Microeconomic

Theory

Andreu Mas-Colell Michael D. Wlt.inston

and

Jerry R. Green

New York Oxford OXFORD UNIVERSITY PREs;;S 1995

 330.!
'H ~1
OXFORD UNIVERSITY PRESS
Oxford New York
Athens Auckland Bangkok Bombay A Esther, por todo A. M.-c.
Calcutta Cape Town Dares Salaam Deli
Aorence Hong Kong Istanbul Karachi
Kuala Lumpur t.ladras Madrid Melbourne
Mexico City Nairobi Paris Singapore
Taipei Tokyo Toronto
and associated cofllpanies in
Berlin lbadan

Copyright© 1995 by Oxford University Press, Inc. To Bonnie, for keeping me smiling throughout;
Published by Ox ford Univer!.ity Press. Inc., to Noah, for his sweetness and joy at the book's completion;
198 M<.~dison A\'<nue, New York, New York 10016
and to Nan, for helping me get started M. D. w.
Oxford h. a regi~tered trademurk of Oxford Univcnity Prc:o.s

All rights re!ooervcd. No part of this publicouion may he reproduced.

stored in 01 rctrie,·al system, or tnmsmined. by any form or by any mean,,
C'lectroniL". mech:tnical. photocopying. recording, ur otherwi~c.
without the prior permissiun of Oxford Uni\'er~ity Pr~\s.

Library of Congr=s Cataloging·in-Publication Data

Mas-Colell, Andreu. To Pamela, for her kindness, character, strength, and spirit J. R. G.
Micr~conomie lh tory I Andreu Mas-Colell, Michael D. \Yhinston, and Jerry R. Green.
p. em.
lnclud~.s bibliographical references and index.
ISDN 0-t9-507340-l (clotlJ)
ISBN 0-t9-510268-t (paper)
I. Microeconomics. I. \Vhinston, Michael Dennis. II. Green, Jerry R. lli. Title
fm t72.M6247 1~s
338.5-dc20 95-18128

98765432t
Printed in the United States of America
on acid.free paper
Contents

Preface XIII

PART ONE: INDIVIDUAL DECISION MAKING 3

Chapter I. Preference and Choice 5
l.A Introduction 5
I. B Preference Relations 6
I.C Choice Rules 9
I.D The Relationship between Preference Relations and Choice Rules II
Exercises 15

Chapter 2. Consumer Choice 17

2.A Introduction 17
2. B Commodities 17
2.C The Consumption Set 18
2.D Competitive Budgets 20
2.E Demand Functions and Comparative Statics 23
2.F The Weak Axiom of Revealed Preference and the Law of Demand 28
Exercises 36

Chapter 3. Classical Demand Theory 40

3.A Introduction 40
3. B Preference Relations: Basic Properties 41
3.C Preference and Utility 46
3.0 The Utility Maximization Problem 50
3.E The Expenditure Minimization Problem 57
3.F Duality: A Mathematical Introduction 63
3.G Relationships between Demand, Indirect Utility, and Expenditure Functions 67
3.H Integrability 75
3.1 Welfare Evaluation of Economic Changes 80
3.1 The Strong Axiom of Revealed Preference 91
Appendix A: Continuity and Differentiability Properties of Walrasian Demand 92
Exercises 96

Chapter 4. Aggregate Demand 105

4.A Introduction 105
4.B Aggregate Demand and Aggregate Wealth 106

vii
viii CONTENTS C 0 NT EN T S ix

4.C Agpegate Demand and the Weak Axiom 109 Chapter 9. Dynamic Games 267
4.D Aggiegate Demand and the Existence of a Representative Consumer 116
9.A Introduction 267
Appendix A: Regularizing Effects of Aggregation 122 9.B Sequential Rationality, Backward Induction, and Subgame Perfection 268
Exercises 123 9.C Beliefs and Sequential Rationality 282
9.D Reasonable Beliefs and Forward Induction 292
Chapter 5. Production 127
Appendix A: Finite and Infinite Horizon Bilateral Bargaining 296
5.A Introduction 127 Appendix B: Extensive Form Trembling-Hand Perfect Nash Equilibrium 299
5. B Production Sets 128 Exercises 30 I
5.C Profit Maximization and Cost Minimization 135
5.0 The Geometry of Cost and Supply in the Single-Output Case 143
5.E Aggregation 147
5.F Efficient Production 149 PART THREE: MARKET EQUILIBRIUM AND MARKET FAILURE 307
5.G Rem<rks on the Objectives of the Firm 152
Chapter 10. Competitive Markets 311
Appendix A: The Linear Activity Model !54
Exercises 160 IO.A Introduction 311
IO.B Pareto Optimality and Competitive Equilibria 312
Chapter 6. Cfloice Under Uncertainty 167 IO.C Partial Equilibrium Competitive Analysis 316
IO.D The Fundamental Welfare Theorems in a Partial Equilibrium Context 325
6.A Introduction 167 IO.E Welfare Analysis in the Partial Equilibrium Model 328
6.B Expected Utility Theory 168 IO.F Free-Entry and Long-Run Competitive Equilibria 334
6.C Monc:y Lotteries and Risk Aversion 183 lO.G Concluding Remarks on Partial Equilibrium Analysis 341
6.0 Com)larison of Payoff Distributions in Terms of Return and Risk 194
Exercises 344
6.E State-dependent Utility 199
6.F Subjective Probability Theory 205
Exercises 208 Chapter II. Externalities and Public Goods 350
II.A Introduction 350
PART TWO GAME THEORY 217 II.B A Simple Bilateral Externality 351
Chapter 7. Basic Elements of Noncooperative Games 219 II.C Public Goods 359
11.0 Multilateral Externalities 364
7.A Introoduction 219 I I.E Private Information and Second-Best Solutions 368
7.B What Is a Game? 219
Appendix A: Nonconvexities and the Theory of Externalities 374
7.C The Extensive Form Representation of a Game 221
Exercises 378
7.0 Strategies and the Normal Form Representation of a Game 228
7.E Rand<1mized Choices 231
Exercises 233 Chapter 12. Market Power 383

Chapter 8. Simultaneous-Move Games 235 12.A Introduction 383

12.B Monopoly Pricing 384
S.A Introduction 235 12.C Static Models of Oligopoly 387
8.B Oomimnt and Dominated Strategies 236 12.0 Repeated Interaction 400
8.C Ratiomlizable Strategies 242 12.E Entry 405
8.0 Nash Equilibrium 246 12.F The Competitive Limit 411
8.E Games of Incomplete Information: Bayesian Nash Equilibrium 253 12.G Strategic Precommitmcnts to Affect Future Competition 414
8.F The Possibility of Mistakes: Trembling-Hand Perfection 258 Appendix A: Infinitely Repeated Games and the Folk Theorem 417
Appendix A: Existence of Nash Equilibrium 260 Appendix B: Strategic Entry Deterrence and Accommodation 423
Exercises 262 Exercises 428
I NT 5
C 0 NT EN T S Xi

~ apter 13. Adverse Selection, Signaling, and Screening 436

17.D Local Uniqueness and the Index Theorem 589
13.A Introduction 436 17.E Anything Goes: The Sonnenschein-Mantel-Debreu Theorem 598
13.B Informational Asymmetries and Adverse Selection 437 17.F Uniqueness of Equilibria 606
13.C Signaling 450 17 .G Comparative Statics Analysis 616
13.D Screening 460 17.H Tiitonnement Stability 620
Appendix A: Reasonable-Beliefs Refinements in Signaling Games 467 17.1 Large Economies and Nonconvexities 627
Exercises 473 Appendix A: Characterizing Equilibrium through Welfare Equations 630
Appendix 8: A General Approach to the Existence of Walrasian Equilibrium 632
Exercises 641
1
1pter 14. The Principal-Agent Problem 477
14.A Introduction 477
'4.B Hidden Actions (Moral Hazard) 478 Chapter 18. Some Foundations for Competitive Equilibria 652
'4.C Hidden Information (and Monopolistic Screening) 488
14.D Hidden Actions and Hidden Information: Hybrid Models 501 18.A Introduction 652
!8.B Core and Equilibria 652
Appendix A: Multiple Effort Levels in the Hidden Action Model 502
18.C Noncooperative Foundations of Walrasian Equilibria 660
-\ppcndix 8: A Formal Solution of the Principal-Agent Problem with Hidden Information 504
18.D The Limits to Redistribution 665
•:xercises 507
18.E Equilibrium and the Marginal Productivity Principle 670
Appendix A: Cooperative Game Theory 673
'RT FOUR: GENERAL EQUILIBRIUM 511 Exercises 684

"apter 15. General Equilibrium Theory: Some Examples 515

15.A Introduction 515 Chapter 19. General Equilibrium Under Uncertainty 687
15.B Pure Exchange: The Edgeworth Box 515
19.A Introduction 687
5.C The One-Consumer, One-Producer Economy 525
19.8 A Market Economy with Contingent Commodities: Description 688
: 5.D The 2 x 2 Production Model 529
19.C Arrow-Debreu Equilibrium 691
: 5.E General Versus Partial Equilibrium Theory 538
!9.D Sequential Trade 694
Exercises 540
19.E Asset Markets 699
!9.F Incomplete Markets 709
:.. ,pter 16. Equilibrium and Its Basic Welfare Properties 545 19.G Firm Behavior in General Equilibrium Models Under Uncertainty 713
19.H Imperfect Information 716
16.A Introduction 545
Exercises 725
16. B The Basic Model and Definitions 546
! 6.C The First Fundamental Theorem of Welfare Economics 549
16.D The Second Fundamental Theorem of Welfare Economics 551
16.E Pareto Optimality and Social Welfare Optima 558 Chapter 20. Equilibrium and Time 732
!6.F First-Order Conditions for Pareto Optimality 561 20.A Introduction 732
16.G Some Applications 566 20. B Intertemporal Utility 733
\ppendix A: Technical Properties of the Set of Feasible Allocations 573 20.C Intertemporal Production and Efficiency 736
!2xercises 575 20.D Equilibrium: The One-Consumer Case 743
20. E Stationary Paths, Interest Rates, and Golden Rules 754
20.F Dynamics 759
l1apter 17. The Positive Theory of Equilibrium 578
20.G Equilibrium: Several Consumers 765
17.A Introduction 578 20.H Overlapping Generations 769
17.B Equilibrium: Definitions and Basic Equations 579 20.I Remarks on Nonequilibrium Dynamics: Tatonnement and Learning 778
17.C Existence of Walrasian Equilibrium 584
Exercises 782
xii C 0 NT EN T S

PART FIVE: WELFARE ECONOMICS AND INCENTIVES 787

Chapter 21. Social Choice Theory 789
2l.A In traduction 789 Preface
21.B A Special Case: Social Preferences over Two Alternatives 790
21.C The General Case: Arrow's Impossibility Theorem 792
2l.D Some Possibility Results: Restricted Domains 799
21.E Social Choice Functions 807
Exercises 812 Microeconomic Tlteory is intended to serve as the text for a first-year graduate
course in microeconomic theory. The original sources for much of the book's material
Chapter 22. Elements of Welfare Economics and Axiomatic Bargaining 817
arc the lecture notes that we have provided over the years to students in the first-year
22.A Introduction 817 microeconomic theory course at Harvard. Starting from these notes, we have tried
22.B Utility Possibility Sets 818 to produce a text that covers in an accessible yet rigorous way the full range of topics
22.C Social Welfare Functions and Social Optima 825 taught in a typical first-year course.
22.D Invariance Properties of Social Welfare Functions 831 The nonlexicographic ordering of our names deserves some explanation. The
22.E Tile Axiomatic Bargaining Approach 838 project was first planned and begun by the three of us in the spring of 1990.
22.F C<>alitional Bargaining: The Shapley Value 846 However, in February 1992, after early versions of most of the book's chapters had
Exercises 850 been drafted, Jerry Green was selected to serve as Provost of Harvard University,
a position that forced him to suspend his involvement in the project. From this
Chapter 23. Incentives and Mechanism Design 857
point in time until the manuscript's completion in June 1994, Andreu Mas-Colell
23.A Introduction 857 and Michael Whinston assumed full responsibility for the project. With the conclusion
23.B Tile Mechanism Design Problem 858 of Jerry Green's service as Provost, the original three-person team was reunited for
23.C Dominant Strategy Implementation 869 the review of galley and page proofs during the winter of 1994/1995.
23.D Bayesian Implementation 883
23.E Participation Constraints 891 Tile Organization of the Book
23.F Optimal Bayesian Mechanisms 897
Microeconomic theory as a discipline begins by considering the behavior of individual
Appendix A: Implementation and Multiple Equilibria 910
agents and builds from this foundation to a theory of aggregate economic outcomes.
Appendix B: Implementation in Environments with Complete Information 912
Exercises 918
Microeconomic Tlteory (the book) follows exactly this outline. It is divided into five
parts. Part I covers individual decision making. It opens with a general treatment of
individual choice and proceeds to develop the classical theories of consumer and
MATHEMATICAL APPENDIX 926
producer behavior. It also provides an introduction to the theory of individual choice
M.A Matrix Notation for Derivatives 926 under uncertainty. Part II covers game theory, the extension of the theory of
M.B Homogeneous Functions and Euler's Formula 928 individual decision making to situations in which several decision makers interact.
M.C Concave and Quasiconcave Functions 930 Part Ill initiates the investigation of market equilibria. It begins with an introduction
M.D Matrices: Negative (Semi)Definiteness and Other Properties 935 to competitive equilibrium and the fundamental theorems of welfare economics in
M.E The Implicit Function Theorem 940 the context of the Marshallian partial equilibrium model. It then explores the
M.F Continuous Functions and Compact Sets 943 possibilities for market failures in the presence of externalities, market power, and
M.G Convex Sets and Separating Hyperplanes 946 asymmetric information. Part IV substantially extends our previous study of
M.H Correspondences 949 competitive markets to the general equilibrium context. The positive and normative
M.l Fixed Point Theorems 952 aspects of the theory are examined in detail, as are extensions of the theory to
M.J Unconstrained Maximization 954 equilibrium under uncertainty and over time. Part V studies welfare economics. It
M.K Constrained Maximization 956 discusses the possibilities for aggregation of individual preferences into social
M.L The Envelope Theorem 964 preferences both with and without interpersonal utility comparisons, as well as
M.M Linear Programming 966 the implementation of social choices in the presence of incomplete information
M.N Dynamic Programming 969 about agents' preferences. A Mathematical Appendix provides an introduction to
Index 971 most of the more advanced mathematics used in the book (e.g., concave/convex

xiii
PREFACE PREFACE XV

functions, constrained optimization techniques, fixed point theorems, etc.) as well as breaking it up into segments that are discussed right before they are used (e.g.,
references for further reading. Chapter 7, Chapter 8, and Sections 9.A-B before studying oligopoly, Sections 9.C-D
before covering signaling). Some other possibilities include teaching the aggregation
Tile Style of tile Book of preferences (Chapter 21) immediately after individual decision making and
covering the principal-agent problem (Chapter 14), adverse selection, signaling, and
In choosing the content of Microeconomic Theory we have tried to err on the side screening (Chapter 13), and mechanism design (Chapter 23) together in a section of
of inclusion. Our aim has been to assure coverage of most topics that instructors in the course focusing on information economics.
a first-year graduate microeconomic theory course might want to teach. An inevitable In addition, even within each part, the sequence of topics can often be altered
consequence of this choice is that the book covers more topics than any single easily. For example, it has been common in many programs to teach the preference-
first-year course can discuss adequately. (We certainly have never taught all of it in based theory of consumer demand before teaching the revealed preference, or
any one year.) Our hope is that the range of topics presented will allow instructors "choice-based," theory. Although we think there are good reasons to reverse this
the freedom to emphasize those they find most important. sequence as we have done in Part 1, 2 we have made sure that the material on demand
We have sought a style of presentation that is accessible, yet also rigorous. can be covered in this more traditional way as well. 3
Wherever possible we give precise definitions and formal proofs of propositions. At
the same time, we accompany this analysis with extensive verbal discussion as well
as with numerous examples to illustrate key concepts. Where we have considered a
On Mathematical Notation
proof or topic either too difficult or too peripheral we have put it into smaller type
to allow students to skip over it easily in a first reading. For the most part, our usc of mathematical notation is standard. Perhaps the most
Each chapter offers many exercises. ranging from easy to hard [graded from A important mathematical rule to keep straight regards matrix notation. Put simply,
(easiest) to C (hardest)] to help students master the material. Some of these exercises vectors are always treated mathematically as column vectors, even though they are
also appear within the text of the chapters so that students can check their often displayed within the written text as rows to conserve space. The transpose of
understanding along the way (almost all of these are level A exercises). the (column) vector x is denoted by xT. When taking the inner product of two
The mathematical prerequisites for use of the book are a basic knowledge of (column) vectors x and y, we write x · y; it has the same meaning as x Ty. This and
calculus, some familiarity with linear algebra (although the use of vectors and other aspects of matrix notation are reviewed in greater detail in Section M.A of the
matrices is introduced gradually in Part 1), and a grasp of the elementary aspects of Mathematical Appendix.
probability. Students also will find helpful some familiarity with microeconomics at To help highlight definitions and propositions we have chosen to display them
the level of an intermediate undergraduate course. in a different typeface than is used elsewhere in the text. One perhaps unfortunate
consequence of this choice is that mathematical symbols sometimes appear slightly
differently there than in the rest of the text. With this warning, we hope that no
Teaching the Book confusion will result.
The material in this book may be taught in many different sequences. Typically we Summation symbols <L:l
are displayed in various ways throughout the text.
have taught Parts I-III in the Fall semester and Parts IV and V in the Spring Sometimes they arc written as
N
(omitting some topics in each case). A very natural alternative to this sequence (one
used in a number of departments that we know of) might instead teach Parts I and IV L:
r~= 1

in the Fall, and Parts II, Ill, and V in the Spring.' The advantage of this alternative (usually only in displayed equations), but often to conserve space they appear as
sequence is that the study of general equilibrium analysis more closely follows the "L~ = 1 , and in the many cases in which no confusion exists about the upper and lower
study of individual behavior in competitive markets that is developed in Part I. The limit of the index in the summation, we typically write just L. •. A similar point
disadvantage, and the reason we have not used this sequence in our own course, is applies to the product symbol n.
that this makes for a more abstract first semester; our students have seemed happy
to have the change of pace offered by game theory, oligopoly, and asymmetric
2. In particular. it is much easier to introduce and derive many properties of demand in the
information after studying Part I.
choice-based theory than it is using the preference-based approach; and the choice-based theory
The chapters have been written to be relatively self-contained. As a result, they gives you almost all the properties of demand that follow from assuming the existence of rational
can be shifted easily among the parts to accommodate many other course sequences. preferences.
For example, we have often opted to teach game theory on an "as needed" basis, 3. To do this, one introduces the basics of the consumer's problem using Sections 2.A-D and
3.A··D, discusses the properties of uncompensated and compensated demand functions, the indirect
utility function, and the expenditure function using Sections 3.0-1 and 2.E, and then studies revealed
I. Obviously. some adjustment needs to be made by programs that operate on a quarter preference theory using Sections 2.F and 3.J (and Chapter I for a more general overview of the
system. two approaches).
XVi P A E FACE
PAEFACE XVii

Also described below are the meanings we attach to a few mathematical symbols
staff at Keyword Publishing Services did an absolutely superb job editing and
whose use is somewhat less uniform in the literature [in this list, x = (x 1, ... , xN)
producing the book on a very tight schedule. Their complete professionalism has
andy= (y 1 , ••. , YN) are (column) vectors, while X and Yare sets]:
been deeply appreciated.
Symbol Meaning The influence of many other individuals on the book, although more indirect, has
been no less important. Many of the exercises that appear in the book have been
x;::y x. ;:.: y. for all n = I, ... , N.
conceived over the years by others, both at Harvard and elsewhere. We have indicated
X»y x. > y. for all n = I, ... , N.
our source for an exercise whenever we were aware of it. Good exercises are an
XcY weak set inclusion (x e X implies x e Y).
enormously valuable resource. We thank the anonymous authors of many of the
X\Y The set {x:xeX but x¢ Y).
exercises that appear here.
E,[f(x, y)] The expected value of the function f( ·) over realizations of the
The work of numerous scholars has contributed to our knowledge of the topics
random variable x. (When the expectation is over all of the
discussed in this book. Of necessity we have been able to provide references in each
arguments of the function we simply write E[f(x, y)].)
chapter to only a limited number of sources. Many interesting and important contri-
butions have not been included. These usually can be found in the references of the
Ack11owledgments works we do list; indeed, most chapters include at least one reference to a general
survey of their topic.
Many people have contributed to the development of this book. Dilip Abreu, Doug
We have also had the good fortune to teach the first-year graduate microeconomic
Bernheim, David Card, Prajit Dutta, Steve Goldman, John Panzar, and David Pearce
theory course at Harvard in the years prior to writing this book with Ken Arrow,
all (bravely) test-taught a very early version of the manuscript during the 1991-92
Dale Jorgenson, Steve Marglin, Eric Maskin, and Mike Spence, from whom we
academic year. Their comments at that early stage were instrumental in the refinement
learned a great deal about microeconomics and its teaching.
of the book into its current style, and led to many other substantive improvements
We also thank the NSF and Sloan Foundation for their support of our research
in the text. Our colleagues (and in some cases former students) Luis Corch6n, Simon
over the years. In addition, the Center for Advanced Study in the Behavioral Sciences
Grant, Drew Fudenberg, Chiaki Hara, Sergiu Hart, Bengt Holmstrom, Eric Mask in,
provided an ideal environment to Michael Whinston for completing the manuscript
John Nachbar, Martin Osborne, Ben Polak, Ariel Rubinstein, and Martin Weitzman
during the 1993/1994 academic year. The Universitat Pompeu Fabra also offered its
offered numercus helpful suggestions. The book would undoubtedly have been better
hospitality to Andreu Mas-Colell at numerous points during the book's development.
still had we managed to incorporate all of their ideas.
Finally, we want to offer a special thanks to those who first excited us about
Many generations of first-year Harvard graduate students have helped us with
the subject matter that appears here: Gerard Debreu, Leo Hurwicz, Roy Radner,
their questions, comments, and corrections. In addition, a number of current and
Marcel Richter, and Hugo Sonnenschein (A.M.-C.); David Cass, Peter Diamond,
former students have played a more formal role in the book's development, serving
Franklin Fisher, Sanford Grossman, and Eric Maskin (M.D.W.); Emmanuel
as research assistants in various capacities. Shira Lewin read the entire manuscript,
Drandakis, Ron Jones, Lionel McKenzie, and Edward Zabel (J.R.G.).
finding errors in our proofs, suggesting improvements in exposition, and even (indeed,
often) correcting our grammar. Chiaki Hara, Ilya Segal, and Steve Tadelis, with the
A.M.-C., M.D.W., J.R.G.
assistance of Marc Nachman, have checked that the book's many exercises could be
Cambridge, MA
solved, and have suggested how they might be formulated properly when our first
March 1995
attempt to do so failed. Chiaki Hara and Steve Tadelis have also given us extensive
comments and corrections on the text itself. Emily Mechner, Nick Palmer, Phil Panet,
and Billy Pizer were members of a team of first-year students that read our early
drafts in the summer of 1992 and offered very helpful suggestions on how we could
convey the material better.
Betsy Carpenter and Claudia Napolilli provided expert secretarial support
throughout the project, helping to type some chapter drafts, copying material on
very tight deadlines, and providing their support in hundreds of other ways. Gloria
Gerrig kept careful track of our ever-increasing expenditures.
Our editor at Oxford, Herb Addison, was instrumental in developing the test
teaching program that so helped us in the book's early stages, and offered his support
throughout the book's development. Leslie Phillips of Oxford took our expression
of appreciation for the look of the Feynman Lectures, and turned it into a book
design that exceeded our highest expectations. Alan Chesterton and the rest of the
Microeconomic Theory

Ii
'

...... I
,
 P A R T 0 N E

Individual
Decision Making

A distinctive feature of microeconomic theory is that it aims to model economic

activity as an interaction of individual economic agents pursuing their private
interests. It is therefore appropriate that we begin our study of microeconomic theory
with an analysis of individual decision making.
Chapter I is short and preliminary. It consists of an introduction to the theory
of individual decision making considered in an abstract setting. It introduces the
decision maker and her choice problem, and it describes two related approaches to
modeling her decisions. One, the preference-based approach, assumes that the decision
maker has a preference relation over her set of possible choices that satisfies certain
rationality axioms. The other, the choice-based approach, focuses directly on the
decision maker's choice behavior, imposing consistency restrictions that parallel the
rationality axioms or the preference-based approach.
The remaining chapters in Part One study individual decision making in explicitly
economic contexts. It is common in microeconomics texts-and this text is no
exception-to distinguish between two sets of agents in the economy: individual
consumers and firms. Because individual consumers own and run firms and therefore
ultimately determine a firm's actions, they are in a sense the more fundamental
element of an economic model. Hence, we begin our review of the theory of economic
decision making with an examination of the consumption side of the economy.
Chapters 2 and 3 study the behavior of consumers in a market economy. Chapter 2
begins by describing the consumer's decision problem and then introduces the
concept of the consumer's demand function. We then proceed to investigate the
implications for the demand function of several natural properties of consumer
demand. This investigation constitutes an analysis of consumer behavior in the spirit
of the choice-based approach introduced in Chapter I.
In Chapter 3, we develop the classical preference-based approach to consumer
demand. Topics such as utility maximization, expenditure minimization, duality,
integrability, and the measurement of welfare changes are studied there. We also
discuss the relation between this theory and the choice-based approach studied in
Chapter 2.
In economic analysis, the aggregate behavior of consumers is often more

I
important than the behavior of any single consumer. In Chapter 4, we analyze the

3
 4 PART 1: INDIVIDUAL DECISION MAKING C H A P T E R

extent to which the properties of individual demand discussed in Chapters 2 and 3

also hold for aggregate consumer demand.
In Chapter 5, we study the behavior of the firm. We begin by posing the firm's
decision problem, introducing its technological constraints and the assumption of
profit maximization. A rich theory, paralleling that for consumer demand, emerges.
Preference and Choice 1
In an important sense, however, this analysis constitutes a first step because it takes
the objective of profit maximization as a maintained hypothesis. In the last section
of the chapter, we comment on the circumstances under which profit maximization
can be derived as the desired objective of the firm's owners.
Chapter 6 introduces risk and uncertainty into the theory of individual decision
making. In most economic decision problems, an individual's or firm's choices do
not result in perfectly certain outcomes. The theory of decision making under
uncertainty developed in this chapter therefore has wide-ranging applications to
economic problems, many of which we discuss later in the book.
'"
' .
l.A Introduction
In this chapter, we begin our study of the theory of individual decision making by
considering it in a completely abstract setting. The remaining chapters in Part I
develop the analysis in the context of explicitly economic decisions.
The starting point for any individual decision problem is a set of possible (mutually
exdusit•e) altanatives from which the individual must choose. In the discussion that
follows, we denote this set of alternatives abstractly by X. For the moment, this set
can be anything. For example, when an individual confronts a decision of what career
path to follow, the alternatives in X might be: {go to law school, go to graduate
school and study economics, go to business school, ... , become a rock star}. In
Chapters 2 and 3, when we consider the consumer's decision problem, the elements
of the set X are the possible consumption choices.
There are two distinct approaches to modeling individual choice behavior. The
first, which we introduce in Section l.B, treats the decision maker's tastes, as
summarized in her preference relation, as the primitive characteristic of the individual.
The theory is developed by first imposing rationality axioms on the decision maker's
preferences and then analyzing the consequences of these preferences for her choice
behavior (i.e., on decisions made). This preference-based approach is the more
traditional of the two, and it is the one that we emphasize throughout the book.
The second approach, which we develop in Section l.C, treats the individual's
choice behavior as the primitive feature and proceeds by making assumptions directly
concerning this behavior. A central assumption in this approach, the weak axiom of
rel'ealed prej~rence, imposes an element of consistency on choice behavior, in a sense
paralleling the rationality assumptions of the preference-based approach. This
choice-based approach has several attractive features. It leaves room, in principle,
for more general forms of individual behavior than is possible with the preference-
based approach. It also makes assumptions about objects that are directly observable
(choice behavior), rather than about things that are not (preferences). Perhaps most
importantly, it makes clear that the theory of individual decision making need not
be based on a process of introspection but can be given an entirely behavioral
foundation.

...:..a·------------------
 SECTION 1.8: PREFERENCE RELATIONS
6 CHAPTER 1: PREFERENCE AND CHOICE

Understanding the relationship between these two different approaches to rationality. Transitivity implies that it is impossible to face the decision maker with
modeling individual behavior is of considerable interest. Section l.D investigates this a sequence of pairwise choices in which her preferences appear to cycle: for example,
feeling that an apple is at least as good as a banana and that a banana is at least as
question, examining first the implications of the preference-based approach for choice
good as an orange but then also preferring an orange over an apple. Like the
behavior and then the conditions under which choice behavior is compatible with
completeness property, the transitivity assumption can be hard to satisfy when
the existence of underlying preferences. (This is an issue that also comes up in evaluating alternatives far from common experience. As compared to the complete-
Chapters 2 and 3 for the more restricted setting of consumer demand.) ness property, however, it is also more fundamental in the sense that substantial
For an in-depth, advanced treatment of the material of this chapter, sec Richter portions of economic theory would not survive if economic agents could not be
(1971). assumed to have transitive preferences.
The assumption that the preference relation ;:: is complete and transitive has
l.B Preference Relations implications for the strict preference and indifference relations >- and -.These are
summarized in Proposition l.B.l, whose proof we forgo. (After completing this
In the preference-based approach, the objectives of the decision maker are summar- section, try to establish these properties yourself in Exercises !.B. I and 1.8.2.)
ized in a preference relation, which we denote by ;::. Technically, ;:: is a binary
Proposition 1.8.1: If ;:: is rational then:
relation on the set of alternatives X, allowing the comparison of pairs of alternatives
x. y EX. We read x ;=: y as "x is at least as good as y." From ;::, we can derive two (i) >- is both irreflexive (x >- x never holds) and transitive (if x >- y and y >- z,
other important relations on X: then x >- z).
(i) The strict preference relation, >-.defined by (ii) - is reflexive (x - x for all x), transitive (if x- y and y- z, then x- z).
and symmetric (if x - y, then y - x).
x >- y = x ;:: y but not y ;=: x y;::
(iii) if x >- z, then x >- z.
and read "x is preferred to y."'
(ii) The indijference relation, -, defined by The irreflexivity of>- and the reflexivity and symmetry of - are sensible properties
for strict preference and indifference relations. A more important point in Proposition
X - }' <'> X ;:: y and }' ;:: X
I. B. I is that rationality of ;:: implies that both >- and - are transitive. In addition,
and read "x is indifferent to y." a transitive-like property also holds for >- when it is combined with an at-least-as-
In much of microeconomic theory, individual preferences are assumed to be good-as relation, ;::.
rational. The hypothesis of rationality is embodied in two basic assumptions about
the preference relation ;=:: completeness and transitivity. 2 An individual's preferences may fail to satisfy the transitivity property for a number or
reasons. One difficulty arises because of the problem of just JWrceptible differenas. For
Definition 1.8.1: The preference relation ;:: is rational if it possesses the following two
example, if we ask an individual to choose between two very similar shades of gray for painting
properties:
her room, she may be unable to tell the difference between the colors and will therefore be
(i) Completeness: for all x. y EX, we have that x;:: y or y;:: x (or both). indifferent. Suppose now that we offer her a choice between the lighter of the two gray paints
(ii) Transitivity: For all x, y, z EX, if x;:: y and y;:: z, then x;:: z. and a slightly lighter shade. She may again be unable to tell the difference. If we continue in
this fashion, letting the paint colors get progressively lighter with each successive choice
The assumption that ;=: is complete says that the individual has a well-defined
experiment, she may express indifference at each step. Yet, if we offer her a choice between
preference between any two possible alternatives. The strength of the completeness
the original (darkest) shade of gray and the final (almost white) color, she would be able to
assumption should not be underestimated. Introspection quickly reveals how hard distinguish between the colors and is likely to prefer one of them. This, however, violates
it is to e\·aluate alternatives that are far from the realm of common experience. It takes transitivity.
work and serious reflection to find out one's own preferences. The completeness Another potential problem arises when the manner in which alternatives are presented
axiom says that this task has taken place: our decision makers make only meditated matters for choice. This is known as the framing problem. Consider the following example,
choices. paraphrased from Kahneman and Tversky (1984):
Transitivity is also a strong assumption, and it goes to the heart of the concept of
lmagmc that you arc about to purcha~e a stereo for 125 dollars and a calculator for 15
dollars. The salesman tells you that the calculator is on sale for 5 dollars less at the other
1. The symbol -=- is read as "if and only ir." The literature sometimes speaks of x;;: y as "x branch of the store. located 10 minutes away. The stereo is the same price there. Would
is weakly preferred to .\' .. and x >- y as ··x is strictly preferred to y:· We shall adhere to the you make the trip to the other store?
terminology introduced above.
It turns out that the fraction or respondents saying that they would travel to the other store
2. Note that there is no unified terminology in the literature; weak order and complete preorder
are common alternatives to the term rational preference relation. Also, in some presentations, the
for the 5 dollar discount is much higher than the fraction who say they would travel when the
assumption that ;:: is reflex ire (defined as x ~ x for all x EX) is added to the completeness and question is changed so that the 5 dollar saving is on the stereo. This is so even though the
tr;:msitivity assumptions. This property is, in fact, implied by completeness and so is redundant. ultimate saving obtained by incurring the inconvenience of travel is the same in both
8 CHAPTER 1: PREFERENCE AND CHOICE SECTION 1.C: CHOICE RULES 9

cases.' Indeed, we would expect indifference to be the response to the following question: elements of X in accordance with the individual's preferences. This is stated more
Because of a stockout you must travel to the other store to get the two items, but you will precisely in Definition I.B.2.
receive S dollars off on either item as compensation. Do you care on which item this S
dollar rebate is given? Definition 1.6.2: A function u: X-+ R is a utility function representing preference
relation ;:::; if, for all x, y eX.
If so, however, the individual violates transitivity. To see this, denote
x):;y = u(x) ~u(y).
x =Travel to the other store and get a 5 dollar discount on the calculator.
}' = Travel to the other store and get a 5 dollar discount on the stereo. Note that a utility function that represents a preference relation ;:::; is not unique.
z = Buy both items at the first store. For any strictly increasing function f: R -+ R, v(x) = f(u(x)) is a new utility function
The first two choices say that x >- z and z >- y, but the last choice reveals x- y. Many problems representing the same preferences as u( · ); see Exercise 1.8.3. It is only the ranking
of framing arise when individuals are faced with choices between alternatives that have
uncertain outcomes (the subject of Chapter 6). Kahneman and Tversky (1984) provide a
number of other interesting examples.
At the same time, it is often the c-.tse that apparently intransitive behavior can be explained
fruitfully as the result of the interaction of several more primitive rational (and thus transitive)
of alternatives that mailers. Properties of utility functions that are invariant for any
strictly increasing transformation are called ordinal. Cardinal properties are those not
preserved under all such transformations. Thus, the preference relation associated
with a utility function is an ordinal property. On the other hand, the numerical values
associated with the alternatives in X, and hence the magnitude of any differences in
•
preferences. Consider the following two examples
the utility measure between alternatives, are cardinal properties.
(i) A household formed by Mom (M), Dad (D), and Child (C) makes decisions by majority
The ability to represent preferences by a utility function is closely linked to the
voting. The alternatives for Friday evening entertainment are attending an opera (0), a rock
assumption of rationality. In particular, we have the result shown in Proposition
concert (R), or an ice-skating show (1). The three members of the household have the rational
individual preferences: 0 >-u R ;.., I, I >-o 0 >-oR, R >-c I >-cO, where ;..,., >- 0 , >-care the I.B.2.
transitive individual strict preference relations. Now imagine three majority-rule votes: 0 versus Proposition 1.6.2: A preference relation ;:::; can be represented by a utility function
R, R versus I, and I versus 0. The result of these votes (0 will win the first, R the second, and only if it is rational.
I the third) will make the household's preferences ~have the intransitive form: 0 >- R >-I >- 0.
(The intransitivity illustrated in this example is known as the Condorcet paradox, and it is Proof: To prove this proposition, we show that if there is a utility function that
a central difficulty for the theory of group decision making. For further discussion, see represents preferences ;:::;, then )::; must be complete and transitive.
Chapter 21.)
(ii) Intransitive decisions may also sometimes be viewed as a manifestation of a change of Completene.~s. Because u( ·)is a real-valued function defined on X, it must be that
tastes. For example, a potential cigarette smoker may prefer smoking one cigarette a day to for any x, y eX, either u(x) ~ u(y) or u(y) ~ u(x). But because u( ·)is a utility function
not smoking and may prefer not smoking to smoking heavily. But once she is smoking one representing ;:::;, this implies either that x)::; y or that y;:::; x (recall Definition I.B.2).
cigarette a day, her tastes may change, and she may wish to increase the amount that she Hence, ;:::; must be complete.
smokes. Formally, letting y be abstinence, x be smoking one cigarette a day, and z be heavy
smoking, her initial situation is y, and her preferences in that initial situation are x >- y >- z. Tramitivity. Suppose that x)::; y and y;:::; z. Because u( ·) represents ;:::;, we must
But once .< is chosen over y and z, and there is a change of the individual's current situation have u(x) ~ u(y) and u(y) ~ u(z). Therefore, u(x) ~ u(z). Because u(·) represents)::;,
from y to .<, her tastes change to z >- x >- y. Thus, we apparently have an intransitivity: this implies x;:::; :. Thus, we have shown that x;:::; y and y;:::; z imply x;:::; z, and so
z >- x >- z. This change-of-tastes model has an important theoretical bearing on the analysis transitivity is established. •
of addictive behavior. It also raises interesting issues related to commitment in decision making
(see Schelling (1979)]. A rational decision maker will anticipate the induced change of tastes At the same time, one might wonder, can any rational preference relation )::; be
and will therefore attempt to tie her hand to her initial decision (Ulysses had himself tied to described by some utility function? It turns out that, in general, the answer is no. An
the mast when approaching the island of the Sirens). example where it is not possible to do so will be discussed in Section 3.G. One case
It often happens that this change-of-tastes point of view gives us a well-structured way to in which we can always represent a rational preference relation with a utility function
think about nonrational decisions. See Elster (1979) for philosophical discussions of this and arises when X is finite (see Exercise I.B.5). More interesting utility representation
similar points. results (e.g., for sets of alternatives that are not finite) will be presented in later
chapters.
Utility Functions
In economics, we often describe preference relations by means of a utility function.
l.C Choice Rules
A utility function u(x) assigns a numerical value to each element in X, ranking the In the second approach to the theory of decision making, choice behavior itself is
taken to be the primitive object of the theory. Formally, choice behavior is
3. Kahneman and Tversky aHribute this finding to individuals keeping .. mental accounts" in represented by means of a choice structure. A choice structure (&il, C( · )) consists of
which the savings are compared to the price of the item on which they are received. two ingredients:

'
 SECTION 1.0: RELATIONSHIP BETWEEN PREFERENCE RELATIONS AND CHOICE RULES
10 CHAPTER 1: PREFERENCE AND CHOICE

(i) !!I is a family (a set) of nonempty subsets of X; that is, every element of !JI is Note how the assumption that choice behavior satisfies the weak axiom captures the
a set B c X. By analogy with the consumer theory to be developed in Chapters 2 consistency idea: If C({x, y}) = (x}, then the weak axiom says that we cannot have
and 3, we call the elements Be !JI budget sets. The budget sets in !JI should be thought C({x,y,z}) = (y}.'
of as an exhaustive listing of all the choice experiments that the institutionally, A somewhat simpler statement of the weak axiom can be obtained by defining
physically, or otherwise restricted social situation can conceivably pose to the decision a revealed preference relarion ;::;• from the observed choice behavior in C( · ).
maker. It need not, however, include all possible subsets of X. Indeed, in the case of
Definition 1.C.2: Given a choice structure (!JI, C( ·))the revealed preference relation
consumer demand studied in later chapters, it will not. ;::;• is defined by
(ii) C( ·) is a choice rule (technically, it is a correspondence) that assigns a x ;::;• y = there is some 8 e !JI such that x, y E 8 and x E C(8).
nonempty set of chosen elements C(B) c B for every budget set Be !fl. When C(B)
contains a single element, that element is the individual's choice from among the We read x ;::;• y as "xis revealed at least as good as y." Note that the revealed
alternatives in B. The set C(B) may, however, contain more than one element. When preference relation ;::;• need not be either complete or transitive. In particular, for
it does, the elements of C(B) are the alternatives in B that the decision maker mig/11 any pair of alternatives x andy to be comparable, it is necessary that, for some Be !JI,
choose; that is, they are her acceptable alternatives in B. In this case, the set C(B) we have x, y e B and either x e C(B) or y E C(B), or both.
can be thought of as containing those alternatives that we would actually see chosen We might also informally say that "xis revealed preferred to}'" if there is some
if the decision maker were repeatedly to face the problem of choosing an alternative BE :Jl such that x, y E B, x e C(B), and y f C(B), that is. if x is ever chosen over y
from set B. when both are feasible.
With this terminology, we can restate the weak axiom as follows:" If x is re••ealed
Example I.C.I: Suppose that X= (x,y,z} and !JI = {(x,y}, {x,y,z}}. One possible
at least as yoocl as J', tll<'ll y ca1111ot be revealed preferred to x."
choice structure is (!JI, C 1(·)), where the choice rule C 1('} is: C 1((x,y}) = {x} and
C 1 ({x, y, z}) = {x}. In this case, we see x chosen no matter what budget the decision Example I.C.2: Do the two choice structures considered in Example I.C.I satisfy the
maker faces. weak axiom? Consider choice structure (91, C 1( ·)).With this choice structure, we have
Another possible choice structure is (!!1, C2 ( • )), where the choice rule C 2 ( ·) is: x ;::;• .rand x ;::;• z. but there is no revealed preference relationship that can be inferred
C2 ({x, y}) = {x} and C 2 ({x, y, z}) = {x, y}.ln this case, we see x chosen whenever the between y and z. This choice structure satisfies the weak axiom because y and z are
decision maker faces budget {x, y}, but we may see either x or y chosen when she never chosen.
faces budget (x, y, z}. • Now consider choice structure (9?, C 2 ( ·)).Because C,({x, y, z}) = {x, y). we have
When using choice structures to model individual behavior, we may want to y ;::;• x (as well as x ;::;• ,r. x ;::;• :, and y ;::• :). But because C 2 ({x, y}) = {x}, x is
impose some "reasonable" restrictions regarding an individual's choice behavior. An revealed preferred to y. Therefore, the choice structure (!!I, C 2 ) violates the weak
important assumption, the weak axiom of revealed preference [first suggested by axiom. •
Samuelson; see Chapter 5 in Samuelson (1947)], reflects the expectation that an
We should note that the weak axiom is not the only assumption concerning choice
individual's observed choices will display a certain amount of consistency. For
behavior that we may want to impose in any particular setting. For example, in the
example, if an individual chooses alternative x (and only that) when faced with a
choice between x and y, we would be surprised to see her choose y when faced with consumer demand setting discussed in Chapter 2. we impose further conditions that
arise naturally in that context.
a decision among x, y, and a third alterative z. The idea is that the choice of x when
The weak axiom restricts choice behavior in a manner that parallels the use of
facing the alternatives (x, y} reveals a proclivity for choosing x over y that we should
expect to see reflected in the individual's behavior when faced with the alternatives the rationality assumption for preference relations. This raises a question: What is
{x, y, z}. 4 the precise relationship between the two approaches? In Section I.D, we explore
this matter.
The weak axiom is stated formally in Definition I.C.I.
Definition 1.C.1: The choice structure (!fl. C( ·)) satisfies the weak axiom of revealed
preference if the following property holds: l.D The Relationship between Preference Relations and
If for some 8 E !JI with x, y e 8 we have x e C(8), then for any 8' e ~with Choice Rules
x, y E 8' and y e C(8'), we must also have x E C(8').
We now address two fundamental questions regarding the relationship between the
In words, the weak axiom says that if xis ever chosen when y is available, then there
two approaches discussed so far:
can be no budget set containing both alternatives for which y is chosen and x is not.

4. This proclivity might reflect some underlying "'preference" for x over}' but might also arise 5. In f<~ct, it says more; We must have C(:x.y.:}) = {x}, ={:},or ={x.:~. You are asked to
in other ways. It could. for example, be the result of some evolutionary process. show this in Exercise l.C.J. See also Exercise t.C.2.
12 CHAPTER 1: PREFERENCE AND CHOICE SE c TI0 N 1 . D: R E LA T I 0 N S H I P 8 E T W E EN P REF E R EN C E R E LA T I 0 N S AN 0 CH0 I C E RULES 13

(i) If a decision maker has a rational preference ordering ~. do her decisions all budget sets in !Jl. In a sense, preferences explain behavior; we can interpret the
when facing choices from budget sets in £11 necessarily generate a choice decision maker's choices as if she were a preference maximizer. Note that in general,
structure that satisfies the weak axiom? there may be more than one rationalizing preference relation ~ for a given choice
(ii) If an individual's choice behavior for a family of budget sets £11 is captured structure (!14, C( · )) (see Exercise l.D.l).
by a choice structure (£11, C( ·))satisfying the weak axiom, is there necessarily a Proposition l.D.l implies that the weak axiom must be satisfied if there is to be
rational preference relation that is consistent with these choices? a rationalizing preference relation. In particular, since C*( ·, ~) satisfies the weak
axiom for any ~. only a choice rule that satisfies the weak axiom can be rationalized.
As we shall see, the answers to these two questions are, respectively, "yes" and
It turns out, however, that the weak axiom is not sufficient to ensure the existence
"maybe".
of a rationalizing preference relation.
To answer the first question, suppose that an individual has a rational preference
relation ~ on X. If this individual faces a nonempty subset of alternatives B c X, Example 1.0.1: Suppose that X = {x, y, z}, !J4 = { {x, y}, {y, z}, {x, z} }, C({x, y}) = {x},
her preference-maximizing behavior is to choose any one of the elements in the set: C({y, z}) = {y}, and C({x, z}) = {z}. This choice structure satisfies the weak axiom
(you should verify this). Nevertheless, we cannot have rationalizing preferences. To
C*(8, ~) = {x e 8: x ~ y for every y e 8}
sec this, note that to rationalize the choices under {x, y} and {y, z} it would be
The elements of set C*(8, ~)are the decision maker's most preferred alternatives in necessary for us to have x >- y and y >- z. But, by transitivity, we would then have
B. In principle, we could have C*(B, ~) = 0 for some B; but if X is finite, or if x >- z, which contradicts the choice behavior under {x, z}. Therefore, there can be no
suitable (continuity) conditions hold, then C*(8, ~)will be nonempty. 6 From now rationalizing preference relation. •
on, we will consider only preferences ~ and families of budget sets £11 such that
C*(8, ~)is nonempty for all Be J!. We say that the rational preference relation ~ To understand Example J.D. I, note that the more budget sets there are in 94, the
generates the choice structure (91, C*( ·, ~ )). more the weak axiom restricts choice behavior; there are simply more opportunities
The result in Proposition J.D. I tells us that any choice structure generated by for the decision maker's choices to contradict one another. In Example I.D.l, the set
rational preferences necessarily satisfies the weak axiom. {x, y, z} is not an element of 94. As it happens, this is crucial (see Exercises 1.0.3). As
we now show in Proposition 1.0.2, if the family of budget sets 94 includes enough
Proposition 1.0.1: Suppose that ~ is a rational preference relation. Then the choice subsets of X, and if (!14, C( ·)) satisfies the weak axiom, then there exists a rational
structure generated by ~. (91, C*( ·, ~)), satisfies the weak axiom.
preference relation that rationalizes C( ·) relative to £JI [this was first shown by Arrow
Proof: Suppose that for some Be Yd, we have x, y e 8 and x e C*(B, ~). By the (1959)].
definition of C*(B, ~).this implies x ~ y. To check whether the weak axiom holds, Proposition 1.0.2: If (.c?J, C( ·)) is a choice structure such that
suppose that for some B' e !J4 with x, y e 8', we have y e C*(B', ~).This implies that
y ~ z for all z e 8'. But we already know that x ~ y. Hence, by transitivity, x ~ z for (i) the weak axiom is satisfied,
all z e 8', and sox e C*(B', ~).This is precisely the conclusion that the weak axiom (ii) 94 includes all subsets of X of up to three elements,
demands. • then there is a rational preference relation ~ that rationalizes C( ·) relative to £11;
that is, C(8) = C*(8, ~) for all 8 e !Jl. Furthermore, this rational preference
Proposition I.D.l constitutes the "yes" answer to our first question. That is, if
relation is the only preference relation that does so.
behavior is generated by rational preferences then it satisfies the consistency
requirements embodied in the weak axiom. Proof: The natural candidate for a rationalizing preference relation is the revealed preference
In the other direction (from choice to preferences), the relationship is more subtle. relation <:::;•. To prove the result, we must first show two things: (i) that ;::• is a rational
To answer this second question, it is useful to begin with a definition. preference relation, and (ii) that ;::• rationalizes C( ·)on il. We then argue, as point (iii), that
;::• is the unique preference relation that does so.
Definition 1.0.1: Given a choice structure (.M, C(·)), we say that the rational prefer-
(i) We first check that ;::• is rational (i.e., that it satisfies completeness and transitivity).
ence relation ~ rationalizes C( ·) relative to Jl if
C(8) = C*(8, ~) Complereness By assumption (ii), {x, y} E il. Since either x or y must be an element of
C({x, y}), we must have x <:::;• .v. or y <:::;• x, or both. Hence ;::• is complete.
for all 8 e -~. that is, if ~ generates the choice structure (Yd, C( · )).
Transitiviry Let x <:::;• y and )' <:::;• z. Consider the budget set {x, y, z} E 91. It suffices to
In words, the rational preference relation ~ mtionalizes choice rule C( ·)on 94 prove that .< e C({x, y, :)), since this implies by the definition of <:::;• that x <:::;• z. Because
if the optimal choices generated by <:::; (captured by C*( ·, ~))coincide with C( ·)for C({x, y, z}) # 0. at least one of the alternatives.<, y, or z must be an element of C({x, y, z}).
Suppose that y e C({x. y, :)). Since x <:::;• y, the weak axiom then yields x E C({x, y, z}), as we
6. Exercise I.D.2 asks you to establish the noncmptim.-ss of C*(B, ~} for the case where X is
want. Suppose instead that z E C({x, y, z}); since y <:::;• z, the weak axiom yields ye C({.<, y, z}),
fmite. For general results. See Section M.F of the Malhcmatical Appendix and Section 3.C for a and we are in the previous case.
specific application. (ii) We now show that C(B) = C*(B, <:;*) for all BE il; that is, the revealed preference
14 CHAPTER 1: PREFERENCE AND CHOICE EXERCISES

relation ;::• inferred from C( ·) actually generates C( · ). Intuitively, this seems sensible. REFERENCES
Formally, we show this in two steps. First, suppose that x e C(B). Then x ;::• y for all y e B;
so we have x e C*(B, l:*). This means that C(B) c C*(B, l:*). Next, suppose that x e C*(B, l:*). Arrow. K. (1959). Rational choice functions and orderings. Economerrica 26: 121-27.
This implies that x ;::• y for ally e B; and so for each y e B, there must exist some set B, e !!I Elster, J. (t979). Ulys.<es and th• Sir<n•. Cambridge, U.K.: Cambridge Univenity Pres..
such that x, y e B, and x e C(B,). Because C(B) ¢. 0. the weak axiom then implies that Kahneman. D., and A. Tvenky. (1984~ Choices. values, and rrames. Am<rican P•ychologi•t 39: 341-50.
x e C(B). Hence, C*(B, l:*) c C(B). Together, these inclusion relations imply that C(B) = Ploll, C. R. (1973). Path independence. rationality and social choice. Eco110111<trica ~I: 1075-91.
C*(B, l:*). Richter. M. (1971). Rational choice. Chap. 2 in Preferenct5, Utility and Demand, edited by J. Chipman,
(iii) To establish uniqueness, simply note that because iJI includes all two-element subsets L. Uurwicz. and H. Sonnenschein. New York: Harcourt Brace Jovanovich.
of X, the choice behavior inC(·) completely determines the pairwise preference relations over Samuelson. P. ( 1947). Fmmdmions of Economic Analysis. Cambridge. Mass.: Harvard University Press.
Schelling. T. (1979). Micrumotil'es and Maaobelzat'ior. New York: Norton.
X of any rationalizing preference.
Thurstonc, L. L. ( 1927). A law of comparative judgement. Psychological Review 34: 275-86.
This completes the proof. •

We can therefore conclude from Proposition 1.0.2 that for the special case in EXERCISES
which choice is defined for all subsets of X, a theory based on choice satisfying the
weak axiom is completely equivalent to a theory of decision making based on rational 1.8.1 8 Prove property (iii) of Proposition 1.8.1.
preferences. Unfortunately, this special case is too special for economics. For many 1.8.2' Prove properties (i) and (ii) of Proposition 1.8.1.
situations of economic interest, such as the theory of consumer demand, choice is
1.8.38 Show that iff: R- R is a strictly increasing function and u:X - R is a utility function
defined only for special kinds of budget sets. In these settings, the weak axiom does
representing preference relation ;::, then the function v: X - R defined by v(x) = f(u(x)) is
not exhaust the choice implications of rational preferences. We shall see in Section 3.J,
also a utility function representing preference relation ;::.
however, that a strengthening of the weak axiom (which imposes more restrictions
on choice behavior) provides a necessary and sufficient condition for behavior to be 1.8.4' Consider a rational preference relation ;::. Show that if u(x) = u(y) implies x - y and
capable of being rationalized by preferences. if u(x) > u(y) implies x :> y, then u( ·) is a utility function representing ;::.

1.8.58 Show that if X is finite and ;:: is a rational preference relation on X, then there is a
Definition I. D. I defines a rationalizing preference as one for which C(B) = C*(B, l:). An utility function u: X - R that represents ;::. [Him: Consider first the case in which the
alternative notion of a rationalizing preference that appears in the literature requires only that individual's ranking between any two elements of X is strict (i.e., there is never any indifference),
C(B) c C*(B, l:); that is, ;:: is said to rationalize C( ·) on iJI if C(B) is a subset of the most and construct a utility function representing these preferences; then extend your argument to
preferred choices generated by ;::, C*(B, 1:::; ), for every budget Be !!1. the general case.)
There are two reasons for the possible use of this alternative notion. The first is, in a sense,
I.C.I 8 Consider the choice structure(~. C( ·))with .lit=({.~. y}, {x, y, zj) and C({x, y}) = {.<}.
philosophical. We might want to allow the decision maker to resolve her indifference in some
Show that if (.JI. C( ·))satisfies the weak axiom, then we must have C({x, y, :j) = {x}, = {:l·, or
specific manner, rather than insisting that indifference means that anything might be picked.
The view embodied in Definition 1.0.1 (and implicitly in the weak axiom as well) is that if
={x,:J.
she chooses in a specific manner then she is, de facto, not indifferent. I.C.2 11 Show that the weak axiom (Defmition I.C.I) is equivalent to the following property
The second reason is empirical. If we are trying to determine from data whether an holding:
individual's choice is compatible with rational preference maximization, we will in practice Suppose that IJ. /J' E .JI, that .<, y E B, and that x. r E 8'. Then if x E C(B) and y E C(IJ'), we
have only a finite number of observations on the choices made from any given budget set B. must have {x • .rl c C(B) and {x. y} c ((8').
If C(B) represents the set of choices made with this limited set of observations, then because
these limited observations might not reveal all the decision maker's preference maximizing I.C.3" Suppose thai choice structure(.-'.!, C( ·))satisfies the weak axiom. Consider the following
choices, C(B) c C*(B, l:) is the natural requirement to impose for a preference relationship two possible revealed preferred relations. ::>* and >-**:
to rationalize observed choice data. x ;>* !' = there is some /J e .11 such that x. y e B, x e C(B), and y ¢ C(B)
Two points are worth noting about the effects or using this alternative notion. First, it is x >-** _r <:::> x ;::-• .r but not J' ~· x
a weaker requirement. Whenever we can find a preference relation that rationalizes choice in where 2::* is the revealed at-least-as-good-as relation defined in Definition I.C.2.
the sense of Definition J.D. I, we have found one that does so in this other sense, too. Second, (a) Show that :> • and :>** give the same relation over X: that is, for any x, y EX.
in the abstract setting studied here, to find a rationalizing preference relation in this latter x :>* r = x :>** y. Is this still true if (.JI. C( ·))does not satisfy the weak axiom''
sense is actually trivial: Preferences that have the individual indifferent among all elements of
(b) Must >-*be transitive'!
X will rationalize any choice behavior in this sense. When this alternative notion is used in
the economics literature, there is always an insistence that the rationalizing preference relation (c) Show that if .JI includes all three-element subsets of X, then :>* is transitive.
should satisfy some additional properties that are natural restrictions for the specific economic 1.0.1 8 Give an example of a choice structure that can be rationalized by several preference
context being studied. relations. Note that if the family of budgets .lit includes all the two-element subsets of X, then
there can be at most one rationalizing preference relation.
16 CHAPTER 1: PREFERENCE AND CHOICE
C H A P T E R

1.0.2• Show that if X is finite, then any rational preference relation generates a nonempty
choice rule; that is, C(B) # 0 for any B c X with B # 0.

1.0.3" Let X = {x, y, z}, and consider the choice structure (oil, C( ·)) with
oil= {{x, y}, {y, :}, {x, z}, {x, y, z}}
Consumer Choice 2
and C({x,y})= {x}, C({y,z})= {y}, and C({x,z})= {z}, as in Example 1.0.1. Show that
(J1, C( ·)) must violate the weak axiom.

1.0.4" Show that a choice structure(.;!, C( ·))for which a rationalizing preference relation <::;
e>ists satisfies the patll-invariana property: For every pair 8 1 , 8 1 e 94 such that 8 1 u 8 1 e :il
and C(8 1) u C(B 1 ) e.;!, we have C(B 1 u 8 1 ) = C(C(8 1) u C(B,)), that is, the decision problem
can safely be subdivided. See Plott (1973) for further discussion.

I.I>.Sc Let X = {x, y, z} and.;! = {{.<, r}. {y, z), {:, x} }. Suppose that choice is now stochastic
in the sense that, for every 8 e 94, C(B) is a frequency distribution over alternatives in B. For
example, if B = {x. y}, we write C(B) = (C.(B), C,(B)), where C.(B) and C,(B) arc nonnegative
numbers with C,.(B) + C,(B) = I. We say that the stochastic choice function C( ·) can be
mtionali:ecl hy preji•mtce;· if we can find a probability distribution l'r over the six possible 2.A Introduction
(strict) preference relations on X such that for every Be.;!, C(B) is precisely the frequency of
choices induced by Pr. For example, if B = {x. y). then C.(B) = Pr ({ >-: x >- y}). This concept The most fundamental decision unit of microeconomic theory is the consumer. In
ot ;ginates in Thurstone ( 1927), and it is of considerable econometric interest (indeed, it provides this chapter, we begin our study of consumer demand in the context of a market
a theory for the error term in observable choice). economy. By a market economy, we mean a setting in which the goods and services
(a) Show that the stochastic choice function C(!x, ylJ = C({y, z}) = C({:, x}) = (!, iJ can that the consumer may acquire are available for purchase at known prices (or,
be rationalized by preferences. equivalently, are available for trade for other goods at known rates of exchange).
(b) Show that the stochastic choicefunction C({ x, y}) = C( {)',:}) = C({:, x}) = <1.1) is not We begin, in Sections 2.6 to 2.D, by describing the basic elements of the
rationalizable by preferences. consumer's decision problem. In Section 2.6, we introduce the concept of commodities,
(c) Determine the 0 < 2 < I at which C({x. y}) = C({J', z}) = C({z. x}) =(>,I ->)switches the objects of choice for the consumer. Then, in Sections 2.C and 2.D, we consider
from rationalizable to nonrationalizable. the physical and economic constraints that limit the consumer's choices. The former
are captured in the consumption set, which we discuss in Section 2.C; the latter are
incorporated in Section 2.D into the consumer's Walrasian budget set.
The consumer's decision subject to these constraints is captured in the consumer's
Walrasian demand functiott. In terms of the choice-based approach to individual
decision making introduced in Section I.C, the Walrasian demand function is the
consumer's choice rule. We study this function and some of its basic properties in
Section 2.E. Among them are what we call comparative statics properties: the ways
in which consumer demand changes when economic constraints vary.
Finally, in Section 2.F, we consider the implications for the consumer's demand
function of the weak axiom of revealed prejere11ce. The central conclusion we reach
is that in the consumer demand setting, the weak axiom is essentially equivalent to
the compensated law of demaml, the postulate that prices and demanded quantities
move in opposite directions for price changes that leave real wealth unchanged.

2.B Commodities
The decision problem faced by the consumer in a market economy is to choose
consumption levels of the various goods and services that are available for purchase
in the market. We call these goods and services commodities. For simplicity, we
assume that the number of commodities is finite and equal to L (indexed by
t=l, .. . , L). _.\

17