Man& on methods of estimating pupulatbn

MANUAL VIII

Methods for Projections

of Urban and Rural Population
Department of Economic and Social Affairs
POPULATION STUDIES, No. 55

Manuals on methods of estimating population

MANUAL VIII

Methods for Projections

of Urban and Rural Population

United Nations
New York, .I 974
 NOTE

Symbols of United Nations documents are composed of capital letters combined

with figures. Mention of such a symbol indicates a reference to a United Nations
document.
The designations employed and the presentation of the material in this publica-
tion do not imply the expression of any opinion whatsoever on the part of the
Secretariat of the United Nations concerning the legal status of any country or
territory or of its authorities, or concerning the delimitation of its frontiers.
Population Studies Nos. 1 to 54 were issued under the series symbol ST/SOA/
Series All-54.

Sales No. E.74.XIII.3

Price: $U.S. 7.00

(or equivalent in other currencies)
 FOREWORD

Pursuant to the recommendations of the Population Also, within the context of this coherent and cumulative
Commission, the United Nations Secretariat has been programme, two other publications should be mentioned;
preparing several manuals describing methods of demo- namely, Estimating Future School Enrolment in Develop-
graphic analysis, estimation and projection needed for ing Countries, a Manual of Methodology, published
economic agd social policy purposes and suitable for use jointly by the United Nations and UNESCO, and a
in many countries, including those where demographic technical report entitled The Concept of a Stable Popula-
statistics and methods of analysis are not yet well advanc- tion: Application to the Study of Populations of Countries
ed. Some of those manuals deal with the analysis and with Incomplete Demographic Statistics, lo which presents
evaluation of basic statistics, notably those of population the theoretical background of part of the aforementioned
censuses, and others are concerned with the projection Manual IV.
of various population quantities which are needed in
diverse fields of economic and social planning. This In this manual, projections of urban and rural popula-
manual, concerned with the projection of urban and rural tion are dealt with on the assumption that methods of
population, is part of this longer-range programme. projection of a country's total population, or its total
population by groups of sex and age, are already known,
The following manuals have been published so far in and that such projections have in fact already been
the Manuals on Methods of Estimating Population series : carried out. Those methods have been dealt with in
Manual I: Methods of Estimating Total Population for Manual IZI of the present series. It is also assumed here
Current Dates ;l that the reader is somewhat familiar with the appraisal
Manual 11: Methods of Appraisal of Quality of Basic of accuracy in basic statistics, a subject developed at
Data for Population Estimates; some length in the previous Manual II.
Manual 111: Methods for Population Projections by This manual has been drawn up especially with a view
Sex and Age ; to its uses in less developed countries or countries whose
Manual IV: Methods of Estimating Basic Demographic population statistics are not very detailed. It is addressed
Measures from Incomplete Data; mainly to population analysts possessing limited technical
Manual V: Methods of Projecting the Economically means, and it does not consider the possible uses of
Active Population ; computer methodology. The methods are accordingly
Manual VI: Methods of Measuring Internal Migration ; simple, but, depending on opportunities, may be elab-
orated further.
Manual VII: Methods of Projecting Households and
Families; and, related to this series, The United Nations is indebted to the Demographic
Methods of Analysing Census Data on Economic Research and Training Centre (CELADE) in Santiago,
Activities of the Population. Chile; the United States Bureau of the Census; K. V.
-- Ramachandran of the Regional Institute for Population
I United Nations publication, Sales No. 52.XIII.5. Studies in Accra, Ghana; and D. Courgeau of the
a United Nations publication, Sales No. 56.XIII.2. Institut National d'Etudes Demographiques in Paris for
United Nations publication, Sales No. 56.XIII.3. their review of the draft of this manual, and the many
United Nations publication, Sales No. 67.XIII.2. useful suggestions which have been incorporated in its
United Nations publication, Sales No. E.70.XIII.2 (in co- final version.
operation with the ILO).
United Nations publication, Sales No. E.70.XIII.3.
United Nations publication, Sales No. 73.XIII.2. United Nations publication, Sales No. 66.XIII.3.
8 United Nations publication, Sales No. E.69.XIII.2. lo United Nations publication, Sales No. 65.XIII.3.
 CONTENTS

Chapter
I . PROBLEMS OF DEFINITION OF URBAN POPULATION ........................ 9
Administrative definition: type of local government ...................... 9
Administrative definition: seats of district government .................... 10
Definition by size of administrative units .............................. 10
Economic definition. applied to administrative units .................... 11
Geographic definition : agglomerations ................................ 11
Geographic definition: metropolitan areas .............................. 12
Otherdefinitions .................................................... 12
Purposes served by various definitions .................................. 13

11. COMPONENTS OF URBAN AND RURAL POPULATION CHANGE ..............

Differences between urban and rural sex-age structures ..................
Urban residence ratios by sex and age ..................................
Differences between urban and rural crude death rates ..................
Differences between urban and rural crude birth rates ....................
Differences between urban and rural rates of natural increase ..............
Differences due to the effects of international migration ..................
Differencesdue to internal rural-to-urban migration ....................
Effects of rural-to-urban area reclassification ............................
Net rural-to-urban population transfers. i.e. combined effects of migration
and area reclassification ............................................
Implications for projection methodology ..............................

.
I11 TEMPOOF URBANIZATION AND URBAN CONCENTRATION .................. 25
General considerations .............................................. 25
Concept of a growth rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Tempo of urbanization .............................................. 26
Reasons for use of the urban-rural growth difference as a measure of the tempo
of urbanization .................................................. 27
Tempo of urban concentration ........................................ 30

IV. PROJECTION OF URBAN AND RURAL POPULATION TOTALS USING THE SIMPLEST
METHODS ........................................................ 32
Use of urban growth rates ............................................ 32
Use of rural growth rates ............................................ 33
Ratio method ...................................................... 34
Summary .......................................................... 35

V . UNITEDNATIONS METHOD OF URBAN AND RURAL POPULATION PROJECTIONS . 36

Actual observations concerning urban-rural growth differences . . . . . . . . . . . . 36
Formula for iterative calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Year-to-year interpolation of a five-yearly projection .................... 38
Applications of the method (by annual rates) .......................... 38
Chapter Page
Same method. using exponential rates .................................. 42
Flexible assumptions ................................................ 43

V1. PROJECTIONS
FOR INDIVIDUAL CITIES. GROUPS OF CITIES AND DISTINCT GROUPS
OF LOCALITIES .................................................. 45
Methods not dealt with in this manual .................................. 45
Ratio method ...................................................... 46
United Nations method .............................................. 46
Fixed groups of cities .............................................. 48
Expanding groups of cities .......................................... 48
Projection of population by size group of locality ........................ 50

VII . SUPPLEMENTARY
ESTIMATION OF SEX-AGE COMPOSITION FOR PROJECTED TOTALS
OF URBAN AND RURAL POPULATION ................................ 55
A. Preliminary considerations ........................................ 55
Use of supplementary methods .................................... 55
Adjustment of defective age data for urban and rural projections . . . . . . . . 55
B. Method of difference elimination .................................... 56
Explanation of the method ........................................ 56
Application to a projection ........................................ 57
Further uses of this method ........................................ 59
Additional considerations ........................................ 62
A shorter method ................................................ 63
C. Estimating sex-age groups with the use of the logistic table ............ 63

VIII. MIXEDPROJECTION METHODS .......................................... 69

Crude components methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Composite method employing urban residence ratios for fixed sex-age groups 70
Composite method employing urban residence ratios by cohort ............ 72
Migration-survival method as applied to the rural population ............ 78

A . Derivation of basic data and estimates .............................. 83

B. Standard projection .............................................. 95
C. Variant projections ................................................ 101

ANNEXES

I . Table of logistic curve .............................................. 117

I1. Table of survival ratios (P. ) of model life tables for five-year age groups and
five-year intervals of time ............................................ 123
 INTRODUCTION

Although the concept of urban as distinct from rural fields of economic, social, administrative and physical
places has existed since ancient times, urban and rural development, and in the maintenance of environmental
classifications were introduced into the compilations of quality, which have to be investigated with reference to
European population statistics only during the nine- current and future estimates of urban and rural popula-
teenth century. Most of the statistics then available on tion.
an international basis, including many countries of the
world, were compiled and discussed at the end of the USESAND APPLICATIONS OF URBAN AND
nineteenth century by Adna Ferrin Weber. Although RURAL POPULATION PROJECTIONS
some very famous cities had arisen even in ancient Many detailed planning problems have arisen in
times, most cities were relatively small at the opening connexion with the huge increases in both urban and rural
of the nineteenth century, and the bulk of the world's population and the large transfers of population from
population was for the most part rural. In 1800 it is rural to urban areas. As a minimum, it has become
estimated that there were only about 750 places with necessary to be able to estimate and to projedt, in each
5,000 or more inhabitants in the world, and that these country, the total urban and rural populations. For
places contained only 3 per cent of the world's popula- many purposes there is the further need to anticipate
tion. The number of cities with 100,000 or more what the sex and age compositions of urban and rural
inhabitants in 1800 may have been only 45. By contrast, populations will be, as these factors affect greatly such
a recent compilation lists 1,777 places with 100,000 or things as the need for schools and services for children,
more inhabitants in the world in 1970. The percentage jobs, housing, medical facilities and so on for the working-
of the world's population which is now urban is approxi- age population; the need for special services and facili-
mately 37, and by the end of this century the world is ties for the elderly; and many other important necessities
expected to be at least half urban. Moreover, the of various age groups. In this context, Shryock and
absolute quantities of both urban and rural populations Siegel have suggested that in many instances it would
have been swelled by the rapid rate of total population often be sufficient to project urban and rural population
increase since 1800. It has become evident that rising at least in the following age categories: under 15, 15-44,
levels of urbanization pose increasing problems in the 45-64, and 65 and over,8 identifying approximately the
school population, the child-bearing 'population; the
Henry S. Shryock and Jacob S. Siege], The Materials and potential labour force population and the elderly popula-
Methods of Demography (Washington, D.C., United States Bureau tion. They took note that projections have often been
of the Census, 1971), vol. I, p. 151. prepared in considerable age detail (usually in five-year
a see his he Growth of Cities in the Nineteenth Century (Ithaca, intervals), when the quality of the data available does not
New York, Cornell University Press, 1963). Originally published in permit an accurate projection in great detail, and the
1899 for Columbia University by the Macmillan Company, New
York, as volume XI of Studies in History, Economics and Public Law. intended users the projections may require it.
See, for example, Wolf Schneider, Babylon is Everywhere: However, where suitable data are available, it may
The City as Man's Fate, translated from the German by Ingeborg often be useful to calculate the projection in greater
Sammet and John Oldenburg (New York, McGraw-Hill Book
Company, Inc.), published in Germany under the title uberall ist detail than is intended for the purposes of an efficient
Babylon by Econ Verlag G.m.b.H.; L. Hilberseimer, The Nature of presentation of results. In some instances, results of a
Cities (Chicago, Paul Theobald and Co., 1955); and Gideon Sjoberg, projection by five-year age groups may have to be further
The Preindustrial City (Glencoe, Illinois, The Free Press, 1960). interpolated with respect to single years relevant for
See estimates of Kingsley Davis and Hilda Hertz, as published instance to school enrolment, voting rights or old-age
in Philip M. Hauser, ed., Urbanization in Asia and the Far East
(Calcutta, UNESCO Research Centre on the Social Implications pensions. Likewise, there may be a need for projections
of Industrialization in Southern Asia, 1958), pp. 56-57. to be presented for individual calendar-year intervals,
Kingsley Davis, World Urbanization 1950-1970: Analysis of though the projection was originally calculated by time
Trends, Relationships, and Development, vol. 11, Population Mono- intervals of five years. Again, the required results may
graph Series No. 9 (University of California, Berkeley, 1972). be obtained by interpolation with respect to time.
Monthly Bulletin of Statistics, November I971 (United Nations
publication), p. xxxvi. The following can be found in this publica- Educational, occupational, residential and public-
tion for all countries of the world for dates beginning with 1960 service requirements are usually quite different in urban
and projected to 1985: (a) estimates of urban and rural population; as contrasted with rural areas, on the one hand because of
(b) percentage of urban population; (c) annual rates of growth of the differencesin physical, as well as economic and social,
urban and rural population. "National definitions" were used in
these estimates and projections, i.e., the definition accepted by environment; and on the other hand because of different
official statisticians within each country. sex and age compositions and different population distri-
'Growth of the World's Urban and Rural Population, 1920-2000
(United Nations publication, Sales No. E.69.XIII.3). p. 71. 8 Shryock and Siegel, op. cit., vol. 11, p. 843.
butions in space. These factors give cause to different units whichcan come into use indifferentcountries, aUnited
types of investment with different amounts of expenditure. Nations manual cannot describe all the combinations
For example, only minimum fire and police protection which might become relevant in the particular instances.
are offered in rural areas, and the per capita quantity and The reader may nevertheless observe that the methods
organization of medical facilities often has to be quite described herein can be adapted to many special purposes.
different between rural and urban areas. The occupa-
tional or industrial composition of the labour force, of ELEMENTS
OF PROJECTION : STRUCTURE AND
course, differs immensely between the two areas of resi- TREND COMPONENTS
dence.
Urban population projections are one variety of sub- There has been much hesitation in the calculation of
national population projections. The general importance urban and rural population projections by simple methods,
of a number of types of subnational projections was especially because they do not usually yield directly the
recognized at a conference devoted to that subject held corresponding sex-age structures, for which there is also
at Bangkok in 1969, but provisions for implementing much interest. It will be demonstrated in this manual
such projections by means of a demographically trained that simple methods are quite adequate to the estimation
staff still appear to be scant. Several of the methods of sex and age composition in projected urban and rural
here discussed may also be applicable in the population population totals. For good reasons, the use of simple
projections for regions, provinces or districts within a methods deserves encouragement, at least for simple
country. Furthermore, in some of the most urbanized forecasting purposes.
countries with highly developed transportation facilities, In this manual it will be assumed that methods of
the traditional distinction between urban and rural local- projecting national populations by groups of sex and age
ities has lost much of its relevance in describing salient are already known, and that such projections have in
economic and social features and needs. In some of these fact been carried out. The task, then, is that of projecting
countries, statistics are now being collected, distinguishing the urban and rural segments of a population with refer-
city-dominated regions such as metropolitan areaslo ence to the national projections already calculated. Only
within which there can be a further differentiation between in the last chapter will urban and rural population
the core city and the suburbs and satellite cities within projections be considered, from which a national projec-
the heavily urbanized periphery. Such areas will often tion is subsequently derived as the sum of the two.
include some rural population devoted to agriculture Elsewhere it is a matter of projecting only the urban (or
primarily for local metropolitan consumption. The rural) population and obtaining the corresponding rural
projection techniques described herein for urban and rural (or urban) population as a result of subtraction from
populations may often be equally useful in the projection projected national totals.
of metropolitan and non-metropolitan populations.
Within an individual country, innumerable combina- As distinct from most national projections, migratory
tions of subnational projections are possible. For transfers are a major factor affecting both urban and rural
instance, there might be lower-level projections of urban population structures and growth. Additional compli-
and rural population within major administrative terri- cations result from the fact that net migration is a balance
torial units or within territorial units defined by other of in-migration and out-migration, in each of which the
criteria such as communications linkages or population trend and composition can vary. Other rural-to-urban
density. There might also be interest in some countries population transfers result without migration where
in the further classification of urban and rural populations previously rural territory is reclassified as urban, whether
by ethnic composition. In highly urbanized countries, for administrative or other reasons. Births and deaths
there may be good reason to project rural population or occur among migrants, and in reclassified areas, and the
small-town population in two categories: (a) urban and latter areas are also affected by migration. Methods
rural population within the large metropolitan regions, taking all these involved demographic factors into account
and (b) urban and rural population outside those metro- can become unduly complex, hence the need for more
politan regions. l1 Because of the great diversity of area summary methods.
As has just been mentioned, urban and rural population
See Projections of Populations of Sub-National Areas: Report changes are heavily affected by migration between these
of a Working Group (Bangkok, 1969) (E/CN.11/897). areas of residence, and such migration can fluctuate
lo For example, in 1960 nearly two-thirds of the population of the widely and rather unpredictably in both volume and
coterminous United States lived within metropolitan areas, and structure. In the resulting projections the element of
93 per cent of the total population lived within 100 miles of a metro- uncertainty can be so great that the numerical results are
politan area. The latter included 96 per cent of the urban, 88 per
cent of the rural-non-farm, and 82 per cent of the rural-farm of a low order of reliability. "High" and "low" alter-
population. See Dale E. Hathaway, J. Allan Beegle and W. Keith natives, accordingly, should also be calculated and
Bryant, People of Rural America, a 1960 census monograph (Wash- should be set rather far apart. Since the margins of
ington, D.C. United States Bureau of the Census, 1968), p. 35. error will hardly be reduced by the employment of
l1 Such a classification scheme was employed in a study of small
towns in the United States (in this case, incorporated places with
less than 10,000 inhabitants) during the period 1940-1960. Within by small towns in other parts of metropolitan areas, and finally
each region, small towns were classified by size of place and by by towns in other non-metropolitan parts of the country. See
location with respect to major cities. Is was found that within each Glen V. Fugitt and Donald W. Thomas, "Small-town growth in the
region, small towns in each size of place group grew most rapidly if United States: An analysis by size class and by place", Demog-
located within "urbanized areas" (as defined in the census), followed raphy (Chicago), 1966, vol. 3, No. 2, especially p. 525.
elaborate methods, forecasts by simple methods economic and social action, hence a variety of economic
are often quite sufficient. The frank use of simple and social policies may be contemplated in terms of their
methods has the further advantage that no illusion is possible demographic consequences.
generated as though the forecast could be precise. In All projections which assume some continuation of
fact, in many countries the use of elaborate methods is past demographic trends, irrespective of any influence
unwarranted because of the almost complete lack of which other factors might exert on these trends, are also
detailed statistics bearing on the migration between urban called "autonomous" projections. In contrast with
and rural areas. these, there is one variety of theoretical population
The matter is different, of course, when the projection projections, namely the so-called "normative" projec-
is theoretical. Because of the nature of the arguments tions, in which it is assumed that an already approved
involved, precise (but not necessarily realistic) assump- programme of national economic and social plans will
tions are then needed concerning each factor of urban and be carried out with precision. The demographic conse-
rural population change: fertility, mortality, migration quences of the programme are then incorporated in the
and perhaps also the reclassification of previously rural population projection. As the programme progresses,
areas as urban. The absolute numbers resulting in such of course, targets may have to be modified on the basis
a projection are of less significance, since attention is of practical experience, and new demographic projections
focused on the comparative results of alternatives. may then have to be computed in which the change of
Uncertain or missing statistics may be substituted by targets is also reflected. This manual is mostly concerned
reasoned assumptions, but the method of calculation with "autonomous" projections.
must reflect the comparative dynamics of the particular Ideally, demographic, economic and social projections
factors under consideration. can be merged into one combined system where by
means of computerized multiple feedbacks all possible
PROJECTIONS
SERVING AS FORECASTS AND interactions, causal as well as consequential, between
PROJECTIONS OF MODELS modifiable demographic, economic and social conditions
are dealt with simultaneously. For practical purposes,
Population projections can be made for two kinds of however, such a system will usually be too cumbersome
purposes which are logically distinct. The purpose of and too unreliable, although it can constitute a precious
most population projections is to provide, however conceptual exercise. In practice the nature of most of
roughly, a forecast. Such projections are calculated on the interactions cannot be measured in isolation, and
the assumption of some reasonable continuation or many missing statistical data will have to be substituted
modification of observed population trends, to indicate by guess-work whose combined errors can have cumulative
the approximate future population magnitudes to be effect. Because of those practical limitations of a more
taken into account so that human needs may be met in complete system, most population projections will remain
the various fields of economic and social policy where of the forecasting variety, while some may be specialized
they have relevance. Implied in such projections is for the illustration of possible demographic consequences
the assumption that population trends may change of selected lines of economic and social action.
gradually but, barring catastrophic developments, not
abruptly. Catastrophic future events are of course Whether serving as forecasts or for purposes of theore-
possible, but cannot be reasonably foreseen in ongoing tical debate, the projections should be prepared in more
policies and plans. Reasonable forecasts can orient the than one alternative. The "medium" or, with momenta-
targets to be included in rational economic and social rily available knowledge, "best" forecast will probably
programmes and can sometimes also indicate priorities err, and for purposes of safe planning the likely magnitude
among those programmes. of the error will also have to be appreciated. This can
be done by supplementing additional "high" and "low"
Some projections, on the other hand, are of a more forecasts whose possible future realization should not
theoretical nature, and these are often referred to as be cause for surprise. The theoretical projections will
"models". These provide material for the discussion of necessarily be calculated in several alternatives specified
the possible demographic consequences which might according to the nature of the argument for which illustra-
result from particular economic and social policies among tion is being sought.
which there is a choice. They may assist in arguing that
the demographic consequences of one line of policy STRICTLYDEMOGRAPHIC PROJECTION TECHNIQUES
might be preferable to those of another, providing a
scenario from which policy strategy can be debated, In the projection of urban and rural population, four
possibly leading to the selection, among alternatives, of types of methods can be distinguished, depending on the
the policy to be adopted. detail which enters the calculation, namely:
Sometimes one set of projections is used for both (a) Global methods; in these neither the sex-age compo-
purposes, as forecasts and as theoretical models, but care sition of the population nor the component trends of
must then be taken to avoid fallacious reasoning because urban and rural growth (fertility, mortality, migration and
the nature of the argument differs in the two instances. area reclassification) form part of the basic calculation;
In the forecast, the population trend is the independent (b) Composite methods: here, the calculation considers
variable from which certain economic and-social impli- the sex-age composition of the urban and rural popula-
cations may follow. In the theoretical projection, the tion, but not the separate effects of component trends
population trends are assumed to be modifiable through (fertility, mortality, migration and area reclassification);
 (c) Crude component methods: urban and rural trend but the detailed growth components are disregarded.
components (fertility, mortality, migration and area For instance, there is what one might call the "migration-
reclassification) are considered separately, but not the survival method", most suitable if at least two censuses
sex-age composition of the population; and have been taken at an interval of five or ten years. In
( d ) Cohort-survival methods, where use is made of the the rural population, identical cohorts change from one
sex-age detail of the urban and rural population and the date to the other under the combined influence of mortality
incidence of fertility, mortality and migration (and and net migration (including area reclassification), and
possibly area reclassification) by groups of sex and age. one may project on the assumption that the combined
effect of these two components (though they are not
The methods most often used are either of type (a) or
distinguished) may remain the same; as for children, one
type (d), sometimes described as "mathematical" and may assume that child-woman ratios remain the same
"demographic" methods, respectively. If so described,
(combined effect of fertility, mortality and migration).
the methods of types (b) and (c) can be regarded as The urban population can be obtained by subtraction
"mixedy'. To serve as forecasts, methods of types (a),
of the projected rural from the projected total population.
(b) and (c) may be adequate, being comparatively simple.
But if a theoretical model is required, the method will OTHERTECHNIQUES RELEVANT TO THE PROJECTION
have to be of type (d), for only then can the precise OF URBAN AND RURAL POPULATION
effects of alternatives in fertility, mortality and migra-
tion trends upon the size and structure of the resulting Except for the last chapter, where the comparison of
populations be followed through. l2 theoretical projections is envisaged, this manual is confined
to "autonomous" demographic techniques. In these it is
The use of a logistic transformation is an important assumed that urban and rural populations will continue
feature of this manual. In some of the methods of type to grow and change on the basis of an inherent demog-
(a) and (b), a logistic progression in the percentage of raphic momentum as measured in actual observations.
urban population is used in the projections. In direct Unless a large amount of geographic detail is sought, the
calculations of percentage growth over time, it is expected corresponding calculations are not very complex and can
that relative growth will be uneven for equal time periods: be calculated with simple mechanical equipment. As
slow at first, then more rapid and finally slower. In the manual is intended largely (though not entirely)
other words, the rise in the percentage over time follows for application in less developed countries, where electro-
an S-shaped curve. However, it is difficult to determine nic computer facilities are not always available, it has not
systematically what rise in percentage should be expected been carried to an advanced level of complexity. True,
in any given time period. With the logistic transfor- the methods here shown, if they are to be applied to
mation of percentage figures, a reasonable pattern of numerous subcategories (e.g. urban population in each
percentage growth can be projected by assuming equal of numerous provinces and so on) may be advantageously
amounts of change in the transformed logistic values. adapted to computer programmes.
It will be noted that only methods (b) and (d) result It is fully recognized, however, that methods of project-
directly in projections of population detailed by groups ing urban and rural population are not thereby exhausted.
of sex and age. However, auxiliary methods can be Depending on special conditions, other methods such as
employed, with fairly adequate results, permitting the those listed below may sometimes be more advantageous.
estimation of the sex-age detail in projections calculated It is certainly true that urbanization is interrelated with
by methods (a) and (c). l3 As illustrated in chapter VIII social and economic factors, though these can be numer-
of this manual, there are also some useful methods in ous and also interrelated with each other.
which age groups of urban and rural population are used In special instances, population growth can be closely
la For discussion of a variety of types of demographic techniques
dependent on one predominant function and should
currently used in urban and rural projections, see the following: therefore be projected in relation to that function. This
Shryock and Siegel, op. cit., vol. 11, pp. 843-846; Jacob S. Siegel, may be especially the case with residential suburbs where
"Some principles and methods of projections of urban-rural a given amount of housing capacity can be foreseen and
population by age and sex" in Proceedings of the World Population where it is most reasonable to link future numbers of the
Conference 1965 (United Nations publication, Sales No. 66. XIII.7),
vol. 111, pp. 91-96; and Statistical Commission and Economic population to the foreseen dwelling space. Similarly
Commission for Europe, Conference of European Statisticians, the population of a new mining town or town dominated
Methods and Uses of Projections of Urban and Rural Populations in by one major industry may develop in proportion to
European Countries (Conf. Eur. StatslWG.3514). foreseen employment in that basic industry; it is then
la John V. Grauman, "Population estimates and projections", in reasonable to multiply the foreseen basic employment with
P. M. Hauser and 0. D. Duncan, eds., The Study of Population
(Chicago, University of Chicago Press, 1959), pp. 565-569;Population additional employments in complementary services and
Growth and Manpower in the Philippines, 1960 (United Nations numbers of dependents in relation to the combined
publication, Sales No. 61 .XIII.2); R a d Benitez Zenteno and Gustavo numbers of workers who presumably will be employed.
Cabrera Acevedo, Proyecciones de Iapoblacidn de Mkxico 1960-1980 At the national level or in large and industrially diversified
(Mexico D.F., Banco de Mkxico, 1966); Alberto Cataldi D.,
La situacidn demogrhjica del Uruguay en 1957 y proyecciones a cities, urban populations will also be influenced by
1982 (Santiago de Chile, Nationes Unidas Centro Latinoamericano economic and social changes. But the large city or the
de Demografia, 1964) (E/CN.CELADE/C.l5 B.62.3.111) (series C); combined urban population of the entire country exercise
and Human Resources of Central America, Panama and Mexico, more numerous functions than can easily be taken into
1950-1980, in Relation to some Aspects of Economic Development
(United Nations publication, Sales No. 60.XIII.l), pp. 39 and account in a simple calculation. A simplification is then
94-95. , needed to establish a possible relationship between
anticipated national development and the corresponding It must be noted, however, that even the highest of
urbanization trends. these correlations, .87, explains only about three-fourths
of the variance, and .70 would explain only about half
Paired projections of the variance. Moreover, the variables correlated with
In instances where a projection exists of a variable urbanization are often more highly correlated with one
known to be correlated with urbanization, projections can another than with urbanization, l7 SO that rather little
be made of future urban and rural population by assum- additional explained variance should be expected from a
ing some constant multiplier or difference relationship multiple regression approach.
between the level of urbanization and the correlated Even though paired projections alone should be used
variable. l4 Similar results could be obtained by using with caution, they can still serve a useful function by
regression equations. l6 The accuracy of such population providing a check on autonomous demographic projec-
projections depends on two factors: tions for what can happen at least demographically may
(a) The accuracy of the projection of the correlated be constrained by environmental limitations, both eco-
variable. Here it must be noted that although accurate nomic and spatial. It would be desirable from the point
population projections are difficult to make, the future of view of economic and social as well as demographic
behaviour of correlated economic variables may be even projections to assemble and analyse the entire correlation
more uncertain. Population growth and distribution matrix of the many interdependent variables so that the
tend to be more constrained by biological and traditional reactions of the individual variables and combinations
factors which change rather slowly by comparison with of variables upon each other could be evaluated.
the more volatile economic conditions. Computerized projections
(b) The degree of correlation between urbanization and
the paired variable. Correlations with urbanization Economy-based population projections can be calcul-
ranging between .70 and .87 have been observed in various ated without explicit consideration of probable changes in
studies for the following variables: males in non-agricul- demographic trend components, provided that the
turd activities, males in transport and communications, assumption can safely be made that the population
life expectancy, fertility, school enrolment, per capita required for urban economic growth will be drawn to
consumption of energy from commercial sources, motor urban areas without exhausting its source of supply in
vehicles registered per capita, daily newspaper circulation the rural areas. However, changes in fertility and mortal-
per 1,000 inhabitants. le ity are likely to modify the ratio of dependants in each
household. Also female labour fource participation may
l4 Probably the variable most often paired with the level of
be affected. le
urbanization is the proportion of the labour force in agriculture. In projections of urban and rural population for
In this approach, the rural population is implicitly estimated from numerous regions in the Soviet Union, explicit account
the agricultural labour force and the urban population is obtained
as a residual. For instance, in certain projections of urban and was taken of fertility and mortality which were considered
rural population in several Latin America countries a constant separately for urban and rural areas. The models were
absolute difference was assumed between the proportion of the total based on a projection technique which has been in use
labour force in agriculture and the proportion of total population in the Soviet Union since the first five-year economic
in rural areas. This assumption was based on an examination of a
time series of United States census data which showed that, although development plan. The volume of migration was
the proportion of rural had fallen from 92.8 per cent in 1820 to determined by the planning agencies on the basis of
43.8 per cent in 1930 and the porportion of labour force in agricul- plans for the distribution of productive forces and man-
ture had fallen from 71.8 per cent to 21.4 per cent, still the difference power. l9 In its present detail, the method is very
between the two variables at each decade had remained almost
constant at about 21 or 22 percentage points. See Human Resources labour-consuming. For example, to estimate the popula-
of Central America, Panama and Mexico, 1950-1980, in Relation to tion of a particular territorial unit twenty years ahead
Some Aspects of Economic Development (United Nations publication, under such headings as urban and rural population, male
Sales No. 60.XIII.l), p. 37. Again in a projection of urban and and female population and so on, taking migration into
rural population in the Philippines, it was assumed that the propor- account, it is necessary, with 2,000 items of basic data, to
tion of the population living in rural areas would decrease by the
same ratio (from 65 per cent in 1957 to 56.5 per cent in 1977) as the perform over 100,000 calculations. 20 However, with
decrease in labour force engaged in agriculture and related activi- the use of computers such methods can now be used with
ties (from 59 per cent in 1957 to 51 per cent in 1977). See Population considerable detail, which was not possible previously.
Growth and Manpower in the Philippines (United Nations publica-
tion, Sales No. 61 .XIII.2), p. 12.
l7 Population Bulletin of the United Nations, No. 7 (United Nations
l6 See, for example, regression techniques for projecting eco-
publication, Sales No. 64.XIII.2), pp. 145-147.
nomically active population used in various studies described in
Manual V: Methods of Projecting the Economically Active Population Is John Stuart MacDonald, "Anticipating city growth and popula-
(United Nations publication, Sales No. E.70.XIII.2), pp. 22-34. tion projections for urban development planning", in Proceedings
But the regression approach has seldom been used so far to project of the World Population Conference 1965, vol. I11 (United Nations
urban and rural population. population, Sales No. 66.XI11.7), pp. 72-73.
Urbanization: Development Policies and Planning, International lo P. G. Podyachikh, "Population projections in which allowance

Social Development Review, No. 1 (United Nations publication, is made for migration", in Proceedings of the World Population
Sales No. E.68.IV.l), p. 27; Kingsley Davis and Hilda Hertz Conference 1965, vol. I11 (United Nations publication, Sales
Golden, "Urbanization and the development of pre-industrial No. 66.X111.7), pp. 83-90; and A. F. Pobedina, "The use of electro-
areas", Economic Development and Cultural Change (Chicago), nic computers for population projections", in Proceedings of the
vol. 111, No. 1, October 1954, pp. 8 and 16; Leo F. Schnore, "The World Population Conference 1965, vol. I11 (United Nations publica-
statistical measurement of urbanization and economic development", tion, Sales No. 66.XIII.7), pp. 34-39.
Land Economics (Madison, Wisconsin), August 1961, p. 234. a. A. F. Pobedina, op. cit.
LCon Tabah has described a computer model based on uses of the land exceed the expected returns to urban
survival ratios which produces simultaneously projections uses. On the other hand, where cities grow by merger
with cross-classifications of population by- sex, age, with other cities, there may remain pockets of agricultural
economically active and inactive, and urban and rural land area within the legal city limits. For this reason,
areas. 21 the United States Bureau of the Census has designated
certain cities as "extended cities", and in the 1970 census
Geographic projections figures both an urban and a rural population are shown
Anticipated area reclassifications can become difficult for such cities. 25 Since conditions in different places
to introduce into the calculations of certain urban popula- may vary greatly, no generalized methods can as yet be
tion projections. At least one detailed attempt has been proposed. 26
made by Jerome P. Pickard, using his own definition of
"urban regions" which are subject to geographic expan-
sion. The definitions include the rural population
expected to fall within the transportation, communica- This manual, being developed by the United Nations
tions and market networks of the expanding city-dominat- Secretariat, naturally favours those methods of projection
ed areas. Pickard projects not only future population of of urban and rural population which it has found most
urban regions, but also the number of square miles that useful in some of its own work. For its own purposes,
will be contained within these areas at future dates. A the United Nations requires methods having widest
map area computer was used in this work as well as international applicability, including those countries for
census map records and enumeration books for the which statistical documentation is only scant. These
United States going back to 1920. 22 methods may be too crude for some countries and too
Even without detailed geographic information, it is refined for others since available statistics and actual
possible to arrive at approximations of the amount of conditions vary widely from one country to another.
land which will come under urban settlement in the The user of this manual is encouraged to make his own
future. The simplest approach would be to project adaptations, which, according to judgment, may seem
the future density of total urban population or of a indicated in the case of his particular country. While,
particular city and to calculate the amount of land that in these methods, extensive use is made of a logistic-scale
would be necessary to accommodate the projected urban curve (tablulated in annex I), the use can be flexible,
population at the projected density. 25 In places where permitting accelerations and slow-downs which might
improvements in transportation are expected, it is often sometimes be reasonably expected. Reasons for the use
reasonable to assume a future decrease in density. " of such a curve are discussed in chapter 111. No less
important than the particular schemes of calculation are
However, in cases where heavy in-migration is expected
the density may increase. There is also the possibility problems of definition and formulation of suitable assump-
that spatial merging of two or more formerly separate tions. The first three chapters accordingly deal with the
urban areas will alter the average density. There could clarification of some basic concepts.
even be an agricultural limitation if the urban agglomera- Simple methods for projections of urban and rural
tion spreads so far that it becomes difficult for the inner populations (relative to national population totals) are
city population to be supplied with perishable foodstuffs discussed in chapter IV; the United Nations method is
and consequently the expected returns to agricultural introduced in chapter V; and its extension to mutually
consistent projections for individual cities is illustrated
21 U o n Tabah, "Reprbentations matricielles de perspectives de in chapter VI. The simple application of the United
population active", Population (Paris), May-June 1968, pp. 437-476. Nations method is facilitated by the logistic model
Pickard's most extensive study is described in Dimensions of tabulated in annex I.
Metropolitanism, research monograph 14 (Washington, D.C.,
Urban Land Institute, 1967). Other published studies by Pickard All these are global methods of type (a), but they need
can be found in: "Future growth of major U.S. urban regions", not be limited to population totals. The consistent
Urban Land (Washington, D.C.), February 1967; "U.S. urban estimation of sex-age composition is always possible,
regions: Growth and migration patterns", Urban Land (Washington,
D.C.), May 1966; "Urban regions of the United States", Urban whether by conventional methods (chapter VII, section A)
Land, (Washington, D.C.), April 1962; and Metropolitanization of or by an extension of the United Nations method (chap-
the United States, research monograph 2 (Washington, D.C., ter VII, section B); methods of types (b) and ( c ) may
Urban Land Institute, 1959).
Such an approach was taken in Department of the Environment,
Long-Term Population Distribution in Great Britain-A Study: United States Bureau of the Census, 1970 Census of Population:
Report by an Interdepartmental Study Group (London, Her Majesty's Number of Inhabitants: United States Summary (December 1971),
Stationary Office, 1971). p. ix (preliminary report).
~4 Kingsley Davis, "Conceptual aspects of urban projections in Kingsley Davis lists several types of criteria which could be
developing countries", in United Nations, Proceedings of the World used in evaluating which area units should be included within the
Population Conference 1965 (United Nations publication, Sales boundaries of metropolitan areas. See his "Conceptual aspects
No. 66.XIII.7), vol. 111, pp. 61-65. Davis found, in a long-run of urban projections in developing countries" in Proceedings of the
study of the San Francisco Bay Area, that the population rose World Population Conference 1965 (United Nations publication,
from 364,000 in 1890 to 3,217,000 in 1960, but the over-all density Sales No. 66.XIII.7), vol. 111, pp. 61-65. Even more numerous
declined from 5,643 to 2,501 per square mile, because the territory criteria have been used on occasion in the efforts of the United States
expanded at a faster pace than the population. Reference is made Bureau of the Census to delineate the contours of metropolitan
in the article to a book which he published with Eleanor Langlois, areas. See United States Bureau of the Census, United States
entitled Future Demographic Growth of the San Francisco Bay Area Census of Population: 1960 (Washington, D.C., United States
(Berkeleys University of California, 1963). Government Printing Office, 1961), vol. 1, part A, p. xxiv.
nevertheless often be preferable, and these are reviewed nary, those readers who need only a simple method for
briefly in chapter VIII. projecting urban and rural totals which yields at least
Chapter IX, finally, introduces the more detailed model reasonably useful results under a wide variety of circum-
of urban and rural population change by demographic stances may wish to move directly to chapter V. Chap-
components corresponding to the method of type (d), ter VII explains simple methods for adding sex and age
which may not be necessary for population forecasts, but detail to these projections. Chapter VI tells how indivi-
is indispensable for comparative theoretical projections. dual cities may be projected. Both chapters IV and VI
A table of survival ratios (Pz)of model life tables for are relevant also in the case of a small country most or all
five-year age groups and five-year intervals of time is of whose urban population is that of one city. Finally,
provided in annex I1 for use in this model. those readers who require component methods of varying
The manual is designed for a variety of needs. As the complexity may move directly to the last two chapters
discussions in chapters I through IV are largely prelimi- of the manual.
 Chapter I
PROBLEMS OF DEFINITION OF URBAN POPULATION

1. Before proceeding to a projection, it is necessary to tive boundary, such areas as may be called suburbs,
examine the definitions according to which data on possibly including administratively separate towns which
"urban" and "rural" population have been determined. have come to form conurbations. The administrative
Under some definitions, for instance, "urban" areas area can be said to be "over-bounded" when it is more
remain within constant geographic boundaries, and extensive than the dense13 built-up area and includes also
"urban" population, so defined, can grow by births, some population settled at much lower densities. The
deaths and migration only. According to some other administratively "urban" population can thus be smaller
definitions, the geographic extent of "urban" areas can than the population inhabiting urbanized terrain, and in
expand continuously; in such a case, the "urban" popula- some instances it can also be larger. The extent of
tion grows by area reclassifications as well as by births, coincidence between the two depends largely on adminis-
deaths and migration. trative flexibility and on the speed with which built-up
2. Definitions of "urban" localities vary from country areas have expanded geographically. The administra-
to country, and within the same countries from time to tive reforms in Japan in 1953, for instance, have had the
time. In some countries, moreover, two or more defi- effect that most cities, which previously were rather
nitions are maintained side by side. Urbanization being "under-bounded", suddenly became "over-bounded"
both a quantitative and qualitative process, different with reference to contiguous built-up areas. As a conse-
criteria of "urbanism" gain or lose relevance as time quence, 37.5 per cent of the population were classified as
progresses. "urban" in the 1950 census, as against 56.3 per cent only
five years later, at the census taken in 1955. While
3. A bewildering variety of criteria have been used, undoubtedly the urban settlements gained population
singly or in combination, to define "urban" localities. 27 rapidly in that period, the apparent growth would not
Essentially, the definitions are mostly of the following have been so extraordinary if the administrative areas had
types: administrative, economic and geographic. The remained the same, or if "urban" population had been
type of definition must be ascertained when a projection defined on the two occasions in geographic or other
is made since it determines whether or to what extent terms. 28
"urban" populations can also grow through reclassifica-
tion of areas previously defined as "rural". Different 6. Countries vary considerably in the territorial
types of definition, moreover, can be suitable according flexibility of administrative areas. In some countries,
to the purposes which the forecast is intended to serve. local administrative boundary adjustments are made very
rarely if at all, in others they happen from time to time,
and in still others they occur with great frequency. This
ADMINISTRATIVE
DEFINITION :
concerns both the recognition of additional areas as
TYPE OF LOCAL GOVERNMENT
"urban" (through establishment of a new local govern-
4. In many countries the minor territorial adminis- ment of an "urban" type) and the geographic expansion
trative units have local governing bodies of diverse forms of existing "urban" areas through annexation of surround-
or degrees of authority, making obvious the distinction ing terrain previously under "rural" forms of government.
of those which are of an "urban" type. The adminis- 7. Where administrative boundaries remain unchang-
trative boundaries of urban local governments then ed, the growth of population in densely built-up areas is
contain the population defined as "urban", all other areas not fully reflected by the increases noted in the population
being classed as "rural". The effects of such a definition of administratively "urban" areas. This is especially
vary, however, depending on the relationship between the case where, with urban growth, there is an increasing
administrative area and the area built up at an urban overspill of urbanized settlement beyond the admini-
density, and depending also on the frequency with which strative city limits, as happens generally where the city
administrative boundary adjustments may occur.
as The 1955 census of Japan included a table showing the 1950
5. The administrative area can be said to be "under- population within areas classified as "urban" in 1955. In all,
bounded" where there are additional, inmmediately 53.5 per cent of Japan's 1950 population inhabited the "urban"
adjacent, densely built-up areas outside the administra- areas of 1955. The comparison of figures suggests that between
1950 and 1955 as much as 16 per cent of the population of Japan
were reclassified from rural to urban, and that between 1950 and
For a brief survey of definitions of "urban" population in the 1955 the "urban" population ratio rose only slightly. However,
censuses of 123 countries taken around 1960, see Growth of the without doubt, effectively urbanized areas within the "urban"
World's Urban and Rural Population, 1920-2000 (United Nations areas of 1955 expanded significantly within the 1950-1955 period,
publication, Sales No. E.69.XIII.3), pp. 7-10. a circumstance not reflected in this comparison of percentages.
is "under-bounded". In manv countries there are now and the population of these cities or towns is considered
extensive urban agglomerations whose populations are as the "urban" population, the remainder being "rural".
multiples of those of the administrative central cities. 13. In actual fact, the seats of regional (district,
In some large cities, because of city-to-suburb migration, municipality, canton and so on) government are usually
the population of the agglomeration keeps growing even also the centres to which certain particular administrative
though the population within the administrative municipal measures apply, hence in at least that sense they have a
limits decreases. In the London region population different type of "local" government. This is typically
increases notably in areas situated outside the Green Belt, the case in Latin American countries. The effects of such
whereas in the agglomeration within the Green Belt it a definition can be similar to those already discussed in
has been decreasing for some time. the previous subsection, except under certain limiting
8. To a lesser extent this is also true of "over-bounded" conditions. It can fairly be assumed that, in view of the
cities within fixed limits. Here, the expansion of the degree of regional autonomy, the outer boundaries of the
inner agglomeration is associated with some displacement "urban" centres are usually adjusted to coincide approxi-
or absorption of a population living at hitherto rural mately with those of the built-up urban terrain.
densities. However, since the displaced or absorbed 14. Whereas the delimitation of such urban centres may
rural population (with its loher densities) is usually be flexible, that of the districts (of which they are the
comparatively small, the underestimation of rates of centres) is usually quite rigid. New "urban" centres will
urban growth in most such instances is only slight. rarely appear because the establishment of new districts
9. The estimation and projection of urban population probably occurs rather seldom. This despite the fact
trends as defined within unvaried territorial limits has that new compact settlements other than district centres
the advantage that change occurs only as a result of can also emerge. A new mining town, for instance, may
births, deaths and migration, but not from the reclassifica- fail to make its appearance among administrative urban
tion of previously "rural" areas. Numerous other settlements. Some urban centres, furthermore, may
statistics referring to such constant areas can often also grow so large that their physical limits encroach on the
be assembled. There is no break of continuity in the territory of neighbouring districts. Unless a territorial
statistical time series. It is convenient in many studies reorganization takes place, part of the growth of large
to compare population and other statistics which are cities will then fail to be reflected in the population trend
available for the same fixed area. of the "urban" centres of which one is assumed to lie
10. In a large country where local administrative within each of the districts.
boundaries are flexible and frequently adjusted, the trend
in "urban" population (administratively defined) may DEFINITION
BY SIZE OF ADMINISTRATIVE UNITS
closely parallel the trend in the population of physically
urbanized terrain. It can be assumed, in such a country, 15. In many countries most of the basic administrative
that administrative changes are made approximately in units are quite small, and there is no formal distinction
proportion to the geographic expansion of densely of those of an "urban" type, all local governments being
built-up areas. It can also be assumed that territorial similarly constituted (in communes, circumscriptions
adjustments will be made continuously in the future, and so on). This situation is found especially in some
having similar effect. Differences between administered rather densely populated European countries. Since the
"urban" terrain and the areas under dense settlement, in area units are mostly small, the presence of a sizable
such an instance, may remain unimportant or negligible. population in any one of them usually indicates the
For certain studies, however, the fact that continuous area presence of an urban cluster. Only in exceptional
changes are also involved must be borne in mind. circumstances would an entirely rural commune comprise
a large population. Moreover, where administrative
11. A problem arises especially in the instance of a areas are generally small, it can often happen that two or
country where administrative adjustments are made several neighbouring units are merged into one, permitting
only from time to time. For in this case the adjustments an expanding city to continue being administered as one
can be abrupt and discontinuous, and it is difficult to unit. Territorially large units may emerge in time
predict what other modifications of "urban" boundary representing in each instance an expanding major city.
will be made in the future, if any. Projections of adminis- Annexation of neighbouring communes into a city
tratively "urban" population can then be scarcely made usually occurs only after some time lag, surrounding urban
except on the uncertain assumption that administrative terrain or "suburbs" still being administered in the interim
limits will remain unchanged. In relation to such an as separate units; but these also, for the most part, would
assumption, the past trend of administratively "urban" be of greater population size than the more strictly rural
population would also have to be re-estimated in accord- units. The size of individual cities may often not be
ance with the most recent boundaries. sufficientlyreflected, but the size of the combined "urban"
population (including separate suburbs which are also
ADMINISTRATIVE DEFINITION :
"urban") can well be represented by the population sum
of all those units in which some minimum population size
SEATS OF DISTRICT GOVERNMENT
is exceeded.
12. In a number of countries basic administrative 16. The minimum size of units to be recognized as
units of one single type cover the entire territory. Each "urban", however, differs greatly between countries,
unit is governed from one city or town situated within it, varying in extreme instances from as little as 100 inhabi-
tants to as much as 30,000. Often the limit is 2,000, contours of dense settlement, irrespective of adminis-
2,500 or 5,000. A uniform limit cannot be prescribed for trative deliminations. This concerns localities of every
all countries since typical qualitative features or residen- size and economic type. The "locality" may consist of
tial densities in settlements of a given size may differ a small grouping of rural houses (e.g. a hamlet as distinct
among the countries. A few European countries have from completely dispersed rural population) and, at the
adopted a definition by which units with 10,000 or more other extreme, of an urbanized region comprising several
inhabitants are considered as "urban", those with administrative cities, assuming that the urbanized areas
2,000-9,999 inhabitants as "semi-urban", and those with have coalesced. Among localities so delimited one may
less than 2,000 as "rural". select those above a certain size, or those having mostly
17. The trend of "urban" population defined in this non-agricultural activities, to define "urban" agglomera-
fashion may well parallel that of population in built-up, tions. It is evident that agglomerations expand geo-
urbanized areas. Part of the "urban" population growth graphically whenever there is building activity at their
will result from the surpassing of the minimum "urban" periphery. Upon occasion, two or several neighbouring
population size by additional administrative units, as a agglomerations merge into one.
result of which those units are reclassified. But the 22. In some countries the delimination of agglomera-
growth of individual big cities may be inadequately tions on the occasion of a census is made on the basis
reflected so long as many suburbs continue being adminis- of detailed map work including, as in Sweden, the meas-
tered separately. urement of distance between individual houses to deter-
mine which of them should be considered as included in
ECONOMIC DEFINITION, APPLIED TO the same agglomeration. But simpler determinations
ADMINISTRATIVE UNITS can be made, for instance, by using as units the groups
of contiguous census enumeration districts in which a
18. In some countries, where basic administrative units minimal population density is exceeded. The boundaries
are small and uniform, "urban" administrative units are of the agglomerations may then not be determined with
defined as those in which at the most a small percentage- geographic precision but nevertheless accurately enough
less than some stated maximum-of the economically for demographic studies. In the United States, "urban-
active population (or active males) are engaged in agri- ized areas" are established at each census, adjacent to
culture. There are also some countries where units are administrative cities of at least 50,000 inhabitants. In
defined as "urban" by combined criteria of size and type Japan, any group of contiguous census enumeration
of economic activity. districts having more than a certain high popuIation
19. Such a definition is of value particularly in those density and a combined population of at least 5,000
countries where some agricultural villages are apt to inhabitants are referred to as "densely inhabited districts".
grow larger than some small towns in which urban It is possible of course that both in Japan and in the
features predominate (e.g. mining towns, small suburbs, United States densely urbanized areas smaller than entire
local trading centres). In such countries, evidently, a census districts are not included, hence that the urbanized
definition based on an economic criterion has greater area is somewhat under-bounded. Other methods for
relevance than one based on population size only. an approximate delimitation of agglomerations have
Though it is recognized that non-agricultural activities come into use elsewhere, such as the multicommunal
exist in areas settled in a rural fashion, and that there can agglomerations which have been defined in France.
also be a small amount of agriculture among urban 23. Speaking somewhat schematically, one may say
residents, the coincidence of a prevalence of agriculture that a major agglomeration, or "urbanized area", ordi-
with other features describing a rural environment narily consists of a central city (i.e. an administrative
usually remains considerable. urban unit) together with densely urbanized "tentacles"
20. The effect of such a definition on population trends branching out alongside the major routes of travel,
in "urban" places is similar to that where the definition whether roads, rails or waterways. Such an urban
merely involves a size limit. Administrative units are "octopus" may with time absorb within it other urban
added to the "urban" category, in this case, when non- centres, e.g. satellite towns, or it may come in contact
agricultural activities in them come to outnumber the with one or several neighbouring "octopuses" thus
agricultural ones to the extent implicit in the definition. forming a conurbation. Between the extended "ten-
It is difficult, however, to assess the past and future trend tacles" there remain areas inhabited at lower densities
for "urban" areas so defined. Where it is a matter of which must still be regarded as rural. Geographic
size limit only, the population estimates themselves growth of the "octopus" occurs through a lengthening and
express whether an "urban" size has been or probably thickening of the "tentacles" and lateral outgrowths,
will be attained at given dates. It is not so easy to from them which, with time, form a tissue of lateral
estimate and project trends in the agricultural and non- connexions. Actually, of course, particular geographic
agricultural composition of the labour force in each features, if not also deliberate action, for instance the
individual locality. construction of a circular road around the city, will
prevent agglomerations from attaining precisely such a
schematic shape.
24. The noteworthy fact remains that the growth of
21. The internationally recommended definition of agglomerations is an accurate reflection of the areas
"localities" is that of population clusters within the inhabited under physical urban conditions, i.e. at urban
densities. The trends of growth include the transforma- ment are included where proximity to the city presumably
tion of previously "rural" into "urban" areas, hence they exerts a major influence on their types or methods of
imply area reclassifications. land use; hence it is not illogical to include them within
25. The analysis and projection of urban populations the wider urban periphery. In a few instances, groups
in terms of all agglomerations above some minimum of contiguous metropolitan areas have been described as
size has certain drawbacks, however. On the occasion a "megalopolis".
of each population census the areas have to be delimited 30. Again, the determination of metropolitan areas
afresh. It is not possible to follow their continuous is usually confined to major cities only (e.g. those whose
expansion through intercensal or postcensal time periods. centres have at least 50,000 or 100,000 inhabitants),
Nor is it possible, for areas whose limits differ from any leaving numerous separate towns which form no part of
administrative boundaries, and which vary continuously, the metropolitan areas. On the other hand, partly
to organize the collection of other statistics, such as because of actual internal economic interdependence and
those on births, deaths, school enrolment, retail trade, partly because of statistical convenience, metropolitan
motor vehicle licenses, and so forth. areas have often become official planning regions in
26. Since usually only major cities and towns are programmes of regional development.
demarcated in a census as agglomerations, small towns, 31. When compared with the corresponding agglomer-
though they may have markedly urban features, remain ations, the metropolitan areas can be considered as
omitted from this special "urban" category, unless they "over-bounded" urban entities. As has already been
are already absorbed in a major agglomeration. The noted, in "over-bounded" cities rates of city growth
combined population of the selected major urbanized are more adequately reflected than in "under-bounded"
areas is generally less than the country's combined ones, owing to the comparative smallness of the rural
"urban" population under a definition which can include population which becomes displaced or absorbed as a
far more numerous smaller cities and towns. result of the urban expansion. The extensive delimitation
makes it possible in many countries, where urban growth
is not too rapid, to maintain a constant demarcation of
GEOGRAPHIC DENNITION : METROPOLITAN AREAS metropolitan areas over extended periods of time.
Eventually, however, a redefinition of metropolitan
27. As has already been noted, administrative defi- areas can become necessary because of continued geo-
nitions often remain confined to constant areas, reflecting graphic expansion of the corresponding agglomerations
only part of the urban growth, with consequent insuffi- and a further widening of their spheres of influence.
ciencies in the measurement of the urban phenomenon. Where this happens, "geographic" methods of population
The definition of geographic agglomerations reflects the projection may have to be adopted.
expansion of urban environments quite realistically, but 32. Because of the emerging new settlement patterns
has the disadvantage that no statistics other than those around the cities of industrialized countries, it will
determined at each censuses can be collected for them, probably become increasingly useful to distinguish four
since their contours change continuously. Mainly to types of areas, namely urbanized and non-urbanized areas
circumvent the disadvantages inherent in both types of within metropolitan regions, and urbanized and non-
definition, another concept has been formulated, namely urbanized areas outside such regions. So far, few
that of metropolitan areas. censuses permit the distinction of areas according to
28. The metropolitan areas are usually conceived in such a four-fold classification.
such wide limits that they include all, or virtually all,
of each agglomeration "octopus" and, in addition, the
non-urbanized areas lying in the interstices between its
"tentacles". The boundaries are drawn in such a way 33. In some countries urban areas have been defined
that they coincide with the outer limits of a group of as administrative units presenting certain "urban features".
existing minor administrative areas which are unlikely Among such features there might be the network of paved
to change. These boundaries can, for a period of time, streets, numbered houses, streetlights, domestic electricity,
be left unaltered, and various types of statistics for each sewerage, mail delivery, the presence of a secondary
such group of administrative units can also be contin- school, a church, a medical establishment, a police station
uously secured. Reclassification of areas from "non- and so forth. The use of qualitative data describing
metropolitan" to "metropolitan", under these conditions, urbanism has much conceptual appeal since, according
need not occur for some time. Of course, the average to the conditions in a country, those might indeed appear
population density of metropolitan areas within their to be the most distinguishing features. On the other
wide limits, is much lower than that of the more strictly hand, in the context of trend study and projection, the
urbanized terrain. "urban" significance of any such features cannot be
29. The determination of administrative units to be considered as permanent. Whereas at one time electric
included in a metropolitan area often involves some quite light or police stations may be mostly confined to urban
precise criteria, such as the frequencies in travel and areas, eventually such features can also spread over the
communications to and from the central city, establishing countryside. The significance of an "urban" definition
a high degree of economic and social interdependence expressed in terms of such descriptive attributes can
with the city. Some areas under rural forms of settle- change with the passing of time.
 PURPOSES
SERVED BY VARIOUS DEFINITIONS For the formulation of social policies, therefore, it is
preferable to define "urban" populations as those con-
34. The projections of "urban" and "rural" population tained within "urban" administrative areas. Since the
may have to serve various planning and policy purposes, boundaries of these areas may or may not be flexible,
among yhich may be mentioned, very roughly: economic as has been explained, the implications for a population
plans, social measures and physical plans. projection can be various.
35. For the geographic disaggregation of economic 37. Phy6ical plans, finally, are much concerned with
plans, recourse is made increasingly to regional planning, environmental and traffic-flow management within areas
including those regions which comprise each of the major inhabited at an "urban" density. In such contexts,
cities. Such purposes are usually best served where the it is probably most useful to project the population of
estimates and projections are made in the rather extensive agglomerations (or "urbanized areas") as distinct from
terms of "metropolitan" and "non-metropolitan" popu- other areas of lesser population size or density. But
lation, preferably within constant boundaries. Internal more than one purpose may have to be served by the same
linkage by an interconnected system of transportation projections, and statistics on "urban" and "rural" popu-
can usually be assumed in such urban regions. Geo- lation are not usually available in every convenient form.
graphic priorities in particular investments can be Depending on the nature of available statistics, there
reasonably determined in such a regional context. are usually inevitable constraints to the definitions
36. Social policies depend for their implementation on which can practically be adopted for a population
existing organs of local government as described by law. projection.
 Chapter II
COMPONENTS OF URBAN AND RURAL POPULATION CHANGE

38. Even though it may be preferable to use only rural migrants, once settled in urban localities, prefer
crude methods in forecasts of urban and rural population, to remain there also at advanced ages.
some knowledge of the underlying dynamics will help 41. The ratio of males to females, of all ages, is much
in exercising good judgement. It is perhaps best to greater in urban than in rural areas in South Asia and
begin a discussion of components of population growth Africa, while the reverse is true in Northern America,
with a consideration of sex-age structures. The composi- Latin America and Oceania, and to a lesser extent also
tion of populations by sex and age is a consequence of in Europe. Below age 15, except those aged 5 to 14 in
preceding population trends (births, deaths and migra- Latin America, urban and rural sex ratios are nearly
tion). In its turn, this structure partly determines the same as might be expected since a majority of children
current and future demographic trends because each of of either sex will ordinarily reside with their parents. The
the trend components, births, deaths or migration, have high rural sex ratio in Northern America at ages 15 to
a varied incidence among population groups distinguished 24 may in part be attributed to the stationing of military
by sex and age. Reclassification is an additional factor troops in rural localities. But probably for different
which can affect the population composition. reasons the excess of males over females aged 15 to 24
in rural areas is also considerable in Latin Anierica and
DIFFERENCES
BETWEEN URBAN AND
Oceania. In South Asia and Africa, by contrast, there
RURAL SEX-AGE STRUCTURES
is a considerable excess of males in urban areas at ages
15 to 24, and an even greater excess at ages 25 to 44.
39. The United Nations Secretariat has made a But despite such regional diversity, and for unknown
survey of sex-age patterns in urban populations through- reasons, at ages 65 and over all major areas show a lower
out the world. It was found that each national popu- urban than rural sex ratio, and the difference between the
lation is urbanized to a varied extent when every sex-age two is greatest in Northern America, Latin America and
group is considered separately. It was found, further- Oceania.
more, that the sex-age patterns of urbanization vary 42. The economic, social and cultural reasons for such
among different parts of the world, no doubt owing to a diversity in regional modes of urbanization have not
a diversity of social, economic and cultural conditions yet been systematically studied, 30 but it is evident that
underlying urbanization trends, in factors causing the age structures reflect urban-rural differences in demo-
differences between urban and rural fertility and mortal- graphic trends (fertility, mortality and migration). Since
ity, and in the selection of rural migrants proceeding urban and rural structures differ so much, it is of consider-
to urban areas by sex and age.29 able interest to detail urban and rural population projec-
40. Estimates have been made for 1960, largely based tions according to sex and age. In the choice of a suitable
on census data, of the sex-age compositions of the urban method to obtain such results, it is of some importance to
and rural populations in eight major areas of the world, appreciate the relative roles of the several demographic
as summarized in table 1. It will be noted that in all trend components in producing such effects.
eight areas the percentages of children, up to the age of
15, are smaller in urban as compared with rural places, URBANRESIDENCE RATIOS BY SEX AND AGE
certainly as a result of lower urban than rural birth 43. There is a more direct way to analyse the structural
rates. Likewise, in all areas the urban population has differences between the urban and rural population. If
larger proportions at ages 15-44 than does the rural we can assume a sufficient degree of population mobility
population because urban places attract especially those between the two types of residence (urban, rural) we can
persons of principal working ages. In Europe, Northern consider the percentage urbanized, in each sex-age group,
America, Latin America and Oceania, but not in the as reflecting a specific preference for residence in urban,
other major areas, middle-aged and elderly adults are rather than rural, areas, varying according to sex and
also relatively more numerous in urban than in rural age. Those sex- and age-specific "urban residence
places. It is probable that at least in these regions ratios" can therefore be plotted in a curve much as one
See "Sex and age patterns of the urban population" (ESA/ One further United Nations study suggests that there is an inter-
PiWP.36); "Comparative regional typology of urbanization patterns relation between patterns of urbanization and patterns of marital
by sex and age" (ESA/P/WP.42); and "Sex-age composition of the status, both of which may be partly determined by the role and
urban and rural population of the world, major areas, regions and status of women in a society. See "Urban-rural differences in the
individual countries in 1960" (ESA/P/WP.44). marital status composition of the population" (ESA/P/WP.Sl).
 TABLE1. PERCENTAGE AGE COMPOSmON OF MALE AND FEMALE POPULATIONS. AND AGE-SPECILlC SEX RATIOS.
IN URBAN AND RURAL AREAS IN EACH OF EIGHT MAJOR AREAS OF THE WORLD. AS ESTIMATED FOR 1960

East Asia South Asia Europe Soviet Union Africa Northern America Latin America Oceania
.
.
Group Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural

Males
Allages ....................... 100.0
0 4 ......................... 12.8
5-14 ......................... 21.0
15-24 ......................... 22.1
25-44 ......................... 28.8
45-64 ......................... 12.7
65 and over .................... 2.6
Females .
Allages ....................... 100.0
WI ......................... 12.9
5-14 ......................... 21.1
15-24 ......................... 21.2
25-44 ......................... 28.3
45-64 ......................... 13.0
65 and over .................... 3.7
Males per 100 females
Allages ....................... 103
0-4 ......................... 103
5-14 ......................... 103
15-24 ......................... 108
25-44 ......................... 105
45-64 ......................... 101
65 and over .................... 74
can plot age-specific labour force participation rates, 45. An exact interpretation of the graphs is not possible
or age-specific percentages of married population, owing to the accumulation of various effects in the course
entries and departures from the category being possible of time. Nor are the data very accurate: zig-zags at
at each age so as to result in a curve. There is, however, advanced ages can be the result of a different incidence of
an important exception in the factors causing the varied age misstatements in urban as compared with rural areas.
urban residence ratios: the children of either urban or A few facts are evident nevertheless: the young migrants
rural parents have no free choice of residence and ordi- to urban areas are predominantly female in Colombia
narily stay with their parents, or at least with theirmothers. and predominantly male in Iran: in both countries, female
Only as age progresses into adolescence and beyond can migrants are somewhat younger, on an average, than
one note the effects of the varying balance between male migrants; in both countries, as witnessed by child-
entries into and departures from urban residence. woman ratios, urban fertility is lower than rural fertility;
44. Almost invariably the urban residence ratios and whereas probably most female migrants in ~olombia
attain a peak near those ages at which migration is most establish permanent residence in urban areas, in Iran the
frequent. 31 The age of greatest peak comes a few years residence of many young male migrants in urban areas
of age sooner for females than for males. The height of is probably more temporary, leaving smaller balances
the peak, on the whole, is greater for men than for women past the age of 25. The impressions may be partly
in Africa and Asia, but it is greater for women than for falsified by inaccuracy in the age data, and by possible
men in Latin America and in most of the more developed and unknown fluctuations in net migration during
countries. Countries differ in the shape of the curve decades of the past.
at ages past the peak. In some countkes, for instance,
a second peak appears at some more advanced ages, DIFFERENCES
BETWEEN URBAN AND
after an intervening depression. Probably the shape of RURAL CRUDE DEATH RATES 32
the curve reflects to a considerable extent the past trends
of net rural-to-urban migration which, of course, could 46. Conditions affecting mortality are apt to differ
have fluctuated. To an undetermined extent, however, between urban and rural areas. For instance, in urban
the curve is also influenced by comparative levels and places medical facilities are more readily accessible. On
trends in urban and rural fertility, and mortality, respec- the other hand, risks of contagion or motor vehicle
tively. As is known, age structure can be greatly affected accidents are probably less in the rural areas. Because
by varying fertility trends of the more or less distant past, of a differing incidence of various forms of pathology,
but only to a minor extent by varying mortality trends. it is possible that one of the two environments has relative
advantages for persons of one or the other sex, or for
TABLE2. PERCENTAGE
OF NATIONAL SEX-AGE
GROUPS RESIDING individuals of different age. Whether, on balance,
IN URBAN AREAS,
COLOMBIA1964 AND IRAN 1966 mortality is heavier in cities or in the countryside is a
Co!ombia 1964 Iran 1966
little-researched subject. It is believed that, because of
the relative lack of health services, rural mortality appre-
Age (years) Males Females Males Females
ciably exceeds the urban in the less developed countries,
but it is not known how large the difference may be in
0-4 .............. 49.1 49.5 35.0 35.6
5-9 .............. 48.5 49.8 35.9 36.3
the typical instances. In the more developed countries
10-14.. ............ 49.3 53.7 42.0 42.6 medical and sanitary services are very widely available,
15-19 .............. 50.7 60.3 46.9 42.9 and mortality is generally so low that the differences
20-24.. ............ 51.4 58.5 51.2 42.2 between urban and rural areas cannot be very wide.
25-29.. ............ 51.8 57.0 41.5 38.2 47. The United Nations has carried out some calcula-
30-34.. ............ 52.4 57.0 38.7 37.5 tions of what the urban and rural crude death rates could
35-39.. ............ 49.3 55.1 38.7 40.6
4 0 4 4 .............. 49.1 55.1 36.8 36.6 have been around 1960 if mortality conditions in each
45-49.. ............ 47.8 55.6 39.8 42.2 region had been the same in urban and rural areas. Then
50-54.. ............ 47.3 56.2 39.8 41.8 assuming some differences at least in the less developed
55-59.. ............ 49.0 58.3 37.1 41.3 regions, the United Nations also calculated what the
60-64 .............. 45.3 56.7 37.7 38.7 crude death rates might actually have been under those
6 5 - 6 9 . ............ 49.0 :!Xi) 35.4 36.5 assumption^,^^ with further adjustments in these rates
70 and over ........ 46.2 to make them compatible with a basic set of United
Nations estimates pertaining to regionalrates of population
Almost invariably, the urban residence ratios for small growth. The two sets of results are shown in table 3.
children fall below the level for women of ages at which
they might be their mothers; this is a reflection of a Sa In this and the following two sections, estimates are discussed
usually lower fertility in urban areas as compared with as calculated in "The components of urban and rural population
rural ones. To illustrate this rather summary statement, change: tentative estimates for the world and twenty-four regions
the relevant ratios for Colombia (1964) and Iran (1966) for 1960" (ESA/P/WP.46).
are shown in table 2 and figure I. It was assumed that urban expectation of life at birth exceeded
the rural by 5 years in East Asia (except Japan), South Asia, Africa.
Melanesia, Polynesia and Micronesia, and by 2.5 years in Latin
81 AS further examined from intercensal balances in "The dynamics America. In Japan, Europe, the Soviet Union, Northern America
of rural-to-urban population transfers by sex and age" (ESA/P/ and Australia and New Zealand urban and rural mortality were
WP.48). assumed to be the same.
 FIGURE I
Percentage of national sex-age groups residing in urban areas,
Colombia 1964 and Iran 1966
COLOMBIA 1964

Males

40J I I I I I I I
1
0-4 10-14 20-24 30-34 4044 50-54 60-64 70 and over
,Age (years)

IRAN 1966

2 40

n
Males

I I I I 1 I I 1
04 10-14 20-24 30-34 4044 50-54 60-64 70 and over
Age (years)

TABLE3. URBANAND RURAL CRUDE DEATH RATES WHICH WOULD RESULT IF URBAN AND
RURAL MORTALITY WERE THE SAME, AND AS ACTUALLY ESTIMATED, AROUND 1960

Assuming equal urban and

rural mortality in each As actunIIy estimated
mqior area (with adjustment)
Urban Rural Urban Urban Rural Urban
death death minus death death minus
Major area rate rate rural rate rate rural

East Asia ...................... 14.9 18.6 -3.7 12.8 19.2 -6.4

South Asia .................... 20.5 22.3 -1.8 17.1 22.8 -5.7
Europe ........................ 10.4 10.4 0.0 10.1 10.0 0.1
Soviet Union .................... 6.5 8.3 -1.8 6.5 8.4 -1.9
Africa .......................... 20.9 24.3 -3.4 17.9 25.0 -7.9
Northern America .............. 9.2 9.6 -0.4 9.0 9.3 -0.3
Latin America .................. 11.6 12.0 -0.4 10.8 12.5 -1.7
Oceania ........................ 9.7 12.3 -2.6 8.7 13.1 -4.4

48. In the first three columns of table 3, differences and Latin America, however, the differences are slight or
between urban and rural crude death rates are due negligible because in these major areas the urban localities
only to differences between urban and rural sex- contain also comparatively large shares of the aged
age compositions. Since young adults predominate population. In the last three columns of the table,
more in urban than in rural localities, and death risks however, it becomes evident that probably in many
for young adults are always comparatively low, these regions the difference between urban and rural crude
hypothetical death rates are generally lower in the urban death rates is actually larger than it would be for reasons
areas, the differences being most considerable in East of age structure only. As estimated, the urban death
Asia, Africa and Oceania; in Europe, Northern America rate may have been almost 8 points less than the rural in
 TABLE4. URBANAND RURAL CRUDE BIRTH RATES WHICH WOULD RESULT IF URBAN AND
RURAL FERTILITY WERE THE SAME, AND AS ACTUALLY ESTIMATED, AROUND 1960

Assuming equal urban and

ruralJkrtility in each As actually estimated
major area (with adjustment)
Urban Rural Urban Urban Rural Urban
birth birth minus birth birth minus
Major area rate rate rurrrl rate rate rural

East Asia ...................... 40.0 35.2 4.8

South Asia .................... 44.6 44.5 0.1
Europe ........................ 19.5 17.9 1.6
Soviet Union.. .................. 27.2 22.4 4.8
Africa .......................... 47.3 47.4 -0.1
Northern America .............. 25.4 22.1 3.3
Latin America .................. 43.7 37.0 6.7
Oceania ........................ 28.1 26.5 1.6

Africa, about 6 points less in East Asia and South Asia, sex-age compositions. On this hypothetical basis, the
and considerably less also in Oceania; in Europe and urban birth rates would generally exceed the rural birth
Northern America, on the other hand, the difference, if rates because of the prevalence of young adults in the
any, could have been only slight. urban areas. However, as already noted, in some
regions the urban population has comparatively few
DIFFERENCES
BETWEEN URBAN AND women, and mainly for this reason equal urban and rural
RURAL CRUDE BIRTH RATES fertility would produce only a negligible difference
49. Virtually throughout the world the fertility of between the hypothetical urban and rural birth rates of
urban women is lower than that of rural women, and in South Asia and Africa. In Latin America, where women
many instances the difference is large. Fertility decline, predominate in the urban population, equal fertility
where it has occurred, usually began first in urban areas would cause the urban birth rate to exceed the rural by
and spread to rural areas later on. Thus, the urban- nearly 7 points. Considerable excesses of the urban over
rural fertility difference can widen for a time, to narrow the rural crude birth rate would have occurred also in
down again when the general decline in fertility comesto an East Asia, the Soviet Union and Northern America.
end. The reasons for such differences between urban and 52. Actually, however, as calculated by reverse-survival
rural fertility appear to be manifold. Various factors of sex-age data, the urban crude birth rate was in general
usually associated with lowered fertility, such as education, considerably below the rural crude birth rate. Only in
public health, non-agricultural activities, the economic Northern America was the difference rather slight. In
activity of women, the prevalence of salaried employ- Europe, the rural birth rate exceeded the urban by 4 points,
ments and so forth, are associated with urbanization; in the Soviet Union and Africa by about 6 points, in
hence it cannot be said to what extent the urban habitat East Asia and South Asia by about 7 points, in Latin
itself, as an isolated factor, contributes to this fertility America by 9 points, and in Oceania by 14 points.36
difference. One factor, which has not been sufficiently
explored may be the different incidence of marriage among DIFFERENCES BETWEEN URBAN AND
young urban or rural women: since urban women generally RURAL RATES OF NATURAL INCREASE
marry at a later age or more often remain single, the
general fertility of urban women can be expected to be 53. On the basis of age composition only, as has been
lower than that of the rural for this reason alone. 34 argued, the urban areas would have higher crude birth
rates and lower crude death rates than the rural areas.
50. Without going into the discussion of possible Therefore, on this hypothetical basis, the urban areas
causes, we refer to United Nations estimates of urban would have a greater potential for natural increase,
and rural crude birth rates around 1960, analogous to the assuming that urban and rural fertility and mortality
estimates of death rates already discussed. Comparison could be equal. In actual fact, however, both the birth
is made of the figures shown in table 4. rates and the death rates are lower in urban than in rural
51. If fertility, calculated as the average number of places, and often the excess of the rural over the urban
births to women aged 15 to 44 years, were the same in birth rate is greater than the excess of the rural over the
urban and rural areas, differences between urban and rural urban death rate, with the consequence that the rate of
crude birth rates would result as shown in the first three natural increase is generally somewhat lower in urban,
columns of table 4. These differences would have been as compared with rural, populations. This reasoning
due entirely to differences between urban and rural is illustrated in table 5, showing the rates of natural
increase which result from subtraction of the death rates
In fact, women's age at marriage is almost everywhere higher
in urban as compared with rural areas; and the proportions of 56 In the case of Oceania it should be noted that most of the
married women are markedly lower in urban as compared with urban population is that of Australia and New Zealand, whereas
rural areas: "Urban-rural differences in the marital status composi- much of the rural population is that of Melanesia, Polynesia and
tion of the population" (ESA/P/WP.51). Micronesia.
 TABLE
5. URBANAND RURAL RATES OF NATURAL INCREASE WHICH WOULD RESULT IF URBAN AND
RURAL FERTILITY AND MORTALITY WERE THE SAME, AND AS ACTUALLY ESTIMATED,
AROUND 1960

Assuming equal urban and

rural fertrlity and As actually estimated
mortality in each major area (with adjustments)
Urban Rural Urban Urban Rural Urban
natural natural minus natural natural minus
Mojor area increase increase rural increase increase rural
--
East Asia ........
South Asia ......
Europe ..........
Soviet Union ......
Africa ............
Northern America
Latin kmerica ....
Oceania ..........

in table 3 from the birth rates in table 4. Potentially, in rates of natural increase or to international migration.
the urban natural increase could have exceeded the rural It is therefore obvious that net migration between rural
by more than 8 points in East Asia, by about 7.points in and urban areas plays a large role. The relative effect
the Soviet Union and Latin America, by about 4 points of internal migration will be greater on the urban popula-
in Northern America and Oceania, by 3 points in Africa tion while this is still a minority in the national population,
and by smaller amounts in South Asia and Europe. and greater on the rural population once the urban popu-
However, as actually estimated, the reverse was the case, lation makes up the majority. The sex-age selectivity
and the rural natural increase probably exceeded the of migration is important because of the resulting effects
urban by almost 10 points in Oceania, 7 points in Latin on urban and rural population structures. Unfortun-
America, 4 points in Europe and the Soviet Union, and tunately, difficulties are involved in the more detailed
smaller amounts in East Asia and South Asia; in Northern calculation of these migratory effects.
America the estimated difference was slight, and in Africa 56. In a calculation of rates it is logical to attribute
the urban natural increase was estimated higher than the migration to the population in the areas of origin,
rural, though not by a large amount. namely the population at risk of sending migrants.
Actually, however, the matter is more complicated since
DIFFERENCES
DUE TO THE EFFECTS migration occurs in both directions, rural-to-urban and
OF INTERNATIONAL MIGRATION urban-to-rural. It is possible, for instance, that in
54. At least in a few countries international migration certain age groups the balance of all movements is in
can have a significant effect on the growth and composition favour of urban areas, while in some other age groups
of urban and rural populations. In earlier times there the balance is in the opposite direction. However, the
was much international migration originating from rural separate calculation of the components of each migratory
areas in one country, destined to settle in rural areas balance is a refinement depending on detailed statistics
of another country. This may still be the case in some which are rare, and it may be unnecessary in projections
countries, such as in Africa. But elsewhere in recent made only for the purpose of forecasts. The more
times most international migration has had both an direct estimation of migratory balances by sex and age
urban origin and an urban destination. The composition is discussed in chapter IX.
of the international migratory stream will have some 57. An important question, in assessing the net effects
effect on the sex-age structures of the populations affected. of rural-to-urban migration, is the extent to which
Countries where these considerations are of some impor- migrants establish a permanent urban residence. Where
tance are fewer than they used to be, nor are the effects they do, they contribute to urban population growth also
as large as they were in earlier times. Unfortunately, in the long run, founding new urban families and pro-
in many countries the statistics on international migration ducing offspring. In other instances, many young
are quite unsatisfactory, and even where they are fairly persons in urban areas constitute a floating population
accurate the urban or rural origin of the migrants remains that is in a continuous state of turnover. The pool of
undetermined. short-term migrants is replenished by a continuous inflow
of young persons, but considerably drained, at the same
DIFFERENCES
DUE TO INTERNAL time, by a continuous return flow to rural areas. 36 The
RURAL-TO-URBAN MIGRATION floating migratory population may increase at times when
55. The most conspicuous cause of differences between urban areas offer a particular attraction, but shrink again
urban and rural rates of population growth is the geo- at other times when they do not. It may leave only a
graphic shift of persons from rural to urban places of small residual of persons remaining in the urban area
residence within the same countries. Throughout the
The large effect of return migration from an Asian city is docu-
world, urban populations grow with considerably greater mented by K. C. Zachariah, "Bombay Migration Study: A Pilot
speed than do the corresponding rural populations. Analysis of Migration to an Asian Metropolis", Demography
Little of that difference can be attributed to differences (Chicago), vol. 3, No. 2, 1966, pp. 378-392.
up to an advanced age. But where migration leads involved is of a type intermediate between the typical
largely to continued urban residence, migrants of earlier compositions of urban and rural populations, respect-
date remain in the urban population at all ages following ively. This can be approximately the case because the
the age of peak migration. If a survey were to be made, localities concerned may usually have characteristics
it would be difficult to distinguish between the "floating" which place them near the border line between the two
migrants and the more permanent migrants; persons types of settlement. In a country whose population is
who have migrated to town will often not know themselves about 50 per cent urbanized, it is possible that the sex-
whether they are likely to leave soon, or whether they are age composition of the reclassified areas resembles that
apt to stay a longer time or even indefinitely. In those of the combined national population. In less urbanized
few countries where the census draws a distinction be- or more highly urbanized countries this should generally
tween de jure and de facto population, the differences not be expected.
between the two categories in urban areas reflect to some 62. An example of sex-age compositions of populations
extent the presence of temporary migrants. in localities of differing size groups is provided in a special
58. The type of migration to urban places will also report of the 1965 population census of Japan. Table 6
depend in part on the census definition of "urban" areas, shows percentage compositions of Japan's total popula-
already discussed. For instance, in many countries tion and of its population in different community size
there is a considerable migration from city centres groups. 3s In examining these data it should be borne in
toward the suburbs. Often this is a migration of mature mind that Japan is now a highly urbanized country. Two
adults, perhaps aged in their thirties, who, having made significant observations can be made. First, the percent-
a successful beginning in an economic career, wish to age compositions vary in almost continuous fashion, as
settle their young families in the more favourable sub- we proceed from settlements of the largest to those of
urban housing conditions. Where suburbs do not the smallest size. For instance, males aged 20 to 24 years
form part of the definition of "urban" localities, the constituted 7.3 per cent in cities of 500,000 or more
so-called "urban" places may gain young adult migrants inhabitants, while in successively smaller localities they
(aged, say, in their twenties), while losing those aged in constituted 4.9,4.2,3.4,3.0,2.6 and 2.4 per cent. Second-
their thirties together with their young children. ly, the national sex-age composition of the population is
59. A desirable type of cegsus data for population most nearly approximated in towns having around 50,000
projections (usually carried out in five-year time inter- inhabitants. As a matter of fact, in 1965 nearly half the
vals) would be those on places of residence five years population of Japan inhabited towns of this size and larger.
previous to a census in relation to their places of current 63. If it is a case of annexation of suburban areas,
residence. Five-yearly migration rates, immigration as the matter may be quite different. Aside from migratory
well as emigration, would then be directly calculable. balances between the urbanized areas and the rest of the
The data would be even more fruitful if it could be country, important balances also usually exist between
determined whether the previous place of residence the central parts of each city and its suburbs. The latter
(such as, five years ago) was urban or rural, 37 but such migration is usually highly selective of groups of sex and
figures are seldom secured as well as difficult to obtain age. In the United States, for instance, the suburban
because of possible definitional changes in urban and rural areas tend to be settled by men of career ages and their
localities. A question remains also whether the places respective families, hence the suburbs have higher
of previous residence have been stated accurately in the percentages of small children and persons in their thirties
census. than do the city centres, whereas the latter have higher
percentages of adolescents and young adults, as well as
EFFECTSOF RURAL-TO-URBAN AREA
RECLASSIFICATION
of adults in mature and advanced ages.
64. Census data for 1960 have been provided for major
60. Depending on the kind of definition involved, cities in the United States whose territory had been
rural-to-urban reclassifications can occur especially in enlarged as a result of annexations between 1950 and
two types of areas: (a) in villages or other settlements 1960.39 Percentages in each sex-age group of the
attaining the status of "towns"; and (b) in suburban combined city population residing in amexed areas can
areas being annexed by the administrative authorities of be calculated, and table 7 presents the results for the
geographically expanding cities. The consequence in city of Dallas. The same data are also charted in
either event is a transfer of population from the rural figure 11. As can be seen, the above statement regarding
to the urban category. Census data describing the suburban population is well borne out. It is rather
sex-age composition of the transferred population are probable, however, that in a city like Dallas much of
rarely available. The study of data in those countries the migration to the annexed areas occurred subsequently
where they are provided gives some indication of pos-
sible effects of these transfers upon the sex-age compo- Japan, Bureau of Statistics, 1965 Population Census of Japan:
sition of the rural and the urban population. Results by Population Size of Shi, Machi and Mura and of DID'S
(Officeof the Prime Minister). The calculation was made for shi,
61. If it is only a matter of rural localities attaining machi and mura, those being the minor administrative divisions of
urban status, it can be conjectured, very roughly, that the country.
the sex-age composition of the marginal population 38 Miller and Varon, Population in 1960 of Areas Annexed to
Large Cities of the United States between 1950 and 1960 by Age,
Such data have been found in censuses of India and Greece. Sex and Color, Analytical and Technical Reports No. 1 (Phila-
See Manual VI: Methods of Measuring Internal Migration (United delphia, University of Pennsylvania, Population Studies Center,
Nations publication, Sales No. E.70.XIII.3), p. 37. November 1961).
 TABLE .
6 PERCENTAGE
COMPOSlTION BY SEX AND AGE OF POPULATION IN JAPANESE
ADhlINBTRATlVE UNITS
ACCORDING TO POPULATION SIZE. 1965

Size of administrative unit (number of inhabitants)

Japan 500.000 100.000 - 50.00.0 30.O W . 10.000 . 5.000 . Smaller than
Sex and age total population and over 499.999 99.999 49.999 29.999 9.999 5.

Males ................... 49.1 50.8 49.1 48.6 48.2 48.3 48.4 49.0
0-4 ................. 4.2 4.2 4.5 4.4 4.2 4.1 3.9 3.9
5-9 ................. 4.1 3.3 3.9 4.1 4.4 4.6 4.7 5. 0
10-14 ................. 4.8 3.4 4.3 4.7 5.3 5.8 6.2 6.4
15-19 ................. 5.6 6.3 5.8 5.5 5.3 5.1 4.7 4.0
20-24 ................. 4.6 7.3 4.9 4.2 3.4 3.0 2.6 2.4
25-29 ................. 4.2 5.7 4.6 4.1 3.6 3.2 2.9 2.9
30-34 ................. 4.2 4.7 4.5 4.3 4.0 3.7 3.6 3.6
35-39 ................. 3.8 3.8 3.9 3.9 3.8 3.7 3.7 3.9
40-44 ................. 2.8 2.7 2.8 2.8 2.8 2.8 2.9 3.1
45-49 ................. 2.3 2.1 2.2 2.3 2.4 2.4 2.4 2.5
50-54 ................. 2.2 2.1 2.1 2.2 2.3 2.3 2.4 2.5
55-59 ................. 2.0 1.7 1.8 2.0 2.1 2.1 2.3 2.4
60-64 ................. 1.7 1.4 1.5 1.6 1.8 1.9 2.1 2.2
65-69 ................. 1.2 1 .0 1.1 1.2 1.3 1.5 1.7 1.8
70-74 ................. 0.8 0.6 0.7 0.8 0.9 .
10 1.1 1.2
75+ .................. 0.8 0.4 0.6 0.7 0.8 1 .0 1.1 1.3
Females ................. 50.9 49.2 50.9 51.4 51.8 51.7 51.6 51.0
0-4 ................. 4.1 4.0 4.3 4.2 4.1 3.9 3.7 3.8
5-9 ................. 3.9 3.2 3.7 3.9 4.2 4.5 4.6 4.8
10-14 ................. 4.6 3.3 4.1 4.5 5.1 5.7 6.0 6.3
15-19 ................. 5.5 5.6 5.8 5.8 5.5 5.2 4.7 3.7
20-24 ................. 4.7 6.0 5.1 4.7 4.1 3.6 3.2 2.8
25-29 ................. 4.3 5.1 4.7 4.4 3.9 3.5 3.2 3.0
30-34 ................. 4.2 4.5 4.5 4.2 4.1 3.8 3.6 3.6
35-39 ................. 3.8 3.8 3.9 3.8 3.9 3.8 3.8 3.8
40-44 ................. 3.3 3.1 3.2 3.3 3.4 3.4 3.5 3.5
45-49 ................. 2.7 2.5 2.6 2.8 2.9 2.9 3.0 3.0
50-54 ................. 2.5 2.3 2.4 2.5 2.7 2.7 2.9 2.9
55-59 ................. 2.1 1.8 2.0 2.1 2.2 2.3 2.5 2.5
60-64 ................. 1.7 1.5 1.6 1.7 1.9 2.0 2.2 2.2
65-69 ................. 1.4 1.1 1.2 1.3 1.5 1.6 1.8 1.8
. 70-74 ................. 1.0 0.7 0.8 1 .0 1.1 1.2 1.3 1.4
75+ .................. 1.2 0.7 1 .0 1.2 1.3 1.5 1.7 1.9

TABLE7. POPULATION
OF DALLAS
(UNITED STATES).1960: rather than prior to their annexation. with the implication
PERCENTAGEOF EACH SEX-AGE GROUP CONTAINED IN AREAS that not so much adjustment is needed for this case of
ANNEXED BY THE CITY BETWEEN 1950 AND 1960. AND WITHIN M E reclassification as the population data for the annexed
1950 BOUNDARIES (WHITE POPULATION ONLY)
areas would seem to suggest.
Annexed areas Within 1950 baunrinries NET RURAL-TO-URBAN POPULATION TRANSFERS. I.E. COM-
Ages Males Females Males Females BINED EFFECTS OF MIGRATION AND AREA RECLASSIFI-
CATION
65. Whereas census data reflecting directly on rural-
to-urban migration are rare. it is often possible to calcu-
late the combined balance of net rural-to-urban popu-
lation transfers resulting from the migratory balance as
well as area reclassification. Given the rates of urban
and rural population growth and the rates of urban and
rural natural increase. simple subtraction can yield the
rates of rural-to-urban transfers. It can usually be
presumed that the greater part of the transfers is due to
migration. and only a smaller portion to area reclassi-
fication. depending on whether the definition of urban
areas is more or less flexible.*O
40 But this does not seem to hold true in Brazil. where the analysis
of data suggests that reclassification contributed in a large measure
to urban growth (see chap. IX) .
 FIGURE
I1
Population of Dallas (United States), 1960: percentage of each sex-age group
contained in areas annexed by the city between 1950 and 1960

\---
\ --I Males

--a

Females

0J ,
0-4
I
10-14
I

20-24
I

30-34
I

40-44
I
50-54
I
60-64
I
70-74
Age (years)

TABLE8. RATESOF URBAN POPULATION GROWTH AND ITS COMPONENTS, AND RURAL RATES
OR NATURAL INCREASE AND ITS COMPONENTS,
AROUND 1960

Urban populatiorr growth rate Rural natural increase rate

and its components and its components
(per 1,000 urban poprrlation) (per 1,000 rirral population)
Growth Natural Transfers Natural Growth Transfers
Major area rate increase from rural increase rate to urban

East Asia ...................... 47.0 17.1 29.8 17.6 8.8 8.7

South Asia .................... 37.5 23.0 14.5 24.5 21.2 3.2
Europe ......................... 18.1 7.7 10.4 11.7 -4.3 16.0
Soviet Union.. .................. 35.1 14.4 20.7 18.2 -2.1 20.3
Africa .......................... 45.8 24.0 21.9 23.0 18.2 4.8
Northern America . . . . . . . . . . . . . . . 24.6 15.4 9.2 15.6 -1.8 17.4
Latin America .................. 44.8 24.6 20.3 32.0 12.7 19.3
Oceania ........................ 26.5 13.7 12.9 23.6 13.2 10.3

66. The United Nations estimates of urban and rural and Oceania, and 18 per 1,000 in Europe. Urban rates
rates of natural increase around 1960 in eight major of natural increase were 23 to 25 per 1,000 in South
areas of the world, already discussed (see table 5), have Asia, Africa and Latin America, 14 to 17 per 1,000 in
been ajusted to correspond with another set of estimates East Asia, the Soviet Union, Northern America and
relating to urban and rural rates of population growth, 41 Oceania, and 8 per 1,000 in Europe. It follows that the
with results shown in table 8. The interdependence of rates of net incoming population transfers, relative to
these comparative figures is of course conditioned by the the size of the urban population, were about 30 per 1,000
levels of urbanization. As estimated for 1960, the urban in East Asia, 20 to 22 per 1,000 in the Soviet Union,
population constituted 70 per cent of the total in Northern Africa and Latin America, and 9 to 14 per 1,000 in South
America, 66 per cent in Oceania, 58 per cent in Europe, Asia, Europe, Northern America and Oceania. The
49 per cent in the Soviet Union, 48 per cent in Latin proportion of urban population growth accounted for by
America, 23 per cent in East Asia, and 18 per cent each transfers from rural areas was thus 64 per cent in East
in South Asia and Africa. Asia, 59 per cent in the Soviet Union, 58 per cent in
67. As estimated for 1960, urban rates of population cent Europe, 47 per cent each in Africa and Oceania, 46 per
in Latin America, 38 per cent in South Asia, and
growth were 45 to 47 per 1,000 in East Asia, Africa and 37 per cent in Northern America.
Latin America, 35 to 37 per 1,000 in South Asia and the
Soviet Union, 25 to 27 per 1,000 in Northern America 68. Rural rates of natural increase were 32 per 1,000
in Latin America, 23 to 24 per 1,000 in South Asia,
The growth rates were calculated as exponential rates (not Africa and ~ c e a n i a ,16 to 18 per 1,000 in East Asia, the
compound rates) to maintain comparability with birth rates, death and America, and 12 per 1,000 in
rates and rates of natural increase (ESA/P/WP.46). Europe. The rates at which rural populations grew
were less, namely 21 per 1,000 in South Asia, 18 per subtracted; in the third event, he is not registered in the
1,000 in Africa, 13 per 1,000 in Latin America and urban population at all; and in the fourth event, he is
Oceania, and 9 per 1,000 in East Asia; in Europe, the again added. The net result is the addition of one
Soviet Union and Northern America, the rural population individual to the urban population, but the events have
underwent slow decreases. Relative to the size of the been several. The fictitious example illustrates that the
rural population, outgoing population transfers occurred several events depend not only on the initial urban popu-
at rates of 16 to 20 per 1,000 in Europe, the Soviet Union, lation, but on an urban population which changes while
Northern America and Latin America, at 10 per 1,000 in these and other events, some of them partly interdepen-
Oceania, at 9 per 1,000 in East Asia, and at comparatively dent, occur. A complete independence of different
low rates in South Asia and Africa whose rural popula- events from each other can be postulated only when the
tions, relative to the size of the urban, were still very time interval is infinitesimally short.
large. 42 The proportion of rural natural increase disposed
of by transfers to urban areas was thus 137 per cent in
73. Some degree of statistical stability-whether a
constant level, a continuous trend or fluctuations-can be
Europe, 112 per cent each in the Soviet Union and
Northern America, 60 per cent in Latin America, 51 per imputed to the flow of events when these are taken in
proportion to the population at risk of experiencing
cent in East Asia, 43 per cent in Oceania, 21 per cent in these events. Annual events occurring per thousand of
Africa, and 12 per cent in South Asia.
the population at risk constitute demographic rates. 43
However, some difficulties occur in determining which is
the population "at risk".
69. The foregoing review of components of urban and 74. Clearly it is the urban population itself that is at
rural population change and of their influence upon urban risk of experiencing urban births and urban deaths.
and rural population structures demonstrates a complexity On the other hand, it is the rural population (and also the
of factors difficult to isolate from each other unless the population of other countries) which is at risk of sending
available statistics reflect all that detail with great accu- migrants to the urban areas. Furthermore, the rural
racy. population, or at least some sections of it are at risk of
70. To recapitulate, the urban and the rural popu-
being reclassified to an urban status. "
lations can change by the incidence of urban and rural 75. In this connexion, it must be pointed out that
mortality, urban and rural fertility, internal migration the population "at risk" of sending emigrants from the
from rural to urban and from urban to rural areas, urban areas is not necessarily the urban population itself.
international migration from abroad to urban and rural Many of the persons leaving an urban area are previous
areas, and the population of previously rural localities immigrants, either from a rural area or from abroad, who
being reclassified to an urban status. The incidence of after a period of urban residence return to their places of
all these events varies to an important degree by groups origin. In so far as this is the case, the migrant popu-
of sex and age, and it will be recalled that age itself lation is the one "at risk". And since the risk of generat-
changes as time progresses. ing this migrant population (at least the internal migrants)
71. If we consider the urban population only we can pertains to the rural population of origin, it might even
derive a scheme of its change as follows. Let Uoand Ut be be said that the rural population itself is at risk of receiv-
the urban population at the beginning and at the end of a ing back some of its previous emigrants.45
time interval t ; B the number of births occurring during 76. Statistics detailed and accurate enough to trace
that interval; D the number of deaths; Zi internal migrants the frequency of every type of demographic event exist
from rural to urban areas; Ze internal migrants from urban in some of the countries which have systems of population
to rural areas; Ei migrants to urban areas from abroad; registers. However, where registers are maintained on
Ee urban emigrants to other countries; and R the popu- a local basis, the statistics are usually biased: whereas
lation being reclassified; we arrive at the equation most migrants register their arrival at the place of destin-
Ut=Uo+B-D+Ii-ZefEi-Ee+R ation, some always fail to register their departure at
the place of origin. Only in a centralized registration
72. If the time interval t is very short, only a few system is it possible to ensure that the resulting errors
individuals will undergo more than one of the indicated will be detected and reduced to a minimum.
events. Over a longer period of time, such as in the
course of five years, many individuals may experience 77. The existence of a vital statistics system and
two or several of the events noted in the equation. repeated censuses of sufficient accuracy can suffice for
Thus, one person may move from the rural to the urban calculations by which the balance of rural-to-urban
area, then emigrate abroad and then return to a rural 43 As explained in the second section of chapter 111, correct demog-
place which in turn undergoes urban reclassification still raphic rates are those related to the mean population of each time
within the same time period. In the first event, he is period, ordinarily represented by the mid-year population. The
added to the urban population; in the second event, he is mid-year population, of course, is already affected by about one-
half the events occurring in each year.
44 In special instances, it is also possible that part of the urban
AS-it happens (see also the first section of this chapter), the
highest rates of rural out-migration occurred mainly in those population is at risk of being reclassified as rural.
regions where women predominated among the migrants; where the 46 This would not be the case, however, where suburbs receive
migration consisted mostly of men, the rates of rural out-migration many migrants directly from the central city and the suburbs con-
were lower. tinue to be governed by administrations of a rural type.
migration and reclassification can be estimated. 46 In without the risk of great error, to estimate at least the
a few countries the censuses also contain data concerning net migratory balance between urban and rural areas.
place of residence at some prior fixed date, such as one In view of what has been said, the rural population can be
or five years prior to the census, and these data are directly considered at risk, not only of sending migrants to urban
relevant to the calculation of migration in both the rural- areas, but also of receiving some of them back as they
urban and urban-rural directions. However, such data return to their places of origin. This consideration can
are often believed to be rather deficient and, even in the make it legitimate to attribute the net balance resulting
few instances where the urban or rural character of the from migration and return migration to the rural popula-
previous place of residence is recorded, a residual uncer- tion as the one "at risk". One can go further and ascribe
tainty remains because some of those places may mean- the combined balance of migrationandareareclassification
while have become reclassified. The vital statistics, even to the rural population, i.e. to consider the rural popu-
if accurate and detailed, may record births and deaths by lation "at risk" of producing a net rural-to-urban popu-
place of occurrence rather than by place of usual residence, lation transfer with these combined effects. Projection
with the result that births and deaths occurring to rural methods therefore are applicable in which use is made of
residents at an urban hospital are attributed to the urban some combinations of the components of urban and rural
area. population change, instead of the use of detailed assump-
78. In the absence of vital statistics, it is still possible to tions concerning every one of the numerous components.
estimate levels of fertility and mortality from the relation- 80. Unfortunately, the countries lacking sufficient
ship between a population's age structure and the esti- statistics for urban and rural population projections
mated rate of population growth. 47 In that situation, by components of population change are more numerous.
one can no longer detect what differences in mortality A large part of the manual is concerned with situations
conditions might exist between urban and rural areas, where detailed statistics are rather scant. It is obvious
but in actual fact it is possible that those differences are that in such countries the projections can be made only
not quantitatively important in a population projection. by rather summary methods. However, it does not
On the other hand, by the method of reverse survival the follow that responsible projections can be made by the
level of urban and rural fertility can be estimated separ- mechanical or indiscriminate application of a simple
ately, often with a tolerable degree of accuracy, even if the formula. Judgment is needed in selecting the formula
level of child mortality is estimated only by a rough most appropriate to the particular conditions. Such
approximation. The accuracy of these estimating pro- judgment depends on some knowledge and experience of
cedures, however, can be much affected if the census age factors likely to affect urbanization and its demographic
statistics are inaccurate. components, as well as of the calculable interaction of
79. All things considered, there are a fair number of the demographic components in some other countries
countries in which the statistical data situation permits, where the statistical documentation is more satisfactory.
48 In this context it is worth mentioning that those areas often
46 See United Nations, Manual VI: Methods of Measuring Internal tend to be reclassified as urban where as a result of much recent
Migration (United Nations, Sales No. E.70.XIII.3). migration a population size or density has been attained which can
47 See United Nations, Manual IV: Methods of Estimating Basic be assimilated with urban conditions elsewhere in the same country.
Demographic Measures from Incomplete Data (United Nations, In the circumstances, reclassification is partly determined by migra-
Sales No. 67.XIII.2). tion.
 Chapter I11
TEMPO OF URBANIZATION AND LTRBAN CONCENTRATION

tions are clearly unpredictable. A useful forecast

should reflect a future long-run trend which, with cur-
81. After examining the effect of definitions on the rently available judgment, appears plausible. Actual
urban and rural population data and the likely influence future developments, subject to fluctuations, should not
of the components of population change, it is generally be expected to coincide closely with the forecast. For
useful to study the trends in the two population categories such reasons, a forecast should preferably be made for
(urban and rural) which have occurred in the past. In a period of at least twenty years, and with three variants,
the absence of other criteria, it may sometimes be assumed "high", "low" and "medium", such that it appears more
in the population forecast that similar changes will likely that future trends will lie within the range between
continue in the future. There are several ways in which the "high", and "low" variants than outside it. The
past urbanization trends can be measured, hence, depend- "medium" series should be the one most recommended
ing on the measure used, the future forecast may differ. for actual use, and the "high" and "low" series should be
Different measurements can be relevant depending on presented to indicate probable errors of a magnitude that
whether the urbanization level is comparatively high, should be no cause for surprise.
comparatively low or in some intermediate range.
82. Future assumptions, however, should not be
derived uncritically from past observations. However
it may be measured, the urbanization trend can vary 86. Perhaps the best known concept of a growth rate
from one period to another under the influence of numer- is that of the annual, or compound-interest rate. Because
ous developments. The past period for which observa- of its widespread use, it will also be employed in the
tions can be made may have been affected by special calculations of this manual unless otherwise stated.
circumstances unlikely to be repeated in the future. 87. The compound interest rate rests on an assumption
Again, for future purposes government policy favouring that by the end of each year an increment is added to
either urban or rural developments (for instance, industry, an initial amount (such as interest to capital) which stands
agriculture or housing policy) may have effects on urban- in a fixed proportion to the size of that amount at the
ization trends for which the past period does not consti- beginning of each year. Thus, if r is the proportional
tute a suitable precedent. But even then the past period annual increment, and t the time interval measured in
can provide a standard of comparison with assumptions
drawn up more freely for the future. For instance, it
+
years, then Pt = Po(l r)t, Po and Pt being the magni-
tudes (of capital; of population; etc.) at the beginning
may be assumed that the urbanization tempo (however and at the end of the time period. If r is unknown but
measured) will either accelerate or slow down. Poand Pt are given, r can be determined by transposition
83. It should also be borne in mind that urbanization
v;
I-

trends, despite their long-term inertia, are subject to of that formula into the form r = - - 1. The
fluctuations. While it is true that throughout the world
the urbanization levels are continuously rising-apart time interval t need not be an integral number of years
from temporary effects of a disastrous period, e.g. a and can also include fractions of a year. The relevant
destructive war-it is also true that the rises differ among calculations are performed easily with reference to a
countries and from one decade to another. conventional table of logarithms, i.e. logarithms with
84. There is no objective test for the distinction base 10, and such tables are widely available. It is not
between trend and fluctuations in the urbanization tempo. intended to discourage their use wherever appropriate.
If the past observations cover a long period, they may 88. For demographic purposes, however, the use of an
well be indicative of a trend; if they extend over a short annual rate is afflicted by some imperfections. First, it
period only, such as the most recent decade, they may be is to be noted that other demographic rates, such as
considerably influenced by temporary fluctuations; a birth or death rates, are not calculated with reference to
very long past period, however, is not necessarily indica- the population at the beginning of each year but, ordi-
tive of a future trend, because much time has elapsed narily, with reference to mid-year or mid-period popu-
since the beginning of the past period and relevant lations (being the best substitutes for period-mean
conditions may meanwhile have changed significantly. populations). In growing populations, each mid-year
85. Even for future periods it should be expected that figure is likely to be somewhat larger than the figure for
the urbanization tempo will fluctuate again; but fluctua- the beginning of each year, some growth having occurred
already within the first half of each year. It follows that rate it is. most convenient to make use of a table of the
birth rates and death rates are ordinarily slightly smaller exponential function if available. 49 Since the user of
than they would have been if (which is not the practice) this manual may find no table of the exponential function
they had been calculated in proportion to the population readily at hand, in so far as possible the calculations have
at the beginning of each year. If a growth rate is to be been performed according to the compound interest rate,
consistent with other demographic rates, therefore, it whose mathematical shortcomings are here admitted.
should not be calculated on a compound interest basis. Certain methods, however, will be explained further on,
89. Secondly, the compound interest rate lacks desir- which are derived from an exponential function. In the
able mathematical properties. With this type of rate, use of those methods, however, only one derived function
for instance, different results are obtained if a given is needed, and this is tabulated in annex I to this manual.
amount increases at a certain rate for two co~isecutive 92. For purposes of quick calculation, a third method
years, and if it increases at twice that rate for only one is worth mentioning which is entirely independent of
year. In the one instance, the relative increment becomes either logarithms or any other tabulated function, but
+ + +
(1 r)2, which is 1 2r r2, while in the other instance it must be stressed that satisfactory results will be obtained
+
it becomes only 1 2r, leaving a difference of r2, which only so long as the growth rate is not high and the time
may be quite small so long as the rate is not high. At period under consideration is not long. Within such
higher rates, or over extended periods of time, however, limits it can be demonstrated that a rough approximation
this internal inconsistency can become considerable. to a growth rate is obtained in the formula Pt = Po_- 2 rt +
90. These drawbacks are avoided when growth is L - rt
measured by instantaneous, also called "exponential" and its transformation r = 2(Pt - The first
rates, as expressed in the formula Pt = Poect, where c is t(Pt Po) +
the instantaneous rate and e is 2.71828, the basis of formula can serve to determine Pt when r is known,
Napierian, also called "natural" logarithms. It may be and the second to determine r when Pt is known. But
noted that the relationship between the compound despite its "mathematical" appearance, this formula is
interest rate and the exponential rate takes the form a simplification of only limited use. It should be employ-
+
ec = 1 r. The exponential increment, namely c - 1, ed only to establish quickly an approximate order of
if positive, is always at least slightly less than r, and if magnitude, for instance to check whether no error was
the rate is high it can be considerably less (in cases of made in the more precise calculation of either a compound
decrease, the negative rate c - 1 is always at least slightly interest or an exponential rate.
more than the rate r). This is so because in a steadily
increasing population the amount of increment, per
instant of time, rises continuously in proportion to the 93. The word "urbanization" has a double meaning:
population increase itself. For some selected instanta-
neous increments the corresponding compound increments it can indicate a situation existing at any particular
moment, in which instance it is best to speak of a level
are as follows: of urbanization; or it can be used to express the trend of
Instantaneous Conlpound urban developments, in which instance it is better to
increment (c - 1) increment (r) speak of the tempo of urbanization.
0.01 0.01005 94. The level of urbanization is measured simply, and
0.02 0.02020 for present purposes satisfactorily, by the percentage of
0.03 0.03045 urban in the total population at any fixed date. True,
0.04 0.04081 some more complex concepts have been brought into
0.05 0.05 127 use, such as indices of urbanization level weighted by the
0.06 0.06184 size classes of cities and towns,50 but those special
0.07 0.07251 purpose measurements need not concern us here. It is
0.08 0.08329 also to be noted that different levels of urbanization will
0.09 0.09417 result where urban population is defined in two or more
0.10 0.10517 alternative ways.
In some countries, urban populations grow at rates of 95. There are several alternative ways in which the
5, 6 or 7 per cent per year, and it is evident that an tempo of urbanization can be measured. The most
appreciable discrepancy can result between the two
49 See for instance, United States Department of Commerce,
methods of measuring such a high growth rate. Table of the Exponential Function ex (Washington, Government
91. The mathematical interaction between two or more Printing Office, 1961). It is also possible to operate with conven-
instantaneous rates is evident from the fact that the +
tional logarithms using the formula logl,Pt = 1 0 g , ~ P ~ c . l ~ g , ~ e . t ,
where logloe = 0.4342945.
variable is in the exponent. Thus, if c is the product n
of a and b, it must be true that ect equals eabt. It also C Cf2
follows that the same increase is achieved whether growth
occurs at a given rate over a given period of time, or at 60 For instance, U = - , where Cr is the population of city i,
P
any multiple of that rate over a corresponding fraction P is the total population of the country, and n is the number of cities.
1 Eduardo E. Arriaga, "A New Approach to the Measurement of
of that time period, because nc times ,t is the same as ct. Urbanization", Economic Development and Cultural Change
Rut to operate with the instantaneous, or exponential, (Chicago), vol. 18, No. 2, January 1970, pp. 206-218.
 9. ALTERNATIVE
TABLE MEASURES OF THE TEMPO OF URBANIZATION AS APPLIED TO ESTIMATES OF TOTAL
AND URBAN POPULATION OF EIGHT MAJOR WORLD AREAS FOR 1960 AND 1970

Tempo of urbanization, as measured by

(per cent per year):
Annual gain Annual rate
Total population Urban population Level of urbanization Annual rate in of gain in
(in thousands) (in thousands) (per cerrt urban in total) of growth percentage percentage
of urban urban urban
Major area 1960 1970 1960 1970 1960 1970 population population population

. .
East Asia . .. . . . . ..
.
South Asia . . . . . . ..
.. . .
Europe . . . . .. . ..
.
Soviet Union . .. . ..
.. .. .
Africa . . . . . . . ..
Northern America . ..
Latin America . . . ...
Oceania . . . .. .. . ..
. .
Note: Estimates according to Monthly Bulletin of Statistics. November 1971 (United Nations publication).

common-sense measures might seem to be these: the while much of the rural population is that of Melanesia
annual rate at which the urban population is growing; (where natural increase is high).
the annual amount by which the (percentage) level of
urbanization is rising; and the annual rate at which the TABLE
10. TEMPO
OF URBANIZATION AS MEASURED BY THE DIF-
level of urbanization is rising. When applied to FERENCE BETWEEN URBAN AND RURAL RATES OF POPULAnON
several populations, differing in level of urbanization or GROWTH (ACCORDING TO THE SAME E S ~ A T E S AS SHOWN IN
TABLE 9)
in rate of population growth, comparative results will
differ according to the measure used. Projections will Rate of growth Rare of growth Urban rate
in urban in rural minus
also differ if the tempo of urbanization is assumed to population population rural rate
continue but is measured in varied ways. The variety Mojor area
(per cent
per year)
(per cent
per year)
(per cent
per year)
of results obtained by different measurements is illustrated
in table 9. .. . . . . .
East Asia . . . . . . . 4.54
96. Judging by rates of growth in urban population, South Asia .. . . . .
.. .. . . 4.10
the tempo of urbanization was highest in Africa (4.69), .. . . .
Europe . .. . .. . . . . . 1.73
East Asia (4.54) and Latin America (4.42), and compara- . ..
Soviet Union . . . . .. . 2.71
tively high also in South Asia (4.10), whereas in Europe it Africa .. . ..
. . .. . . . . . . . 4.69
was the lowest (1.73). According to absolute rises in the Northern America . . . ... 2.03
percentage levels of urbanization, the tempo was highest
.
Latin America .. . . . . . 4.42
in Latin America (0.78) and the Soviet Union (0.77),
.. .. .
Oceania.. . .. . . . .. . 2.43

followed by East Asia (0.70) and Europe (0.54), whereas

it was lowest in Oceania (0.22) and South Asia (0.27). REASONS FOR USE OF THE URBAN-RURAL GROWTH DIF-
According to rates of rise in those percentage levels, the FERENCE AS A MEASURE OF THE TEMPO OF URBANIZATION
tempo was highest in East Asia (2.72) and Africa (2.15), 98. The measure illustrated in table 10, for simplicity,
considerably lower in Latin America (1.51), and lowest in will be referred to as the urban-rural growth difference,
Oceania (0.34) and Northern America (0.65). Each or URGD for short. It has several interesting advantages
of these different sets of measurement can have relevance, over the measures illustrated in table 9, especially in its
depending on the manner in which future projections are range of applicability. In a wide variety of circumstances,
to be formulated. Their limitations in projections are comprising virtually all those which will ever occur, the
illustrated in chapter IV. assumption can be made that an URGD observed in
97. If the United Nations method of projection is used, the past may also be maintained for an indefinite future
it is directly relevant to measure the tempo of urbanization period without leading to absurd results. This remains
as the net difference between the rate of growth in urban true irrespective of the current level of urbanization, the
population and that in rural population, with the results rate of growth in total population52 or whether rural
shown in table 10. According to this measurement, population is increasing or diminishing.
the tempo is highest in East Asia (3.72), followed by 99. The avoidance of absurd results does not, of course,
Latin America (3.22) and the Soviet Union (3.10), guarantee that such a projection will be an accurate
being lowest in Oceania (1.03) and South Asia (1.78). prediction of the future course of events. No formula
The peculiar result for Oceania is to be attributed to the of any kind can offer a key to unlock the mystery of
fact that most of the urban population is that of Australia future human developments. These are always subject
and New Zealand (where natural increase is moderate), to the influence of unforeseeable changes in social
61 Annual, i.e. compound interest, rates are used here, for simpli- Assuming, of course, that the total population will not grow
city. For a further explanation, see beginning of chapter IV. forever which, in the final result, eventually becomes impossible.
behaviour, difficult to identify and recognizable only in reflect a fortuitous coincidence because, as the level of
their consequences. But it can be argued on several urbanization rises very high, or the acceleration in the
grounds that the approach to the analysis of urbanization growth of total population subsides, rural population
trends by means of the URGD is reasonable. growth is bound to slow down. In fact, in nearly all
100. Let us first examine the three measures of the highly urbanized countries the previous growth in rural
tempo of urbanization which have been illustrated in population has come to an end and has given way to rural
table 9. population decrease, a possibility not allowed for in an
101. Almost universally urban populations are now assumption of constant rates of rural population growth.
growing faster than the corresponding rural populations 106. A measure might also be devised relating growth
and therefore faster also than the total populations in the urban population to growth in the total population.
(urban and rural). If the rate of growth in urban For instance, there are now many countries, with high as
population, higher than that in the total population, well as low rates of growth in total population, where the
remains constant, sooner or later a time will come when rate of growth in urban population is approximately
the calculated future urban population begins to exceed twice that in total population. But again it is evident
the calculated future total population, a result which that such a situation cannot continue indefinitely because
would obviously be absurd. Hence this measure of an urban population growing faster than the total would
urbanization can have only limited use. eventually come to exceed the total.
102. As regards the second measure (absolute rises, 107. All the alternative measures of urbanization
in the percentage of urban population), it is likewise tempo, discussed in the foregoing, have limitations either
evident that a trend, measured in these terms, cannot in terms of eventually absurd consequences or in terms
continue indefinitely, for eventually it would lead to the of common sense. This is not to say that they are use-
attainment and surpassing of a level of 100 per cent, less. In the analysis of past or current observations they
which is inherently impossible. may have much illustrative value. They may also serve
103. As regards the third measure (relative rises in the in simple population projections over limited future
percentage of urban population), an absurd result would periods during which unlikely eventual consequences
appear at an even earlier date, because then the rises in would not yet occur. The remote future is extremely
percentage level would accelerate in proportion to the uncertain, no matter what method of projection may be
levels attained at each future moment. used, and simplicity of future assumptions is in any event
104. Other measures, not illustrated here, can also be desirable. In many situations, and for limited periods,
imagined. One might measure the tempo of urbanization, the use of one of those measures need not be discouraged.
for instance, in terms of rural population growth, ordi- 108. By contrast it can be argued that the URGD
narily slower than the growth in urban and total popu- measure offers at least three types of advantage: the
lation. A constant rate of rural population growth, avoidance of absurd consequences, the compatibility
slower than the total, can in fact continue indefinitely, with common sense and the consistency with a logistic
or at least as long as total growth continues, for it would curve describing the rise of percentage in urban popu-
remain consistent with rising levels of urbanization with- lation. These three advantages are now discussed, each
out ever reducing the rural population to zero or below. in turn.
A constant amount of decrease in the percentage of
rural population, however, would produce an impossible 109. In the most diverse combinations of population
result, namely an eventual percentage of zero or less. trends, indefinite continuation of a given level of URGD
On the other hand, no necessary absurdity is involved never produces an absurd result. While the percentage
if the percentage of rural population is permitted to level of urbanization is low, the rural population increases
decrease at a constant rate: the absolute size of the rural almost as rapidly as the total population, and the corre-
population may eventually diminish-and this is not sponding urban population can increase considerably
contradictory to observations already made in highly more rapidly than the total; when the percentage level of
urbanized countries-but it would diminish by decreasing urbanization rises high, the urban population increases
amounts without ever completely exhausting a rural only slightly more rapidly than the total, and the rural
residual. population can increase considerably less rapidly, and
105. Two measures in terms of rural population, possibly decrease. While total population increases at
therefore, are theoretically compatible with possible a fast rate, urban population may grow with very great
developments up to an indefinite future. But it has speed, and rural population may also grow considerably;
to be admitted that the measurement of urbanization in when growth in total population slows down, the growth
terms of rural population change does not suggest a in urban population may also slow down, while rural
common-sense approach. Except under very special population growth may be much reduced, or give way to
conditions, unlikely to persist, it would be difficult to decrease. Such would be the consequences of an inde-
argue that the tempo of rural population change, by finite continuance of a given URGD, in diverse situations;
itself, determines how urbanization might progress. It is they are at least possible, and also in rough conformity
true, for instance, that in several Latin American coun- with observations actually made.
tries; with accelerating growth in total population, rural 110. In terms of common sense, the urbanization
populations in recent decades have grown at compara- process is influenced by comparative advantages or
tively constant rates, whereas urban growth was all the disadvantages perceived in both the urban and the
more accelerated. But these observations may merely rural environment, often referred to as relative "push" and

28
 "pull". This makes it reasonable to i m a ~ n ethe usually 114. The relationship between the assumption of a
greater attraction of urban over rural areas as a residual constant URGD and that of a logistic curve in the per-
force reflected in the difference of growth rates; this can, centage of urban population can be demonstrated as
of course, fluctuate within short time periods, but on an follows.
average is likely to change only very slowly within The general formula for a logistic curve is
extended periods of time. A temporary unusually rapid
transfer of people from rural to urban areas may produce 1
Yo =
an increased pressure on urban employment opportunities k abz ' +
and other facilities while reducing that on rural facilities, which can also be put in the form
and the opposite may happen in the contrary event.
Regulatory mechanisms may then operate to maintain
urban-rural growth differences more nearly at constant
levels over the long run.
in either case k is a constant, and e is the basis of natural
111. From another point of view, it can be argued that logarithms, leaving a and b to be determined. 63a
there are varying restraints on the rate of rise in the
percentage level of urbanization, depending on the level 115. Now, let To, Uo and Ro be the total urban and
attained at any moment. The level is likely to rise by rural populations at the beginning of a period (t = O),
smaller amounts per unit of time when it is either very Tt, Ut and Rt the same populations after t years, u and r
low or very high, than when it is in some intermediate the exponential rates of growth in the urban and rural
range. Rises in level, therefore, are apt to accelerate at population, and d the URGD, i.e. the difference between
first, to reach a maximum rate, and then to slow down the rates, namely u-r. Then, at any given moment
again. Perhaps the simplest curve in which such a Ut = Uoeut, and Rt = Roert, so that
relationship between given level and rate of rise can be
expressed is a logistic curve.
112. An analogy may help to suggest why rises in a Therefore,
growing proportion could generally be expected to be
slow at first, then more rapid and eventually again slow.
Let us consider the rates at which the percentage of
literates in a population (say, aged 10 years and over)
may rise. When the level is very low, only few literate
persons will be available to teach literacy to others, which is a simple formula for the logistic curve. In the
hence a considerable effort is then involved in raising
the general literacy level by any considerable amount. Ut
above, of course, 100- is the urbanization level (urban
When a substantial proportion of the population, perhaps Tt
about one-half, are literate, the remaining illiterates that as a percentage of total population). The formula holds
can be reached are still numerous, but so are also the true in the course of time though both u and r may vary
potentially available teachers, hence, at this level, progress with time, provided that the difference u-r remains
can be quite rapid. At a high level of literacy, most of constant. The formula is further simplified if we take
the remaining illiterates will be those difficult to reach, the logistic's point of origin (t = 0) at the point where the
whether because of their geographic remoteness, cultural urbanization level is 50 per cent. This is the point of
resistance or physical and mental handicaps; at such a inflexion, or of maximum rate of rise in the urbanization
level, a large effort can secure only a comparatively level, and about .thispoint the logistic curve is symmetrical.
small amount of additional progress.
Here, Uo = Ro, and the fraction - Uo is unity.
113. Similarly the growth of urban residence will The
depend both on urban facilities already in existence Ro
(these would be in the role of "potential teachers", to use formula then becomes
the analogy) and the comparative size of the rural reser- Ut 100 eat
100-
voir from which additional urban population can be
recruited (in the analogy, the remaining "illiterates").
Tt
=
1 eat , +
Though the subject matter is entirely different, varied t being positive where the urbanization level is higher
rates of rise in urbanization can be expected as logically than 50 per cent, and negative where it is lower. If it
as varied rates of rise in literacy, depending on levels is further assumed that d (the level of URGD, in per cent
already attained. 65 per year) is unity, the formula is reduced to
The notion that the interaction of "push" and "pull" factors
in the process of urbanization may tend to be reflected in the rise
of the urbanization level in accordance with a logistic curve has
also been discussed in another source, including some empirical Values of this curve ranging from less than one per cent
evidence. It was noted that other variables, such as school enrol- to more than 99 per cent are tabulated in annex I.
ment ratios or labour force participationratios may also beconsidered
from this point of view. M. Sivamurthy and K. V. Ramachandran,
"An empirical investigation into the evolution of certain demo- 53aFor methods of computation see Croxton and Cowden,
graphic variables", Journal of Social Sciences, Karnatak University Applied General Statistics, 2d. ed. (Englewood Cliffs, N.J., Prentice,
(Dharwar), vol. IV, April 1968. Hall Inc.), p. 310.

19
 116. An additional observation is worth recording, TEMPOOF URBAN CONCENTRATION
making it possible to determine the value of ed or, in
other terms, (1 + URGD), without use of the logistic 119. Generalizing further from the observations already
made, we may also subdivide the urban population into
reference table. This can be shown as follows. One
of the formulas in paragraph 115 was two subpopulations, for instance the population of cities
(larger than some minimum size, or a given list of cities)
and that of towns (smaller than that size, or all urban
localities outside a given list). The percentage of city
from which it also follows that population in the combined urban population, at any
Ut Un given time, may be called the level of concentration of
the urban population, and increases (possibly also declines)
in that level may be considered as measurable in terms
of a tempo of concentration (or, in the event of decline,
deconcentration). It is evident that similar types of
expressing the same relationship in terms of proportions relationships can be assumed to exist between city
of the total population urban and rural, instead of population and towns population within the combined
absolute numbers urban and rural. By cross-multi- urban population, as between urban population and rural
plication from both sides of the equation we obtain both population within the combined total population.
that 120. It is evident also that one can subdivide the urban
Ut .Ro = edt, population into that of the country's largest, or capital,
Rt . Uo city and the remainder of the urban population. Addi-
tional steps are possible, making it feasible to measure
and that successively the tempos of concentration for each of
a list of cities, for instance beginning with the largest
city, then, after eliminating it from the total, proceeding
to the next largest city and so forth. One may also deal
similarly dividing a single city into segments, for instance
distinguishing its core and its suburbs.
In either instance the exponential growth-difference rate ed 121. Where the URGD may serve in the projection
is obtained as the t-th root of the expression to the left of urban, relative to an already projected total, popu-
side of the equation. This mathematical relationship lation, an analogous type of measurement can serve
may in some instances find a useful practical application. likewise in the projection of city, relative to an already
The extraction of the root, of course, will usually have projected urban population. If individual cities are taken
to be performed with the use of conventional logarithms. up successively, this leads to the possibility of projecting
117. To provide a simple numerical example, let us their populations individually. The advantage of the
suppose that a population is exactly 20 per cent urban URGD type of measurement is again that absurd results
at the date of one census, and exactly 25 per cent urban can be avoided, and that projections for individual cities
at the next census, taken ten years later. In this case remain consistent with a projection for the combined
urban population.
122. Where such detailed use is made of this method
t = 10. We find that, in this case, ed is the 10th root of of measurement it is necessary, of course, to bear in
mind how the urban population and that of individual
0.25 x 0.8
i.e. the tenth root of - Oe2 or the tenth cities are defined, and whether the definitions are com-
0.75 x 0.2 ' 0.15 ' parable in their geographic rigidity or flexibility. It
root of 1.333333. With the help of logarithms we find would be unreasonable, for instance, to compare the
that this comes to 1.0292, and we can say that the URGD, growth of cities larger than, say, 100,000 inhabitants with
as an exponential rate, comes to 2.92 per cent per year. the growth of the urban population of a country if the
118. As a final note, it is admitted that the assumption number of such cities is small: the fortuitous attainment
of the 100,000 size limit by an additional previously
that urbanization may ordinarily proceed according to smaller, city is then a discontinuous event, by which such
a logistic curve is arbitrary, and that it commends itself measurement might be easily upset. In a large country
chiefly because of its simplicity of application. If there having numerous cities of such size, the inclusion of
were more knowledge on the subject one might perhaps additional cities attaining that size is a fairly continuous
postulate, that urbanization tends to progress, say, like process, and there is no incongruity in assuming that an
a normal probability curve (which resembles the logistic
rather closely), or like some asymmetric curve, whether observed tempo of concentration, for such cities within
an asymmetric logistic, a modified exponential or a the combined urban population, can continue.
Gompertz curve. Any of these curves, and perhaps some 123. On the other hand, the boundaries of some cities
others as well, might be compatible with the observations may have been widened, while in other cities this was not
made so far. But the fact is that our knowledge is not the case. In such instances it would not be reasonable to
so precise, hence it is advisable to seek the solution to assume that cities of the first type will continue to grow
the problem in the simplest terms. faster than cities of the second type. The opposite may
happen. Cities with recently widened boundaries may in large cities or in individual cities of the urban popu-
for some future time experience few territorial additions, lation may also fluctuate in time, depending on local
whereas in cities where boundary changes have not developments, for instance government policies affecting
occurred recently the likelihood of this happening in the geographic distribution of investments which, because
the near future may be greater. Such considerations of resulting employment opportunities, may cause the
have implications for the validity of calculated differential population of some cities to grow faster than that of some
rates of the URGD type. The tempo of concentration others, at least at certain times.
 Chapter IV
PROJECTION OF URBAN AND RURAL POPULATION TOTALS
USING THE SIMPLEST METHODS

leave the following future estimates of rural population

as residuals :
124. The simplest projection is, perhaps, that resulting Urban rate of increase (assumed)
from an assumption of a constant rate of growth. In
Year At 5 per cent At 6 per cent At 7 per cent
the examples which follow it will be assumed that the
urban population grows at a constant (annual) rate. It 1965 .................. 10,997 10,997 10,997
will be assumed, however, that a projection of the corre- 1970 .................. 12,372 12,330 12,286
sponding total population already exists, and that in the 1975 .................. 14,047 13,938 13,818
latter projection the rate of growth in (total) population 1980 .................. 16,067 15,853 15,607
is not necessarily constant. The urban population, as 1985 .................. 18,490 18,086 17,667
projected by a constant rate, will then be subtracted from
the total population, as projected by other means, to 128. Depending on the urban rate of increase, the
ascertain whether the growth in the residual, i.e. in the rural population would grow at the following average
hypothetical rural, population remains plausible. rates :
Urban rate of increase (assumed)
125. Three examples have been selected for an examin-
Years At 5 per cent At 6 per cent At 7 per cent
ation of how well the simplest mathematical methods
may apply to populations of low, intermediate and high
levels of urbanization. The examples concern the
populations of the United Republic of Tanzania, Iran
and Canada, and use is made of the latest estimates and
projections calculated in the United Nations Secretariat. 64
126. To take first the case of the United Republic of There is nothing implausible in the implied future rates
Tanzania, utilizing the United Nations projections of of growth of rural population hence, in a country like
total population and a 1965 estimate of the size of the the United Republic of Tanzania, to assume a constant
urban population, it will be assumed that the urban popu- growth rate in urban population over a stretch of twenty
lation may grow in the future at rates of 5, 6 or 7 per cent years is not necessarily unrealistic.
per year, resulting in increases by 27.62,33.82 or 40.32 per 129. Using the same methods (assuming somewhat
cent, respectively, per five-year period. Estimated as lower rates) in the case of Iran, we obtain the following
677,000 in 1965, the urban population would increase as projections of urban population :
follows (figures in thousands): Urban rate of increase (assumed)
Year At 4 per cent At 5 per cent At 6 per cent
Urban rate of increase (assumed)
Year At 5 per cent At 6 per cent At 7 per cent
1965 .................. 9,172 9,172 9,172
1970 .................. 11,159 11,706 12,274
1975 .................. 13,577 14,940 16,425
1980 .................. 16,518 19,068 21,980
1985 .................. 20,097 24,336 29,414

130. Here, the total population was estimated as

24,549,000 in 1965, and projected to 28,358,000,
127. The total population of the United Republic of 33,152,000, 38,769,000 and 45,050,000 for the subsequent
Tanzania has been estimated as 11,674,000 in 1965 and dates, By subtraction, we obtain the following implied
has been projected by the component method to projections of rural population :
13,236,000, 15,150,000, 17,475,000 and 20,282,000 for Urban rate of increase (assumed)
1970, 1975, 1980 and 1985, respectively, that is at succes-
Year At 4 per cent At 5 per cent At 6 per cent
sive annual average rates of 2.54, 2.74, 2.90 and 3.02 per
cent. The above projections of the urban population,
when subtracted from the projected total populations,
64 World Population Prospects as Assessed in 1968 (United Nations
publication, Sales No. 72.XIII.4).
 131. The following rates of increase in the rural 134. In conclusion, the assumption of a constant urban
population are implied : growth rate remains useful only so long as the level of
Urban rate of increase (assumed) urbanization is rather low. At intermediate or higher
Years At 4 per cent At 5 per cent At 6 percent
levels such an assumption soon tends to become unreal-
istic.

135. The same three examples are now worked through

in terms of assumed rates of growth in rural population.
First, for the United Republic of Tanzania it will
As can be noted, the implied rates of growth in rural be assumed that the rural population may grow at
population are far apart, hence at this level of urban-
ization the rural population is quite sensitive to the the constant rates of 2.0, 2.5 or 3.0 per cent. The
following future estimates of rural population are
particular assumption concerning urban growth. It is obtained :
surprising, furthermore, that the continuance of a high
rate of growth in the urban population may soon result Urban rate of increase (assumed)
in absolute decreases in the rural population. This makes Year At2.0percent At2.5percent At3.0percent
it evident that at an intermediate-level of urbanization
the assumption of a constant rate of growth in the urban
population should not be carried very far. It is more
likely then that, with time, urban growth will slow down
somewhat.
132. Taking, finally, the case of Canada, as a country
at a very high level of urbanization, we shall assume that 136. Subtracting from the projected total population,
the urban population may grow at rates of 2.0, 2.5 or we obtain the following implied projections of urban
3.0 per cent, increasing by 10.41, 13.14 or 15.93 per cent population :
in each five-year period. The urban population is Rural rate gf increase (assumed)
estimated at 14,333,000 in 1965, and we obtain these Year A t 2 . 0 p e r c e n t At 2 . 5 p e r c e n t A t 3 . 0 p e r c e n t
results :
Urban rate of increase (assumed)
Year A t 2 . 0 p e r cent A t 2 . 5 p e r c e n t A t 3 . 0 p e r cent

137. Following are the implied annual rates of urban

growth:
Rural rate of increase (assumed)
133. The total population is estimated at 19,644,000 Years At2.Opercent At2.5percent At3.0percent
in 1965 and projected to 21,426,000, 23,284,000,
25,299,000 and 27,348,000 at those future dates. Conse-
quently, the following future rural populations areimplied :
Urban rate of increase (assumed)
Year A t 2 . 0 p e r cent At 2 . 5 p e r c e n t A t 3 . O p e r c e n t
The constraints on the assumed rate of growth in rural
population, at this low urbanization level, are noticeable.
Urban population in the United Republic of Tanzania
can be expected to grow at a fast rate, though perhaps not
so fast as 9 per cent per year. This makes the rural
rates of 2.0 and 3.0 per cent both rather improbable.
Most likely, a rural rate of growth between 2.0 and 2.5
In the circumstances of Canada it is unlikely that the per cent can be sustained. For instance, if the rural rate
rural population will increase much, hence the assumption were a constant 2.25 per cent, urban population would
of a constant 2 per cent growth in urban population is not increase in the successive quinquennia at rates of 7.9,
very probable. But it is just as unlikely that the rural 8.5, 8.5 and 8.0 per cent. Provided that the rate of
population will decrease at a steep and accelerating rate, growth in rural population is selected with care, therefore,
soon dwindling to quite small numbers. Therefore, to the assumption of constant growth rates in the rural
assume a 3 per cent growth in urban population is also population need not lead to implausible results for the
unrealistic. It can be seen that, at this high level of urban population. Admittedly, this is a trial-and-error
urbanization, the constraints for a plausible assumption procedure, and it would be more logical to base the
of a constant rate of increase in the urban population projection directly on an assumed rate of growth in the
become narrow. urban population, as was done before.
 138. Next, we take again the example of Iran. Here There may be many situations in which such an assump-
it will be assumed that the rural population may grow tion produces satisfactory results. It is understood, of
at constant rates of 1.5, 2.0 or 2.5 per cent. The follow- course, that the rural population increases more slowly
ing future rural populations are obtained:. than the urban.

Rural rate of increase (assumed)

Years At 1 . 5 p e r cent At 2 . 0 per cent At 2 . S p e r cent 143. The ratio method rests on the assumption that
an observed trend in the percentage of a subnational
1965 .................. 15,377 15,377 15,377
population to a country's total population will continue.
1970 .................. 16,566 16,978 17,398
1975 .................. 17,847 18,745 19,684 It is a highly practical method in the estimation and
1980 .................. 19,227 20,696 22,270 projection of regional or provincial populations in
1985 .................. 20,713 22,850 25,196 relation to estimated and projected national totals. The
method is also useful in the projection of urban and rural
139. By subtraction from projected total populations, populations, subject to certain constraints, as will be
the implied urban populations are the following: shown.
144. The case of Iran is taken first. Here, according to
Rural rate of increase (assumed) United Nations estimates, 27.13 per cent of the total
Year AtI.5percent At2.0percent At2.5perrent
population was urban in 1950, and 33.18 per cent in
1960. Thus, the percentage had risen by 6.05 points
in ten years, and it may go on rising by 3.025 points
every five years. Projecting these percentages linearly
and using the United Nations projection of the total
population, we obtain the following projections of urban
and rural population (in thousands):
140. The implied rates of growth in urban population Per cent
Year Total urban Urban Rural
are as below:
Rural rate of Increase (assumedl
Years At1.5percent At2.0percent At2.5percent

145. The implied annual rates of increase in urban and

rural population are the following:
As can be seen, none of these results are necessarily
implausible. Under any of the three assumptions Years Urban Rural

regarding rural growth, a fairly steady rate of growth in

urban population is obtained, first accelerating and then
slowing down.
141. Coming finally to the example of Cahada, we
shall assume that the rural population remains constant,
that it decreases annually by 0.5 per cent or that it There is nothing implausible in such rates of growth of
decreases by 1.0 per cent. The following rural popu- the urban and rural population, and the projection may
lations are thereby projected : be acceptable.
146. Now, let us use the same method with respect to
Assumed rural rate of increase
(or decrease)
the United Republic of Tanzania. Here, according to
the estimates, the population was 3.50 per cent urban in
Year AtO.Oper cent At -0.5percent At -1.Oper cent 1950 and 4.98 per cent in 1960. The level had risen by
1.48 in ten years and it may rise by 0.74 every five years.
In conjunction with the projected total population, we
obtain the following urban and rural projections:
Per cent
Year Total urban Urban Rural

It is not necessary to adduce the implied urban popu-

lations and their rates of growth, as it can be readily
seen that at this high level of urbanization no implausible
result will be implied.
142. It can be concluded that constant rates of change
in the rural population have a somewhat wider applica- 147. In these projections the annual rates of growth in
bility than constant rates of change of urban population. urban and rural population would be the following:
 Years Urban Rural urbanization level, one of intermediate level and one of
advanced level, some general conclusions emerge.
151. At the low level of urbanization, the most ade-
quate of the three methods appeared to be the one employ-
ing constant rates of growth in the urban population.
At the intermediate level, the ratio method seemed to
Again, the results may be acceptable. One is led to be most satisfactory; and at the advanced level it appeared
wonder, however, whether the rise in the percentage of best to make plausible assumptions concerning the
urban population, at this low level of urbanization, growth (or decline) in the rural population. Not all
would not have a tendency to accelerate, as higher and three methods can fit any given situation equally well.
higher levels of urbanization are being reached. It should 152. One is led to think that some modification of
be recalled that in Iran the percentage level was estimated the ratio method may have the widest applicability if it
to rise 3.025 points every five years, whereas in the makes provision for an acceleration in the rise of the
United Republic of Tanzania it rises only 0.740 points. urbanization level while it is low, a peak rate while it is
148. Moving now to the example of Canada, we note intermediate, and a slow-down when it is high. These
that 60.66 per cent of the population was estimated urban conditions can be met in a logistic curve, or some other
in 1950, and 68.45 per cent in 1960, a rise of 7.79 points curve having both an upper and a lower asymptote.
in ten years, or 3.895 points in five years. Using the As was shown in chapter 111, the assumption of a constant
projection of total population, we obtain these results: difference between urban and rural rates of growth
coincides with a logistic curve in the level of urbanization
Per cent within the outer limits of zero and 100 per cent.
Year Total urban Urban Rural
153. It would be difficult, however, and often impos-
sible, to calculate the several parameters of a complex
curve on the basis of limited past observations; and it
would be inadvisable to make a forecast depend on
parameters easily affected by temporary fluctuations or
accidents. True, urbanization throughout the world
progresses in the same direction, but the phenomenon is
It can be noted that in this projection the rural population subject to variations in time and place which often elude
decreases at a rapidly accelerating tempo. Furthermore, measurement or prediction. To arrive at judgment it
the constant rise in the percentage cannot continue very is important to bear in mind pertinent observations made
long because then 100 per cent would be reached and in numerous countries.
surpassed already in the year 2001, which is absurd. 154. It should also be recalled that the use of a mathe-
At such a high level of urbanization, therefore, the use of matical curve courts the risk of misinterpretation on
the ratio method encounters a more severe limitation. the part of the users of a forecast. It may seem to them
It is evident that the rise in the ratio must eventually that a forecast pretends to be exact because a precise
slow down. formula was used. It will have to be stated clearly that
149. In conclusion, the ratio method yields acceptable the use of a formula does not imply accuracy in the
results at intermediate levels of urbanization, but it results. Future developments will never conform to a
should not be used where the urbanization level is very mathematical artefact. To guard against misinter-
low or very high. pretation, the assumptions should be presented flexibly
and with an allowance for an error range. This is
probably best done by calculating "high" and "low"
alternatives in addition to a "medium" forecast.
150. Three methods have been examined, namely the 155. A curve lending itself readily to comparison of
method employing constant rates of growth in the urban observations throughout the world and to flexible appli-
population, that using constant growth (or decline) in cation in alternative forecasts is therefore useful. The
the rural population, and the ratio method. From the following chapter will show the ease with which a logistic
observations made with estimates for a country of low curve can be applied.
 Chapter V
UNITED NATIONS METHOD OF URBAN AND RURAL POPULATION PROJECTIONS

ACTUALOBSERVATIONS CONCERNING the world for 1960-1970,65 now being revised on the
URBAN-RURAL GROWTH DIFFERENCES basis of new data.
157. It will be noted that levels of urbanization in
156. As already explained, the United Nations method 1960 were distinctly higher in more developed regions,
of measuring the tempo of urbanization and projecting ranging from 44 to 80 per cent, than in less developed
it is based on the difference between urban and rural regions, where the range was from 2 to 46 per cent.
population growth and its logistic transformation. For Rates of population growth were distinctly lower in more
an appreciation of possible magnitudes in this measure-
U R for ~short--reference is made to table 11, presenting 66 ~ ~ B ~ ~ Iof&tisks,
I ~~ ~ ~ hNovember
~ l 1971
~ (United Nations
estimates of urbanization level in twenty-four regions of publication), special tables B-I1 and B-111.

TABLE11. LEVELS
OF URBANIZATION, 1960 AND 1970, RATES OF GROWTH IN URBAN AND RURAL
POPULATION,
1960-1970, AND DIFFERENCES BETWEEN THESE RATES, IN NINE MORE DEVELOPED AND
FIETEEN LESS DEVELOPED REGIONS OF THE WORLD

Percentage Annual rates of growth,

urban population 1960-1970 (per cent)
-
Region 1960 1970 Urban Rural Difference

More developed regions a .......... 61.0 67.5

Australia and New Zealand ..... 79.9 83.4
Northern Europe .............. 72.3 76.2
Temperate South America b ..... 71.1 78.3
Northern America ............. 69.7 74.3
Western Europe ............... 68.8 74.4
Soviet Union .................. 49.5 57.1
Eastern Europe ................ 48.8 54.0
Southern Europe .............. 45.0 50.8
Japanc ....................... 44.3 58.9
Less developed regions a ........... 24.6 29.6 4.7 1.8 2.9
Tropical South America d ....... 45.8 55.5 5.0 1. O 4.0
Middle America ............... 44.9 51.4 4.8 2.1 2.7
Southern Africa. ............... 41.7 45.6 3.2 1.6 1.6
Caribbean .................... 36.7 41.6 3.6 1.5 2.1
Other East Asia e .............. 32.9 42.7 5.3 1.2 4.1
South West Asia .............. 31.5 37.4 4.5 1.9 2.6
Northern Africa ............... 29.9 34.9 4.4 2.1 2.3
China ........................ 18.1 25.5 4.9 1 .O 3.9
South East Asia ............... 17.5 21 .O 3.7 2.3 1.4
Middle South Asia ............. 16.9 18.9 3.8 2.4 1.4
Polynesia and Micronesia ....... 15.8 20.1 5.4 2.7 2.7
Western Africa ................ 15.3 19.9 5.4 1.9 3.5
Middle Africa ................. 11 .5 16.5 5.7 1.4 4.3
Eastern Africa ................ 7.5 10.1 5.6 2.1 3.5
Melanesia .................... 2.3 3.1 5.5 2.8 2.7

SOURCE: Monthly Bulletin of Statistics, November 1971 (United Nations publication). A few regions
have been regrouped as indicated in foot-notes.
a Unweighted average.
* Not including Paraguay.
"Urban" population defined as that of densely inhabited districts; including former Ryukyu Islands.
a Including Paraguay.
Republic of Korea. Democratic People's Republic of Korea. Hong Kong. Mongolia and Macao.
developed than in less developed regions. Thus, urban 1800-1900. The long continuation of urbanization with
population grew at rates between 1.2 and 3.1 per cent in nearly the same momentum (as measured by URGD)
the first group, and between 3.6 and 5.7 per cent in the is impressive, and it can be concluded that long-run
second group of regions. In more developed regions, urbanization trends have great inertia. Within shorter
the rural population declined (or did not change much, time periods, however, this process can fluctuate.
as in Australia and New Zealand), while in less developed 162. A long period of short-time changes can be
regions, rural population grew at rates between 1.0 and observed in the decennial census data of the United
2.8 per cent per year, and this range is similar to the rates States (see table 12). A break in the series occurs
of growth of urban population in the more developed between 1940 and 1950, as the urban population was
regions. then redefined to take into account some of the more
158. Despite these differences in urbanization level and recent developments resulting mainly from suburban-
growth rates, however, the differences between urban and ization; the difference between the two rates, urban and
rural growth rates in more developed and less developed rural, therefore is perhaps too low in 1930-1940 (redefi-
regions are of the same order of magnitude. In the nition having become due) and perhaps too high in
first group of regions, URGD ranges from 2.1 to 3.9 per 1940-1950 (an overdue redefinition having at last been
cent, and in the second group, from 1.4 to 4.3 per cent. adopted). There are also reasons to doubt the accuracy
The unweighted average in both groups is nearly the of the 1870 census, taken after the Civil War.
same, 2.8 and 2.9 respectively. One can say that regard-
less of levels of development, levels of urbanization or
the tempo of population growth, URGDs between about TABLE12. RATESOF OROWTH IN URBAN POPULATION, IN RURAL
2.0 and 4.0 are now typical throughout the world. Where POPULATION, AND DIFFERENCES BETWEEN THE TWO RATES, IN THE
there is no other detailed knowledge, projections may be UNITEDSTATES, 1800-1960 (PERCENT PER YEAR)
made with reference to these observed ranges.
Annual growth rate Difference
159. For the combined world population, the URGD between the
is smaller as a result of distortions affecting the weighted Period Urban Rural two rates
averages. In 1960-1970, the more developed regions
still included a majority of the world's urban population, Decades
1800-1810 ..........
but in these regions rates of population growth were 1810-1820 ..........
comparatively moderate; the great majority of the 1820-1830. . . . . .. .. .
world's rural population is in less developed regions 1830-1840. . . . . .. .
..
where population growth is rapid. During 1960-1970,
1840-1850. . . . . . . . . .
according to the estimates, the world's urban population 1850-1860. . . .. . .
. ..
grew from 986 to 1,358 million, and the world's rural 1860-1870a ........
population from 1,995 to 2,277 million, representing 1870-1880a ........
an urban rate of 3.3 per cent, a rural growth rate of 1880-1890 ..........
1.3 per cent and an URGD of only 2.0 per cent per year. 1890-19OO..........
This represents a slight acceleration when compared with 1900-1910 ..........
world-wide estimates for the nineteenth and the earlier 1910-1920. . .. .
. . .. .
part of the twentieth century. 1920-1930 ..........
160. Davis and Hertz have estimated the world's .
1930-1940 b . . . . . . .
population, and the world's urban population (defined 1940-195Oc ........
as localities with at least 5,000 inhabitants) in 1800 and 195@-1960 ..........
1900. According to them, the urban population grew 40-year periods
in that century from 27 million to 219 million, and the 1800-1840 ..........
remaining, or rural, population from 879 to 1,389 million. 1840-1880 ..........
Average growth rates of the nineteenth century therefore 1880-1920 ..........
amount to 2.1 per cent for the urban population, and 1920-1960C ........
0.5 per cent for the rural population, leaving an URGD
of 1.6 per cent. In view of the distorting effects already a Probable undernumeration of rural population in 1870,
8 Period of economic depression,
mentioned in the combined world figures for 1960-1970, e Including change-over to a new definition of urban population
URGD in individual countries or regions may for the which had become necessary by 1950.
most part have been higher.
161. Linking the estimates of Davis and Hertz for 163. Over the entire 160-year period here considered,
1900 with our present estimates for 1960, we arrive at an the difference between urban and rural rates of increase
average 1900-1960 urban growth rate of 2.5 per cent, a averages 2.4 per cent per year, including both favourable
rural growth rate of 0.8 per cent and an URGD of 1.7 periods and others which were unfavourable to rapid
per cent, only slightly more than the world estimate for urban growth. The average URGD was as high as
3.0 per cent during 1840-1880. It is noteworthy that
66 See M. Hauser, ed., Urbanization in Asia and the Far East, during 1920-1960, a period when the level of urbanization
Proceedings of the Joint United Nations/UNESCO Seminar (in co- was high, the difference between the two rates was no
operation with the ILO) on Urbanization in the ECAFE Region, less than it had been during 1800-1840, a time when
Bangkok, 8-18 August 1956, Tensions and Technology Series No. 7
(Calcutta, 1957), pp. 56-57. urbanization was still at a very low level.
 164. It is evident that considerable fluctuations can and the final formula is
occur in individual decades. Nevertheless, even at
widely different urbanization levels, the span between urban
and rural rates of population growth can be quite similar.
Where no adverse conditions retard the tempo of The rural population is then obtained by subtraction
urbanization, both across the world and in the long time of U' from T'. The computation can be repeated for
perspective, a 2.5 or 3.0 per cent excess of urban over each single year of the projection period.
rural growth rates appears to be fairly normal, while
growth differences between 2.0 and 4.0 per cent need not YEAR-TO-YEAR
INTERPOLATION OF
be regarded as unusual. A FIVE-YEARLY PROJECTION
165. These general observations, however, do not 169. There should be little problem in the interpolation
dispense with the need to arrive at an independent of a projection of the total population by individual
judgement as regards the situation in each particular calendar years. The easiest method is perhaps the
country, considering the probable or possible influence graphic one, where the pro-jected population, at five-
of varying economic and social conditions or policies. yearly intervals, is plotted, and the points are connected
It is an advantage, and not a defect, of the method that free-hand or with the use of French curves to obtain a
varied assumptions of growth differences can be drawn smooth transition among gradually changing growth
up quite freely. rates for the intervening individual years.
170. The use of a simple mathematical interpolation
can also be recommended. The population of Iran, for
166. The urban and rural population growth rates, and instance, has been projected for five-yearly intervals with
the difference between them, can be calculated either as these results :
instantaneous (exponential) or annual (compound inter-
est) rates. Accordingly, there are two alternative modes Year Population (in thousands)

of application of the growth-difference method for pro- 1965 .......................... 24,549

jections of urban and rural population. In either instance 1970 .......................... 28,358
it is assumed that a projection of the national total 1975 .......................... 33,152
1980 .......................... 38,769
population already exists. Often that projection may 1985 .......................... 45,050
be available by five-year time intervals only.
167. If five-year time intervals only are required for the A series of this type can usually be simply interpolated
urban and rural projection, then the recommended method with the use of first and second differences, as shown in
may very well be that of instantaneous rates of growth, table 13. The detailed procedure is lengthy to explain
for direct use can then be made of the logistic curve in a text, but will be quite evident as described in the
tabulated in annex I. But if projections of urban and foot-notes to the table. This procedure is known as the
rural population are desired for a series of individual "central difference method".
calendar years, there may sometimes be an advantage
in the use of annual rates, applied by a simple formula
year by year. But then it is also necessary to refer to
a projection of total population by individual calendar
years. If that initial projection proceeds by five-year 171. For Iran, an urban population of 9,172,000 has
intervals only, a series of year-by-year interpolations is been estimated for 1965. Urbanization has been pro-
also needed. ceeding speedily in Iran, hence the URGD may perhaps
168. The formula for the year-by-year projection of be put at 3.0 per cent on a medium assumption. Alter-
urban and rural population in relation to an existing native projections may put it as low as 2.0 and as high
projection of total population, employing annual rates as 4.0 per cent. In applying the formula
of growth, can be derived as follows.
Let T, U and R be the total, urban and rural population
for the year t, T', U' and R' the same populations for
U' = ( T;'dR )U, we obtain the following three series
the year t+ I, u and r the urban and rural rates of increase, implying a high, medium or low tempo of urbanization,
and d the difference between them; then the following respectively. 67
relationships obtain :
+
T = U R, and T' = U' + R', 67 To illustrate the first step of computation assuming the "high"
tempo, we have the following quantities: T = 24,549, T' = 25,225,
U' = U(l +u) = U(1 +r+d), and R' = R(l +r), U = 9,172, R = 15,377 (i.e. 24,549 minus 9,172) and d = 0.04
whence (high assumption).
T' = U(l+r+d) + R(l+r) = (U+R)(l+r) + Ud =
(-
T' ;dR )U we have
so that
T(l+r) + Ud,
Substituting in the formula U' =

T' - Ud = T(l+r).
Therefore, Subtracting U' from T', we obtain
R' = 25,225 - 9,654 = 15,571;
likewise for each successive year of the projection and for alternative
assumptions.
 TABLE13. ANNUALINTERPOLATION FOR A QUINQUENNIAL SERIES BY MEANS OF FIRST AND SECOND DIFFERENCES

Mean Mean Second First

first second difference difference First trial. Adiustments. Results.
Year Data difference difference interpolated interpolated results interpolated adiusted
(1) (2) (3) (4) (5) (6) (7) (81 f 9)

a Equals 115 of (28.358 . .

24.549) Other first differences are calculated in the same manner .
6 Equals 115 of (959 - 762). Other second differences'are calculated in the same manner .
Extrapolated. in these instances equal 39 +(39.33). and 27 +
(27.33) .
6 Trial results.
Difference between trial result (in parenthesis) and original data (in italics). 28. 358 - 28. 354 in this instance .

URBAN
AND RURAL POPULATION OF IRAN.1965.1985. ACCORDING TO A PROJECTION OF TOTAL POPULATION
AND THREE ASSUMED TEMPOS OF URBANIZATION

"High" tempo (4 per cent) "Medium" tempo ( 3 per cent) "Low" tempo ( 2 per cent)
Year Projected total Urbarz Rural Urban Rural Urban Rural
 172. In accordance with these projections, the follow- these figures. A more detailed knowledge of relevant
ing percentage levels of urban population would be circumstances in Iran, however, might give cause to
attained at future dates: modify such a tentative judgment.
Year "High" tempo "Medium" tempo "Low" tempo 175. One might wonder whether the same method also
leads to acceptable results when the level of urbanization
is either very low or very high. It has therefore also been
applied to the urban and rural populations of the United
Republic of Tanzania and those of Canada. Again,
URGD has been assumed as 4 per cent for the "high",
3 per cent for the "medium", and 2 per cent for the "low"
A noteworthy observation can here be made. At the projection. The calculations were also carried out from
low tempo (URGD = 2 per cent), a level of urbanization interpolated figures by individual calendar year. For
(46.8) is obtained after 20 years, which at the high tempo brevity, the results are shown for five-year intervals only.
(URGD = 4 per cent) is obtained after 10 years. At the 176. In the case of the United Republic of Tanzania
medium tempo (URGD = 3 per cent), the level of (low level of urbanization), a relatively wide divergence
urbanization after 20 years (51.6) is that attained at the in projected urban populations is obtained, depending on
high tempo (URGD = 4 per cent) after 15 years. the assumed URGD. Relative to its size, the rural popu-

Average annual rate of growth

"Highh"tempo "Medium" tempo "Low" tempo
- .
Years Urban Rural URGD Urban Rural URGD Urban Rural URGD

Evidently, the time taken to attain a given level of urban- lation is projected within narrower limits, being more
ization is inversely proportional to the tempo in URGD. 58 closely dependent on the projection of the total popu-
173. On each of the three assumptions, the quinquen- lation. The following levels of urbanization are obtained
nial gains in percentage levels would be by nearly equal (per cent of total):
amounts in each five-year period. Accordingly, the Year "High" tempo "Medium" tempo "Low" tempo
simple method might have yielded fairly similar results.
As indicated by the results for the "medium" tempo, the
use of a constant rate of growth in the rural population
might also have yielded satisfactory results. Results
will differ to a greater extent from those of the simpler
methods when the level of urbanization is either very
low or very high, not intermediate as in the example given.
It can be seen that the level of urbanization rises by
174. As a further check on the consistency of these increasing amounts. This is reasonable where urbaniza-
projections, one may also consider average annual rates tion is still in a phase of increasing momentum. As
of growth (per cent) in urban and rural population, shown on the next page both the urban and rural popula-
respectively, resulting from the projections, as shown tions increase at accelerating rates, but the acceleration
above. There is nothing inherently implausible in is moderate. One might have obtained similar results
by assuming rates of growth in the urban population,
The findings are slightly inexact when made on the basis of such as 5.0, 6.0 and 7.0 per cent per year, respectively,
annual rates. With exponential rates these findings would be
mathematically precise. as was already done in chapter IV.

URBANAND RURAL POPULATION OF THE UNITEDREPUBLIC OF TANZANIA, 1965-1985, IN FIVE-YEAR

INTERVALS OF TIME, ACCORDING TO A PROJECTION OF TOTAL POPULATION AND THREE ASSUMED
TEMPOS OF URBANIZATION (URGD)

"High" tempo "Medium" tempo "Low" tempo)

(4 per cent) (3 per cent) ( 2 per cent)
Projected
Year total Urban Rural Urban Rural Urban Rural
 Average annual rate of growth
p~

"High" tempo "Medium" tempo "Low" tempo

Years Urban Rural URGD Urban Rural URGD Urban Rural URGD

177. In the case of Canada (high level of urbanization), The method may be equally useful in the projection of
the relative divergence of results is most marked in the urban and rural populations among provinces or regions
projected rural population, as shown below. of the same country, assuming that provincial or regional
178. In these projections, the rise in the percentage projections of total population already exist.
level of urbanization gradually slows down, as can be 181. The method raises, however, an important
noted in the following figures: problem. In the foregoing examples, a URGD of 4 per
cent was assumed as a "high", 3 per cent as an inter-
Year "High" tempo "Medium" tempo "Low" tempo mediate, and 2 per cent as a "low" assumption. These
three tempos can be related to world-wide average
observations. In the particular countries in question,
however, developments may be of a kind deviating more
widely from those average circumstances. Assumptions
must then be drawn up independently, in the light of
detailed knowledge concerning each particular country.
This slowing-down in the rise of level is necessary at Past trends noted in the same country provide an impor-
high levels so that an absurd result of 100 per cent or tant background for reference.
more will never be attained. 182. When applying the same method to a wide
179. The average annual increases or decreases in variety of examples, one arrives invariably at the follow-
urban and rural population are as shown on the next page. ing experience:
It can be noted that the urban population increases at (a) With any given URGD, the percentage level of
diminishing rates and the rural population decreases at urbanization rises in the manner of a logistic curve, with
increasing rates, except in the "low" tempo where the a continuous acceleration at the low levels, a point of
rural population also increases at a diminishing rate. inflection about 50 per cent and a continuous slow-down
The sensitivity of the rural population to the assumptions thereafter;
selected is evident, since the projection of the urban (b) These rises in percentage level are independent
population depends ever more closely on the projection of the rate ,of growth in total population; and
of the total population. (c) The time taken to attain a particular urban-
180. As can be seen, the URGD method has the merit ization level, starting from the same initial level, will be
of equal applicability under a wide range of conditions inversely proportionate to the URGD.
and is therefore suitable for international comparisons. 183. With the use of annual rates of growth, these
features are obtained to a high degree of approximation;
The meaning to be attached to the rural trend may also have with the use of instantaneous (exponential) rates, the
to be judged in the light of the changing features of the "rural observations are precise because, as already explained in
population". There is probably a tendency towards suburbaniza-
tion in areas of residence sometimes defmed as "rural", depending chapter 111, the curve in the level of urbanization is then
on local administrative circumstances. an exact logistic.

URBANAND RURAL POPULATION OF CANADA,

1965-1985, IN FIVE-YEAR INTERVALS OF TIME,
ACCORDING TO A PROJECl'ION OF TOTAL POPULATION AND THREE ASSUMED TEMPOS OF URBAN-
IZATION

"High" tempo "Medium" tempo "Law" tempo

( 4 per cent) ( 3 per cent) ( 2 per cent)
Projected
Year total Urban Rural Urban Rural Urban Rural
 "High" tempo "Mediuin" tenrpo "Low" tenrpo
Years Urban Rural URGD Urban Rural URGD Urban Rural URGD

actual dates of the past (e.g. the interval between the two
censuses) and the interval in the fictitious time scale of
Ut the table. For instance, the population of Colombia was
184. A table of the logistic curve, 100- =
Tt 29.1 per cent urban at the census of 5 August 1938, and
is tabulated in annex I. A computation formula for the 38.0 per cent urban at the census of 9 May 1951, i.e.
URGD method, using exponential rates, is therefore not after an actual time interval of 12.68 years; in the table,
needed. All pertinent calculations can be carried out an urbanization level of 29.1 per cent corresponds to the
very simply by reference to that table. The table is fictitious data of -89, and a level of 38.0 to the fictitious
arranged in two columns, one identified as lOOdt (a com- date of -49, a fictitious time interval of 40 years; divid-
bination of time and the urban-rural growth rate differ- ing 40 by 12.68 we obtain 3.15, and this is the value of
Ut URGD for Colombia during 1938-1951.
ence), and the other as 100- (the percentage level of
urbanization). Tt 189. Given URGD, on the other hand, we can readily
project the urbanization levels. For instance, the urban-
185. It should be noted that the first column of that ization level of Colombia in mid-year 1960 has been
table represents a fictitious time scale, measured from the estimated as 47.8 per cent. Let it be assumed that
point where the urbanization level is exactly 50 per cent, during 1960-1970 URGD was at the same level as during
with negative values where urbanization is lower and 1938-1951, namely 3.15. Ten years later, in mid-year
positive values where it is higher. The intervals in that 1970, the urbanization level should have progressed by
time scale are exact calendar years if URGD is exactly 31.5 years (10 times 3.15) in the fictitious time scale of
1.0 per cent (d=0.01). If the URGD is some multiple the table. The level of 47.8 per cent (Colombia, in
of 1.0 per cent, the intervals in the fictitious time scale 1960) corresponds to the year -9 in the table. Ten years
are corresponding fractions of calendar years. For later, i.e. progressing by 31.5 years in the table, we arrive
instance, if the URGD is exactly 2.0 per cent (d=0.02), at the year 22.5 in the table. At that point in the table,
the intervals are half-years, hence exactly two successive the urbanization level is 55.6 per cent (by interpolation),
intervals represent an interval of one year in actual time. and this, according to assumption, might have been the
186. The URGD, of course, is not always an exact urbanization level of Colombia in 1970.
integral multiple of 1.0, and it may include further
decimals. Interpolations will then have to be made in 190. The table thus makes it possible to project urban-
the fictitious time scale to obtain the corresponding ization levels directly and to calculate urban and rural
points on the scale of actual time. populations according to those levels from a projection
of the total population. This is a considerable simpli-
187. The fictitious time scale, for instance, is about fication of the procedure.
-219.7 at the point where the urbanization level is 10 per
cent, and about +219.7 where it is 90 per cent. Thus, 191. As an illustration, we may take the case of Canada
about 439.4 years would have to elapse for a 10 per cent and assume, for this purpose, that URGD = 2.5. The
urbanized population to become 90 per cent urbanized, annual rises in urbanization level will then correspond
if URGD is at a constant 1.0 per cent. Only a quarter to rises noted every 2.5 intervals in the table (where
of that time, about 110 years, would be needed if URGD URGD = 1.0).
were at a constant 4.0 per cent. Assuming that an actual 192. According to the available estimates, 72.96 per
population has progressed from 10 per cent to 90 per cent of the population of Canada was urban in 1965, a
cent urban in the space of, say, 250 years, the URGD level which occurs in the table (with an interpolation)
needed to reach this effect is 439.4 divided by 250, which at the point 99.25. The future levels, year by year, will
is about 1.76 per cent. The relationship between actual be those occurring in the table at the points 101.75,
time, urbanization levels and URGD, represented by the 104.25, and so forth, each time advancing by 2.5 intervals.
table in annex 11, is thus illustrated. The following urbanization levels will then be attained.
188. The table thus facilitates two types of procedure. 193. Applying now these urbanization levels to avail-
Given the urbanization levels at two dates in the past, able population estimates for the years 1965-1969 and
e.g. as observed in the results of two successive censuses, to a projection of the total population from 1970 up to
the URGD for that time interval can be calculated directly: 1985, which was interpolated for individual calendar
it is the ratio between the time interval between the years, we obtain the following results.
 Calendar year Point in the table Percentage .
195 In our example of the United Republic of Tanza-
1965 ...................... 99.25 72.96 nia. for instance. let us assume that the Government
1966 ...................... 101.75 73.45 entertains plans for a continuously accelerating industrial-
1967 ...................... 104.25 73.93 ization . In this connexion. perhaps a flexible URGD
1968 ...................... 106.75 74.41 should be assumed. rising continuously with time . For
1969 ...................... 109.25 74.88 the sake of the example. let it be assumed that the URGD
1970 ...................... 111.75 75.35 is initially 2.5 and that it may rise to the level of 4.5 in
1971 ...................... 114.25 75.81 the course of twenty years . It would then rise by 0.1 each
1972 ...................... 116.75 76.27 year . Since the urbanization level of the United Republic
1973 ...................... 119.25 76.72 of Tanzania in the year 1965 is estimated as 5.8 per cent.
1974 ...................... 121.75 77.16 which occurs in the table near the point - 279. the follow-
1975 ...................... 124.25 77.60 ing future urbanization levels can be estimated .
1976 ...................... 126.75 78.03
1977 ...................... 129.25 78.46 Percentage
1978 ...................... 131.75 78.88 Calendar year URGD Point in table urban
1979 ...................... 134.25 79.29
1980 ...................... 136.75 79.70
1981 ...................... 139.25 80.10
1982 ...................... 141.75 80.49
1983 ...................... 144.25 80.88
1984 ...................... 146.75 81.27
1985 ...................... 149.25 81.65
1971 .............. 3.05 262.20
. 6.77
Total Percentage 1972 .............. 3.15 -259.05 6.97
Year population urban Urban Rural 1973 .............. 3.25 - 255.80 7.19

196. Applying now these urbanization levels to a

projection of the total population interpolated by indivi-
dual years. we obtain the following results.
Total Percentage
Year population urban Urban Rural

1965 .............. 11. 674 5.79 676 10. 998

1966 .............. 11. 966 5.92 708 1 1 . 251
1967 .............. 12. 267 6.11 750 11. 517
FLEXIBLE
ASSUMPTIONS 1968 .............. 12. 578 6.24 785 11. 793
1969 .............. 12. 900 6.40 826 12. 074
194. Under certain conditions it may be reasonable 1970 .............. 13. 236 6.59 872 12. 364
to assume. instead of a constant URGD. that the URGD 1971 .............. 13. 585 6.77 920 12. 665
will change in the course of time. This may be the case 1972 .............. 13. 951 6.97 972 12. 979
where there is a development plan intended to shift the 1973 .............. 14. 333 7.19 1. 031 13. 302
balance of future developments between urban and rural 1974 .............. 14. 733 7.42 1. 093 13. 640
areas. If rural developments are emphasized. the URGD 1975 .............. 15. 150 7.66 1. 160 13. 990
may decline with time . If the stress is on urban or indus- 1976 .............. 15. 584 7.91 1. 233 14. 351
trial developments. the URGD may increase 60 Assump- . 1977 ..............
..............
16. 034 8.18 1. 312 14. 722
tions of URGD may then be drawn up accordingly. The 1978 16. 500 8.47 1. 398 15. 102
table in annex I is well suited for the derivation of a
population projection in which URGD undergoes a
1979
1980
..............
..............
..............
16. 980
17. 475
8.77
9.10 .
1.489
1 590
15. 491
15. 885
change. 1981 17. 985 9.43 1. 696 16. 289
1982 .............. 18. 514 9.79 1. 813 16. 701
But URGD may decrease if there occurs much settlement of 1983 .............. 19. 066 10.17 1. 939 17. 127
suburban places which remain under a "rural" type of local adminis- 1984 .............. 19. 653 10.58 2. 079 17. 574
tration . 1985 .............. 20. 287 11.01 2. 234 18. 053
 .
197 To vary the example. let us suppose that the Year
Total
population
Percentage
urban Urban Rural
Canadian Government. concerned over an expected
decline in rural population. engages in a large programme
.
of rural rehabilitation and development If this should
be the case. the URGD can be assumed to diminish.
Let us assume that it declines from 3.0 in 1965 to 1.5 in
1980. and then remains at that level. The following
percentage levels of urbanization can then be read from
the table in annex I1.
Percentage
Calendor year URGD Point in the table urban

can serve to guide policy. The fictitious rural rehabili-

tation and development programme would have to be on
a sufficient scale to reduce the URGD at least to 2.0,
and preferably to 1.5, as assumed in the above. The
approximate implication would be a halving in the average
annual rural-to-urban transfers (net migration and
reclassifications). Estimates can then be made of the
expenditures required to achieve such an effect. In
practice. however. the meaning of such a model will also
198. In relation to the interpolated projection of the depend very much on the definition of the "rural" popu-
total population. this leads to the following results. lation. whose living conditions in an advanced country
199. In the above example. the assumptions suffice can become qualitatively similar to those of the urban
to produce a stabilization in the rural population by 1974. population. except for densities of settlement. In fact.
By 1981 the 1965 size of the rural population would be in the absence of administrative change. an increasing
.
recuperated Such a calculation need not be carried out proportion of the population may come to inhabit
as a forecast. but rather to provide a rough model which suburbs which are technically classified as "rural" .
 Chapter VI
PROJECTIONS FOR INDIVIDUAL CITIES, GROUPS OF CITIES
AND DISTINCT GROUPS OF LOCALIT.IES

METHODSNOT DEALT WITH IN THIS MANUAL changes, including the correlated growth of employment
in industry and services, available means of transport,
200. Forecasts of the population for one or several taxes and subsidies, rising levels of education, discontent
distinct cities are usually made by economic and geo- with conditions in rural areas and small towns and so
graphic methods. These methods can comprise much forth. Cities containing certain key industries may
detail, but they must be adapted in each instance to grow faster than other cities. Chain effects of migration
known economic and geographic circumstances, and these are also known, such as the attraction to a city of addi-
can differ remarkably from one city to another. Since tional migrants once a viable local community has been
a general discussion of such locally specialized adapta- formed by previous migrants pertaining to a particular
tions would be difficult, the presentation of such methods ethnic group, caste or region of origin. The subject is
is outside the scope of this manual. It is to be doubted ramified and cannot be dealt with at length in this techni-
whether a set of methods suitable in the forecast for a cal manual. As will be sen in this chapter, several
particular city can be applied without considerable alternative methods of projecting city populations in
transformations to a city in another country where consistency with national total and urban populations are
circumstances and available information are of a different possible, and they can lead to a range of alternative
kind. results. The elements of judgement here suggested will
201. The economic method is probably most suitable have a bearing on the selection of those alternatives which
in countries with central economic plans where those plans might be of best practical use to the authorities in need
include expectations of labour force in the chief branches of a plausible forecast.
of industry to be employed at future dates in each of the 204. In this chapter, methods are proposed which
various locations. Multipliers can then be applied to ensure that the population projections for one or several
represent ratios of total population (including labour cities will not be inconsistent with those for other cities
force in complementary industrial and service sectors and or the rest of the country's population. Such assurance
the dependents of the labour force in every sector) to would be lacking if population projections were decentral-
arrive at corresponding total populations. These fore- ized and carried out independently in each city. The
casts for individual regions and cities may have to be procedures suggested here are sensitive to the relative
reconciled with the over-all population projections for position of each city within the urban system. But they
the whole nation since, logically, the sum of regional should not be used too mechanically. Where elements
forecasts should agree with the national totals. It is then of judgement indicate it, a corresponding adjustment
possible to foresee also the migratory movements neces- should be introduced in an otherwise too systematic
sary to reach the planning targets. procedure.
202. An alternative method, also making use of multi- 205. The methods suggested here may also serve in
pliers, is pertinent in particular to the foreseeable popu- projections for portions of a city, for instance its centre
lation of individual city quarters, residential towns or and its periphery, but care should be taken in such a
suburbs for which the construction of a given quantity case that well-known circumstances are not being dis-
of residential dwelling space can be foreseen. Given the regarded. In population forecasts with respect to
amounts of dwelling space estimated for the future and individual city quarters geographical considerations
assuming full or a high degree of occupancy, one can should perhaps be given some weight. Use can be made,
calculate the corresponding estimates of future population. for instance, of the empirical observation of gradients
Again, the method is most pertinent where housing of residential housing density as a function of distance
construction is centrally planned. And future population from the city's core and progressive changes therein.
estimates according to housing space may have to be In virtually all large cities it is observed that in the most
reconciled with future population estimates derived from central areas the resident population tends to decline.
projected employment in basic industries to arrive at Immediately surrounding the centre are areas of very
consistency in the respective plans. high residential density, but densities decrease with
203. However, elements of judgment concerning com- distance from that area, and rates of population growth
parative rates of city growth will not be lacking in coun- reach a maximum further and further away from the city
tries whose economy is not centrally planned. City centre. Forecasting methods based on such observa-
growth has been studied in relationship to numerous other tions are of special interest where separate population
estimates for the central parts and the peripheries of a question which would have to be resolved with more
city are desired. The geographic method can be made detailed knowledge of relevant local circumstances).
more detailed having regard to the shifting distribution But a projection in which growth at first slows down,
of areas used for business, industry, residence, transport, and thereafter gives way to decline, may have to be
recreation and so forth. Such forecasts depend, of questioned seriously. The fault may lie, in part, with
course, on detailed knowledge of each city's geography. the method of calculation, making it desirable to calculate
206. In preceding chapters, urban populations have with a different method, and then compare the results.
been projected with reference to the projected total popu- 209. Perhaps an assumption should be made that
lation. The same methods can be employed to project changes in percentage level, as extrapolated linearly in
the population of a leading city or of a group of cities table 14, will gradually slow down until some future date
with reference to the projected urban population. The when all cities grow at the same rate as the combined
point of interest in these methods is that consistency is urban population. The interpolated and extrapolated
maintained between the projected population of individual percentage levels of table 14 will then be modified,
cities and the combined urban population. changing more and more slowly, eventually becoming
constant. The growth of Tehran, Esfahan and Mashdad,
which occurred in 1956-1966 at faster rates than the
combined urban growth, could be made to slow down
somewhat, while, in the same procedure, the slower
207. For the projection of individual city populations
growth of Tabriz, Abadan and Shiraz would be corre-
in relation to a projection of the combined urban popu-
spondingly accelerated. Eventually, depending on the
lation, the ratio method is sometimes most suitable, at
least so long as the unusual growth rate of a particular date assumed for the convergence of growth rates, all
city does not lead to extremes where the assumption of cities would grow at the same rates as the combined
urban population. It will have to be recognized, how-
a constant linear change in a percentage can become
ever, that the assumption of a future date of conver-
unreasonable.
gence of growth rates is arbitrary and serves only to
208. A projection of the urban population of Iran has avoid eventually unreasonable results.
already been presented in chapter IV. Table 14 illus-
trates how the population of individual cities (the six
largest in 1966) may be projected, in relation to the pro-
jected urban population, with the use of the ratio method. 210. A more satisfactory alternative method, then,
As a result of these calculations, the population of Tehran may be the URGD method which has already been illus-
may grow almost threefold between 1965 and 1985, the trated in the projection of urban relative to total popu-
populations of Esfahan, Mashdad and Shiraz may grow lation. Proceeding in much the same way, the popu-
more than twofold, the population of Tabriz may grow lation of individual cities can be projected in relation
by two-thirds, but the population of Abadan may decline to the urban population as already projected. The
considerably from 1975 onward. It is to be questioned steps involved in such a calculation are illustrated in
whether the latter result can be accepted. Since Abadan table 15, on page 47. Linear changes are now interpolated
has grown at a comparatively low rate between the and extrapolated, not in terms of the percentages them-
censuses of 1956 and 1966, it is possible that prospects selves, but rather in terms of logistic levels corresponding
for growth are comparatively limited in that city (a to those percentages as tabulated in annex I.

- - - - -

Cafegory Urban popularion Tehran Esfahan Mashdad Tabriz Abadan Shiraz

Census, November 1956 . . . . . . . 5,954 1,512

Census, November 1966 . . . . . . . 9,794 2,720
Percentage of urban, 1956 . . . . . 100.0 25.40
Percentage of urban, 1966 . . . . . 100.0 27.77
Annual change in percentage . . . ... 0.237
Percentages, interpolated and extrapolated to mid year
1965 ...................... ... 27.44
1970 ...................... ... 28.63
1975 ...................... ... 29.82
1980 ...................... ... 31.01
1985 .......... . . . . . . . . . . . . ... 32.20
Population projections, mid year
1965 ...................... 9,172 2,517
1970 ...................... 11,576 3,314
1975 ...................... 14,708 4,386
1980 ...................... 18,598 5,767
1985 ...................... 23,250 7,486
 TABLE
15. CALCULATION
OF POPULATION PROJECTIONS FOR SIX PRINCIPAL CITIES
OF IRAN,
1965-1985, WITH THE URGD METHOD
Category Tehran Esfahan Mashdad Tabriz Abadan Shiraz

Census, November 1956

Urban population a . . . . . . . 5,954
Population of city . . . . . . . . 1,512
Percentage in city ......... 25.40
Census, November 1966
Urban population a ....... 9,794
Population of city ........ 2,720
Percentage in city ......... 27.77
Logistic level of percentage
November 1956 .......... - 107.2
November 1966 .......... - 95.6
Interpolation and extrapolations to mid year
1965.. ................... -97.2
1970.. . . . . . . . . . . . . . . . . . . . -91.4
1975.. ................... -85.6
1980 ..................... -79.8
1985 ..................... -74.0
Correspondingpercentages, mid year
1965.. ................... 27.45
1970.. . . . . . . . . . . . . . . . . . . . 28.62
1975.. ................... 29.82
1980.. ................... 31.05
1985.. ................... 32.30
Population projections, mid year
1965
Urban population a . . . . .
Population of city ......
1970
Urban population a . . . . . . .
Population of city ........
I975
Urban population a .......
Population of city ........
1980
Urban population a .......
Population of city ........
1985
Urban population .......
Population of city ........

a Each time excluding cities listed in preceding columns,

211. To be more specific and to maintain consistency population of Tehran is projected to increase, progres-
with projections already made, the procedure will have sively, from 2,518,000 to 7,5 10,000.
to be applied to one city at a time, preferably beginning 213. When the projection for the first city is completed,
with the largest, proceeding to the second largest, and so it is possible to proceed to the next city, excluding this
forth. time the first city from the combined urban population.
212. In table 15, the percentage of Tehran's popu- In the urban population from which Tehran is excluded,
lation in the combined urban population rose from Esfahan occupied 5.73 per cent in 1956, and 5.99 per cent
in 1966, and the logistic level rose from -280.0 to
25.40 in 1956 to 27.77 in 1966, that is from logistic level -275.3. Again, the logistic levels can be extrapolated
- 107.2 to logistic level -95.6, or by 1.16 steps in logistic linearly, corresponding percentage levels can be estab-
level per year. Interpolating and extrapolating, we lished, and future populations of Esfahan can be calcu-
obtain logistic levels progressing from -97.2 in 1965 to lated from the projection of urban population from which
-74.0 in 1985. The percentages corresponding to the projection for Tehran has already been subtracted.
those levels progress from 27.45 to 32.30. When these The same procedure can be repeated for each subsequent
percentages are applied to the projected urban popu- city, each time relating that city to the urban population
lation (9,172,000 in 1965, and 23,250,000 in 1985), the from which all the preceding cities have been subtracted.
 214. The results obtained in table 15 may now be It will be noted that a slightly faster population growth
compared with those in table 14. For the year 1985 results in the combination of individually projected cities
they differ by less than one per cent in the cases of than in the direct projection of the combined cities.
Esfahah, Mashdad and Shiraz, and in view of possible This follows from the varying speed with which the
errors in any projection such differences are negligible. percentages advance or fall off on a logistic scale, depend-
As compared with table 14, the results in table 15 are ing on their levels. The combined population of the
slightly larger for Tehran, noticeably larger for Tabriz six cities is nearly 50 per cent of the combined urban
and considerably larger for Abadan. Even so, Abadan population, and at that level hardly any acceleration
is still projected to grow much more slowly than the other
cities, and it may be overtaken to an increasing extent, occurs in the logistic rise of a percentage. On the other
for instance, by the population of Shiraz. But this is not hand, cities smaller than Tehran each time constitute
the only plausibleexpectation. Special circumstances may only a small percentage of the urban population (after
have accounted either for the slow growth of Abadan or exclusion of cities of larger size, as in table 15), and at
the rapid growth of Shiraz in the 1956-1966 period. such low levels, according to the logistic, the rise of the
Perhaps the boundaries of one city were widened while percentage accelerates. At any rate, the difference
those of the other were not. Compensatory future resulting between these two methods of calculation is
developments may perhaps have to be foreseen as a only rather slight.
result of which Abadan may again increase more rapidly
or Shiraz more slowly. 218. The foregoing is an example in the projection for
a fixed group of cities, the cities remaining the same
215. There is every freedom to modify the implied throughout the projection period. The matter is differ-
assumptions to take such specific considerations into ent when it is desired to project, to any future date, the
account. Knowledge of recent conditions and prospects combined population of all cities having at least some
in individual cities of Iran would be necessary if assump- minimal size. Since with time additional cities come to
tions were to be modified. In the absence of such surpass that minimal size, the group of cities so defined
detailed knowledge, only the rather mechanical result
of the projection in table 15 can be presented here. is an expanding one, comprising more numerous cities
as time progresses.

FIXEDGROUPS OF CITIES
EXPANDING
GROUPS OF CITlES
216. While the projections for individual cities may
be considerably in error, it is possible that the errors are
partly compensated, some cities being over-estimated 219. In a country like Iran it may be inadvisable to
while some others are underestimated. It may seem make a projection for the expanding group of cities above
safer, therefore, to make a forecast for the combined a high size limit, such as 100,000. At the 1956 census,
population of a group of cities. As regards the six apart from Tehran, Iran had eight cities of such size,
largest cities of Iran, this may be done as follows. and at the 1966 census it had thirteen such cities, since
217. At the 1956 census, the six cities combined had five additional cities had meanwhile entered the group.
a population of 2,696,000, which was 45.28 per cent But two of the additional cities had almost 100,000
within an urban population of 5,954,000; at the 1966 inhabitants already in 1956, namely Hamedan with
census, they had 4,500,000 within an urban population of 99,909 and Ghom with 96,499. The exact number of
9,794,000, which is 45.95 per cent. Using the logistic additional cities that may enter the group, say, in future
scale in annex I and interpolating and extrapolating, we five-year periods will depend much on accident; by
arrive at the following percentages that may be comprised coincidence, there may be some time periods when many
by those six cities in mid year 1965, 1970, 1975, 1980 and cities happen to enter the group, and others when few
1985: 45.70, 45.99, 46.26, 46.52 and 46.73 per cent. or perhaps no additional city happens to reach such size.
Applying these percentages to the urban population, we It is preferable to make calculations for all individual
arrive at the totals shown below; in addition, the sums cities of such size that their attainment of 100,000 by
are shown resulting from the individual city projections the end of the projection period (e.g. by 1985) is possible.
of table 15. In this way one can also estimate how many cities of
Population of six principal lranian czties
such size may exist at any future date, and what the
combined population of the expanding group might be.
Sum of city
populations,
As projected directly as projected 220. When a lower size limit is chosen sufficient to
Year
in relation to the
urban population
individuallv
(see table 141
make the cities of the group quite numerous, these
reservations regarding the direct projection for the group
become unnecessary. In Iran, for instance, there were
49 cities and towns of at least 20,000 inhabitants in
1956, and 73 in 1966. While there may be yearly fluc-
tuations in the number of additional towns which attain
 Percentage of
urban population
Population (thouwmak) (excluding Tehran) Loaistic level Urban
places
Urban Cities Cities Cities Cities Cities Cities Combined Cities New cities smaller
(excluding 100,000;t 100,000+ 100,000;f 100,000+ 100,000: 100,000f urban 100,000f 100,000+ than
Date Tehran) in 1956 at any date in 1956 at any date in 1956 at any date Date population Tehran in 1956 at any date 100,000

A. Projection B. Composition of urban population (including Tehran)

Censuses 1. In thousands
1956 ................ 4,441 1,538 1,538 34.63 34.63 -63.5 -63.5 1965 ................ 9,172 2,518 2,198 512 3,944
1966 ................ 7,075 2,318 2,947 32.76 41.66 -71.9 -33.3 1970 ................ 11,576 3,313 2,652 1,019 4,592
1975 ................ 14,708 4,386 3,219 1,753 5,349
1980 ................ 18,598 5,775 3,885 2,776 6,162
1985 ................ 23.250 7,510 4,631 4,136 6,973
Mid-year estimates 2. Percentage
1965 ................ 6,654 2,198 2,710 33.03 40.73 -70.7 -37.5 1965 ................ 100.0 27.4 24.0 5.6 43.0
1970 ................. 8,263 2,652 3,671 32.10 44.43 -74.9 -22.4 1970 ................ 100.0 28.6 22.9 8.8 39.7
1975 ................ 10,322 3,219 4,973 31.19 48.18 -79.1 - 7.3 1975 ................ 100.0 29.8 21.9 11.9 36.4
1980 ................ 12,823 3,885 6,661 30.30 51.95 -83.3 + 7.8 1980 ................ 100.0 31.0 20.9 14.9 33.2
1985 ................ 15,740 4,631 8,767 29.42 55.70 -87.5 +22.9 1985 ................ 100.0 32.3 19.9 17.8 30.0
Including Hamedan (99,909 inhabitants in 1956), excluding Tehran.

TABLE17. MOD- PROJECTION FOR THE GROUP OF CITES IN IRAN WHICH HAD AT LEAST 100,000 INHABITAMS IN 1956,
AND FOR THE GROUP OF CITIES WHICH MAY HAVE AT LEAST 100,000 AT ANY SUBSEQUENT DATE, 1965-1985
--

Percentage o.
urban population
Population ( t h o ~ ~ ~ d s ) (excluding Tehran) Logistic level Urban
places
Urban Cities Cities Cities Cities Cities Cities Combined Cities New cities smdIer
(excluding 100,000f 100,000+ 100,000: IOO,OW+ 100,000: 100,000+ urban 100,000f 100,000+ than
Date Tehran) in 1956 at any date in 1956 at any &re in 1956 at any date Date population Tehran in 1956 at any date 100,000
1

Censuses
A. Projection
I B. Composition of urban populatwn (including Tehran)

1956 ................ 4,441 1,638 1,638 36.88 36.88 -55.9 -55.9

1966 ................ 7,075 2,442 2,947 34.52 41.66 -64.0 -33.3

Mid-year estimates 2. Percentage

1965 ................ 6,654 2,314 2,728 34.78 41.00 -62.90 -36.4 1965 ................ 100.0 27.4 25.2 4.5 42.9
1970 ................ 8,263 2,798 3,616 33.86 43.76 -66.95 -25.1 1970 ................ 100.0 28.6 24.2 7.1 40.1
1975 ..............-. 10,322 3,402 4,806 32.96 46.56 -71.00 -13.8 1975 ................ 100.0 29.8 23.1 9.5 37.6
1980 ................ 12,823 4,112 6,332 32.07 49.38 -75.05 - 2.5 1980 ................ 100.0 31.0 22.1 11.9 35.0
1985 ................ 15,740 4,911 8,216 31.20 52.20 -79.10 + 8.8 1985 ................ 100.0 32.3 21.1 14.2 32.4
NOTE: This is a modification of the projection in tabie 16. The modification consists in including the city of Hamedan among those with at least 100,000 inhabitants already in 1956.
Including Hamedan (99,909 inhabitants in 1956), excluding Tehran.
20,000 inhabitants, the fluctuations in added population supply of cities with hitherto less than 100,000 inhabitants,
are probably quite small in relation to the increase in which can reasonably be expected to surpass that limit
size of the combined population of all localities of at least by 1985.
20,000. No appreciable error is introduced, in such a 224. Still considering table 16, we note that from 1965
case, if the addition of new localities is regarded as a to 1985 Tehran may increase 3.0-fold; the fixed group of
fairly continuous process. cities, 2.1-fold; the expanding group of cities, 3.2-fold
221. The foregoing considerations will now be illus- (not much more rapidly than Tehran, despite increase in
trated with the results of actual calculations. First, the number of cities); and the population of towns
two calculations are made for Iran's cities other than smaller than 100,000, 1.8-fold, which is not much slower
Tehran with 100,000 or more inhabitants, distinguishing than the growth of the fixed group of cities, despite the
cities which already had at least 100,000 inhabitants at the fact that cities attaining 100,000 inhabitants are con-
1956 census from those additional cities which may tinuously being lost to that last-named group. The
attain such size at any subsequent date. The first several projections are not necessarily inconsistent with
calculation (table 16) shows the results strictly according each other.
to the census data. Rut the same type of calculation 225. The fact that five new cities attained 100,000 each
is repeated (see table 17) to show the difference in results between the censuses of 1956 and 1966, however, may
when one additional city (Hamedan) is also included have to be regarded as rather accidental, especially in
among those with 100,000 inhabitants already in 1956: view of the fact that at the 1956 census two of these cities
according to the census, it then had 99,909 inhabitants, already had very nearly 100,000 inhabitants (Hamedan
i.e. virtually 100,000. with 99,909, and Ghom with 96,499). A more plausible
trend of developments may be suggested by counting at
222. Let us consider first the calculations in table 16. least the city of Hamedan among cities with 100,000
Cities other than Tehran with 100,000 inhabitants already inhabitants already in 1960, so that only four additional
in 1956 had a population of 1,538,000 in 1956 and cities join the group between 1956 and 1966. The
2,318,000 in 1966; thus their proportion in the combined calculations in table 17 are the same as those in table 16,
urban population (without Tehran) fell from 34.63 per with this modification. The resulting projection is
cent to 32.76 per cent. After interpolation and extra- perhaps somewhat more realistic than the one previously
polation on the logistic scale, conversion into percentages shown.
and application of the percentages to the projected urban
population (without Tehran), we find that the combined 226. As compared with table 16, we now find (in
population of this fixed group of cities may grow from table 17) that the 1965-1985 growth in Tehran may be
2,198,000 in 1965 to 4,631,000 in 1985. The same type 3.0-fold (as before); the growth in the fixed group of
of calculation was made for the population of all cities cities (including Hamedan), 2.1-fold (approximately as
other than Tehran having at least 100,000 inhabitants before); the growth in the expanding group of cities,
at any time. This is an expanding group, and its 3.0-fold (slightly less than before); and the growth in
combined population rose from 1,538,000 in 1956 to towns smaller than 100,000, 1.9-fold (slightly more than
2,947,000 in 1966; the projection indicates a rise in before). For reasons stated, the projection in table 17
population from 2,710,000 in 1965 to 8,767,000 in 1985. should perhaps be preferred. It may be noted that the
modification is not very substantial. It can be roughly
223. Subtracting the projection for the fixed group estimated6= that the calculation implies eighteen new
from that of the expanding group of cities we find, as cities reaching 100,000 between 1956 and 1985.
shown in part Bl of table 16, that new cities of 100,000 and
more inhabitants which had not yet had such a size in PROJECTION
OF POPULATION BY
1956 may rise from a combined population of 512,000 in SIZE GROUP OF LOCALITY
1965 to 4,136,000 in 1985. Such rapid growth in the
population of new cities reaching 100,000 at any time, 227. Before proceeding to the projection of Iranian
while not impossible, appears rather surprising, for it is towns and cities greater than 20,000 inhabitants, the
calculated that by 1985 these new cities will combine general line of approach should first be discussed. As
almost as much population as the eight cities which had has been shown, the consistency with other projections,
such size already in 1956. It can be roughly estimated notably projections for cities of other size groups,
that the number of new cities of such size would rise to remains a consideration of interest, especially as a test
about twenty-three by 1985. It is a matter of specu- of the plausibility of the results. Separate projections
lation whether such a development appears plausible, have already been made for the population of Tehran
and one may have to examine whether there is a sufficient (much larger than any other city, hence a category apart),
of the urban population without Tehran, and of the popu-
Allowing for growth of already existing new cities in any
lation of cities either larger or smaller than 100,000.
five-year period at rates similar to the growth of old cities, we obtain A projection of the rural population is also shown in
balances for additional new cities of any five-year period. For chapter V. With the further introduction of a lower
instance, the population of 512,000 (new cities of 1965) may repre- size-limit, we can obtain, relative to the projection of
sent four cities having each recently surpassed 100,000; these may Iran's total population, the following component cate-
increase to about 650,000 by 1970, leaving a further increase by
about 369,000 (1,019,000 minus 650,000), the equivalent perhaps of gories: Tehran; other cities of at least 100,000; towns and
another three new cities; and so forth. The calculation is hypo-
thetical and very rough. 62 See foot-note 61 for the method of estimation.
cities with 20,000-99,999 inhabitants; towns smaller urban population grew from 5,954,000 to 9,794,000,
than 20,000; and rural areas; altogether five categories, that is at an instantaneous rate of 4.98 per cent; and the
which comprise between them the total population of rural population increased from 14,362,000 to 16,950,000,
Iran. i.e. at an instantaneous rate of 1.62 per cent; the differ-
228. When dealing with such successive categories, ence between these two rates, or URGD, is 3.36 per cent.
it becomes desirable to apply a systematic procedure. All the calculations which follow will be derived strictly
This can be done in two alternative ways. In one from the census data, hence the results will differ from
procedure, which we may call the downward procedure, those in the foregoing calculations.
we begin with the total population and make the appro- 230. The data which will be used are the available
priate calculations in respect to Tehran; then we take up projection of the total population for 1965-1985, the
the total population without Tehran and carry out the 1956 and 1966 populations in each of the five categories,
same type of exercise with regard to other cities of at and the cumulative populations of successive categories
least 100,000 inhabitants; and so forth with the remaining including either all preceding categories or all subsequent
categories, until we arrive at the rural population. The categories.
opposite, or upward procedure, would be as follows: 231. To recapitulate, the total population of Iran has
again we begin with the total population and make a been projected as follows (in thousands):
calculation concerning the rural population; then we
PopulaIion
take up the urban population and do the respective Year (in thousands)
calculation for towns smaller than 20,000 inhabitants; 1965 .......................... 24,549
and so forth with other categories, until finally we arrive 1970 .......................... 28,358
at Tehran. 1975 .......................... 33,152
229. A11 the foregoing experiments in projecting 1980 .......................... - 38,769
Iran's urban population and that of cities or groups of 1985 .......................... 45,050
cities have been based on the arbitrary assumption (see
chapter V) that the urban-rural growth difference, or The populations enumerated in 1956 and 1966 in each
URGD, is at the level of 3.0 per cent. Actually, between category, and in cumulated categories, were as follows
the censuses of November 1956 and November 1966, the (in thousands) :O3

Towns Towns
Other with smaller
cities 20,OW - than Rural
Census date Tehran 100,000+ 99,999 20,000 population

Population in each category

1956....................... 1,512 1,538& 1,543 1,361 13,001
1966....................... 2,720 2,947 2,462 1,665 15,285
Cumulative population, including all subsequent categories
1956....................... 18,955 17,443 15,905 14,362 13,001
1966....................... 25.079 b 22,359 19,412 16,950 15,285
Cumulative population, including all preceding categories
1956... . . . . . . . . . . . . . . . . . . . . 1,512 3,050 4,593 5,954C 18,955b
1966....................... 2,720 5,667 8,129 9,794c 25,079b
- -- - -

a Includes Hamedan. b i.e. the total population. i.e. the urban population.

232. To illustrate the first two steps in the downward 235. A comparison of the results of the two procedures
procedure, the figures on the next page show how the (tables 18 and 19) shows that in the "downward" proce-
population of Tehran can be projected in relation to the dure (calculations beginning with Tehran and ending
country's total population and how the population of with the rural population), the city of Tehran is projected
other cities can be calculated with reference to the coun- to grow with appreciably greater speed than in the
try's total population without Tehran. "upward" procedure (calculations beginning with the
233. Continuing these operations for the remaining rural population and ending with'Tehran). On the other
categories, we obtain the results shown in table 18 in the hand, in the "downward" procedure the rural population
order in which they were calculated. is projected to grow somewhat less than in the "upward"
234. In the inverted, or "upward" procedure, first a procedure. The projections for cities of at least 100,000
projection of the rural population is obtained in relation inhabitants (without Tehran) and for towns smaller
to the projection of the total population; then a popu- than 20,000 inhabitants do not differ much in the two
lation for towns smaller than 20,000 is derived in relation procedures; for towns with 20,000-99,999 inhabitants,
to the projection for the urban population; and so forth. the two procedures produce virtually identical results.
The manner of calculation is the same and need not be
illustrated. The results are as shown in table 19, in the 68 For reasons already discussed, the city of Hamedan is included
order of the successive calculations. among those with 100,000 or more inhabitants in 1956.
 Percentage Total
Total Population Tehran Logistic without
population of Tehran in total level Tehran

Censuses
1956.. ................... 18,955 1,512 7.98 -244.5 17,433
1966.. ................... 25,079 2,720 10.85 -210.5 22,359
Mid-year estimates
1965.. ................... 24,549 2,556 10.41 -215.2 21,993
1970.. ................... 28,538 3,434 12.11 -198.2 24,924
1975.. ................... 33,152 4,655 14.04 -181.2 28,497
1980.. ................... 38,769 6,288 16.22 -164.2 32,481
1985 ..................... 45,050 8,406 18.66 -147.2 36,644
Percentage Total
Total 100,000+ without
population Other in total Tehran and
without cities without Logistic cities
Tehran 100,000+ Tehran Level 1~,Ooo+

Censuses
1956 ..................... 17,443 1,538 8.82 -233.6 15,905
1966 ..................... 22,359 2,947 13.18 -188.5 19,412
Mid-year estimates
1965.. ................... 21,993 2,747 12.49 - 194.70 19,246
1970 ..................... 24,924 3,781 15.17 - 172.15 21,143
1975 ..................... 28,497 5,215 18.30 -149.60 23,282
1980.. ................... 32,481 7,120 21.92 - 127.05 25,361
1985.. ................... 36,644 9,535 26.02 - 104.50 27,109

236. Whether the results are internally consistent 238. At first it may seem difficult to decide which of the
may be tested by comparing the calculated 1965-1985 two sets of projections should be preferred, that obtained
increases in each category. In the "downward" proce- by the "downward" procedure, or that resulting from the
dure, Tehran increases 3.3-fold; other cities of 100,000 "upward" procedure. But note should be taken of the
inhabitants or more increase 3.5-fold; towns with percentage which each category represents in relation
20,000-99,999 inhabitants increase 2.3-fold; small towns to the combined category from which it was projected.
increase 1.4-fold ; and the rural population grows 1.3-fold.
In the "upward" procedure the growth of Tehran is 239. As for the downward procedure, the following
3.0-fold; that in other cities of 100,000 or more inhabi- percentages are taken from the 1966 census data: Tehran
tants, 3.4-fold; that in towns of 20,000-99,999 inhabitants, (2,720,000) then contained 10.85 per cent of the country's
2.3-fold; that in small towns, 1.4-fold; and that in the total population (25,079,000); other cities of 100,000
rural population, 1.3-fold. or more inhabitants (2,947,000) then contained 13.18
per cent of the total population after exclusion of
237. It is reasonable to expect that the group of cities Tehran (22,359,000) ;towns with 20,000-99,999 inhabitants
of 100,000 or more inhabitants-which is an "expanding" (2,462,000) contained 12.68 per cent of the total popu-
group, growing more numerous with time-may grow lation after exclusion of Tehran and other cities of at
faster than Tehran, though not all individual cities will least 100,000 (19,412,000); and towns smaller than
grow so fast. It is also reasonable to expect that with 20,000 inhabitants (1,665,000) contained 9.82 per cent
dwindling size classes rates of growth become progres- of the total population after exclusion of Tehran, cities
sively lower. It appears somewhat peculiar, nevertheless, of at least 100,000, and towns of 20,000-99,999 inhabitants
that the population of towns smaller than 20,000 is (16,950,000); the rural population (15,285,000) consti-
projected to increase hardly any faster than the rural tuted the residual.
population. With time, new small towns may come into
existence or villages may acquire the size or character- 240. As for the upward procedure, the following
istics which make them classifiable as towns. It appears percentages are taken from the 1966 census data: The
that, with unchanging census definitions of "urban" rural population (15,285,000) comprised 60.95 per cent
localities, little expansion of the small-town category is of the country's total population (25,079,000); towns
provided for. Perhaps the census definition has iden- smaller than 20,000 (1,665,000) comprised 17.00 per
tified in 1956 and 1966 the same identical towns, making cent of the country's urban population (9,794,000);
no allowance for additional towns that might have come towns with 20,000-99,999 inhabitants (2,462,000) com-
into existence during the census interval, with the conse- prised 30.29 per cent of the urban population excluding
quence that the projection likewise fails to provide for small towns (8,129,000); and cities other than Tehran
the possible emergence of more numerous small towns. with 100,000 or more inhabitants (2,947,000) comprised
On the other hand, it is possible that small towns are in 52.00 per cent of the population of all cities including
fact economically stagnant, and grow with hardly any Tehran (5,667,000); the population of Tehran (2,720,000)
greater speed than the strictly rural localities. constituted the residual.
 Towns Towns
Other with srnalIer
Total cities 20,000- than Rural
Mid-year date population Tehran IW,000+ 99,999 20,000 population

1. In thousands
24,549 2,556 2,747
28,358 3,434 3,781
33,152 4,655 5,215
38,769 6,288 7,120
45,050 8,406 9,535
2. Percentage
100.0 10.4 11.2
100.0 12.1 13.3
100.0 14.0 15.7
100.0 16.2 18.4
100.0 18.7 21.2

Towns Towns Cities

smaller with 100,000 +
Total Rural than 20,000- without
Mid-year date population population 20,000 99,999 Tehran Tehran

I . In thousands

2. Percentage

241. It will be noted that all the percentages used in The results of the two procedures illustrated in the
the "downward" procedure are rather small, while in the foregoing and those of the .two simple methods obtained
"u~ward"procedure most of the percentages are rather for the year 1985 are compared on the next page.
-
substantial: On the logistic scale, small percentages rise
at accelerating rates, but much less acceleration occurs 243. The ratio method and constant rates (prorated)
in percentages at more substantial levels. It is possible, produce larger results for small and medium-sized towns
therefore, that more reliance should be placed on the than the foregoing procedures. Otherwise, results of
projections arrived at with the "upward" procedure and the ratio method are generally closer to those of the
presented in table 19. "upward" procedure, while those of constant rates
(prorated) are intermediate between those of the "down-
242. As a further check one may compare the present ward" and "upward" procedures. Each method may be
results with those which would have been obtained with affected by a bias of its own, and no definite conclusion
the use of simpler methods. The drawbacks of simple can be reached as to which method produces the "best"
methods have been discussed in chapter IV. Never- results. If no other source of judgement is available,
theless, within certain limits their use is quite expedient perhaps one might suggest to use an average of the
where a population is subdivided into severalcategories. For results obtained by the "downward" and "upward"
instance, the sum of extrapolated percentages for every procedures, respectively.
category will always remain 100 per cent. If constant
rates of growth are calculated for each category, on the 244. This tentative suggestion might be modified in the
other hand, the sum of results will exceed the projection light of more knowledge of the circumstances prevailing
of the total population, but an adjustment can be made in the urbanization process in Iran. For instance, it
through prorating, maintaining the projected proportions, might be argued that urbanization occurs either more in
but reducing numbers in each category by a constant. response to a "pull" (i.e. the attraction exerted by big
 Towns Towns
Other with smaller
Total cities 20,000 - than Rural
Procedure population Tehran laO,000+ - 99,999 20,000 population

1. In thousands
"Downward" procedure . . . . . . . 45,050 8,406 9,535
"Upward" procedure . . . . . . . . . . 45,050 7,565 9,212
Ratio method . . . . . . . . . . . . . . . . 45,050 7,294 8,348
Constant rates :
(Unprorated) . . . . . . . . . . . . . . (47,075) (8,117) (9,891)
Prorated.. ................. 45,050 7,768 9,466
2. Percentage
"Downward" procedure.. . . . . . . 100.0 18.7 21.2 12.2 5.0 43.0
"Upward"procedure . . . . . . . . . . 100.0 16.8 20.4 12.2 5.1 45.5
Ratiomethod . . . . . . . . . . . . . . . . 100.0 16.2 18.5 13.0 5.6 46.7
Constant rates ............... 100.0 17.2 21.0 12.5 5.4 43.9

cities) or more in response to a "push" (i.e. the repulsion the results of the "upward" procedure. Thus, specific
of the rural and small-town environments). If "pull" is information regarding conditions in Iran may assist in
a prevalent factor, the "downward" procedure may be making a choice, in the particular instance. It does not
more to the point. If "push" is the dominant pheno- follow that the same choice would be equally pertinent
menon, then there may seem to be more reason to accept in other countries where conditions may differ.
 Chapter VII
SUPPLEMENTARY ESTIMATION OF SEX-AGE COMPOSITION
FOR PROJECTED TOTALS OF URBAN AND RURAL POPULATION

A. PRELIMINARY CONSIDERATIONS comparative results which would follow if one or several

of the underlying demographic conditions were modified.
Use of supplementary methods Such methods are considered in chapter IX. Chapter
245. Several methods of projecting urban and rural VIII considers certain other methods yielding directly
totals have been illustrated in previous chapters. In projections of urban and rural population by groups of
this chapter, mechanical methods are described for deriv- sex and age, though in their case the analysis of under-
ing sex-age detail to be used with the projected urban lying factors is not so complete.
and rural totals. In these methods, an observed age 249. For the use of the mechanical methods described
structure in the past or present is adjusted to agree with in the present chapter, it will have to be assumed that the
the projected urban and rural totals calculated by factors causing a disparity between urban and rural
methods previously described. It will have to be population structures undergo little change. This could
recognized that such a procedure is "mechanical" in the be approximately the case where the relative difference
sense that it is not responsive to the influence which between urban and rural fertility levels remains fairly
particular factors may exert on urban and rural age constant, migratory movements do not fluctuate much,
structures. and especially where the greater part of the migrants'
246. A11 the methods of projection described in residence in urban areas is of a rather temporary charac-
chapters 1V to VI lead to projected totals only, whether ter. For in that case the urban population comprises
for the urban and rural population, cities or size classes what one may call a "floating population" consisting of
of communities. The estimated compositions of these a pool of mostly young persons, many of them soon
totals by sex-age groups are nevertheless of much import- departing to be replaced by other young persons who are
ance because of their economic and social implications, again of the same age range. Where migrants tend to
for they determine to a large extent the available labour take up permanent urban residence, time variations in
force, the formation of households or families and needs sex-age structure can be introduced as they advance in
arising in the areas of education, employment, housing age, volumes of migration having fluctuated in the past.
and so forth. It is therefore useful to estimate in some 250. As surveyed in chapter 11, two prevailing modes
detail the respective population structures, provided this of urbanization can be recognized. In Africa and
can be done at least with a tolerable degree of approxima- South Asia the rural-to-urban migrants are mostly
tion. young men and their duration of urban residence is often
247. The demographic causes of variations in urban rather short. In Latin America, Northern America,
and rural population structures are complex. They Oceania and Western Europe, young women are more
include urban-rural differences in fertility level and trend, numerous than young men among the migrants, and
to a minor degree also the mortality differences, volumes there is the apparent tendency of many migrants to take
and time trends in rural-to-urban migration, as well as up permanent urban residence as witnessed by relatively
urban-to-rural return migration, and the sex-age com- high degrees of urbanization even at advanced ages.
position of these migratory streams, itself also suscep- Intermediate, or mixed, conditions can be found in some
tible to variation in the course of time. These strictly other parts of Europe, in the Soviet Unionandapparently
demographic causes are further conditioned by economic, also in East Asia. In view of such diversity of actual
social and cultural background factors, often difficult conditions, the use of the mechanical estimating methods
to identify, and themselves also apt to change as time proposed in this chapter may be more satisfactory in
progresses. Because of the combined effect of these some countries than in others.
several factors and their accumulation with time, the
action of each particular factor cannot easily be identified Adjustment of defective age data for urban
by analytic means. and rural projections
248. Where it is possible, or reasonable, to calculate 251. Since the accuracy of age records is questionable
urban and rural population projections with the cohort- in many censuses-and these data are needed for the
component method, the attempt should be made. In supplementary estimates discussed here- it is appropriate
that method, the particular demographic factors resulting to review briefly how the raw census data on sex-age
in the varied sex-age structures are specifically taken composition of the urban and rural populations might
into account. It is then also possible to calculate the sometimes have to be adjusted. Methods of detecting
inaccuracies in age data and of making rough adjustments sex-age data of the urban and rural population, each
with regard to national total populations have already taken separately. Since rural-to-urban migration is
been dealt with in other manuals. 64 It is known that in most intensive in late adolescence or early adulthood, true
many censuses significant proportions of small children, irregularities of urban and rural age composition are
notably in the 0 to 4-year age group, are omitted; and reflected even in the most accurate data: relative bulges
furthermore that owing to preferences for particular at ages 15 to 19 or 20 to 24 in the urban population, and
final digits in age statements, excessive numbers tend to relative deficits at the same ages in the rural population.
be reported in some age groups and deficient numbers The effect of applying a smoothing formula directly to the
in other age groups. urban and rural populations, each taken separately,
252. As regards the first defect, namely the under- would be to diminish considerably those irregularities
enumeration of small children, it can usually be assumed which are normally to be expected; the consequence may
that the enumeration of children aged 5 to 9 years is very well be that the smoothed data are even less accurate
substantially more accurate and that (unless there is than the defective data without adjustment. Smoothing
information to the contrary) the fertility of women of of data in the national total population (i.e. the urban
child-bearing ages may have been about the same in and rural populations combined) nevertheless remains
both the recent five-year periods, that when children aged justifiable because international migration only seldom
5 to 9 years were born, and that when those aged 0 to 4 produces any marked irregularities in national age
years were born. To correct the 0 to 4 group with structure, and sharp fluctuations in past birth rates
reference to the 5 to 9 group, the following steps are (another possible cause of irregularity in age structure)
then necessary. Numbers of children aged 5 to 9 years are ordinarily not very likely.
are reverse-survived to the period of five to ten years 255. Accordingly, to correct age data of the urban and
prior to the census, namely when they were born, so as to rural population, the adjustments should first be made in
obtain an estimate of numbers of births in that period. 65 the national total population. The relative amounts by
Numbers of women of ages where they might nave been which each sex-age group has been adjusted in the
their mothers, e.g. ages 15 to 44 years, are reverse- national total can then be allocated proportionately in
survived for the middle of that period of births. The the same sex-age groups of the urban and rural popu-
ratio of numbers of births to numbers of women, of lation. 67
that period, provides a rough measure of fertility.
Assuming that in the most recent five-year period fertil- 256. After adjustments have been made for incomplete
ity has been the same, numbers of women reverse- child enumeration and for inaccuracies of age statement,
survived to the middle of the more recent period can be it is advisable to prorate both the adjusted urban and
multiplied by that same measure in order to obtain an rural populations so as to make them coincide with the
estimate of births which occurred within the most recent original census totals. Given the fact that census totals
five years. By forward survival, this number of births can usually have a certain publicity, it is desirable that the
finally be converted into numbers of children (boys and population projections remain consistent with them, so
girls) aged 0 to 4 years at the date of the census. as to avoid possible confusions in the comparison of the
results of projections with the basic census data.
- 253. A S regards the second defect, namely inaccuracy
of age statements, an adjustment can be made by
applying a smoothing formula to the age data by five- B. METHOD OF DIFFERENCE ELIMINATION
year groups. The particular formula which was sug- Explanation of the method
gested in Manual lZZs6 was
x = ,Ils (-S-, +4 S-, + +
10 S 4 Sl - S,),
257. The difference elimination procedure described
here is an application of a general matrix solution for
where Z is the adjusted number of persons in one five- adjusting the information in the cells of a matrix in
year group, to be computed; S is the reported number of such a way that it conforms to an independent set of
persons in the same five-year group; S-, and S-, are horizontal and vertical totals. In the application
reported numbers in the two preceding five-year groups; described here two kinds of information are utilized:
and Sl and S, are reported numbers in the two subsequent (a) base period data for the sex-age composition of the
five-year groups. urban and rural population arranged in matrix form;
254. It must be pointed out, however, that this latter and (b) projections of independent totals of population
adjustment is not recommended for direct application to by sex and age (the horizontal totals) and urban and rural
population (the vertical totals). From these, a new
64 Manual ZZ: Methods of Appraisal of Quality of Basic Data for
Population Estimates (United Nations publication, Sales No.
matrix of projected sex-age composition of the urban
56.XIII.2); and Manual 111: Methods for Population Projections by and rural population is calculated, which balances both
Sex and Age (United Nations publication, Sales No. 56.XIII.3). horizontally and vertically, with the projected totals.
In forward survival, numbers in each age group are multiplied 258. The method is perhaps best illustrated in a
with the corresponding survival ratios. In reverse-survival, the
numbers are divided by the corresponding ratios, thus "bringingto simplified and fictitious example, with a matrix distin-
life" those who have died in the-interim. With the appropriate
ratios, numbers aged 5-9 at a given date can be reverse-survived to 67 he implied assumption is, of course, that inaccuracy of age
numbers aged 0-4 five years previously, and the latter to numbers statement has the same proportionate effects in both the urban and
born within the period of five to ten years ago. For details of rural population, though it can sometimes be expected that ages are
procedure, see Manual ZZZ, op. cit. reported with somewhat greater accuracy in the urban (usually
Zbid., p. 12. more literate) population than in the rural population.
 20. FICT~~IOUS
TABLE EXAMPLE TO ILLUSTRATE THE METHOD OF DIFFERENCE ELIMINATION

Panel A (base period data) Panel B (projected totals)

Age Urban Rural Total Age Urban Rural Total

Panel C (horizontal prorating) Panel D (vertical prorating)

Age Urban Rural Total Age Urban Rural Tofa1

0-14.. ...... 143 357 500 0-14.. ...... 200 295 495
15-44. ....... 200 400 600 15-44.. ...... 280 331 611
45+. ........ 50 150 200 45+. ........ 70 124 194
TOTAL 393 907 1,300 TOTAL 550 750 1,300

Panel E (horizontal prorating) Panel F (arbitrary adjustment)

Age Urban Rural Total Age Urban Rural Total

guishing only a few group cells. Reference may be made the predetermined totals can be tolerated, it is not
to table 20, in which panel A represents the base data, necessary to carry the procedure to its end.
and panel B, the results of the projections (total popu- 263. The method is similar to, and yields similar
lation by age, and urban and rural totals). results as, another method which has long been used by
259. As a first step (panel C), the figures in each row the Registrar General's office of England and Wales. 68
of panel A are prorated in the ratio of the marginal total The present method is preferred here because it is easier
of panel B to the marginal total of panel A. Thus, to manipulate on worksheets. Differences in the results
100 times 500/350 is 143, and so forth. But, when added of the two methods are virtually negligible.
up vertically, the results in panel C do not agree with the 264. The mathematical properties of the Registrar
vertical totals of panel B. General's method have been studiedYB9 and it can be
260. As a next step (panel D), the figures in each presumed that the present method has the same properties,
column of panel C are prorated in the ratio of marginal at least to a high degree of approximation. For instance,
totals of panel B to.margina1 totals of panel C. For the same results are obtained irrespective of whether the
instance, 143 times 5501393 is 200, and so forth. But, first prorating was done horizontally or vertically.
when added up horizontally, the results again show some Also, the results are the same whether the method is
discrepancies from the horizontal totals of panel B. used repeatedly (e.g. first prorating from data for a date A
to totals for a date B, and then prorating from the results
261. The procedure is repeated twice or three times, for date B to totals for a date C), or directly (prorating
by prorating alternatingly in the horizontal and vertical at once from date A to date C).
directions. In each prorating, the cells of the last matrix
are multiplied by a ratio obtained by dividing the mar- Application to a projection
ginal totals of the projection in panel B by the marginal
totals of the last matrix. At each successive step the 265. The method of difference elimination is now
remaining discrepancies are smaller, and soon they applied to a projection of Iran's urban and rural popu-
become negligible. When only a few insignificant lation by a global method, and a projection of Iran's
discrepancies remain-as already happens in the example total population by sex-age groups. Original data on
in panel E-these can simply be adjusted by hand. In
the end, a matrix is obtained whose marginal totals, both In that method, the increments (gains or losses in each group
cell) of each row or column are prorated, each time, in the propor-
horizontal and vertical, agree perfectly with those of the tion of the initial data (panel A).
projections (panel F in the example). D. F. Friedlander, "A technique for estimating a contingency
262. How often the procedure has to be repeated will table, given the marginal totals and some supplementary data",
Journal of the Royal Statistical Society (London), series A (general),
depend on the number of significant digits in the figures. vol. 124, part 3, 1961, pp. 412-420. In this method, the modification
In actual practice, three horizontal and three vertical of the initial data to the final results, measured in terms of the sum
proratings usually suffice. If small discrepancies from of chi-square, is minimized.
 2 1 ~ . APPLICATIONOF THE METHOD OF DIFFERENCE ELIMINATION TO A PROJECTION OF URBAN
TABLE
AND RURAL POPULATION, AND OF TOTAL POPULATION BY SEX-AGE GROUPS (EXAMPLE OF IRAN):
DATA

Census, November 1966

Projection
Sex and ase Total Urban Rural to 1975

Males
0-4 .................... 2,301 805 1,496 3,096
5-9 .................... 2,160 776 1,384 2,464
10-14. ................... 1 ,609 676 933 2,114
15-19. ................... 1,068 501 567 1,790
20-24. ................... 777 398 379 1,531
25-29 .................... 789 327 462 1,196
30-34. ................... 864 335 529 911
35-39. . . . . . . . . . . . . . . . . . . . 763 295 468 750
40-44. ................... 760 280 480 649
4 5 4 9 . ................... 483 192 291 553
50-54. ................... 362 1 44 218 467
55-59. ................... 216 80 136 384
60-64. . . . . . . . . . . . . . . . . . . . 342 129 21 3 301
+
65 ..................... 504 178 326 523
Females
0-4 . . . . . . . . . . . . . . . . . . . . 2,142 762 1,380 3,007
5-9 .................... 1,988 722 1,266 2,402
10-14 .................... 1,438 612 826 2,064
15-19. ................... 1,055 452 603 I , 746
20-24. . . . . . . . . . . . . . . . . . . . 887 375 512 1,494
25-29. . . . . . . . . . . . . . . . . . . . 848 324 524 1,166
30-34. . . . . . . . . . . . . . . . . . . . 803 301 502 886
35-39. ................... 670 272 398 730
40-44 .................... 578 212 366 634
45-49. ................... 365 154 21 1 542
50-54. . . . . . . . . . . . . . . . . . . . 382 160 222 462
55-59. ................... 201 83 118 385
60-64 .................... 326 126 200 312
+
65 ..................... 463 169 294 592

TOTALS,
ORIGINAL 1966 25,144 9,840 15,304
PROJECTED
TOTALS, 1975 14,707 18,4448 33,151 a

a Originally 33,152,000 and 18,445,000. Adjustment by 1,000 was made to make totals agree with
the sum of rounded figures for sex-age groups,

the urban and rural population, detailed by sex-age successive age groups. 70 For the purpose of the pro-
groups, are available from the census taken in Novem- jection of the total population by sex-age groups, how-
ber 1966. The method of difference elimination is used ever, the age data were smoothed and a correction was
to obtain similarly detailed results for 1975. The entire made for the underenumeration of small children.
calculation can be carried out on one large worksheet. 267. The method of difference elimination can be
For reasons of space it is illustrated here in three succes- applied directly to the raw data: since the total popu-
sive segments (tables 21a, 21b, and 21c). lation used for reference is smoothed, the prorating
266. The census age data suffer from large inaccuracies. procedures have the effect of similarly smoothing the
Probably the enumeration of small children was incom- age distributions of the urban and rural populations. In
plete, since nearly the same number of children were fact, direct smoothing of the original urban and rural age
reported at ages 0 to 4 and 5 to 9. Probably there was data is inadvisable: there are genuine irregularities in
also a large heaping on ages such as 30, 40, 50 or 60, as urban and rural sex-age compositions, mainly owing to
indicated by the irregular progression of numbers in the age-specific incidence of migration, and these would
be caused to vanish if a smoothing formula were applied
Methods of examining the accuracy of census statistics by age
have been described in Manual 11: Methods of Appraisal of Quality l1 By means of methods described in Manual 111: Methods of
of Basic Data for Population Estimates (United Nations publication, Population Projections by Sex and Age (United Nations publication,
Sales No. 1956.XIII.2). Sales No. 1956.XIII.3).
 TABLE21~. APPLICATION
OF THE METHOD OF DIFFERENCE ELIMINATION TO A PROJECTION OF URBAN
AND RURAL POPULATION,
AND OF THE TOTAL POPULATION BY SEX-AGE GROUPS (EXAMPLE OF IRAN):
FIRST (HORIZONTAL)
AND SECOND (VERTICAL) PRORATING

Horizontal resulrs Vertical results

Projection
Srx and age Urban Rural Urban Rural to 1975

Males
0-4 ..................
5-9 ..................
10-14 ..................
15-19 ..................
20-24 ..................
25-29. . . . . . . . . . . . . . . . . .
30-34 . . . . . . . . . . . . . . . . . .
35-39. . . . . . . . . . . . . . . . . .
40-44. .................
45-49 ..................
50-54. .................
55-59 ..................
60-64 ..................
65+ ...................
Females
0-4 ..................
5-9 ..................
10-14. . . . . . . . . . . . . . . . . .
15-19.. ................
20-24 ..................
25-29 . . . . . . . . . . . . . . . . . .
3cL34. . . . . . . . . . . . . . . . . .
35-39 ..................
40-44 ..................
45-49 ..................
50-54. .................
55-59. .................
60-64 ..................
65+ ...................

CALCULATED
TOTALS, 1975
TOTALS, 1975
PROJECTED
- -

a Discrepancy from projected total due to accidents of rounding,

directly. In this respect, the method of difference needed to agree with a twofold classification. One such
elimination offers a distinct advantage. 72 use is to estimate, if there are some data for another
268. Since the method has already been illustrated in date and the marginal totals for the given date, numbers
a fictitious example, the procedure involved in tables 21a, of the population by sex and age and by marital status. 73
21b and 21c needs no further comment. After two Another use is to estimate age distributions for numerous
horizontal and two vertical proratings, the results come regions of a country, if there are sex-age totals for
so close to the predetermined totals that only very slight another date. The method is likewise useful when three
adjustments remain to be made, requiring no further or more categories of the population are considered,
calculation. e.g. the population in different size classes of localities,
Further uses of this method or the urban and rural populations in each of several
regions of a country. One further use of the method is
269. The method of difference elimination can be its applicability in the establishment of sex-age estimates
useful for many other purposes where cell frequencies are for an "urban" popul?tion defined differently than in the
The assumption implicit in this use of the method is that errors census.
of age statement are of equal direction and intensity in the urban
and rural populations. If-as may often be-urban age statements 270. To provide an example and at the same time a
are more accurate than the rural, the consequence is an overadjust- test, the case of Japan is taken. Here, the 1960 census
ment of the urban data and an insufficient adjustment of the rural
data. These residual errors, however, would usually have only
secondary importance. 7S As illustrated by Friedlander, op. cit.
 TABLE
2 1 ~ . APPLICATIONOF THE METHOD OF DIFFERENCE ELIMINATION TO A PROJECTION OF URBAN
AND RURAL POPULATION, AND OF THE TOTAL POPULATION BY SEX-AGE GROUPS (EXAMPLE OF
IRAN): THIRD (HORIZONTAL) PRORATING AND FINAL RESULTS

Horizontal results Final results a

Projection
Sex and age Urban Rural Urban Rural to 1975

Males
04 .................. 1,228 1,868 1,231 1,865 3,096
5-9 .................. 1,001 1,463 1,004 1,460 2,464
10-14. ................. 992 1,122 994 1,120 2,114
15-19. ................. 907 883 909 881 1,790
20-24. ................. 861 670 863 668 1,531
25-29. ................. 555 641 556 640 1,196
30-34. ................. 397 514 398 513 911
35-39. ................. 326 424 327 423 750
4 0 4 .................. 270 379 27 1 378 649
4549. ................. 247 306 248 305 553
50-54. ................. 209 258 210 257 467
55-59. ................. 160 224 160 224 384
60-64. ................. 128 173 128 173 301
65+ ................... 210 313 21 1 312 523
Females
0-4 .................. 1,212 1,795 1,215 1,792 3,007
5-9 .................. 986 1,416 988 1,414 2,402
10-14. ................. 980 1,084 982 1,082 2,064
15-19. ................. 834 912 836 910 1,746
20-24 .................. 706 788 708 786 1,494
25-29. ................. 502 664 503 663 1,166
30-34. ................. 375 511 376 510 886
35-39 .................. 332 398 333 397 730
40-44. ................. 263 371 264 370 634
45-49. ................. 256 286 257 285 542
50-54 .................. 217 245 218 244 462
55-59 .................. 178 207 178 207 385
60-64 .................. 136 176 136 176 312
65+ ................... 204 388 203 389 592
- - - -
CALCULATED
TOTALS, 1975 14,672 18,479 14,707 18,444 33,151
TOTALS, 1975
PROJECTED 14,707 18,444 14,707 18,444 33,151
- - - -

Results of one further (vertical) prorating, not shown here, and final adjustments by hand.

defined as "urban" the population of administrative 271. As shown in table 22, the method of difference
municipal districts (shi), some of them territorially very elimination leads to a fairly good estimate of the sex-age
extensive and including much population which, because composition of the "urban" population when defined
of lower densities of settlement, might otherwise be re- differently. The errors of estimate are comparatively
garded as rural. At the same time, the census determined small. In the particular instance numbers of children
the population of urban clusters as defined by popu- and older persons have been somewhat over-estimated
lation size (at least 5,000 inhabitants) and density (at for densely inhabited districts, while young adults have
least 4,000 persons per square kilometre within each been somewhat underestimated. 74 If the sex-age com-
position of densely inhabited districts had not been
respective enumeration district), under the term "densely ascertained in the census, the estimate could have served
inhabited districts". Thus, the "urban" and "rural" as a rough substitute.
population amounted in 1960 to 59,239,000 and
272. The method of difference elimination, finally,
34,108,000 inhabitants, respectively, the definition of can be of some help in countries where the sex-age compo-
"urban" being rather extensive in that case; and the
population within and outside "densely inhabited 74 under the more intensely "urban" conditions within the nar-

districts", a rather narrow definition of the "urban" rower confines of densely inhabited districts, migrants of early adult
age are more heavily represented than they are in municipal districts
population, amounted to 40,807,000 and 52,540,000 which include some population that in other respects perhaps may
inhabitants, respectively. be classifiable as "rural".
 TABLE22. ESTIMATE OF SEX-ME COMPOSITION OF D.I.D. (DENSELY INHABITED DISTRICTS) POPU-
LATION FROM DATA ON SEX-AGE C O M P O ~ O NOF "URBAN" AND "RURAL" POPULATION AS DEFINED
ADMINISTRATIVELY. JAPAN. 1960. USING THE METHOD OF DIFFERENCE ELlMINATION

Municipal Esrimated D.I.D.

population D.I.D. population Error of
Sex and age (census data) population a (census data) estimate

Males
0-4 ................ 2. 435 1. 635 1. 615 +20
5-9 ................ 2. 708 1. 766 1. 710 +56
10-14 ................ 3. 337 2. 220 2. 168 +52
15-19 ................ 3. 286 2. 394 2. 433 -39
20-24 ................ 2. 959 2. 195 2. 264 -69
25-29 ................ 2. 824 2. 033 2. 070 -37
30-34 ................ 2. 466 1. 723 1. 743 -20
35-39 ................ 1. 774 1. 226 1. 243 -17
4044 ................ 1. 461 1. 006 1. 007 -1
45-49 ................ 1. 433 987 995 -8
50-54 ................ 1. 281 870 873 -3
55-59 ................ 1. 103 743 737 +6
60-64................ 856 567 549 +18
65-69 ................ 585 330 363 +17
70-74 ................ 386 246 228 $18
75-79 ................ 205 128 119 +9
80-84 ................ 84 52 42 +lo
85 and over .......... 26 16 12 +4
Females
0-4 ................ 2. 325 1. 555 1. 535 +20
5-9 ................ 2. 609 1. 705 1. 653 $52
10-14 ................ 3. 226 2. 145 2. 112 +33
15-1 9 ................ 3. 155 2. 267 2. 293 -26
20-24 ................ 2. 891 2. 081 2. 133 -52
25-29 ................ 2. 809 2. 006 2. 060 -54
30-34 ................ 2. 455 1. 716 1. 756 -40
35-39 ................ 2. 107 1. 456 1. 488 -32
40-44 ................ 1. 758 1. 212 1. 229 -17
45-49 ................ 1. 607 1. 095 1. 098 -3
50-54 ................ 1. 339 910 910 0
55-59 ................ 1. 137 764 750 +14
60-64................ 890 590 582 +8
65-69 ................ 660 434 420 +14
70-74 ................ 496 322 301 +21
75-79 ................ 319 204 186 +18
80-84 ................ 173 111 92 +19
85 and over .......... 74 47 40 +7
-
TOTAL 59. 239 40. 807 40. 807 0

Calculated from "urban" (municipality) and "rural" (other areas) population census data by method
of difference elimination .

sitions of the urban and rural population have not yet and Sierra Leone. however. are heavily affected by age
been ascertained. Data for another country where some misstatement. But in the case of Sierra Leone a smooth
of the conditions might be similar can then be used and population model has been substituted. and estimates
prorated to the summary data available for the particular and projections have been made with that model . This
country. makes it possible to prorate the data for Ghana directly
273. For instance. the sex-age composition of the total to smoothed data for Sierra Leone. In the given example.
population of Sierra Leone. but not of its urban and the data for Ghana are brought into line with Sierra
rural population. has been obtained in the 1963 census. Leone's national sex-age composition as estimated for
Relevant data exist. however. in the 1960 census of Ghana 1960. For that year. Sierra Leone's urban population
for an urban population of 1.551. 360 and a rural popu- is estimated as 250.000. and the rural population as
lation of 5.179.460 . Both sets of data. those for Ghana 1.886.000.
 Sierra Leone,
Ghana (census) final adiustment
Sierra Leone
Sex and age (estimate) Urban Rural Urban Rural

Males
0-4 ..............
5-9 ...............
1C14. . . . . . . . . . . . . . .
15-19. ..............
20-24. ..............
25-29 . . . . . . . . . . . . . . .
30-34 ...............
35-39. ..............
4 0 4 4 ...............
45-49 ...............
50-54. . . . . . . . . . . . . . .
55-59. ..............
60-64. ..............
65 f................
Females
0-4 ...............
5-9 ...............
10-14 . . . . . . . . . . . . . . .
15-19 ...............
20-24. ..............
25-29. ..............
30-34. ..............
35-39. ..............
40-44. ..............
45-49. ..............
50-54. ..............
55-59. ..............
60-64 ...............
65+ ................

274. Table 23 shows the original data for Ghana, the and in the estimation of sex-age composition for "urban"
marginal totals for Sierra Leone, and the results obtained and "rural" populations defined differently than at the
after two horizontal and vertical proratings, rounded to census, or even belonging to another country. The
the nearest hundred. In the absence of the pertinent question remains whether, in a projection, the adjust-
information, these results may perhaps constitute an ments obtained by this method are plausible in terms of
acceptable estimate of the sex-age composition of the the dynamics of the growth components.
urban and rural population in Sierra Leone, despite the 276. The differences between urban and rural sex-age
roughness of the original data for either country, and the structures stem mainly from urban-rural differences in
likelihood that conditions affecting urbanization are fertility and from net rural-urban migration. As
not quite the same in both countries. compared with the rural, the urban population pyramid
has usually a somewhat narrower base (because of lower
Additional considerations urban fertility), and "bulges" about the ages at which
most migrants arrive in the cities and towns. The
275. Several advantages in the use of the method of method of difference elimination preserves these peculiari-
difference elimination have been noted. It is applicable ties in the urban and rural population structures, in so far
not only in the projection of the urban and rural popu- as they remain compatible with projected changes in the
lation, but also in the calibration of urban and rural sex-age composition of the total population. In other
sex-age data to smoothed data of the total population, words, implicit in such projections are the rough assump-
tions that the urban-rural fertility difference will remain the projection of total population, applying these per-
unchanged, and that migration, both in its relative centages.
magnitude and in its composition by sex and age groups, 281. To obtain sex and age detail for future urban and
will continue to have similar effects as in the past. rural population, the following steps were taken. First,
277. It can be demonstrated that quite similar results the percentage distributions by sex-age groups in 1957
are often obtained whether the urban and rural popu- were calculated separately for the total, urban and
lations are projected globally and the age composition rural population. Next, ratios were computed for the
estimated with the present method, or whether the urban percentages in the urban and rural population by sex
and rural populations are each projected with the cohort- and age, relative to the percentages in the total popu-
survival method. For in the latter method it will usually lation. These steps are illustrated in table 24.
also have to be assumed that fertility differences between 282. These ratios were then maintained, as "correction
urban and rural areas tend to persist, and that the rela- factors" so to speak, to adjust any projected percentage
tive volume and composition of the migratory stream, sex and age distributions of the total population so as
with its dominance of young adults, will remain similar to yield corresponding distributions for the urban and
to that for which a relevant estimate can be made, i.e. rural populations, respectively. 76 The projected total
according to past experience. The method of difference population had the following percentages in different
elimination simulates the combined effects of those sex-age groups (table 25). Multiplication of the percen-
fertility differences and sex-age compositions at the base tages in table 25 by the urban and rural "correction
date, without analysing and applying them in any factors" of the last two columns of table 24, led to the
specific detail. Therefore, there may sometimes be following percentage sex and age distributions for the
little difference in the results of both types of projections, projected urban and rural population (table 26). These,
except that the global projection leaves the analytic in turn, were then applied to projected actual numbers of
detail unspecified. urban and rural population so as to obtain the absolute
278. Under more special circumstances, where the numbers to be estimated. 77
inherent conditions are expected to undergo some change
or where past fluctuations are implicit in the initial age
structure, the global method with difference elimination C. ESTIMATING SEX-AGE GROUPS WITH THE
can become less adeauate. For instance. a fertilitv USE OF THE LOGISTIC TABLE
decline might be foreslen for urban areas, 'but not yet
for rural areas, with consequent widening in the urban- 282. Levels or degrees of urbanization can be measured
rural difference. Or the tempo of urbanisation, espe- Separately for each sex-age group of the national popu-
cially its migratory component, might be expected to tion, in terms of urban residence ratios specific by sex
accelerate or to slow down with consequent intensi- and age (see chapter 11, second section). or reasons
fication or diminution of the effects of migration upon the discussed in chapter 111, third section, it is useful to
urban and rural age structures. Where such consider- transform these Sex-age-specific percentage levels to a
ations are important a cohort-survival projection is logistic scale. For, as argued, the flexibility of a percen-
of course preferable. tage is under varied constraints, depending on whether
the percentage is at an intermediate or extreme level.
A shorter method Here again the particular parameters entering into the
calculation of a logistic curve are immaterial, hence the
279. The method of difference elimination discussed simple curve tabulated in annex I can serve as reference
in the foregoing may sometimes appear more cumber- also for the purpose of estimating sex-age composition.
some than is warranted by the approximate nature of the 283. Table 27 shows urban percentages of the p o p -
forecasts. A shorter method is therefore also to be lation of Iran in 1966, specific by groups of sex and age.
recommended in which, however, the differences in age Thus, children were urbanized to an extent of 35 to 36 per
group totals are not entirely eliminated. Among the cent, while at adult ages the percentage fluctuated about
many uses which have been made of the shorter method 40, reaching peaks among adolescents and young adults,
one may cite a demographic study undertaken jointly and also among women about the age of 50. Compara-
by the United Nations and the Cbvernment of the tively high percentages of urbanization appear among
Philippines. 76 males aged 15 to 19 and those aged 20 to 24. It is not
280. The base data for the projection were those unlikely that many young men of rural origin spend their
obtained in household surveys around 1957, including period of military service at urban duty stations.
separate data by sex and age for the urban and rural
population. The total population was projected by 7e The majority of the initial population was still rural, hence the
groups of sex and age for five-yearly intervals up to 1977. rural correction factors differ less from unity than the urban correc-
percentages of rural population were in a trend tion factors. With time, an increasing proportion of the population
will be urban, but the present method makes no provision for the
parallel to a projected trend in the percentage of labour effectof shift in weights on the correction factors.
force engaged in agriculture. The expected future sizes 77 With this procedure, the totals of urban and rural population
of urban and rural population were then derived from within each sex-age group do not exactly agree with the projected
total population, although the discrepancies are generally trivial.
76 Population Growth and Manpower in the Philippines, a joint The sum of the urban and rural population may therefore be
study by the United Nations and the Government of the Philippines accepted as a modified projection of total population, or an adjust-
(United Nations publication, Sales No. 61.XIII.2). ment may be made in the urban and rural figures.

63
TABLE24. SEX-AGE
DISTRIBUTIONS OF THE TOTAL POPULATION OF THE PHILIPPINES AND OF URBAN
AND RURAL POPULATION. 1957 (DATA FROM 1956 AND 1957 PSSH ROUNDS. REFERRING TO
PRIVATE HOUSEHOLD POPULATION)
- - -

Per cent distribution by sex and age Ratio. Ratio.

.
col 3
-
c01. 4
-
Total Urban Rural
Sex and age (years) population population population col.'2 '
(1) (2) (3) (4) (5)

Both sexes.
allages .................. 100.00 100.00 100.00
Males
0-4 .................... 9.32 8.94 9.53
5-9 .................... 7.29 6.91 7.50
10-14 .................... 6.58 6.36 6.70
15-19 .................... 5.12 5.17 5.08
20-24 .................... 4.10 4.19 4.05
25-34 .................... 5.89 6.10 5.78
3 5 4 .................... 4.80 4.79 4.81
45-54 .................... 3.39 3.43 3.36
55-64 .................... 1.85 1.70 1.94
65 and over .............. 1.57 1.34 1.70
Females
0-4 .................... 8.97 8.79 9.07
5-9 .................... 7.04 6.76 7.18
10-14 .................... 6.26 6.07 6.36
15-19 .................... 5.32 6.01 4.93
20-24 .................... 4.47 4.91 4.23
25-34 .................... 6.51 6.98 6.26
3 5 4 .................... 4.98 4.97 4.98
45-54 .................... 3.28 3.30 3.27
55-64 .................... 1.73 1.73 1.73
65 and over .............. 1.54 1.55 1.54
- - -

a This will be considered as the urban "correction factor" .

This will be considered as the rural "correction factor" .

TABLE25. PERCENTAGE
SEX-AGE C O M P O ~ O NOF THE TOTAL POPULATM~N
OF THE PHILIPPINES AS PROJECTED FROM 1957 TO 1977

Sex and age

(years) 1957 1962 1967 1972 1977

Males
0-4 .................... 9.32 9.46 9.57 9.59 9.58
5-9 .................... 7.29 7.48 7.60 7.70 7.74
10-14 .................... 6.58 6.17 6.21 6.36 6.43
15-19 .................... 5.12 5.57 5.19 5.27 5.31
20-24 .................... 4.10 4.28 4.63 4.30 4.36
25-34 .................... 5.89 6.08 6.37 6.74 6.70
3544 .................... 4.80 4.27 4.00 4.11 4.29
45-54 .................... 3.39 3.38 3.12 2.77 2.60
55-64 .................... 1.85 1.89 2.00 2.00 1.85
65 and over .............. 1.57 1.38 1.25 1.23 1.27
Females
0 4 .................... 8.97 9.22 9.32 9.34 9.31
5-9 .................... 7.04 7.22 7.41 7.51 7.55
1&14 .................... 6.26 5.95 6.07 6.21 6.27
15-19 .................... 5.32 5.29 4.99 5.07 5.18
20-24 .................... 4.47 4.45 4.40 4.14 4.21
25-34 .................... 6.51 6.66 6.75 6.67 6.41
35-44 .................... 4.98 4.62 4.42 4.51 4.56
45-54 .................... 3.28 3.43 3.31 3.06 2.93
55-64 .................... 1.73 1.81 2.03 2.11 2.05
65 and over .............. 1.54 1.40 1.28 1.29 1.40
 TABLE26. P R O J E ~ O
OFNSEX-AGE D I S T R I B ~ O NOF URBAN AND RURAL P O P U L A ~ O NOF THE PHILIPPINES.
1957-1977
(Percentage)
1957 1962 1967 1972 1977
Sex and clge
(years) Urban Rural Urban Rural Urban Rural Urbon Rural Urban Rural

Both sexes. all ages ... 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00
Males
0-4 ............. 8.94 9.53 9.07 9.67 9.18 9.78 9.20 9.80 9.19 9.79
5-9 ............. 6.91 7.50 7.09 7.69 7.21 7.81 7.30 7.92 7.34 7.96
10-14 ............. 6.36 6.70 5.97 6.28 6.09 6.40 6.15 6.47 6.22 6.54
15-1 9 ............. 5.17 5.08 5.62 5.53 5.25 5.16 5.33 5.24 5.37 5.28
20-24 ............. 4.19 4.05 4.37 4.23 4.73 4.57 4.40 4.25 4.46 4.31
25-34 ............. 6.10 5.78 6.29 5.97 6.59 6.25 6.98 6.61 6.93 6.57
35-44 ............. 4.79 4.81 4.26 4.28 4.00 4.01 4.11 4.11 4.29 4.29
45-54 ............. 3.43 3.36 3.42 3.35 3.16 3.09 2.80 2.74 2.63 2.58
55-64 ............. 1.70 1.94 1.73 1.98 1.83 2.10 1.84 2.10 1.70 1.94
65 and over ........ 1.34 1.70 1.17 1.49 1.06 1.35 1.05 1.33 1.08 1.37
Females
@-4 ............. 8.79 9.07 9.04 9.32 9.14 9.42 9.16 9.44 9.13 9.41
5-9 ............. 6.76 7.18 6.93 7.38 7.12 7.57 7.22 7.67 7.25 7.71
10-14 ............. 6.07 6.36 5.76 6.05 5.88 6.17 6.02 6.31 6.08 6.37
15-19 ............. 6.01 4.93 5.98 4.92 5.64 4.64 5.74 4.72 5.85 4.81
20-24 ............. 4.91 4.23 4.89 4.21 4.83 4.16 4.55 3.92 4.63 3.98
25-34 ............. 6.98 6.26 7.14 6.40 7.23 6.49 7.15 6.41 6.87 6.16
35-44 ............. 4.97 4.98 4.61 4.62 4.41 4.42 4.50 4.51 4.55 4.56
45-54 ............. 3.30 3.27 3.44 3.42 3.33 3.30 3.08 3.05 2.95 2.92
55-64 ............. 1.72 1.73 1.81 1.81 1.29 2.03 2.12 2.11 2.05 2.05
65 and over ........ 1.55 1.54 1.41 1.40 1.29 1.28 1.30 1.29 1.41 1.40

Population
Per cent Logistic Same.
Sex and age Total Urban urban level smoothed

Both sexes ................

Males .....................
0-4 ....................
5-9 ....................
10-14 ....................
15.19 ....................
20-24 ....................
25.29 ....................
30-34 ....................
35-39 ....................
40-44 ....................
45-49 ....................
50-54 ....................
55-59 ....................
60-64 ....................
65 and over ..............
Females ...................
0-4 ....................
5-9 ....................
10-14 ....................
15-19 ....................
20-24 ....................
25-29 ....................
30-34 ....................
35-39 ....................
40-44 ....................
45-49 ....................
5&54 ....................
55-59 ....................
60-64 ....................
65 and over ..............
 TABLE
28. PROJECTION
OF LOGISTIC LEVEL,
URBAN RESIDENCE RATIOS AND ACTUAL NUMBERS OF
URBAN POPULATION TO 1975, IN CONFORMITY WITH PROJECTED POPULATION OF IRANBY SEX-AGE
GROUPS, AND URBAN TOTAL
Logistic level Population, 1975
-- Per cent --
urban Urban,
Sex and age 1966 1975 1975 Total Urban prorated

Both sexes ...

Males .....................
0-4 . . . . . . . . . . . . . . . . . . . .
5-9 ....................
10-14. ...................
15-19. . . . . . . . . . . . . . . . . . . .
20-24. ...................
25-29. ...................
30-34. ...................
35-39. ...................
40-44. ...................
4 5 4 9 . ...................
50-54. . . . . . . . . . . . . . . . . . . .
55-59. ...................
60-64 ....................
65 and over ..............
Females ...................
0-4 ....................
5-9 ....................
10-14 ....................
15-19 ....................
20-24. . . . . . . . . . . . . . . . . . . .
25-29. ...................
30-34 . . . . . . . . . . . . . . . . . . . .
35-39 . . . . . . . . . . . . . . . . . . . .
40-44. . . . . . . . . . . . . . . . . . . .
4549. ...................
50-54 . . . . . . . . . . . . . . . . . . . .
55-59. . . . . . . . . . . . . . . . . . . .
60-64 ....................
65 and over ..............

284. With reference to the table in annex I, these 285. These smoothed logistic levels of urbanization
specific urban residence ratios can be transformed to the were entered in the first column of table 28. In a pre-
logistic levels also shown in table 27. Some incon- vious projection of urban and rural totals, the percentage
gruities emerge at middle and advanced adult ages, and urban in Iran was estimated to rise from 39.1 per cent at
these are most pronounced among women where the the 1966 census to 44.6 per cent in 1975. By reference
logistic level fluctuates from -51 at ages 30 to 34 to to the logistic table in annex I it can be seen that this
-38 at ages 35 to 39, back to -55 at ages 40 to 44, and corresponds to a rise along the logistic curve of 22 points
again to -29 at ages 45 to 49. It is unlikely that in (from a value of -44 to -22). For purposes of the
actual fact womei of adjacent five-year age groups
projection it will be assumed that the increase in logistic
would be urbanized to such an unequal extent. Most
likely, these fluctuations are due to unequal accuracy level of each sex-age group was the same as the increase
in age statements for urban women as compared with in logistic level previously projected for the total popu-
those for rural women. To counteract this effect, the lation, i.e. 22 points as shown in the second column of
logistic levels were smoothed for age groups between table 28. The corresponding percentage levels (as read
25 and 64 years of age, using a three-term moving average. from the logistic table in annex I) are shown in the third
Another slight modification, also in the last column of column. These percentages are then applied to the
table 27, was made by introducing the arbitrary assump- projection of the total population to 1975, already
tion that at ages 0 to 4 and 5 to 9 boys and girls may be calculated by sex-age groups. The results add up to an
urbanized to the same extent. 78 urban population of 14,949,000, which is more than the
figure of 14,707,000 already projected. In the last
This adjustment is debatable. In Latin America, for instance, column, therefore, the figures for each sex-age group are
small girls are more often taken by their migrant mothers to the prorated so as to coincide with the latter total. (The
towns, whereas small boys are more often left in the countryside,
as can be borne out by analysis of census data. rural population, by sex and age, can then be obtained
 TABLE 29. CALCULATION OF SEX-AGE SPECIFIC URBAN RESIDENCE RATIOS
AND OF CORRESPONDING LOGISTIC LEVELS FROM 1950 CENSUS DATA FOR BRAZIL

Population
-- Per cent Logistic
Sex and age Total Urban urban level

Both sexes.................
Males .....................
0-4 ....................
5-9 ....................
10-14.. ..................
15-19.. . . . . . . . . . . . . . . . . . .
20-24. ...................
25-29. ...................
30-39. ...................
4 0 4 9 ....................
50-59. ...................
60-69 ....................
70 and over a .............
Females ...................
0-4 ....................
5-9 ....................
10-14. ...................
15-19 ....................
2&24. ...................
25-29. ...................
30-39. ...................
40-49 ....................
50-59. ...................
60-69 ....................
70 and over a ............
a Includes persons of unknown age.

~ by subtracting estimated numbers in sex-age groups of

the urban population from those already projected for
the total population.)
interval in the table in annex I, but no smoothing was done
among those levels.
288. The comparison of the projection with the 1960
286. The reader may wish to compare these results with census result is carried out in the last three columns of
the ones which had been obtained in the preceding table 30. For males and females alike, the following
chapter, in table 21c, by means of the method of difference observations can be made. The projection results in
I
elimination. It will be noted that the results do not an underestimation of the population aged 0 to 14 years,
I differ greatly. Part of the difference in results may be an over-estimation of nvmbers aged 15 to 29 years, an
I attributable to the smoothing of the logistic levels, as underestimation of the population aged 30 to 39 years,
done in table 28. At any rate, census age statements an over-estimation at ages 40 to 69 and a nearly correct
have been rather inaccurate, hence the estimates are estimation at ages 70 and over. Evidently the compara-
rather uncertain, and it would be futile to argue which tive urbanization levels did not, in actual fact, remain the
of the two sets of results are likely to be the more accu- same between the two censuses, 1950 and 1960.
rate ones. It will be noted that less calculating work is 289. The underestimation in projected numbers of
involved when the logistic table is used than in the method children suggests that the relative fertility level of the
of difference elimination. urban population (as compared with the rural population)
287. To give an idea of the possible errors of such a was higher in the 1950-1960 period than in a period
calculation, tables 29 and 30 present the results of the preceding 1950. The over-estimation in projected num-
same type of procedure for Brazil. The calculation for bers of young adults, aged 15 to 29, suggests that the
1960 proceeds from the census of 1950 and results are intensity of rural-to-urban migration was less during
compared with corresponding data of the 1960 census. 1950-1960 than during years preceding 1950. The alter-
For the calculation it is assumed that only the sex-age nation between over-estimates and underestimates, as we
composition of the total population is known for 1960, proceed through the entire range of ages, may even
as well as the totals of urban and rural population. suggest that the intensity of urbanization trends could
Since the age declarations are probably somewhat more have fluctuated cyclically over extended periods of the
accurate in Brazil than in Iran, percentages have been past. Of course, the effects of variations in fertility
calculated with an additional decimal, and logistic levels differences and migration trends cannot be disentangled in
have been interpolated to the nearest tenth of each such an analysis, and alternative interpretations are pos-
 30. PROJECTION
TABLE OF LOGISTIC LEVEL. URBAN RESIDENCE RATIO. AND ACTUAL NUMBERS OF
URBAN POPULAnON IN 1960. AND COMPARISON WITH 1960 CENSUS DATA FOR BRAZIL
Population. 1960
Logistic Per cent Total Urban Urban Error
level urban as per pro- as per in
Sex and age 1960 1960 census jected census projection

Both sexes ..........

Males ...................
0-4 ..................
5-9 ..................
10-14 ..................
15-19 ..................
20-24 ..................
25-29 ..................
30-39 ..................
4 0 4 9 ..................
50-59 ..................
60-69 ..................
70 and over 8 ...........
Females .................
0-4 ..................
5-9 ..................
10-14 ..................
15-19 ..................
20-24 ..................
25-29 ..................
30-39 ..................
40-49 ..................
50-59 ..................
60-69 ..................
70 and over a ..........

a Includes persons of unknown age.

sible: both urban fertility and migration may have ten years more advanced at each of the successive decen-
fluctuated in the course of time. nial censuses. In so far as this may have to be expected.
290. As was noted in chapter 11. the mode of urban- the mechanical methods proposed in both parts of the
ization in Latin America is characterized by higher present chapter have only limited validity. The case of
urbanization levels of women as compared with men. and Brazil will be further investigated in the next chapter .
rising urbanization levels with advancing age. Accord- 291. The improved estimation of sex and age groups
ingly. one may assume that in Brazil a large proportion in a projected urban and rural population depends.
of rural-to-urban migrants settle in the urban areas however. on additional statistical information . In the
indefinitely. If sex-age specific urban residence ratios absence of sufficient information. it would be difficult
are affected by fluctuations from past trends. it is prob- to make better estimates than those possible with the
able that the same fluctuations will reappear at ages procedures of both parts of the present chapter.
 Chapter VIII
MIXED PROJECTION METHODS

292. As explained in the Introduction, the methods of present little advantage over methods discussed in preced-
projecting urban and rural populations can be of four ing chapters. Their use depends on the availability of
types, namely : accurate statistics on births and deaths occurring in
(a) Global methods (using neither sex-age groups of urban and rural areas, respectively.
the population nor the component trends); 297. In at least one instance, nevertheless, this method
(b) Composite methods (using sex-age groups of the has been used with good effect. 7B In Jamaica the statis-
population, but not the component trends); tics on births and deaths are accurate, and successive
censuses provide comparable data on parish of birth of
(c) Crude component methods (using component migrants, permitting calculation of migratory balances
trends, but not the sex-age groups of the population); and for the intercensal time interval. The most significant
(d) Cohort-survival methods (using both the sex-age distinction between parishes is made when the "metro-
groups of the population and the component trends). politan" parishes of Kingston and St. Andrew, containing
293. Chapters IVYV and VI dealt with global methods, the sprawling capital city, are singled out because the
while chapter VII illustrates how age groups can be remaining parishes are mostly rural.
estimated for globally projected urban and rural popu- 298. Aside from internal migration between the metro-
lations. Chapter IX deals with cohort-component politan and the rural parishes, it was necessary also to
methods. The present chapter is concerned with methods calculate balances of international migration, as there
of the above types (b) and (c), i.e. methods which are has been a considerable outward movement of Jamaicans.
detailed in one respect, but not in another. Type (c) Statistics on international migration are deficient, but
will be dealt with in the first section, while the rest of the the net movement can be calculated by means of inter-
chapter is concerned with methods of type (b). censal balancing equations, once the components of
294. Crude component methods present an advantage natural increase and internal migration have been deter-
over global methods only in so far as they can show to mined. The Jamaican census data are accurate and detailed
what extent urban and rural population growth is apt enough for such a purpose.
to be affected by births, deaths and migration, respec- 299. The particular data situation made it advan-
tively. This leads to the possibility of varying the tageous to calculate presumable future trends in natural
assumptions concerning these growth components. increase, internal migration and international migration
Otherwise, this type of method has little to commend it, as continuations from past trends for the metropolitan
nor does it lead to estimates of urban and rural sex-age area of Kingston and St. Andrew and for the remaining
structures. parishes. A projection of this type is interesting because
295. The composite methods, on the other hand, one can deduce from it how much of the expected local
produce directly projections of urban and rural popu- population growth may be due to natural increase, how
lation by groups of sex and age. In a sense, therefore, much to the internal migratory balance and how much
they are almost as useful as the cohort-survival methods, to the external balance. In the given instance, the calcu-
but they lack flexibility because no clear distinction can lation was also made of the population resulting in the
be made between the effects of fertility, mortality and metropolitan parishes if internal migration were to end
migration, respectively (and area reclassification where suddenly. Other assumptions might also be varied in
this is relevant). However, their use depends on the order to study the respective consequences.
availability of relevant and comparable statistics from 300. There may be other countries or cities where the
two successive censuses. statistical data situation is similar so that this type of
calculation becomes relevant. But it would seem unwise
CRUDECOMPONENTS METHODS
to carry such a calculation over a period longer than, say,
ten years. With time, age structures vary, and these
296. In special circumstances it may be commendable variations will have effects on vital rates and migration
to project urban and rural populations on the basis of which this simple method does not make it possible to
crude birth, death and migration rates. Those should ~alculate-
be projections over short periods only, since no attention
78 G. W. Roberts, "Provisional assessment of growth of the
is given changes in the crude rates which may Kingston-St.An&ew area, 1960-70", Social and Economic Studies,
eventually result from the modification of age Structures. University of the West Indies (Kingston, Jamaica), December 1963,
Because of this limitation, the crude component methods pp. 432-441.
 3 1 ~ . PROJECTION
TABLE OF IRAN'S
URBAN AND RURAL POPULATION ACCORDING TO AGE GROUPS:
CALCULATION OF SEX-AGE SPECIFIC URBAN RESIDENCE RATIOS FROM IRANIAN
CENSUS DATA OF
1956 AND 1966

1956 population 1966 population

(thousands) Per cent (thousandsJ Per cent
urban urban
Sex and age a Total Urban in 1956 Total Urban in 1966
-
Both sexes . . . . . . . . . . . . . . . 31.4
Males .................. 31.9
0-4 ................. 28.0
5-9 . . . . . . . . . . . . . . . . . 30.0
10-14 . . . . . . . . . . . . . . . . . 34.1
15-19.. ............... 37.6
20-24. ................ 41.3
33.1
30.7
31 . O
31.3
31.5
50-54 . . . . . . . . . . . . . . . . 31.3
55-59 . . . . . . . . . . . . . . . . 30.5
60-64. . . . . . . . . . . . . . . . 29.7
65 and over .......... 28.5
Females . . . . . . . . . . . . . . . . . 31.0
'0-4.................. 27.3
5-9 .................. 29.6
10-14.. ................ 35.0
15-19.. ................ 35.9
20-24. ................. 33.5
29.9
29.6
29.6
32.6
33.1
50-54.. . . . . . . . . . . . . . . 33.0
55-59. ............... 31.2
60-64. ............... 30.8
65 and over .......... 28.9

a Excluding population of unknown age.

* Age groups from 25 to 29 upward obtained by halving the corresponding ten-year age groups by
means of Newton's formula.

COMPOSITE
METHOD EMPLOYING URBAN RESIDENCE scale. The reader will recall that there are varying
RATIOS FOR FIXED SEX-AGE GROUPS constraints on the rise in a percentage, depending on
301. In this method, it is considered that the pro- whether this is at a low, intermediate or high level, and
the logistic scale takes this fact into account.
pensity of the national population to reside in urban areas
varies by groups of sex and age. An exception will have 303. Tables 31a, 31b and 31c illustrate the procedure
to be made for children at those ages where they would involved with the use of 1956 and 1966 census data for
normally reside with their mothers, but this complication Iran. In this connexion, it is pointed out that some
will be disregarded at first. From the data of two suc- modifications were made in the published census data.
cessive censuses, sex-age specific urban residence ratios The data for 1956 have been published by five-year age
can be calculated. It can then be assumed that in future groups up to age 25, and by ten-year groups for ages 25 to
periods the rise in these ratios, as inferred from the two 34, 35 to 44, and so forth. To obtain estimates by
censuses, will continue at the same tempo. five-year groups, the ten-year groups were halved with
302. The residence ratios are best expressed as percen- the use of Newton's formula. To maintain compar-
tages in each national sex-age group residing in urban ability, age groups 25 to 29, 30 to 34 and so forth up to
areas. Because of the considerations first introduced
in chapter 11, second section, and also used in part B
80f -
na - 112 [fn + 118 (fn-1 - fn+i)], where fna is the number in
the first half of the given ten-year group, to be computed; fn is the
of chapter VII, the "tempo" of rise in these ratios given number in the entire ten-year group; and fn-, and fn+l are
(expressed as percentages) will be measured on the logistic numbers in the preceding and following ten-year groups.
 TABLE
318. PROJECTION
OF IRAN'S
URBAN AND RURAL POPULATION ACCORDING TO AGE GROUPS:
DERIVATION OF LOGISTIC LEVELS OF URBAN RESIDENCE RATIOS IN 1956 AND 1966, AND PROJECnON
OF THESE LEVELS TO MID-YEAR 1975

Logistic level
Level
Census Census Ten-jvar 8.62-year projected to
Sex and age 1956 1966 rise in level rise in level mid-year 1975

Both sexes ................. ... ... ... ...

Males ..................... ... ... ... ... ...
0-4 . . . . . . . . . . . . . . . . . . . . - 94 - 62 32 28 - 34
5-9 .................... - 85 - 58 27 23 - 35
10-14 .................... - 66 - 32 34 29 -3
15-19. ................... -51 - 12 39 34 +22
20-24 . . . . . . . . . . . . . . . . . . . . - 35 f5 40 34 -t39
25-29. . . . . . . . . . . . . . . . . . . . - 70 - 34 36 31 -3
30-34. ................... - 81 - 47 34 29 - 18
35-39. ................... - 80 -51 29 25 - 26
40-44 .................... - 78 - 49 29 25 - 24
45-49. ................... - 78 -43 35 30 - 13
50-54. ................... - 79 - 40 39 34 -6
55-59 .................... - 82 -49 33 28 - 21
60-64. ................... - 86 - 52 34 29 - 18
65 and over .............. - 92 - 61 31 27 - 30
Females ................... ... ... ... ... ...
0-4 .................... - 98 - 59 39 34 - 25
5-9 .................... - 87 - 56 31 27 - 29
10-14 .................... - 62 - 30 32 28 -2
15-19 .................... - 58 - 29 29 25 +4
20-24. ................... - 69 - 31 38 33 + 2
25-29. ................... - 85 -47 38 33 - 14
30-34. ................... - 87 - 52 35 30 - 22
35-39 .................... - 87 -47 40 34 - 13
40-44 .................... - 73 -43 30 26 - 17
4549. ................... - 70 - 34 36 31 -3
50-54. . . . . . . . . . . . . . . . . . . . - 71 - 30 41 35 $5
55-59 .................... - 79 - 38 41 35 -3
60-64 .................... -81 -46 35 30 - 16
65 and over .............. - 90 - 55 35 30 -25

age 65 were recalculated for the 1966 census by forming and age. The corresponding rural population is then
ten-year groups and halving them according to the same obtained by subtraction of the urban from the projected
formula; the noted fluctuations among adjacent five- total.
year groups were thereby mostly eliminated. 305. It will be noted that the levels or percentages
304. Table 31a presents the data of the censuses held projected to 1975 are affected by considerable fluctu-
in November 1956 and in November 1966 and the ations. These may in part be due to inaccuracies in the
corresponding urban residence ratios. These ratios original sex-age data, but to a large extent they may also
correspond to levels in the logistic curve of annex I reflect the effects of past fluctuations in urban birth rates
and the logistic levels are entered in the first two columns (relative to the rural rates), in migration, or both. Since
of table 31b. Differences between them indicate the the effects of true fluctuations move up in age as time
"tempo" at which rises have occurred during the exact progresses, the method of projection is partly at fault.
ten years of the census interval. These rises are then On the other hand, the comparatively high levels of
multiplied by 0.862, to correspond to the 8.62-year urbanization among persons aged 15 to 24, especially
interval between the census of November 1966 and the males, are likely to reproduce themselves also at the
date of mid-year 1975. Added to the levels of Novem- same ages in the future: it is highly likely that much of
ber 1966, they indicate levels to be expected in 1975. the urban residence of such young persons has only a
In table 31c, the logistic levels projected to 1975 are temporary character, and while many return sooner or
converted to the corresponding percentages (see annex I). later to rural localities, they tend to be replaced by other
The country's total population, projected by sex-age young persons, again of the same ages.
groups to 1975, is then multiplied with these percentages 306. The reader may wish to compare the projection
to yield projected numbers of urban population, by sex in table 31c with projections of the Iranian urban popu-
 TABLE
3 .
1 ~PROJECTION OF IRAN'S URBAN AND RURAL POPULATION ACCORDING TO AGE GROUPS:
PROJECTED LOGISTIC LEVELS. URBAN RESIDENCE RATIOS AND NUMBERS OF TOTAL. URBAN AND RURAL
POPULATION. MID-YEAR 1975

Population (thousands)
Logistic Per cent
Sex and age level urban Total Urban R~rral

Both sexes ..................

Males .....................
0-4 ....................
5-9 ....................
10-14 ....................
15-19 ....................
20-24 ....................
25-29 ....................
30-34 ....................
35-39 ....................
40-44 ....................
45-49 ....................
50-54 ....................
55-59 ....................
60-64 ....................
65 and over ..............
Females ...................
0-4 ....................
5-9 ....................
10-14 ....................
15-19 ....................
20-24 ....................
25-29 ....................
30-34 ....................
35-39 ....................
40-44 ....................
45-49 ....................
50-54 ....................
55-59 ....................
60-64 ....................
65 and over ..............

lation by sex and age groups presented in the two proced- years as the time interval of the calculation . Persons
ing chapters. in tables 21c and 28 . Whereas the methods of the same cohort may become urbanized to an increas-
differ. the results do not differ very greatly . Which of ing extent. as time passes. if more of them migrate
the three projections is the "best" cannot be ascertained from rural to urban areas than from urban to rural. also
until a new census is taken . if localities in which some of them reside become reclassi-
fied from rural to urban . This method of calculation
COMPOSITE
METHOD EMPLOYING URBAN is slightly more laborious than that of the foregoing
RESIDENCE RATIOS BY COHORT example. and will presently be illustrated .
.
307 In the foregoing example. urban residence ratios 308 . The question arises. of course. whether projeo
specific by age group were projected to rise. but they were tions by cohort are more pertinent than those with age-
assumed to remain applicable to the same age groups specific rates of urbanization . The answer depends on
also in the future. The population of persons of the the degree of permanence of urban residence once a
same age at different dates. however. does not comprise migrant has entered the urban area . As was discussed
the same individuals because. with time. a population in chapter 11. the process of urbanization has varying
of given individuals advances in age. The method can. features in different parts of the world . By and large.
however. be modified so that persons of the same cohort the settlement of rural migrants in urban areas tends to be
are followed whose ages advance by the same number of of longer duration. or greater permanence in Latin
America and in the more developed regions. than in
many parts of Africa and Asia . In Africa and Asia
.
.

el BY "cohort" we understand a group of persons born in the same

period of time and advancing in age as time progresses. young men typically migrate temporarily or seasonally
 TBLE 3 2.~ PROJECTIONOF BRAzL'SURBAN AND RURAL POPULATION ACCOR~INGTO AGE GROUPS:
CALCULATION OF SEX-AGE SPECIFIC URBAN RESIDENCE RATIOS FROM BRAZILIAN
CENSUS DATA OF
1950 AND 1960
1950 population 1960 population
(thousands) Per cent (thousands) Per cent
urban urban
Sex and age Total Urban in 1950 Total Urban in 1960

Both sexes............... 51. 944 18.784 36.16

Males ................... 25. 884 8. 971 34.66
0 4 .................. 4. 236 1. 273 30.05
5-9 .................. 3. 561 1. 026 28.81
10-14 .................. 3. 165 976 30.84
15-19 .................. 2. 645 922 34.86
20-24 .................. 2. 384 901 34.79
25-29 .................. 2. 030 779 38.37
30-34a ................ 1. 702 662 38.89
35-39 .................. 1. 444 563 38.99
40-44 .................. 1. 235 489 39.35
45-49 .................. 1. 011 403 39.86
50-54 .................. 775 309 39.87
55-59 .................. 584 234 40.07
60-64.................. 426 169 39.67
65-69 .................. 303 119 39.27
70 and over b .......... 383 146 38.12
Females ................. 26. 060 9. 813 37.66
0-4 .................. 4. 135 1. 256 30.37
5-9 .................. 3. 455 1. 027 29.73
10-14 .................. 3. 144 1. 047 33.30
15-19 .................. 2. 858 1. 093 38.24
20-24 .................. 2. 607 1. 069 41.00
25-29 .................. 2. 102 877 41.72
30-34a ................ 1. 732 734 42.38
35-39 .................. 1. 408 608 43.18
40-44 .................. 1. 175 515 43.83
45-49 .................. 944 42 1 44.60
50-54 .................. 732 331 45.22
55-59 .................. 558 258 46.24
60-64.................. 412 196 47.57
65-69 .................. 311 151 48.55
70 and over b .......... 487 230 47.23

a Age groups from 30 to 34 upward obtained by halving the corresponding ten-year age groups by
means of Newton's formula .
* Including persons of unknown age .

to urban areas in search of cash income needed by their of a cohort will occur . By middle age. persons in
families who remain in the rural areas. Usually they these areas are more or less permanently committed to
return to their rural residences when sufficient cash has urban or rural occupations and urban or rural life
been accumulated. Hence. fixed age calculations may styles.
be more appropriate for many countries in Africa and 309. Data on urban and rural population by groups
Asia where migrations are often temporary. while cohort of sex and age are now available from three Brazilian
calculations may be more appropriate for other areas censuses. those of 1950. 1960 and 1970. This circurn-
where migration (or the lack of it) is more often a per- stance makes it possible to submit the projection methods
manent experience which depends on the particular to a test . In what follows. two composite projections
circumstances existing during the young adult ages of of the urban population of Brazil are made. both based
each individual cohort . Such occurrences as economic on the same data of the 1950 and 1960 censuses. and the
depressions may permanently curb urban migration in results of both for the year 1970 will then be compared
the contemporary cohort of young adults. while wars with census data for 1970 in order to ascertain which of
or economic prosperity in the cities at another date may
sa There is. however. a strong indication that nearly everywhere
swell the tide of urban migration in another cohort of rural widowed women tend to migrate to urban places See .
young adults. Once having reached middle age. there "Urban-rural differences in the marital status composition of the
is rather less chance that a change in the residence pattern population" (ESA/P/WP.51).
 328. PROJECTION
TABLE OF BRAZIL'S
URBAN AND RURAL POPULATION ACCORDING TO AGE GROUPS:
DERIVATION OF LOGISTIC LEVELS OF URBAN RESIDENCE RATIOS IN 1950 AND 1960, AND PROJECTIONS

Percentage urban Logistic level

Rise in 1970
Sex and age 1950 I960 1950 1960 level projected

Both sexes ...............

Males ...................
0-4 . . . . . . . . . . . . . . . . . .
5-9 ..................
10-14 ..................
15-19 ..................
20-24. .................
25-29.. ................
30-34. .................
35-39 . . . . . . . . . . . . . . . . . .
40-44. . . . . . . . . . . . . . . . . .
45-49. . . . . . . . . . . . . . . . . .
50-54. . . . . . . . . . . . . . . . . .
55-59. . . . . . . . . . . . . . . . . .
60-64 ..................
65-69. .................
70and over ............
Females .................
0 4 ..................
5-9 ..................
10-14 ..................
15-19. .................
20-24 ..................
25-29 . . . . . . . . . . . . . . . . . .
30-34. . . . . . . . . . . . . . . . . .
35-39. . . . . . . . . . . . . . . . . .
40-44 ..................
45-49. .................
50-54. .................
55-59. . . . . . . . . . . . . . . . . .
60-64. .................
65-69 ..................
70 and over ............

the two methods has yielded the more accurate results. at different dates. How this can be overcome is illus-
The first projection is made according to fixed age groups, trated in table 33a, part of which is a rearrangement of
and the second according to cohorts. table 32b. Three sets of age groups apply to the same
310. The Brazilian census data are by five-year age individuals according to the three dates, 1950, 1960 and
group up to age 29, and then by ten-year groups such 1970. 8Q The logistic levels for 1950 and 1960 are the
as 30 to 39, 40 to 49 and so forth. Again, Newton's same as those of table 32b, but they have now been
formula * is used to halve the ten-year groups and placed in those lines where the age groups apply at the
estimate the population in each five-year group. particular date. Thus, the logistic urbanization level of
males aged 0 to 4 in 1950 was -84.5 in that year. And
311. The projection by specific age groups is taken for the same cohort, i.e. males aged 10 to 14 in 1960,
up first, and it is illustrated in tables 32a, 32b and 32c. the level in 1960 was -33.9, this being an advance by
Little explanation is needed because the same method +50.6 for the identical cohort, as followed through time
was used in the projection for Iran, in tables 31a, 31b and while their age advances.
31c. The Brazilian case is simpler because both the
first census interval (1950-1960) and the postcensal 313. The assumption of the projection by cohorts is
period (1960-1970) comprise exactly ten years. that rises in each cohort's urbanization level which
occurred during 1950-1960 may occur again in the same
312. If the procedure is to be carried out by identical
cohorts-and not by fixed age groups-a complication 84 A simpler method of presentation would be to identify cohorts
arises because the same individuals are of different age according to the years in which they were born. For instance,
those aged 15-19 in 1950, 25-29 in 1960 and 35-39 in 1970 have
88 See foot-note 80. originated from births which occurred during 1930-1935, etc.
 3 2.~ PROJECTION
TABLE OF BRAZIL'S
URBAN AND RURAL POPULATION ACCORDING TO AGE GROUPS:
PROJECTED LOGISTIC LEVELS. URBAN RESIDENCE RATIOS AND NUMBERS OF TOTAL. URBAN AND RURAL
POPULA~ON. 1970

1970 population (thousands)

Logistic Per cent
Sex and age level urban Total a Urban Rurol

Both sexes ............... ... ... 93. 204 53. 282 39. 922
Males ................... ... ... 46. 330 25. 668 20. 662
04 .................. +15.7 53.92 7. 020 3. 785 3. 235
5-9 .................. +-16.5 54.12 6. 730 3. 642 3. 088
10-14 .................. +13.0 53.25 5. 850 3. 115 2. 735
15-19 ..................
20-24 ..................
-2.1 49.48 4. 934 2. 441
.
2. 493

25-29 ..................
30-34 ................
$28.2
+32.6
$42.6
57.00
58.08
60.49
.
4. 063
3 202
2. 857
2. 316
1. 860
1. 728
1 747
1. 342
1. 129
35-39 .................. $47.0 61.54 2. 460 1. 514 946
40-44 .................. +27.7 56.88 2. 209 1. 256 953
45-49 .................. $20.7 55.00 1. 879 1. 033 846
50-54 .................. +30.1 57.47 1. 510 868 642
55-59 .................. $32.3 58.01 1. 193 692 501
60-64 .................. +28.3 57.03 894 510 384
65-69 .................. f33.6 58.32 656 383 273
70 and over .......... +40.8 60.06 874 525 349
Females ................. ... ... 46. 874 27. 614 19. 260
0-4 .................. +22.6 55.63 6. 879 3. 808 3. 071
5-9 .................. $24.2 56.02 6. 571 3. 681 2. 890
10-14 .................. +25.1 56.24 5. 816 3. 271 2. 545
15-19 . . . . . . . . . . . . . . . . . . +20.9 55.21 5. 270 2. 910 2. 360
20-24 .................. 1-35.8 58.86 4. 359 2. 566 1. 793
25-29 .................. $48.8 61.96 3 . 344 2. 072 1. 272
30-34 ll ................ +60.3 64.63 2. 963 1. 915 1. 048
35-39 .................. $64.0 65.48 2. 502 1. 638 864
40-44 .................. +45.8 61.25 2. 179 1. 335 844
45-49 .................. t40.9 60.08 1. 828 1. 098 730
50-54 .................. $50.6 62.39 1. 482 925 557
55-59 .................. f57.1 63.90 1. 170 748 422
60-64 .................. +56.3 63.71 869 554 315
65-69 .................. +59.4 64.43 649 418 231
70 and over .......... +75.1 67.94 993 675 318

Census results (for totals only) .

a
Age groups from 30 to 34 upward obtained by halving the corresponding ten-year age groups by
means of Newton's formula .
Including persons of unknown age .

age groups (i.e. in the cohort born ten years more recently. 314. Two observations must be made. however. one
hence ten years younger) during 1960.1970 . In table 33a. trivial and the other rather important . It will be noted
the assumed 1960-1970 rises in logistic level are the same that calculations are carried out up to a terminal age
as those observed in 1950.1960. except that they are moved group of 70 years and over . Individuals of such ages
up by ten years. For example in the 1945-1950 cohort. at each given date were aged 60 years and over at a date
next to the last column of table 33a. the 1960-1970 ten years previously. To be exact. the calculation should
difference of 63.5 was assumed to be equal to the 1950- have been carried out in terms of rises of logistic levels
1960 difference for the 1935-1940 cohort (shown in the
previous column). since those two cohorts were the same from a combined age group 60 years and over at one
age at dates ten years apart . The 1960-1970 rises are date (i.e. the sum of those aged 60 to 64. 65 to 69 and
then added to the levels ascertained for each cohort 70 and over) to a terminal group of 70 years and over
in 1960 so as to arrive at projected levels in 1970 when each at the next date . This was not done here mainly for
cohort is again ten years older . The procedure is reasons of simplicity. and because the resulting error is
completed in table 33b. where the logistic levels are unlikely to be very significant.
converted to percentage levels with the use of the table 315. More important. it will be noted that this method
in annex I. and the resulting percentages are applied to of projection cannot be applied to individuals aged
the census data on sex-age composition of the total popu- 0 to 4 and 5 to 9 years in 1970. because none of them had
lation in 1970. been born as yet in 1960. and there are no usable national
 3 3 ~ . PROJECTION
TABLE OF BRAZIL'S
URBAN AND RURAL POPULATION ACCORDING TO COHORTS:
LOGISTIC LEVELS OF URBAN RESIDENCE RATIOS IN 1950 AND 1960,DIFFERENCES, AND PROJECTIONS OF LEVELS TO 1970
Logistic level
Sex and age
Rise in level
Age in Age in Age in 1970,
Year of birth 1950 1960 1970 1950 1960 1950-1960 a 1960-1970 projected

Both sexes ............

Males ................
1965-1970 ..........
1960-1965 ..........
1955-1960 ..........
1950-1955 ..........
1945-1950 ..........
1940-1945 ..........
1935-1940 ..........
1930-1935 ..........
1925-1930 ..........
1920-1925 ..........
1915-1920 ..........
1910-1915 ..........
1905-1910 ..........
1900-1905 ..........
1895-1900 ..........
1890-1 895 ..........
1885-1890 ..........
Females ..............
1965-1970 ..........
1960-1965 ..........
1955-1960 ..........
1950-1955 ..........
1945-1950 ..........
1940-1945 ..........
1935-1940 ..........

* Observed rise.
b Assumed rise.
Assumed rise added to level observed in 1960.
a For simplicity, the assumed rise is added to level of 60 to 64-year age group observed in 1960.

statistics on births in Brazil. 86 For these groups, there- would remain about the same. 86 The following ratios
fore, special calculations had to be made, which are were defined :
explained below. Children (males and females separately) aged 0 to 4
years per woman aged 20 to 39 years; and
316. The assumption was made that urban levels of
child-woman ratios, relative to rural levels, would be 86 The assumption is not exact because child-woman ratios,

similar in 1970 to what they were in 1950 and 1960. however defined, do not measure fertility accurately. Some of the
children have died, and rates of infant and child mortality may differ
This is to say, very roughly, that the ratio of fertility in between urban areas and the rest of the country. More important,
urban areas relative to fertility in the whole country not all children necessarily reside with their mothers. Detailed
observations lead to the impression that some women who have
migrated from rural to urban areas tend to leave some of their
The matter would be different in a country with good birth children under the care of rural relatives; this seems to be more often
registration statistics, where urban and rural birth cohorts of pre- the case with boys than with girls, and more often with bigger
censal and postcensal periods can be compared with eventual children (aged 5 to 9) than with small children (aged 0 to 4). This
survivors aged 0 to 4 and 5 to 9 years in the urban and rural areas. possibly interesting topic, however, has so far been little investigated.
 3 3. ~ PROJECTION
TABLE OF BRAZE'SURBAN AND RURAL POPULATION ACCORDING TO COHORT: PROJECTED LOGISTIC LEVELS.
URBAN RESIDENCE RATIOS AND NUMBERS OF TOTAL. URBAN AND RURAL POPULATION. 1970

Population (thousands)
Sex and age
Year of birth in 1970 Logistic level Per cent urban Total a Urbon Rural

Both sexes ...............

Males ................... ...
1965-1970 ............. 0-4
1960-1965 ............. 5-9
1955-1960 ............. 10-14
1950-1955 ............. 15-19
1945-1950 ............. 20-24
1940-1945 ............. 25-29
1935-1940 ............. 30-34
1930-1935 ............. 35-39
1925-1930 ............. 40-44
1920-1925 ............. 4549
1915-1920 ............. 50-54
1910-1915 ............. 55-59
1905-1910 ............. 60-64
1900-1905 ............. 65-69
Up to 1900 ............ 70 and over
Females .................
1965-1970 ............. 0-4
1960-1965 ............. 5-9
1955-1960 ............. 10-14
1950-1955 ............. 15-19
1945-1950 ............. 20-24
1940-1945 ............. 25-29
1935-1940 ............. 30-34
1930-1935 ............. 35-39
1925-1930 ............. 40-44
1920-1925 ............. 4549
1915-1920 ............. 50-54
1910-1915 ............. 55-59
1905-1910 ............. 60-64
1900-1905 ............. 65-69
Up to 1900 ............ 70 and over

a Census results; age groups from 30 to 34 upward obtained by halving the corresponding ten-year age groups by means of Newton's formula.
For method of calculation. see text.

Children (males and females separately) aged 5 to 9 years and again in respect of data for 1960. It was then
per woman aged 25 to 44 years. assumed that in 1970 the ratio between residence ratios
317. Actually the calculations were carried out in might be an average of those in 1950 and 1960. The
terms of relative urban residence ratios . The urban urban residence ratios of women aged 20 to 39 or 25 to
residence ratios of children (separately for males and 44 have already been projected to 1970. and by applying
females. and for 0 to 4 and 5 to 9 age groups) were the assumed ratios between residence ratios the urban
calculated. and also the average of urban residence residence ratios for children (males and females. 0 to 4
ratios of women who might be their mothers (aged and 5 to 9) were obtained . These are entered in the
20 to 39 and 25 to 44). Then a ratio between the resi- appropriate place in table 33b. 89
dence ratios of the children and their mothers was It was assumed that. in the case of ratios. the appropriate
calculated. This was done in respect of data for 1950. average should be the geometric mean; actually. an artihmetic mean
might have sufficed. and results would have been nearly the same.
Take. for example. the case of males in the age group 0 to 4 .
the mothers Of may in some instances be From table 32a it can be calculated that the arithmetic average of the
: ~ ~ ~ ~ ~ ~ ~ r$ccdsEh
t GF :$:
~ ~ l
calculation might be further refined by assigning weights to numbers
urban residence ratios (URR). i.e. the percentage urban. for women
~ aged20
~ ~
to39is~ 42.07~ in 1950
h and$ 52.52
~ in ,1960". For themalesaged
to it is 30.05 in 1950 and 41.49 in 1960. The ratio of the ratios.
of women of different ages according to the comparative likelihood i.e. the ratio of the URR. of the children over the mothers.
that they might be mothers of children aged either 0 to 4 or 5 to 9
years. But such refinement would be of interest only if there were 41 49
more detailed information about possible trends and age patterns in (%42.07
in 1950 and - 52.52 in 1960) equals .7143 in 1950 and
urban and rural fertility. which is not the case in the present example. .7900 in 1960 . (Continued on next page.)
 1950 1960
- --
Male Female Male Female
Residence ratio o/. children children children children
- -- ---

Children aged 0 to 4 . .................... 30.05 30.37 41.49 42.47

Women aged 20 to 39 a .................. 42.07 42.07 52.52 52.52
Ratio between residence ratios . . . . . . . . . . . 0.7143 0.7219 0.7900 0.8086
Children aged 5 to 9 . .................... 28.81 29.73 40.86 42.34
Women aged 25 to 44 a .................. 42.78 42.78 53.20 53.20
Ratio between residence ratios ........... 0.6734 0.6950 0.7680 0.7959

a Arithmetic mean of residence ratios in the corresponding five-year age groups.

318. This auxiliary calculation is interesting for its own erred considerably with respect to children aged 0 to 4
sake, because it indicates, however roughly, that relative and 5 to 9 in 1970. The projection by fixed age groups,
urban fertility (i.e. relative to fertility in the whole because of an apparent urban fertility rise between 1950
country) evidently underwent some change between 1950 and 1960, implies a continuous urban fertility rise until
and 1960, with implications for 1970 which, in the 1970, and therefore over-estimates urban children in
absence of the new census, would have been difficult to 1970, especially the children aged 0 to 4. The projection
foresee. For this reason, the detail is presented below. by cohorts assumes that during 1960-1965 and 1965-
319. For children aged 0 to 4, the ratio between child- 1970, fertility would be at levels averaging those of
ren's and women's residence ratios in 1950 was 0.7143 1940-1945 and 1950-1955, and those of 1945-1950 and
for boys and 0.7219 for girls; by 1960, these ratios had 1955-1960. In this projection, urban children aged
risen to 0.7900 and 0.8086, respectively. There was a 0 to 4 in 1970 were slightly underestimated, and those
similar rise in the ratio between children's and women's aged 5 to 9 in 1970 were considerably underestimated.
residence ratios in respect of children aged 5 to 9. Aside It becomes evident that in actual fact, urban fertility
from the varying frequency with which migrant women relative to the fertility of the whole country continued
may leave their children with rural relatives and varia- to rise during 1960-1965 (hence the large underestimate
tions in infant mortality, this rising tendency, from of children aged 5 to 9), but declined significantly during
1950 to 1960, reflects a rise in urban fertility relative to 1965-1970 (hence the smaller underestimate of children
the fertility level of the whole country. The calculation aged 0 to 4). But this trend would have been difficult to
presents an opportunity to make varied assumptions. foresee in a projection when the 1970 census data were
Urban fertility (relative to that of the entire country) not yet available.
may be assumed to rise even further until 1970; it may be 322. Past the age of 10 years in 1970, with a few excep-
assumed to decline again by 1970, possibly below the tions, the errors in the projection by cohorts are markedly
1950 level; and it may be assumed to reach a level in smaller than the errors in the projection according to
1970 which is intermediate between the levels of 1950 and fixed age groups. It is worth while to consider the distri-
1960. Actually, this latter assumption was made to bution of errors according to sex and age and to reflect
arrive at the results presented in table 33b. on their possible causes.s1 On the whole it appears
320. Two projections of Brazil's urban population in that the cohort method is the more reliable of the two,
1970 have now been presented, one in terms of fixed age though actual events will always deviate more or less
groups (table 32c), the other in terms of cohorts (table 33b). from any projection. Rural-to-urban migration and
Both were calculated in relation to 1970 census data on rural-to-urban area reclassification may at times gather
sex-age composition of the total national population. momentum and ebb off at other times, and the selection
Actually, however, 1970 census data detailed by sex and of migrants by sex and age may vary as time progresses.
age are also available separately for the urban and rural
population. It is possible, therefore, to compare the MIGRATION-SURVIVAL
METHOD AS APPLIED
projections with actual data for 1970, and to indicate TO THE RURAL POPULATION
which of the two projections has been more accurate. 323. The method now to be described resembles the
The comparison is carried out in table 34. cohort-survival method in so far as use is made of a
321. Discrepancies between the projections and actual type of ratio similar to "survival ratios". But survival
observations are indicated in the last two columns of ratios, as used in the cohort-survival method, indicate
table 34. It will be noted that both projections have only the progressive decimation of each cohort as a
result of deaths; as regards the consequences of migration
(Continued)
and of rural-to-urban area reclassification, separate
The geometric mean of these = 1/(.7143) (.7900) = .751199. Then
the arithmetic mean of URR's for women aged 20 to 39 in 1970 can The trend cannot be measured accurately because of possible
be calculated from table 32b to be 62.20. The ratios of ratios are changes in infant mortality and in the completeness of enumeration
URR, males 0 to 4,1970 = .751 99) of small children (defective in most censuses).
then set equal to each other
( .6220
and the solution of 46.72 per cent urban is posted as the first figure
O1 Another source of error difficult to evaluate is the inaccurate
reporting of age at each census; the accuracy of age reporting may
of the second column in table 33b. also vary from one census to another.
 TABLE
34. BRAZIL'S
URBAN POPULATION BY SEX AND AGE ACCORDING TO THE CENSUS OF 1970
AND ACCORDING TO TWO PROJECTIONS

Urban population, 1970 Error of projection

(thousands) (thousands)
-- -- - --
Projected by Projection by
Year of birth Sex and age in 1970 Census Age groups Cohorts Age groups Cohorts
-

Both sexes............. 52,098 53,282 51,616 +1,184 - 482

Males ................. 25,213 25,668 24,882 -1455 - 331
1965-1970 ........... 0 4 3,449 3,785 3,280 +336 - 169
1960-1965 ........... 5-9 3,489 3,642 3,036 +I53 - 453
1955-1960 ........... 10-14 3,130 3,115 3,162 - 15 + 32
1950-1955 ........... 15-19 2,668 2,441 2,728 - 227 + 60
1945-1 950 ........... 20-24 2,273 2,316 2,330 + 43 -43
1940-1945 ........... 25-29 1,795 1,860 1,783 + 65 - 12
1935-1940 ........... 30-34 a 1,650 1,728 1,739 + 78 + 89
1930-1935 ........... 35-39 1,470 1,514 1,479 +44 +9
1925-1930 ...........
1920-1925 ...........
40-44
4549
1,297
1,098
1,256
1,033
1,302
1,105
-41
- 65
++5
7
1915-1920 ...........
1910-1915 ...........
50-54
55-59
858
668
868
692
867
676
++10
24
++9
8
1905-1910 ........... 60-64 500 510 511 +10 +11
1900-1905 ........... 65-69 372 383 379 +11
+ +
7
Up to 1900 .......... 70 and over b 496 525 505 29 +
9
Females ............... 26,885 27,614 26,734 +729 - 151
1965-1970 ........... 0-4 3,388 3,808 3,269 +420 - 119
1960-1965 ........... 5-9 3,437 3,681 3,065 + 244 - 372
1955-1960 ........... 10-14 3,227 3,271 3,347 f 44 + 120
1950-1955 ........... 15-19 3,081 2,910 3,048 - 171 - 33
1945-1 950 ........... 20-24 2,586 2,566 2,683 - 20 $-97
1940-1945 ........... 25-29 2,010 2,072 2,019 +62 +9
1935-1940 ........... 30-34 a 1,826 1,915 1,851 +89 + 25
1930-1935 ........... 35-39 1,564 1,638 1,612 +74 +48
1925-1930 ........... 40-44 1,360 1,335 1,387 - 25 +27
1920-1925 ........... 4549 1,143 1,098 1,160 -45 + 17
1915-1920 ........... 50-54 923 925 926 + 2 f 3
1910-1915 ........... 55-59 732 748 735 +16 +3
1905-1910 ........... 60-64 551 554 557 + 3 + 6
1900-1905 ........... 65-69 417 418 423 + 1 +6
Up t o 1900 .......... 70 and over b 640 675 652 + 35 + 12

a Age groups from 30 to 34 upward obtained by halving the corresponding 10-year age groups by means of Newton's formula.
Including persons of unknown age.

calculations would have to be made in the cohort- 325. Table 35 presents the results of such an assuinp-
survival method. In the method now presented, the tion with respect to the rural population of Brazil.
"migration-survival ratios" incorporate the combined From numbers in identical rural cohorts in 1950 and 1960,
consequences of mortality, net rural-to-urban migration migration-survival ratios for the 1950-1960 period are
and rural-to-urban reclassification of areas without obtained by division of numbers for 1960 by numbers
considering their several effects separately. for 1950. The migration-survival ratios are then moved
324. As may be recalled from chapter IV, third section, up by ten years to apply to the same age-intervals again
constant rates of change in the rural population have a in the 1960-1970 period. Multiplication of numbers in
somewhat wider applicabiIity than constant rates of each cohort in 1960 by the rearranged migration-survival
change of urban population. There may be many ratios then yields a projection of the rural population
situations in which such an assumption produces satis- by sex-age groups for 1970. Again, age groups 0 to 4
factory results. What is true of the entire rural popu- and 5 to 9 require a separate calculation, the numbers
lation is also valid for individual rural population of births in the time interval being unknown. In the
cohorts. While advancing in age, each cohort tends to present instance, that calculation has been performed as
be depleted by the combined action of mortality and follows.
net transfers (migratory or by classification) to the urban 326. Ratios were calculated between numbers of child-
category, and it is at least conceivable that the combined ren aged 0 to 4 (males and females) and numbers of
rates of depletion remain approximately the same in the women aged 20 to 39, both in 1950 and in 1960; and, for
course of time. the same dates, ratios of children aged 5 to 9 (males,
 TABLE
35. PROJECITON
OF BRAZIL'SRURAL POPULATION WITH THE MIGRATION-SURVIVAL METHOD

Rural popr~lation Projected

Sex and age (thousands) Migration-survival ratios rural
population,
Age in 1950 .4ge in 1960 Age in 1970 1950 1960 1950-1960 1960-1970 1970
--

Both sexes............... 33,159 37,563 41,452

Males ................... 16,912 19,389 21,526
.................... ... 0-4 ... ... ... ... 3,5468
.................... ... 5-9 ... ... ... ... 3,1Wa
.................... 0-4 10-14 ... 3,342 ... 0.845 2,824
.................... 5-9 15-19 ... 3,051 ... 0.789 2,407
0 4 ................. 10-14 20-24 2,963 2,503 0.845 0.735 1,840
5-9 ................. 15-19 25-29 2,535 1,999 0.789 0.759 1,517
10-14 ................. 20-24 30-34 2,189 1,610 0.735 0.766 1,233
15-19 ................. 25-29 35-39 1,723 1,308 0.759 0.776 1,015
20-24 ................. 30-34 40-44 1,483 1,136 0.766 0.833 946
25-29 ................. 35-39 4549 1,251 971 0.776 0.826 802
30-34 ................. 40-44 50-54 1,040 866 0.833 0.755 654
35-39 ................. 45-49 55-59 881 728 0.826 0.717 522
40-44 ................. 50-54 60-64 746 563 0.755 0.751 423
4549 ................. 55-59 65-69 608 436 0.717 0.697 304
50-54 ................. 6044 70 + 466 335 0.751 0.438 384
55-59 ................. 65-69 ... 350 244 0.697 ... ...
60+ b ................ 70f ... 678 297 0.438 C ... ...
Females ................. 16,247 18,174 19,926
.................... ... 0-4 ... ... 3,3968
.................... ... 5-9 ,.. ... ... ... 2,957a
.................... 0-4 10-14 ... 3,155 ... 0.827 2,609
.................... 5-9 15-19 ... 2,884 ... 0.813 2,345
0 4 ................. 10-14 20-24 2,879 2,380 0.827 0.763 1,816
5-9 ................. 15-19 25-29 2,428 1,973 0.813 0.732 1,444
10-14 ................. 20-24 30-34 2,097 1,601 0.763 0.690 1,105
15-19 ................. 25-29 35-39 1,765 1,292 0.732 0.712 920
20-24 ................. 30-34 40-44 1,538 1,061 0.690 0.761 807
25-29 ................. 35-39 4549 1,225 872 0.712 0.776 677
30-34 ................. 40-44 50-54 998 759 0.761 0.721 547
35-39 ................. 45-49 55-59 800 621 0.776 0.686 426
4M4 ................. 50-54 60-64 660 476 0.721 0.666 317
4549 ................. 55-59 65-69 523 359 0.686 0.660 237
50-54 ................. 60-64 70 + 401 267 0.666 0.436 323
55-59 ................. 65-69 ... 300 198 0.660 ... ...
60fb ................ 70f ... 633 276 0.436C ... ...

a For method of calculation, see text.

i.e. 60 to 64, 65 to 69, 70 and over and persons of unknown age.
Ratio from combined ages 60 and over to ages 70 and over.

females) to numbers of women aged 25 to 44. The the urban population. In table 36, the projection for
average of ratios for 1950 and 1960 rsepectively were the rural population and the implied projection for the
then assumed to apply in 1970. Since numbers of women urban population are compared with the Brazilian census
of ages when they might be mothers of young children results of 1970. Reference to table 34 suggests that the
have already been projected, corresponding numbers of results of the present projection (by migration-survival
children can be obtained by multiplying the projected ratios in the rural population) are no worse than those
numbers of women of the relevant age groups with these obtained previously by the composite method using
average ratios. 92 levels of urban residence ratios by fixed age groups;
327. If either a projection or census data of the total but they are perhaps less accurate than most of the
population by sex-age groups in 1970 exist, subtraction of results obtained previously by the composite method
projected numbers in the rural population produces a using urban residence ratios by cohort.
corresponding forecast for the sex-age composition of 328. More in particular, the following errors appear to
be involved. Rural fertility has been considerably
gZ Alternative assumptions might have been made, implying
underestimated for 1960-1965 and appreciably also for
either a rise or a fall in rural fertility between 1960 and 1970. 1965-1970, and a change may also have occurred in the
 TABLE .
36 BRAZIL'S
TOTAL URBAN AND RURAL POPULATION BY SEX AND AGE IN 1970 ACCORDING TO CENSUS.
AND RURAL POPULATION AS PROJECTED BY MEANS OF MIGRATION-SURVIVAL RATIOS

Population according to 1970 census Plojected population Error in

(in thousands) (in thousands) projected
urban
Sex and age Total Urban Rural Urban Rural population

Both sexes...............
Males ...................
0-4 .................
5-9 .................
10-14 .................
15-19 .................
20-24 .................
25-29 .................
30-34 ................
35-39 .................
40-44 .................
45-49 .................
50-54 .................
55-59 .................
60-64 .................
65-69 .................
70 and over ..........

Females .................
0-4 .................
5-9 .................
10-14 .................
15-1 9 .................
20-24 .................
25-29 .................
30-34 ................
35-39 .................
40-44 .................
45-49 .................
50-54 .................
55-59 .................
60-64 .................
65-69 .................
70 and over ..........
The "projected" urban population is obtained by subtracting the projected rural population from the total population enumerated at the
census.
* Age groups from 30 to 34 upward obtained by halving the corresponding 10-year age groups by means of Newton's formula .
Including pmons of unknown age .

relative frequency (impossible to calculate) with which ages past 30. male rural-to-urban migration has been
migrant mothers leave either their sons or their daughters underestimated slightly. and female rural-to-urban migra-
under the care of rural relatives. Rural-to-urban migra- tion has been over-estimated slightly. The inferences
tion of young persons has been underestimated. which are. however. debatable. partly owing to the influence of
seems in actual fact to have intensified during 1960.1970. additional factors (e.g. area reclassification and urban-
especially at the ages 15 to 19 and 25 to 29.93 At most rural differences in mortality). and partly owing to
possible errors in census enumeration and age reporting.
That the apparent underestimate in the intervening 20 to 24 age Perhaps the original age data should have been subjected
group of females was less is probably a somewhat complex conse
quence of variations in the accuracy of age statements from one
to some smoothing before being used in the calculation
census to another . of migration-survival ratios.
 Chapter IX
COHORT-SLRVIVAL METHOD

329. In this chapter, projections of urban and rural and all procedures involved has rarely if ever been
population are developed in such a way that it becomes published. The reason is simple. Usually many experi-
possible to compare the demographic consequences of mental calculations are needed and many auxiliary
alternative assumptions regarding each of the component estimates must be constructed, especially if the method
factors of urban and rural population change: initial is a fairly refined one. The complete description of all
size and sex-age composition of the urban and rural these operations would require an undue amount of
population; urban and rural fertility and its incidence space. But although a full description of methods can
by age groups of women; urban and rural mortality and be space-consuming, it does not follow that the methods
its incidence by groups of sex and age; and rural-to-urban are mathematically complicated. With adaptations as
population transfers, whether by migration or area needed, these methods can be used for the projection of
reclassification, their volume and sex-age composition. urban and rural populations also in countries with less
Valid comparisons of the results of modification in any detailed or less accurate statistics, although actually
one of these factors are possible only if the projections this has been done very seldom.
are calculated on the basis of such detail. In short, the 332. In the case of Brazil, selected for this chapter, it
cohort-survival method is indispensable for such a is to be noted that there exist no comprehensive statistics
purpose. on births and deaths. Such statistics would be valuable
330. It is recognized, however, that without recourse to for a projection, particularly if births and deaths were
electronic computers the calculations cannot be carried registered separately in urban and rural areas and if
to a high degree of precision. As has been explained in those areas were not affected by reclassification. But even
chapter 11, more than one demographic event can occur some of the most accurate vital statistics systems distin-
to the same individual within any given time period, hence guish these vital events only by place of occurrence,
fertility, mortality and migration (also reclassification) rather than by place of usual residence. Because of the
are partly interdependent functions whose precise inter- urban location of many hospitals, births and deaths to
action, moment by moment of time, would have to be rural residents may often occur within an urban area,
simulated in a computer model. In calculations with thus falsifying somewhat the statistics for purposes of a
desk equipment, the interaction of the several factors projection of the urban and rural resident population.
must be artificially simplified. As will be shown, it is Nor can there be any continuous series of birth and death
most convenient to begin with mortality, making use of the statistics, except within constant administrative boun-
continuous life-table function. Next, it is advisable to daries.
measure and project the action of rural-to-urban popu- 333. On the other hand, the case of Brazil has the
lation transfers as already affected by mortality interven- advantage that censuses have been taken every ten years;
ing within the period of the calculation. Finally, it is that the definition of "urban" localities, while geograph-
practical to deal with an imperfect measurement of ically flexible, has remained the same; 94 and that the
fertility, as ascertained for the end of each period after accuracy of age reporting is at least tolerable. In coun-
the intervening effects of mortality and population trans- tries where these conditions are not met, various adjust-
fers. The rates of transfer and the rates of fertility, both ments to data and auxiliary estimates would first have to be
as calculated for past periods and as projected, lack preci- made, requiring a lengthy description. gS
sion, but the consequences of such imperfect measurement
upon the projection are only of small importance since 334. In judging these projections, it should also be
the method of measurement, for the past as well as the borne in mind that the case of Brazil may be fairly typical
future, remains the same. of Latin America, but not of other parts of the world.
As noted in chapter 11, in Latin America a majority of the
33. Because of the space needed for exposition, a rural-to-urban migrants are female, and many former
variety of possible results is illustrated in this chapter
for one case only, namely Brazil. More cases might have O4 In the Brazilian censuses, localities are defined as "urban" if
been worked out, including countries with more detailed they constitute urban and suburban zones of administrative centres
and accurate statistics, and others whose data are less of municipios or districts. The geographical boundaries of these
detailed and less accurate, but the essential methods "zones" are flexible and apparently undergo much expansion from
would have been nearly the same also in other examples. one census to another.
9S The estimation and adjustment of basic data for a population
Such detailed projections of urban and rural population projection has been discussed in Manual 111: Methods for Population
have been carried out in a number of countries with Projections by Sex and Age (United Nations publication, Sales
excellent statistics, but a full exposition of all data used No. 56.XIII.3).
migrants reside in the cities indefinitely. In these respects, more exact level will not cause any large error in the corres-
urbanization in Africa or Asia has quite different charac- ponding population projections. For this reason, it was
teristics. Similar calculating methods may be used for considered sufficient to estimate mortality in Brazil only
projections in other countries, but the comparison of rather roughly. The urban and rural expectations of life
results would lead to quite different conclusions. through the 1965-1970 period shown below were derived
335. The present chapter is organized in three parts. by arbitrary estimation from the life expectations of the
In part A, it is shown how the detailed estimates and total population. The following considerations were
assumptions underlying the projections have been derived. observed :
Part B then describes the calculation procedures for one (a) Rural expectation of life was assumed to be lower
particular combination of assumptions, which will be than the urban expectation. Advanced economic and
referred to as the "standard projection". In part C , social conditions are found, for instance, in Brazil's
finally, the results of the "standard projection" are south, much of whose population is urban. In Brazil's
compared with those of eight "variant projections", north, much of whose population is rural, conditions are
each of which differs from the "standard" in one or an- still comparatively backward. With the lack of pertinent
other single respect. Many possible refinements of detailed statistics, it is a matter of conjecture by how
calculation have been omitted in order to facilitate a much urban expectation of life may exceed the rural,
fairly simple exposition. and it is at least conceivable that the difference may now
be of the order of five years.
A. DERIVATION O F BASIC DATA (6) For the purpose of the projection, it was assumed
AND ESTIMATES that both the rural and urban expectations of life would
Sex-age composition rise according to the United Nations sequence of model
life tables. 98
336. In chapter VIII, the Brazilian census data bn (c) The life expectancy of the total population was
sex-age composition of the urban and rural population assumed to be the approximate average of the urban and
were accepted without adjustment. For the most part, rural rates.
the age statements are tolerably, but not highly, accurate.
In one respect, however, it is necessary to introduce a Expectation of life at birth (years)
modification for the present purpose. As in the cen-
Total Urban Rural
suses of many countries, the enumeration of small children, Date pooulation a population oopulation
aged 0 to 4 years, has been rather incomplete in the
Brazilian censuses. To avoid inconsistency with levels
of fertility estimated for past periods and to be assumed
in the projection and in the consequent trends of the child
population, it was found necessary in this chapter to
modify the numbers of children aged 0 to 4 years on the
approximate assumption that only 95 per cent of such
children were enumerated in urban areas, and only 90 per
cent in rural areas. 96
Mortality
337. In order to project urban and rural mortality, it a Estimates according to Wprld Population Prospects as Assessed in
1968 (United Nations publication, Sales No. 72.XIII.4).
is necessary as a minimum to have a recent estimation of
the level of mortality in the total population as expressed
in the expectation of life at birth. Methods have been Rural-to-urban population transfers, by sex and age
described in a previous manual permitting the estimation 338. In the calculation of rural-to-urban population
of mortality level, given only some data on age compo-
sition and the rate of growth of the total population. 97 transfers, which follows, the distinction between urban
and rural mortality will be dropped for the sake of
These methods are now applied to available data for simplicity. True, a more refined calculation might be
Brazil for several past dates, and the estimated expecta-
tions of life in the total population will be presently carried out in which urban and rural mortality levels are
distinguished, but the presentation of such a detailed
shown. The methods described in the previous manual
provide only approximate estimations. However, where procedure would require much space. Furthermore,
since age statements are not highly accurate, errors of
mortality is rather low, even an inaccurate estimate of the calculation due to inaccurate age statement are likely
ga How this rough estimate was arrived at is explained further on,
to be larger than those due to the lack of distinction
in paragraph 371, in connexion with estimation of fertility trends. between urban and rural mortality. Finally, where
The correction introduced here consists in multiplying numbers of mortality is fairly low, errors in calculated numbers of
urban children aged 0 to 4 by 100195 or 1.0526, and of rural the population due to an inaccurate estimation of the
children by 100/90 or 1.111 1. This correction has also been taken mortality level, at the most, are quite slight (except perhaps
into account in the ensuing calculations of apparent survival ratios
and rural-to-urban population transfers. in earliest childhood and at advanced ages).
87 Manual IV: Methods of Estimating Basic Demographic Meas-
ures from Incomplete Data (United Nations publication, Sales Five-year survival ratios in the successive model tables are
No. 67.XIII.2). reproduced in annex 11.
 TABLE
37. APPARENTDECENNIAL SURVIVAL RAnOS IN TOTAL POPULATION AND HYPOTHETICAL
SURVIVAL OF RURAL POPULATION,
1960-1970, AGES 10 AND OVER (AGES AS OF 1970)

Rural population
Census data on (thousands)
total population Apparent
(thousands) intercensals Hypothetical
survival ratio, As of 1960 survivors
Sex, and aqe os of 1970 1960 1970 1960-1970 (census) in l9W
----- -.--------.
-- -. --
Males
10-14. ................... 6,208 b 5,850 0.9423 3,713 3,499
15-19 .................... 5,159 4,934 0.9564 3,051 2,918
20-24.. .................. 4,287 4,063 0.9477 2,503 2,372
25-29.. .................. 3,446 3,202 0.9292 1,999 1,857
30-34.. .................. 2,964 2,857 0.9639 1,610 1,552
35-39.. .................. 2,522 2,460 0.9754 1,308 1,276
40-44.. .................. 2,257 2,209 0.9787 1,136 1,112
45-39.. .................. 1,953 1,897 0.9621 971 934
50-54.. .................. 1,667 1,510 0.9058 866 784
55-59.. .................. 1,385 1,193 0.8614 728 627
60-64.. .................. 1,096 894 0.8157 563 459
65-69 .................... 855 656 0.7673 436 335
70 and over .............. 1,707 C 874 0.5120 876 449
Females
10-14.. .................. 5,958 5,816 0.9762 3,506 b 3,423
15-19.. .................. 5,002 5,270 1.0536 2,884 3,039
26-24.. .................. 4,287 4,359 1.0168 2,380 2,420
25-29.. .................. 3,697 3,344 0.9045 1,973 1,785
30-34.. .................. 3,197 2,963 0.9268 1,601 1,484
35-39.. .................. 2,687 2,502 0.9311 1,292 1,203
40-44 .................... 2,291 2,179 0.9511 1,061 1,009
45-49 .................... 1,919 1,828 0.9526 872 831
50-54.. .................. 1,602 1,482 0.9251 759 702
55-59.. .................. 1,305 1,170 0.8966 621 557
60-64.. .................. 1,033 869 0.8412 476 400
65-69 .................... 802 649 0.8092 359 29 1
70andover .............. 1,717C 993 0.5783 741 429

a Obtained by multiplying each sex-age group (according to the 1960 census) with the corresponding
apparent survival ratio obtained from the intercensal comparison of the total population.
G e e . aged 0 to 4 in 1960. Figure corrected for assumed incomplete census enumeration of small
children.
i.e. sum of those aged 60 to 64, 65 to 69 and 70 and over in 1960.
A "survival ratio" greater than unity is in theory impossible, but age misstatements can produce an
"apparent" ratio greater than unity (see text). For the same reason, calculated "survival ratios" at ages
25 to 29 and 30 to 34 are probably too low.

339. Two assumptions are made in the following calcu- balance having been small as compared with the size
lation of rural-to-urban population transfers : namely, of Brazil's population.
that the effects of age misstatement are the same in the 340. The first few steps of the calculation are illustrated
rural and urban populations; and that for practical in table 37. In the first column, data of the 1960 census
purposes mortality conditions, and hence the survival concerning the total population are arranged according
ratios, were also the same in the rural and urban popu- to ages 10 years and higher, i.e. ages which those popu-
lations. Both assumptions are admittedly inaccurate, lation groups, if surviving, were to attain by 1970. As
but since the calculations can at best only be approximate. can be expected, with two exceptions, smaller numbers
the resulting errors are probably not of much i m p o r t a n ~ e . ~were
~ actually enumerated in 1970 because at least some
Furthermore, a smoothing procedure will be applied at persons in each age group had died in the time interval.
the end of the calculation which, it can be presumed, The exceptions are females aged 15 to 19 and 20 to 24 in
eliminates at least some of the errors resulting in the 1970 (or 5 to 9 and 10 to 14 in 1960), either because too
calculation. Some international migration also occurred few were enumerated in 1960, around the ages of 10 years,
during recent decades, which is disregarded here, the or too many in 1970 around the ages of 20 years; the
latter is more likely to have been the case, as there is
Age reporting is probably more accurate in the urban than in reason to believe that some young prefer to
the rural population. Mortality is probably higher in the rural than
in the urban population. The effects of resulting errors in the calcu- report an age between and 25 years rather than an age
lation are complex and cannot be easily assessed. below 15 or above 25, with the consequence that women
 TABLE
38. HYPOTHETICAL
SURVIVORS OF RURAL POPULATION IN 1970, RURAL P O P U L A ~ O NENUMER-
ATED IN 1970 AND APPARENT NET RURAL-TO-URBAN POPULATION TRANSFERS OF 1960-1970 (NUM-
BERS AND PERCENTAGES OF CALCULATED RURAL SURVNORS)

Apparent net transfers, 1960-1970

Rural population, 1970
(thousands) Transfer rate Rate of
- Numbers (i.e. per cent non-transfer
Hypothetical Actually (in thou- of hypot.ketica1 (i.e. 100 minus
Sex, and age as of 1970 survivors enumerated sands) rural survivors) transfer rate)

Males
10-14.. .................. 3,499 2,720 779 22.3 77.7
15-19. ................... 2,918 2,266 652 22.3 77.7
20-24.. .................. 2,372 1,790 582 24.5 75.5
25-29.. .................. 1,857 1,407 450 24.2 75.8
30-34.. .................. 1,552 1 ,207 345 22.2 77.8
35-39.. .................. 1,276 990 286 22.4 77.6
40-44.. .................. 1,112 912 200 18.0 82.0
4 5 4 9 .................... 934 791 143 15.3 84.7
50-54 .................... 784 652 132 16.8 83.2
55-59 .................... 627 525 102 16.3 83.7
60-64.................... 459 394 65 14.2 85.8
65-69 .................... 335 284 51 15.2 84.8
70 and over .............. 449 378 71 15.8 84.2
Females
10-14.. .................. 3,423 2,589 834 24.4 75.6
15-19.. .................. 3,039 2,189 850 28.0 72.0
20-24.. .................. 2,420 1,773 647 26.7 73.3
25-29.. .................. 1,785 1,334 451 25.3 74.7
30-34.. .................. 1,484 1,137 347 23.4 76.6
35-39.. .................. 1,203 938 265 22.0 78.0
40-44.. .................. 1,009 819 190 18.8 81.2
4 5 4 9 . . .................. 831 685 146 17.6 82.4
50-54 .................... 702 559 143 20.4 79.6
55-59 .................... 557 438 119 21.4 78.6
60-64.................... 400 318 82 20.5 79.5
65-69.. .................. 291 232 59 20.3 79.7
70 and over .............. 429 353 76 17.7 82.3

were over-reported within those particular age groups result in hypothetical numbers of survivors of that rural
and under-reported in adjacent groups (10 to 14, or population by 1970, as is done in the last two columns of
25 to 29). No exact determination can be made, but if table 37.
it can be assumed that age misstatements in urban and 342. The last column of table 37 is re-entered as the
rural areas had a similar tendency, part of the error of first column of table 37, where the calculation continues.
calculation is cancelled out in the subsequent procedures Of course, fewer individuals than those which survived
(see further on). are to be expected as still residing in rural areas by 1970,
341. Continuing with the calculation illustrated in owing to the intervening rural-to-urban transfers of
table 37, we obtain in the third column the "apparent" population, whether as a consequence of migration or
intercensal survival ratios as a result of dividing the area reclassification. In fact, as can be seen from a
numbers in each sex-age group obtained at the 1970 comparison of the first two columns in table 38, fewer
census by the numbers aged ten years less at the 1960 rural inhabitants were enumerated at all ages in rural
census (who, if surviving, would by 1970 have attained areas at the 1970 census, and the difference between
the same ages). If international migration had been hypothetical rural survivors and actually enumerated
negligible and age reporting at each census had been rural population can be attributed to population trans-
accurate, the sequence of survival ratios would permit a fers, provided the net errors of calculation are not too
detailed calculation of mortality conditions and the con- severe.
struction of a life table. Actually, because of inaccuracies 343. The net transfers which appear in the third
in the data, this is not quite the case. The "apparent" column of table 38 do not include persons who may have
survival ratios are partly false, but it will be assumed been involved in transfers during 1960-1970 but have
that the distortions are proportionately the same in the died before 1970. More accurately speaking, those are
total as well as in the rural population. On this rough net transfers during the period of those persons only who
assumption, we can now apply those "apparent" survival also survived. This manner of calculation is necessary
ratios to the rural population enumerated in 1960 to for the eventual purposes of the projection where also
 TABLE
39. INTERPOLATION
OF DECENNIAL RURAL-TO-URBAN TRANSFER RATES,
1960-1970,
FOR FIVE-YEAR PERIODS

Decennial rate of
Sex, and age non-transfer Quinquennial rate
at the end of (per 100 of rural Log. of same Interpolation Same, -
each period population) (add. 2 ) of log. smoothed " Non-transfer Transfer

Males
5-9 . . . . . . . . . . . . . ... ... .9452 .9452 88.1 11.9
10-14 ............. 77.7 1 ,8904 .9452 .9452 88.1 11.9
15-19 ............. 77.7 1.8904 .9452 .9421 87.5 12.5
20-24 ............. 75.5 1.8779 ,9327 .9394 87.0 13.0
25-29 ............. 75.8 1 .8797 .9470 .9427 87.6 12.4
30-34 ............. 77.8 1.8910 .9440 ,9452 88.1 11.9
35-39 ............. 77.6 1.8899 .9459 .9509 89.3 10.7
40-44 ............. 82.0 1.9138 .9679 .9604 91.3 8.7
4549 ............. 84.7 1 .9279 .9600 .9620 91.6 8.4
50-54 ............. 83.2 1 .9201 .9601 ,9607 91.4 8.6
55-59 . . . . . . . . . . . . . 83.7 1 .9227 .9626 .9640 92.0 8.0
60-64 ............. 85.8 1.9335 .9709 .9655 92.4 7.6
65-69 ............. 84.8 1 .9284 ,9575 .9634 91.9 8.1
70 and over ........ 84.2 1 .9253 .9678 .9634 91.9 8.1
Females
5-9 .............. ... ... .9392 .9392 86.9 13.1
10-14 ............. 75.6 1.8785 ,9393 .9340 85.9 14.1
15-19 ............. 72.0 1 .8573 .9180 ,9306 85.2 14.8
20-24 . . . . . . . . . . . . . 73.3 1.8651 .9471 .9346 86.0 14.0
25-29 . . . . . . . . . . . . . 74.7 1.8733 .9262 ,9394 87.0 13.0
30-34 ............. 76.6 1.8842 .9580 .9441 87.9 12.1
35-39 . . . . . . . . . . . . . 78.0 1.8921 .9341 .9504 89.2 10.8
40-44 . . . . . . . . . . . . . 81.2 1 .9096 .9755 .9564 90.4 9.6
4549 ............. 82.4 1.9159 .9405 9542 90.0 10.0
50-54 . . . . . . . . . . . . . 79.6 1 .9009 .9604 .9491 88.9 11.1
55-59 ............. 78.6 1 .8954 .9350 .9490 88.9 11.1
60-64 ............. 79.5 1.9004 .9654 .9505 89.2 10.8
65-69 ............. 79.7 1.9015 .9361 .9467 88.4 11.6
70 and over ........ 82.3 1.9154 .9793 .9467 88.4 11.6

a Smoothed by formula + +
2a, a+l), where a, is the unsrnoothed number in the given age group, and a - , and a+, are the unsmoothed
numbers in the preceding and subsequent age groups.
* Assumed same as unsmoothed number.
Assumed same as smoothed number at ages 65 to 69.

first the effects of mortality are calculated, and then the are converted into logarithms to which the amount of
effects of transfer (migration and reclassification) among- 2 is added for convenience of calculation.
those who survive (see section B of this chapter). 346. It will be recalled that the decennial survival
344. The transfer rates shown in the fourth column ratios were for persons passing through two successive
of table 38 are calculated as percentages relative to calcu- five-year age groups. Thus, persons aged 0 to 4 years
lated numbers of hypothetical survivors; those are the in 1960 were affected by certain quinquennial rates (still
proportionate deductions which will have to be made in to be determined) to bring them to ages 5 to 9 years in
a projection of the rural population according to survival 1965, and then by certain other quinquennial rates to
rates. When these rates are subtracted from 100 per bring them to ages 10 to 14 years in 1970. Thus, the
cent, complementary rates, or rates of non-transfer, are decennial non-transfer rate for those aged initially 0 to 4
obtained, which appear in the last column of table 38. and finally 10 to 14 are the product of a quinquennial
345. These rates, it will be recalled, have been obtained rate for persons passing from ages 0 to 4 to 5 to 9, and
for the decennial time interval between the censuses of another quinquennial rate for persons passing from ages
1960 and 1970. Projections, on the other hand, are 5 to 9 to 10 to 14; and similarly for all other age groups.
usually calculated by five-year time periods. The In addition, it is to be noted that the next decennial rate,
operations necessary to transform rates for ten-year bringing individuals from ages 5 to 9 to eventual ages
periods into rates for five-year periods are illustrated in 15 to 19, is the product of quinquennial rates for persons
table 39. Logarithms are used here for convenience, passing from 5 to 9 to 10 to 14 and rates for persons
although this is not strictly necessary. The calculation passing from 10 to 14 to 15 to 19. Consequently,
begins with the decennial rates of non-transfer, as obtained among adjacent age cohorts, decennial rates each time
in table 38. In the second column of table 39, these rates contain one common multiplier, in this instance the
 Quinquennial transfer Quinquennial transfer
rates calculated rates cnlculated
Age at end from 1950-1960 data from 1960-1970 data
of
each period Males Females Males Females

5-9 ..........................
10-14. .........................
15-19. .........................
20-24. ............... . . . . . . . .
.
25-29 ..........................
30-34. .........................
35-39. .........................
40-44 ..........................
45-49. . . . . . . . . . . . . . . . . . . . . . . . . .
50-54 ..........................
55-59. . . . . . . . . . . . . . . . . . . . . . . . . .
60-64 . . . . . . . . . . . . . . . . . . . . . . . . . .
65-69. .........................
70 and over ....................

Comparison of quinquennial transfer rates by sex and age as calculated from data
for 1950-1960 and 1960-1970 (ages a t the end of a five-year period)
' 1960-1970

Females
'-\ Females

Males
Males

quinquennial rate for persons passing from 5 to 9 to third column, and so on. Thus, one-half of 1.8904
10 to 14.1m equals 0.9452 for males aged 5 to 9. Successive sub-
347. Since in the logarithmic scale the successive tractions then yield a sequence of logarithms for quin-
multiplications are represented by successive additions, quennial rates, admittedly seesawing to some extent
the interpolation can be performed very easily, as will partly as a result of errors of calculation due to inaccurate
be seen in the comparison between the second and third age statements. lol The zigzags are eliminated after the
columns of table 39. Somewhat arbitrarily, the first smoothing procedure which is carried out in the next
quinquennial rate (for persons passing from 0 to 4 to column according to a simple formula (see table foot-
5 to 9) is taken as the square root (i.e. one-half of the note a). The resulting logarithms are then converted
logarithm) of the corresponding decennial rate. The into the corresponding quinquennial non-transfer rates,
first quinquennial rate in the third column is then sub- and these, after subtraction from 100 per cent, yield
tracted from the 10 to 14 age group figure in the second the quinquennial transfer rates appearing in the last
column to obtain the second quinquennial rate in the column of table 39.
looAmong the adjacent age groups, however, survival from 5 to 9
to 10 to 14 occurs in different time periods. To simplify, it is here lol If the rates themselves had been used in the calculation instead
assumed that mortality in both periods was the same. The error of their logarithms, successive divisions would have had to be made
of calculation due to this simplification is trivial. instead of successive subtractions.
 348. The foregoing calculation is for the intercensal 351. The Brazilian census data provide no direct clue
period of 1960-1970. To gain some judgement regarding for the distinction of transfers caused by reclassification
the possible validity of the results, the same calculation from those caused by migration. Rough estimates can
has also been carried out for the 1950-1960 period. nevertheless be attempted if there is sufficient ground for
The details of calculation are not shown here to save space, two assumptions, both of them somewhat plausible.
but the procedure has been the same. A comparison of First, since much of the net migration is caused by the
the results of calculations for both periods appears in search for employment, affecting especially adolescents
table 40 and in figure. 111. The following observations and young adults, it can be assumed that past a certain
can be made. age, when most persons are rather comniitted to
349. In both periods, female transfer rates at all ages their existing livelihood and residence, net migration
exceeded the male transfer rates, female rates reaching a becomes slight if not negligible; since it was observed
peak in the group passing from ages 10 to 14 to ages that the net transfer rates reach a minimum about
15 to 19, whereas the male peak occurs in the group the age of 40, one may assume, very roughly, that
passing from ages 15 to 19 to ages 20 to 24. Both males rural-to-urban transfers past the age of 40 years
and females reach minimal transfer rates about the age are mostly the consequence of area reclassification.
of 40 years, but the rates of females rise again with Secondly, as regards the reclassified areas themselves, it
advancing age. loz For both sexes and all age groups the may perhaps be reasonable to assume, in the case of
rates were higher in 1960-1970 than in 1950-1960. Some Brazil, that the sex-age composition of their population
change of detailed pattern also seems to have occurred. resembles that of the population which has already been
For instance, in the more recent period the transfer urban during the same time. lo4 This is a very hazardous
rates for males aged between 30 and 40 years almost assumption whose possible validity must first be tested.
attained the levels for females at the same ages. Never- The following analysis does not prove thecorrectness of such
theless, in view of the general consistency of the results an assumption, but it shows that it may be at leasttenable.
obtained for both periods, the calculated transfer rates 352. As shown in table 40 and figure 111, when calcu-
appear to be adequate for use in a projection. lated relatively to the rural population, transfer rates of
ageing females rise until they exceed considerably those
The separate efects of rural-to-urban area of ageing males. The matter becomes quite different,
reclasszjication and migration however, when the transfer rates are calculated relatively
to the urban population, in which case they might be
350. The foregoing calculations comprise net transfers termed "rates of accretion", in view of the fact that the
of population from rural to urban areas which can occur transferred population is added to that of urban areas.
for two distinct reasons: the reclassification as urban of 353. The calculation of these urban accretion rates,
areas previously designated as rural; and the physical based on the quinquennial rural transfer rates arrived
migration from areas designated at the time as rural to at in table 40, is shown in the first three columns of
areas designated at the time as urban. Some inter- table 41. It is assumed that quinquennial transfer rates
action between the two phenomena exists, since presum- on the basis of 1960-1970 data are also applicable to
ably the reclassified areas also have experienced much the 1965-1970 period. Multiplying these rates with the
recent immigration. For the purpose of the projections rural population of 1970 in each age group, we obtain
it is important to consider these two processes separately, estimated amounts of transfer during that quinquennium.
for only then can variations be introduced in the assump- Dividing these amounts by the urban population of 1970
tions with respect to reclassification, migration or both. in each age group, we obtain the corresponding urban
The contribution made by net migration as such to urban accretion rates. To save space, the detailed steps of the
and rural population change is also a subject of separate calculation for 1955-1960 are not presented, but the result-
interest. lW ing urban accretion rates of 1955-1960 are shown in the
last column for comparison with those of 1965-1970.
lo2 In this comexion, it should be noted that the calculation
implies equal mortality levels in urban and rural areas. It is pos- The results for both periods are also shown in figure IV
sible, however, that urban females of mature and advanced ages had on page 90.
significantly lower mortality than the corresponding rural females, 354. Though rural transfer rates rose from 1950-1960
and that this circumstance may have affected the results of the calcu- to 1960-1970 (figure 111), urban accretion rates fell
lation. However, as shown further on, another explanation can be
found for this observation, which may be of ,greater significance. because of the faster growth of urban population as
loS Only the balance, or net migration, is considered here, which compared with the rural. The pattern of urban accretion
in turn results from the difference between in-movements and out- rates remained similar in these respects: female accretion
movements. Very detailed statistics, found in rather few countries, rates exceeded the male only at early ages, whereas
would be needed to assess the composition and trend of movements between the ages of 20 and 40 years male rates exceeded
in either direction separately. The question may be of some impor-
tance in countries where suburban population continues to be the female (slightly in 1955-1960, and more considerably
classified as "rural", as there may be at certain ages a net movement
from cities to the suburbs; and also in African and Asian countries, lo4 Depending on the type of flexibility of the definition of urban
where a good deal of urbanization is caused by the influx of short- areas, quite different assumptions may have to be made in other
term residents who, after some period of time, generally return again countries. As discussed in chapter 11, sex-age compositions of
to their rural places of origin. In the case of Brazil, as already localities in Japan vary systematically with the size of the localities;
noted, the boundaries of "urban" areas are flexible and can include and in the United States, the structures of suburban populations
expanding zones having, as defined, a "suburban" character. Hence differ greatly from the populations of cities within their adminis-
the large importance of area reclassification in the case of Brazil. trative boundaries.
 41 . CALCULATION
TABLE OF QUINQUENNIAL URBAN ACCRETION RATES FOR 1965.1970.
AND COMPARISON WITH SIMILARLY CALCULATED RATES FOR 1955-1960

Estimated
Rural net Rural amount Urban Urban accretion rate
Sex and age transfer popula~ion of transfer. population (percentage)
at end of each rate. 1970 1965.1970 1970
period 1960-1970 a (thousands) (thousands1 (thousands) 1965-1970 1955-1960 o

Males
5-9 ................ 11.9 3. 241 386 3. 489 11.1 13.5
10-14 ................ 11.9 2. 720 324 3. 130 10.4 13.3
15-19 ................ 12.5 2, 266 283 2. 668 10.6 14.5
20-24 ................ 13.0 1,790 233 2, 273 10.3 13.2
25-29 ................ 12.4 1 ,407 174 1. 759 9.9 11.2
30-34 ................
35-39 ................
11.9
10.7
.
1 207
990
144
106
1. 650
1. 470
8.7
7.2
10.3
8.9
40-44 ................ 8.7 912 79 1. 297 6.1 8.1
4 5 4 9 ................ 8.4 791 66 1. 098 6.0 8.4
50-54 ................ 8.6 652 56 858 6.5 8.3
55-59 ................ 8.0 525 42 668 6.3 7.9
60-64 ................ 7.6 394 30 500 6.0 8.0
65-69 ................ 8.1 284 23 372 6.2 8.6
70 and over .......... 8.1 378 31 496 6.3 8.4
Total. 40 and over ...... 327 5. 289 6.2 8.2
Females
5-9 ................ 13.1
10-14 ................ 14.1
15-19 ................ 14.8
20-23 ................ 14.0
25-29 ................ 13.0
30-34 ................ 12.1
35-39 ................ 10.8
40-44 ................ 9.6
4 5 4 9 ................ 10.0
50-54 ................ 11.1
55-59 ................ 11.1
60-64 ................ 10.8
65-69 ................ 11.6
70 and over .......... 11.6

Total. 40 and over .......

Both sexes

Sum.40 and over .....

a Taken from table 40. per 100 of the rural population .
CaIculated by the same procedure as the rates for 1965.1970 .

in 1965.1970) . Past the age of 40 years. the rates for be used to distinguish the two types of population
either sex fluctuated about the same values. without any transfer. that due to reclassification and that due to
trend. namely about 8.1 per cent in 1955.1960. and about migration .
6.2 per cent in 1965.1970 . The fluctuations past the 355. As assumed. past the age of 40 all transfers are
age of 40 years may be due as much. if not more. to the now ascribed to reclassification. hence net migration
effects of inaccurate age statements upon the calculation rates of significance can be calculated only up to the age
(with advancing age the accuracy of age reporting dete- of 40 years. The assumption now takes the form that
riorates). as they may have been due to any significant urban accretion rates due to reclassification are the same
difference between the sex-age composition of the urban at all ages. below 40 and above 40 years; namely. as
population and that of the population which became calculated in table 41. 6.2 per cent in the 1965-1970
reclassified as urban . The two assumptions which have period. and 8.1 per cent in the 1955-1960 period . Since
been made-namely that past the age of 40 years migra- the population at risk of net migration is the rural one.
tory balances may be negligible. and that the reclassified migration rates need to be calculated relative to the rural
population may resemble that which is already urban- population . The calculation pertains only to ages below
though not proven. are significantly strengthened . 40 years and is shown in table 42 . First. from the size
While admittedly debatable. these assumptions will now of sex-age groups in the urban population and the
 Comparison of quinquennial urban accretion rates by sex and age as calculated for the periods 1950-1960 and 1960-1970
(ages at the end of each five-year period) and assumed applicable in 1955-1960 and 1965-1970

16 -, Females 1955-1960

Females

6.2 level

TABLE
42. CALCULATION OF QUINQUENNIAL NET RURAL-TO-URBAN MIGRATION RATES, 1965-1970,
AND COMPARISON WITH SIMILARLY CALCULATED RATES FOR 1955-1960

Amount of Amount of Amount of Net migration rate

Urban reclassified translerred net Rural (per 100 of the rural population)
Sex, and age population, population, population, migration, population,
at end of each period 1970 1965-1970 1965-1970 1965-1970 1970 1965-1970 1955-1960 r

Males
5-9 ..................
10-14 ..................
15-19 ..................
20-24 ..................
25-29 ..................
30-34 ..................
35-39 ..................
Females
5-9 ..................
10-14 ..................
15-19 ..................
20-24 ..................
25-29 ..................
30-34 ..................
35-39 ..................
* Assuming an urban accretion of 6.2 per cent due to reclassification (for 1955-1960, the urban accretion due to reclassification is estimated as
8.1 per cent).
Taken from table 40.
Calculated by the same procedure as the rates for the 1965-1970 period.

reclassification accretion rate ascertained for ages above 356. First it will be noted that the reclassification rates
40, the amount of population reclassified in a five-year are surprisingly high and the migration rates surprisingly
period is calculated for ages below 40 years. When this low in comparison with the combined transfer rates.
amount, in each sex-age group, is subtracted from esti- The explanation of this circumstance consists in the
mated amounts of transfer, the estimated amount of net likelihood that much migration occurs in the direction
migration is obtained. And when this is related to the of areas which are about to be or have recently been
size of the rural population, in each sex-age group, net reclassified in the period of observation. This must be so,
migration rates are ascertained. The calculation is not only owing to the present assumption that the sex-age
illustrated for the 1965-1970 period, and for the sake of structure in reclassified areas is apt to resemble that in
comparison results of the same procedure for the 1955- areas already urban; but also because much migration
1960 period are presented in the last column. The occurs to precisely those areas-surroundings of growing
calculated migration rates are plotted also in figure V. big cities and new emerging towns-which have the
 Comparison of quinquennial rural-to-urban migration rates by sex and age as assumed applicable
in the periods 1955-1960 and 1965-1970 (ages at the end of each five-year period)

'1 Females

6- Females
/'\ en
0

2-

1-
0 - I I I

greatest likelihood of becoming reclassified. In fact, to an extent difficult to ascertain). lo5 The same order of
there may even be a significant outward migration from calculations-mortality first, migration and reclassi-
established urban centres to the areas in process of reclas- fication second, and fertility last-will also be maintained
sification. If, in a special projection, an arbitrary assump- in the population projections.
tion is to be made that no additional areas will be reclassi- 359. Any calculation of fertility generally relates the
fied in the future, a large part of the urbanization process number of children born in a period (boys and girls)
will be missed owing to considerable migration to new to numbers of women of reproductive ages during the
areas of an urban type which, on such an assumption, period of time when those births occurred. The cal-
would remain classified as rural. culation is usually detailed by age-specific fertility rates
357. Of course, the foregoing calculation can contain of women in each of the age groups from 15 to 19 to
large errors. Whereas at advanced ages the population 45 to 49. One prerequisite is to estimate the mean
structure of reclassified areas may indeed resemble that number of women in each age group during the period
of areas already urban, it is far from certain whether this under consideration, defined here as the arithmetic
is also the case at early ages. Fertility rates, for instance, average between numbers of such women at the beginning
may differ between established urban areas and areas in and numbers at the end of the period. lo6 The numbers
process of reclassification, and conceivably they may be at the end of each period, of course, are already reduced
higher in the latter areas (which are not yet so congested, by intervening rates of infant and childhood mortality,
perhaps). In such a case, the reclassification may com- and augmented (in the urban population) or reduced
prise a larger number of children than have here been (in the rural population) by intervening rates of popu-
calculated, with the consequence that the presently lation transfer. The fertility estimate for the rural popu-
estimated migration rates for children are somewhat too lation then excludes, and that for the urban population
large. However, in the absence of pertinent detailed includes, that of women in process of transfer from rural
information no further tests can be made. It would be to urban areas. Separate calculation of the fertility
difficult to improve upon the present calculation with the of migrant or reclassified urban women is therefore
limited information at hand. unnecessary. One of the questions often mentioned
in the projection of urban population is an insufficiency
Fertility of knowledge concerning the fertility of migrants. In
our present calculations, however, the fertility of the
358. It will be noted that in the foregoing, mortality urban population includes at any time, past or future,
was considered first, and rural-to-urban transfers were the fertility of previouk rural-to-urban migrants so that
calculated thereafter in relation to age groups at the
end of each period, so that transferred persons who lo6 There is a further separate effect, also disregarded here, of

had died in the period were no longer included. The migrant mothers leaving some of their children under the care of
their rural relatives.
difficulty of estimates of deaths occurring to the trans- '06 It is a simplification to use the arithmetic mean, since it is
ferred population itself (whether reclassified or migrants) not believed that population growth in each period occurs at an
is thereby obviated. For similar reasons, the calculation arithmetic rate. The survival ratios from birth to ages 0 to 4 during
of fertility is postponed to the end to avoid the complexity five-year periods, taken from model life tables, likewise refer to an
annual arithmetic average of children born. It can therefore be
of separate calculation of the fertility o f migrants or considered that the "errors" in calculating from arithmetic means
fertility of the population in reclassified areas (each of (of number of women and of births) are mostly cancelled out in the
which may differ from both the urban and rural fertility, calculated fertility rates.
at each projection date the urban fertility likewise 363. Depending on the fertility age pattern, the same
includes that of recent in-migrants. number of births can result from different gross reproduc-
360. An important consideration to take into account, tion rates. Usually, women of earlier reproductive ages
however, pertains to the accuracy with which children are more numerous than those of later reproductive ages.
aged 0 to 4 years may have been enumerated. In the If the fertility age pattern gives much weight to the
censuses of most countries a more or less significant younger groups of reproductive women, therefore, a
proportion of small children tend to be omitted, and there lower GRR suffices to produce the same number of
is reason to suspect that the enumeration of small children births as would result if the age pattern gave more weight
was also deficient in the censuses of Brazil.loB This to older age groups of reproductive women, and if these
consideration makes it advisable to estimate fertility were somewhat less numerous. If it is important to
levels and trends over periods longer than the five years measure the reproductivityof women as such, then much
preceding the census. accuracy is also needed in the estimation of the fertility
age pattern.
361. The relationship between women's age-specific
fertility rates and the gross reproduction rate (GRR for 364. In a population projection, however, the levels
short) also needs to be elucidated. Given a set of age- and trends in numbers of births only are required,
specific fertility rates covering the reproductive age span whatever the detailed age pattern of fertility or the precise
of women, one can calculate the average number of level of GRR might be. An inaccurately estimated
births a woman may ultimately undergo as she passes fertility age pattern, then, does not introduce any appre-
through all these consecutive age groups. To ascertain ciable errors in the corresponding projection of the child
the rate at which female generations replace each other, population. Both the level of GRR in past periods and
the GRR is expressed in terms of the number of girls that projected to future periods may (with an inaccurate
born. loo In most countries with good statistics, about age pattern) be either over-estimated or underestimated,
105 boys are born per 100 girls born, hence the GRR is but the error of estimation is largely cancelled out in
100 terms of the projected numbers of births.
the sum of age-specific fertility rates multiplied by - 365. The age pattern of fertility may differ significantly
205
or, approximately, 0.4878. For the purposes of a between urban and rural areas, and each of the two
projection, it is a frequent practice to calculate the past patterns may change with time in different ways. If
and conjecture the future trend of fertility in terms of there are sufficientstatistical data, it is desirable to estimate
GRR. The detailed calculation can therefore be con- both fertility age patterns with some degree of accuracy,
fined to the ascertainment of the number of girls born; perhaps also to make separate assumptions of future
the number of boys born may then be obtained by changes in both the urban and rural patterns. If he has
multiplying the combined number of girls by 1.05. 11° access to the requisite statistics, the user of this manual
362. The projection in terms of the GRR is simplified may refine the calculations as he sees fit. But since the
if, instead of explicit age-specific fertility rates, a set of present example is largely confined to an illustration of
comparative rates only are considered, i.e. the so-called basic methods, while statistical data for Brazil are not so
refined, and simplicity is preferable, only one uniform
"age pattern of fertility", or the percentage contributions
fertility age pattern will be assumed for both urban and
made by each age-specific rate to their sum, which is the rural areas, and it will be kept constant also in future
GRR (after multiplication by 0.4878). By applying
these percentages to numbers of women in the corre- periods. This age pattern was arrived at partly through
analogy with other countries where conditions might
sponding age groups, a "weighted sum" of women of have been similar, in a population projection for Brazil
reproductive age is obtained; this "weighted sum" of recently calculated by the United Nations.111 It is
women can then be multiplied directly by the GRR to represented by the figures below, but no claim is made
calculate the corresponding number of girls born, and that these figures reflect the actual Brazilian fertility
then by 1.05 to obtain corresponding numbers of boys pattern with any degree of accuracy.
born. One assumption of the vroiection. then. consists
of the "age pattern of fertility3- conveniently expressed Age of women Percentage contribution
l o GRR
in percentage terms. (years)
15-19 .............................. 7
'"If migration should intensify or diminish considerably and the 20-24 .............................. 23
fertility of migrants differs noticeably from that of non-migrants, 25-29 .............................. 25
the average level of urban fertility (migrants and non-migrants) 30-34. ............................. 21
may thereby be affected. 35-39 .............................. 14
AS already mentioned, for the purpose of the projection the 40-44 .............................. 8
numbers of children aged 0 to 4 years had to be modified, taking 45-49 .............................. 2
into account an estimate of the probable degree of completeness of -
their enumeration. 100
loo The net reproduction rate, or NRR, equals the GRR multiplied
by the probability of survival to the mean age of reproductivity. 366. To estimate fertility in past periods in the absence
It indicates the rate at which successive generations of women tend
to replace each other. of suitable birth registration statistics, the procedure of
110 Slightly different ratios between numbers of boys and girls population projection must now be reversed. Numbers
born have also been observed, for instance in the Caribbean region,
but there is no evidence that the ratio in Brazil may differ from 'I1 World Population Prospects as Assessed in 1968 (United Nations
1.05 to any significant extent. publication, Sales No. 72.XIII.4).
 TABLE
43~.REVERSE-SURVIVAL
OF NUMBERS OF CHILDREN AND REPRODUCTlVE WOMEN FROM BRAZILIAN
CENSUS DATA FOR THE URBAN POPULATION IN 1970, AND ESTIMATION OF URBAN GROSS REPRO-
DUCTION RATE IN 1965-1970 AND 1960-1965

Ferli1it.v 1970 urban Survival 1965 urban Survivdl 1960 urban

age population ratio popuhtion ratio population
Sex and age pattern (census) foe. = 63.2) (estimate) (Oer = 60.4) (estimate)

Boys
(to be born) ..................
0-4..........................
5-9 ..........................
Girls
(to be born) . . . . . . . . . . . . . . . . . .
0-4 ..........................
5-9 ..........................
Women
15-19 ................ 0.07
20-24.. .............. 0.23
25-29.. .............. 0.25
30-34.. .............. 0.21
35-39.. .............. 0.14
40-44 ................ 0.08
45-49.. .............. 0.02
50-54. ...............
55-59. ...............
Weighted sum .......... 1.00

Births of period ..................

Mean weighted sum of women .....
Births per woman ................
Gross reproduction rate b ..........

a Uncorrected figure, as provided in the census.

Births per woman times 0.4878.

of children are projected backward to arrive at numbers ratio of survival from births occurring during a five-year
of births in past periods; numbers of reproductive women period to the eventual age group 0 to 4 at the end of those
are projected backward to those ages when they might five years. Thus, in the same table, the 3,449,000 urban
have given birth; and a relationship is established between boys aged 0 to 4 in 1970, by virtue of a survival ratio of
the two sets of numbers so as to estimate the level of 0.9262, may have resulted from a generation of
GRR. Use is made of model life table survival ratios 3,724,000 births which occurred in the period from 1965
corresponding to the levels of mortality already indicated to 1970. By the same procedure, estimated numbers of
near the beginning of this chapter. But whereas in a the population in 1965 are reverse-survived to yield
projection into the future, numbers in each age group are estimated numbers of births during 1960-1965 and num-
multiplied by corresponding survival ratios, in a projec- bers of women in 1960.
tion into the past, numbers of each age group are divided
by the survival ratios, as a result of which additional 368. In the next step, numbers of women in each of
individuals, who had died in the period are, so to speak, the age groups from 15 to 19 to 45 to 49 are multiplied
brought back to life. The procedure is often referred with the proportions contributed to their gross reproduc-
to as "reverse survival". The steps of the calculation tivity according to the assumed fertility age pattern,
are illustrated in tables 43a and 43b for the urban and yielding the weighted sums of 2,047,000 in 1970,
rural populations, respectively. 1,755,000 in 1965 and 1,540,000 in 1960 in the case of the
urban population.
367. In the first step, numbers of the 1970 census are
divided by survival &ios from model life tables as 369. In the final step, births of each period are related
tabulated in annex 11, but each time situated one line to the mean of weighted. sums of women of the same
above, to obtain numbers in the preceding age group who period. As calculated for the urban population, during
might have been alive in 1965. Thus, in table 43a, the 1965-1970, 3,724,000 boys and 3,612,000 girls were
3,489,000 urban boys aged 5 to 9 in 1970 when divided born, giving a sum of 7,336,000 births. These are related
by 0.9765, indicate that there might have been 3,573,000 to a weighted sum of 1,901,000 women, being the arith-
boys of the same generation aged 0 to 4 years in 1965; metic mean of 2,047,000 in 1970 and 1,755,000 in 1965.
and similarly with other age groups. The first survival Division shows that during the period there were 3.86
ratio is shown in parenthesis to indicate that this is the births per woman, or a gross reproduction rate of 1.88
 TABLE
4 3 ~ . REVERSE-SURVIVAL
OF NUMBERS OF CHILDREN AND REPRODUCTIVE WOMEN FROM BRA-
ZILIAN CENSUS DATA FOR THE RURAL POPULATION IN 1970, AND ESTIMATION OF RURAL GROSS
REPRODUCTION RATE IN 1965-1970 AND 1960-1965
Fertility 1970 rural Survival 1965 rural Survival 1960 rural
age population ratio popuhfion ratio population
Sex and age pattern (census) (Oea= 57.6) (estimate) foeo = 55.0) (estimate)

Boys
(to be born) .................. ... (0.8877) (4,023) (0.8703) (3,860)
04.......................... 3,571 a 0.9648 3,359
5-9.. ........................ 3,241
Girls
(to be born) .................. ... (0.9036) (3,863) (0.8882) (3,649)
0-4.......................... 3,491a 0.9669 3,241
5-9. ......................... 3,134
Women
15-19.. .............. 0.07
20-24.. . . . . . . . . . . . . . . 0.23
25-29 ................ 0.25
30-34.. . . . . . . . . . . . . . . 0.21
35-39.. .............. 0.14
40-44.. .............. 0.08
45-49 ................ 0.02
50-54. ...............
55-59. ...............
Weighted sum .......... 1.00 1,344 1,117 939
1965-1970 1960-1065
Birthsofperiod ................................ 7,886 7,509
Mean weighted sum of women ................... 1,230 1,028
Births per woman .............................. 6.41 7.30
Gross reproduction rate . . . . . . . . . . . . . . . . . . . . . . . . 3.13 3.56

Uncorrected figure, as provided in the census.

Births per woman times 0,4878.

(assuming that of all births a proportion of 0.4878 were much the low figures of 1945-1950, 1955-1960 and
those of girls). 1965-1970 have to be raised so that, as a result, a fairly
370. The same calculations were performed for the continuous time trend is obtained for both the urban and
rural population of 1970 (table 43b) as well as for the rural series of GRR. It was found that this is approxi-
urban and rural populations in earlier periods on the mately the case when the assumption is made that the
basis of earlier census data (not shown here to save censuses enumerated only 95 per cent of such children in
space). Thus, from census data of 1960 urban and rural urban areas and 90 per cent in rural areas. For when the
GRR's were calculated for 1955-1960 and 1950-1955; and corresponding figures are raised in such proportions, we
from census data of 1950, for 1945-1950 and 1940-1945. obtain the following, fairly continuous series. 112
GRR
371. As the provisional result of all these calculations,
Period Urban Rural
the following time series of urban and rural GRR were
obtained : 1940-1945 .................. 1.84 3.46
1945-1950 .................. 1.94 3.58
GRR
1950-1955 .................. 2.22 3.66
Period Urban Rural 1955-19 60 .................. 2.16 3.45
1960-1965 .................. 2.29 3.56
1965-1970. ................. 1.98 3.48
The number of children aged 0 to 4 years in each of the
censuses was accordingly modified, as already indicated
at the beginning of this chapter (see paragraph 336).
372. Considering the estimated trend of rising GRR
When these series are plotted on a graph, it is noted that during 1940-1955 and falling GRR in more recent years,
there are zigzags with comparatively low figures in 1945- it is somewhat probable that urban fertility may decline
1950, 1955-1960 and 1965-1970, i.e. in those periods for fairly consistently in coming years, and that a rural fertil-
which births had been reverse-survived from census data
for the age group 0 to 4 years. Since it is to be suspected These estimates, because of their uncertainty, are deliberately
rough. Actually, the smoothest trends in GRR result if it is assumed
that in each of the censuses the enumeration of this age that in both urban and rural areas about 88 per cent of children
group was incomplete, an estimate has to be made by how aged 0 to 4 years had been enumerated.
ity decrease, gradual at first but gathering speed with time, In the standard projection no distinction will be made
may also come under way. Assumptions of future between transfers due to migration and those due to
trends in GRR will be made accordingly. the reclassification of areas.
4. Seemingly plausible future trends in urban and rural
B. STANDARD PROJECTION gross reproduction rates have been conjectured,
Assumptions somewhat arbitrarily, for the purpose of the standard
projection. It is assumed that the GRR of urban
373. In the preceding part of the chapter, trends areas, near 2.0 in 1970, may decrease linearly to a
of mortality, rural-to-urban population transfer and level of 1.2 by the year 2000; and that the GRR of
fertility have been examined, leading to some plausible rural areas, near 3.5 in 1970, may decrease with gra-
basic estimates and assumptions. A detailed assessment dually increasing momentum to reach about 1.6 by
of the "most probable" future population prospects in the year 2000. For the intervening periods of the
Brazil may have to depend on various other evaluations projection, the levels of urban and rural GRR would
concerning the nation's economic and social development then be as follows:
and the popular propagation of fertility regulation Assumed GRR
practices, which perhaps only the national experts can Period Urban areas Rural areas
make. Without such pretension, some selected assump- 1970-1975. ................. 1.88 3.35
tions are here combined to serve as the basis fora standard 1975-1980 .................. 1.74 3.15
projection, i.e. a projection which, with the admittedly 1980-1985 .................. 1.62 2.90
limited information at hand, may appear to have some 1985-1990 .................. 1.50 2.60
degree of likelihood that actual future population trends 1990-1995. ................. 1.38 2.25
may in fact approximate it. The set of assumptions is 1995-2000. ................. 1.26 1.85
recapitulated below:
374. For simplicity and owing to the lack of more
1. The sex-age data of the census of 1960 were accepted detailed information, a uniform age pattern of fertility
as given, except for a modification in the 0 to 4 year age is maintained, indicated by the following percentage
group for which census enumeration had probably contributions by individual age groups to GRR:
been incomplete.
Age of women Contribution to GRR
2. For future periods five-year survival ratios will be
assumed according to the set of United Nations model 15-19. ............................. 7
20-24 .............................. 23
life tables in annex 11, selected according to the follow-
ing expectations of life at birth: 25-29 .............................. 25
30-34 .............................. 21
Model life table to be selected according
to expectation of life at birth (years) 35-39 .............................. 14
40-44 .............................. 8
Period Urban areas Rural areas
45-49 .............................. 2
1970-1975 .................. 65.8 60.4
1975-1 980. ................. 68.2 63.2 Sum .............................. 100
1980-1985 .................. 70.2 65.8
1985-1990.. ................ 71.7 68.2
1990-1995. ................. 73.0 70.2 Manner of calculation
1995-2000. ................. 73.9 71.7
375. To calculate the actual projection, three different
3. It will be assumed that the quinquennial rural-to-urban worksheets have to be used concurrently, the beginning
transfer rates, which have been calculated from data parts of which are shown in tables 44a, 44b and 44c. In
of the 1960-1970 period, will remain constant in all one worksheet, the rural population is calculated; in the
future periods of the projection. According to age other, the urban population, and in the third, the num-
at the end of each five-year period, the transfer rates bers of births. To facilitate ready recognition of different
(in per cent) are the following: columns of figures, the analyst carrying out the calcu-
Quinquennial tranqfer rate lation is advised to use pencils of different colours.
(per 100 of the rural population)
376. Beginning with the first column in table 44a
Age group Males Females
(rural population) we note that the numbers "to be born"
5-9 ...................... 11.9 13.1 are not yet determined and will be calculated in the third
10-14 ...................... 11.9 14.1 worksheet. The survival ratios of the next column are
15-19.. . . . . . . . . . . . . . . . . . . . . 12.5 14.8
those of the appropriate model life table, including the
20-24. ..................... 13.0 14.0
special survival ratios from births to ages 0 to 4, and the
25-29. ..................... 12.4 13.0 ratio assumed to apply for survival from ages 65 and over
30-34. ..................... 11.9 12.1
to ages 70 and over. Multiplication of figures of the
35-39. ..................... 10.7 10.8
40-44. ..................... 8.7 9.6 first two columns results in estimated survivors at the
4 5 4 9 . ..................... 8.4 10.0 end of the five-year period, shown in the third column and,
because of advancing age, placed one line beneath. Thus,
50-54. ..................... 8.6 11.1
55-59. ..................... 8 .o 11.1 the 3,968,000 males aged 0 to 4 years in 1970, if surviving
60-64. ..................... 7.6 10.8 at the rate of 0.9708, result in 3,852,000 males aged 5 to 9
65-69.. .................... 8.1 11.6 years in 1975; and so forth. The transfer rates of the
70 and over ...........;.... 8.1 11.6 next column are now multiplied with corresponding
 Rural Survival Transfer Trans- Rural
population ratio Survivors rate ferred populafion
Sex and age 1970 (Oro= 60.4) 1975 (per cent) populafion 1975

Both sexes...............
Males ...................
(to be born) ............
04 ..................
5-9 ..................
10-14 ..................
15-19 ..................
20-24 . . . . . . . . . . . . . . . . . .
25-29 ..................
30.34 ..................
35.39 . . . . . . . . . . . . . . . . . .
40-44 . . . . . . . . . . . . . . . . . .
4549..................
50-54 ..................
55.59 ..................
60-64 ..................
65.69 ..................
70 and over . . . . . . . . . . . .
Females .................
(to be born) . . . . . . . . . . . .
0-4 ..................
5-9 ..................
10-14 . . . . . . . . . . . . . . . . . .
15-1 9 ..................
20-24 ..................
25-29 ..................
30-34 . . . . . . . . . . . . . . . . . .
35-39 ..................
40-44 ..................
45.49 ..................
50-54 ..................
55-59 . . . . . . . . . . . . . . . . . .
60-64 . . . . . . . . . . . . . . . . . .
65-69 ..................
70 and over ............

aTo be born during 1970-1975; see table 44c .

aSurvival ratio from a five-year period of births to ages 0 to 4 .
To be born during 1975.1980 .
Determined after calculation of number of births.
Assumed survival ratio from ages 65 and over to ages 70 and over; actually. the survival ratio from
ages 70 to 74 to ages 75 to 79 according to model life table .

numbers of survivors in 1975 to result in estimated 377. Proceeding to table 44b (urban population). we
numbers of transferred population (whether owing to note that the mode of calculation is essentially the same.
to migration or reclassification) at the same date. Thus. except as regards the transferred population . This latter
if 3.852. 000 male survivors aged 5 to 9 are subject to category is simply copied from table 44a (rural popu-
transfer to urban areas at the rate of 11.9 per cent. then lation). where the transfers were calculated in relation
the rural population will lose 458.000 males aged 5 to to numbers of rural survivors. The same numbers are
9 years; and similarly with other age groups. When the now added to urban survivors of each age group. resulting
transferred population is subtracted from calculated in the projected numbers which appear in the last column .
survivors. finally. we arrive at the projected numbers of Again. as in table 44a. numbers of births and survivors to
the rural population for 1975. shown in the last column . ages 0 to 4 are still missing.
At this stage of the calculation. numbers aged 0 to 4 years 378. Table 44c. finally. shows how the births of inter-
are still missing because the births of the preceding vening periods are calculated . Numbers of women of
period remain to be calculated. reproductive age at the various dates are copied from the
 44s. STANDARD
TABLE PROJECTION OF URBAN AND RURAL POPULATION FOR BRAZIL
(PART OF WORKSHEET IN WHICH URBAN POPULATION IS CALCULATED

Urban Survival Trans- Urban

population ratio Survivors /erred population
Sex and age 1970 foe, = 65.8) 1975 population 1975
-.

Both sexes ................. 51,872 64,379

Males ..................... 24. 966 31. 304
(to be born) . . . . . . . . . . . . . . (4. 583) a (0.9438) b (5. 300)
0-4 .................... 3. 631 0.9818 4. 325
5-9 .................... 3. 489 0.9937 3. 565 458 4. 023
10-14 .................... 3. 130 0.9928 3. 467 382 3. 849
15.19 .................... 2. 668 0.9893 3. 107 337 3. 444
20-24 .................... 2. 273 0.9873 2. 639 290 2. 929
25.29 .................... 1. 795 0.9866 2. 244 218 2. 462
30-24 .................... 1. 650 0.9849 1. 771 164 1. 935
35.39 .................... 1. 470 0.9808 1. 625 126 1. 751
40-44 .................... 1. 297 0.9728 1 . 442 84 1. 526
45.49 .................... 1. 098 0.9596 1. 262 74 1. 336
50-54 .................... 858 0.9400 1. 054 65 1. 119
55.59 .................... 668 0.9098 807 49 856
60-64 .................... 500 0.8639 608 36 644
65-69 .................... 372 (0.7012) a 432 27 459
70-74 .................... 496 609 37 646
Females . . . . . . . . . . . . . . . . . . .
(to be born) ..............
0-4 ....................
26. 906
(4. 365)
3. 566
. (0.9535)
0.9844
33. 075
(5.048)
4. 162 d
5-9 .................... 3. 437 0.9948 3. 510 495 4. 005
10-14 .................... 3. 227 0.9941 3. 419 438 3. 857
15-19 .................... 3. 081 0.9914 3. 208 380 3. 588
20-24 .................... 2. 586 0.9896 3. 055 302 3. 357
25.29 .................... 2. 010 0.9884 2 . 559 227 2. 786
30-34 ....................
35.39 ....................
40-44 ....................
1. 826
1. 564
1. 360
0.9868
0.9841
0.9788
.
1. 987
1 802
1. 539
159
121
88
2. 146
1. 923
1. 627
45-49 .................... 1. 143 0.9700 1. 331 80 1. 411
50-54 .................... 923 0.9570 1. 109 73 1. 182
55.59 .................... 732 0.9351 883 59 942
60-64 .................... 551 0.8971 684 44 728
65-69 .................... 417 0.7422) 494 33 527
70 and over .............. 640 785 49 834

To be born during 1970-1975; see table 44c.

a
Survival ratio from a five-year period of births to ages 0 to 4 .
To be born during 1975.1980 .
6 Determined after calculation of number of births .

Assumed survival ratio from ages 65 and over to ages 70 and over . Actually the survival ratio
from ages 70 to 74 to ages 75 t o 79 according to model life table .

worksheets on rural and urban population for the begin- another time interval up to 1970. but the projected num-
ning and the end of each period . Multiplication with bers of reproductive women are not available until the
the coefficients describing the fertility age pattern and calculations of tables 44a and 44b are likewise extended.
addition of the products results in the weighted sums The numbers of births calculated in this fashion are then
shown below those numbers . The weighted sums are inserted in the appropriate places in tables 44a and 44b
then averaged for each period to represent reproductive and multiplied with the corresponding survival ratios to
women during the periods for which births are to be yield numbers of eventual survivors at ages 0 to 4 .
calculated. Multiplication of these mean weighted sums
with the GRR assumed for the same period directly Presentation of results for 1970-2000
yields the projected numbers of girls to be born; and 379 . The projection. as described. has been calculated
multiplication of the latter by 1.05 yields the correspond- by repeating the same procedure in each successive
ing numbers of boys . The calculation for both segments five-year period up to the year 2000. Projected numbers
of the population. urban and rural. proceeds in the same of the urban and rural population by sex and age. at
fashion. In table 44c. the worksheet is extended for time intervals of five years. are shown in tables 45a and
 TABLE
4 4 ~ . STANDARD
PROJECTION OF URBAN AND RURAL POPULATION FOR BRAZIL
(PART OF WORKSHEET IN WHICH BIRTHS ARE CALCULATED)

Fertility
ase
nattern

Rural population
Women
15-19.. ................ 0.07
20-24.. ................ 0.23

Weighted sum ............ 1.00

Mean of period .....................
GRR ..............................
Girls born a . . . . . . . . . . . . . . . . . . . . . . . . .
Boysborna .........................

Urban population
Women
15-19 . . . . . . . . . . . . . . . . . . 0.07 3,081 3,588
20-24.. . . . . . . . . . . . . . . . . 0.23 2,586 3,357

Weighted sum ............ 1.00 2,047 2,598 3,203

Mean of period . . . . . . . . . . . . . . . . . . . . . 2,322 2,901
GRR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.88 1.74
Girls born . . . . . . . . . . . . . . . . . . . . . . . . . 4,365 5,048
Boys born b ......................... 4,583 5,300

a To be entered in table 44a. TO be entered in table 44b.

45b. It will be seen that the combination of assump- 1970 to 74.1 per cent in 2000. Shifts in the age composi-
tions adopted for the standard projection causes the tions of the urban and rural populations are also indicated.
urban population to increase from 52 million in 1970 to At all dates of the projection, the rural population is more
143 million in 2000, while the rural population, 42 million youthful and has smaller proportions of middle-aged and
in 1970, would grow at a decreasing rate reaching a elderly persons than does the urban population. In
maximum of about 50 million around 1995, and diminish both populations, the projection implies decreases in
slowly thereafter. Mainly because of declining fertility the proportions of children and increases in the proportion
rates, the age structures of both the urban and the rural of the middle-aged and elderly. As projected, the age
population will be modiiied. For instance, in the urban composition of the rural population in the year 2000 may
population numbers of small children will merely double resemble that of the urban population in the year 1970.
between 1970 and 2000, while numbers at the most Finally, an appreciable surplus of women is maintained
advanced ages will increase fivefold; in the rural popu- in the urban population, and a surplus of men in the rural,
lation between 1985 and 2000 numbers below the age of throughout the projection period.
15 would diminish, while at subsequent ages numbers
would continue to increase. Because of the greater The components of change
intensity of female rates of rural-to-urban population 381. From the worksheets used in the calculation of
transfer, the urban population at all dates from 1970 to the projections, it is possible to deduce the components
2000 maintains a relative surplus of females over males of of demographic change in the total, urban and rural
almost two million, whereas in the rural population there population. The following sequence of calculations is
continues to be a relative deficit of females of nearly the involved.
same magnitude. 382. First, the amounts of population growth can be
380. Table 46 summarizes some further results of the deduced for each time interval of the projection by
projection. It can be seen, for instance, that the level of subtracting the population total at the end from that of
urbanization is projected to rise from 55.6 per cent in the beginning of each period.
 4 5. ~ STANDARD
TABLE PROJECTION FOR THE URBAN AND RURAL POPULATION OF BRAZIL. 1970-2000;
URBAN POPULATION. BY SEX AND AGE. AT RVE-YEAR TIME INTERVALS

Sex and age 1970 1975 1980 1985 1990 1995 2WO

Both sexes............. 52.458 64.379

Males ................. 25. 395 31. 304
0 4 ............... 3. 631 4. 325
5-9 . . . . . . . . . . . . . . . 3. 489 4 . 023
10-14 . . . . . . . . . . . . . . . 3. 130 3. 849
15-19 . . . . . . . . . . . . . . . 2. 668 3. 444
20-24 ............... 2 . 273 2. 929
25-29 ............... 1. 795 2. 462
30-34 . . . . . . . . . . . . . . . 1. 650 1 . 935
35-39 ............... 1. 470 1. 751
40-44 ............... 1. 297 1 . 526
4 5 4 9 ............... 1. 098 1. 336
50-54 ............... 858 1. 119
55-59 ............... 668 856
60-64 ............... 500 644
65-69 ............... 372 459
70 and over .......... 496 646
Females ............... 27. 063 33. 075
0 4 ............... 3. 566 4 . 162
5-9 ............... 3. 437 4 . 005
10-14 ............... 3. 227 3. 857
15-19 ............... 3. 081 3. 588
20-24 ............... 2. 586 3. 357
25-29 ............... 2. 010 2. 786
30-34 ............... 1. 826 2. 146
35-39 ............... 1. 564 1. 923
40-44 ............... 1. 360 1. 627
45-49 ............... 1. 143 1. 411
50-54 ............... 923 1. 182
55-59 ............... 732 942
60-64 ............... 551 728
65-69 ............... 417 527
70 and over .......... 640 834
Males per 100 females ... 93.8 94.6

383. Secondly. the number of births can be roughly births of small children to transferred (migrant or reclas-
ascertained for each time period. as calculated in the sified) women at the end of any five-year period may
projection worksheet (table 44c). subject to a correction have occurred before the transfer (e.g. before migration).
which will ~resentlvbe discussed. and therefore still in the rural area; or after the transfer.
384. Thirdly. the numbers of rural-to-urban transfers and therefore already in the urban area . The transferred
(ages 5 and over) can be added up for each period from population itself must have been slightly more numerous
than the survivors of the transfer calculated in the work-
the corresponding worksheet (table 44a). again subject
to further corrections. sheets because some individuals. alive at the time of the
transfer (or migration). may have died subsequent to
.
385 Then the amount of natural increase can be that transfer. yet within the same five-year period . A
calculated as the difference between the amount of popu- precise calculation of all the quantities involved would be
lation growth and the amount of rural-to-urban transfer . feasible only in a simulation model for an electronic
And finally the number of deaths can be derived as the computer. and this cannot be considered in the present
difference between births and natural increase. All .
context But the complicating instances (e.g. migrants
these amounts-growth. natural increase. births. deaths who died or children born to women before migration)
and transfers- can be transformed into crude rates by are relatively not very numerous. hence a few rough
dividing one-fifth the numbers of each five-year period adjustments will suffice. In this connexion. it will be
by the mean population (total. urban. rural) of the period. assumed that each five-year period's transferred popu-
for simplicity taken as the arithmetic mean between the lation. on an average. spent one-half of that period in the
population at the period's beginning and that at its end . rural area and the other half in the urban area.
386. Complications arise from several sources. but 387. To correct the number of births occurring in
these can be dealt with by fairly easy adjustments . The rural areas. the calculations of births (worksheet. see

99
 4 5 .~ STANDARD
TABLE PROJECTION FOR THE URBAN AND RURAL POPULATION OF Bruzn. 1970-2000;
RURAL POPULATION. BY SEX AND AGE. AT FIVE-YEAR TIME INTERVALS

Sex and age

........ -

Both sexes .............

Males .................
0-4 ...............
5-9 ...............
10-14 ...............
15-19 ...............
20-24 ...............
25-29 ...............
30-34 ...............
35-39 ...............
4 w ...............
45-49 ...............
50-54 ...............
55-59 ...............
60-64 ...............
65-69 ...............
70 and over ..........
Females ...............
0-4 ...............
5-9 . . . . . . . . . . . . . . .
10-14 ...............
15-19 ...............
20-24 ...............
25-29 . . . . . . . . . . . . . . .
30-34 ...............
35-39 ...............
40-44 ...............
45-49 ...............
50-54 ...............
55-59 ...............
60-64 ...............
65-69 ...............
70 and over ..........
Males per 100 females ...

table 44c) are repeated by substituting. for the end of worksheets (table 44a or 44b) indicates that an adequate
each period. the numbers of women surviving from the correction is made if it is estimated that the actual trans-
rural population (i.e. numbers prior to their reduction fers (including persons who died shortly thereafter) were
by transfer) to obtain numbers of rural births which about one per cent more numerous than the surviving
would have occurred if there had been no transfer at all transferred population originally calculated in the
(third column in table 44a). Then the difference between projection . 113 In individual circumstances. the reader
the results of the two calculations of births is halved so may wish to apply other correction techniques not
as to represent those numbers of births still occurring in described here .
the rural area among the women transferred to the urban 389. These corrections apply in the urban and rural
population within the same period . This number is population only. The total population is unaffected.
added to the original calculation of rural births and since no assumption has been made with regard to inter-
subtracted from the original calculation of urban national migration. Table 47 shows the amounts
births. involved in the components of population change. and
388. The amount of correction for the number of births table 48 shows the corresponding rates (per 1.000 of the
is also added to the first results of the calculation of mean population of each period).
'
rural-urban transfers since children of transferred women 390. From table 47 it can be deduced that the growth
still born in the rural area were presumably (at least for in total population will be by slowly increasing amounts
the most part) also transferred. A further adjustment
remains to be made in calculated numbers of transfers
'la This would imply a crude death rate among migrants of 2 per
to take account of additional transfers of persons followed. 1. 000 per year. which is possible considering that most migrants are
within the same period. also by the deaths of those same young adults. However. the reclassified population might well
persons. A rough calculation utilizing figures in the have a somewhat higher crude death rate.
 TABLE
46. STANDARD
PROJECTION FOR THE URBAN AND RURAL POPULATION OF BRAZIL,1970-2000;
SELECTED RESULTS

Population (thousands)
Total population ........ 94,360 108,719 124,411 141,260 158,686 176,157 193,017
Urban population .... 52,458 64,379 77,791 92,765 108,891 125,790 143,044
Rural population ..... 41,902 44,340 46,620 48,495 49,795 50,367 49,973
Percentage urban ..... 55.6 59.2 62.5 65.7 68.6 71.4 74.1

Percentage age composition of the urban population (both sexes)

All ages . . . . . . . . . . . . . . . 100.0 100.0 100.0 100.0 100.0 100.0 100.0
0-4 ............... 13.7 13.2 12.8 12.2 11.5 10.7 9.9
5-14 . . . . . . . . . . . . . . . 25.3 24.4 23.5 23.0 22.3 21.3 20.1
15-24 . . . . . . . . . . . . . . . 20.2 20.7 20.6 20.0 19.7 19.6 19.3
25-44 . . . . . . . . . . . . . . . 24.7 25.1 26.0 27.2 28.5 29.1 29.7
45-64 . . . . . . . . . . . . . . . 12.3 12.8 13.0 13.1 13.2 14.0 15.3
65 and over .......... 3.7 3.8 4.1 4.4 4.9 5.3 5.8

Percentage age composition of the rural population (both sexes)

All ages ............... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
0-4 ............... 18.7 19.5 18.8 17.6 16.4 14.9 13.1
5-14 ............... 27.9 27.4 28.3 28.8 28.0 26.8 25.3
15-24 ............... 19.1 18.8 18.4 18.5 19.6 20.6 20.7
25-44 . . . . . . . . . . . . . . . 20.9 20.7 21.1 21.4 22.1 23.0 24.7
45-64 ............... 10.4 10.4 10.3 10.3 10.3 10.9 11.9
65 and over .......... 3 .O 3.0 3.2 3.4 3.6 3.9 4.2

Males per 100 females in the population (all ages)

Total population ........ 98.9 99.2 99.3 99.6 99.8 100.1 100.3
Urban population .... 93.8 94.6 95.1 95.8 96.5 97.0 97.5
Rural population ..... 105.6 106.2 106.7 107.1 107.6 108.2 108.8

until about 1990, and then by slowly decreasing amounts. but the projected decline in the birth rate will be more
Likewise, numbers of births will rise to a maximum about rapid than that in the death rate, which is already quite
1990 and then slowly diminish. The numbers of deaths low.
in the total population will decrease somewhat to reach 395. The urban birth rate may decline from 29.4 to
a minimum about 1985, and then begin to increase, 21.0 per 1,000 in the projection period, and the rural
mainly as a result of changes in age structure. birth rate, from 45.7 to 27.7 per 1,000. The urban
391. The urban population will grow by increasing death rate may fall from 6.8 to 4.6 per 1,000, and the
amounts to the end of the century. Of the amount of rural death rate from 9.6 to 4.4. Population transfers
growth, 45 per cent will be due to rural-to-urban transfers will make a decreasing contribution to the urban popu-
in 1970-1975, and 55 per cent to natural increase. By lation because of its rapid growth, as compared with the
1995-2000, 36 per cent of the urban growth will be much slower growth of the rural population whence the
caused by transfers, and 64 per cent by natural increase. transfers come. In the rural population, the assumption
of the projection has the effect that the net transfer rates
392. Growth of the rural population will slow down stays at an almost constant level.
and give way to a slight decrease by the century's end.
The natural increase of the rural population would attain
its largest amount in 1975-1980 and thereafter it would C. VARIANT PROJECTIONS
diminish. Two-thirds of the natural increase are offset Population projections as part of systems analysis
by rural-to-urban population transfer in 1970-1975,
and at the century's end the net transfer will exceed the 396. The availability of several alternative population
rural natural increase. projections detailed by urban and rural areas (if not
also in other respects: sex, age, marital status, economic
393. Other inferences can be deduced from table 48, activity, level of education and so on, can make a valuable
showing the components of population change in terms contribution to decision-making processes concerned
of crude rates. with the adoption of one or another course of action in
394. The entire country's population growth will fields of economic, social, cultural or administrative
occur at diminishing rates, from a rate of 28.4 per 1,000 policy, regional and physical planning and so forth. It
in 1970-1975, to one of 18.3 per 1,000 in 1995-2000. is to be noted that for such purposes the absolute magni-
Both the national birth rates and death rates will diminish, tudes reached in the population projections are only of
 47. STANDARD
TABLE PROJECTION FOR THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000.
AMOUNTSOF GROWTH. NATURAL INCREASE. BIRTHS. DEATHS AND NET POPULATION TRANSFERS
IN TOTAL. URBAN AND RURAL POPULATION. BY FIVE-YEAR PERIODS

Quantity 1970-1975 1975-1980 1980-1985 1985-1990 1990-1995 1995-2000

Total population
Growth .............. 14. 397 15. 665 16. 849
Natural increase . . . . . . . . . . 14. 397 15. 665 16. 849
Births ................. 18. 440 19. 737 20. 717
Deaths ................ 4. 043 4. 072 3. 868
Net transfer .............. 0 0 0
Urban population
Growth ............. 11. 921 13. 412 14. 974
Natural increase ...... 6. 609 7. 908 9. 095
Births ............. 8. 584 9. 992 11. 350
Deaths ............ 1. 975 2. 183 2. 255
Net transfer .......... 5. 312 5. 603 5. 879
Rural population
Growth ................. 2. 476 2. 253 1. 875
Natural increase .......... 7. 788 7. 856 7. 754
Births ................ 9. 856 9. 745 9. 367
Deaths ................ 2. 068 1. 889 1. 613
Net transfer .............. -5. 312 -5. 603 -5. 879

TABLE48 . STANDARDPROJECTION FOR THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000.

RATESOF GROWTH. NATURAL INCREASE. BIRTHS. DEATHS AND NET POPULATION TRANSFERS IN
TOTAL. URBAN AND RURAL POPULATION. BY FIVE-YEAR PERIODS. PER 1.000 INHABITANTS

Rate 1970-1975 1975-1980 1980-1985 1985-1990 1990-1995 1995-2000

Total population
Growth ................. 28.4 26.9 25.4 23.2 20.9 18.3
Natural increase .......... 28 .A 26.9 25.4 23.2 20.9 18.3
Births ................. 36.3 33.9 31.2 28.4 25.6 22.8
Deaths ................ 8.0 7.0 5.8 5.1 4.7 4.5
Net transfer .............. 0.0 0.0 0.0 0.0 0.0 0.0
Urban population
Growth ................. 40.8 37.7 35.1 32.0 28.8 25.7
Natural increase . . . . . . . . . . . 22.6 22.0 21.3 19.9 18.2 16.4
Births ................. 29.4 28.1 26.6 24.8 22.9 21. 0
Deaths ................ 6.8 6.1 5.3 4.9 4.7 4.6
Net transfer .............. 18.2 15.8 13.8 12.1 10.6 9.3
Rural population
Growth . . . . . . . . . . . . . . . . . 11.5 9.9 7.9 5.3 2.3 -1.6
Natural increase .......... 36.1 34.5 32.6 30.1 27.1 23.3
Births ................. 45.7 42.8 39.4 35.8 32.0 27.7
Deaths . . . . . . . . . . . . . . . . 9.6 8.3 6.8 5.7 4.9 4.4
Nettransfer .............. -24.6 -24.6 -24.7 -24.8 -24.8 -24.8
secondary importance. Comparisons have to be made security services and so forth, without a commensurate
among alternative projections, hence what is really increase in the tax-paying capacity of the urban popu-
important is that the dzyerences between the results of lation. Rising expenditures may drain the municipal
such projections, their direction and magnitude, are valid budget, and some of the expected services including that
enough. The forecasting value of the projections in of subsidized housing may then have to be cut back.
this context is not of particular relevance. The assembly The initial action in its eventual results may thus defeat
of several alternative projections for comparative pur- its own purpose and others as well. 114
poses, preferably projections in which one basic assump-
tion is varied at a time, constitutes what is called a 400. To vary the example, an argument like the follow-
scenario. ing may also be conceivable. To relieve unemployment
in a depressed area, an industrial estate may be established
397. In this discussion it is necessary at once to point at public cost, including the provision of some infra-
at the great complexity of background factors underlying structure (e.g. a power supply, improved transport
a scenario for policy decisions. True, specific fiscal facilities, a training institute enabling adolescents to
measures, programmes in housing, public health, educa- acquire economic skills or some other projects). Fiscal
tion, employment and so on, each can have effects on one measures (subsidies, tax relief) may then be taken to
if not several of the population trend components attract an initial group of entrepreneurs who establish
(fertility, mortality, migration and reclassification), and some local industries. Once the initial project is launched,
the calculated interaction of the demographic components additional enterprises may be attracted without further
with the dynamics of the evolving age structure results in public expenditure because of their complen~entaritywith
population projections relevant to these considerations. industries already established, and benefiting also from
However, each of the various lines of policy also affect a labour supply whose skills have been improved. The
economic, social, cultural and other conditions, require- incomes generated create demands for additional local
ments and trends, and these also interact among each services, and the economic activities in the area may then
other quite apart from their interaction with population multiply and diversify. The new industries may include
trends. The population trends themselves have further the processing of agricultural products raised inthevicinity
consequences on economic, social, fiscal, housing and and contribute to the prosperity of the farmers. Not
other conditions. To take account of the entire inter- only is a high level of employment achieved within the
play of multiple factors and their multiple interactions, area, but skilled workers are also attracted from other
a complex model may have to be built for computer areas. The eventual outcome may be a prosperous town
simulation. A large amount of specific information is and rural hinterland capable of supporting a large tax
needed for the construction of such a model, including revenue, far in excess of the public expenditure involved
absolute magnitudes, trends and the presumed inter- in the initial project which, so to speak, merely primed
actions (correlations, regressions) within any pair of the pump. With such eventual results, the initial action
the component trends. Because of its complexity, such can be said to have been catalytic. 116
an interactive model is referred to as a system.
401. Even though a line of public action need not be
398. Systems analysis is an important tool for reasoning expected to have extreme (self-defeating or catalytic)
where the eventual effects of any particular line of action effects, it is evident that a full appreciation of all its
are far from obvious. There can be multiple feedbacks important consequences, including perhaps many favour-
leading to eventual results which are rather different from able or unfavourable side-effects, depends on a consider-
those expected as the seemingly more direct outcome. ation of a complex interplay among numerous factors.
Depending on likely further repercussions within the
system, a given line of action may even prove to be even- 402. In passing, it may be mentioned that even the
tually self-defeating. In the opposite case, where desir- demographic factors can have effects upon each other.
able eventual effects can become much larger than those Motivation to reduce family size may be increased when
immediately following the action, the particular measures infant and child mortality reaches a low level. Migrants
taken can have a catalytic effect. Without implying may have a lower fertility than persons remaining in
that such consequences will necessarily result in any given their places of residence. Migration can become a
instance, reasoning as in the following two examples is at major cause for the urban reclassification of hitherto
least conceivable. rural areas; and so forth.
399. A programme to increase the urban housing
supply could actually have the unintended consequenceof ""nother argument of this type has been illustrated in a compu-
increasing the housing shortage and producing also some terized system in Jay W. Forrester, Urban Dynamics (The M.I.T.
other undesirable effects. The subsidized housing pro- Press, Cambridge, Mass., 1969). Forrester arrives at the pessi-
ject can encourage an increased fertility among the mistic conclusion that self-defeat may be inherent in most social
measures. "In a situation where coincident symptoms appear to
resident families. Rural-to-urban migration can be be causes, a person acts to dispel the symptoms. But the under-
stimulated as more potential migrants entertain an expec- lying causes remain. The treatment is either ineffective or detri-
tation that their urban housing might be facilitated by mental. With a high degree of confidence, we can say that the
the project. The demand for broader welfare measures intuitive solutions to the problems of complex social systems will
be wrong most of the time" (Ibid., p. 110).
may increase and the urban government may have to 116 This line of reasoning, more optimistic than Forrester's conclu-
make the corresponding concessions. Again, more sion (see the preceding foot-note), is implicit in the presently evolving
migrants may be attracted to the city, increasing also theories of regional planning, especially in the concepts of satellite
overhead charges concerning transport, sewage, schools, towns, growth poles, service centres and so forth.
 403. The con~plexcomputer models used for an anal- realism the situations which various quite conceivable
ysis of the response of an entire system of factors to developments may give rise to. The simple model may
any given line or amplitude of policy action require, aside sometimes suffice to indicate the desirability of particular
from mathematical and programming skill, a large amount instruments of economic, social, administrative and some-
of research into the possible interrelation between any times also population policy. The accuracy of such a
two factors. The apparent interplay between two factors model may be sufficient to indicate the direction as well
can only seldom be observed in isolation because in as the magnitude of the impact to be expected if certain
actual experience there would usually be a further elements of the given situation are permitted to change.
interference by additional factors which cannot be con- The simple model reflecting variability in demographic
trolled as they might be in a laboratory experiment. factors only can also be regarded as a subsystem suitable
Many points of detail will probably remain debatable. for integration into a wider system, in which the interplay
Nevertheless, the exercise can lead to much valuable of other, non-demographic factors is included.
insight. For instance, there is the possibility that short- 407. The authors of this manual are not thoroughly
run consequences may differ from those in the long run, familiar with the relevant economic, social, administra-
and decisions may have to be formulated accordingly. tive and other circumstances in Brazil, nor have they been
A given line of action, aimed at short-run effects, may able to consult with the respective national experts.
then have to be followed by suitably timed corrective The calculations, furthermore, are carried out with
action in the longer run to neutralize or counteract even- simple calculating equipment and without the use of
tual undesirable side effects. computers. The aim of comparing the standard projec-
tion (reviewed in the preceding section) with each of
The present purpose several variants is strictly methodological, namely to
show what differences in results are obtained if only
404. The methods presented in this manual are pri- one or another of the demographic assumptions in the
marily intended for relatively short- to medium-term standard projection is varied. It is not implied that any
usage, such as periods from ten to thirty years. The one of the calculated variants has a likelihood of approxi-
methods presented in chapters IV to VIII have been in mating the as yet unknown future course of events, and
the nature of simple forecasts rather than models. 116 some of the modified assumptions are quite frankly
Forecasts are calculated on the assumption that some unrealistic, as can at once be recognized.
reasonable continuation or modification of an observed 408. The following variant projections will be con-
past trend will assert itself also in the near future. Simple
sidered :
forecasts may serve well where reasonably dependable
results are required for immediate practical purposes, as Variant I: it is assumed-quite unrealistically-that
past trends are likely to change only gradually, barring urban and rural populations grow by natural increase
catastrophic developments. only, and that there are no rural-to-urban population
transfers;
405. Projections calculated on the basis of specified
assumptions for a small number of trend components Variant 11: it is assumed that there will occur no rural-
have the advantage of permitting an evaluation of what to-urban area reclassifications, i.e. urban and rural areas
the comparative consequences might be if one or another will remain the same, although migratory transfers will
of the component assumptions has to be modified. The occur between the two sets of areas;
reasons why they have to be modified, under the influence Variant IZI: it is assumed that rural-to-urban popu-
of what development in the total situation and to what lation transfers will occur at the lower rates calculated
extent remain unexplained. The choice of modifications for the 1950-1960 period, rather than at the rates of the
in the specific assumptions is left as a mere matter of 1960-1970 period assumed in the standard projection;
judgement, preferably judgement based on some exper- Variant IV: it is assumed that rural-to-urban popu-
ience through past studies of the manner in which demo- lation transfers will occur with similar intensity as in the
graphic phenomena may respond to changing conditions standard projection, but with a different incidence by
or may be influenced by specific measures. This subject sex and age; the particular (unrealistic) assumption made
matter lies outside the scope of this manual. 117 is that transfers of males will occur at the rates originally
406. In the sections which follow, the attempt has been assumed for females, and transfers of females at the rates
made to develop a population model limited to demo- originally assumed for males;
graphic factors only and permitting the insertion of Variant V: it is assumed that mortality will not
arbitrary changes in the component assumptions. Sub- improve, but will remain constantly at the levels estimated
ject to judgements of detail developed by other means, for 1970;
the simple model also facilitates some manipulation and Variant VI: it is assumed that fertility will not decrease,
permits at least some limited conclusions. With enough but will remain constantly at the levels estimated for
knowledge, detailed assumptions might be delineated 1970;
which can be believed to reflect with some degree of Variant VZI: as compared with the standard projec-
tion, it is assumed that urban fertility will decline more
ll6 For a discussion of the differences between forecasts and rapidly and rural fertility less rapidly;
models, see the introduction to this manual.
Variant VIII: as compared with the standard projec-
117 It is reviewed at length, for instance, in The Determinants and
Consequences of Population Trends (United Nations publication, tion, it is assumed that urban fertility will decline less
Sales No. E.71.XIII.S). rapidly and rural fertility more rapidly.
 TABLE49. VARIANT 1 AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATIONOF BRAZIL.1970-2000;
COMPARISON OF SELECTED RESULTS

Standard projection Variant I

Base data .
1970 1980 1990 2000 1980 1990 2000

Population (thousands)
Total population .............. 94. 360 124. 411 158. 686 193. 017 125.572 163.499 202. 887
Urban population .......... 52. 458 77. 791 108. 891 143. 044 66. 221 81. 339 95. 758
Rural population ........... 41. 902 46. 620 49. 795 49. 973 59. 351 82. 160 107. 129
Percentage urban ........... 55.6 62.5 68.6 74.1 52.7 49.7 47.2
Percentage age composition of the urban population (both sexes)
All ages ..................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 13.7 12.8 11.5 9.9 13.2 11.5 9.6
5-14 ..................... 25.3 23.5 22.3 20.1 22.7 21.9 19.4
15-24 ..................... 20.2 20.6 19.7 19.3 19.8 18.4 18.5
25-44 ..................... 24.7 26.0 28.5 29.7 26.3 28.5 28.9
45-64 ..................... 12.3 13.0 13.2 15.3 13.7 14.4 16.9
65 and over ................ 3.7 4.1 4.9 5.8 4.3 5.4 6.7
Percentage age composition of the rural population (both sexes)
All ages ..................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
0 4 ..................... 18.7 18.8 16.4 13.1 18.3 16.5 13.2
5-14 ..................... 27.9 28.3 28.0 25.3 28.1 27.8 25.8
15-24 ..................... 19.1 18.4 19.6 20.7 19.3 20.1 21.1
25-44 ..................... 20.9 21.1 22.1 24.7 21.4 22.9 25.5
45-64 ..................... 10.4 10.3 10.3 11.9 9.8 9.5 10.8
65 and over ................ 3.0 3.2 3.6 4.2 3.1 3.3 3.6
Males per 100 females in the population (all ages)
Total population .............. 98.9 99.3 99.8 100.3 99.3 99.9 100.5
Urban population .......... 93.8 95.1 96.5 97.5 95.4 96.9 98.0
Rural population ........... 105.6 106.7 107.6 108.8 104.0 103.1 102.8

Standard projection Variant I

Rate 1970-1980 1980-1990 1990-2000 1970-1980 1980-1990 1990-2000

Total population
Growth ................. 27.6 24.3 19.6
Natural increase .......... 27.6 24.3 19.6
Births ................. 35.1 29.8 24.2
Deaths ................ 7.5 5.5 4.6
Net transfer .............. 0.0 0.0 0.0
Urban populatioiz
Growth ................. 39.3 33.6 27.3
Natural increase .......... 22.3 20.6 17.3
Births ................. 28.7 25.7 21.9
Deaths ................ 6.4 5.1 4.6
Net transfer .............. 17.0 13.0 10.0
Rural population
Growth ................. 10.7 6.6 0.4
Natural increase .......... 35.3 31.4 25.2
Births ................. 44.3 37.6 29.8
Deaths ................ 9.0 6.3 4.6
Net transfer .............. - 24.6 -24.8 -24.8
 409. For each variant, summary tables of results will standard projection, but that no reclassification of areas
be presented, making comparisons with some of the would occur. In the calculation, net migration rates
figures already shown in tables 46 and 48 of the preceding are used as determined in table 42, for the 1965-1970
section for the standard projection. period. Theoretically, this projection should result in
future population estimates within and outside those
Variant I: No rural-to-urban population transfers areas which in 1970 were classified as "urban". The
projection does not take into account those aspects of the
410. This variant differs from the standard projection
urbanization process which result in new areas settled
in so far as it is assumed that no rural-to-urban popu-
under urban residential densities. Furthermore, since
lation transfers would occur from 1970 onward. Both
probably much migration of rural origin is actually
the urban and rural populations would grow indepen-
directed at areas in process of reclassification, rather
dently of each other according to rates of natural increase
than at areas already classified as "urban", only a dimin-
The assumption is obviously unrealistic since in actual
fact the urbanization trend can hardly come to an abrupt ishing proportion of the combined rural-to-urban migra-
tory flows is reflected in this variant. The assumption
end. But comparison of results of this variant with
those of the standard projection permits an appraisal of that the migration pattern to areas already classified as
urban in 1970 will remain constant is also debatable.
the extent to which the growth in urban, rural and total
population tends to be modified by migration and reclas- With the geographic expansion of existing urban agglo-
merations, a redistribution of population between the
sification. The comparison is made in tables 49 and 50.
"old" urban centre and the "new" urban periphery
41 1. As can be seen in table 49, under this variant the occurs in actuality, and it is possible that eventually,
total population of Brazil may grow to 203 million by owing to increasing suburbanization, a negative migratory
the year 2000, rather than 193 million as in the standard balance develops within the "old" city cores, more
projection. This additional population growth would be persons moving from the city centres into suburbs than
the consequence of more women remaining in the rural new migrants moving into the central parts of the city.
areas where their fertility is at a higher level. Another Despite these possible incongruities in the assumption,
consequence would be a more rapid population growth the calculation is worth while because it permits compari-
in rural areas than in urban areas, because in the rural sons to be made both with the standard projection and
areas the rate of natural increase also is higher. In with variant I.
fact, under this assumption the level of urbanization 415. Table 51 again shows the surprising result that
would decline, from 55.6 per cent in 1970 to 47.2 per cent urban and rural age compositions are little affected by
in 2000. rural-to-urban migration, although they are responsive
412. Surprisingly, despite the greatly modified amounts to the fertility and mortality changes assumed in both
of growth in urban, rural and total population, under this variant I1 and the standard projection. Another sur-
variant, the age compositions of urban and rural popu- prising result is that, with the apparently prevailing
lation will resemble closely those calculated in the tempo of net rural-to-urban migration, the urban popu-
standard projection. Ageing of the population, as lation grows only slightly faster than the rural population,
measured by the percentage aged 65 years and over, bringing the urbanization level from 55.6 per cent in
would be slightly more rapid in the urban population in 1970 to only 57.7 per cent in 2000. As shown in table 52,
the absence of migration and reclassification and slightly the "net transfer" rates due to migration to established
slower in the rural population. It is evident that the urban areas only are comparatively slight, whereas rates
fertility and mortality trends assumed both in the stan- of natural increase differ appreciably between urban and
dard projection and in this variant have far greater effect rural areas. The reader may also wish to compare
on age composition than the combined effects of migra- tables 51 and 52 with tables 49 and 50 (variant I), the
tion and area reclassification. There would however be difference of results being due to the effects of reclassi-
a marked effect on sex composition, especially in the fication, as well as the inclusion of migration occurring
rural population. In the standard projection, because from rural places to new areas in process of being
more females than males are involved in the population reclassified.
transfer, the rural sex ratio would rise from 105.6 in
1970 to 108.8 in 2000, whereas in variant I it would Variant 111: Lower rates of rural-to-urban
diminish to 102.8 in 2000. populution trunsfers
413. As shown in table 50, the assumption of no rural-
416. Since it is uncertain whether rural-to-urban popu-
to-urban transfers would have the effect of a more rapid lation transfers will continue to occur at the rates cal-
decline of the rate of natural increase in urban areas and culated (in section A of this chapter) for the 1950-1970
a slower decline in rural areas. Because of the greater period, a variant is here introduced to show the effects
proportion of rural in total population, the rate of of a lower intensity of such movements, as calculated
natural increase in the combined total population would for the 1950-1960 period (also shown in section A).
be significantly higher under this variant than according By the same token, higher rates of rural-to-urban popu-
to the standard projection. lation transfers might also have been assumed, in which
case the comparison of results with the standard projection
Variant 11: No reclassification would lead to the opposite observations. Other assump-
414. In this variant it is assumed that rural-to-urban tions might also have been made, such as a change in
migration would continue at the same rates as in the transfer rates in the course of time.
 TABLE51 . VARIANTU AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000;
COMPARISON OF SELECTED RESULTS

Standard projection Variant 11

Base data .
1970 1980 1990 2000 1980 1990 2000

Population (thousands)
Total population .............. 94. 360 124.411 158.686 193.017 125.317 161.863 199. 107
Urban population .......... 52. 458 77. 791 108. 891 143. 044 70. 104 91. 496 114.973
Rural population ........... 41. 902 46. 620 49. 795 49. 973 55. 213 70. 367 84. 134
Percentage urban ........... 55.6 62.5 68.6 74.1 55.9 56.5 57.7
Percentage age composition of the urban population (both sexes)
AN ages ..................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 13.7 12.8 11.5 9.9 13.0 11.6 9.9
5-14 ..................... 25.3 23.5 22.3 20.1 23.5 22.6 20.4
15-24 ..................... 20.2 20.6 19.7 19.3 20.5 19.7 19.6
25-44 ..................... 24.7 26.0 28.5 29.7 26.0 28.4 29.5
45-64 ..................... 12.3 13.0 13.2 15.3 12.9 13.0 14.9
65 and over ................ 3.7 4.1 4.9 5.8 4.1 4.7 5.7
Percentage age composition of the rural population (both sexes)
AN ages .................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
C 4 ..................... 18.7 18.8 16.4 13.1 18.5 16.2 12.8
5-14 ..................... 27.9 28.3 28.0 25.3 27.5 27.2 24.7
15-24 ..................... 19.1 18.4 19.6 20.7 18.5 19.0 20.2
25-44 ..................... 20.9 21.1 22.1 24.7 21.6 22.8 25.0
45-64 ..................... 10.4 10.3 10.3 11.9 10.5 10.9 12.7
65 and over ................ 3.0 3.2 3.6 4.2 3.3 3.9 4.6
Males per 100 females in the population (all ages)
Total population .............. 98.9 99.3 99.8 100.3 99.5 100.0 100.5
Urban population .......... 93.8 95.1 96.5 97.5 95.7 97.1 98.2
Rural population ........... 105.6 106.7 107.6 108.8 104.5 103.9 103.6

TABLE52. VARIANT11 AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION
OF BRAZIL.1970-2000; COMPAIUSON OF RATES OF GROWTH AND THEIR COMPONENTS

Standard projection Variant 11

Rate 1970-1980 1980-1990 1990-2000 1970-1980 1980-1990 1990-2000

Total population
Growth ................. 27.6
Natural increase ......... 27.6
Births ................. 35.1
Deaths ................ 7.5
Net transfer .............. 0.0
Urban population
Growth ................. 39.3 33.6 27.3
Natural increase .......... 22.3 20.6 17.3
Births ................. 28.7 25.7 21.9
Deaths ................ 6.4 5.1 4.6
Net transfer .............. 17.0 13.0 10.0
Rural population
Growth ................. 10.7 6.6 0.4
Natural increase .......... 35.3 31.4 25.2
Births ................. 44.3 37.6 29.8
Deaths ................ 9.0 6.2 4.6
Net transfer .............. -24.6 -24.8 -24.8
 TABLE53 . VARIANT
111 AND STANDARD P R O J E C ~ NOF THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000;
COMPARISON OF SELECTED RESULTS

Standard projection Varmt 111

Base dala
1970 1980 1990 2000 1980 1990 2000

Population (thousands)
Total population .............. 94. 360 124.411 158.686 193.017 124. 779
Urban population .......... 52. 458 77. 791 108.891 143.044 76. 589
Rural population ........... 41. 902 46. 620 49. 795 49. 973 48. 190
Percentage urban ........... 55.6 62.5 68.6 74.1 61.4
Percentage age composition of the urban population (both sexes)
AN ages ..................... 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 13.7 12.8 11.5 9.9 12.8
5-14 ..................... 25.3 23.5 22.3 20.1 23.5
15-24 ..................... 20.2 20.6 19.7 19.3 20.4
25114 ..................... 24.7 26.0 28.5 29.7 26.1
45-64 ..................... 12.3 13.6 13.2 15.3 13.1
65andover ................ 3.7 4.1 4.9 5.8 4.1
Percentage age composition of the rural population (both sexes)
AN ages ..................... 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 18.7 18.8 16.4 13.1 18.6
5-14 ..................... 27.9 28.3 28.0 25.3 28.5
15-24 ..................... 19.1 18.4 19.6 20.7 18.7
25-44 ..................... 20.9 21.1 22.1 24.7 21 .0
45-64 ..................... 10.4 10.3 10.3 11.9 10.1
65 and over ................ 3.0 3.2 3.6 4.2 3.1
Males per 100 females in the population (all ages)
Total population .............. 98.9 99.3 99.8 100.3 99.7
Urban population .......... 93.8 95.1 96.5 97.5 95.3
Rural population ........... 105.6 106.7 107.6 108.8 107.3

TABLE54. VARIANT 111 AND STANDARD PROJECnON OF THE URBAN AND RURAL POPULATION
OF BRAZIL.1970-2000; COMPARISON OF RATES OF GROWTH AND THEIR COMPONENTS

Standard projection Variant 111

Rate 1970-1980 1980-1990 1990-2000 1970-1980 1980-1990 1990-2000

Total population
Growth ................. 27.6 24.3 19.6 27.9 24.4 19.7
Natural increase .......... 27.6 24.3 19.6 27.9 24.4 19.7
Births ................. 35.1 29.8 24.2 35.2 30.0 24.4
Deaths ................ 7.5 5.5 4.6 7.3 5.5 4.6
Net transfer .............. 0.0 0.0 0.0 0.0 0.0 0.0
Urban population
Growth ................. 39.3 33.6 27.5 37.7 32.5 26.8
Natural increase .......... 22.3 20.6 17.3 22.8 20.6 17.2
Births ................. 28.7 25.7 21.9 28.9 25.7 21.9
Deaths ................ 6.4 5.1 4.6 6.1 5.1 4.7
Net transfer .............. 17.0 13.0 10.0 14.9 11.9 9.6
Rural population
Growth ................. 10.7 6.6 0.4 14.0 10.0 4.0
Natural increase .......... 35.3 31.4 25.2 35.1 31.2 25.3
Births ................. 44.3 37.6 29.8 44.0 37.5 29.9
Deaths ................ 9.0 6.2 4.6 8.9 6.3 4.5
Nettransfer .............. -24.6 -24.8 -24.8 -21.1 -21.2 -21.3
 417. As can be seen in table 53, transfer rates lower 423. Some modification of age structure is implied
than those of the standard projection would cause a in the mortality declines foreseen in the standard pro-
somewhat smaller increase in urban and larger increase jection. As compared with variant V, the eventual
in rural population, but the effect on age structures would urban and rural populations of the standard projection
be almost negligible. As can be deduced from table 54, have somewhat lower proportions of adults between the
the rates of growth in urban population would only be ages of 25 and 64, and somewhat higher proportions of
slightly less rapid, whereas the growth in rural popu- either children or aged persons. The increase in the
lation would be appreciably more rapid. Rates of latter proportions is due to the fact that the largest
natural increase would be scarcely affected. decreases are possible in the specific death rates in infancy
and early childhood, and at relatively advanced ages:
Variant ZV: Transfers involving more at intermediate ages, death risks are already so low that
males and fewer females mortality reductions no longer result in any large increase
in the number of survivors.
418. In this variant, the unrealistic assumption is 424. As can be inferred from table 58, while birth
made that, beginning in 1970, the rates of transfer for rates decline (as assumed in both projections), death
males will be those calculated for females in the preced- rates in both the urban and rural population under
ing period, and those for females will be those previously constant mortality would have a tendency to rise because
calculated for males. the decline in fertility modifies the age structure.
419. As shown in table 55, the effects of this varied
assumption on population growth will be quite slight. Variant VZ: Constant fertility
Since more females remain in rural areas, where fertility
is higher, the rural population will grow somewhat more 425. It is not probable that fertility levels as high as
rapidly, and the urban population somewhat more slowly, they were estimated for the 1960-1970 period will long
than in the standard projection. By the end of the cen- continue in Brazil. Urban fertility has already declined
tury, the proportion of children will be slightly higher in far below the rural level, and the general level of fertility
rural areas and slightly lower in urban areas. The sex has recently been decreasing at least in Brazil's south.
ratios in the two sets of areas would, however, be signif- As has been the experience of other countries in their
icantly altered. Beginning about 1985, a male surplus process of economic and social development, it is natu-
would develop in urban areas and a female surplus in rally to be expected that within the coming thirty years
rural areas, reversing the situation observed in 1970. fertility in Brazil will undergo a substantial decline,
420. As appears in table 56, rates of growth and natural both in urban and in rural areas. One particular form
increase would not be significantly modified in the urban of this assumption has been incorporated in the standard
population, but would be noticeably higher in the smaller projection, implying a linear decline in urban fertility
population of rural areas. and a first gradual but then accelerating decline in rural
fertility. But the future course of events cannot be
Variant V : Constant mortality estimated with such exactitude, and many alternative
developments are entirely possible. The comparison of
421. For the purpose of this variant, it has been variant VI with the standard projection permits an
assumed that both urban and rural mortality will remain appreciation of the amount of population changes due
at the levels estimated for 1970, typified by expectations to the particular fertility assumption of the standard.
of life at birth of 64.5 years in urban areas and 59.0 years The assumption in variant VI is that the gross reproduc-
in rural areas. An abrupt halt of progress in public tion rate (GRR) in urban areas will remain at the level
health is, of course, not a realistic assumption, though it of 2.0, and in the rural areas at 3.5. As can be seen in
is reasonable to foresee the possibility that actual future table 59, a very large reduction of population growth is
progress may be somewhat slower or even somewhat implied in the assumptions of the standard projection.
more rapid than assumed in the standard projection. If fertility were to remain at the pre-1970 levels, the natio-
The purpose of this variant is to permit an inference of nal population would grow to 234 million by the year 2000,
the possible effects of different mortality trends on the whereas in the standard the total population then attains
resulting population. 193 million, which is 41 million less. As compared with
422. As of 1970, mortality in Brazil is estimated at constant fertility, the standard projection results by
such a low level that the gain in additional lives saved 2000 in an urban population that is 27 million smaller
by further progress in health will no longer be exceedingly and a rural population about 14 million smaller. The
large. As can be seen in table 57, the mortality decline level ot urbanization, however, advances with almost the
assumed in the standard projection will result in an same speed in both projections.
additional population of 13 million inhabitants in 2000, 426. Also shown in table 59, changes in age composi-
when the national population may total 193 million, tion in both urban and rural areas which might occur from
instead of the 180 million in the event of constant mor- 1970 to 2000 are much more pronounced in the standard
tality. As compared with variant V, the standard pro- projectioil (declining fertility) than in variant VI. Sub-
jection results in an eventual urban population that is stantially smaller proportions of the population will be
9 million larger, and a rural population that is larger by children and appreciably larger proportions will be in
4 million. The eventual level of urbanization, however, advanced ages if fertility declines than if it remains
would be virtually the same. constant. (Text continues on p. 115.)
 55. VARIANTIV
TABLE AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000;
COMPARISON OF SELECTED RESULTS

Standard projection Variant IV

Base data - .
1970 1980 1990 2Om 1980 1990 2000
. - -
A

Population (thousands)
Total population .............. 94. 360 124.411 158.686 193. 017 124. 614 158. 563 193.269
Urban population .......... 52.458 77. 791 108. 891 143.044 77. 830 107. 965 141. 673
Rural population ........... 41. 902 46. 620 49. 795 48. 973 46. 784 50. 598 51. 596
Percentage urban ........... 55.6 62.5 68.6 74.1 62.5 68.1 73.3
Percentage age composition of the urban population (both sexes)
Allages ..................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
04 ..................... 13.7 12.8 11.5 9.9 12.7 11.2 9.7
5-14 ..................... 25.3 23.5 22.3 20.1 23.5 22.2 19.9
15-24 ..................... 20.2 20.6 19.7 19.3 20.6 19.9 19.4
25-44 ..................... 24.7 26.0 28.5 29.7 26.1 28.4 30.0
45-64 ..................... 12.3 13.0 13.2 15.3 13.0 13.4 15.2
65 and over ................ 3.7 4.1 4.9 5.8 4.1 4.9 5.8
Percentage age composition of the rural population (both sexes)
Allages ..................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
04 ..................... 18.7 18.8 16.4 13.1 19.0 17.0 13.8
5-14 ..................... 27.9 28.3 28.0 25.3 28.2 28.2 26.0
15-24 ..................... 19.1 18.4 19.6 20.7 18.3 19.3 20.6
25-44 ..................... 20.9 21.1 22.1 24.7 21 .0 21.7 24.0
45-64 ..................... 10.4 10.3 10.3 11.9 10.2 10.2 11.5
65 and over ................ 3.0 3.2 3.6 4.2 3.2 3.6 4.1
Males per 100 females in the population (all ages)
Total population .............. 98.9 99.3 99.8 100.3 99.5 100.5 100.8
Urban population .......... 93.8 95.1 96.5 97.5 98.3 101.3 102.2
Rural population ........... 105.6 106.7 107.6 108.8 101.4 98.8 97.1

TABLE56 . VARIANTIV AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION
OF BRAZIL.1970-2000 . COMPARISONOF RATES OF GROWTH AND THEIR COMPONENTS

Standard projection Voriant IV

Rate 1970-1980 1980-1990 1990-2000 1970-1980 1980-1990 1990-2000

Total population
Growth ................. 27.6 24.3 19.6
Natural increase .......... 27.6 24.3 19.6
Births ................. 35.1 29.8 24.2
Deaths ................ 7.5 5.5 4.6
Net transfer .............. 0.0 0.0 0.0
Urban popuIation
Growth ................. 39.3
Natural increase .......... 22.3
Births ................. 28.7
Deaths ................ 6.4
Net transfer .............. 17.0
Rural population
Growth ................. 10.7 6.6 0.4
Natural increase .......... 35.3 31.4 25.2
Births ................. 44.3 37.6 29.8
Deaths ................ 9.0 6.2 4.6
Net transfer .............. -24.6 -24.8 -24.8
 . 57. VARIANT
TABLE V AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000;
COMPARISON OF SELECTED RESULTS

Standard projection Variant V

Base data
1970 1980 1990 2000 1980 1990 2000

Population (thousands)
Total population .............. 94. 360 124. 411 158. 686 193. 017 123. 108
Urban population ..........
Rural population ...........
Percentage urban ...........
.
52. 458
41 902
55.6
77. 791
46. 620
62.5
108. 891
49. 795
68.6
143.044
49. 973
74.1
77. 119
45. 989
62.6
Percentage age composition of the urban population (both sexes)
All ages ..................... 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 13.7 12.8 11.5 9.9 12.6
5-14 ..................... 25.3 23.5 22.3 20.1 23.5
15-24 ..................... 20.2 20.6 19.7 19.3 20.7
25-44 ..................... 24.7 26.0 28.5 29.7 26.3
45-64 ..................... 12.3 13.0 13.2 15.3 13.0
65 and over ................ 3.7 4.1 4.9 5.8 4.0
Percentage age composition of the rural population (both sexes)
All ages ..................
0-4 ..................
5-14 ..................
15-24 ..................
25-44 .....................
45-64 .....................
65 and over ................
Males per 100 females in the population (all ages)
Total population .............. 98.9 99.3 99.8 100.3 99.4
Urban population .......... 93.8 95.1 96.5 97.5 95.3
Rural population ........... 105.6 106.7 107.6 108.8 106.6

Standard projection Variant V

.
.
..
.
.
Rate 1970-1980 1980-1990 1990-2000 1970-1980 1980-1990 1990-2000

Total population
Growth ................. 27.6 24.3 19.6 26.6 21.7 16.3
Natural increase .......... 27.6 24.3 19.6 26.6 21.7 16.3
Births ................. 35.1 29.8 24.2 35.2 30.1 24.7
Deaths ................ 7.5 5.5 4.6 8.6 8.4 8.4
Net transfer .............. 0.0 0.0 0.0 0.0 0.0 0.0
Urban population
Growth ................. 39.3 33.6 27.3 38.4 31.2 24.2
Natural increase .......... 22.3 20.6 17.3 21.5 18.4 14.5
Births ................. 28.7 25.7 21.9 28.8 26.0 22.3
Deaths ................ 6.4 5.1 4.6 7.3 7.5 7.8
Net transfer .............. 17.0 13.0 10.0 16.9 12.8 9.7
Rural population
Growth ................. 10.7 6.6 0.4 9.3 3.3 -3.8
Natural increase .......... 35.3 31.4 25.2 33.9 28.0 20.9
Births ................. 44.3 37.6 29.8 44.5 38.1 30.7
Deaths ................ 9.0 6.2 4.6 10.5 10.1 9.8
Net transfer .............. - 24.6 -24.8 - 24.8 -24.6 -24.7 -24.7
 TABLE59 . VARIANTVI AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000;
COMPARISON OF SELECTED RESULTS

Standard projection Variant VI

Base data
1970 1980 1990 2000 1980 1990 200

Population (thousands)
Total population .............. 94. 360 124.411 158.686 193.017 127. 788 173.486 234. 231
Urban population .......... 52. 458 77. 791 108. 891 143. 044 79. 863 118. 339 170. 030
Rural population ........... 41. 902 46. 620 49. 795 49. 973 47. 925 55. 147 64. 201
Percentage urban ........... 55.6 62.5 68.6 74.1 62.5 68.2 72.6
Percentage age composition of the urban population (both sexes)
All ages ..................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 13.7 12.8 11.5 9.9 14.3 14.2 13.8
5-14 ..................... 25.3 23.5 22.3 20.1 23.6 24.3 23.8
15-24 ..................... 20.2 20.6 19.7 19.3 20.0 18.7 19.2
25-44 ..................... 24.7 26.0 28.5 29.7 25.4 26.2 25.4
45-64 ..................... 12.3 13.0 13.2 15.3 12.6 12.1 12.9
65 and over ................ 3.7 4.1 4.9 5.8 4.0 4.5 4.8
Percentage age composition of the rural population (both sexes)
All ages ..................... 100.0 100.0 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 18.7 18.8 16.4 13.1 20.3 19.9 20.2
5-14 ..................... 27.9 28.3 28.0 25.3 28.2 29.3 29.0
15-24 ..................... 19.1 18.4 19.6 20.7 17.9 18.2 18.8
2 5 4 ..................... 20.9 21.1 22.1 24.7 20.5 20.0 19.6
45-64 ..................... 10.4 10.3 10.3 11.9 10.0 9.3 9.2
65 and over ................ 3.0 3.2 3.6 4.2 3.1 3.3 3.3
Males per 100 females in the population (all ages)
Total population .............. 98.9 99.3 99.8 100.3 99.4 100.2 .
101 0
Urban population .......... 93.8 95.1 96.5 97.5 95.4 97.0 98.4
Rural population ........... 105.6 106.7 107.6 108.8 106.6 107.4 108.3

Standard projection Variant VI

Rote 1970-1980 1980-1990 1990-200 1970-1980 1980-1990 1990-200

Total population
Growth ................. 27.6 24.3 19.6
Natural increase .......... 27.6 24.3 19.6
Births ................ 35.1 29.8 24.2
Deaths ................ 7.5 5.5 4.6
Net transfer .............. 0.0 0.0 0.0
Urban population
Growth ................. 39.3 33.6 27.3
Natural increase .......... 22.3 20.6 17.3
Births ................. 28.7 25.7 21.9
Deaths ................ 6.4 5.1 4.6
Net transfer .............. 17.0 13.0 10.0
Rural population
Growth ................. 10.7
Natural increase .......... 35.3
Births ................. 44.3
Deaths ................ 9.0
Net transfer ............... 24.6
 TABLE61 . VARIANT
VII AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION OF BRAZIL.1970-2000;
COMPARISON OF SELECTED RESULTS

Standard projection Variant VII

Base data
1970 I980 1990 2000 1980

Population (thousands)
Total population ............ 94.360 124. 411 158. 686 193.017 123. 270
Urban population ........ 52. 458 77. 791 108.891 143.044 75. 874
Rural population ......... 41. 902 46. 620 49. 795 49. 973 47. 396
Percentage urban ......... 55.6 62.5 68.6 74.1 61.6
Percentage age composition of the urban population (both sexes)
AN ages ..................... 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 13.7 12.8 11.5 9.9 11.3
5-14 ..................... 25.3 23.5 22.3 20.1 23.4
15-24 ..................... 20.2 20.6 19.7 19.3 21.1
25-44 ..................... 24.7 26.0 28.5 29.7 26.7
45-64 ..................... 12.3 13.0 13.2 15.3 13.3
65 and over ................ 3.7 4.1 4.9 5.8 4.2
Percentage age composition of the rural population (both sexes)
AN ages ..................... 100.0 100.0 100.0 100.0 100.0
0-4 ..................... 18.7 18.8 16.4 13.1 19.6
5-24 ..................... 27.9 28.3 28.0 25.3 28.3
15-24 ..................... 19.1 18.4 19.6 20.7 18.1
25-44 ..................... 20.9 21.1 22.1 24.7 20.7
45-64 ..................... 10.4 10.3 10.3 11.9 10.1
65 and over ................ 3 .0 3.2 3.6 4.2 3.1
Males per 100 females in the population (all ages)
Total population ........... 98.9 99.3 99.8 100.3 99.3
Urban population ....... 93.8 95.1 96.5 97.5 94.9
Rural population ........ 105.6 106.7 107.6 108.8 106.6

TABLE62. VARIANTVII AND STANDARD PROJECTION OF THE URBAN AND RURAL POPULATION
OF BRAWL.1970-2000; COMPARISON OF RATES OF GROWTH AND THEm COMPONENTS

Standard projection Variant VZI

Rate 1970-1980 1980-1990 1990-2000 1970-1980 1980-1990 1990-2000

Total population
Growth ................. 27.6 24.3 19.6
Natural increase .......... 27.6 24.3 19.6
Births ................. 35.1 29.8 24.2
Deaths ................ 7.5 5.5 4.6
Net transfer .............. 0.0 0.0 0.0
Urban population
Growth ................. 39.3 33.6 27.3
Natural increase .......... 22.3 20.6 17.3
Births ................. 28.7 25.7 21.9
Deaths ................ 6.4 5.1 4.6
Net transfer .............. 17.0 13.0 10.0
Rural population
Growth ................. 10.7 6.6 0.4
Natural increase .......... 35.3 31.4 25.2
Births ................. 44.3 37.6 29.8
Deaths ................ 9.0 6.2 4.6
Net transfer .............. - 24.6 .-24 .8 '-24.8
 Standardprojecrion Variant VII Variant V I l l
Period Urban Rural Urban Rural Urban Rural

427. As can be seen in table 60, the assumption of to whether such changes could actually occur or, if they
constant fertility (but declining mortality) has the effect did, whether they would be the result of specific economic,
of gradually rising crude birth rates, both in urban and social or administrative programmes. The comparison
rural areas, because of modifications in the population's is of methodological interest, but does not constitute a
age structure. scenario. If the alternative results of a scenario were to
be surveyed, it would be necessary to select one or a few
Variants VZZ and VZZI: Varied assumptions of fertility characteristic findings and compare the results of all
decline in both urban and rural areas projections in those particular respects. For the sake
of illustration, a few over-all comparisons of the present
428. Whereas it is most probable that fertility will projections are shown in tables 65 and 66. The eight
decline in both the urban and rural areas of Brazil, it variants have been separated into two groups, since
remains realistic to assume that the decline may be either variants I to IV differ in respect of assumed population
slow or rapid, and that the tempo of decline may differ transfers, whereas variants V to VIII differ in assumed
between urban and rural areas. While in all the preced- mortality and fertility trends.
ing variants, I to VI, there is some degree of unrealism,
variants VII and VIII, showing alternative assumptions
for future fertility trends, may be regarded as somewhat TABLE65. STANDARD
PROJECTION, AND EIGHT VARIANTS, OF THE
realistic. For it is to be admitted that, among all the URBAN AND RURAL POPULATION OF BRAZIL, 1970-2000; INHABI-
assumptions, that of the future fertility trend is still TANTS IN 2000 PER 100 INHABITANTS IN 1970, IN THE TOTAL, URBAN
affected by great uncertainties. AND RURAL POPULATION

429. As compared with the standard projection,

Total Urban Rural
variant VII assumes a more rapid decline in urban fertil- Projection population population population
ity and a less rapid decline in rural fertility, whereas in
variant VIII the opposite combination is made. In Standard .......... 205
particular, the assumptions are as shown in the table Variant I .......... 215
above, in terms of the gross reproduction rate (GRR). Variant I1 .......... 21 1
430. Since the urban population is of greater and Variant 111 ........ 206
increasing weight in the national total, the trend in urban Variant IV ........ 205
fertility has more effect on national population growth Variant V .......... 191
than does the trend in rural fertility. For this reason, Variant VI ........ 248
variant VII (rapidly declining urban fertility, slowly Variant VII ........ 195
declining rural fertility) results in a smaller national total Variant VIII ........ 210
population, and variant VIII (with the opposite assump-
tions) in a larger national total, than does the standard
projection (see tables 61 and 63). As can be expected,
the level of urbanization rises less rapidly under variant VII 432. As appears in table 65, the growth in total popu-
and more rapidly under variant VIII. Age structures are lation, from 1970 to 2000, varies rather little, although
considerably affected by the assumed fertility trends. the rural-to-urban transfers assumed may vary greatly.
Tables 62 and 64 permit comparisons of the corresponding On the other hand, constant mortality (variant V) and
growth rates, reflecting for instance the different speeds especially constant fertility (variant VI) have a large
with which crude birth rates might decline. effect on total population growth. The growth in urban
and rural population depends considerably on assumed
Relative effects of different population transfers (none assumed in variant I) or at
demographic assumptions least on assumed reclassification of areas (none assumed
in variant 11), but little on modified assumptions con-
431. As was pointed out at the beginning of this cerning transfers. Constant mortality and especially
section, the several variants represent the effects of arbi- constant fertility have large effects on the growth of both
trarily selected demographic assumptions without regard urban and rural populations.
 433. An element of age structure is reviewed in table 66. less than 25 years). The effect of varied mortality on
It will be noted that varied assumptions concerning popu- age structure (variant V) is also rather slight. Both the
lation transfers (variants I to IV) have surprisingly little urban and rural age structures, however, are greatly
effect on resulting urban and rural age structures (as influenced by the particular assumptions made in respect
measured here by the proportion of the population aged of fertility (variants VI to VIII).

TABLE66. STANDARD
PROJECTION, AND EIGHT VARIANTS, OF THE URBAN AND RURAL POPULATION OF BRAZIL,1970-2000;
COMPARISON OF PERCENTAGES OF POPULATION AGED UNDER 25 YEARS

Urban population Rural population

Projection 1970 1980 1990 2000 1970 1980 1990 7000

Standard .......................... 60.2 56.9 53.5 49.3

Variant I .......................... 60.2 55.7 51.8 47.5
Variant I1 ......................... 60.2 57.0 53.9 49.9
Variant 111 ........................ 60.2 56.7 53.2 49.1
Variant IV ......................... 60.2 56.8 53.3 49.0
Variant V ......................... 60.2 56.8 53.1 48.7
Variant VI ......................... 60.2 57.9 57.2 56.8
Variant VII ........................ 60.2 55.8 49.5 43.8
Variant VIII ....................... 60.2 57.3 54.9 52.0
 ANNEX I

Ut looe"
Table of logistic curve 100 - = for values of lOOdt from -480 to 480
Tt 1 edt +
Note: The logistic transformation column is equivalent to an abstract "table year" of time at
which the correspondingpercentageurban would beexpected to occur. The value of lOOdt is defined
as the table year (assuming the year zero at the point where 50 per cent of the population is urban,
and also that the difference between the exponential rates of urban and rural growth equals one per
cent a year).
Ut
The value of 100 -
Tt
is defined as the percentage urban in the total population at the given

moment of time, and presented under the heading "percentage urban". For more explanations,
see the penultimate section in chapter I11 and the penultimate section in chapter V.

Table year Percentage urban 1 Tabk year Percentage urban 1 Table year Percentage urban I Table year Percentage urban
 ANNEXI (continued)

Table year Percentage urban Table year Percentage urban I Table year Percentage urban Table year Percentage urban

- 340 3.2295 9.1123 - 175 14.8047

- 339 3.2609 9.1955 - 174 14.9313
- 338 3.2926 9.2793 - 173 15.0588
- 337 3.3246 9.3638 - 172 15.1871
- 336 3.3569 9.4490 -171 15.3164
 ANNEXI (continued)

Table year Percentage urban Table year Percentage urban 1 Table year Percentage urban I Table yeat Percentage urban

61.0639
61.3014
61.5384
61 .7748
62.0106
 ANNEXI (continued)

Table year Percentage urban Table year Percentage urban 1 Table year Percentage urban Table year Percentage urban
-
93.401 1
93.4625
93.5233
93.5836
93.6434
 ANNEXI (concluded)

Table year Percentage urban Table year Percentage urban I Table year Percentage urban I Table year Percentage urban
 ANNEX 11

Table of survival ratios (P,) of model life tables for five-year age groups
and five-year intervals of time

Note: This and related tables were originally published in Manual 111: Methods for Population
Projections by Sex and Age to serve as estimates of mortality conditions applicable in population
projections calculated by the cohort-component method. In particular, an annual gain of 0.5 years
in expectation of life at birth was assumed wherever the expectation is less than 55 years. When
an expectation of 55 years is attained, the model assumption implies a slight acceleration in gains
until the expectation approaches 65 years. Thereupon, the rate of gain slows down and becomes
slight when the expectation has risen substantially higher than 70 years. These assumptions were
based on observed world-wide averages at about mid century.
The table is arranged by five-year intervals of time. Survival ratios for other time intervals
can be calculated from the table by interpolation. Interpolations may also be made to serve in
assumptions of either a slower or a more rapid decline in mortality.

a United Nations publication, Sales No. 56.XIII.3.

 Sex and age
(percentage) in years
Level 0
foe,, =
20)
Level 5
foe,,
22.5)
= (OC
2-71
.
Level I0
=
Level 15
(0e0 =
27.5)
I.evcl20
foeo =
30)
Level 25
("eo =
32.5)
Level 30
fOco=
35)
Level 35
(oro =
37.5
.
Level 40
(oe =
40)
Level 45
foeo =
42.5)
Level 50
foeo =
45)
Level 55
(%a =
47.5)
- --
Males
(Births) b .......... (0.6092) (0.6334) (0.6622) (0.6853) (0.7058) (0.7248) (0.7440) (0.7626) (0.7789) (0.7950) (0.8107) (0.8256)

Females
(Births) b ............ (0.6290) (0.6558) (0.6801) (0.7024) (0.7242) (0.7447) (0.7627) (0.7802) (0.7969) (0.8135) (0.8292) (0.8443)

.
Manual 111: Methods for Population Projections by Sex and Age (United Nations publication. Sales No 56.XIII.3), pp . 80.81
SOURCE: .
a Equivalent values of Oe. shown in parentheses refer to expectation of life at birth. for both sexes. in years.
A ratio of survival from births occurring in the course of a five-year period to age group 0 to 4 by the end of the period .
Ratio of survival from terminal age group 80 and over to terminal age group 85 and over.
FIST-YEAR INTERVALS OF TIME; MORTALITY LEVEL (OR TIME REFERENCE IN YEARS) a

Level 60
(Oe, =
50)
Level 65
(Oe, =
52.5)
Level 70
(Oe, =
55)
Level 75
(aeo =
57.6)
Level 80
(Oeo =
60.4)
Level 85
(Oe0 =
63.2)
.
Level 90
(oe =
65.8)
Level 95
Oeo =
68.2)
Levcl 100
foea =
70.2)
Level 105
(Oeo =
71.7)
Level 110
fOeo=
73.0)
Levcl 115
leeo =
73.9)
Sex and age
(percentage) in years)

Males
(0.8406) (0.8557) (0.8703) (0.8877) (0.9070) (0.9262) (0.9438) (0.9580) (0.9678) (0.9744) (0.9788) (0.9817)............... .(Births)

Females
(0.8594) (0.8739) (0.8882) (0.9036) (0.9208) (0.9380) (0.9535) (0.9660) (0.9744) (0.9801) (0.9838) (0.9864)............... .(Births)
 &SOWTO OBTAIN UNITED NATIONS PUBLICATIONS
Wbd Hatiam. p u b l i u w may be obtained from beokstoracl m d diatributom
-kt t h e - d d dCauult your bodutore or write to: United Nations, Snlu
Ssctli#l, N e w Y & 4 i C t f G ~ .

e-.SE PBOCUBER PUBLICATIONSDES NATIONS UMES

pllblicrt&xs &as Nab Unies m t en venta dam les librairiea et l a agencas
d$pp.itains du monde entiFr. Inform~z-vowauprbs do votre libmire ou ad--vouu
1 :Nations U W ,Sectiondea ventea, New York ou Gonave.

' S b g a ~ x xO ~ ~ E H S 06-bcsnnenx1,ac
~ ~ U R Bnuuii x o l ~ ~rynma
o s xxnxshxx nur-
amax x ao ECEX pattoxiax xnpa. EIaso~mecnprmxa 06 XI3AUIPrnX m
namem ~~RNXUOY Y ( L P I I J ~unx
~ ~ nnnlunre
~ no wpecy: Opmnmakien Obaen~tnrrtaux
'EIrcmB,Cexqxs no nponaxe raa~rurnfi,Xblo-lilopx unx Xeadaa.

COMO CONSEGUIE PUBLICACEONES DE LAS NACIONES UN1[DAS

U m publicrcianss ds las N ~ c h c Urnidas
s eatdn en venta an lihrerh y c u m dirtri-
buidom en tadu pa- dsl mundo. C a k l ( e a au librsro o did* 8: N.ciooso
Unidu, Seeci6a ds Vanha, Nwva Yo* o Gin&=

Printad in Wgium Price: $U.S. 7.00 Waited N W puMicrtfon

74-OS99LDecembcs 1974--5m (or eq-t in ot+ cumndcs) Srks No.E.74.XIII.3