shows that an increase in the price of shirts
from $20 to $30 decreases the quantity demand from 260 (point c) to 200 (point e).
According to the law of demand, an increase in price decreases the quantity de-
manded, ceteris paribus (Latin for "'everything else remaining the same"). Therefore,
the demand curve is negatively sloped.
What sort of changes would cause the market demand curve to shift to the right
or left? A change in any one of the variables held fixed in drawing the demand curve
would cause the entire curve to shift.
701
702 Appendix
Figure A-l Market Demand Curve
"tt 20
Demand
200 260
Quantity of shirts per minute
The demand curve shows the relationship between price and quantity demanded, ceteris paribus
(everything else equal).
Income. In the jargon of economics, a "normal" good is defined as a good

for which there is a positive relationship between income and consumption:
an increase in income increases the quantity of the good demanded. In
Figure A-2, an increase in income would shift the market demand curve for
a normal good to the right, from D' to D". In other words, an increase in
income means that a larger quantity will be demanded at a given price of
shirts. For an "inferior" good, there is a negative relationship between
income and consumption: an increase in income decreases the quantity of

the good demanded. In Figure A-2. an increase in income would shift the
market demand curve for an inferior good to the left, from D" to D'
Price of a substitute. An increase in the price of a substitute good will shift
the demand curve to the right. Suppose that the curve shown in Figure A-2
is the demand curve for cotton shirts. An increase in the price of polyester
shirts would shift the demand curve for cotton shirts to the right: at every
price, more cotton shirts would be demanded. This is sensible because an

increase in the price of polyester shirts will cause some polyester
consumers to switch to cotton shirts.
Price of a complement. An increase in the price of a complementary good

will shift the demand curve to the left. If the curve shown in Figure A-2 is
the demand curve for cotton shirts, an increase in the price of pants will
Figure A-2 Shifting the Market Demand Curve

n D"
200
The demand curve shifts as a result of a change in income, a change in the price of a related good, or a
change in the number of people in the relevant market area.
shift the demand curve to the left (from D" to D')\ at every price, fewer
cotton shirts will be consumed. This is sensible because an increase in the
price of pants increases the cost of a dress outfit (shirt and pants), so
consumers will demand fewer outfits and thus fewer shirts.
4. Number of people. One of the variables held fixed in drawing the market
demand curve isnumber of people in the relevant market area. In
the
Figure A-2, an increase in the number of people will shift the demand curve
to the right, from D' to D": at every price, more shirts will be demanded.
The Market Supply Curve

The market supply curve shows the relationship between the price of a particular
good and the quantity of the good supplied by firms. To draw the market supply curve,
we change the price and observe how producers respond, assuming that everything
else that affects the production of the good remains fixed. In Figure A-3, the market
supply for shirts shows that an increase in the price of shirts (from $20 to $30)
increases the quantity supplied from 100 (point b) to 200 (point e). According to the
law of supply, an increase in price increases the quantity supplied, ceteris paribus.
Therefore, the supply curve is positively sloped.

704 Appendix
Figure A-3 Market Supply Curve
Supply
; 20
100 200
The suppl) curve shows the relationship between pnee and quantitv supplied, ceteris paribus.
What sort of changes would cause the market supply curve to shift? A change
in any one of the variables held fixed in drawing the supply curve would cause the
entire curve to shift.
1 Input prices. An increase in the price of an input used to produce a

particular good will increase the cost of production, causing firms to supply
a smaller quantity at each price. In Figure A—4. the supply curve for cotton
shirts would shift to the left (from S" to S') if (a the price of labor )
increased, (b) the cost of cotton increased, or (c) the cost of shirt-making
machinery increased.
2. Production technology. A technological innovation that decreases the cost
of production will shift the market supply curve to the right: at each price,
firms will supply a larger quantity because it is less costly to produce the
good. In Figure A-4. a cost-saving innovation in the production of shirts
would shift the supply curve from 5' to S"
Elasticities of Demand and Supply

Economists use two simple formulas to measure the responsiveness of consumers and
producers to changes in price. The price elasticity of demand for a particular good
Figure A-4 Shifting the Market Supply Curve
100 200 250

The supply curve shifts as a result of a change in input prices or a change in production
technology.
is defined as the percentage change in quantity demanded divided by the percentage
change in the price of the good. The demand elasticity measures the responsiveness
of consumers to changes in price: the larger the computed elasticity (in absolute
value), the more responsive are consumers to a change in price. Similarly, the price
elasticity of supply for a particular good is defined by the percentage change in
quantity supplied divided by the percentage change in price. The supply elasticity
measures the responsiveness of producers to changes in price: the larger the computed
elasticity, the more responsive are producers to a change in price.
Market Equilibrium
An equilibrium is a situation in which there
no tendency for things to change
is
over time. In the market for shirts, equilibrium means that there
is no tendency for
the price of shirts to change over time: tomorrow's price will be the same as today's
price. At the equilibrium price, firms produce the quantity of shirts that consumers
want to buy: the quantity demanded (by consumers) equals the quantity supplied (by
producers). Once the market reaches the equilibrium price, there is no tendency for
the price to change because there is neither a shortage nor a surplus of shirts.
We can use demand and supply curves to depict the market equilibrium. In
Figure A-5, the demand curve shows the negative relationship between the price of
706 Appendix
Figure A-5 Market Equilibrium and Disequilibrium

Supply
100 140 200 260 300

At the market equilibrium (point e, with priee = $30 and quantity = 200), the quantity supplied
equals the quantity demanded. At a priee less than the equilibrium price (e.g.. $20), there is a
shortage: the quantity supplied 100 at point b) is less than the quantity demanded (260 at
(
point c). At a priee exceeding the equilibrium price (e.g., $40), there is a surplus: the quantity
supplied exceeds the quantity demanded.
shirts and the quantity of shirts demanded: an increase in price decreases the quantity
of shirts demanded. The supply curve shows the positive relationship between the
price of shirts and the quantity of shirts supplied: an increase in price increases the
quantity of shirts supplied. The market equilibrium is shown by the intersection of
the supply and demand curves. At a price of $30, the quantity supplied (from the
supply curve) equals the quantity demanded (from the demand curve), so there is
neither a shortage nor a surplus of shirts: consumers want to buy 200 shirts, which
is exactly the quantity that producers want to sell.
Shortage or Surplus
If the price is less than the equilibrium price, there will be a shortage of shirts. For
example, if the price is $20, the market will be at point c on the demand curve
(consumers would like to buy 260 shirts) and point b on the supply curve (producers
are willing to sell only 100 shirts). In this case, there is a shortage or an excess
demand of 160 shirts: the quantity demanded (260) exceeds the quantity supplied
( 1 00) by 1 60 shirts. Because consumers want to buy more shirts than firms are willing
to supply, we would expect the price of shirts to increase over time: tomorrow's price
will be higher than today's price.
An increase in the price of shirts will cause the market to move upward along
both the supply and demand curves. An increase in the price of shirts allows firms
in the shirt industry to outbid other industries for the inputs required to produce
shirts (machines, labor, land, raw materials), so the output of the shirt industry
will increase. In other words, the market moves upward along the supply curve:
as the price increases, the quantity supplied increases. On the demand side, the
increase in price decreases the quantity of shirts demanded: the market moves upward
demand curve. In combination, the movements along the supply curve and
along the
thedemand curve narrow the gap between the quantity supplied and the quantity
demanded. The market continues to move upward along the two curves until the
shortage is eliminated. This occurs at point e, where the quantity supplied equals
the quantity demanded at a price of $30.
If the initial price exceeds the equilibrium price, there will be a surplus of shirts.
For example, if the price is at point j on the demand curve
$40, the market will be
(consumers would like to buy 140
and point k on the supply curve (producers
shirts)
are willing to sell 300 shirts). In this case, there is a surplus or an excess supply of
160 shirts: the quantity supplied (300) exceeds the quantity demanded ( 140) by 160
shirts. Because producers want to sell more shirts than consumers are willing to buy,
we would expect the price of shirts to decrease over time: tomorrow's price will be
lower than today's price.
As the price decreases, the market moves downward along both the supply and
demand curves. A decrease in the price of shirts means that the shirt industry will be
outbid by other industries for some of the resources used to produce shirts (machines,
labor, land, raw materials), so the output of the shirt industry will decrease. In other
words, the market moves downward along the supply curve: as the price decreases,
the quantity supplied decreases. On the demand side, the decrease in price increases
the quantity of shirts demanded: the market moves downward along the demand
curve. In combination, the movements along the supply curve and the demand curve
narrow the gap between the quantity supplied and the quantity demanded. The market
will continue to move downward along the two curves until the surplus is eliminated.
This occurs at point e, where the quantity supplied equals the quantity demanded at
a price of $30.
Market Effects of Changes in Demand or Supply

A change on either side of the market will change the equilibrium quantity and
price. Figure A-6 shows the effect of an increase in the demand for shirts. The initial
equilibrium shown by point e: price is $30 and quantity is 200 shirts per minute.
is
Suppose demand for shirts increases, shifting the market demand curve to the
that the
right. As explained earlier, this increase in demand could be caused by an increase
in income, a change in the price of a related good, or an increase in population. At
the initial price, there is a shortage of shirts, so the equilibrium price will increase.
708 Appendix
Figure A-6 The Market Effects of an Increase in Demand

New demand
Original demand
Supply
200 250 270

An increase in the demand for shirts shifts the demand curve to the right. At the initial price (S30).
there is a shortage: the quantity demanded < point /) exceeds the quantity supplied ( point e). The
resulting increase in price causes the market to mo\e upward along the supply curve and the new
demand curve until equilibrium is restored at point g.
We could also use the supply and demand curves to show the market effects
of a change in supply. In Figure A-7. a rightward shift of the supply curve would
generate a surplus of shirts at the original price: the quantity supplied will exceed the
quantity demanded. The new equilibrium (shown by the intersection of the demand
curve and the new supply curve) has a lower price and a larger quantity.
Theory of Consumer Choice

This part of the appendix provides an overview of indifference-curve analysis, which
is used in Urban Economics to analyze consumers' choices among alternative goods.
This material has been written for students who have not studied indifference-curve
analysis in other courses, but could also be used as a refresher for students with more
experience in economic analysis.
Tools of Microec onom it \s 709
Figure A-7 The Market Effects of an Increase in Supply

Demand
Original supply
New supply
200 230
An increase in the supply of shirts shifts the supply curve to the right. At the initial price (S30l. there is a
surplus: the quantity supplied (point /) exceeds the quantity demanded (point e). The resulting decrease in
price causes the market to move downward along the demand curve and the new supply curve until
equilibrium is restored at point g.
A consumer must decide how to allocate a fixed budget among a large num-
ber of consumer goods. Indifference-curve analysis is based on the notion that the
consumer has an objective (to maximize his satisfaction or "'utility") and a budget
constraint (a fixed income to spend on consumer goods). The consumer's prob-
lem is to find the affordable consumption bundle that maximizes his utility. The
analysis of the consumer-choice problem can be divided into two parts. An indif-
ference curve summarizes the consumer's subjective preferences about alternative
consumer goods. The budget line shows the limits (constraints) imposed on the
consumer.
Constructing an Indifference Curve
The simplest way to explain the notion of an indifference curve is to construct one.
Suppose that Mary gets satisfaction or utility from consuming hamburgers and hot
dogs. Figure A-8 shows the quantity of hamburgers on the horizontal axis and the
quantity of hot dogs on the vertical axis. To construct an indifference curve, we
710 Appendix
Figure A-8 Indifference Curve and Marginal Rate

of Substitution
Quantity of hamburgers
The indifference curve shows the different combinations of hamburgers and hot dogs that
generate the same utility level.
pick a point (e.g.. point C. with two hamburgers and five hot dogs) and pose two
questions:
1 Which combinations of hamburgers and hot dogs would Mary prefer to the
combination at point CI In other words, which combinations would
generate a higher level of satisfaction or utility ?
2. Which combinations of hamburgers and hot dogs would Mary consider to

be inferior to the combination at point C? In other words, what
combinations would generate lower utility?
Once we've established the combinations that are superior to point C (those that
generate higher utility) and inferior to point C (lower utility), the remaining combi-
nations are equivalent to point C in the sense that they generate the same level of sat-
isfaction or utility. If we connect these combinations, we have an indifference curve.
Mary will be indifferent among all the combinations along the indifference curve.
The indifference curve that passes through point C divides the combinations of
hamburgers and hot dogs into three groups.
1 . Superior combinations. All combinations above the indifference curve

generate more satisfaction (higher utility) than combination C. For
example. Mary would prefer point E or F to point C.
2. Inferior combinations. All combinations below the indifference curve

generate less satisfaction (lower utility) than combination C. For example.
Mary would prefer point C to point H or point /.
3. Equivalent combinations. All combinations along the indifference

curve generate the same satisfaction (the same utility) as combination C.
For example, Mary would be indifferent between combinations B, C,
and D.
Because indifference curves show the subjective preferences of consumers, the

shapes of indifference curves vary from one consumer to another. Nonetheless, the
indifference curves of all consumers share two characteristics: they are negatively
sloped and convex.
Why are indifference curves negatively sloped? If the two commodities are
"goods" (more of the commodity is better), not "bads" (more of the commodity is
worse), the indifference curve will be negatively sloped. In other words, if we give
Mary an additional hamburger, we must take away some hot dogs to restore her
original utility level. The slope of the curve is the marginal rate of substitution
(MRS) between the two goods. It shows the rate at which Mary is willing to substitute
one good for another. In Figure A-8, if Mary starts at point C and we give her one
more hamburger, we must take away one hot dog to keep her on the same indifference
curve. Therefore, the marginal rate of substitution near point C is 1 .0. Mary is willing
to trade one hot dog for one hamburger, so her subjective trade-off between the two
goods is 1 .0.
The slope of the indifference curve decreases (in absolute value) as the quantity
of hamburgers increases. In other words, the indifference curve becomes flatter as
Mary's hamburger consumption increases. In Figure A-8, the slope (the rise over
the run) near point B is 2.5: if we give Mary one more hamburger, we must take
away 2.5 hot dogs to keep her on the same indifference curve (to keep her utility
constant). Therefore, the MRS near point B is more than twice the MRS near point
C. The MRS near point D is 0.30: if we give Mary one more hamburger, we must
take away only 0.30 hot dog to keep her utility constant.
Because the slope of the indifference curve decreases as hamburger consumption
increases, the indifference curve is convex to the origin. The convexity results from
the assumption of diminishing marginal utility. The marginal utility of hamburgers
is defined as the increase in Mary's utility per additional hamburger. Suppose that
as Mary's hamburger consumption increases, the marginal utility of hamburgers de-

creases. As we move down Mary's indifference curve (increasing hamburgers and
decreasing hot dogs to keep her utility constant), we must take away progressively
fewer hot dogs to keep her at the same utility level because she gains progres-
sively less utility per additional hamburger. At the same time, as Mary's hot-dog
consumption decreases, the marginal utility of hot dogs increases, so a given de-
crease in the number of hot dogs has a larger effect on her utility level. Therefore,
it takes progressively fewer hot dogs to offset a given increase in utility from an
additional hamburger. The combined effect of decreasing marginal utility of ham-

burgers (as hamburger consumption increases) and increasing marginal utility of hot
dogs (as hot-dog consumption increases) means that the subjective trade-off between
712 Appendix
hamburgers and hot dogs decreases. In other words, the indifference curve becomes
flatter and the marginal rate of substitution decreases.
An Indifference Map
An indifference map is a set of indifference curves. To construct an indifference
map. we pick several starting points and then draw indifference curves through each
point. In Figure A-9. the starting points J. K. and L generate indifference curves
I\, h. and 73. As Mary moves from any point on the indifference curve 7] to any
point on h_. her utility increases. To show that her utility is higher along 7;. all
we need to show is that she prefers one point on h to one point on 7] Point K .
(on 7;) is obviously better than point J (on I\) because K provides more hot dogs
and more hamburgers. Under the assumption that both commodities are "goods."
she will prefer the combination with larger quantities of both commodities shown (
by point K In general. Man's utility increases as she moves in the northeasterly

).
direction to higher indifference curves (from 7 to 7; to 7;. and so on).

t
The Budget Set and Budget Line

Mary's ability to purchase hamburgers and hot dogs is limited by her income and
the prices of the two goods. Mary's budget set includes all the combinations of
Figure A-9 Indifference Curve Map
\n indifference map shows a set of indifference cur\es. with utilitj increasing

as we move nonhcasiorh to hisher indifference curves
hamburgers and hot dogs that she can afford, given her income and the prices of
hamburgers and hot dogs. Figure A- 10 shows Mary's budget set as the shaded area.
The budget set is drawn under the following assumptions:
1 Mary's income is $ 20, and she spends
1 all of her income on the two goods.
2. The price of a hamburger is $3 and the price of a hot dog is $2.
Given these numerical assumptions, Mary can afford all the combinations in the
shaded area.
The budget line shows all the combinations of hamburgers and hot dogs that
exhaust Mary's budget. In Figure A-10, the budget line is the line connecting points
M and N. Point M
is the combination that occurs if Mary spends her entire budget
on hot dogs: if the price is $2 and she spends her entire $ 20 budget on hot dogs, she
1
will get60 hot dogs. Point N is the combination that occurs if Mary spends her entire
budget on hamburgers (40 hamburgers from spending $120 on $3 hamburgers). If
Mary spends some money on each good, she can reach the points between and N M
For example, she could reach point Q (20 hamburgers and 30 hot dogs) by splitting
her budget equally between the two goods.
The slope of the budget line is the market trade-off between hamburgers and
hot dogs. It shows the rate at which the consumer can trade hamburgers for hot dogs,
given the market prices of the two goods. Starting from point M, if Mary gives up
one hamburger, she frees up $3 that can be spent on hot dogs. If the price of hot
dogs is $2, the $3 will buy .5 hot dogs, so the slope of the budget line is — .5: the
1 1
Figure A-10 Budget Set and

Budget Line
The budget set shows

all the affordable combinations of
hamburgers and hot dogs, and the budget line shows the
combinations that exhaust the budget.
714 Appendix
"rise" is 1.5 hot dogs and the "run" is 1 hamburger. In general, the slope is equal to
the price of hamburgers (the good on the horizontal axis) divided by the price of hot
dogs (the good on the vertical axis). If the prices of the two goods do not change as
Mary changes her consumption pattern, the budget line will be a straight line (the
slope will be constant).
Changes in the prices of the two goods will change the budget line. In Fig-
ure A-l 1, a decrease in the price of hamburgers from $3 to $2 will shift the line
from MN to MT. The original vertical intercept (point M) is still in the budget set: if
Mary spends her entire budget on hot dogs, she will be unaffected by the decrease in
the price of hamburgers. The horizontal intercept moves from TV (40 hamburgers) to
T (60 hamburgers) because at the lower price, a given budget buys more hamburgers.
If Mary buys some hamburgers and some hot dogs (she chooses some point between
the horizontal and vertical intercepts), she can afford more combinations because
she spends less per hamburger. The same argument applies to a decrease in the price
of hot dogs, shown in Figure A-l 1 as a shift from MN to R N (the price of hot dogs
decreases from $2 to $1.50).
How does an increase in income affect the budget line? For a given set of prices,
an increase in income increases the size of the budget set and shifts the budget line
outward. In Figure A-l 2. an income from $120 to $180 shifts the

increase in Mary's
budget line from MN to ST. She can now afford to buy 90 hot dogs if she buys only
Figure A-l 1 Changes in Prices Shift the

Budget Line
20 40 60
A change in the price of either good changes the slope of the budget line.
Figure A-12 Changes in Income Shift the

Budget Line
30 40 60
An increase in income shifts the budget line to the right.
hot dogs (point S, given a price of $2) or 60 hamburgers if she buys only hamburgers
(point T, given a price of $3). If she spends half of her income on each good, she
can buy 30 hamburgers and 45 hot dogs (point U).
Maximizing Utility
The consumer's objective is to maximize utility, subject to the budget constraint. In

other words, Mary will choose the affordable combination of hamburgers and hot
dogs that generates the highest utility. In graphical terms, Mary will choose a point
on the highest feasible indifference curve.
In Figure A-13, Mary will choose point V (22 hamburgers and 27 hot dogs)
and will achieve the utility level associated with indifference curve I5. Why does
she choose point V instead of point W, X. or K?
Point W: Mary doesn't choose this point for two reasons. First, it is not on the
budget line, so it does not exhaust her budget. Second, it is on a lower
indifference curve (and thus generates less utility) than point V.
Point X: Although point X exhausts Mary's budget. X lies on a lower indifference

curve than V. so it generates less utility than V (another point on the
716 Appendix
Figure A-13 Consumer Maximizes Utility at Tangency of

Indifference Curve and Budget Line
B 27 -
To maximize utility, the consumer finds the combination of hamburgers and hot dogs at which an
indifference curve is tangent to the budget line, meaning that the marginal rate of substitution
equals the price ratio.
budget line). Starting from point X, Mary could reallocate her budget
and buy more hamburgers and fewer hot dogs. As she moves down her
budget line, she moves to progressively higher indifference curves,
ultimately reaching point V and indifference curve I5.
Point Y: Although point Y generates a higher utility level (it's on a higher

indifference curve), it lies outside Mary's budget set, so she cannot
afford it.
At point V. Mary reaches the highest indifference curve possible, given her
budget set. The indifference curve touches —
but does not pass through the budget —
line. In other words, the indifference curve is tangent to the budget line. If the
indifference curve cut through the budget line at some point (e.g., at point X). Mary
could increase her utility by reallocating her budget between the two goods. A
tangency occurs at the utility-maximizing combination.
What is the economic interpretation of the tangency condition? At the tangency.
the slope of the indifference curve (the marginal rate of substitution) equals the slope
of the budget line (the price ratio). Therefore, the consumer's subjective trade-off
Tools of Microeconomics 1\1
between the two goods (the consumer's MRS) equals the rate at which the two
goods can be traded off in the consumer's budget (the price ratio). At any other
combination, the MRS would be unequal to the price ratio, so the consumer could
reallocate her budget and increase her utility. For example, starting from point X,
Mary's indifference curve suggests that she is willing to give up about six hot dogs
to get one hamburger (the slope of the indifference curve (the MRS) is about 6.0).
Given the prices of the two goods, the price ratio is 1.5: Mary must give up only 1.5
hot dogs to get one hamburger. Therefore, she will increase her utility by reallocating
her budget to consume more hamburgers and fewer hot dogs. She will continue to
reallocate her budget until the MRS equals the price ratio (until the slope of the
indifference curve equals the slope of the budget line). This occurs at point V, where
MRS = 1.5.
Changes in Price and Income

How do changes in prices affect the utility-maximizing consumption bundle? As
shown in Figure A-l 1, a decrease in price changes the shape of the budget set and
the slope of the budget line.The consumer will find the highest indifference curve
within thenew budget set, going to the point at which an indifference curve is tangent
to the new budget line. The consumer is likely to change the consumption of both
goods. The law of demand states that (a) an increase in the price of a good will
decrease the consumption of that good and (b) a decrease in price will increase the
consumption of the good.
How do changes in income affect the utility-maximizing consumption bundle?
As shown in Figure A- 12, an increase in income shifts the budget line outward, pro-
viding the consumer with more budget options. The consumer will find the highest
indifference curve within the new budget set, finding the point at which an indif-
ference curve is tangent to the new budget line. The consumer is likely to change
the consumption of both goods. A good is considered "normal" if an increase in
income increases consumption of the good (and if a decrease in income decreases
consumption). A good is considered "inferior" if consumption is negatively related
to income (an increase in income decreases consumption and a decrease in income
increases consumption).
Theory of Worker's Choice

Indifference-curve analysis can also be used to explore a worker's choice between
leisure time and labor time (income-earning time). In this case, the two items that
generate satisfaction or utility (the two "goods") are leisure time and income. People
get utility from leisure activities (relaxing, reading, bowling, watching movies, play-
ing sports) and also from the consumer products that can be purchased with wage
income. The trade-off is obvious: the more leisure time you take, the less income
you earn, so the smaller the quantity of consumer products you can purchase.
718 Appendix
Indifference Curves for Worker
The worker's preferences concerning leisure time and income are summarized in
indifference curves. The slope of an indifference curve equals the marginal rate of
substitution between leisure and income. In other words, the slope indicates the rate
atwhich the worker is willing to trade off income and leisure time. Figure A-14
shows three indifference curves, with leisure time on the horizontal axis and income
on the vertical axis.
Budget Line for Worker

The worker has a fixed number of hours to allocate between leisure time and income-
earning time. The budget line (BC in Figure A-14) shows the rate at which he can
trade leisure hours for income. The slope of the budget line is the wage: a one-hour
decrease in leisure time increases income by the hourly wage. In Figure A-14. we
assume that the worker has 100 hours per week to allocate between leisure time and
income-earning time, and the wage is $4 per hour. The horizontal intercept (point C)
is the maximum leisure time 100 hours). As leisure time decreases, income increases
(
by $4 per hour, from $40 at point D (90 hours of leisure and 10 hours of labor) to
$ 1 20 at point E (70 hours of leisure and 30 hours of labor), and so on.
Figure A-14 Worker's Choice between Leisure and Income
Leisure time per week
In maximize utility, the worker finds the combination of leisure time and income ut which
an indifference curve is tangent to the budget line, meaning thai the marginal rale of substitution
equals the wage
How would a change in the wage affect the budget line? A decrease in the wage
would decrease the slope of the budget line, causing it to pivot downward. At point
C, labor time and income are zero, so a decrease in the wage has no effect. As
leisure time decreases (work time increases), a given reduction of leisure time (a
given increase in labor time) generates less labor income. Therefore, the decrease in
the wage decreases the slope of the budget line. For example, if the wage drops to
$3 per hour, the income earned from 10 hours of work (90 hours of leisure) would
decrease from $40 to $30, so the budget point associated with 90 hours of leisure
would drop by $10.
Maximizing Utility
The worker will maximize utility at point E because it is the combination of leisure
and income that lies on the highest feasible indifference curve. The indifference curve
is tangent to the budget line, so the worker's subjective trade-off between leisure
and income (the marginal rate of substitution) equals the wage. At the optimum
combination, the marginal rate of substitution will equal 4.0 (the slope of the budget
line). For any other point along the budget line, the worker could reallocate his time
and increase his utility. For example, starting with point F, the marginal rate of
substitution is about 9.0, meaning that the worker is willing to give up $9 of income
to get one more hour of leisure time. Because he gives up only $4 for every additional
hour of leisure time, he will move down the budget line toward point E. He will
continue to trade income for leisure until his subjective trade-off (the MRS) equals
the wage, meaning that he will stop at point E.
How would an increase in the wage affect the amount of leisure time? An increase
in the wage has two effects on the worker's choice between leisure and labor time.
The first is the income effect: an increase in the wage means that the worker has a
higher real income. The worker will consume more of all "normal" goods, including
leisure. Therefore, the income effect provides upward pressure on leisure time and
downward pressure on work time. The second effect of an increase in the wage is the
substitution effect, or the opportunity-cost effect. An increase in the wage increases
the opportunity cost of leisure time: given the higher wage, each hour of leisure time
causes a larger reduction of income. As the opportunity cost of leisure time increases
(as the benefit of work time worker will tend to spend less time on
increases), the
leisure and more time working. In other words, the substitution effect provides down-
ward pressure on leisure time and upward pressure on labor time. It is not possible
to predict, on theoretical grounds, which effect will dominate, so it's impossible to
predict whether an increase in the wage will increase or decrease labor time.
Theory of Producer Choice

The basic logic underlying the consumer-choice model is also used in the economist's
model of a firm's decision concerning the choice of inputs. Consider a firm that
produces a particular good with two inputs, labor and capital. An isoquant shows
720 Appendix
the different combinations of labor and capital that will produce a given quantity of
output: iso means "equal" and quant means "quantity." In other words, the firm's
isoquants are analogous to the consumer's indifference curves. Figure A- 15 shows
three isoquants. each of which is associated with a different quantity of output.
The firm's budget set shows the affordable combinations of capital and labor,
given a fixed production budget. Like the consumer's budget line, a firm's budget
line shows the bundles that exhaust the budget. To draw a budget line, we need to
know the price of each input and the production budget. Let's assume the price of
labor is $4. the price of capital is $6, and the production budget is $240. Figure A-15
shows the firm's budget line: the slope is 1.5 (the price of capital divided by the price
of labor); the vertical intercept is 60 units of labor ($240/$4); and the horizontal
intercept is 40 units of capital ($240/$6).
The firm's objective is to maximize output, subject to its budget constraint. In
other words, the firm wants to reach the highest isoquant possible, given its budget
set. In Figure A-15. output is maximized at point E. where the isoquant is tangent to
the budget line. The slope of the budget line is the market trade-off between the two
inputs (the ratio of the price of capital to the price of labor). The slope of the isoquant
is the marginal rate of technical substitution (MRTS) between the two inputs, the rate
Figure A-15 A Firm Maximizes Output at the

Tangency of the Isoquant and the Budget
Line
Output = 140
Output = 120
Output = 100
22 40
Quantity of capital
In maximize output, the firm finds the combination oi capital and labor al which an
isoquant i*. tangent to the budget line, meaning that the marginal rate of technical
substitution equals the price ratio
at which one input can be substituted for another without changing total output. The
tangency condition means that the MRTS equals the price ratio, that is, the technical
trade-off between the two inputs equals the market trade-off.
Marginal Decision Making

The marginal principle provides a general rule for making decisions.
The Marginal Principle: Pick the level of an activity at which the marginal benefit of the
activity equals its marginal cost.
If the marginal benefit of some activity exceeds the marginal cost, you will be better
off if you do more of the activity. You should continue to increase the level of an
activity as long as the marginal benefit exceeds the marginal cost.
To explain the marginal principle, consider a problem facing Betty the beauti-
cian.Suppose that on a particular day, Betty must decide whether to keep her beauty
shop open for an extra hour, meaning that her shop would be open for 10 hours
instead of 9. To use the marginal principle, Betty must compare the marginal benefit
of staying open to the marginal cost: if the marginal benefit exceeds the marginal
cost, it will be sensible to remain open for the extra hour.
Marginal Benefits and Costs

On the benefit side, the marginal benefit of remaining open is the revenue generated
during the extra (tenth) hour. The total benefit of operating the beauty shop is the
revenue from beauty treatments, so the marginal — or incremental — benefit is the
increase in total revenue from the tenth hour of operation. For example, if Betty
expects to give five treatments at a price of $8 each during the tenth hour, the
marginal benefit of remaining open is $40.
On the cost side, the marginal cost of remaining open is the extra cost incurred
during the tenth hour. The costs associated with operating the beauty shop include
the monthly rent and the costs of electricity to power and light the shop. To use
the marginal principle, Betty must determine which of her costs increase when she
remains open for the extra hour. In this case, keeping the shop open for an extra hour
increases electricity costs and Betty's opportunity cost. The rent on the beauty shop
is irrelevant to the decision because the monthly rent is independent of the number
of hours the beauty shop is open.
Using the Marginal Principle

After Betty has estimated the marginal benefit and cost of operating the beauty
shop, she can make a decision about whether to remain open for an additional hour.
Suppose that Betty's opportunity cost is $25 per hour. In other words, the value of
her time in the next-best use is $25 per hour. If the cost of electricity is $5 per hour.
722 Appendix
the extra cost for the tenth hour (the marginal cost of remaining open) is S30. If the
marginal benefit (the additional revenue from treatments) is $40. the extra hour is
clearly profitable: the marginal revenue exceeds the marginal costby $10.
The marginal principle suggests that Betty should keep the beauty shop open as
long as the marginal benefit exceeds the marginal cost. She could use the same
marginalist logic for the eleventh, twelfth, and thirteenth hour. Eventually, the
marginal benefit will be lower than the marginal cost, and then Betty can close
the shop and go home. For example, suppose that the marginal benefit drops to $32
for the eleventh hour (she does only four treatments during the eleventh hour) and
then drops to $24 for the twelfth hour (three treatments during the twelfth hour). In
this case, the marginal benefit of the twelfth hour ($24) is lower than the marginal
cost ($30), so she should remain open for only 1 1 hours. To maximize her welfare,
the beautician picks the number of hours (11) at which the marginal benefit ($32) is
just above the marginal cost ($30).

It is worth noting two important ideas underlying the marginal principle. To use
the marginal principle. Betty must be careful how she measures her costs.
1 Betty ignores the fixed costs of the beauty shop, defined as costs that do not
varj with the time she remains open. For example, her monthly rent is
fixed, so she ignores this cost in calculating the marginal cost. A common
error in economic analysis is to include fixed costs in the calculation of
marginal cost.
2. Betty includes the opportunity cost of her time. Although Betty does not
receive an hourh wage, the labor cost associated with remaining open
equals the opportunity cost of her time. A common error in economic
analysis is to include only explicit monetary costs in the computation of
costs.
£T ndex
Aaron, Henry J., 442^44, 519, Alonso-Muth theory of income Associated Home Builders Inc. v.
540, 541 segregation, 230-234 City ofLivermore, 323

Abandonment (housing), ALS, 554 Athenian empire, 88
375-377 Altshuler, Alan A., 580, 582, 608, Atkinson, Reilly, 265
Absolute advantage, 19 612 Atlantic City, 605
Accelerated depreciation, 385 Amenities, 61 Automobile assembly, 53
Ackerman, Bruce, 582 American fur trade, 47, 48 Automobiles
Acs, Gregory, 365 Anas, Alex, 290, 300, 301 air pollution, and, 569-571
Administrative cities, 77, 79 Anderson, Edward J., 617 optimum speed, 572-574
AFDC, 357, 358 Anderson, John E., 76, 82 safety regulations, 572
Affirmative marketing policy, 398 Anderson, Martin, 445 small cars, 574, 575
Agglomeration economies Andes, Alberto F., 131, 137 suburbanization, and, 272, 275,
comparison shopping, 37 Anguish cost, 660 280
empirical estimates, 33, 34 Annas, Alex, 616 Average road cost, 561
localization economies, 26-32 Apartment complex, 312 Average travel cost, 549
office sector, 33, 34 Apgar, William C, 404, 406
shopping externalities, 36-38 Area licensing system (ALS),
urban growth, 147-149 554 Babcock, Richard R., 331, 332
urbanization economies, 32 Armington, Catherine, 177 Bahl, Roy, 477
Agricultural preserves, 318, 319 Arnott, Richard J., 266, 290, 300, Bailey, Martin J., 407
Ahlburg, Dennis A., 691 403 Baird, J. D., 582
Aid to Families with Dependent Aronson, J. Richard, 470, 478, Bakery, 53
Children (AFDC), 357, 358 540,541 Balancing-means rule, 324
Air pollution, 569-571 Arrest ratio, 669 Balkin, S., 692
Airport shuttles, 604 Arrow, Kenneth, 354, 365 Baltimore, 59,431
Alcaly, Roger E., 245 Aschauer, David A., 176 Bane, Mary Jo, 356, 365
Allen, Garland E.,444 Ashenfelter, Orley, 357, 365, 366, Banfield, Edward C, 350
Allsop, Kenneth, 696 502 Banta, John, 329
Alonso, William, 81, 99, 103, 136, Assessment of property tax, Baratz, Morton, 406
177, 187,204,230,246,330 517-519 Barber. E., 557,581
723
724 Index
Barr,J.,50] Birch. David, 177,265 Buchanen. James. 696

BART (San Francisco). 592. 596, Birdzell. L.E.. 103 Budget elections. 481—+88
597, 604, 608 Black. David E.. 501. 519. 541 Budget line. 709. 713
Bartel. Ann P.. 669, 678. 693 Black-capitalism approach. 346 Budget set. 712. 713
Bartik. Timothy J.. 72. 73. 75. 77, Blinder. Alan. 176 Buenos Aires, 130
81.82. 145. 146. 163. 165. Block grants. 530-532 Buffalo. 59
166. 175. 177.352.359.360. Blumstein. Alfred. 691. 696 Building permits. 305-308
364. 366 Boarding up. 374. 375 Burchell. Robert. 404
Barton. David M.. 416. 443 Boarnet. Marlon. 346. 350 Burg. A., 582
Barzel. Y„ 652 Boehm. Thomas P.. 41 1 . 41 2. 443 Burt. Martha. 361.362. 366
Basic sector. 1 40 Bogart. William T. 346. 350 Bus system. 592. 596. 603
Baskin.C.W., 136 Bogue. Donald. 298 Bush. John. 72. 73. 81
Bassi. Laurie. 357. 366 Boland. Barbara. 669. 692. 694. Business location decisions. See
Bateman. Worth. 350 696 Location decisions
Bauden. D. Lee. 365 Bollens. John C. 477 Business taxes. 144
Baumol. William. 466. 478 Bollinger. Christopher R.. 300. Buswa\. 593
Baumol model. 466-468 608.616 Button. Kenneth J.. 580
Becker. Charles. 72. 73. 81 Bone. A. J.. 299
Becker. Gary S.. 353. 365. 695. Boorstin. Daniel. 83
696 Border cities. 77. 79 California
Beckmann. M. J.. 136 Boskin. Michael. 540 impact fees. 315. 316
Beesley. Michael E.. 588. 613. Bosselman. Fred. 329 responsive public services. 431
614.618 Boston. 94. 430 taxpayer revolt. 470
Beet-sugar factories, 51 Bourne. L. S.. 137.405 tenant bribery. 439
Beltways. 271.299 Bowes. David 300R.. Callies. David. 329
Bendick, Marc. 175.445 Boyum. David. 684. 696 Campbell. E. Q.. 625. 650
Benefit analysis. 322 Bradbury. Katharine. 81. 277. 298. Canning. 51
Benefit principle. 483 444. 445. 479 Capitalization of taxes. 492
Bergstrom, T. 501 Bradford. David F. 346. 463. 464. Capitalized. 187.492
Berne. Robert. 65 465. 477. 478. 479. 643. 652 Carbon tax. 571
Berry, Brian J. L„ 125. 126. 136. Break. George. 477 Card. David. 628. 650
137.407 Brennan. Geoffrey. 478 Carelli. Richard. 679. 690
Bewley. T„ 501 Bridge. Gary, 653 Carlino. Gerald A.. 34. 43. 44
Bianchi. Suzanne M.. 353. 365 Brimmer. Andrew F.. 350 Carlton. Dennis W.. 74. 75. 82.
Bid-rent function Broker discrimination. 395 266
flexible/inflexible farmers. 191. Brooks. Mary E., 331 Cam. N.. 331
192 Brower. David J.. 331 Carol. H.. 137
income, and. 235-237 Brown. A. Theodore. 83 Carr-Hill. R.. 602
increased supply, and. 194 Brown. B.. 650 Carroll. Stephen J.. 652
manufacturers (monocentric Brown. C. E.. 693 Case. H. W.. 582
city), 206 209 Brown. James H.. 405 Castells. Manuel. 70. 82
office firms (monocentric cit) I, Brown. L.. 649 Cataldo. Everett F. 653
:i> l >-211 Broun v. Board of Education of Categorial grant. 526. 527
residential, 219-223 Topeka, 639 CDBG. 430
transport cost. and. 193 Brueckner. Jan K.. 235. 246. 265 Central county, 8
typical farmer. 1 88. 1 89 Brueggeman. William, 406 Central-citv education. 343
Index 725
Central-city problems, 276, 277 Coase solution, 320 Coons, John E., 651,654
Central place theory, 1 19-129 Coase, R. H., 320 Core-dominated city. See
Cervero, Robert, 28 1 , 299, Cobb, William, 691 Monocentric model of land
604-606,608,612,615,616 Coleman, James S., 625, 650, 652, use
Chaiken, Jan M., 690 653 Corn Laws debate, 196, 197
Chaiken, Marcia, 690 Coleman report, 625, 626 Corporate headquarters, 27
Changes in demand or supply, Colorado, 316 Cost-burdened households, 410
707, 708 Commercial firms, 47 Cotton baling, 51
Cherwony, W., 612 Communication technology, 282, Coulson,N. Ed, 607, 617
Cheslow, Melvyn D., 613 283 Courant, Paul N., 332, 407, 470,
Chicago Community development, 471,478,541,627,652
corporate service linkages, 291 429^31 Court system, 670-672
Gautreaux program, 343 Community Development Block Craig, Steven G., 693
subcenters, 287-289 Grants (CDBG), 430 Crandall, Robert W., 572, 574,
Chinitz, Benjamin, 83, 177, 297, Community service system, 670 582
300 Commuting, 226, 227, 274-276, Crane, Jonathan, 356, 365
Chipty, Tanseem, 366 589. See also Mass transit Crecine, John, 372, 405
Christaller, Walter, 106, 136 Comparative advantage, 19-21 Crime and punishment, 655-697
Churchill, Winston, 249 Comparison shopping, 37 burden of proof, 673
Ciccone, Antonio, 44 Compensation rules, 323, 324 city size, 680, 681
Citrin. Jack, 470, 478 Competition and land rent, 185 costs of crime, 658
City, 7 Complementary goods, 38 court system, 670-672
Clague, Christopher, 696 Computer industry, 32 crime-supply curve, 661, 662
Clapp, J. M., 43, 245, 299 Computer manufacturers, 70, 7 decrease value of expected
Clark, Peter, 480 Conditional grants, 526, 527 loot, 667, 668
Clark, Thomas A., 407 Congestable public good, 457 deterrence, 678-680
Clark, William A., 340, 349 Congestion, 546-550 drug usage, 682-685
Clawson, Marion, 247 Congestion externality, 549 errors, 673
Clean Air Act, 569 Congestion tax, 550-556, facts/statistics, 655-658
Clearance rate, 670 565-569 hardening the target, 667
Cleveland Conley, J., 696 housing prices, and, 657, 658
affirmative marketing policy, Consolidated metropolitan incapacitation, 679
398 statistical area (CMSA), 9, 10 increase value of legal
improved housing, 430 Construction subsidies, 380-382 activities, 666
neighborhood racial change, Consumer choice, 708-717 index crimes, 655
397, 398 Consumer surplus, 551 insurance, 668
Von Sweringen brothers Consumption effect, 274 juvenile vs. adult penalties,
(streetcar line), 275 Contracting for transit services, 680
Clinton, Bill, 358 603, 604 moral-hazard problem, 668
Closed-ended matching grants, Conversion (housing), 374 optimum amount of crime,
529, 530 Convex housing-price function, 662-665
Clotfelter, Charles T., 639, 642, 218,219 parole system, 677, 678
692 Convex land-rent function, 1 89 plea bargaining, 672,
Cloth factory, 21-23 Cook, Philip J., 692 673
Cluff, George S., 580, 582 Cook, Thomas D., 695 police, 669, 670
Clune, William H., 651 Cooley, Thomas F., 330 prevention, 666-680
726 Index
principle of marginal Department of Housing and Urban Drug usage. 682-685

673-676
deterrence. Development (HUD). 409, Duncombe. William. 637. 652
prison system, 676-680 410 Dyett. Michael V, 608, 617
rational criminal model, Design zoning. 316-319 Dylan. Bob. 35
659-662 Deterrence. 678-680
rehabilitation, 677 Deutsch. Joseph, 691
three-strikes laws, 679, 680 Development approach. 346. 347 E-mail. 282, 283
typical dispositions. 670-672 Development coupons. 317. 318 Eaton. Susan. 45. 343. 349
victim cost /prevention cost, Development tax. 315 Ebel, Robert. 247. 298. 299. 300
662, 663 Development zone. 3 1 Eberts, Randall W.. 146. 147.
victims. 657 Dewees, Donald N., 612, 617 167-169. 175-177.501.
why commit crime? 661 Dial-a-ride vans, 604 651
\\ h\ study crime? 655 Diamond. Douglas B., 408 Echols. James C. 603. 615
Cuciti, Peggy, 480 Diamond lanes. 554, 593 Eckert, Ross, 616
Curtis. B., 45 Dicken. Peter, 137 Economic base study, 153-157.
Cutler. David M., 341. 342, 346 Dickens, William T.. 695 160-162
Dieckman. D., 693 Economic city. 7
Dilulio, John J., Jr.. 679. 690. Economic growth. 139. See also
Dahlgren, Joy. 593. 612 696 Urban economic growth
Dallas, 559 Diminution-of-value rule. 323 Economies of scale. See Scale
Danzinger. Sheldon H., 365. 366 DiPasquale, Denise, 176 economies
Darden, Joe T.. 407 Direct effect, 144 Edel. Matthew. 177. 204. 265.
Darity, William A.. 354. 356, Direct income transfers, 357, 358 298.350.579.614
365 Directed development. 316-318 Edelson. N.. 501
Das id. Philip, 245 DiRenzo. J.. 557. 581 Edge cities. 289. 290
Davidoff. L.. 33 Discrimination Edson. Charles L.. 442
Davidoff, P., 33 broker. 395. 396 Education. 621-654
Davis, Kingsley, 86. 90. 103 labor market. 393-396 class size. 627. 628
Davis, Otto. 37*2. 405. 445. 501 racial. See Racial Coleman report, 625. 626
De Leeuw, Frank, 378, 405 discrimination, curriculum. 627
De Salvo, Joseph S., 405. 438, Residential segregation desegregation policy. 637-643
445 Discrimination in labor markets. foundation grants. 632-634
Dear. Michael, 137 353-356 GTB plan. 634-636
Deaton, B. J.. 693 Diseconomies of scale. 472 home environment. 624
Decoupling, 283 Dispersal policy. 346 instructional equipment. 627
Deen, Thomas B., 617 Distribution cost. 48 legal mandate to equalize
Defensive city. 86, 87 Distribution phase,589 expenditures. 630. 631

Delafons. John. 329 power equalizing, 634
District peer group effects. 624. 64
Demand and supply. 701-708 Domenich. Thomas A., 579, 614 private schools, 639-643
Demand density. 1 1 Dorfman. Robert, 582 production function. 622-628
Demand elasticity, 705 Dowall, David E.. 331. 332. 346. reform, 631. 637
Demand experiment. 427 350 school resources. 624
Demise of small stores. 118, 119 Downs. Anthony, 81, 277. 298. spending inequalities. 629-637
Denike. K. G.. 136 406, 444, 445, 466. 477. 50 1 teachers. 626. 627
Density function. 272, 273 580.581.617 Tiebout process, 628. 629
Density gradients. 273 Dressmaking industry. 27 voucher program. 644-646
Index
727
white flight to private schools, Exclusionary zoning, 314 Fiscal zoning, 313-316
639-643 Expected crash cost, 573 Fischel, William, 322, 323, 330
Education externalities, 455, 456 Experimental Housing Allowance Fisher, P., 594, 595, 596, 614
Education vouchers, 644-646 Program (EHAP), 427, 428 Fisher, Ronald C, 493, 502
Educational achievement, 343 Export sector, 140 Fisher, Ronald J., 617
Effective tax rate, 5 1 Express lanes, 554, 593 Fitts, Jerry J., 444
Effluent fees, 309-31 1, 570, 571 Externalities,455^457 Fixed costs, 722
EHAP, 427, 428 Externality zoning, 308-31 Fleisher, Belton, 666, 692
Ehrilich. Isaac, 669, 693 Florida, 316, 359
Ekanem, Nkanta, 378, 405 Florida, Richard, 71,82
Elasticities of demand and supply, Face-to-face communication. Flypaper effect, 525, 526
704, 705 34-36 James R., 392, 404, 405
Follain,
Electricity, 60 Factor specialization, 22 Food stamps, 358

Electronic mail, 282, 283 Facts/statistics Foot voting (Tiebout model),
Electronic road pricing (ERP), 554 crime and punishment, 488-494
Ellwood, David T., 349, 356, 365, 655-658 Forst, Brian, 670, 671,694
392, 406 local government, 449^152 Forth Worth, 430, 43
Emerson, Ralph Waldo, 47 mass transit, 585-589 Foster, CD., 618

Employment poverty, 335-338 Foundation grants, 632-634
monocentric city, 224 residential segregation, 339 45-degree line, 62
suburbanization of, 277, 278 suburbanization, 267-269 Fox, James A., 691
Employment discrimination, 353 Fagin, Henry, 330 Franklin, H. M, 331
Employment growth, 162-169 Fair-housing audit, 395 Free-rider problem, 484

Employment multiplier, 142 Fair market rent, 420 Freeman, A. Myrick, 405, 582
Energy inputs, 59, 60 Falk,D., 331 Freeman, Richard B., 343, 349,
Energy-intensive firm, 59 False negative result, 673 468, 479, 540, 658, 666, 677,
Engel, Kathleen, 696 Farb, Warren E., 404 690, 692
Engle, Robert F, 519, 541, 607, Fare-box ratio, 601 Frey, W. H., 277, 298
617 Farley, John E., 407 Frieden, Bernard J., 279, 297,
English woolen industry, 24, 25 Fauth, Gary R., 582 444

Enterprise zones, 346 Feldstein, Martin, 693 Friedlander, Ann F, 582
Environmental quality, 150-153 Female-headed households, 356, Friedlander, Daniel, 357, 366
Epple, Dennis, 645, 654 357 Friedman, John, 136, 177

Equal protection, 322 Fertility analysis, 183-187 Fringe development, 315, 316
Equalization of spending, 464 Feudal cities, 89-101 Fringe housing, 315

Equalization of taxes, 463, 464 Field, David Dudley, 198 Fullerton, Don, 389, 407
Equilibrium, 705 Fielding, Gordon J., 601, 614 Fur trade, 47, 48
Erickson, R., 298 Figlio, David, 470, 479
ERP, 554 Filer, Randall K., 360, 362, 366
Escuerdo, Emilio, 608, 617 Filimon, R., 526, 541 Gabriel, Stuart A., 340, 344, 349
Euclid v. Ambler, 321 Filtering model, 378-382 Gaitlin, Douglas, 653

Evans, Catherine, 62 Finke, Jeffrey K., 580, 582 Galster, George C, 397, 399,
Evans, Robert, 695 Fire protection, 457 408
Evans, William N., 629, 636, 637, First cities, 85-87 Gambling industry, 77
652 Fiscal problems (central city), Gamoran, Adam, 642, 653
Everett, Carol T., 556, 581 471^174 Gardner, John L., 176,478
728 Index
Garms, Walter I., 649 Gomez-Ibanez, Jose A.. 555. 565. Gyimahbrempong. K... 478
Garnick, Daniel H., 99, 101. 566. 574. 575. 579. 580, 582, Gyourko. Joseph. 434. 436, 437,
103 597.598,606,607.613-615 445, 472, 479
Garreau, Joel. 289. 300 Goodman. R.. 501
Garrison. William L.. 125. 126. Gordon, David. 697
136 Gordon. Peter. 597. 615 Haggler. 118
Gasoline prices. 607, 608 Gordon. Roger H.. 389. 407 Hakim. Simon. 691,693
Gasoline tax. 556. 570. 571 Gorham. William. 479. 613. 653, Hall. Peter. 43
Gaspar. Jess. 34, 35, 45 692, 694, 696 Hall, Robert E.. 44
Gender discrimination, 354-356 Gottman, J., 35, 45 Hamer. Andrew Marshall. 614.
General Assistance, 358 Gottschalk, Peter. 352. 365 617
General-equilibrium analysis, Gould. P.. 137 Hamilton, Bruce W.. 136. 145.
249 Government loans and loan 175.490.501.502
General-equilibrium land use. guarantees. 65, 150 Hanson, Gordon H.. 43. 71
249-266 Graham. Stephen. 45, 300 Hanushek. Eric A.. 444. 622. 623.
conditions, 249-25 Gramlich, Edward M.. 352. 365. 627.651
increase in export sales, 470.471.478.492.493.502. Harberger. Arnold C. 404
253-256 541.652 Harm-prevention rule. 324
numerical examples, Grants. See Intergovernmental Harney. Kenneth R.. 442
251-259 grants Harnischfeger. Annegret. 651
property tax. 259-262 Graves. Phillip E.. 61,81. 176.478 Harrison. Bennett. 175. 297. 350
streetcar system, 256-259 Gray. Charles M.. 657. 691. 692 Hartmann. David J.. 408
Gentrification. 279 Greek cities. 87. 88 Heavy industry. 3 1
George. Henry, 198,204 Green belts. 318. 319 Heavy-rail (tixed-rail)

Getz. Malcolm. 137 Greenberg. David F. 695 systems. 592. 596. 597. 608.
Giant cities. 129-132 Greenberg. David H.. 357. 366 609
Giarratani. Frank. 176 Greene. Kenneth V.. 477 Heckman. James J.. 355. 365
Giles. Michael W.. 653 Greenwood. Michael, 81. 297 Hedonic approach. 368-370
Gini curve, 62 Greenwood, Peter. 695 Heenan. Warren G.. 617
Ginsberg. A.. 649 Griesenger, Harriet. 366. 694 Heikkila. Eric J.. 493. 502
Giuliano. Genevieve. 284-286, Grieson. Ronald E.. 247. 265, 266, Heilbrun. James. 406
300.546.580.593.603.613. 332. 405. 480 Hekman. John S.. 70. 82
615.617 Grisby. William. 406 Helms. L. Jay. 76. 82
Glaab. Charles. 83 Grogger. Jeffrey. 666. 678. 690. Helsley. Robert. 300
Glaeser, Edward L., 32, 34. 35, 44, 694 Hendershott. Patric H.. 388. 407
45. 131. 137.341.342.346. Gronberg. Timothy J.. 501 Henderson. J. Vernon. 33. 44. 175.
680.681.690 Growth of U.S. cities, 94-101 176.290.300.502
Glaister, Stephen. 615 Grubb. W. N.. 298 Hernderson. R., 32, 44
Glaser, Daniel. 695 Gruenscpecht, Howard. 572. 574. Hemstein, Richard J.. 691
Glatt. Evelyn S., 444 5S2 Hesletine. P. M.. 607. 615
Glaze, Nathan, 479. 613, 653. 692. GTB plan. 634-636 Hewings. Geoffrey J. D.. 81
694. 696 Guaranteed tax base (GTB). Hierarchal system of cities. 122
GM's Saturn plant, 73, 74 634-636 Higgs. Robert. 43
Goldberg, I., 694 Gunning. J. Patrick. Jr.. 691 High-density housing, 313
Goldhaber, Dan D.. 642. 653 Guthrie, James W., 631, 649, 651, High-occupancy and toll (HOT)
Goldstein, Gerald. 617 654 lanes. 555
Index 729
High-occupancy vehicle (HOV) community development, taxes, 385-389

lanes, 554, 593, 594 429-43 Urban Renewal, 429, 430
High-technology firms, 27, 28 condition of housing stock, used vs. new, 378-380
Highes, James, 82 409-111 vouchers, 423^126
Highways. See Automobiles, construction subsidies, Housing allowances, 427
Urban transportation 380-382 Housing-price function, 216-219
Hilley. John L.. 470, 478, 540, 541 conversion, 374 Housing-price gradient, 219
Hilton, George W., 614, 615, 616, cost-burdened households, 410, Housing vouchers, A23-A26
617 411 Houston
Hinds, D., 331 crime, and, 657, 658 city without zoning, 320
Hines, James R., 389, 407 demand-side policy, 421-429 HOV system, 593, 594
Hirsch, Werner, 478 deterioration/maintenance, 373, subcenters, 287
History of Western urbanization, 374 HOV lanes, 554, 593, 594
85-103 durability, 372-378 Hoynes, Hilary W., 356, 365
feudal cities, 89-101 Experimental Housing Hoyt, Homer, 245
first cities, 85-87 Allowance Program, 427, Huang, J. C., 137
Greek cities, 87, 88 428 Huckster, 118
industrial revolution, 92-94 fair market rent, 420 HUD, 409, 410
mercantile cities, 91, 92 filtering model, 378-382 Hughes, James W., 299, 300
Roman cities, 88, 89 hedonic approach, 368-370 Humphrey, Nance, 479
United States, 94-101 heterogeneity/immobility, Hutchinson, Bruce G., 33, 44
Hitchcock, J. R., 405 368-372
Hobson, C. J., 625, 650 inadequate dwellings, 410
Hochman, Harold, 103, 330, 350, income/price elasticity, 391, Ibsen, Henrik, 449
477, 652 392 Ichniowski, Casey, 470, 479
Hoffer, Thomas, 652 mortgage deduction, 385-389 Ihlanfeldt, Keith R., 300, 345,
Hohenberg, Paul M., 89, 90, 103, moving cost, 390, 391 349,392,406,411,412,443,
235, 246 neighborhood dissatisfaction, 519,541,608,616
Holland, Edward P., 580 411-413 Illegalgoods, 682-685
Holly, Brian, 43 neighborhood effects, 372 Impact fee, 315, 316
Holzer, Harry J., 343, 349, 352, neighborhood transition Imperfect substitutes, 36-38
359, 366, 666, 690 (mixed-race Import substitution, 161
Homelessness, 360-363 neighborhood), 396-399 Impure public goods, 457
Honig, Marjorie, 360, 362, 366 overcrowding, 410 Imputed rental income, 388
Hoover, Edgar M., 43, 81, 136, public, 413-121 In-kind programs, 358
176,298 rapid depreciation, 385 Inadequate dwellings, 410
Hoselitz, Bert, 103 rent control, 43 1—139 Incapacitation, 679
HOT lanes, 555 renting vs. homeownership, Inclusionary zoning, 314, 315
Hotelling, Harold, 501 383-389 Income effect, 469, 532
Housing, 367—146. See also retirement decision, 374, 375 Income multiplier, 142
Residential segregation section 8 rent certificate, Income redistribution, 452, 453
abandonment, 375-377 421-423 Income segregation, 230-237
allowance, 427 submarkets, 371, 372 Income tax, 521, 522
boarding up, 374, 375 supply elasticity, 377, 378 Incubator process, 31, 32
choosing a dwelling, 370, supply-side policy, 413—121, Index crimes, 655
371 428, 429 Index of fiscal health, 472-174
730 Index
Indifference curve. 709 Internal scale economies. 21-24 Kemp. Michael A.. 588. 613-615
Indifference-curve analysis Intracity truck, 269-27 Kendig. Lane. 330
consumer choice. 708-7 1 Intrametropolitan location Kent. Mary M.. 348
worker's choice. 7 7-7 1 1 decisions. 75 Kern. Clifford R.. 279. 298
Indifference map. 712 Invasion rule. 323 Key money. 328. 439
Indivisible inputs. 22 Irby, Iredia. 443 Kilgore. Sally. 652
Industrial bonds. 64. 65. 150 Isserman, Andrew M., 176 Killalea. J.. 649
Industrial firms, 47 Izraeli. Oded, 176 King, Leslie J.. 136
Industrial nuisances. 309 King. Mervyn A.. 389. 407
Industrial revolution. 92-94 King. Thomas. 246. 405. 408
Inferior good. 702 Jackson. Jerry. 246 Kirby. Ronald F. 615. 616
Inflation and homeovvnership. 389 Jackson. John. 372. 405 Kitigawa. Evylyn M„ 298
[nghram, Cheryl A.. 331 Jackson. Kenneth T.. 585 Kleiman. Mark A. R.. 684. 696
Ingram. Gregory K., 404 Jacobs. Barry G.. 442 Knight. Richard. 298
Inman. Robert P.. 476. 487. 501. Jacobs, Jane. 32. 44 Knowledge spillovers. 30, 31
530. 532. 542. 652. 653. 643 Jaffe, A.. 32. 44 Kraft. Gerald. 579. 614
Input coefficients, 158 Japanese automobile firms. 71 Kraft. John. 416. 443
Input-oriented firms. 59-66 Jenkins. D. T., 42 Kraus. Marvin. 582
Input-output study. 157-162 Jiminez. Emmanuel. 392, 405 Krohm. Gregory. 69
Insights. 2 Jitneys. 605 Krueger. Alan B., 628. 650
Institutionalization of mentally ill. Job discrimination. 354 Krugman. Paul. 24. 43. 61. 62. 81.
363 Job-training programs. 357 130. 131. 137
[ntercit) truck. 271,522-534 Joelson, Mitchell. 657. 691 Kulash. Damian. 581
Intergovernmental grants Johnson. Christine M.. 616
block grants. 530-532 Johnson. T. G.. 693
closed-ended matching grants. Johnston. Robert A.. 331 Labor-intensive firm. 60
529. 530 Joint consumption. 457 Labor market and economic
flypaper effect. 525. 526 Joint labor supply. 60 growth. 140-149
grants to high-tax communities. Junction cities. 57-59. 77 Labor-market discrimination.
534 Just compensation. 323. 324 353-356
grants to low -income .luster. F. Thomas. 416. 443 Labor-market pooling. 28-30
communities. 533. 534 Labor productivity, 144
lump-sum conditional grants. LaCivita. C. J.. 330
526. 527 Kain. John F.. 245. 246. 340. 344. Ladd. Helen F. 471-473. 478.
matching grants. 527-530 349. 350. 364. 372. 404-406. 479. 542
open-ended matching grants. 42S. 444. 613. 614. 617 LagO, Armando M.. 614
527-524 Kallal. Heidi D.. 32. 34. 44 Lake. Steve. 72. 73. 81
reasons for, 523. 524 Kanter. Sandra. 1 75 Land rent. Sec also Bid-rent
statistics. 522. 523 Kates. Don B.. Jr.. 697 function
stimulative effects, 530 Katz, Jeffrey, 624. 630 accessibility and. 187-195
unconditional grants. 524-526 Katzman, Martin. 697 competition, and. 185
welfare reform, and. 53 I. 532 Ka/imi. Camilla. 564. 583 Corn Laws debate. 196
Interjurisdictional, 466 Keeler, Theodore E.. 552. 572. defined. 181. 1 82
Intermediate location. 53 574. 580-582. 594-596. 614 fertility. 183-187
Intermetropolitan location [Celling, G. L., 643 housing prices, and. 196
decisions, 75 Kelly, Sara, 653 market interactions. 146. 147
Index 731
public policy, and, 185-187 Latent demand, 558, 559 facts/statistics, 449^452
single tax, 198, 199 Lav, Michael R., 136 fiscal problems, 471^474
198-200
taxes, and, Lave, Charles A., 603, 613, 616 grants. See Intergovernmental
Von Thunen model, 187 Lave, Lester B., 572, 574, 582 grants
Land-rent function, 188, 189 Law of demand, 701,717 growth of local spending,
Land-rent gradient, 221-223 Law of supply, 703 466-471
Land-use controls and zoning, Ledebur, Larry, 1 75 income tax, 521, 522
303-332 Lees, Lynn H., 89, 90, 103, 235, interjurisdictional spillovers,
building permits, 305-308 246 466
commercial/industrial Leftover principle, 185 local public goods, 457^160
development, 316 Legal environments, 77 median- voter model, 481^88
compensation rules, 323, 324 Leroy, S. F., 266 metropolitan consolidation,
constitutionality of zoning, LeRoy, Stephen, 246 463-465
321-324 Letellier, Johnson, 655 metropolitan federation, 465
design zoning, 316-319 Lett, Monica, 445 municipal vs. metropolitan
directed development, 316-31! Leven, Charles L., 301 system, 494-497
equal protection, 322, 323 Levin, A., 331 natural monopoly, 453^455
exclusionary zoning, 314 Levin, Henry, 645, 654 property taxes. See Property
fiscal zoning, 313-316 Levine, James, 693 taxes
fringe development, 315, 316 Levitan, Don, 299 public-safety externalities. 456,
history of zoning, 308 Levitt, Steven D., 669, 676, 680, 457
inclusionary zoning, 314, 315 690 role of, 452^160
314
large-lot zoning, 313. Levy, Daniel C, 653 sales tax, 519-521
legal environment, 321-324 Levy, John M., 64, 81, 149, 175 scale economies vs. demand
nuisance zoning, 308-31 Light rail, 597-599 462
diversity,
open-space zoning, 318, 319 Lillydahl, Jane H., 315, 330 tax expenditures, 534-536
performance zoning, 311, 312 Lindahl approach, 483 tax/spending limitations, 470,
population growth, 303-308 Lindholm, R., 501 471
substantive due process, 321, Line-haul economies, 54 Tiebout model, 488^94
322 Line-haul phase, 589 types of policies, 452, 453
unzoned cities, 320 Linear housing-price function, Local-government bonds
urban growth boundary vs. 216,217 (tax-free), 536
urban service boundary, Linear land-rent function, 188 Local-input costs vs. transport
303-306 Linneman, Peter, 434, 436, 437, costs, 66-69
Land-use patterns 445 Local inputs, 59
congestion, and, 565-569 Lipsey, Mark, 677, 695 Local public goods, 457^60
fixed-rail systems, and, 608, Lipsey, Richard G., 45 Local public services and taxes,
609 Lloyd, Peter E., 137 63-65
gasoline prices, and. 607, 608 Local government, 447-542 Local sector, 140
Landes, William, 694 advantages/disadvantages, 460, Local taxes
Landis, John, 608, 616 461 deductibility of, 534, 535
Lane, Bruce S., 442 Baumol model, 466-468 income tax, 521, 522
Laren, Deborah, 352, 365 education externalities, 455 456 , property taxes. See Property
Large-lot zoning, 313, 314, 491 externalities, 455^57 taxes
Laska, Shirley, 298 externalities vs. demand sales tax, 519-521
Lassar, Terry J., 331 diversity, 461,462 Localization economies, 26-32, 61
732 Index
Location decisions. 47-83 Maisel. Sherman, 443 Mass transit. 585-618

case studies. 70-73 Majority voting. 483 British experience. 606, 607
empirical studies. 73-77 Malpezzi. Stephen. 439, 445 bus system. 592. 596. 603
government decisions, and, 77 Malt. R. A.. 478 choosing a travel mode.
input-oriented firms, 59-66 Manhattan dressmaking industry, 589-597
market-oriented firms, 5 1-53 27 commuting trip. 589
port cities. 57-59 Mann. Taynia L., 348 contracting for transit services.
principle of median location, Manufacturing and 603. 604
55-59 suburbanization. 269-272 deregulation. 604
resource-oriented firms. 49-5 Marginal benefit. 721 facts/statistics. 585-589
trading cities. 47. 48 Marginal-benefit curve, 548 fare increases. 588
transfer-oriented firms. 48-55 Marginal cost. 721 fare-box ratio. 60
transport costs vs. local-input Marginal cost of speed. 573 fiscal problems. 601, 602
costs. 66-69 Marginal decision making. 721. fixed-rail systems. 592. 596.
Location-quotient approach. 722 597. 608. 609

154-156 Marginal principle. 721 HOV lanes. 593. 594
Locational equilibrium. 217 Marginal rate of substitution. 7 1 land-use patterns. 607-609
Locational Gini coefficient. Marginal travel cost. 549 light rail. 597-599
62-64 Mariano. Ann, 442 natural monopoly, as. 600
Loeb. Susanna, 637. 652 Market area. 22. 106 paratransit, 604-606
London. 130.273 Market-area analysis, 106-119 peaking problem. 601. 602
Los Angeles demise of small stores, 1 18, price elasticity. 588
corporate service linkages. 241 119 restricting entry into the
impact fees. 316 determinants of market area market. 602. 603

parking tax, 557 (fixed demand). 110-116 scale economies. 603
ramp lights. 559 efficiency trade-offs. 109. 1 10 service improvements. 588. 589
Red Line. 597 entry and competition. 108, 109 subsidies. 599-602
subcenters, 284-287 industry-to-industry variations. time elasticity. 588
Losch, August. 106. 136 119 value of travel time. 588
Lot size. 224 law of demand. 117. 118 Matching grants. 527-530
Low-density zoning, 3 1 market areas. 1 10 Mathewson, Kent. 652
Lovvry. Ira S., 406 pricing with a monopolist. 106, Mathur. V. K.. 694
Lubuele. Luan Sende, 322. 33 1 107 Maxwell. Harold. 666. 692
338, 349 Market centers. 55 Mayer. Neil. 387. 407. 442
Lump-sum conditional grants. Market demand curve. 701-703 Mayo. Stephen K.. 406
526, 527 Market equilibrium. 705, 706 Mayworm, Patrick. 614
Lustig. Morton. 332 Market-oriented firms. 51-53 McConnell-Fay, Natalie. 552. 580
Lynn, Lawrence E., 101, 103 Market supply curve. 703. 704 McCullough. William. 603. 616
Lyons. Donald J.. 43 Market trade-off. 713 McDonald. John F.. 278. 287. 289.
Market value of land. 181. 182 300. 301.692
Markusen. Ann. 43 McDonald. M. Brian. 652
MA, 7-9 MARTA 608
(Atlanta). McElhiney. William. 581
MacKinnon. .1. (!.. 266 Martinson. Robert. 645 McEnroe. J. Matthew. 614
Maclennan, Duncan. 406 Marvel. Thomas B.. 679. 690 McFadden. Daniel. 6 14
Macrae. Norman. 301 Mason. John M.. 405 McGeary. Michael G.. 101. 103
Madison, 556 Mason. Patrick L., 354. 356. 365 Mclnnis v. Ogilvie, 631
1 8 1 1
Index 733
McKay sewing machine, 26 Mills, Florence, 299 Moore, Mark, 696

McLuhan, Marshall. 34, 45 Mishman. E. J., 177 Moore, Terry, 330, 580, 617
McMillen, Daniel, 278, 287, 289, Mismatch problems, 523, 524 Moral-hazard problem, 668
291,300,301 Mitchell, J. Paul, 442, 446 Morlok, Edward K., 603, 613, 616
McPartland, J. M., 625, 650 Mitra. Arindam. 290, 300 Mortgage deduction, 385-388
McQueen, Steve, 1 Model of rural region, 1 7, 1 Moses, Leon N.. 269, 298, 301,
Median transport location. 55 Moffitt, Robert, 356, 366 616.617
Median- voter model, 481-488, Mohring, Herbert, 581. 582. 593. Moss. Mitchell. 300
525 613 Moving cost, 390, 391
Median-voter result, 483 Monetary weight, 49 MSA. 9, 1
Medicaid, 358 Monocentric model of land use, Multicentric city. See

Medieval cities, 90, 91 205-247 Suburbanization
Mehay, Stephen, 247. 581, 618. assumptions, 206 Multiplier effect, 144
693 commuting assumptions, 226, Multiplier process, 141-144
Meiszkowski, Peter, 365 227 Mumford, Lewis, 86, 103
Meldrum, Christina, 343, 349 empirical estimates of Mun, Se-il, 33, 44
Mendel, Richard A.. 677, 695 rent/density functions, Mundle. S. R.. 603, 612. 616
Mercantile cities, 91, 92 237- 239 Municipal bonds (tax-free), 536
Merewitz, L.. 594-596. 614 housing-price function, Municipal government. See Local
Merit good, 465 216-219 government
Merriman, Dwight, 331 income segregation, 230-237 Municipal income tax, 521, 522
Meshenberg, Michael J., 330 land use, 224-226 Municipal sales tax, 5 1 9-52
Metro (Washington), 597 land use (central business Municipality, 7
Metropolitan area (MA). 7-9 district), 211-213 Munnell. Alicia H., 76. 82
Metropolitan federation, 465 manufacturer bid-rent function, Murnane, Richard J., 627, 641,
Metropolitan statistical area 206-209 651,653
(MSA), 9, 1 office bid-rent function, Murray, Charles, 366
Mexican garment industry, 71, 72 209-211 Murray. Michael P., 416, 443
Meyer, John R., 245, 555, 565, pollution, 229, 230 Murray, Sheila E., 629, 636, 637,
574, 575, 579, 580, 582, 606, public goods/taxes, 229 652
607,613,614,618 relaxing the assumptions, Musgrave, Peggy B., 452, 477
Miami, 593, 605 226-230 Musgrave. Richard A., 452, 477,
Microeconomic tools. See Tools of rent gradient, 221-223 479
microeconomics residential bid-rent function, Mushkin, Selma. 581
Miernyk, William H., 176 219-223 Muth, Richard F, 230, 246, 404,
Mieszkowski, Peter. 82, 175, 247. residential density, 223, 224 444
265,287,288,301,330,443, residential land use, 215 Myers, S. L., 694
476,479,501,502,540,541, retailers, location of, 213-215
581,583,613,615 spatial arrangement, 224, 226
Migration effect, 144 tastes for housing, 227-229 Natural monopoly, 453-455
Miller, Gerald K., 556, 581 transportation cost, 224 Navarro, Peter, 436, 445
Mills, Edwin S., 136,204, Monopolistic competition, 108 Neal. Derek, 642, 653
237-239, 246, 247, 265, 266, Mood, A. M., 625, 650 Neenan, William B., 652
269, 272, 273, 297, 299, 322, Moody, Carlisle E., 679, 690 Negroponte, Nicolas, 34, 45
330,331,338,340,349, Moomaw, Ronald L., 44 Neighborhood dissatisfaction.
403^05, 444,502, 582 Moore, Christy, 335 411-413

734 liu/cx
Neighborhood effect. 372 Offices and suburbanization. Pelt/man, Sam, 572, 574, 582
Neighborhood transition 281-283 Pennsylvania Coal v. Mahon, 323
(mixed-race neighborhood). Ohls, James C. 332, 406, 420, Performance zoning, 311.312
396-399 426. 444 Perloff. H., 204, 246
Nelson, Arthur C. 315, 330 Olmstead, Alan, 103 Persky. Joseph J„ 350
Net external benefit, 465 Olsen, Edgar O.. 378, 403. 4 6, 1 Personal crime. 656
Netzer, Richard, 519, 541 434, 443, 445 Personal Responsibility and Work
Neutze, G. M.. 478 Olson. Mancur. 477 Opportunity Reconciliation
Nevada. 77 Onion dehydrators. 5 I Act, 358
New York City Open-ended matching grants. Peskin. R., 603, 616
abandonment (housing). 375. 527-529 Petersilia, Joan, 692. 696
376 Open-space zoning, 318. 319 Peterson, George E., 404. 477,
corporate service linkages, 291 Opportunity cost, 722 479.541,652
fiscal problems. 47 Opportunity-cost effect, 274 Peterson, Paul. 349. 617
growth (trading activity). 48 Ordway, 0..331 Peterson. Thomas C, 618
Manhattan dressmaking Ore processors, 5 Phelps, Edmund, 354, 365. 477
industry, 27 Oregon, 316 Philbrick. Allen. 136
private bus lines, 606 Orfield. Gary, 343. 349. 350 Phillips, Llad. 666, 676, 692-694,
public transit usage, 588 Omstein. Norman J.. 542 697
rent controls. 436. 438 Oron, Yitzhak, 404 Phillips. Patrick L.. 331
Noise pollution, 309 Orsagh, Thomas, 692 Phoenix. 292. 431
Nold, F. C, 694 Oshkosh, 430 Physical invasion, 323
Nonbasic sector, 140 Ottawa. 556. 593 Pickerell. Don H„ 601.614
Nonexcluadability, 458 Output effect. 143 Pierce, Lawrence C. 649
Nonrivalrous consumption, 457 Ovadia. A.. 693 Pilarsky, Milton, 616
Norfolk. 343 Ozanne, L.. 378. 405 Pirenne, Henri, 103
Normal good, 702 Pittsburgh. 200. 593
North, Douglass C, 177 Pivo, Gary. 281.299
Norton, R. D., 82. 83 Pack. H.. 502 Plea-bargaining process. 672. 673
Novel le. Thierry, 44, 136 Pack, Janet Rothenberg. 332. 502 Pleeter. Saul. 176
Nuisance zoning. 308-31 I Pagano. Anthony M,

616 Plessy v. Ferguson, 639
Nunn, Geoffrey, 247, 581. 618 Papke. Leslie. 346. 350 PMSA. 9
Pans. 273 Pogue, Thomas F., 693
Parker. Theodore. I Polenske, Karen. 176
O'Hara. D. J..43. 245 Parking tax. 556. 557 Police and crime. 669. 670
O'Hare, William P.. 348 Parliament, Thomas J., 477 Police power. 32
O'hUallachain, Breandan, 33. 44 Parody, Thomas E., 556, 581 Police services. 456. 457
O'Regan, Katharine M.. 344. 345, Parole system, 677. 678 Polinski. A. M.. 265
350 Parr. J. B„ 136 Polinski. Mitchell. 392. 406
Oakland. William H., 479, 541, Partial land tax. 199 Pollard. Kelvin ML, 348
652 Pate, T., 693 Pollution
Oaks, Wallace E., 204, 460, Patterson, T. 330 air. 569 571
463 465. 477. 478. 492. 502. Peaking problem. 601. 602 monocentric city, 229. 230
541.643. 652 Pechman, Joseph A.. 512. 540 noise. 309
Odle, Marjorie, 1 77 Peleponnesian War, 88 nuisance zoning, 309 312
Off-street parking. 312 Peltz, Michael. 175 Pollution tax. 151. 152.570
1
Index 735
Ponting. Kenneth G., 43 Principle of marginal deterrence, economic growth, and, 149-153
Population density, 223 673-676 education spending equalities,
Population growth Principle of median location, 629-637
land-use controls, 303-308 55-59 educational segregation, 637,
retail suburbanization, 280 Prison system, 676-680 638
Porell, Frank W., 61,81 Private schools, 639-643 filtering, 382
Port cities, 57-59 Private trip cost, 549 land rent, and, 185-187
Porter, Douglas R., 331 Procurement cost, 48 suburbanization, 278
Positive externality, 26 Producer choice, 719-721 Public-safety externalities, 456,
Poterba, James, 350 Product cycle theory, 31 457
Poverty, 335-366 Property crime, 656 Public transit. See Mass transit
discrimination in labor Property taxes Publishing, 28

markets, 353-356 assessment process, 517-519 Pucher, John, 614
economic growth, and, changing the assumptions, Pure local public good, 457
351-353 509-512 Puryear, David, 477, 502
educational achievement, 343 general-equilibrium analysis,
facts/statistics, 335-338 507-509
female-headed households, incidence of the tax (who Quandt, Richard. 613
356, 357 pays), 512-514 Queralt, Magaly, 366
homeless, 360-363 mobile consumers, 51 Quigley, John M., 246, 344, 345,
municipal spending, and, 472 more than two municipalities, 349, 369, 373, 404, 405, 443,
public policy, 346, 347, 509,510 444
357-360 partial-equilibrium analysis,
residential segregation. 505-507
338-346 regressive assessment, 519 Rabushka, A., 479
358
social assistance, 357, residential, 503-514 Racial discrimination. See also
spatial mismatch, 343-346 tax in both cities, 510 Residential segregation
used housing, and, 378-380 tax on commercial/industrial defined. 393
welfare reform, 358-360 property, 514-516 housing, 392-399
working poor, 337. 338 518
tax rates (selected cities), labor market, 353-356
Poverty budget, 335 Tiebout model, and, 490-492, Racial steering, 395
Pozdena, Randall J., 552, 580 516,517 Radio industry, 31
Prejudice, 393 variable supply of dwellings, Rafael, Steven, 344, 350
Present value, 181, 182,385 510,511 Railroad, 48
Preservation zone, 317 Proudfoot, M. J., 137 Ramis, Timothy V., 315, 330
Preston, Anne E., 470, 471, Public housing, 413^-21 Rank-size rule, 127-129
479 housing consumption, and, Rapid depreciation, 385
Preston, Richard E., 137 413,414 Rasmussen, David W., 175, 407
Prevention cost. 662-664 living conditions, 418 Rational criminal model, 659-662
Price elasticity of demand, 704, market effects, 419, 420 Rauch, J., 32, 44
705 recipient welfare, and, Reasonable beneficial use, 323
Price elasticity of supply, 705 414-417 Reasonable-doubt rule, 673
Primary cities, 129-132 Public policy Recidivism rate, 677
Primary metropolitan statistical abandonment, 375-377 Red Line (Los Angeles), 597
area (PMSA), 9 central-city poverty, and, 346, Rees, Albert, 365
Principle of fiscal neutrality, 630 347, 357-360 Rees, John, 81,83
736 Index
Regional center, 59 white flight to private schools, Rothman. Stanley. 696

Regional system of cities. 105 639-643 Rottenberg. Simon. 691. 694-697
Regressive assessment, 519 Resnick. Mitchell. 129. 137 Rouse, Cecilia. 642. 653
Rehabilitation. 677 Resource allocation. 452. 453 Rubin. Paul. 697
Reischauer. Robert. 480 Resource-based cities. 55. 59 Rubinfeld. Daniel L.. 265. 349.
Religious city, 86. 87 Resource-oriented activities, 5 443.470.471.478.492,493,
Reno. 77 Resource-oriented firms. 49-5 501.502,532.541.542.652
Renovation. 279 Retail clusters. 38 Rubinstein. Jonathan. 697
Rent adjustments. 438 Retailers Rush hours. 552
Rent certificates. 421-423 location of (monocentric city ). Ryan. P.. 479
Rent control. 431—139 213-215
alternatives to. 439 nuisances. 312
bribes, 439 suburbanization, 279, 280 Sacerdote. B.. 680. 681. 690
deterioration/maintenance, 435, Retirement decision (housing). Sachs. D.. 650
436 374. 375 Sagalyn, Lynne B.. 279. 297
435
distributional effects. 434. Retribution. 676 Sagi. E.. 693
housing submarkets. 436, 437 Reuter. Peter. 697 Sales tax. 519-521
key money. 438. 439 Reynolds. Morgan. 670. 691, 693 Salomon. I.. 300
market effects. 432^434 Rhodes, W. M., 694 Salt Lake City, 43
New York City. 436. 438 Ricardo, David. 183.204 San Antonio Independent School
owner perks, 438 Richardson. Harry. 176. 597. 615 District v. Rodriguez, 63
supply response. 437. 438 Richmond. Jonathan E.. 593. 594. San Diego. 555
winners/losers. 433 598.599.613 San Francisco. 3 1
Rent gradient. 221-223 Ricketts. Erol R.. 349 Satterthwaite. Mark A.. 33. 44
Renter externality. 384 Rivasplata. Antero. 315, 330 Savas. E. S.. 457. 480. 615. 653
Renting vs. homeovvnership, Riverside Freeway (California). Sawhill. Isabel V.. 349. 364
383-389 555 Sa.xenian. AnnaLee. 3 1 . 43
Representative democracy, 485 Robak, Jennifer. 146. 175 Scale economies
Reschovsky, Andrew. 298. 477 Robins. Philip K.. 357. 366 cloth factor). 21-23
Residential density. 223, 224 Roddewig. Richard J.. 33 diversity. 123
Residential nuisances. 312 Rogers. Will. 181 intermediate inputs. 26-28

Residential propern tax. 503-514 Roman cities, 88. 89 local government. 462
Residential segregation, 33S-346. Romano. Richard E.. 645. 654 market area. 113. 1 14. 118
392-399 Rome. 88. 132 mass transit. 603
affirmative marketing. 398 Romer. T. 526. 541 transportation. 20. 53. 54
broker discrimination. 395. 396 Rosen. Harvey S.. 387. 407 urban area. 23. 24
consequences of, 341-346 Rosen. Kenneth T.. 129. 137 Schapiro. Morton Owen. 691
deli nod. 393 Rosen. Sherwin. 175 Scheinkman. Jose A.. 32. 34. 44
educational achievement, and. Rosenberg. Nathan. 103 Schell. Steven R.. 315. 330
343 Rosenbloom. Sandra. 603. 616 Schelling. Thomas C. 697
facts/statistics. 339 Rosenthal. H.. 526. 541 Schiller. Bradley. 343. 348. 354.
neighborhood transition. Rosenthal, Stuart S., 340. 344. 349 35S. 364. 650
$96-399 Rossi. Peter H., 445 Schlesinger. Steven R.. 695
reasons for, 340. 341 Rosthal, R., 649 Schmandt. Henry J.. 477
schools. 637-643 Kothenberg. Jerome. 177. 204. Schmenner, Roger W., 81, 82,
spatial mismatch. 343-346 265. 298, 350.445.579.614 176
1 1
Index 137
Schnidman. F.. 331 Skunk deodorizing, 51 Statistics. See Facts/statistics
Schools. See Education Small cars, 574, 575 Steam engine, 60

Schwab, Robert M., 145, 175, Small stores, 118, 119 Stegman, Michael A., 407
204, 629, 636, 637, 652 Small, Kenneth A., 277, 284-286. Stepan, Jan, 695
Schwartz, Seymour I., 331 290, 298, 546, 552, 557, 558, Stephens, John D., 43
Seattle, 59 569-572, 580-583, 588, 593, Stern, N., 692
Section 8 rent certificates, 595, 596, 601, 603, 606-608, Sternlieb, George, 82, 299, 300,
421-423 613,614,617 404
Seeley, Robert D., 692 Smeeding, Timothy M., 416, 444 Stevens, Barbara J., 457, 480
Segal. David, 44 Smerk, George M., 617 Stevens, Benjamin, 146, 175
Segregation. See Residential Smith, Adam, 17 Stiefel, Leanna, 651
segregation Smith, Barton A., 287. 288, 301, Stigler, George J., 477, 695
Semiconductor industry, 70 405, 408 Stone, Joe A., 146, 147, 167-169,
Senia. Al, 176 Smith, Donald E, 71,82 175-177,651
Serrano v. Priest, 630, 631 Smith, James P., 365 Storper, Michael, 81
Sesnowitz, Michael, 691 Smith, Lawrence B., 433, 438, 445 Straszheim, Mahlon R., 82, 175,
Sewing machines, 25, 26 Smolensky, Eugene, 103, 444 246. 247, 265, 330, 365, 404,
Shared-ride taxis, 604 SMSAs, 9 406, 408, 443, 445, 476, 477,
Sherman, Lawrence, 669, 693 Social trip cost, 549, 564 479,501,502,552.581,583,
Shilling, James, 388, 407 Solomon, Arthur, 99, 103, 332, 613-615,618.653
Shinnar, Reuel, 696 404 Streetcars, 256-259. 275
Shinnar, Shlomo, 696 Solow, Robert M., 266 Struyk, Raymond J., 298, 378,
Shleifer, Andrei, 32, 34, 44 Sonstelie, Jon, 246 387, 405, 407, 442, 443, 445
Shopping, 492 Sorting, 492, 493 Subcenters, 284-292
Shopping externalities, 215 Southern Burlington Country Subscription commuter services,
Shopping externality, 36 NAACP v. Mount Laurel, 322 605, 606
Shopping paths, 123 Spain, Daphne, 298, 353, 365 Subsidies
Shortage, 706 Spatial effluent fees, 309-31 construction, 380-382
Shoup, Carl, 697 Spatial mismatch theory, 343-346 housing, 421-428
Shoup, Donald C, 557. 581 Split tax, 199 mass transit, 599-602
Siegan, Bernard, 320, 330 Sprull, N. L., 684, 696 new firms, 64, 65, 149, 150
Siemon. Charles, 331 SSI, 357, 358 transit, 559-561,571
Silcock, D. T., 607, 615 St. Louis, 43 Substantive due process, 321
Silicon Valley, 31,70 Stabilization policy, 452, 453 Substitution effect
Silverman, L. P., 684, 696 Stack, Linda, 265 increases in quantity, 469
Simmons. J. W., 137 Stafford, Howard, 81 labor-demand curve, 143
Simon, Carl P., 697 Stahl, Konrad, 45 welfare spending, 532
Simpson, Anthony U., 616 Stanback, Thomas M., 44, 136, Suburban airports, 272
Singapore, 554 292,298,301 Suburban exploitation hypothesis,
Single-story plants, 272 Standard consolidated statistical 463
Single tax, 198, 199 areas (SCSAs), 9 Suburban subcenters, 284-292
Sirmans, C, 407 Standard metropolitan statistical Suburbanization, 267-301
Site development, 65, 150 (SMSAs), 9
areas airports, 272
Sjoberg, Gideon, 103 Standard State Zoning Enabling automobiles, 272, 275, 280
Sjoquist, David L., 345, 349, 694 Act, 321 central-city black households,
Skinner, Robert E., 617 Statistical discrimination, 354 346, 347
738 Index
central-city problems. 276. 277 gasoline. 556. 570. 571 Tideman. Nicholas. 478. 479
Chicago, 287-289 housing, and, 385-389 Tidewater, Virginia. 603
commuting 274—276
cost. income, 521. 522 Tiebout. Charles ML. 155. 176.
edge cities, 289. 290 land. 198-200 177.488.502
employment, 277, 278 local. See Local taxes Tiebout model. 488^494
facts/statistics. 267-269 location decisions, and, 64. 75. assumptions, 489
Houston. 2S7 76 capitalization, 492
intercity truck, 27 parking. 556. 557 criticisms of, 493, 494
intracity truck, 269-27 partial land, 199 education, and. 628. 629
Los Angeles, 284-287 pollution. 151. 152.570 efficiency effects. 489. 490
manufacturing. 269-272 property. See Property taxes property taxes, 490-492, 5 1 6.
offices,281-283 single. 198, 199 517

population. 272-279 split, 199 pros/cons, 494
public policy. 278 tax-free municipal bonds, 536 racial segregation, and. 643
real income. 274 two-rate. 199 shopping, 492
retailers. 279. 280 Tax abatement. 64, 149 sorting. 492. 493
subcenters. 284-292 Tax expenditures, 534—536 Tipping point. 643
urban villages. 292, 293 Tax exporting, 516 Tobin, Gary A.. 407. 408
Sugarman. Stephen D., 651, 654 Tax-free municipal bonds, 536 Toby, Jackson, 697
Sullivan, Arthur M., 261,265, Taylor. Brian D.. 603. 616 Toffler. Alvin. 34. 45
300. 444. 568. 580 TDR policy, 317, 318 Toll booths. 553. 554
Summers, Anita A., 472, 479, 627. Teachers, 626. 627 Tolley. George S.. 176. 408. 478
628. 65 Teal. Roger F.. 615 Tomlinson. Peter. 433. 438. 445
Summers. Gene. 177 Tegeler. Philip D.. 331 Tools of microeconomics. 699-722
Summers. Lawrence. 389. 407 Telecommunications, 34—36 changes in demand or supply.
Sundstrom, Sven, 265 Teleconferencing, 283 707. 708
Supplemental security income Telephone, 35 consumer choice, 708-7 1
(SSI). 357. 358 Temporary Aid to Needy Families elasticities of demand and
Supply and demand. 701-708 (TANF). 358 supply. 704. 705
Supply elasticity. 705 Tenant bribery. 439 marginal decision making. 72 1
Supply experiments, 427 Terlecky, Nester, 246. 405 722

Surplus. 707 Terminal costs, 53 market demand curve. 701-703
Sveikauskas, L.. 44, 45 Terminology, 1 market equilibrium. 705, 706
Swallow, B. M., 693 Terrell. Henry S.. 350 market supply curve. 703, 704
Swart/. Alex. 291.301 Thaler. Richard. 658. 691. 693 producer choice. 7 9-72 1
Sweeney. James L.. 406 Theory of consumer choice, shortage/surplus. 706. 707
S\ stems development charges. 316 708-7 1 supply and demand. 701-708
Theory of producer choice, worker's choice. 7 7-7 1 1
719-721 Toronto
Taking clause. 323 Theory of risk compensation. 572 express toll road, 554
TANF. 358 Theory of worker's choice. federated government, 465
Tauchen. Helen. 43, 245. 669. 694 717-719 subsidies for new construction,
Tax llusse. Jacques-Francois. 235. 246 438
carbon. 571 Thorsnes. Paul. 580. 617 Toynbee, Arnold. 45
congestion, 550-556 Three-strikes laws. 679. 680 Trading cities. 47. 48
development. 315 Thurow. Lester. 697 Trading firms. 47
1 1
Index 739
Traffic accidents, 571 public policy, 149-153 Vacancy decontrol, 438

Trajtenber, M., 32, 44 urban labor market, 140-149 Varma, P. K., 603, 616
Transactions table, 157 Urban economics, 2, 3 Vaughn, Bill, 267
Transfer-oriented firm, 48-55 Urban fringe, 7 Vehicle-identification system
Transferable development rights Urban giants, 129-132 (VIS), 554

(TDRs), 317, 318 Urban growth boundary, 303 Vehicle Safety Act, 572
Transit subsidies, 559-561, 571 Urban place, 9 Venables. Anthony J., 45
Transport costs vs. local-input Urban population, 9 Veno, Makiko, 443
costs, 66-69 Urban renewal, 429, 430 Verhoef. Erik T. 580
Transportation. See Urban Urban service boundary, 303-306 Vernon, Raymond, 27, 3 1 , 298
transportation Urban transportation, 543-618 Vickrey, William 580S.,
Transportation firms, 47 air pollution, 569-571 Victim cost, 662-664

Transshipment points, 57 auto safety, 571-575 Village of Arlington Heights v.
Treyz, George, 146, 175 capacity expansion, 557-559 Metropolitan Housing

Tron, E., 649 congestion, 546-550 Corporation, 322
Trucks congestion tax, 550-556, Viscusi, W. Kip, 694
intercity, 27 565-569 Viton, Philip A., 603, 613, 616
intracity, 269-271 equilibrium vs. optimum traffic von Furstenberg, George M., 443
Tuccillo, John A., 387, 407, 442 volume, 549, 550 Von Sweringen brothers, 275
Tullock. Gordon, 694 flow of traffic, 559 Von Thunen model, 187
Turner, Margery A., 443 gasoline tax, 556, 570, 571 Votey, Harold, 666, 676, 692, 693,
Twinkie effect, 627 HOT lanes, 555 697
Two-rate tax, 199 HOV lanes, 554, 593, 594 Voting
land-use patterns, 565-569, foot (Tiebout model), 488-494
607-609 median-voter model, 481-488
UDAG, 430 latent demand, 558, 559
Union power, 468 long-run analysis of highway
Unions, 76, 77 pricing/investment, Wachs, Martin, 603, 616
Unzoned cities, 320 561-565 Wage discrimination, 353
Urban area, 7 mass transit. See Mass tran.sit WAGES. 359
Urban Development Action optimum speed, 572-574 Wages, 76
Grants (UDAG), 430 parking tax, 556, 557 Walder, Jay H., 616
Urban economic growth, 139-177 peak vs. off-peak travel, 552, Waldfogel, Jane, 355, 365
composition of changes in 559 Walters, A. A., 580, 603, 613
employment, 167-169 safety regulations, 572 Walzer, Norman, 478
economic base study, 153-157, small cars, 574, 575 Warner. Charles Dudley, 503
160-162 toll booths, 553, 554 Warner, Samuel B., Jr., 617
effects of employment growth, traffic accidents, 57 Warth v. Selden, 322
162-167 transit subsidies, 559-561, 571 Washington, D. C, 556
environmental quality, 150-153 vehicle identification system, Wassmer, Robert W., 76, 82, 493,
forecasting techniques, 554 502
153-162 widening the highway, Wasylenko, Michael J., 237, 247,
input-output study, 157-162 557-559 277. 298, 299
location-quotient approach, Urban villages, 292, 293 Waterwheel. 59
154-156 Urbanization economies, 32, 61 Watson, Peter L., 580
poverty, and, 351-353 Urbanized area, 7 Weapons, 53
740 Index
Webber. Melvin W.. 595, 608. White, Larry J., 407 Work and Gain Economic Self-
614,617 White, Lawrence J., 582 Sufficiency (WAGES). 359
Weicher, John. 443, 445 White, Michelle J., 330, 332. 375. Worker's choice, 7 7-7 1 1
Weight-gaining activity, 52 404, 407 Working poor. 337. 338

Weight-losing activities, 51 Widening the highway. 557-559 Wright, Frank Lloyd, 205
Weight-losing firms. 51 Wiley. David E.. 651 Wright, James D.. 697
Wei mer. George, 176 Williams. Stephen R.. 581
Weinberg, Daniel H., 366 Williamson. Harold. 269. 298
Ybarra v. Town of Los Alton Hills,
Weinblatt, J.. 691.693 Willson, Richard W.. 581
322
Weinrield, F. D.. 625. 650 Wilson, James Q„ 268. 615, 669.
Yinger. John. 340. 349. 393-395.
Weisberg. Richard C, 332 671.677.684.690-696
407.408.471-473.479.637.
Weiss. Leonard, 137 Wilson, Julie Boatwright, 479
652
Weiss. Marc A.. 175. 446 Wilson. Peter. 479
York. R. L.. 625. 650
Welch, Finis. 365 Wilson. William Julius, 349
Welfare reform, 358-360, 531, Witte. Ann Dryden, 43, 245. 366.
532 669, 692. 694. 697 Zarkin. Gary A.. 692
Wells, H. G„ 621 Wohl. Martin. 245. 299. 581. 613. Zenou. Yves, 235. 246
Wheaton. William C. 82, 233. 616 Zodrow, George. 502
247. 265 Wolfe. Barbara. 627. 628. 651 Zoning. See Land-use controls and
Whinston. Andrew B., 445 Wolpin. V.. 694 zoning
White, James R., 332 Woolbright, Louie Albert. 408 Zoning plan. 308
Hh
ISBN D-ESb-Et331-D
90000
McGraw-Hill Higher Education g?

A Division of The McGmw-Hitt Companies
9 780256"263312
www. nihhe.com

Urban Economics Arthur OSullivan

Uploaded by

Copyright:

Available Formats

Urban Economics Arthur OSullivan

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Urban Economics Arthur OSullivan

Uploaded by

Copyright:

Available Formats

J MjHWjjyj'

Digitized by the Internet Archive

Oregon State University

This book is printed on acid-free paper:

domestic 12 3 4567890 DOC DOC 9098765432109

Editor-in-chief: Michael \Y. Junior

Library of Congress Cataloging-in-Publication Data

International Edition ISBN 0-07-116974-1

Arthur O'Sullivan is a professor of Economics at Oregon

^^ 4 ^^^^ then spent eleven years at the University of California. Davis.

BaJI nomic and urban/regional economics. Professor O" Sullivan's

What's New in the Fourth Edition?

• Part III: Poverty and Housing

Part I: Market Forces in the Development of Cities

• Includes two new case studies of firms' location decisions: Japanese

developing countries, including trade, trade restrictions, infrastructure

Part II: Land Rent and Urban Land-Use Patterns

• Explores the effects of Pittsburgh's graded property-tax policy (under which

• Includes new material on the development of employment subcenters

Part III: Poverty and Housing

schools, differences in educational achievement, and the spatial mismatch

Part IV: Local Government

• Includes the latest data on government taxes and expenditures

Part V: Urban Transportation

and heavy rail) (Chapter 20)

Part VI: Education and Crime

Randall Bartlett Vijay Mathur

Charles Berry Steven Raphael

Market Forces in the Development of Cities 15

2 Why Do Cities Exist? 17

Land Rent and Urban Land-Use Patterns 179

Poverty and Housing 333

Local Government 447

Urban Transportation 543

What Is Urban Economics? 2

Market Forces in the Development of Cities 15

2 Why Do Cities Exist? 17

A Model of a Rural Region 17

Agglomerative Economies Production 26

3 Where Do Cities Develop? 47

Commercial Firms and Trading Cities 47

4 The History of Western Urbanization 85

The First Cities 85

5 How Many Cities? 105

The Analysis of Market Areas 106

Central Place Theory 119

6 Urban Economic Growth 139

The Urban Labor Market and Economic Growth 140

Land Rent and Urban Land-Use Patterns 179

7 Introduction to Land Rent and Land Use 181

Land Rent and Fertility 183

Land Rent and Accessibility 187

Market Interactions 196

8 Land Use in the Monocentric City 205

Commercial and Industrial Land Use 206

Residential Land Use 215

9 General-Equilibrium Land Use 249

General-Equilibrium Conditions 249

10 Suburbanization and Modern Cities 267

Suburbanization Facts 267