(A. P. Thirlwall (Auth.) ) Growth and Development (B-Ok PDF

GROWTH AND DEVELOPMENT
GROWTH AND
DEVELOPMENT
with special reference to
developing economies
A. P. THIRLWALL
Reader in Economics, University of Kent at Canterbury
Macmillan Education
ISBN 978-0-333-12207-5 ISBN 978-1-349-15472-2 (eBook)
DOI 10.1007/978-1-349-15472-2
© A. P. Thirlwall 1972
Reprint of the original edition 1972
All rights reserved. No part of this publication
may be reproduced or transmitted, in any form
or by any means, without permission.
First published 1972 by

THE MACMILLAN PRESS LTD
London and Basingstoke
Associated companies in New York Toronto
Dublin Melbourne Johannesburg and Madras
SBN 333 12206 2 (hard cover)
333 12207 0 (paper cover)
The paperback edition of this book is sold subject to the condition

that it shall not, by way of trade or otherwise, be lent, re-sold,
hired out, or otherwise circulated without the publisher's prior
consent in any form of binding or cover other than that in which
it is published and without a similar condition including this
condition being imposed on the subsequent purchaser.
TO GIANNA
FOR LORENZO
Contents
PREFACE XI
1 DEVELOPMENT AND UNDERDEVELOPMENT
Current Interest in Development Economics

The World Distribution of Income 6
The Development Gap 9
Per Capita Income as an Index of Development 19
The Measurement and Comparability of Per Capita Incomes 24
The Stages of Development 27
Rostow' s Stages of Growth 31
2 THE PRODUCTION FuNCTION APPROACH TO THE

STUDY OF THE CAUSES OF GROWTH 38
The Anafysis of Growth 38
The Production Function 40
The Cobb-Douglas Production Function 43
Embodied Technical Progress 51
Improvements in the Quality of Labour 55
Resource Shifts 56
Empirical Evidence 57
Appendix 2.1: Denison on' Why Growth Rates Differ' 61
3 LAND, LABOUR AND AGRICULTURE 72

The Role of Agriculture in Development 73
The Growth of the Money Economy 76
Economic Development with Unlimited Supplies of Labour 80
Disguised Unemployment 87
Incentives and the Costs of Labour Tranifers 94
4 CAPITAL AND TECHNICAL PROGRESS 97

The Role of Capital in Development 97
The Capital-Output Ratio 100
viii CONTENTS
Technical Progress 105

Capital- and Labour-saving Technical Progress 105
How Societies Progress Technologically 109
Learning 111
Education 113
5 THE PERSISTENCE OF UNDERDEVELOPMENT 117

Dualism 118
The Process of Cumulative Causation 120
The Population 'Problem' 128
Facts about World Population 133
The ' Optimum' Population 136
A Model of the Low-level Equilibrium Trap 139
The Critical Minimum Effort Thesis 145
6 THE ALLOCATION OF RESOURCES 148

The Broad Policy Choices 148
Industry versus Agriculture 150
The Comparative Cost Doctrine 150
Labour-intensive versus Capital-intensive Technology 151
Balanced versus Unbalanced Growth 153
Unbalanced Growth 158
Investment Criteria 166
The Minimum Capital-Output Ratio Criterion 168
The Social Product Criterion 170
The Marginal Per Capita Reinvestment Quotient Criterion 172
The Labour Absorption Criterion 175
Balance of Payments Considerations 176
The Social Welfare Function 178
7 PLANNING ECONOMIC DEVELOPMENT 180

Arguments for and against Planning 180
Development Plans 183
Policy Models 185
Projection Models 188
Appendix 7. r: Input-Output Analysis 189
Appendix 7.2: The Programming Approach to Development 201
Linear Programming 203
Appendix 7.3: The Planning Wage 212
CONTENTS ix
8 FINANCING DEVELOPMENT 224
A. DOMESTIC RESOURCES FOR INVESTMENT 226
Voluntary Saving 226
Involuntary Saving 228
Taxation 229
Deficit Finance 233
Inflation 236
The Causes if Inflation in Less Developed Countries 240
B. FOREIGN RESOURCES FOR INVESTMENT 244

Dual-gap Anarysis and Foreign Borrowing 244
The Savings-Investment Gap 247
The Export-Import, or Foreign Exchange, Gap 248
Two Gaps in Practice 250
Foreign Borrowing and the Debt-servicing Problem 252
International Assistance 259
Official Bilateral and Multilateral Assistance 262
Private Foreign Investment 263
Criteria for the Allocation if Assistance 265
Technical Assistance 269
Tied Assistance 269
British Assistance to Less Developed Countries 272
Recent Trends in Savings and Investment in Less Developed
Countries 275
9 TRADE AND DEVELOPMENT 277

The Gains from Trade 280
New Trade Theories for Developing Countries: The Prebisch
Doctrine 287
Technical Progress and the Terms if Trade 288
The Income Elasticity if Demandfor Products and the Balance
if Payments 289
Recent Trends in the Terms if Trade 292
Trade Theory and Dual-gap Anarysis 293
Trade Policies 295
Trade Priferences and Effective Protection 297
International Commodity Agreements 300
Buffer Stock Schemes 302
X CONTENTS
Restriction Schemes 302

Price Compensation Agreements 303
Income Compensation Schemes 305
REFERENCES AND FuRTHER READING 307
INDEX 317
Preface
This book is offered as an introduction to certain topics in develop-
ment economics, with particular emphasis on the economic obstacles
to development and the economic means by which less developed
countries may raise their rate of growth of output and living
standards. This does not mean that it provides a recipe or blueprint
for development in the present less developed countries -far from it.
There can be no general recipes of this nature, and even if there were
there would have to be more than economic ingredients. It is
possible, however, to highlight certain fundamental principles and
economic truths common to all countries which have set themselves
the objective of development, and this is the purpose here. The text
is primarily theoretical, and the discussion ranges from the mundane
to the technical. This may be offputting, but is virtually inevitable
with a subject like development which draws on all branches of
economics, and which applies a large chunk of economic analysis
to one particular country or set of countries. Certain obvious things
need saying about the development process, while the analysis of
particular development difficulties and their solution can become
quite complex. The most technical aspects of the book on planning
techniques are relegated to appendices in Chapter 7.
Despite its theoretical emphasis, no more than an elementary
knowledge of economics is required, such as a student may receive
in a first-year economics principles course. Indeed, the course on
which this book is built was originally designed to provide a balanced
diet for students studying the sociology and politics of development
with no more than one year's elementary exposure to macro- and
micro-economic theory. Mathematics is kept to a minimum and
there is no recourse to sophisticated growth models.
Some readers may be worried by the neglect of non-economic
factors in the growth process. The emphasis on the economics of
development is not to deny or demote the importance of non-
economic factors; rather it is a reflection of my interest and com-
petence. Having said this, however, I think the importance of
xii PREFACE
institutional barriers to development is often exaggerated. As
Maddison 1 and others have remarked, the desire for material
improvement in less developed countries is very strong - certainly
strong enough to counter any institutional barriers that may exist-
and the evidence is not very convincing that basic institutional
reform is a necessary precondition of accelerated growth. But, in
any case, it is my firm belief that the economist has something
positive to offer by way of analysis unadulterated by sociological,
political and other non-economic variables. It is a favourite pastime
of other social scientists to criticise economists for ignoring institu-
tional considerations. We now seem to me to be in the opposite
danger of a great deal of woolly thinking about what raising living
standards is all about. In the final analysis, growth must be con-
sidered as an economic process in the important practical sense
that it is unlikely to proceed very far in the absence of an increase
in the quantity and quality of the resources available for production.
Even if we concede that the availability of resources may depend
on non-economic factors such as attitudes towards effort, saving,
risk-taking, and 'maximising' in general, this does not make non-
economic factors prime determinants of development. Their impact
is indirect. This is how economists tend to view the relation between
development and the institutional environment, and this will be the
implicit assumption throughout the present volume.
In preparing this book I have been haunted by that celebrated
review which began: 'There is much in this book which is new
and true; unfortunately that which is new is not true and that
which is true is not new!' There is nothing in this present volume
which is new either of a theoretical or an empirical nature. I hope,
however, that what is not new does not deviate too radically from
'truth' or relevance, and that what is 'true' provides a useful
synthesis for an introduction to a course in development
economics.
Finally, I should make the point that in studying development and
the economic difficulties ofless developed countries one would have
to be extremely insensitive not to form fairly strong views on the
way development problems ought to be tackled. Economists, except
perhaps those who return from less developed countries and become
anecdotal, have tended in the past to be a trifle detached and rather
1 A. Maddison, Economic Progress and Policy in Developing Countries (London:
Allen & Unwin, 1970).

PREFACE xiii
reticent in expressing their views in this controversial area. While
on the whole this book follows the positive tradition, there are
occasions where there is resort to value judgements.
Since this book has arisen out of a series oflectures given to third-
year undergraduates at the University of Kent, I should like to
express a debt of gratitude to at least three cohorts of students who
endured the role of guinea pig so patiently. I am also very grateful
to my colleague Roger Hill who read the entire manuscript in draft
and pointed out many slippages and obscurities·. But I cannot
implicate him in my views or analysis, because I know that he
disagrees with a good deal of the approach I have taken - especially
the presentation of theoretical skeletons without much empirical
flesh, and my neglect of non-economic factors. None the less, I have
valued his discussion enormously. Finally, I must thank the editors
of the National Westminster Bank Review, Moneta e Credito and the
Scottish Journal cif Political Economy for allowing me to reproduce
material previously published in their journals, and Miss Margaret
Hawkins for typing the entire manuscript with her usual cheerful
efficiency.
A.P. T.
Keynes College,
University cif Kent at Canterbury,
1971
The First Law of Development
'For·unto everyone that hath shall be given
and he shall have abundance, but from
him that hath not shall be taken away even
that which he hath.'
(Matthew 25: 29)
I
Development
and Underdevelopment
Current Interest in Development Economics

Current academic interest in development economics is a relatively
recent phenomenon. For the student today it must be difficult to
realise that twenty years ago a course in development economics
was a rare feature of an undergraduate programme in economics,
and that textbooks on development were few and far between.
Similarly, active public concern with the poorer nations of the world
is of equally recent origin. The majority of the national and inter-
national agencies to promote development that we are familiar with
today have been established since the Second World War. Before
the war the poor countries of the world were categorised as un-
developed and relatively neglected. Today the situation is very
different. The blunt title of' undeveloped' has been dropped from
use, to be replaced by a string of euphemistic labels ranging from
'underdeveloped' to 'emerging', expressing faith in the develop-
ment potential of these countries. The development of the so-called
Third World is now regarded as one of the greatest social and
economic challenges facing mankind. What accounts for this ap-
parent sudden change in interest and attitude? Three major stimuli
can be pin-pointed. Firstly, there has been a renewed interest
among professional economists in the process ofgrowth and the theory
of planning. Secondly, the poor countries have become increasingly
aware of their own backwardness, which has led to a natural desire
for rapid development. Thirdly, with the growth of nationalism in
the world, and the cold war, the developed countries have shown a
growing political interest in poor and ideologically uncommitted
nations.
2 GROWTH AND DEVELOPMENT
Academic interest in the mechanics of growth and development

is a renewed interest rather than a new preoccupation of economists.
The progress and material well-being of men and nations has
traditionally been at the centre of economic writing. It constituted
one of the major areas of interest of the classical economists. Smith,
Ricardo, Malthus, Mill and Marx all dealt at some length (with
divergent opinions on many issues) with the causes and conse-
quences of economic advance. It is entirely natural that thinkers of
the day should comment on the contemporary scene, and there is
perhaps an analogy here between the preoccupation of the classical
economists at the time of Britain's industrial revolution and the
concern of many economists today with the economics of develop-
ment and world poverty, which has been brought to the world's
attention so dramatically in recent decades. It is often said that we
owe modern growth theory to the stagnatory state in which Western
nations found themselves between the wars. Whether or not this is
so, the challenge of development represents an equivalent challenge
to economists today to that of depression and mass unemployment
in the thirties. Advances in growth theory, coupled with more
detailed knowledge of the sources of growth, and the refinement of
techniques for planning and resource allocation, have all increased
the possibility of more rapid economic progress than hitherto.
Certain theoretical models and techniques have been used exten-
sively in some countries. For example, models for calculating in-
vestment requirements to achieve a target rate of growth invariably
form an integral part of a development plan, and in some countries
there have been experiments in recent years with such techniques
as input-output analysis and linear programming.
The question is often posed as to what lessons, if any, the present
less developed countries can draw from the first-hand observations
of the classical writers, or more directly from the development
experience of the present advanced nations. One obvious lesson is
that while development can be regarded as a natural phenomenon,
it is also a lengthy process, at least left to itself. It is easy to forget that
it took Europe the best part of three centuries to progress from a
subsistence state to economic maturity. As for classical theory,
however, the gloomy prognostications of Ricardo, Malthus and
Mill that progress will ultimately end in stagnation would seem to be
unfounded. Population growth and diminishing returns have not
been uniformly depressive to the extent that Ricardo and Malthus
DEVELOPMENT AND UNDERDEVELOPMENT 3
supposed. Rising productivity and per capita incomes appear quite
compatible with population growth and the extension of agriculture.
Classical development economics greatly underestimated the role
of technical progress and international trade in the development
process. It is these two factors, above all, which seem to have con-
founded the pessimism of much of classical theory. With access to
modern technology there is hope, and some evidence, that progress
in today's less developed countries will be much more speedy than in
countries in a similar state one hundred years ago. The pool of
technology on which to draw, and the scope for its assimilation, is
enormous. Used with discretion, it must be considered as the main
means of increasing welfare. As for trade, however, the present
less developed countries are probably in an inferior position com-
pared with the present advanced countries at a comparable stage
of their economic history. The dynamic gains are present but the
static efficiency gains are less and the terms of trade undoubtedly
worse. At the present time the gains from trade are very unequally
distributed between rich and poor countries. This, of course, does
not destroy the potential link between growth and trade or con-
stitute an argument against trade; rather it represents a challenge
for altering the terms on which trade takes place.
The greater knowledge and acceptance of planning may also
mean that the development experience of the present less developed
countries will be less protracted and painful than in the past. Plan-
ning can call forth the prerequisites of development more ex-
peditiously than the market mechanism, which takes time to operate,
and provided attention is paid to the income distribution, the sum
total of sacrifice of present generations need be no more severe per
individual. New techniques in economic theory, such as input-
output analysis and mathematical programming, have given a big
boost to development planning. The present generation of develop-
ment economists are no longer involved simply with identifying
strategic factors in the growth process but also with the scope and
role of planning to achieve greater static and dynamic efficiency
in resource allocation. Classical economists were concerned with the
conditions for static efficiency but, by and large, were either anti-
thetical to interference with the market mechanism or regarded
interference as irrelevant and therefore uninteresting. Today there
is much greater acceptance of interference with the market mech-
anism, and planning in less developed countries is seen by many as
the main means by which the development process may be speeded

up. Some would go further and argue for the complete abandon-
ment of the market mechanism for the allocation of resources. The
Soviet Union is frequently pointed to as an example of the contri-
bution that the replacement of the market mechanism by central
direction can make to rapid economic advance.
Planning requires a certain amount of 'model' building and this,
too, has been inspired by economists. The most common type of
model, which forms the basis of much of the model-building that
less developed countries indulge in, is that to calculate the invest-
ment requirements necessary to achieve a target rate of growth of
per capita income - commonly referred to as a Harrod-Damar
model. Neither the models of Harrodl or Domar2 were designed
for the purpose to which they are now put in less developed coun-
tries, but there can be no dispute that their growth equations have
given a big fillip to macro-planning in less developed countries. We
shall consider later the uses of this type of aggregate model in
development planning.
Enough has been said perhaps to indicate the contribution that
the economist has made to the current pervasive interest in the
process of growth and development. A second major factor account-
ing for the upsurge of interest in development economics has been
the poorer nations' accelerated awareness of their inferior economic
and political status in the world, and a desire for greater political
recognition through economic strength. This has been precipitated
by decolonisation and by increased contact with the developed
nations, and strengthened from within by rising expectations as
development has proceeded. The less developed countries have
shown in recent years a marked determination to pull themselves
up by their own bootstraps, assisted, in the words of Professor
Hicks, 'by such crumbs of aid as the richer countries are willing to
spare, and as they themselves are willing to accept'. 3
Lastly, the developed countries in recent years, especially the
two major power blocs, have been compelled out of political neces-
1 R. Harrod, 'An Essay in Dynamic Theory', Economic Journal (Mar 1939),

and Towards a Dynamic Economics (London: Macmillan, 1948).
2 E. Domar, 'Expansion and Employment', American Economic Review
(Mar 1947).
3 J. Hicks, 'Growth and Anti-Growth', Oxford Economic Papers (Nov 1966)
p. 265.
sity to rethink their political and economic relations with the poorer
nations of the world. The East-West divide has virtually forced the
Western capitalist and the Communist countries to compete finan-
cially for the favours of large parts of the Third World for obvious
political motives. One of the side-effects of the urgent desire of
less developed countries for improved living standards is that they
have allowed themselves to become a political battleground for the
great powers in the cold war. On the credit side, though, there has
also been an affirmation by many developed countries of a moral
obligation towards poorer nations. Not all aid and assistance has
been politically inspired. The developed countries, especially over
the last decade, have been showing genuine concern over the plight
of the less developed world, which has resulted in the establishment
and support of several institutions to assist developing countries,
and which led the period 1960-70 to be named the first Develop-
ment Decade. The goal of greater income equality between the
citizens of a nation seems to be gaining support, albeit slowly, as an
ideal among nations. Moreover, the propagation of this ideal is not
confined to the supra-national institutions that have been especially
established to further it. Recent years have witnessed the spon-
taneous creation of several national pressure groups, in different
parts of the world, whose platform is the abolition of world poverty;
and the Church, which has remained silent for so long, is now making
its voice heard. The Roman Catholic Church recently announced
the sale of property in Paris valued at $1 million to establish a
development fund for Latin America to be administered by the
Inter-American Development Bank.l It is now difficult to believe, as
cynics sometimes argue, that indignation over world poverty is
largely a cover for the selfish realisation among developed nations
that their survival depends on racial and economic harmony which
cannot thrive in a world perpetually divided into rich and poor
(with the divide broadly corresponding to the division between
the white and other races). But whatever the motive for concern,
the reality of world poverty and underdevelopment cannot be
escaped. Furthermore, it is likely to persist for many years in the
future. The economist has a special responsibility to contribute to
an understanding of the economic difficulties these nations face and
to point to solutions. Let us start by establishing the magnitude of
economic divisions in the world as precisely as the data will allow.
1 The Times, 27 Mar 1969.
The World Distribution of Income

By any standard one cares to take, the evidence is unequivocal that
the world's income is distributed extremely unequally between
nations and people, and that there exists in the world a broad north-
south divide into rich and poor countries. For example, if figures
are taken for the distribution of gross domestic product per capita
expressed in U.S. dollars, and ignoring measurement problems for
the time being, the following picture emerges for the year 1963 from
the United Nations' Yearbook of National Accounts Statistics, zg66. Of
the 104 countries for which figures are published, from a total of
132 listed, 19 had registered per capita incomes of less than $100;
56 had per capita incomes between $100 and $500; 11 had per capita
incomes between $500 and $1,000; and 18 had per capita incomes in
excess of $1,000. (Those countries for which figures were not pub-
lished for 1963 would have swelled the ranks of the first two distri-
bution brackets.)
Alternatively, considering the distribution of world income in
relation to the distribution of population, and using $500 annual
per capita income as an arbitrary dividing line between rich and
poor countries, we find that approximately 20 per cent of the world's
population is in receipt of approximately 70 per cent of the world's
income and hence that 80 per cent of the world's population
receives only 30 per cent of the world's income. Moreover, there
is little evidence that this distribution has been narrowing over time.
According to Kuznets, if Lorenz curves are drawn by ranking classes
of countries in ascending or descending order of the ratio of their
percentage share of total incomes to their percentage share of total
population, and the cumulative distribution of income is plotted
against the cumulative distribution of the population for the years
1894--5, 1938 and 1949, one would not find the curves shifting
closer to the 45° line (which represents a perfectly equal distri-
bution).! In short, while per capita income has been rising in the
low-income countries, it must have been growing as fast, if not
faster, in the high-income countries during the first half of the twen-
tieth century.
Since 1949, the evidence is not so clear-cut. According to statistics
presented by Mueller to the annual meeting of the American
1 S. Kuznets, 'Regional Economic Trends and Levels of Living', in
Economic Growth and Structure (London: Heinemann, 1965) pp. 142-75.
Statistical Society (see Table 1.1),1 there is some evidence that the
distribution may have narrowed since 1949, at least during the
first half of the 1950s.
Table 1.1 is based on income and population data for seventy
countries covering 80 per cent of the world's population. When
Lorenz curves are constructed from the data (see Fig. 1.1), Mueller
finds they give unambiguous results for 1957 compared to 1949
and 1962, but that the curves cross for 1949 and 1962.
TABLE 1.1
The World Distribution of Income
% ofworld % of world income

population 1949 1957 1962
Low-income countries 67 15 17 15
Middle-income countries 15 18 18 19
High-income countries 18 67 65 66
When two Lorenz curves cross, precluding a definite conclusion

on distribution from a visual inspection of the curves, a more pre-
cise measure of distribution is required. A common procedure is to
express the area enclosed between the Lorenz curve and the 45 °
line as a ratio of the total area below the 45° line. This is the Gini
coefficient of concentration which varies from 0 (complete equality)
to 1 (complete inequality). Calculation of this ratio yielded the
following results: 0·654 (1949), 0·627 (1957) and 0·640 (1962).2
Mueller therefore concludes that the relative income gap between
rich and poor countries declined somewhat between 1949 and 1957,
rose a little between 1957 and 1962, but showed no greater in-
equality in the early 1960s compared with 1949.
Andie and Peacock, however, reach a different conclusion.3
Taking sixty-two countries, they found that the Lorenz curves cross
for 1949 and 1957, and that the concentration ratio for 1957 and
1949 is approximately the same (0·637 and 0·636 respectively).
But while the concentration ratios are roughly the same, Andie
1 Reproduced in a letter to The Economist, 11 Oct 1969.

2 Information supplied by Mueller.
3 S. Andie and A. Peacock, 'The International Distribution of Income,
1949 and 1957', Journal cifthe Royal Statistical Sociery, part 2 (1961).
and Peacock concluded that the relative position of the less de-
veloped countries must have worsened considerably because of the
much faster growth of countries in the upper quintile of the income
distribution than in the lower quintile. This is certainly true of the
IOOr---------------------------------------------------------------------~
"'E
0
u
c:
...
-"'
0
c:
u
...
a.."' ................ \
...-~
1957~ ...
,•':;:/
;7
;:.:::----1962
100
Per cent of population
FIG.l.l
countries of Western Europe compared to the less developed coun-
tries, which is illustrated in Table 1.2 (p. 12 below).
Turning to the spread of per capita income between countries,
this is also colossal, ranging in 1963 from a recorded $35 per annum
in Malawi to $4,902 per annum in Kuwait. We may also note the
dispersion of the average per capita income figure for the different
continents around the average per capita income for the world,
which was $590 in 1963. The continents are listed in ascending order
of 'richness' : Asia (excluding Japan), $100; Africa, $110 ( 1958) ;
Middle East, $230 (1958); Latin America, $330; Europe, $1,080;
and North America, $2,770 (dates in parentheses refer to the latest
available figures published by the United Nations).
It is easily forgotten that the rich-poor country divide is a
relatively recent phenomenon. All countries were once at subsistence
level, and as recently as two hundred years ago, at the advent of the
British industrial revolution, absolute differences in living standards
between countries on average cannot have been as great as all that.
The average per capita income of the less developed countries today
is approximately $150 per annum, and as far as we know this was
about the average level of real per capita income in Western Europe
in the mid-nineteenth century. If we regard $150 as only barely
above subsistence, the major part of present income disparities
between developed and less developed countries must have arisen
over the last century. Some countries, through a combination of
fortune and design, have managed to grow much faster than others.
The overriding influence has been industrialisation and the tech-
nological progress entailed. The concentrated impact of indus-
trialisation on living standards in the Western world is dramatically
emphasised by Patel's illustration that if six thousand years of man's
'civilised' existence prior to 1850 is viewed as a day, the last century
or so represents less than half an hour; yet in this 'half-hour' more
real output has been produced in the developed countries than in
the preceding period.l It is true that living standards in the less
developed countries are currently rising faster than at any time in the
past; but so, too, are living standards in the developed countries,
and the gap between rich and poor countries continues to widen.
Although development consists of more than a rise in per capita
incomes, income disparities are the essence of the so-called develop-
ment 'gap'. Let us examine the nature and magnitude of the gap
more closely.
The Development Gap2

The statement that 'the rich countries get richer and the poor
countries get poorer' has become a popular cliche in the literature
on world poverty, but without much discussion of the facts or the
precise magnitude of the development task facing the less developed
countries if the per capita income gap between rich and poor nations
is to be narrowed. Indeed, the statement itself is not unambiguous.
1 S. J. Patel, 'The Economic Distance between Nations: Its Origins,
Measurement and Outlook', Economic Journal (Mar 1964).
2 The substance of this section was first published in the National West-
minster Bank Review (Feb 1970), and reprinted in Rivista Internazionale di
Scienze Economiche e Commerciale (Mar 1970).
Since living standards in all countries tend to rise absolutely over

time, it obviously refers to the comparative position of poor countries,
but is the comparative position being measured taking absolute or
relative differences in per capita income? How should the develop-
ment 'gap' be assessed? Unfortunately there is no easy answer to
this question, yet the answer given has a profound bearing on the
growth of per capita income that poor countries must achieve
either to prevent a deterioration of their present comparative posi-
tion or for an improvement to be registered. Relative differences
will narrow as long as the per capita income growth rate of the less
developed countries exceeds that of the developed countries; and
this excess of growth is a precondition for absolute differences to
narrow and disappear in the long run. In the short run, however, a
narrowing of relative differences may go hand in hand with a widen-
ing absolute difference, given a wide absolute gap to start with, and
thus the rate of growth necessary to keep the absolute per capita
income gap from widening is likely to be substantially greater than
that required to keep the relative gap the same. But suppose the
relative gap does narrow, and the absolute gap widens, are the poor
countries comparatively better or worse off?
There is a tendency in economics to measure phenomena,
especially dispersions of income, in relative rather than absolute
terms - to compare differences in the rates of change of variables,
rather than absolute differences, as with Lorenz curves. In com-
paring rich and poor countries, however, it is not difficult to argue
that even if a relative per capita income gap is narrowed, the
comparative position of the poor may have worsened because the
absolute gap has widened. Take for illustration the case of the
average Indian living on the equivalent of $100 per annum com-
pared with the average American living on approximately $4,000.
Suppose the Indian's income rises by 20 per cent and the American's
income by 10 per cent. The Indian is now relatively better off, but
is he not comparatively worse off? The American's increased com-
mand over goods and services (i.e. 10 per cent of$4,000) far exceeds
that of the Indian (i.e. 20 per cent of $1 00), and unless marginal
utilities differ radically, divergences in total utility and welfare will
widen in favour of the American. On welfare grounds there would
seem to be a case for paying as much attention to absolute differences
in per capita income between rich and poor countries as to rates
of growth of per capita income.
In the Development Decade 1960-70, however, both the relative
and absolute per capita income gap between the rich and poor
'continents' seems to have widened, as indicated in Table 1.2, which
shows the rate of growth of gross domestic product (G.D.P.),
population and per capita incomes (P.C.Y.) for the years 1950-60
and 1960-6 for selected areas of the world. There is no way to inter-
pret these figures other than pessimistically. Figures for continents
disguise differences between countries within continents, but for
Mrica, East and South Asia, and Latin America as a whole the
growth of living standards clearly fell short of growth in Europe and
North America in the first half of the 1960s. Only southern Europe
and the Middle East exceeded the performance of the more ad-
vanced countries. And, in general, the experience of the early 1960s
was worse than the decade of the 1950s.
Now let us turn to the future comparative position of the less
developed countries, and the magnitude of the development task
as far as more equitable world living standards are concerned. To
avoid the issue of whether strategy and assessment should be con-
cerned with absolute or relative per capita income differences,
and to facilitate quantification, matters will be made simple by
assuming that the desirable goal is to narrow and eliminate both
the absolute and the relative gap. We shall take as a target the per
capita income of America and the existing countries of the European
Economic Community (E.E.C.), and attempt to answer three
specific questions as reliably as the data will allow:
1. Supposing that the E.E.C. countries and America, i.e. the

United States (as representative of the developed world),
experience per capita income growth of 3 per cent per annum
from now until the year 2000, how fast would certain less
developed countries have to grow for per capita incomes to be
equalised by that date?
2. Given the above assumptions, how fast would the less developed
countries have to grow merely to prevent the present absolute
per capita income gap between rich and poor countries from
being any wider in the year 2000?
3. Given the recent growth experience of certain less developed
countries, and again assuming 3 per cent per capita income
growth in the E.E.C. and America, how many years would it
take for the per capita income gap to be eliminated?
TABLE 1.2
Rates of Growth qf Gross Domestic Product,
Population and Per Capita Income, 1950-60 and 1960-6
1950-60 1960-6
(% p.a.) (% p.a.)
('poor' countries)
Africa
G.D.P. 4·4 3·3
Population 2·2 2·3
P.C.Y. 2·2 1·0
South Asia
G.D.P. 3·6 3-4
P.C.Y. 1·7 0·5
East Asia
G.D.P. 5·1 4·9
P.C.Y. 2·5 2·1
Southern Europe
G. D.P. 5·6 7·7
P.C.Y. 4·1 6·2
Latin America
G.D.P. 4·9 4·7
P.C.Y. 1·9 1·7
Middle East
G.D.P. 5·6 7·2
P.C.Y. 2-4 4·2
('rich' countries)
North America
G.D.P. 3·3 5·0
Population 1-8 1·5
P.C.Y. 1·5 3·4
Western Europe
G.D.P. 4·7 4-4
P.C.Y. 4·0 3·4
Source: International Development Agency, Annual Report, 1968.

By asking the first two questions we get some idea of the growth
task facing the less developed countries in their struggle not only for
parity of living standards with the developed world but also in
preventing the absolute gap in living standards from widening.l
The third question is designed to highlight how lengthy the catching-
up process is likely to be for the poorer countries if they do not achieve
per capita income growth substantially in excess of the growth of
developed nations. There is, of course, no answer to the third ques-
tion if the growth of the poor countries is less than that of the devel-
oped countries (i.e. less than 3 per cent on our assumptions), except
that the relative and absolute per capita income gap between rich
and poor countries will widen for ever.
Given the basic data, the solutions to the questions posed involve
no more than simple manipulation of the formula for compound
interest: X= 0(1r + r)n.2 The basic data have been taken at face
value from the United Nations' Accounts Statistics, and the answers
to the questions are given in Table 1.3 for a wide selection of coun-
tries from different continents.
A few words of caution are in order before commenting on the
results. The answers to the first two questions depend not only on
the choice of the target year in the future but also on the figures for
1 The answers given, of course, will be somewhat sensitive to the target
year chosen. No significance should be attached to the year 2000, however.
Some year has to be taken and the year 2000, apart from marking the end
of one millennium and the start of another, also provides a date far enough
away for action to be taken and results achieved to raise living standards in
the less developed nations.
2 The solution to question 1 is given by
r = (~Xt{To) - 1 X 100;
the solution to question 2 is given by
r = (~[Xt - (Xo - To)/To]) - 1 X 100;
and the solution to question 3 is given by
loge Xo{To
t= X 100
r11 - r.,
where Xt is per capita income in the E.E.C. and America in the year 2000
Xo is base-period per capita income in the E.E.C. and America
To is base-period per capita income of the less developed country
n is the number of years between the base period and the year 2000
r11 is the rate of growth of per capita income in the less developed
country
r., is the rate of growth of per capita income in the E.E.C. and
America.
TABLE 1.3
The Development 'Gap'
(I) (2) (3) (4) (5)
Per Recent Per capita Per capita No. of years
capita annual income growth income growth required to
gross growth required in required to close gap
domestic of less developed prevent from between less
product G.D.P. country to ·widening the developed
atfactor per reach P.C.Y. absolute gap country and
cost in capita of between less E.E.C./U .S.A.
$U.S. (%) E.E.C./U.S.A. developed
(1965) (by year 2000) country and
E.E.C./U.S.A.
(by year 2000)
Latin America
Argentina 783 2·9 4·7 7·3 4·3 6·8 n.s. n.s.
Barbados 377 6·1* 7·0 9·5 6·1 8·3 43 69
Bolivia !53 4·8 9·7 12-4 8·6 IJ.l 124 169
Brazil 232 -1·6 8·5 11·0 7-4 9·8 n.s. n.s.
Chile 579 6·0 5·7 8·1 5·0 7-1 30 57
Colombia 267 2·8 8·0 10·6 7·0 9·3 n.s. n.s.
Costa Rica 382 Q.7 6·9 9·4 6·0 8·3 n.s. n.s.
Dominican Republic 231 0·8 8·5 11·0 7·4 9·8 n.s. n.s.
Ecuador 200 1·6 8·9 11·5 7·8 10·2 n.s. n.s.
El Salvador 252 1·9 8·2 10·8 7-1 9·5 n.s. n.s.
Guatemala 298 2·5 7·7 10·3 6·6 9·1 n.s. n.s.
Guyana 296 4·9 7-7 10·3 6·7 9·1 83 125
Haiti 86 0·1 1J.6 14·2 10·3 12·9 n.s. n.s.
Honduras 207 1·9 8·8 11-4 7·7 10·1 n.s. n.s.
Jamaica 453 3·5 6·4 8·9 5·6 7·8 230 392
Mexico 443 6·2 6·5 9·0 5·7 7·9 37 62
Nicaragua 325 3·4 7-4 10·0 6·4 8·8 370 573
Panama 474 4·3* 6·3 8·7 5·5 7·7 85 147
Paraguay 200 5·8* 8·9 11·5 7·8 10·2 70 99
Peru 238 3·5 8·4 10·9 7-4 9·7 359 520
Puerto Rico 1089 9·4 3·8 6·2 3·5 5·5 4 17
Trinidad and Tobago 661 3·0 5·3 7·9 4·7 6·8 n.s. n.s.
Uruguay 562 2·9 5·8 8·2 5·0 7-1 n.s. n.s.
Venezuela 916 10·9*' 4·3 6·8 3·9 7·8 6 16
Latin America 380 1·8 1 7·0 9·5 6·1 8·3 n.s. n.s.
Africa
Algeria 206• 5·0' 8·6 11·0 7-6 9·9 71 122
Ethiopia 47 2·3 13·5 16·2 12·2 15·0 n.s. n.s.
Gabon 369 6·4 6·9 9·5 6·1 8·4 40 64
Gambia 85 3·0 IJ.6 14-2 lo-7 12·9 n.s. n.s.
Ghana 265 7·8 8·0 10·6 7-1 9·4 35 52
Kenya 86 1·8 11·6 14·2 10·3 12·8 n.s. n.s.
Liberia 251• 7·6' 8·0 IO.S 7·0 9·3 36 54
Libya 707 29·2 5·1 7·8 4·5 6·5 3 7
Malawi 41 3·6 14·0 16·6 12·6 15·2 592 727
Mauritius 228 0·6 8·5 IJ.l 7-4 9·7 n.s. n.s.
Morocco 180 1·8 9·2 11·8 8·1 10·5 n.s. n.s.
Nigeria 68 5·1 12-4 15·0 11·0 13·6 145 184
Sierra Leone 136 1·5*' 10·1 12·7 8·9 11·3 n.s. n.s.
Somalia 5\b 11·0 13-1 10·4 12·3 n.s. n.s.
Sudan 96 2·3 11·2 13·9 10·0 12·5 n.s. n.s.
Tanzania 69 3·0* 12·3 14·9 10·8 13·5 n.s. n.s.
Tunisia 188 3·9 9·1 11·7 7·9 10·4 225 315
Uganda 83 3·8 11·8 14·4 10·4 12·9 356 457
Zambia 206 6·8 8·8 11-4 7-7 10·1 51 73
Africa 110b 1·0' 8·9 IJ.l 8·4 10·3 n.s. n.s.
TABLE I .3-continued
(I) (2) (3) (4) (5)

Per Recent Per capita Per capita No. ofyears
capita annual income growth income growth required to
gross growth required in required to close gap
domestic of less developed prevent from between less
product G.D.P. country to widening the developed
at factor per reach P.C.Y. absolute gap country and
cost in capita of between Jess E.E.C./U .S.A.
$U.S. (%) E.E.C./U.S.A. developed
(1965) (by year 2000) country and
E.E.C.fU.S.A.
(by year 2000)
East and South-east Asia

Mghanistan 49b n.a. 11·1 13·2 10·9 12-4 n.a. n.a.
Burma GOa 1·9' 12·4 15·0 lJol 13-7 n.s. n.s.
Cambodia 120 6·9 10·6 13-1 9·3 11·8 64 84
Ceylon 137 J.4 10·1 12-7 8·9 11-4 n.s. n.s.
Hong Kong 305• 3·5 7·2 9-6 6·4 8·6 309 471
India 92 4·4 11·4 14·0 10·1 12·6 196 254
Indonesia 82 -0·2' 11·8 14·4 10·5 13·0 n.s. n.s.
Iran 240 3·5 8 8·4 10·9 7·3 9·6 357 636
Korea (Republic) 93 4·0' 11·4 14·0 10·1 12·6 273 354
Laos 62b n.a. 10·4 12·6 9·9 11·8 n.a. n.a.
Malaysia 272 3·5 8·0 10·5 6·9 9·3 332 494
Nepal 69 4·7 12·3 15·0 11·0 13·5 178 226
Pakistan 95• 3·2' 10·7 13·9 9·7 12·0 1356 1760
Philippines 237 2·6 8·4 11·0 7-4 9·7 n.s. n.s.
Singapore 527 2·9 6·0 8·4 5·2 7·4 n.s. n.s.
Taiwan 200 6·0 8·9 11·5 7·8 10·2 66 93
Thailand 113 5·0 10·7 13·3 9·5 12·0 127 167
East and South-east IOOa n.a. 10·5 12·9 9·5 11·8 n.a. n.a.
Asia (excluding Japan)
Middle East
Aden 53b n.a. 10·9 13·0 10·3 12·2 n.a. n.a.
Iraq 242d 2·3' 8·4 10·9 7·3 9·6 n.s. n.s.
Jordan 198o 6·3' 8·7 11·2 7-7 10·3 58 83
Lebanon 218b n.a. 7·2 9·3 6·9 8·6 n.a. n.a.
Syria 162• 6·4' 9·1 11·5 8·2 10·4 64 88
Yemen SOb n.a. 11·0 13·2 10·4 12·4 n.a. n.a.
Middle East 230b n.a. H 9·1 6·6 8·4 n.a. n.a.
(including Kuwait)
Key:
Col. (1): a refers to 1964; b refers to 1958; c refers to 1963; d refers to 1965. Source: United Nations,
rearbook of NatioMI Accounts Statistics, 1966, Table 7A, pp. 725-9.
Col. (2): Figures refer to the period 1958-65 unless otherwise stated.
1 refers to 1950-66
1 refers to 1950-8
' refers to 1958-64
• refers to 1963-5
• refers to 1960-6
1 refers to 1958-63
7 refers to 1960-3
8 refers to 1960-5
n.a. = not available.
• denotes figures not comparable from year to year. The growth figures in col. (2), therefore,
are highly unreliable and no importance should be attached to the calculations in col, (5).
Col. (5): n.s. =no solution.
the base-year level of per capita income in the less developed coun-
tries given in the first column of Table 1.3. To the extent that
income statistics invariably understate the value of production in
less developed countries, the calculations in cols (3) and (4) are
overestimates of the magnitude of the growth task. The degree of
overestimation is not likely to be so great, however, as to invalidate
the conclusion that the growth rates required are not feasible given
the past experience of most of the countries and knowledge of their
future prospects.
The second obvious caution is that the recent per capita income
growth of a country may not be a reliable guide to its future per-
formance. The assumed 3 per cent growth of per capita income for
the E.E.C. and America, based on long-run trends, may not be too
unrealistic, but for most of the less developed countries under review
it has only been possible to take the period 1958-65 as a guide to the
future, and growth rates over such a short period, whether high or
low, may obviously have been subject to special factors. In general,
however, the rate of growth of per capita income may be expected
to decelerate with development. Admittedly, population growth
(the denominator in the calculation) is likely to decelerate, but so
too may income growth (the numerator) as the scope for absorbing
the subsistence sector into the money economy diminishes. Growth
rates, as opposed to statistics of the absolute level of income, do
tend to be biased upwards in less developed countries for this
reason and others. The meaningfulness of the calculations in col. (5),
therefore, in answer to question 3, depends on the reasonableness
of the growth rates shown in col. (2). The calculations are highly
sensitive to the growth rates taken, but unless the figures are gross
underestimates, the message of the calculations is likely to be the
same, and it is this in which we are primarily interested rather than
precise calculations for individual countries.
Thirdly, as far as the comparison of living standards between
countries is concerned, there are the exchange problems associated
with the conversion of per capita incomes valued in national
currencies into U.S. dollars. As we shall see later in this chapter,
exchange rates are generally a very poor reflection of relative
prices and purchasing power in different countries. Foreign ex-
change rates basically reflect the relative prices of goods and ser-
vices which enter international trade. Typically, goods and services
produced and used within low-income countries are cheaper, relative
to the same goods in developed countries, than those that enter into
foreign trade, so that the conversion of less developed countries'
national incomes into dollars by the use of the foreign exchange
rate understates their true income. Moreover, the mix of goods
bought and sold in one country may be radically different from the
mix bought and sold in another, so that the equivalent of a dollar's
expenditure in, say, India, at the current rate of exchange, may
'buy' a very different standard ofliving from a dollar in America.
This difficulty must be lived with, but should not preclude analysis
or seriously affect the conclusions.
With the above qualifications in mind, what are the main con-
clusions that emerge from Table 1.3? First, it is clear that the growth
rates that would be necessary in the less developed countries to raise
per capita income to the level in the E.E.C. and America by the
year 2000 are, for most of the countries, considerably in excess of the
past rates of growth achieved. Take, for example, the countries of
Latin America. Excluding Venezuela (for reasons given in the
table), none of the countries examined has experienced growth in
the recent past which would be sufficient for per capita income to
match living standards in either the E.E.C. or America by the year
2000 (except for Puerto Rico, which is virtually a state of the
U.S.A.). For the vast majority of countries an annual rate of growth
of per capita income of 8 per cent would be required to match the
E.E.C. countries, and a corresponding rate of 10 per cent to match
America, compared with actual per capita income growth of between
2 and 4 per cent. The picture is broadly the same for Africa, Asia
and the Middle East. On the continent of Africa, Ghana and Liberia
would just about match the E.E.C. in the year 2000 if the current
high rates of per capita income growth could be maintained. Libya's
recorded growth rate is hard to believe, but if it is correct and was
maintained, living standards there would surpass those in America
within seven years! For most Mrican countries, however, the out-
look is bleak. The scene in East and South-east Asia (excluding
Japan) appears even more depressing, where there is no country
with a recent per capita income growth that would raise the level
of per capita income close to that of the E.E.C. during the next
thirty years. And the story is the same for the Middle Eastern
countries considered.
In answer to the second question the same conclusion emerges as
in answer to the first, because the growth of per capita income
necessary to prevent the present per capita income gap from

widening is almost as high as the growth rate necessary to eliminate
the gap entirely. Unless there is a substantial increase in per capita
income growth in the less developed countries over the next decade,
it can safely be said that with the exception of a handful of countries
the absolute per capita income gap between the developed and less
developed countries will be considerably wider in the year 2000 than
it is now. Here we illustrate the point that the growth required to
prevent an absolute gap from widening may be considerably
greater than that required to keep the relative gap from widening.
(Compare the growth rates in col. (4) with the rate of 3 per cent
which, on our assumptions, would be necessary to prevent relative
gaps from widening.)
In short, the magnitude of the development task is colossal if
defined in terms of achieving roughly comparable living stan-
dards throughout the world in a relatively short space of time. For
most of the less developed countries a doubling or trebling of per
capita income growth would be required over the next three dec-
ades, necessitating net investment ratios of between 20 and 30
per cent of national product, even with population growth substan-
tially reduced. Investment ratios of this magnitude, from domestic
resources alone, would involve intolerable reductions in con-
sumption. The only hope must lie in a massive infusion of resources
from abroad. What form assistance may take we discuss in Chapter 8.
In answer to the third question, there is no solution for the major-
ity of the countries considered because their recent growth of per
capita income has been less than the 3 per cent future growth assumed
for the E.E.C. and America. Where there is a solution, however,
the number of years that it would take some countries to get on a
par with the E.E.C. and America is quite staggering. Take a few
countries at random where recent growth has been modestly in
excess of 3 per cent per annum. In the case of Peru, for example,
given its present per capita income level, and a per capita income
growth rate of 3·5 per cent per annum, it would take 359 years for
per capita income to match the level of the E.E.C. and 520 years
to match the level of America. In the case of Malawi the corre-
sponding figures are 592 years and 727 years, and in the case of
Pakistan 1,356 years and 1,760 years respectively. The precise
accuracy of the growth figures can be quibbled over, but again the
general conclusion cannot be escaped that, with living standards
currently prevailing, it will take centuries, if ever, for living stan-
dards to be equalised in the world unless the growth of per capita
income in less developed countries exceeds by at least 3 per cent the
growth of the developed nations. This in turn implies, among other
things, substantially more resources devoted to investment which
can only come from abroad in the absence of the necessary con-
sumption sacrifices internally.
The arguments and calculations are purposely presented in stark
form to dramatise the size of the development gap for the idealist.
It can be argued, of course, that world income equality is an im-
practicable ideal and that the primary aim is not equality of living
standards throughout the world but 'tolerable' living standards in
all countries, which is a very different matter. The problem is to
define 'tolerable' living standards, and especially to guarantee a
reasonably equitable distribution of a tolerable average per capita
income which does not leave 10-20 per cent of the population in
need of assistance as in some advanced countries. Since primary
poverty still exists in most of today's so-called developed countries,
and is likely to persist for some time, it is not necessary to be an
idealist to treat seriously the calculations in cols (3) and (4) of
Table 1.3. The rates of growth required to match the E.E.C. by the
year 2000 would produce living standards currently enjoyed in
America which, although a wealthy country by existing world
standards, by no means guarantees a comfortable living for everyone.
On the other hand it is true that the answers to question 3 in
col. (5) of the table do lose some of their force if the aim is to achieve
tolerable living standards. Absolute poverty in Pakistan, for ex-
ample, will presumably have been eradicated long before the year
when, on current trends, per capita income in Pakistan and America
would be equalised. No one would dispute, however, the current
existence of acute world poverty, and the foregoing analysis gives
some idea of the height of the hurdles to its solution.
Per Capita Income as an Index of Development

Figures of per capita income are frequently used as an index of
development and for making a distinction between developed and
underdeveloped countries, as well as between rich and poor. While
there may be an association between poverty and underdevelop-
ment and riches and development, there are a number of reasons
why great care must be taken in using per capita income figures as
a criterion of development (unless underdevelopment is defined as
poverty and development as riches, which is not uncommon).
Apart from the difficulties of measuring national income in many
countries, together with the misleading nature of the figures and the
difficulties of making inter-country comparisons, which will be
considered later, a single per capita income figure to divide de-
veloped from less developed countries is somewhat arbitrary, ig-
noring such factors as the distribution of income within countries
and differences in development potential. It is not so much a question
of whether or not low-income countries should be labelled 'under-
developed' but what income level should be used as the criterion for
separating the developed from the less developed countries, and
whether all high-income countries should be labelled 'developed'.
To give just one example of the problems that might arise if the
figures are not used with discretion, suppose $300 annual per capita
income was taken as the dividing line between developed and less
developed countries. If this was done, much of South America
would be classified as developed, which would be absurd. Equally
absurd would be to classify Kuwait as highly developed simply
because it has the highest per capita income in the world.
Furtado! has attempted a structural definition of underdevelop-
ment which has a sufficient degree of generality to cope with cases
which do not accord with definitions of development and under-
development using per capita income as a criterion: 'Underdevelop-
ment is a state of factor imbalance reflecting a lack of adjustment
between the availability of factors and the technology of their use,
so that it is impossible to achieve full utilisation of both capital and
labour simultaneously.' An underdeveloped structure is therefore
a situation in which 'full utilisation of available capital is not a
sufficient condition for complete absorption of the working force at
a level of productivity corresponding to the technology prevailing
in the dynamic sector of the system'. By this criterion, Kuwait is
obviously not a developed country, and Australia and Canada may
be classified as totally developed even though they have under-
utilised natural resources and great development potential. But
bearing in mind the arbitrariness of per capita income, it is still very
convenient to have one readily available, and easily understandable,
1 C. Furtado, Development and Underdevelopment (Berkeley: University of
California Press, 1964) pp. 141-3.

criterion for classifying countries, and perhaps per capita income is
the best single index we have. It also has one positive advantage,
namely that it focuses on the raison d'etre of development which is the
raising ofliving standards and the eradication of poverty. And in the
last resort per capita income is not a bad proxy for the social and
economic structure of societies. Ifless developed countries are defined
on the basis of a per capita income level so as to include most of the
countries of Asia, Africa and South America, striking similarities
are found between the characteristics and development obstacles of
many of the countries in these continents. These characteristics
include a high proportion of the labour force engaged in agriculture
and low agricultural productivity; a high proportion of domestic
expenditure on food and necessities; an export trade dominated
by primary products and an import trade dominated by manu-
factured goods; labour-intensive technology; high birth rates coupled
with falling death rates; and savings undertaken by a small per-
centage of the population. There may, of course, be some countries
which on a per capita income basis are classified as developed and
which possess most of the above-mentioned characteristics, but the
exceptions will be few, and the reverse of this situation is almost in-
conceivable. In general, therefore, we conclude that per capita
income may be used as a starting-point for classifying levels of
development, and can certainly be used for identifying the need for
development. The only major reservation that we shall have to
make later concerns the case of geographically dual economies
where an aggregate per capita income figure can disguise the need
for the development of a sizeable region as great as the need for the
development of a country itself, e.g. southern Italy.
It should be emphasised that we are not at this stage implying
any causal relation between the characteristics of low-income
countries and the extent of their poverty or underdevelopment.
Low per capita incomes do seem to go hand in hand with such
characteristics as high birth rates and an absence of industry, but it
is always dangerous to equate association with causation, and in
this context what is cause and what is effect is by no means certain
without some adequate theorising. As Kuznets reminds us, 'it is
easy to translate close statistical association into significant causal
relationships ... [but] in view of the continuous interplay of income
levels and these associated characteristics this simple translation is
a logical trap that should be avoided lest it lead to intellectual
sterility and to a dangerously mechanistic approach to policy impli-

cations' .1 A simple mechanical argument ascribing poverty to such
factors as low savings and primary product production ignores the
fact that countries may be at different stages of their economic
history and may differ radically with respect to past history and
future prospects. In this respect, four main categories of low-
income countries need distinguishing: firstly, those countries with
low per capita incomes but which are progressing rapidly and with
enormous future potential based on indigenous resources; secondly,
those countries with rising per capita incomes but with less hope of
rapid self-sustaining growth because of resource limitations; thirdly,
those countries rich in resources but with per capita income still
relatively stagnant; and lastly, those countries with a stationary
per capita income and with little prospect ofraising living standards
because of a sheer lack of resources. A low-income country may fall
into any one of these four broad categories and it would obviously
be misleading, without further information, to ascribe low per capita
income in a particular country at any particular point in time to
characteristics it shares with other low-income countries. Association
between low incomes and certain development characteristics is all
that one is really entitled to claim.
There is a difference, however, between using per capita income
as a guideline for classifying countries into developed and under-
developed at a point in time, and using the growth of per capita
income as an index of development over time. The difficulty of using
per capita income for the latter purpose is the obvious one that if,
in a particular period, per capita income did not grow because
population growth matched the growth of a country's total income,
one would be forced into the odd position of denying that a country
had developed even though its national product had increased. This
is an inherent weakness of linking the concept of development to a
measure of living standards.
This leads on to the distinction between growth and develop-
ment. Development without growth is almost inconceivable, but is
growth possible without development? If per capita income is
rejected as an index of development over time, an answer to this
question is not possible without defining terms more precisely.
1 S. Kuznets, 'International. Differences in Income Levels', in B. Okun
and R. Richardson (eds), Studies in Economic Development (New York: Holt,
Rinehart & Winston, 1961) p. 9.
The difficulty is defining 'development'. The meaning of' growth'
is fairly unambiguous; most economists would accept the definition
of a rise in real national income, i.e. a rise in money income deflated
by an index of prices. But 'development' is an elusive term meaning
different things to different groups of social scientists.! Most would
agree that development implies more than just a rise in real national
income; that it must be a sustained, secular rise in real income
accompanied by changes in social attitudes and customs which have
in the past impeded economic advance. But at this point agreement
on what constitutes development would probably end. But whatever
definition of development is given, growth is clearly possible with-
out the broader societal changes referred to. The upswing of the
trade cycle is the most obvious example of the possibility of growth
without development, and examples of abortive 'take-offs' are not
hard to find where countries have grown rapidly for a short time and
then reverted to relative stagnation. Argentina is a case in point.
On the other hand, development is hardly possible without growth;
but development is possible, as we have suggested, without per
capita income rising. It would be a strange, rather purposeless,
type of development, however, which left per capita income un-
changed, unless the stationary per capita income was only tem-
porary and a strong foundation was being laid for progress in the
future. For the ultimate rationale of development must be to
improve living standards and welfare, and while an increase in
measured per capita income may not be a sufficient condition for an
increase in individual welfare, it is clearly a necessary condition in
the absence of radical institutional innovations, e.g. the distri-
bution of' free' goods.
An increase in income is not a sufficient condition for an increase
in welfare, for the obvious reason that an increase in income can
involve costs as well as benefits. It may have been generated at the
expense of leisure or by the production of goods not immediately
consumable. If development is looked upon as a means of improving
the welfare of present generations, probably the best index to take
1 For semantic entertainment on the meaning of 'development' and
'underdevelopment', see F. Machlup, 'Disputes, Paradoxes and Dilemmas

Concerning Economic Development', in Essays in Economic Semantics (New
York: Norton, 1967). Machlup himself defines economic development as
'those changes in the use of resources that result in a potentially continuing
growth of national income per head in a society with increasing or stable
population'.
would be consumption per man-hour worked. This index, in

contrast to an index of per capita income, focuses directly on the
immediate utility derivable from consumption goods in relation to
the disutility of work effort involved in their production.
The Measurement and Comparability of Per Capita Incomes

We turn now to the difficulties of measuring real per capita income
and comparing living standards between countries. These difficul-
ties cannot be ignored any longer and must be continually borne in
mind in using per capita income figures both for classification pur-
poses and for comparing the rate of development in different coun-
tries over time. The difficulties of obtaining meaningful and
accurate measures of real per capita income relate more to the
measurement of real income than to population, and we shall thus
concentrate briefly on some of the problems of national income
accounting and the uses of national income statistics in less developed
countries.
The first point to bear in mind is that if no allowance is made for
the non-monetary sector in the national income accounts of a less
developed country, any long-term growth estimates are bound to
have an upward bias owing to the gradual extension of the money
economy and the shift of economic activities from the household to
the market-place. Furthermore, if no allowance is made for the
subsistence sector in some countries, it may be misleading to com-
pare periods in these countries' history and to compare growth rates
between countries, especially between the developed and the less
developed countries.
Growth rates may also be biased upwards by using prices as
weights in compiling national income totals from the output
statistics of different sectors of the economy (unless the weights are
revised frequently), since goods with high prices, which subsequently
fall, are usually the fastest growing. This is more of a danger in less
developed countries than in developed countries because of less
sophisticated accounting techniques, the greater difficulty in
revising price weights, and the more widespread introduction of
new goods with high initial prices.
A consideration of prices is also necessary in deciding on what
price index to use as a deflator of money national income in order to
obtain an index of real income. The task of converting money in-
come statistics into real income raises all the difficulties, not peculiar
to less developed countries, connected with the use of index numbers,
such as which base Y.ear should be taken, how to take account of
changes in the quality of products, which weighting system to
employ, etc. These are conceptual issues to be sorted out by the
national income statistician rather than the development economist,
but it is important for the economist to know how figures for real
national income, or per capita income, have been arrived at prior
to analysis.
But apart from the problem of bias and the choice of a price
deflator, there is the sheer practical difficulty of measuring money
national income in a rural economy where communications are
bad, illiteracy rife, and in which many goods produced and con-
sumed do not exchange for money. Differences in the extent of the
subsistence economy between less developed countries, and dif-
ferences in the ease and difficulty of collecting data, may markedly
influence estimates of national income, and therefore of per capita
income differences, between these countries and the rest of the
world. Some testimony to the role that the subsistence sector must
play in the economies of most less developed countries is illustrated
by the inconceivability that 60 per cent of the world's population
could remain alive on the equivalent of$200 per annum. But this is
not the whole story.
The problem of inter-country comparisons crops up in its most
acute form in the conversion of the national incomes of diverse
countries into a common unit of account (e.g. the dollar). If the
dollar is used as the unit of account, the national income per head of
country X in U.S. dollars is given by
G.N.P.x
Exchange rate X Population
e.g. ifthe G.N.P. of country X is 100 million ducats, its population

is 5 million, and there are 20 ducats to the dollar, then per capita
income of country X in dollars is
100
20 X 5 = $1.
But if living standards are to be compared by this method, it

would have to be assumed that comparing incomes through the
exchange rate is equivalent to dividing the income of each country
by an index of the cost of living; that is, if $1 = 20 ducats, then

20 ducats can buy the same real standard of living in country X
as $1 in America. It is common knowledge, however, that exchange
rates may be a very poor reflection of relative prices and purchasing
power in different countries. For one thing the exchange rate may
not be in equilibrium in relation to the balance of payments of a
country. But even the equilibrium exchange rate may fail to reflect
purchasing power parity. We have already mentioned that ex-
change rates only reflect the -prices of traded goods, and that the
mix of goods purchased within countries may differ so radically as
to make price comparisons meaningless. What is required is some
measure of purchasing power parity, or a real exchange rate, which
takes differences in the internal price level and the structure of
consumption into account. The practical problem is to distinguish
similar commodities between countries to which to apply price
weights. It is interesting to note that on converting all prices into
U.S. dollars by the foreign exchange rate, it turns out that the price
levels are lowest in the poorest countries.! In other words, real in-
come per head in less developed countries is much higher compared
with America than is suggested by estimates obtained simply by
converting per capita incomes into dollars at the official exchange
rate. The lower the per capita income, the greater tends to be the
disparity between the exchange rate and comparative living costs,
and the wider the difference between actual real income and its
measure in U.S. dollars per capita. This is partly because of the way
untraded goods are priced and partly owing to the fact that goods
traded are typically more expensive than home-produced goods. In
short, the conversion ofpoor countries' national incomes into U.S.
dollars is likely to understate living standards. This, together with
the existence of a large subsistence sector, goes most of the way
towards explaining how it is that people can apparently live on a
per capita income of$200 per annum when in America people would
literally starve on the same amount.
Millikan has suggested that the real incomes of many African and
Asian countries are of the order of350 per cent larger than indicated
by United Nations statistics in U.S. dollars per capita. He estimates
that the real income of Asian countries in 1950 amounted to $58
per capita converted at the appropriate rate of exchange; the real
1 D. Usher, Rich and Poor Countries, Eaton Paper No. 9 (London: Institute
of Economic Affairs, 1966) pp. 10-11.
level, after elimination for bias involved in differences in prices of
non-traded goods and services, was nearer $195. And for Mrica, for
the same year, the respective figures were $48 and $177.1 Usher
estimates that the ratio of British to Thai national income per head
is 13·06 when incomes in local currencies are converted into pounds
by the foreign exchange rate. If the foreign exchange rate is by-
passed, however, and British and Thai incomes are valued directly
at Thai prices, the ratio is 6·27, and when both incomes are valued
at British prices the ratio is only 2·76.2 These discrepancies reflect
the ever-present index number problems, to which there is no a
priori solution, when the composition of output differs radically
between two countries. In making international comparisons of
per capita incomes it makes a great difference, therefore, whether
one accepts the foreign exchange rate as a measure of purchasing
power parity or, if one discards the use of exchange rates, which
countries' prices are used as weights to value both countries' national
income from the statistics of output.
The exchange problem is further compounded in comparing
two countries neither of which uses the common unit of account
as currency. For example, if there are 3 shekels to the dollar and 20
ducats to the dollar, and both countries' incomes have been con-
verted into dollars, the purchasing power of 3 shekels has, in effect,
been equated to 20 ducats. From the point of view of comparing
the basketful of goods that 3 shekels and 20 ducats can buy, the
comparison of per capita incomes expressed in dollars may again be
misleading.
The Stages of Development

It is often argued that countries pass through phases during the
course of development and that by identifying these stages, according
to certain characteristics, a country can be deemed to have reached a
certain stage of development. The simplest stage theory is the sector
1 Statement by M. F. Millikan before the Subcommittee on Foreign

Economic Policy of the Joint Economic Committee on the Economic
Report, Hearings, Foreign Ecorwmic Policy, 84th Congress, 1st Session, pp. 21,
28, cited in C. Kindleberger, Economic Development, 2nd ed. (New York:
McGraw-Hill, 1965) p. 9.
z Usher, Rich and Poor Countries, p. 35.
thesis of Fisherl and Clark,2 who employ the distinction between

primary, secondary and tertiary production as a basis of a theory of
development. Countries are assumed to start as primary producers
and then, as the basic necessities of life are met, resources shift into
manufacturing or secondary activities. Finally, owing to rising
income, more leisure and an increasingly saturated market for
manufactured goods, resources move into service or tertiary
activities producing 'commodities' with a high income elasticity of
demand.
Naturally enough in this schema, the less developed countries
get identified with primary production, the more developed coun-
tries with the production of manufactured goods, and the mature
developed economies with a high percentage of their resources in
the service sector. There can be no dispute that resource shifts are an
integral part of the development process, and that one of the main
determinants of these shifts is a difference in the income elasticity of
demand for commodities and changes in elasticity as development
proceeds. But just as one must beware of equating (without quali-
fication) development and welfare with the level of per capita
income, so, too, caution must be exercised in identifying different
degrees of underdevelopment, industrialisation and maturity with
some fairly rigid proportion of resources engaged in different types
of activity. Such an association would ignore the different types of
service activity which may exist at different stages of a country's
history. It would also ignore the doctrine of comparative advantage
which holds that countries will specialise in the production of those
commodities in which they have a relative advantage as determined
by natural or acquired resource endowments. The fact that one
country produces predominantly agricultural products while
another produces mainly manufactured goods need not imply that
they are at different stages of development on any of the conventional
definitions of development we gave earlier. Mter all, Britain has
about the lowest proportion of the work-force in agriculture in the
world, but this hardly makes her the world's most 'developed'
nation. And compare Australia and Great Britain. Is Australia
less developed than Great Britain because the percentage of its
1 A. G. B. Fisher, 'Capital and the Growth of Knowledge', Economic
Journal (Sep 1933) and 'Production: Primary, Secondary and Tertiary',
Economic Record (June 1939).
2 C. Clark, The Conditions of Economic Progress (London: Macmillan, 1940).
resources engaged in primary product production is ten times
greater?
The same arguments apply to service activities. The view that
a high proportion of labour in tertiary production is a consequence
of, or a pointer to, a high level of development or standard of living
must be severely qualified. In practice, contrary to the Fisher-
Clark prediction, the proportion of a nation's resources devoted to
service activities does not seem to change very much in the course
of development. There are three broad categories of service activities,
and the determinants of resource allocation to services accompanying
development appear to operate differently on each in an offsetting
manner. Newer service activities, linked with the growth of leisure
and high mass consumption, tend to have a very high income elas-
ticity of demand; services linked to the growth of manufacturing
also grow but at a declining rate; and traditional services of
pre-industrial times decline.l In short, tertiary production is an
aggregation of many dissimilar service activities some of which are
related to low per capita incomes and some to high per capita
incomes. Today, there is very little disparity between the proportion
of total resources devoted to services in the developed and less
developed countries. Bauer and Yamey2 have drawn attention to
the large number of people engaged in small-scale distribution and
petty trading in West Africa, which may be taken as fairly typical
of many less developed countries lacking employment opportunities
in the urban areas. As these countries develop there will be a decline
in this type of service activity but a rise in those associated with the
growth of manufacturing.
Ideally, a criterion of development stages is required which leaves
the proportion of resources employed in different activities out of
account. One possibility is to argue that a country has reached a
developed state when productivity in the agricultural sector matches
productivity in the industrial sector, and that it has reached a state
of maturity when productivity in all sectors, including services, is
approximately equal, provided the level is reasonably high. The
alternative is simply to classify countries as industrial, semi-industrial
and non-industrial, using as a criterion for division some level of
1M. A. Katouzian, 'The Development of the Service Sector: A New
Approach', Oxford Economic Papers (Nov 1970).
2 P. Bauer and B. Yamey, 'Economic Progress and Occupational
Distribution', Economic Journal (Dec 1951).
the net value of manufacturing production per head of the total

population, combined, perhaps, with an indicator of the degree of
industrialisation of exports.!
Having said all this, however, the fact remains that there is a
good deal of empirical support for the Fisher-Clark view that the
pattern of development across countries evidences many common
characteristics, especially the shift of resources from agriculture to
industry. Chenery 2 and Maizels,a among others, have documented
the Fisher-Clark thesis in analysing the historical pattern of in-
dustrial growth and trade between countries. Using an estimating
equation of the form log v = log a + b log r, where vis per capita
value added and r is per capita income, it is possible to make
estimates of the income elasticity of demand for different com-
modities (given by b). Taking this basic equation (with some modi-
fications), and applying it to cross-section data for fifty-one countries,
Chenery found that the growth elasticity of the agricultural sector
is less than 0·5, while for industry it is over 1·3 and for services it is
approximately 1·0. Some of the structural transformation suggested
by these estimates may be an exaggeration due to the transference
of some agricultural activities to non-farm enterprises, but even so
it is hard to escape the conclusion that in general resource shifts
out of agriculture are an integral part of the development process
as measured by rising per capita income.
The way industry develops in different countries will depend on a
number of factors including resource endowments, transport costs
and the degree of contact with the outside world, but the broad
development of the industrial sector, as opposed to specific in-
dustrial activities, seems to result from the manner in which demand
changes as income grows. The most noticeable demand shift is the
rapid rise in the demand for capital and consumer durable goods
and a shift away from food, beverages and textiles.
The importance attached to industrialisation by less developed
countries lies in the general association that appears to exist between
industrialisation and real income per head. Technical progress
appears to be more rapid in industry and the scope for economies of
1 This is the approach taken by A. Maizels in his book, Industrial Growth

and World Trade (Cambridge University Press, 1963).
2 H. Chenery, 'Patterns of Industrial Growth', American Economic Review
(Sep 1960).
a Maizels, Industrial Growth and World Trade.
scale greater, the practical importance of which is enhanced by the
higher income elasticity of demand for industrial goods both
domestically and internationally. The scope for productivity in-
crease in industry appears greater overall. In fact, the greater the
proportion of a country's workers in manufacturing, the higher
tends to be the level of manufacturing productivity.
In view of these empirical findings on the pattern of industrial
growth, some would argue that structural transformation in the
present less developed countries ought to be induced rather than
left to take its natural course. Certainly, if the international income
elasticity of demand for agricultural products is less than unity,
there will be growing industrial imports if indigenous industry does
not expand. On the other hand, if there is an extreme comparative
disadvantage in industrial products, real income may rise faster
through the concentration on, and export of, agricultural goods in
exchange for industrial commodities. The problem must still be
faced, though, of the slow growth of demand for agricultural goods
and the possibility that the terms of trade may offset any resource
advantages from specialisation. This is a dilemma in which the
less developed countries find themselves at present, and is a topic
we take up later.
Rostow' s Stages of Growth

Interest in stage theories of development was given new impetus
with the publication ofRostow's book The Stages of Economic Growth,l
which represents an ambitious attempt to provide an alternative to
the Marxist interpretation of history - hence its subtitle 'A Non-
Communist Manifesto'. Rostow presents a political theory as well
as a descriptive economic study of the pattern of the growth and
development of nations. A brief summary of his main points will
provide a useful introduction to the next few chapters on the sources
of growth and development.
The essence of the Rostow thesis is that it is logically and prac-
tically possible to identify stages of development and to classifY
societies according to those stages. He distinguishes five such
stages: traditional; transitional; take-off; maturity; and high mass
consumption.
1 W. W. Rostow, The Stages of Economic Growth (Cambridge University
Press, 1960).
All we need say about traditional societies is that for Rostow the
whole of the pre-Newtonian world consisted of such societies; for
example, the dynasties of China, the civilisations of the Middle
East, the Mediterranean and medieval Europe, etc. Traditional
societies are characterised by a ceiling on productivity imposed by
the limitations of science. Traditional societies are thus recognisable
by a very high proportion of the work-force in agriculture (greater
than 75 per cent), coupled with very little mobility or social
change, great divisions of wealth and decentralised political power.
Today there are very few, if any, societies that one would class as
traditional. Most societies emerged from the traditional stage, as
described by Rostow, some time ago, mainly under the impact of
external challenge and aggression or nationalism. The exceptions to
the pattern of emergence from the traditional state are those coun-
tries which Rostow describes as having been 'born free', such as the
United States and certain British dominions. Here the preconditions
of' take-off' were laid in a more simple fashion by the construction
of social overhead capital and the introduction of industry from
abroad. But for the rest of the world change was much more basic
and fundamental, consisting not only of economic transformation
but also a political and social transition from feudalism.
The stage between feudalism and take-off Rostow calls the transi-
tional stage. The main economic requirement in the transition
phase is that the level of investment should be raised to at least 10
per cent of national income to ensure self-sustaining growth. (On
this particular point, as we shall see, there seems to be very little
difference between the transition stage and the later stage of take-
off.) The main direction ofinvestment must be in transport and other
social overhead capital to build up society's infrastructure. The pre-
conditions of a rise in the investment ratio consist of a willingness
of people to lend risk capital, the availability of men willing and
able to be entrepreneurs and to innovate, and the willingness of
society at large to operate an economic system geared to the factory
and the principle of the division of labour.
On the social front a new elite must emerge to fabric the industrial
society and it must supersede in authority the land-based elite of
the traditional society. Surplus product must be channelled by the
new elite from agriculture to industry, and men must be willing to
take risks and to respond to material incentives. And because of the
enormity of the task of transition, the establishment of an effective
modern government is vital. The length of the transition phase
depends on the speed with which local talent, energy and resources
are devoted to modernisation and the overthrow of the old order, and
in this respect political leadership will have an important part to
play.
Then there is the stage of take-off. The characteristics of take-off
are sometimes difficult to distinguish from the characteristics of the
transition stage, and this has been one bone of contention between
Rostow and his critics. None the less, let us describe the take-off
stage as Rostow sees it - a 'stage' to which reference is constantly
made in the development literature. Since the preconditions of
take-off have been met in the transitional stage, the take-off stage
is a short stage of development during which growth becomes self-
sustaining. Investment must rise to a level in excess of 10 per cent of
national income in order for per capita income to rise sufficiently
to guarantee adequate future levels of saving and investment. Also
important is the establishment of what Rostow calls' leading growth
sectors'. Historically, domestic finance for take-off seems to have
come from two main sources. The first has been from a diversion of
part of the product of agriculture by land reform and other means.
The examples ofTsarist Russia and Meijijapan are quoted, where
government bonds were substituted for the landowner's claim to the
flow of rent payments. A second source has been from enterprising
landlords voluntarily ploughing back rents into commerce and
industry.
In practice the development of major export industries has some-
times led to take-off permitting substantial capital imports. Grain in
the United States, Russia and Canada, timber in Sweden and, to a
lesser extent, textiles in Great Britain are cited as examples. Coun-
tries such as the United States, Russia, Sweden and Canada also
benefited during take-off from substantial inflows offoreign capital.
The sector or sectors which led to the take-off seem to have varied
from country to country, but in many countries railway building
seems to have been prominent. Certainly improvement of the in-
ternal means of communication is crucial for an expansion of mar-
kets and to facilitate exports, apart from any direct impact on such
industries as coal, iron and engineering. But Rostow argues that any
industry can play the role of leading sector in the take-off stage
provided four conditions are met: firstly, that the market for the
product should be expanding fast to provide a firm basis for the
growth of output; secondly, that the leading sector generates

secondary expansion; thirdly, that the sector has an adequate
and continual supply of capital from ploughed-back profits;
and lastly, that new production functions can be continually
introduced into the sector, meaning scope for increased pro-
ductivity.
Rostow contends that the beginnings of take-off in most coun-
tries can be traced to a particular sharp stimulus which has taken
many different forms, such as a technological innovation or more
obviously a political revolution, e.g. Germany in 1848, the Meiji
restoration in Japan in 1868, China in 1949 and Indian indepen-
dence in 1947. Rostow is at pains to emphasise, however, that there
is no one single pattern or sequence for take-off. Thus there is no
need for the developing countries today to recapitulate the course
of events in, say, Great Britain, Russia or America. The crucial
requirement is that the preconditions of take-off are met, otherwise
take-off, whatever form it takes, will be abortive. Investment must
rise to over 10 per cent of national income; one or more leading
sectors must emerge; and there must exist or emerge a political,
social and institutional framework which exploits the impulse to
expansion. The examples are given of extensive railway building
in Argentina before 1914, and in India, China and Canada before
1895, failing to initiate take-off because the full transition from a
traditional society had not been made. The dates of take-off for
some of the present developed countries are given as follows: Great
Britain, 1783-1802; France, 1840-60; the United States, 1843-60;
Germany, 1850-73; Sweden, 1868-90; Japan, 1878-1900; Russia,
1890-1914.
Then there is the stage of maturity which Rostow defines as the
period when society has effectively applied the range of modern
technology to the bulk of its resources. During the period of maturity
new leading sectors replace the old, and Rostow sees the develop-
ment of the steel industry as one of the symbols of maturity. In this
respect America, Germany, France and Great Britain entered the
stage of maturity roughly together.
Accompanying changes in the industrial structure will be struc-
tural changes in society such as changes in the distribution of the
work-force; the growth of an urban population; an increase in the
proportion of white-collar workers; and a switch in industrial
leadership from the entrepreneur to the manager.
Maturity also has important political features. This is the
period when nations grow confident and exert themselves - witness
Germany under Bismarck and Russia under Stalin. This is also the
period when fundamental political choices have to be made by
society on the use to which greater wealth should be put. Should
it be devoted to high mass consumption, the building of a welfare
state, or to imperialist ends? The balance between these possibilities
has varied over time within countries, as well as varying between
countries. Ultimately, however, every nation will presumably reach
the stage of high mass consumption whatever the balance of choices
at the stage of maturity. Since the less developed countries have no
likelihood of reaching this stage in the foreseeable future, however,
and only a handful of countries have reached it already, we shall
not consider this fifth stage here.
Instead, let us evaluate Rostow's thesis, and consider the use-
fulness of this type of stage theory other than in providing a valuable
description of the development process and pin-pointing some of the
key growth variables. Most critics have hinged on whether a valid
and operationally meaningful distinction can be made between
stages of development, especially between the so-called transitional
phase and take-off, and between take-off and maturity. Critics
have attempted to argue that the characteristics that Rostow dis-
tinguishes for his different stages are not unique to those stages. Thus
the demarcation between take-off and transition is blurred because
the changes that take place in the transition phase also seem to take
place in the take-off phase, and similarly with the demarcation
between take-off and maturity.
One of the most outspoken of Rostow's critics is Kuznets, and
some of his criticisms may be quoted as representative of the
criticisms that Rostow has received in general. First, there is the
difficulty of empirically testing the theory, which Rostow himself
makes no attempt to do. For one thing there is a general lack of
quantitative evidence for assertions made, and for another Rostow's
description of the characteristics of some of the stages are not
sufficiently specific to define the relevant empirical evidence even if
data were available. With respect to the take-off stage, for example,
what is a 'political, social and institutional framework which
exploits the impulses to expansion in the modern sector'? Kuznets
argues: 'it seems to me that Rostow ... defines these social pheno-
mena as a complex that produces the effect he wishes to explain and
then treats this identification as if it were a meaningful identi-

fication' .1 Kuznets seems to be calling into question the whole of
Rostow's scientific method and is claiming as unscientific the all too
common practice of observing phenomena, developing hypotheses
on the basis of the phenomena, and then using the phenomena to
support the hypotheses! 2
As regards quantitative evidence that is available for testing
hypotheses, Kuznets questions Rostow's figures of investment and the
incremental capital-output ratio during the take-off period in the
countries studied. He says: 'Unless I have completely misunder-
stood Professor Rostow's definition of take-off, and its statistical
characteristics, I can only conclude that the available evidence
lends no support to his suggestions.'3 And on the concept of the take-
off stage in general Kuznets concludes that lack of common ex-
perience typifying countries in the take-off stage, in relation to
investment, etc., 'casts serious doubt on the validity of the definition
of the take-off as a general stage of modern economic growth, dis-
tinct from what Professor Rostow calls the precondition, or transi-
tion, stage preceding it and the self-sustaining growth stage following
it'. 4 Cairncross echoes these remarks of Kuznets and appears to
deliver a decisive blow when he asks what, if the various stages over-
lap, is the meaning of a 'stage' ?5
Are we to conclude from all this that Rostow's contribution is of
little value? The answer to this must be in the negative; at least,
much can be salvaged. While growth stage theories may be lacking
in analytical power, the purpose of stage theory is not that the stages
distinguished should necessarily have parallels in history, or be
rigidly distinct, but to distinguish the situations in which an economy
may find itself- situations which may merge into one another.
While the concept of a 'stage' may be quibbled with, and stage
theory dismissed as a blueprint for development, Rostow offers
1 S. Kuznets, 'Notes on the Take-off', in W. W. Rostow (ed.), The
Economics of Take-off into Sustained Growth, Proceedings of the Conference of

the International Economic Association (London: Macmillan, 1963);
reprinted in Kuznets, Economic Growth and Structure, p. 219.
2 This, of course, goes on in many branches of economics and the social
sciences, making propositions tautological.
a Kuznets, Economic Growth and Structure, p. 227.
4 Ibid.
5 A. K. Cairncross, 'Essays in Bibliography and Criticism, XLV: The
Stages of Economic Growth', Economic History Review (Apr 1961).
some extremely valuable insights into the development process. As
we have seen from the work of Chenery and Maizels, development
is not entirely haphazard, and there are certain features of the
development process which do follow a well-ordered sequence.
Moreover, there are certain development priorities which countries
planning develgpment may neglect at their peril. The role of in-
vestment in raising the rate of growth is particularly stressed, as
are certain political and sociological preconditions for development
which economists are prone to forget, and which are ignored here.
While emphasis on investment appears to be an unfashionable doc-
trine in the developed countries at present, there exists no satis-
factory counter-argument to the doctrine in less developed countries
if capital is properly defined. IfRostow fails to provide an analytical
breakthrough, he has aroused once again theoretical interest in the
history and causes of the growth of the wealth of nations.
2
The Production Function

Approach to the Study of the
Causes of Growth
The Ana!Jsis of Growth
There are several ways in which the growth of income or output of
a country may be expressed, but frequently they consist of identities
which can tell us very little about the causes of growth without
adequate theorising. For example, growth can be expressed as the
product of a country's ratio of investment to output (IfO) and the
productivity of investment (80/I), i.e.
130 I 130
growth= 0 = 0 X 1 . (2.1)
By definition, slow growth is the product either of a low investment

ratio, or a low productivity of capital, or both. It is this equation
which forms the basis of the view that faster growth in less developed
countries requires more resources for investment, but by itself
this does not constitute a theory of development.
Alternatively, income or output can be expressed as the product
of the total labour force (L) and output per unit of labour (0/L),
so that the growth of output can be expressed as the sum of the rate
of growth ofthe work-force (8LfL) and the rate of growth of output
per unit oflabour, or labour productivity, (8(0/L)f(OfL)), i.e.
80 8L
growth = 0 = L + 8(0/L)
(OfL) • (2.2)
In this formulation slow growth is attributable by definition either

to a slow growth of the work-force, or to a lagging rate of growth of
labour productivity, or both. Expressing growth in this way high-
PRODUCTION FUNCTION STUDY OF CAUSES OF GROWTH 39
lights the dependence of the growth of income per unit of labour
(and, more important still, the growth of per capita income) on the
growth of labour productivity, but again this approach does not
constitute a theory of development. How does labour productivity
grow? Is it by capital accumulation, or is it by technical progress
broadly defined to include such factors as improvements in the
quality oflabour, improvements in the quality of capital, economies
of scale, advances in knowledge, a better organisation of capital
and labour in the productive process and so on? Growth identities
of the type outlined cannot distinguish between such competing
hypotheses. What is required is a testable model of the growth
process.
The production function approach to the measurement of growth
is a response to this challenge. It is a method of approach to facili-
tate an understanding of the sources of growth, and to quantity
the contribution of these sources to any measured growth rate. The
approach has been used extensively and usefully in developed coun-
tries, and some results and conclusions from these studies will be
given later. It is now beginning to be employed in the context ofless
developed countries, although some would say not very fruitfully
because the major sources of growth in less developed countries are
institutional and non-measurable. The justification for introducing
the approach here is twofold. Firstly, since it is a technique that is
likely to be used more and more, students ought to be familiar with
it, and aware of its limitations. Secondly, and more important, if
discussions of development are to advance beyond the stage of anec-
dote, hunch and opinion, growth and development must be placed
in an analytical framework so that hypotheses can be advanced
subject to the possibility of testing. The production function
approach is a first step in this direction. Even if growth is primarily
a question of institutional change, which is doubtful, it is still useful
to know the rough contribution that the growth of, say, capital and
labour has made to measured growth in the past, if only to put the
growth debate in perspective. There are problems of calculation and
estimation but these are not insuperable. Apart from the difficulties
of specifYing the sources of growth precisely, and measuring accu-
rately both the dependent and independent variables, the main
problem is a methodological one of fitting the appropriate production
function to the data; that is, the function relating output to
inputs.
The Production Function

One of the desirable properties of any macro-economic hypothesis,
apart from being consistent with the observed facts, is that it should
be consistent with, and derivable from, micro-economic theory.
What we are calling the production function approach to the analysis
of growth in the aggregate possesses, in part, this desirable property
in that it borrows the concept of the production function from the
theory of the firm. Just as we can say for a firm that output is a func-
tion of the factors of production -land, labour, capital and the level
of technology (or factor efficiency) - so we can write aggregate
output as a function of factor inputs and the prevailing tech-
nology, i.e.
0 =f(R, K, L, T) (2.3)
where R is land, K is capital, Lis labour and Tis technology.
The question is how to separate empirically the contribution to
growth of the growth of factor inputs from other factors which can
lead to higher output, included in T, such as economies of scale (due
both to technical change and to increases in factor supplies), im-
provements in the quality of factor inputs, advances in knowledge,
better organisation of factors, and so on. The task is to fit an appro-
priate, correctly specified, production function which, if possible,
will not only separate the contribution offactor inputs to growth from
the contribution of increases in output per unit of inputs (increases
in 'total' factor productivity) but will also distinguish between
some of the factors that may contribute to increases in the pro-
ductivity of factors such as education, improvements in the quality
of capital and economies of scale.
Before going on to discuss the types of function that may be
employed, however, let us examine in a little more detail the proper-
ties of a production function. We have established so far that the
aggregate production function expresses the functional relation
between aggregate output and the stock of inputs. If land is sub-
sumed into capital, and technology is held constant, we are left with
two factors and the production function may be drawn on a two-
dimensional diagram as in Fig. 2.1. Capital (K) is measured on the
vertical axis and labour (L) on the horizontal axis, and each func-
tion represents a constant level of output that can be produced with
different combinations of capital and labour. The functions slope
negatively from left to right on the assumption that marginal
additions of either factor will increase total output - that is, factors
have positive marginal products - and they are drawn convex to
the origin on the assumption that factors have a diminishing marginal
productivity as their supply increases so that if one unit is with-
drawn it needs to be substituted by more and more of the other
factor to keep output constant. The position of the functions broadly
reflects the level of technology. The more' advanced' the technology,
0~----------------L·
FIG. 2.1
the greater the level of output per unit of total inputs and the closer
will a production function of a given value lie to the origin.
From the simple production function diagram it is easy to see
how output may increase. First, there may be a physical increase in
factor inputs, L and K, permitting a higher level of production.
Either or both factors may increase. If only one factor increases,
the movement to a higher production function will involve a change
in the combination offactors and output will not be able to increase
for ever because ultimately the marginal product of the variable
factor will fall to zero. This is illustrated in Fig. 2.2 where, with a
given stock of capital OK1, output cannot increase beyond 300
with increases in the supply of labour ( OL1, OL2, etc.) beyond the
limit indicated. The diminishing productivity of the variable factor,
labour, with capital fixed, is shown by the flatter and flatter slope
of the production functions at successive points, Lt, L2, until at the
limit the production function is horizontal and the marginal product
of labour is zero.
If both factors increase in supply, however, there is no reason why
output should not go on increasing indefinitely. In fact, if both
factors increase in supply there is the possibility that production
may be subject to increasing returns such that output rises more than
proportionately to the increase in combined inputs. If this is the
case, output per unit of total inputs will increase and the pro-
duction functions representing equal additional amounts of pro-
duction, e.g. 100, 200, 300, etc., must be drawn closer and closer
together as in Fig. 2.3.
Kll-;~\c-_;:,,_ _ _ _ 300
--+----200
----~----100
o~-L~,--L~2--,L~im~it~-------,L
FIG. 2.2
In the opposite case of decreasing returns, the functions would be

drawn further and further apart. Finally, in the case of production
subject to constant returns the functions would be drawn equi-
distant from one another.
Increasing returns may also result from advances in technology,
K
0 L
FIG. 2.3
irrespective of increases in factor supplies. These are called tech-

nological economies of scale. In this case increases in output per
unit of input would have to be represented on a production function
diagram either by a relabelling of the functions or a relabelling of
the axes. That is, either the same amount of factor inputs, measured
on the axes, would have to be shown to be producing a higher out-
put than before, or alternatively the same output could be shown to
be produced by lesser amounts of inputs. If the functions are re-
labelled and not the axes, this is tantamount to a shift in all the
production functions towards the origin. Shifts in the production
function towards the origin are implied by all forms of technical
progress or any factor which increases the productivity of the physi-
cal inputs.
In short, there are three broad sources of growth that can be
distinguished using the production function framework: first, in-
creases in factor supplies; second, increasing returns; and third,
technical progress interpreted in the wide sense of anything
that increases the productivity of factors other than increasing
returns.
The Cobb-Douglas Production Function

The production function most commonly fitted to aggregate data
to distinguish empirically between these three broad sources of
growth has been the unconstrained form of the Cobb-Douglas
production function, named after its two American originators,
Charles Cobb (a mathematician) and Paul Douglas (an economist),
who pioneered research in the area of applied economic growth in
the 1920s and 1930s.l The Cobb-Douglas function may be written
as
(2.4)
where Ot is real output at time t
Te is an index of technology, or 'total' productivity
Kt is an index of the capital stock, or capital services, at
constant prices
Lt is an index oflabour input (preferably man-hours)
ex is the partial elasticity (responsiveness) of output with
respect to capital (holding labour constant)
and (3 is the partial elasticity of output with respect to labour
(holding capital constant).
The assumption is that changes in technology are exogenous and
independent of changes in factor inputs, and the effect of technical
progress is neutral on the factor intensity of production. 2 Tt, ex and
(3 are constants to be estimated empirically if the function is un-
constrained. If ex and (3 are assigned values in advance of the use
1 C. Cobb and P. Douglas, 'A Theory of Production', American Economic
Review, supplement (Mar 1928).
2 See p. 108 for a definition of neutral technical progress.
of the function for estimating purposes, the function is said to be

constrained. Normally, oc and ~will be less than unity on the assump-
tion of diminishing marginal productivity of factors. The sum of the
partial elasticities of output with respect to the factors of production
gives the scale of returns, or the degree of homogeneity, of the
function. a + ~ = 1 represents constant returns, oc ~ > 1 +
represents increasing returns, and oc +
~ < 1 represents decreasing
returns, and the function is said to be homogeneous of degree one,
greater than one and less than one, respectively.
If a and ~ are not estimated empirically but are assumed to sum
to unity, in which case the function would be constrained, then
increasing or decreasing returns will be reflected in the value of Tt
which is the index of total factor productivity. Obviously the
existence of increasing returns would bias the value of Tt upwards,
and decreasing returns would bias the value of Tt downwards.
These points are made because in practice the Cobb-Douglas
function is often employed in this constrained form with the sum of
oc and ~ put equal to unity. Then values are assigned to oc and ~
according to the share of capital and labour in the national income.
The underlying assumption is the perfectly competitive one that if
production is subject to constant returns and factors are paid the
value of their marginal products, then factor shares will reflect the
value of a factor's marginal product. Despite the fact that developed
and less developed economies alike are far removed from perfect
competition, it is interesting to note that when oc and ~ have been
estimated empirically, values have frequently been obtained which
do not diverge markedly from the estimates of factor shares of the
national product. This is used, in fact, as a justification for assigning
values to oc and ~ on the basis of factor shares to save the trouble of
making empirical estimates.
To use equation (2.4) for separating out the influence of the
three broad sources of growth mentioned earlier, we must first take
logarithms of the variables and then differentiate with respect to
time. This gives
dlogOt dlog Tt + dlog Kt + ~ dlog Lt (2.5)

dt dt IX dt dt •
For simplicity, equation (2.5) may be written
ro = rw + arK + ~rL (2.6)

where ro is the rate of growth of output per time period
rT is the rate of growth of total productivity, or technical
progress
fJ( is the rate of growth of capital
rL is the rate of growth of labour

and IX and ~ are the partial elasticities of output with respect
to capital and labour, respectively, as before.

In words, equation (2.6) says that the rate of growth of output is
equal to the rate of growth of' total' productivity plus the rate of
growth of capital weighted by the partial elasticity of output with
respect to capital plus the rate of growth of labour weighted by the
partial elasticity of output with respect to labour. With knowledge
of ro, rK, rL, IX and ~' it becomes possible as a first step to separate
out the contribution of factor inputs to growth from increases in
output per unit of inputs represented by rT. Now let us give a
hypothetical example. Suppose ro = 5 per cent per annum; rK = 5
per cent per annum; rL = 1 per cent per annum; and IX= 0·25
and~ = 0·75 (decided on the basis offactor shares). Substituting in
equation (2.6) we have
5·0 = rT + 0·25 (5·0) + 0·75 (1·0). (2.7)
The contribution of capital to measured growth is 0·25 (5·0) = 1·25
percentage points; the contribution of labour is 0·75 (1·0) = 0·75
percentage points; and rT is left as a residual with a contribution of
3·0 percentage points. If IX and ~ were estimated empirically, and
there happened to be increasing returns (IX + ~ > 1), the signifi-
cance of the factor contribution would be enhanced and rT would be
smaller.
Although rT has been variously called technical progress, ad-
vances in knowledge, etc., definitionally it is that portion of the
growth of output not attributable to increases in the factors of
production, and includes the effects not only of the multifarious
factors which go to increase the productivity of labour and capital
but also measurement errors in the capital and labour input series.
rT is perhaps best described as a residual, or, perhaps more appro-
priately still, a 'coefficient of our ignorance' if the analysis proceeds
no further. One important component ofrT, which can be considered
the result of measurement errors, is likely to be the effects of resource
shifts from less productive to more productive activities. The analysis
is so aggregative that there is bound to be a confounding of changes
in actual output with changes in the composition of output unless

the weights used for aggregating inputs are continually revised.
Resource shifts from agriculture to industry could be expected to
figure prominently in any production function study ofless developed
countries, as they do for studies of many advanced economies.
Before considering some of the results of applying the Cobb-
Douglas function to empirical data, we must briefly mention some
of its limitations. Its use has come under attack on four main
counts. The first criticism is that since only one combination of
factor inputs can be observed at any one time, it is impossible to
distinguish shifts in the function (technical progress) from move-
ments along the function (changes in factor intensity) unless the
assumption of neutral technical progress is made. But technical
progress may not be neutral and therefore the effects of technical
progress and changing factor intensity become confused, biasing the
results of the contribution of factor inputs and technical progress
to growth. Secondly, the assumption that technical progress is
independent of increases in factor inputs has been questioned. This
is not a specification error of the function itself, however, and the
Cobb-Douglas function can be used making technical progress a
function of the rate of growth of in puts - so-called endogenous models
of technical progress. Thirdly, the Cobb-Douglas function cannot
represent a change in the ease of substitution between capital and
labour because it contains the restrictive property of constant
unitary elasticity of substitution between factors whatever the factor
intensity .I
1 The elasticity of substitution (a) relates the proportional change in
relative factor inputs to a proportional change in the marginal rate of
substitution between labour and capital (M.R.S.) (or the proportional
change in the relative factor-price ratio on the basis of marginal productivity
theory). The elasticity of substitution may therefore be written as
lllog (L[K)
a =ll log M.R.S ..
The proof that a = 1 is very simple:
llO/IlO rxL
M.R.S. = IlK IlL = {3K
rx L
log M.R.S. = log ~ + log ]('
Differentiating with respect to log M.R.S. gives
I = lllog (L/K) = a.
lllog M.R.S.
This may be serious if the elasticity of substitution of factors differs
significantly from unity and there are wide discrepancies in the
growth rate of factors. For example, if the elasticity of substitution
between capital and labour is significantly less than unity and capital
grows faster than labour, this will result in an overestimate of the
contribution of capital to growth and a resulting underestimate of
the role of other factors. The simple explanation of this bias is that
the smaller the elasticity of substitution the more difficult it is in
practice to obtain increased output just by increasing one factor
because diminishing returns set in strongly. By assuming the
elasticity is higher than it is, the importance of the fastest-growing
factor is exaggerated. If elasticity is high, diminishing returns are
not a problem, and if both capital and labour expand at the same
rate, growth is obviously independent of the elasticity of substi-
tution.l
A final criticism relates to the measurement of output and inputs.
What, argue some, is the meaning of a function which aggregates
so many heterogeneous items; in particular, what is the meaning of
an aggregation of capital goods built at different times, at different
costs and with varying productivities? How are such capital goods
to be equated in an aggregate measure of capital?
By and large, all the above-mentioned criticisms are theoretical
worries, the practical significance of which is minimal. None is so
damaging as to suggest the abandonment of the function. Studies
of the nature of technical progress, at least in advanced countries,
suggest that the assumption of neutrality is a fair working hypo-
thesis. The fact that technical progress may be dependent on factor
accumulation is easily accommodated within the Cobb-Douglas
framework. Capital and labour would have to grow at very different
rates for the elasticity of substitution to matter very much, but in any
case studies show that it is quite close to unity. Finally, although the
aggregation of heterogeneous outputs and inputs is a worry, especially
1 To overcome the restrictive property of the Cobb-Douglas function

when the growth rates of factors do differ, it has become fashionable in
recent years to use the more general constant elasticity of substitution pro-
duction function, of which the Cobb-Douglas is a special case. We cannot
discuss the function here except to say that it, too, is not without its specifica-
tion errors. The assumption of constancy has the drawback that one may be
ascribing changes in elasticity to changes in technology which are really due
to changes in factor proportions. This limitation can only be overcome with
a function possessing the property of variable elasticity of substitution.
the aggregation of capital, none the less capital and labour aggre-
gates seem to perform well in production function studies as measured
by the correct picture they convey of relative factor shares.
What have been the results of applying the Cobb-Douglas func-
tion to empirical data? First, let us consider its application in de-
veloped countries and consider the conclusions that emerge. We can
start with the pioneer work of Cobb and Douglas themselves. Ironic-
ally, the Cobb-Douglas function as first conceived was not intended
as a device for distinguishing the sources of growth but as a test of
neo-classical marginal productivity theory; that is, to see whether
marginal products corresponded to factor shares. Douglas had
observed that the product curve for American manufacturing in-
dustry for the period 1899-1922 lay consistently between the two
curves for the factors of production, and he suggested to his mathe-
matician friend, Cobb, that they should seek to develop a formula
which could measure the relative effect oflabour and capital on the
growth of output over the period in question. This story is described
by Douglas in his fascinating review article 'Are There Laws of
Production?' ,1 and as an insight into inductive method the relevant
passage is worth quoting in full:
Having computed indexes for American manufacturing of the

number of workers employed by years from 1899 to 1922 as well
as indexes of the amounts of fixed capital in manufacturing
deflated to dollars of approximately constant purchasing power,
and then plotting these on a log scale, together with the Day
index of physical production for manufacturing, I observed that
the product curve lay consistently between the two curves for the
factors of production and tended to be approximately one-quarter
of the relative distance between the curve of the index for labour,
which showed the least increase in the period, and that of the
index of capital which showed the most. I suggested to my friend
Charles Cobb that we seek to develop a formula which could
measure the relative effect of labour and capital upon product
during this period. At his suggestion the sum of the exponents was
tentatively made equal to unity in the formula 0 = TK"-Ll-rt.
[our notation] .... The fact that on the basis of fairly wide
studies there is an appreciable degree of uniformity, and that the
1 P. Douglas, 'Are There Laws of Production?', American Economic Review
(Mar 1948).
sum of the exponents approximates to unity, fairly clearly sug-
gests that there are laws of production which can be approximated
by inductive studies and that we are at least approaching them.l
The actual function derived was 0' = 1·01 K0·25 L0·75 which con-
firmed neo-classical predictions, but there was no discussion of the
relative importance of factors of production and the T parameter in
accounting for measured growth. It was not until Abramovitz in
19562 and Solow in 19573 showed that between 80 and 90 per cent
of the growth of output per head in the American economy over the
century could not be accounted for by increases in capital per head
that the production function started to be used in earnest as a tech-
nique in the applied economics of growth. Abramovitz remarked:
This result is surprising in the lop-sided importance which it

appears to give to productivity increase and it should be, in a
sense, sobering, if not discouraging to students of economic
growth. Since we know little about the causes of productivity
increase, the indicated importance of this element may be taken
to be some sort of measure of our ignorance about the causes of
economic growth in the United States, and some sort ofindication
of where we need to concentrate our attention.4
Abramovitz's findings were supported by Solow who found, in

examining the data for the non-farm sector of the American economy
for the period 1919-57, that approximately 90 per cent of the
growth of output per head could not be accounted for by increases
in capital per head, i.e. from equation (2.6)
rpf(ro- TL) = 0·90 (2.8)
where rp = ~=~~-a;~ (which is Solow's estimating function),

OL is output per man, KL is capital per man, and the dot represents
a time derivative.
The findings of Abramovitz and Solow disturbed economists
1 Ibid., p. 20.
2M. Abramovitz, 'Resource and Output Trends in the United States
since 1870', American Economic Review, Papers and Proceedings (May 1956).
a R. Solow, 'Technical Change and the Aggregate Production Function',
RB11iew of Economics and Statistics (Aug 1957).
4 Abramovitz, op. cit., p. 11.
brought up in the belief that investment and capital accumulation

played a crucial role in the growth process. Even allowing for the
statistical difficulties of computing a series of the capital stock, and
the limitations of the function applied to the data (e.g. the assump-
tions of constant returns and neutral technical progress, plus the
high degree of aggregation), it was difficult to escape from the
conclusion that the growth of the capital stock was of relatively
minor importance in accounting for the growth of total output. It is
true that Abramovitz had stressed that his findings did not imply
that resources were unimportant for growth, because of the inter-
relation between the growth of inputs and factors leading to
increases in output per unit of inputs, but this caution was not suffi-
cient to counter the popular conclusion that capital does not matter.
It would not be misleading to say that much of the subsequent
research effort in this field of growth has been designed to reverse
this conclusion, or rather to 'assign back' to the factors of pro-
duction sources of growth which make up the residual factor but
which are interrelated with, or dependent on, the growth of factor
inputs. Work has proceeded on two fronts. On the one hand, attempts
have been made to disaggregate the residual factor, measuring factor
inputs in the conventional way; on the other hand, attempts have
been made to adjust the labour and capital input series for such
things as changes in the quality of factors and their composition so
that much more measured growth is seen to be attributable to
increases in factor inputs in the first place. For example, the labour
input series has been adjusted for improvements in its quality due
to the growth of education, and for changes in its composition due
to agefsex shifts. Likewise, the capital stock series has been adjusted
to reflect changes in its composition and, more important, to allow
for the fact that new additions to the capital stock in any line of
production are likely to be more productive than the existing capital
stock as a result of technical advance. This is the notion of embodied
or endogenous technical change as opposed to the exogenous tech-
nical change assumption of the original Cobb-Douglas function
which assumes that all vintages of capital share equally in technical
progress. A distinction is now made, therefore, between embodied
and disembodied technical progress - embodied technical progress
referring to technical improvements that can only be introduced
into the productive system by new investment, and disembodied
technical progress which is exogenous and not dependent on
capital accumulation. There are several ways in which embodied
technical progress can be isolated from the residual factor by appro-
priate adjustments to the capital stock series to reflect the greater
productivity of the latest investments. The net result is to increase
the sensitivity of growth to changes in the capital stock, and the role
of capital accumulation in the growth process does not appear
so derisory.
Efforts have also been made towards overcoming the aggre-
gation problem by taking explicit account of shifts of labour and
capital from low-productivity to high-productivity sectors. This, too,
reduces the significance of the residual factor and makes the role
of labour and capital in the growth process look correspondingly
more respectable.
Let us now look in greater detail at the modifications to the
Cobb-Douglas function which can be made to allow for the quality
changes discussed above, and then consider some of the empirical
evidence.
Embodied Technical Progress

If capital is measured net at constant prices, and all ages or vin-
tages of capital are treated alike, technical change associated with
capital investment becomes part of the residual factor in the growth
equation. The ultimate effect of adjusting the capital stock series
for embodied or endogenous technical change is to raise the sen-
sitivity of the growth rate to changes in the growth of capital. The
question is, how should the adjustment take place, and how to
assess the relative importance of technical change that depends on
the growth of capital compared to conventional hypotheses of the
role of capital? Experimenting with the embodied technical pro-
gress hypothesis is, in fact, a complicated procedure because measures
of the capital stock can only be properly corrected for the effect of
technical change if the rate of progress is known. Since this rate, in
general, is unknown, it is necessary to work by a process of trial and
error. Applied studies can be divided into those which attempt to
estimate a fairly precise rate of embodied technical progress and
those which simply try to assess its relative importance. As far as the
latter object is concerned, one device is to measure capital gross at
current prices rather than net at constant prices. If capital is
measured in this way, technical change in capital should be implicit
in the price variable leaving disembodied technical change in the

residual factor. Alternatively, an exponential time trend can be
added to the traditional form of the production function, repre-
senting a constant rate of productivity advance, and this may be
called disembodied technical progress. The residual would then
consist of the effects of embodied progress. The difficulty here is that
embodied technical progress may also grow exponentially if the
growth of the capital stock itself is fairly constant.
A third approach, and one which in theory allows a more precise
measure of the rate of embodied technical progress, is the so-called
vintage approach to the measurement of capital. The procedure
here is more formal and involves a consideration of the model
representing the embodiment hypothesis. The vintage approach to
the consideration of embodied technical progress is most associated
with the name of Professor Solow. He was the first to present the
economic theory upon which the vintage production function is
based, and was among the first to modifY the basic Cobb-Douglas
function to allow for embodied technical change and to estimate its
growth. Basically, the approach consists of giving a separate valua-
tion to each year's addition to the capital stock, with a higher weight
being assigned to the most recent and presumably the more pro-
ductive additions. The problem is deciding the weights.
Solow's original modell has estimation complications, but Nelson2
has produced a similar model incorporating the same features
where the 'effective' capital stock is given as a function of the gross
capital stock, its average age, and the rate of productivity improve-
ment of new capital goods. Denoting the 'effective' capital stock
as -r, the Cobb-Douglas function as modified for changes in the
quality of capital may be written as
(2.9)
where -r is the quality-weighted sum of capital goods

and T' is now an index of total productivity excluding the effect
of technical progress embodied in new capital.
1 R. Solow, 'Investment and Technical Progress', inK. Arrow, S. Karlin

and P. Suppes (eds), Mathematical Methods in the Social Sciences (Stanford
University Press, 1960).
2 R. Nelson, 'Aggregate Production Functions and Medium Range
Growth Projections', American Economic Review (Sep 1964).
Assuming that technical progress improves the quality of new
machines at a constant rate per annum (AK), then
I
Tt = ~ Kvt (1
0
+ AK) 11 (2.10)
where K vt is the amount of capital built in year V (of vintage V)

which is still in use in timet. (Kvt is gross capital ofvintage V, and
the variable Tt is thus an integral value of capital with different
vintages.)
If the rate of growth of capital changes, this will alter the age dis-
tribution of capital and this, too, will affect the productivity of
capital in that the gap between the average technology and the
best-practice techniques will be changing. A decrease in the average
age of capital will improve the productivity of capital by an amount
equal to -'AK8A, where a.A is the change in the average age of
capital (8A is negative if the capital stock is getting younger owing to
a faster accumulation of capital).
The rate of growth of the 'effective' capital stock may therefore
be written as
(2.11)
where 8Kj K is the rate of growth of the actual capital stock
'AK is the rate of growth of improvement in the capital

stock
and 'A1,a.A is the effect of changes in the average age of the

capital stock which is a function of the investment
ratio.
The Cobb-Douglas function with embodied technical progress may

now be written in estimating form as
(2.12)
where r denotes the rate of growth of variables over time.
Equation (2.12) is derived in exactly the same way as equation (2.6)

and can be interpreted in the same way.
But the question remains, how much of the measured growth

rate ro is due to embodied technical change, how much to disem-
bodied, exogenous forces, and how much to changes in the age
distribution of capital if the average age is changing? If we know
the rate of growth of total productivity and the age distribution of
capital, and all technical progress is assumed to be embodied, then
AK and the effects of changing age distribution can be calculated.
But it is clearly unrealistic to assume that all technical progress is
embodied. How, then, is the rate of embodied progress to be esti-
mated using the vintage approach? The method commonly em-
ployed is to experiment with different rates of embodied technical
progress and, on a trial and error basis, choose that rate which gives
the best statistical fit when the function is estimated using empirical
data for the other variables. Needless to say, the technique is arbi-
trary, but given values for J...K and J...Ka.if, the sensitivity of output
with respect to capital is increased. Improvements in the quality of
capital can be regarded as equivalent to physical increases in the
quantity of capital of a few extra per cent. The contribution of
capital to growth is seen to be greater.
As far as the empirical evidence is concerned, however, there
seems to be some difference of opinion among investigators as to
the importance of the embodiment hypothesis. Nelsonl finds that
virtually the whole of variations in the growth of productivity in
America in the twentieth century can be accounted for by variations
in the average age of the capital stock, as opposed to variations in
AK even if full embodiment is assumed. On this finding the pro-
portion of the total output of a country that is invested is crucial to
growth, since the average age of capital is an inverse function of the
investment ratio. On the other hand, Denison has argued on
several occasions 2 that the embodiment effect operates solely through
the age distribution of capital which is subject to very small variation
and cannot be important in practice. But it is not only changes in
the average age of the capital stock that distinguish the 'new view'
of investment from the traditional assumption that all progress is
disembodied. The average age of capital may remain the same,
but at the same time more productive machines may be replacing
1 R. Nelson, 'Aggregate Production Functions and Medium Range

Growth Projections', American Economic Review (Sep 1964).
2 E.g. E. Denison, 'The Unimportance of the Embodied Question',
American Economic Review (Mar 1964).
those that are wearing out. Several studies for America, which have
attempted to estimate embodied technical progress by the trial and
error procedure mentioned, arrive at annual improvement rates
of between 2 and 5 per cent.
Improvements in the Qualiry of Labour

Criticism of the inadequate treatment of changes in the quality of
capital apply equally to labour. To what extent may the role of
labour in the growth process be underestimated, and the amount
of 'unexplained' growth exaggerated, by ignoring changes in the
quality of labour? A model analogous to that embodying technical
progress in capital can be developed which embodies quality im-
provements in labour, although it is not strictly necessary for new
additions to the labour force to be more productive than the
average for the average quality to increase. The sorts of factors that
increase the personal efficiency of labour, and labour productivity,
operate, in general, in a disembodied way.
If we denote the improved quality oflabour as qL, where q stands
for the improvement in the productive efficiency oflabour, changes
in the quality of labour are accommodated by writing the Cobb-
Douglas function as
(2.13)
where T* is an even narrower concept than T' by excluding from
the residual improvements in the quality of labour as well as
capital.
q can stand both for an improvement in the average quality of
labour, and an improvement in the productive efficiency of new
workers due to such things as education or training. If q expresses
an improvement in the productive efficiency of new workers, it is
here that there is an analogy with the vintage model of capital, and
it becomes necessary to express the production function not only in
a form recognising an improvement in the average quality of labour
but which also takes account of changes in the age composition of
labour. Let I..L be the yearly rate of improvement in the average
efficiency oflabour and 8E the change in the average age of the work-
force. We can then write
(2.14)
Like T, the growth of'effective' labour input consists of three parts:

the rate of growth of labour input in physical units (8L/L); the
average rate of growth of its improvement (:AL) ; and the effect of
changes in its average age (!.L8E).
The Cobb-Douglas function, adjusted for changes in the quality
of both capital and labour, now becomes (in estimating form)
The residual term, rp•, is now the rate ofgrowth of total productivity
or technical progress independent of increases in factor inputs. The
effect of making allowance for improvements in the quality oflabour
and changes in its average age is exactly the same as in the case of
capital- to increase the sensitivity of output growth to the growth
of the labour force and to reduce the size ofthe residual factor. The
two most important factors affecting the quality of labour in any
economy are work experience (or learning), which primarily im-
proves the average quality of labour, and formal education and
training which may exert their effect both through changes in the
average quality of labour and its 'age' distribution if education and
training expand, and primarily affect new workers. We shall dis-
cuss education and learning more fully when we consider again
capital and technical progress in Chapter 4.
Resource Shifts
In addition to improvements in the quality of inputs, an important
source of productivity growth may be resource shifts from less pro-
ductive to more productive activities. Unless the weights for aggre-
gating capital and labour are continually revised to reflect their
changing productivities in different occupations, the effects of
resource reallocation will appear independent of the factors of pro-
duction whereas, in practice, growth from this source is intimately
bound up with factor endowments. One measure of the importance
of resource shifts is to take a weighted average of the rates of' tech-
nical progress' within different individual industries and to subtract
this from the aggregate measure of technifal advance, leaving the
difference as a measure of productivity advance due to shifts of
resources between industries. This is not an ideal measure because the
difference is bound to contain the effects of other 'errors' and
omissions, but it does give some idea of the effects of aggregation.
When Massell adopted this procedure in a study of nineteen
American manufacturing industries over the post-war years, he
found that approximately 30 per cent of the aggregate rate of
technical advance was the result of aggregation. Approximately 0·1
percentage point of growth was due to labour shifts and 0·8 per-
centage points to capital shifts.l Denison has made estimates of a
similar order of magnitude for some countries of Western Europe
considering just shifts of resources from agriculture to industry.2
From 1950 to 1962 resource shifts from agriculture to industry con-
tributed 1·04 percentage points per annum to the growth rate of
Italy, 0·76 percentage points in Germany and 0·65 percentage points
in France. Indeed, it is these shifts which account in large part for the
difference in the post-war growth performance of continental
Europe on the one hand and the United States and Great Britain on
the other. The potential scope for growth from this source in less
developed countries must be enormous.
Empirical Evidence
Since Abramovitz and Solow reported their findings in 1956 and
195 7, a fairly substantial body of empirical evidence relating to the
sources of growth has accumulated experimenting with different
specifications of the aggregate production function. Unfortunately,
it is not systematic. The time periods taken, the data used, the
sectors of the economy examined, and the methodology employed,
all vary within and between countries. All that can be done here is
to draw attention to some of the major findings and let the reader
check for himself the nature of the work.3 Until recently most of the
evidence available pertained to fairly advanced economies and it is
largely from this evidence, wisely or not, that conclusions have been
drawn on development strategy for less developed countries. Re-
search in developing countries has been hampered by a lack of
researchers, a shortage of reliable empirical data, and perhaps an
even greater suspicion of the aggregate production function, and its
implicit assumptions, than in developed countries. The assumption
1 B. Massell, 'A Disaggregated View of Technical Change', Journal of
Political Economy (Dec 1961).
2 E. Denison, Why Growth Rates Differ: Postwar Experience in Nine Western
Countries (Washington: Brookings Institution, 1967).
3 For a comprehensive survey, see C. Kennedy and A. P. Thirlwall,
'Surveys in Applied Economics: Technical Progress', Economic Journal
(Mar 1972).
that factor shares measure the relative contribution of labour and

capital to growth is probably more dubious in less developed than
in developed countries. The share of labour in total income almost
certainly exceeds its marginal product, while the share of capital
falls short of it. Secondly, the aggregation of inputs and outputs is
generally more difficult, and there are greater problems of resource
underutilisation to contend with. The recent past, however, has
witnessed a sprinkling of production function studies for less
developed countries,l and there are undoubtedly many more to come
as the fashion spreads. Moreover, the production function, despite
its drawbacks, yields highly useful and verifiable hypotheses.
The studies for advanced countries tend to confirm the relative
unimportance of capital compared with other growth-inducing
variables. Even allowing for changes in the composition of capital
and embodiment, capital growth rarely accounts for more than one-
half of the measured growth of output. Denison, who adjusts
the capital stock for changes in its composition, estimates a con-
tribution of approximately 25 per cent in the United States over
the period 1950-62 and just under 20 per cent in North-west
Europe over the same period.2 Solow, using an embodied model,
finds the weighted contribution of embodied technical progress for
plant and machinery less than that of disembodied progress. a One
noticeable exception to this general rule is the case of Israel, where
Gaathon finds that the growth of capital per head accounted for
60 per cent of the annual average growth per head over the period
1950-9.4 Perhaps this is a good example ofthe type of country that
Hicks has in mind when he says that 'it is very wrong to give the
impression to a [developing] country, which is very far from
equilibrium even on past technology, that capital accumulation is
a matter of minor importance'. 5 We shall discuss later good economic
reasons why the growth of capital may be more important for growth
in developing, as opposed to more mature, economies.
1 See J. G. Williamson, 'Production Functions, Technological Change,
and the Developing Economies: A Review Article', Malayan Economic
Review (Oct 1968).
2 Denison, Why Growth Rates Differ.
a R. Solow, 'Technical Progress, Capital Formation and Economic
Growth', American Economic Review, Papers and Proceedings (May 1962).
4 A. Gaathon, Capital Stock, Employment and Output in Israel, 1950-1959
(Jeruaalem: Bank oflsrael, 1961).
5 J. Hicks, Capital and Growth (Oxford University Press, 1965) p. 304.
Israel experienced a rapid rate of capital formation which un-
doubtedly offset some of the biases against capital inherent in the
production function framework. For equal additional increases in
capital and labour the factor of production showing the biggest
contribution to growth will be the one with the largest partial
elasticity. If labour and capital grow at the same rate, and ex and [3
are decided on the basis of shares of the national income, the con-
tribution of capital to growth will always appear less important than
labour simply because its share of the national income is smaller.
This bias is greater the greater the degree of structural disequili-
brium in the labour market with wages in excess of marginal pro-
duct. Furthermore, when ex and [3 are estimated empirically, there
is also a tendency for the partial elasticity of output with respect to
capital to be biased downwards compared with the elasticity for
labour. The reason is that in estimating ex by regression techniques,
output is normally related to the capital stock or capacity rather than
capital utilisation. Output is obviously less sensitive to actual cap-
acity (which is relatively fixed) than to the utilisation of capacity.
The elasticity of output with respect to labour, on the other hand,
can be more easily related to a measure of labour utilisation by
taking statistics of man-hours worked. It is interesting to note that
a study by Brown and de Cani, which took the utilisation of capital
as the measure of capital input, estimated an elasticity of output
with respect to capital of0·739, which is much higher than normally
estimated or employed in production function studies.l It is to be
hoped that in the course of estimating production functions in the
future much more attention will be paid to capacity utilisation, and
also to possible discrepancies between the marginal products of
factors and factor shares, especially in developing countries.
In general, the growth of the labour force has contributed as much
to growth as capital in developed countries, and probably more in
the less developed countries if embodied technical progress is ig-
nored. Labour-force growth is neutral, of course, on the growth of
income per head except to the extent that there is a relation between
the growth of the work-force and the growth oflabour productivity.
There are conflicting views on this matter and we discuss these
later in considering the so-called 'population problem'. The im-
portance of labour is considerably enhanced when the growth of
1M. Brown andJ. de Cani, 'Technological Change in the U.S., 1950-
1960', Productivity Measurement Review (May 1962).
education is taken into account. In America over the period 1929-58

Denisonl has estimated that increases in education raised the quality
of the labour force by the equivalent of a 0·93 per cent per annum
increase in the quantity of labour. Weighting by labour's share of
the national income gives a contribution of education to measured
growth of0·68 percentage points or 23 per cent. The corresponding
contribution of education to the rise in output per person employed
is 42 per cent.
For the less developed countries, a study of the relative contribution
of labour and capital to growth, using the production function
approach, has recently been made by Maddison taking twenty-
two countries over the period 1950 to 1965.2 There is too much
arbitrariness in his calculations and weighting procedures for the
results to be taken too seriously, but for what they are worth, labour
is found to have contributed about 35 per cent to growth overall and
capital 55 per cent, leaving a residual contribution of 10 per cent
attributable to increased efficiency in resource allocation. Labour's
contribution includes the effects of the growth of education, better
nutrition and resource shifts from agriculture to industry.
Even when allowance is made for improvements in the quality of
factors and resource shifts, many production function studies for
individual countries still leave a large unexplained residual. The
economic forces reflected in the residual will include improvements
in the organisation and management of the factors of production,
advances in knowledge, and the general process of learning through
experience which cannot be measured directly and attributed to the
factors of production. Given the type of factors that comprise the
residual, its size will largely be a function of the economic and
institutional environment. For example, a highly competitive eco-
nomy which is receptive to change might be expected to have a larger
residual than a highly protected economy in which labour and
management alike resist new techniques of production and new
modes of organisation. Britain's slow growth compared to Europe
is largely attributed by Denison to a small residual reflecting an
unresponsiveness to technical change, and it is significant to note the
1 E. Denison, The Sources of Economic Growth in the U.S. and the Alternatives
before Us (New York: Committee for Economic Development, Library of
Congress, 1962).
2 A. Maddison, Economic Progress and Policy in Developing Countries (London:
small residual factor for many of the less developed countries studied
by Maddison. Those who stress the importance of social and insti-
tutional change as a prerequisite to growth in less developed coun-
tries must look at the residual in relation to other countries in support
of their argument.
Appendix 2. I
Denison on 'Why Growth

Rates Differ'
As illustrative of the production function approach to an under-
standing of the growth process, we conclude this chapter with a
review of Denison's massive and meticulous study of the sources of
growth in nine Western countries (including America) for the
period 1950-62.1
Denison distinguishes twenty-three separate sources of growth,
and attempts firstly to quantifY for each country the absolute and
relative contribution of each source; secondly, to establish the main
factors accounting for inter-country differences in the growth and
levels ofincome per person employed (especially between Europe and
America); and thirdly, to determine whether the faster growth
experienced by Europe than America has been a 'natural' pheno-
menon or significantly fostered by public action.
Countries (Washington: Brookings Institution, 1967). The nine countries
examined are Germany, Italy, France, Belgium, Netherlands, Denmark,
Norway, United Kingdom and the United States. North-west Europe refers
to all the European countries excluding Italy. The substance of this review
first appeared in Italian in Moneta e Credito (Banca Nazionale del Lavoro,
July 1969).
The Growth Performance

The comparative growth performance of each country under con-
sideration over the period 1950-62, and the level of income per
person employed relative to that of America in 1960, is shown in
Table A2.1. The growth rates are calculated from figures of net
national product at factor cost, measuring each country's output at
its own base-year prices; while the levels of national income per
person employed as a percentage of the American level are based on
United States price weights.
TABLE A2.1
Growth rates (% p.a.), 1950-62 (3)

- - - - - - - - - - - - - - Level of income
(1) (2) per person
National income Income per employed, 1960
person employed (U.S.= 100)
Germany 7·3 5·2 59
Italy 6·9 5·4 40
France 4·9 4·8 59
Netherlands 4·7 3·5 65
Denmark 3·5 2·6 58
Norway 3·5 3·3 59
United States 3·3 2·1 100
Belgium 3·2 2·6 61
United Kingdom 2·3 1·6 59
The table largely speaks for itself, but two points are worth
emphasis. The first is the uniformity of levels of income per person
employed in Europe (except Italy), and the wide gap that exists
between Europe and America. According to these estimates, North-
west Europe in 1960 was at the level attained by America in 1925,
and the absolute gap in levels of income per person employed con-
tinues to widen. The second point is that when the income growth
figures (col. (I)) are adjusted for increases in the numbers employed,
some of the inter-country differences narrow or disappear (col. (2)).
Labour-force growth is a direct and fairly painless source of
national income growth, and this is why Denison is as much con-
cerned with the growth of income per person employed as with the
growth of income itself.
To separate the contribution to growth of factor inputs and the
growth of total productivity, Denison uses the constrained form of
the Cobb-Douglas function, taking factor shares of the national
income as the partial elasticities of output with respect to the factor
inputs. Excluding property income from abroad and imputed
earnings from residences, the average factor shares of national in-
come in North-west Europe and America for the period 1950-62
were as follows:
North-west Europe United States
(%) (%)
Land 4·0 3·0
Labour 77-6 82·0
Reproducible capital 18·4 15·0
In all countries, except America, Denison finds that a higher
proportion of growth is attributable to increases in output per unit
of total input than to increases in inputs themselves. This is despite
the fact that one of the distinctive features of the analysis is that he
attempts to take account of certain changes in the quality of capital
and labour in estimating the 'effective' growth of factor inputs.
Without these adjustments the 'total' productivity term would
appear even larger. We shall not quibble here with Denison's basic
methodological approach. It will be necessary later, however, to
question some of the assumptions that Denison makes in arriving at
quantitative estimates of the various factors influencing the 'effective'
growth oflabour and capital, as well as the importance of the com-
ponents contributing to the large residual item found for most
countries.
Labour Input
Labour input is measured as the growth of employment adjusted for
three main 'quality' changes: firstly, changes in the distribution
of employed persons between full-time and part-time work together
with the number of hours worked by each group; secondly, dif-
ferences in the distribution of total hours worked among seven
categories of workers classified by sex, age and civilian and military
status; and thirdly, changes in the educational stock. The final
index of labour input for each country can be thought of as one in
which all types of labour are related to the work done by an adult
male with eight years' schooling, working the hours prevailing in the
base year 1950. The index excludes from consideration such factors
as the experience, effort and health of labour; and the stock of
knowledge (e.g. changes in the quali~ of education) is assumed con-
stant. These factors are left to contribute to 'total' productivity.
Separate indexes are computed for the three qualitative changes
mentioned, and a I per cent increase or decrease in the quality
index is treated as equivalent to a I per cent change in the quantity
of labour. In calculating the separate contribution of the quality
changes to growth, changes in the quality indexes are weighted by
labour's share of total income in the same way as the growth of
employment itself. No attempt is made to weight employment (or
man-hours) in separate industries by their own hourly earnings in
order to allow for the effect of shifts in the labour force between
low-productivity and high-productivity sectors. This is perhaps a
strange omission, unless the task proved impossible, in view of
Denison's general desire to adjust labour input for changes in its
quality and, by doing so, to assign to labour sources of growth
traditionally left in the residual item. When he finally comes to dis-
aggregate the residual factor, the only resource shift taken account
of is that from agriculture to industry.
The effect of hours worked appears as a negative item under
labour input since in most countries hours worked have decreased
since 1950. But since a negative causal relationship is assumed
between hours worked and labour's efficiency, labour input is
assumed not to be affected by the full amount of the reduction in
hours. The percentage offset that Denison applies, however, is
somewhat arbitrary.
Education, on the other hand, appears as a positive contributor to
growth. There has been an expansion of the number of years'
schooling per man in all countries, and Denison has done a useful
job in comparing the growth of the stock of education between
countries. To calculate the contribution of the expansion of educa-
tion to growth, earnings weights are applied to the distribution of
the labour force (males and females separately) by amounts of
schooling, making the assumption that 60 per cent of earnings
differentials between people of the same age are due to differences
in the number of years' schooling they have received. His conclusions
about the contribution of education to growth, therefore, hinge
crucially on the admittedly bold assumption that 40 per cent of
differences in earnings between workers of the same age are due to
factors other than differences in the stock of education embodied in
them, e.g. native ability, parental background, etc. Given the present
state ofour knowledge of the causes ofincome differentials, Denison's
opinion can perhaps be allowed to go unchallenged. On the other
hand, it seems a little unlikely that the percentage of earnings dif-
ferentials due to differences in education will be the same in every
country when different attitudes prevail in different countries
towards the reward of ability and educational attainment.
Having adjusted the employment figures.for 'quality' changes,
Denison's conclusion is that labour input as a whole contributed
1·12 percentage points (33 per cent) to the growth of income in the
United States, 0·83 percentage points (18 per cent) in North-west
Europe and 0·60 percentage points (25 per cent) )n the United
Kingdom. The unadjusted growth of employment and education
were of major importance in all countries. Education by itself
contributed 0·49 percentage points (15 per cent) to growth in the
United States, 0·23 percentage points (5 per cent) in North-west
Europe and 0·29 percentage points (12 per cent) in the United
Kingdom; and in the case of income growth per person employed
the relative contribution of education was 22, 6 and 17 per cent
respectively.
Denison estimates the 'educational stock' per person in Europe to
be approximately 50 per cent lower than in America, and this goes
some way to account for differences in the level of income per per-
son employed between Europe and America. It is clear, moreover,
that the growth of education in America has been a much greater
stimulus to growth than in Europe, and faster growth in Europe
than in America cannot be explained by a faster rate of expansion
of education. Education appears to be a relatively untapped source
of growth in Europe.
Capital Input
Four main types of capital input are distinguished: dwellings; inter-
national assets; non-residential structures and equipment; and
inventories. Each is weighted by its own base-year returns in
estimating its contribution to growth. As with employment, no
attempt is made to distinguish classes of capital stock by industry
and to weight the capital stock in each industry by its own earnings,
thus excluding the effect on growth of shifts in the distribution of
capital between activities. Capital is measured gross at constant

prices rather than at current prices, the result of which is to leave
changes in the quality or efficiency of capital goods to influence
changes in output per unit of input. In Denison's own words: 'The
value in base period prices of the stock of durable capital goods (be-
fore allowance for capital consumption) measures the amount it
would have cost in the base year to produce the actual stock of
capital in a given year (not its equivalent in ability to contribute to
production).' 1
Excluded from consideration, therefore, as in the case of labour,
are advances in knowledge, which in this context refer mainly to the
ability to design capital goods. It is assumed that advances in know-
ledge are 'disembodied'. Denison has been a persistent advocate
of measuring capital at cost, rather than by its ability to contribute
to production, in order to isolate the contribution of advances in
knowledge from increases in output that arise from increases in
the quantity of investment and changes in the age distribution
of capital.
By far the most important item in any country's capital stock is
non-residential structures and equipment, and Denison finds that
this type of capital per person employed in Europe is less than 50 per
cent of the American level. Additions to the country's capital stock
come about through the processes of saving and net investment,
but Denison goes out of his way to attack the conventional wisdom
of a causal positive relation between the proportion of national
income invested and the growth performance of countries, especially
since the price of capital goods relative to other goods differs between
countries. He finds almost no correspondence between the growth of
the capital stock and the percentage of gross national product
devoted to non-residential structures and equipment, and no
correlation between the growth of income and the amount invested
per person employed. Denison also doubts whether differences in
the age of the capital stock between countries play an important
part in accounting for inter-country differences in growth. The age
distribution of capital is a part of the embodiment hypothesis which
Denison dismisses as unimportant because in practice the age
distribution of capital is subject to small variation. But even if all
Countries (Washington: Brookings Institution, 1967) p. 134.

advances in knowledge were embodied in new capital, and the
productive efficiency of the capital stock rose by becoming younger,
Denison claims that the contribution of a higher investment ratio
to growth would still be insignificant in view of non-residential
capital's small share of total national income.
Having computed indexes of the capital stock, Denison finds that
the growth of capital as a whole contributed 0·53 percentage points
(25 per cent) to growth in the United States, 0·86 percentage points
(18 per cent) in North-west Europe and 0·15 percentage points
(21 per cent) in the United Kingdom. The corresponding absolute
(and relative) contributions to the growth of income per person
employed are 0·66 percentage points (27 per cent), 0·65 percentage
points (17 per cent) and 0·37 percentage points (27 per cent)
respectively. The contribution of differences in the non-residential
stock of capital between European countries and America to dif-
ferences in the levels of national income per person employed
seems to be quite small. Income per person employed in North-
west Europe would have been only 5·6 per cent lower than in America
in 1960 if the amount of capital per person employed had been the
only respect in which the two areas differed; and the difference
between the United Kingdom and the United States would have
been 6·2 per cent. The major part of the gap in income levels per
person employed between America and Europe is accounted for by
the much higher level of output per unit of input in America.
Output per Unit of Input (the Residual Factor)

The contribution of increases in output per unit of total input to the
growth of income is estimated to have been 1·37 percentage points
(42 per cent) in the United States, 3·07 percentage points (64 per
cent) in North-west Europe and 1·18 percentage points (53 per cent)
in the United Kingdom; and to the growth of income per person
employed, 62, 80 and 74 per cent respectively.
Denison divides changes in output per unit of input into three
main components: firstly, resource shifts (from agriculture to in-
dustry) and economies of scale; secondly, advances in knowledge;
and thirdly, 'residual productivity'.
There is no doubt from Denison's study that the transference of
resources from agriculture to industry has been an extremely valuable
source of growth in Europe as a whole, and especially in Germany,
France and Italy. Moreover, the potential for further reallocation

will provide a substantial impetus to growth in the future. Over the
period 1950-62 it is estimated that migration from agriculture into
industry contributed 1·04 percentage points to growth in Italy,
Q·76 percentage points in Germany and 0·65 percentage points in
France, and the shares of the work-force in agriculture in 1964
were still high at 25, 11·4 and 18·4 per cent respectively. In the
United Kingdom and the United States, on the other hand, the
potential for resource shifts from agriculture had already been
largely exhausted by 1950. The United Kingdom, in fact, gained
virtually no increase in productivity from this source and, compared
with the United States, income per person employed in 1960 would
have been lower than it actually was had 'misallocation' not already
been eliminated.
The allocation and use of resources is not only influenced by the
spontaneous movement of resources from low-productivity to high-
productivity sectors. The state of competition, the tax system,
barriers to international trade, etc., all help to determine the effi-
ciency with which an economy operates. Changes in these factors
will inevitably have some impact on growth, but only guestimates
can be made of their quantitative importance and Denison does not
ascribe much importance to them in practice.
A factor afimportance which he does try to quantify, however, is
the influence of economies of scale, which he considers to have been
a significant source of growth, especially in North-west Europe.
Apart from economies of scale arising from the growth of national
and local markets, which he assumes (without empirical evidence)
yielded advances in output equal to about one-eleventh of the
increase in measured national product, economies of scale are also
used to account for the difference between European and American
growth rates when the national products of the European countries
are calculated using American price weights rather than the coun-
tries' own price weights. Typically, European growth rates are
higher when their own price weights are used because consumer
goods' prices tend to be relatively higher than in America. The justi-
fication for ascribing the difference to economies of scale rests on the
(plausible) proposition that systematic association between high
relative prices and large increases in consumption is due mainly to
economies of scale made possible by the high income elasticity of
demand for consumption goods.
Increases in output per unit ofinput not accounted for by resource
shifts from agriculture to industry and economies of scale are attri-
buted to advances in knowledge and such residual items as lags in
the application of knowledge behind best practices; the general
efficiency of the work-force; and all other unmeasurable factors that
might affect production (plus, of course, the net sum of all errors
and omissions).
To separate 'advances in knowledge' from these other residual
influences, Denison makes his boldest assumption of all. He assumes
that the 0·76 per cent increase in output per unit of input not ac-
counted for by economies of scale and resource shifts from agricul-
ture to industry in America truly reflects 'advances in knowledge'
in America. He then proceeds to take this same figure as the abso-
lute contribution of advances in knowledge to growth in the Euro-
pean countries, presumably on the assumption that advances in
knowledge are international 'commodities', freely available. On this
basis advances in knowledge emerge as the largest single source of
growth in four of the eight other countries under review. But even
if one accepts the view that there is no technological gap between
countries, it is misleading to apply the figure of 0·76 to all other
countries without considerable reservation, especially as by doing
so 'advances in knowledge' appears as a predominant source of
growth in all the countries concerned. Despite Denison's attention
to detail, the residual for America is bound to be a composite of
unspecified factors including advances in knowledge, and ought to
be recognised as such.
Having said this, it is well known that there are many ways in
which advances in knowledge do not manifest themselves in growth
statistics, and the impact of advances in knowledge on the broader
concept of welfare may not be exaggerated. In the first place,
advances in knowledge cannot contribute at all to measured growth
in activities where output is measured by inputs, which is the case
in a wide range of service activities. Secondly, measured output
takes no account of changes in the quality of goods and services or
the range of choice of products that technical progress may permit.
Indeed, it is probably because advances in knowledge have largely
gone to improve the quality and choice of products that the enor-
mous growth of research expenditure in the post-war years appears
to have had such little impact on the growth of productivity, at
least in America.
Whatever the nature of advances in knowledge, and their contri-

bution to growth, increases in output per unit of input not attri-
butable to economies of scale and migration from agriculture
account for about one-half of the gap in income levels per person
employed between Europe and America in 1960. In the United
Kingdom specifically, income per person employed was 41 per-
centage points below the American level in 1960, and 29·3 per-
centage points of this shortfall is the result of unmeasured factors.
The reasons for this unexplained gap can only be guessed at, but
are most likely to be connected with the attitudes and dynamism of
both sides of industry, which in turn must be considered the product
of many influences in the economic and social environment. The
overmanning of industry is one possibility, and Denison suggests
that this type of resource waste may be on a scale comparable to
that associated with agriculture in France and Germany.
The blunt fact is that in the United Kingdom in the post-war
years 'residual productivity' has been much lower than in any other
European country, and Denison concludes by saying that
it is difficult to explain why output per unit of input in the
U.K. economy as a whole was not well above that in France
and Germany in 1960 unless unaggressive management, labour
resistance to change, and restrictive practices have combined to
prevent the U.K. from realising as fully as other countries the
potential for reducing labour requirements as technology has
advanced and the quality of capital and the scale of operations
have increased.l
Conclusion
It emerges from Denison's study, therefore, that 'residual produc-
tivity' is the biggest source of difference in levels of income per
person employed between America and Europe, and between
America and the United Kingdom, while advances in knowledge
have been the largest single source of growth in income per person
employed in the United Kingdom and the second largest source,
following economies of scale, in North-west Europe. It must be
remembered, though, that the conclusion about advances in know-

Countries (Washington: Brookings Institution, 1967) p. 294.
ledge rests exclusively on the unproven assumptions that the
American residual constitutes 'advances in knowledge' and that the
absolute contribution of advances in knowledge to growth has been
the same in all European countries as in America. Because of the
shakiness of these assumptions, any policy conclusions must be of
uncertain value. Denison's frequent caveat that the assumptions
of his analysis can be varied is not exactly helpful at this point.
As to Europe's faster growth than America, Denison believes that
this has been largely the result of greater opportunities to raise
income in Europe than in America. Europe's superior performance
does not seem to him to be the outcome of any positive government
action, except perhaps the greater encouragement of investment.
On the other hand, Denison feels that the scope for Europe to grow
at a faster rate than America in the future may be diminishing,
except in the sphere of education in which Europe lags a long way
behind. Assuming that advances in knowledge are freely available to
all countries, it will be the course of' residual productivity' which will
be crucial and decisive in the decades ahead. In levels of income per
person employed Europe is still a long way behind America, owing
primarily to differences in 'residual productivity'. If this is the
result of differences in attitudes and enterprise on the part of
workers and management, Denison's message to Europe is that the
American model of business is there to be copied. In his view, to
reduce the income 'gap' would not be difficult if businessmen,
workers and governments were determined to do so.
In conclusion, there is no doubt that Denison has performed an
extremely useful task in quantifYing the contribution of increases in
physical factor inputs to growth in so many countries, and that his
estimates can be accepted with some degree of confidence. His
attempt to quantifY the sources of increases in output per unit of in-
put, however, falls short of complete success, and some of his con-
clusions must be considered of doubtful worth. As much as one
admires his attention to detail, the present state of our knowledge
does not permit some of the sweeping assumptions that Denison is
forced to rely on. The residual factor in economic growth remains the
'coefficient of our ignorance'.
3
Land, Labour and Agriculture
In the production function approach to the analysis of the sources
ofgrowth, land as a separate factor of production tends to be assumed
away or subsumed into capital. There are two main reasons for this.
The first is the traditional classical notion ofland as a fixed factor of
production, which in the long run is undoubtedly true. The second
is the practical fact that land without the application of capital is
oflittle use, justifYing the treatment ofland and capital as one factor.
This is not to deny, however, the importance ofland and natural
resource endowments as factors in the growth process. The quality of
land can markedly affect the level of agricultural productivity in
the early stages of economic development, and the importance of
agricultural development, in turn, can be clearly seen within the
production function framework. Rising agricultural productivity
permits the release of labour from agriculture to industry, which in
turn leads to economies of scale, rising income per head and greater
capital accumulation. Given the dominance of the agricultural
sector in the economic structure of less developed countries, such
factors as the physical attributes ofland (topography, fertility, etc.),
the land tenure system, the ratio oflabour to land and the extent of
natural resource endowments are likely to exert a major influence
on the speed of development as determinants of the pace of agricul-
tural advance and the pace of industrialisation based on a healthy
agricultural sector or the exploitation of indigenous resources.
Geographical factors, including the nature of the land, and God-
given factors such as weather and terrain, must be included in any
list of reasons drawn up in answer to such questions as: why have
some countries developed earlier than others, or why have some
countries remained in a traditional or transitional state longer than
others? This is the concept of geographic determinism, which can be
advanced as a hypothesis of underdevelopment in its own right.
LAND, LABOUR AND AGRICULTURE 73
The Role of Agriculture in Development

In this section we shall consider land in relation to agriculture, and
discuss in a general way the role of the agricultural sector in the
development process. In the course of the discussion we shall elabo-
rate on the emergence of an economy from its traditional subsistence
state into Rostow's transitional stage in which the seeds are sown
for 'take-off' into self-sustaining growth. The approach to be adopted
has a common-sense, pragmatic appeal since land primarily affects
the productivity of agriculture, and secondly, the development of
agriculture, owing to its initial importance in a country's economic
structure, must play a crucial role in establishing the framework for
industrialisation. The agricultural sector, for instance, must provide,
in large measure, the factor supplies for industry; it must provide
food for an urban industrial population, and it must contribute to
the market for industrial goods if the demand for goods is to be
sufficient to justify their production domestically. For the agri-
cultural sector to release labour, to provide savings, to supply food
and to contribute to the market for industrial goods, it must generate
a steadily rising surplus of production in excess of subsistence
needs. Since land is relatively fixed in supply, this requires rising
agricultural productivity. The 'grass roots' school of economic
development, which came into fashion as a reaction against the
emphasis on industrialisation at any cost, lays stress on policies to
raise the level of productivity in agriculture as the best long-run
development strategy. Until recently, the performance of the agri-
cultural sector in less developed countries had been very disappoint-
ing. For many years, food production per head hardly rose. Since
the mid-fifties, however, agricultural productivity has picked up and
presents less of a worry than formerly. The quickest and cheapest
way to raise productivity will depend on the reasons for low pro-
ductivity. If, for example, low productivity is associated with a high
ratio of labour to land, productivity could probably be increased
substantially by small applications of capital in the form of drainage
schemes, fertilisers, etc., together with a gradual run-down of the
work-force as productivity growth starts to accelerate. If, on the
other hand, low productivity is associated with the opposite situa-
tion of a high ratio ofland to labour, the solution to low productivity
is likely to involve much larger doses of capital for labour to work
with. Most of the present less developed countries have high ratios
of labour to cultivable land in contrast to many of today's most

advanced countries at an equivalent stage in their economic history,
e.g. America, Canada and Australia.
Some argue, more simply, that what is required is a rise in the
price of agricultural products relative to industrial products to
induce extra supply. There is in fact considerable evidence that
producers in close proximity to large markets with good transport
facilities do respond to price incentives, but through a change in the
output mix of agricultural goods rather than through increased
resource utilisation or shifts in the production function. Ideally,
what is required is a shift in the whole agricultural supply curve
through the introduction of new production techniques and the
reduction of barriers and bottlenecks to the expansion of output at
existing prices.
While the transformation of traditional agriculture is obviously
dependent on new inputs, the task is not exclusively a question of
the supply of capital. The basic problem is one of determining the
form that new inputs should take if agriculture is to attract an
adequate share of investment resources. The first stage of an agri•
cultural development programme must be to identify those inputs
which will raise productivity at the least cost to society.
Schultzl has argued that the low productivity of farm labour is
due more to an absence of specific factor inputs, such as research
and education, than to a shortage of reproducible capital as such.
He contends that the most practical and economical approach to
achieving sizeable increases in agricultural productivity lies in en-
hancing the efficiency of the existing agricultural economy through
improvements in the quality of inputs, and by the application of
advances in knowledge and modern technology on a broad front.
Additional quantities ofexisting inputs will achieve nothing because of
very low rates of return. Indeed, contrary to the conventional view
that a stagnant agricultural sector is the result of the reluctance of
peasant producers to respond to incentives, Schultz argues that it is
traditional agriculture which is the cause of the seeming antipathy to
work and investment because oflow returns at the margin to existing
inputs. The way to transform traditional agriculture into a relatively
cheap source of growth is by investment to produce a supply of
new agricultural inputs that will be profitable for farmers to adopt.
1 't. Schultz, Transforming Traditional Agriculture (New Haven: Yale
Schultz suggests that what has been lacking in the past is not an
unwillingness on the part of the agricultural sector to accept new
ideas but public expenditure and the organisation of particular
public activities to serve the agricultural sector. Agricultural re-
search, and investment in man to improve human capabilities in
agriculture, have been too readily neglected. He points to the rapid
rise of agricultural productivity in Japan and Taiwan in the late
nineteenth and early twentieth centuries which was due to forms of
'technical progress' that are taken for granted in the developed
countries; for example, the application offertilisers and the selection
and cultivation of high-yield crops. With the so-called 'green
revolution', the scope for yield increases has become widespread in
less developed countries, and rates of return on investment in new
technologies and varieties of seed are now enormous. Mexican
'dwarf' wheat is testimony to the impact that technology can have
on the productivity of the agricultural sector.
In seeking reasons for low productivity, the influence of natural
factors must also be borne in mind. Climate and terrain determine
to a large degree what goods a country can produce, the amount
of cultivable land available per inhabitant, and the land's fertility.
To some extent the application of capital to the land can compensate
for unfavourable natural forces, but there are obvious limits. Moun-
tains cannot easily be flattened or deserts readily watered.
The relationship between agricultural productivity and the land
tenure system also needs considering. The land tenure system may
be an impediment to increased productivity independently of en-
vironmental factors, or a general lack of incentive due to the low
return on existing agriculturalinputs. An essential prerequisite of
increased productivity may be comprehensive land reform, or at
least changes in the land tenure system coupled with related measures
to ensure their success. When land is held and worked in the form
of large estates, it is frequently underutilised, and land reform
involving the subdivision of estates could contribute to an increased
intensity ofland use and increased production. If tenant farmers are
without security of tenure, land reforms which provide security can
contribute to greater efficiency and initiative on the part of the
tenant farmer. Where, as a result of reform, tenants are permitted to
become owners of their farms, or to receive a larger return than
hitherto, the will to work should be enhanced, again with a corre-
sponding increase in production. There is evidence that where a
change in the tenure system has permitted the producers them-

selves to reap the rewards of new techniques, peasant farmers have
been surprisingly ready to break with custom and tradition. The
task of persuading producers to adopt more modern methods of
production and to purchase improved seed and fertiliser becomes
much less difficult. Countries such as Egypt, Israel, South Korea,
Yugoslavia, Japan and Mexico, where food production per head has
risen impressively, have all undergone fundamental land reform.
On the other hand, land reform is not always successful, and many
countries have experienced increased productivity with unsatis-
factory land tenure systems. Since land tenure systems vary so
much from country to country it is difficult to generalise, but in
theory, at least, agrarian reform may be a precondition of increased
production.
The Growth of the Money Economy

The question of the willingness to change customs and traditions
leads naturally to a consideration of how subsistence peasant econo-
mies, producing goods for consumption only, typically transform
themselves into money economies with an export and industrial
sector. From historical experience two factors would appear to be
crucial for the expansion of the agricultural sector, and the eventual
production of goods for exchange at home and abroad. One is the
expansion of communications to create outlets and markets for
surplus production (and to encourage the production of the surplus
itself). The other is the emergence of a class of middlemen or export-
import merchants acting as agents between world markets and the
home agricultural sector.
Both the ability to export and the ability to market internally
imply surplus production over subsistence needs, and it is the size of
this surplus that will largely determine the speed with which the
subsistence sector can be drawn into the money economy. Again
we come to the point that unless productivity in agriculture in-
creases, the expansion of the monetised sector will tend to decelerate
as the land for cultivation dries up. When land has been exploited
to the full it acts as a constraint on development unless agricultural
productivity increases or non-agricultural activities can be estab-
lished.
The emergence of an export sector provides a powerful stimulus
to development and the extension of the money economy. Exports
create the capacity to import and the very purchase of foreign
products can encourage further export specialisation. A population
which acquires a taste for imported goods provides the impetus to
producers to export more. In the case of new goods, as well as new
techniques, there is fairly concrete evidence that peasant producers
respond to incentives, and are not so dissimilar to Western economic
man as is sometimes claimed. Imports also provide a stimulus to
industrialisation. If a market for a foreign-manufactured good
becomes established it becomes easier, and less risky, with the aid of
tariff protection, for a domestic manufacturer to set up in business
because the market is assured. Imports can also substitute for domes-
tic capital and raise the growth rate directly.
When peasant producers start to specialise in goods for export,
and rely on other producers for goods which they previously pro-
duced themselves, the money economy will spread from the foreign
trade sector to the rest of the domestic economy. This is nothing
more than the international division of labour giving rise to the
need for a means of exchange within a country as well as between
countries.
The transition from a state of total dependence on the pro-
duction of agricultural goods to a state of dualism, with the co-
existence of a subsistence agricultural sector and a commercialised
industrial sector, is a significant landmark in the development of
an economy. The dual economy sets up incentives which are ab-
sent in the agrarian economy, and provides a new form of saving (the
capitalist surplus). Profit gives the owners of industrial capital the
incentive to innovate, and the possibility of acquiring assets in the
industrial sector gives an incentive to agricultural producers also.
Divergences in the pattern of agricultural productivity are frequently
related to the proximity of an industrial-urban complex.
The emergence of an export sector, the spread of the money
economy, and the establishment of industries typically occur con-
currently. What form industrialisation takes will depend, in the
first instance, on the initial impetus. One stimulus to industrialisa-
tion, which we have already mentioned, is imports creating a market
for goods that can be produced domestically without much diffi-
culty. A more obvious factor leading naturally to industrialisation
is the availability of resources from the land forming an indigenous
industrial base. In this case industrialisation takes the form of the
processing of raw materials. There are few countries which do not

possess some natural resource or other, and every country will have
a comparative advantage in the production of one or other raw
material that can be processed.
In many of the present less developed countries, formerly under
colonial rule, the initiating force behind industrialisation was the
foreign exploitation of resources. Industrial activity took the form
of mining operations and plantation agriculture. The establish-
ment of foreign enclave activities undoubtedly exerted a develop-
ment impact, but it is an arguable point whether development would
have been more rapid had the countries been left to their own de-
vices. Some would claim that the long-run development of the
countries has been impaired because the availability of cheap
labour from the subsistence sector has discouraged the installation
of more modern productive machinery, and also that the foreign
ownership and exploitation of a country's resources has considerably
reduced its potential level of investment through the remittance of
profits. The argument begs and raises many important issues con-
cerning development strategy, which are dealt with in some detail
in Chapter 6.
It is interesting to note that very little foreign capital, private
or institutional, has ever gone into traditional agriculture. One
reason could be the risks involved, and the limitation of the market
compared with industry. Another reason might simply be the low
returns in traditional agriculture, as stressed by Schultz. Shonfield
has criticised the meagre amount of resources invested in traditional
agriculture by international institutions, in particular the World
Bank.l Of $4,000 million of development loans made by the World
Bank between 1947 and 1959, only $124 million went into agri-
culture. There has, however, been some reversal of this policy in
recent years.
In the absence of foreign investment, the sources of capital for the
expansion of both industry and agriculture are relatively limited in
the early stages of development. In a truly subsistence economy, in
the sense of an economy producing only what it needs for itself and
no more, everyone is a Robinson Crusoe supplying his own capital
by refraining from present consumption. With specialisation in the
production of goods for export, and the producer's need for capital
to expand productive capacity, mechanisms grow up spontaneously
1 A. Shonfield, Attack on World Poverty (New York: Random House, 1960).
to meet the need for credit. It is a good market maxim that demand
will create a supplier at a price. The suppliers are generally village
money-lenders, shopkeepers, landlords and not infrequently the
Church - especially in South America - charging rates of interest
often in excess of 50 per cent.
Once agriculture emerges from its stagnatory, subsistence state
and starts to specialise and produce goods for export, and industry
develops under the impact of growth in the agricultural sector, the
two sectors of agriculture and industry become very much inter-
dependent. The industrial sector adds to the demand for goods pro-
duced by agriculture, and absorbs surplus labour which may raise
productivity in agriculture. The agricultural sector, in turn, pro-
vides an outlet for industrial goods out of rising real income, and
makes a factor contribution to development through the release of
resources if productivity rises faster than the demand for com-
modities.
The transfer of resources from agriculture to industry may be
capital or labour or both. Since labour is in abundant supply in
most low-income countries, there is generally no difficulty involved
in releasing labour for industry except at harvest periods. Labour, in
any case, will tend to migrate naturally in response to seemingly
better opportunities in the industrial sector and higher real in-
comes. The earnings oflabour in the industrial sector may be half as
much again as the agriculture wage. If the industrial sector is to be
guaranteed an adequate supply of labour, some money differential
is inevitably required to offset the higher living costs in an urban
environment, to compensate for the forfeit of non-monetary bene-
fits of rural life, and also to compensate for greater job uncertainty
in tht industrial sector. Real earnings may also be higher because
of genuinely higher productivity in the industrial sector where
labour has more of other factors of production to work with. Most
models of rural-urban migration make migration a positive function
of the urban-rural expected wage differential, which is the difference
between the minimum urban wage, adjusted for the proportion of
the total urban labour force employed (as a proxy for the probability
of finding work), and the agricultural real wage expressed in terms
of manufactured goods.l
Capital may be less mobile than labour, and if there is considered
1 E.g. J. Harris and M. Todaro, 'Migration, Unemployment and
Development: A Two-Sector Analysis',AmericanEconomic Review (Mar 1970).
80 GROWTH AND DEVEL.OPMENT
to be insufficient lending from the agricultural sector on a voluntary

basis it may become necessary for a government to transfer savings
compulsorily from the agricultural sector by taxation. This method
was resorted to in a severe manner by Japan at the time of the Meiji
restoration and by Soviet Russia after the revolution. In Japan
between 1880 and 1900 the land tax provided approximately 80
per cent of central government taxation, and in Russia forced
extraction of the agricultural surplus took the form of expropriation
of land and extermination of the work-force. Industrialisation in
Western Europe, and particularly in England, was also financed to
a large extent by surpluses generated on the land, but transference
of these surpluses was on the whole voluntary through a rapidly
expanding banking system. The less developed countries today,
despite their access to foreign sources of capital, must also rely heavily
on extracting the surplus from agriculture to finance industriali-
sation. The difficulty is to decide on the best means of extraction
without impairing the incentive to produce, or damaging the growth
of productivity on which a growing agricultural surplus depends.
The finance of economic development will be discussed more fully
in Chapter 8.
Economic Development with Unlimited Supplies of Labour

The process of the emergence of a money economy from a subsistence
state has been formalised by Professor Lewis in his celebrated paper
'Economic Development with Unlimited Supplies of Labour'.1
There he presents a 'classical' model of a dual economy with the
purpose, as he describes it, of seeing what can be made of the classical
framework in solving the problems of distribution, accumulatiop. and
growth. His ultimate aim is to emphasise the crucial role of the
capitalist surplus in the development process.
The Lewis model starts, therefore, with the assumption of a dual
economy with a modern exchange sector and an indigenous sub-
sistence sector, and assumes that there are unlimited supplies of
labour in the subsistence sector in the sense that the supply oflabour
exceeds the demand for labour at the subsistence wage; that is, the
marginal product of workers in the subsistence sector is equal to or
1 A. Lewis, 'Economic Development with Unlimited Supplies ofLabour',
Manchester School (May 1954), and 'Unlimited Supplies of Labour: Further
Notes', Manchester School (Jan 1958).
below the subsistence or institutional wage so that a decrease in the
amount of workers would not lower the average (subsistence)
product of labour and might even raise it.
It has even been argued that the marginal product oflabour may
be zero or negative in an economy still at a fairly low level of de-
velopment experiencing a rapid growth of population. One of the
distinguishing features of agriculture is that it is an activity subject
to diminishing returns owing to fixity of the supply ofland. If there
is rapid population growth and labour has little employment oppor-
tunity other than on the land, a stage may be reached where the
land cannot give further workers a living unless existing workers
reduce their own hours of work drastically. These propositions are
illustrated in Fig. 3.1 below. The curve drawn represents the
FIG. 3.1
marginal product of successive units of labour added to the land.
After the employment of OX units of labour the marginal product
of labour begins to fall owing to diminishing returns; after oxl
units of labour, labour's marginal contribution to output falls
below the subsistence wage; and after ox2 units oflabour, labour's
contribution to output becomes negative and total product will
decline with successive additions of labour beyond ox2.
The same tendencies can also be represented using the type of
production function diagram introduced in Chapter 2, which is
reproduced in Fig. 3.2. With more than OX2 labour employed with
a fixed amount ofland, OK, the marginal product oflabour becomes
negative, and further additions to the labour supply, without
corresponding increases in the amount ofland, will push the economy
on to a lower production function, i.e. total output will fall from
100 units to, say, 50 units with OXs units of labour.
There are three main means of escape from the tendency towards
diminishing returns and zero marginal product in agriculture:
firstly, by productivity increasing faster than population through
the absorption of more and more ofthe agricultural population into
Labour
FIG. 3.2
industry; secondly, by technical progress in the agricultural sector

increasing labour's marginal product; and thirdly, by capital
accumulation which can raise productivity directly and which can
also be the vehicle for technical progress.
In Lewis's model, labour in excess of OX1 in Fig. 3.1 is in com-
......
Q) ' ..................
~
~
'
p ', ~
> W 1-------""'o;:,-----'T.,t------Jndustrial wage
~ I '
o I ',
c
·~
.g ~~1--------f---i---."l~--'>.-----Subsistence
'
',
"' , wage
\
0'---------:':--~--~---:\::---Units
M M
of labour
1
FIG. 3.3
pletely elastic supply to the industrial sector at whatever the

industrial wage. The industrial or capitalist sector is represented
in Fig. 3.3. above. The curve NR represents the marginal revenue
product oflabour in the capitalist sector, OW is the industrial wage
and, on the profit-maximising assumption, labour is employed in
the capitalist sector up to the point where the marginal revenue
product is equal to the wage rate. That is, OM will be employed.
Workers in excess of OM earn what they can in the subsistence
sector. The industrial wage is assumed to be determined in some
relation to the wage that workers can earn in the subsistence sector.
The differential ( WW1) between the industrial wage and the sub-
sistence wage will be a function of many factors, some of which were
mentioned earlier, e.g. higher living costs in the capitalist sector
and greater job uncertainty. Given that the industrial wage is based
on earnings in the subsistence sector, capitalists obviously have a
direct interest in holding down productivity in the subsistence
sector, and Lewis comments that the record of every imperial power
in Mrica in modern times is one of impoverishing the subsistence
economy.
The total product oflabour, ONPM, in Fig. 3.3 is split between
the payment to labour in the form of wages, OWPM, and the
capitalist surplus, WNP. The expansion of the capitalist sector and
the rate of absorption of labour from the subsistence sector depends
on the use made of the capitalist surplus. If the surplus is reinvested
leading to greater capital formation, this will increase the total
product of labour. The marginal product curve will shift upwards
to the right, say to N 1R1 , which means that if wages remain con-
stant the capitalist sector can now afford to employ more labour and
will do so by drawing on labour from the subsistence sector to the
extent of MM1 workers. The size of the capitalist surplus will in-
crease from WNP to WN1P1 which is available for further reinvest-
ment, and so the process goes on. This for Lewis is the essence of the
development process. The stimulus to investment in the capitalist
sector comes from the rate of profit which must rise over time
because all the benefits of technical progress accrue to capital if
wages are constant.
According to Lewis, the share of profits in the national income
(PfO) will also rise. Firstly, the share of profits in the capitalist
sector (PfC) will increase, and secondly, the capitalist sector rela-
tive to the national income (CfO) will tend to expand, i.e. if
PfO = PfC X CfO, then PjO will rise as PfC and CfO increase.
Whether the share of profits in national income does rise with
development is ultimately an empirical question. At the theoretical
level, however, Professor Enkel has accused Lewis of being led astray
by his diagram, and questions whether the P/Cratio does rise. If the
1 S. Enke, 'Economic Development with Unlimited and Limited Supplies
of Labour', Oxford Economic Papers (June 1962).

successive marginal revenue product curves are drawn parallel to

each other, the capitalist surplus area does rise more than pro-
portionately to the wage-payment area, so that the surplus and
profits as a proportion of income appear to rise. But is it legitimate
to draw the curves parallel? If the capital stock is augmented through
reinvestment of the surplus (say, doubles), then, assuming constant
returns to scale, the marginal product of the nth labourer will be
the same as that of the 2nth labourer, and the marginal product of
the ;th labourer will be the same as that of the nth labourer, and
so on. It is not strictly correct, therefore, to draw the successive
-
( .)
:::l
~a.
Q)
.,:::lc> w2
~
0c
.E
0
::!; Units of labour
FIG. 3.4
marginal product curves as parallel to each other representing dif-

ferent proportionate increases in product per man. Instead, each
curve should be drawn from the same vertical intercept, keeping the
proportionate gap between the two curves the same as in Fig. 3.4.
If the upward shift in the curve, NR to NR1, is proportionately the
same over the whole of its range, the capitalist surplus and the
payment to labour rise at the same rate, leaving the share of profits
in the capitalist sector unchanged, i.e. if Q.2Q./W2Q. = P2PjWP, then
WNPjONPM = WNP1jONP1M1.
This technicality does not detract from the usefulness of the
Lewis model. For one thing, the main function of Lewis's model is
to represent a process, and to contribute to an understanding of the
nature of the economic problem in backward economies. Secondly,
Lewis is more concerned with the size of the capitalist sector relative
to the total economy (C/0), an increase in which can also cause
profits as a share of national income to rise. The evidence for this
comes from Lewis himself when he says: 'If we ask why the less
developed countries save so little the answer is not because they are
so poor but because their capitalist sector is so small.' Accepting the
strict classical assumptions that all wages are consumed and that all
profits are saved, the savings ratio depends not only on the share of
profits in the capitalist sector but on the size of the capitalist sector
as well.
The process outlined by Lewis comes to an end when capital
accumulation has caught up with population so that there is no
longer surplus labour in the subsistence sector left to absorb. When
all the surplus labour is absorbed, the supply of labour to the in-
dustrial sector becomes less than perfectly elastic. It is now in the
interests of producers in the subsistence sector to compete for labour,
since the marginal product of labour is no longer below the insti-
tutional wage. When this point is reached, the agricultural sector
can be said to have become commercialised. This change in pro-
ducer behaviour in the subsistence sector has also been defined as
the end of the take-off stage.l
It is possible, however, that the process of absorption may end
prematurely before surplus labour in the subsistence sector is fully
exhausted, owing to checks to the expansion of the capitalist surplus.
Firstly, wages may rise and profits fall for reasons unrelated to the
expansion of the capitalist sector itself, but which none the less check
the rate of capital accumulation. For example, wages in the
industrial sector may be forced up directly by labour unions, or
indirectly through rising wages in the subsistence sector due to
increased productivity. Secondly, capital accumulation and labour
absorption may be checked for reasons which are related to the
expansion of the capitalist sector. For example, as the capitalist
sector expands, the terms of trade may turn against it. If the demand
for food expands faster than agricultural productivity, the capitalist
sector will be forced to pay higher prices for food in exchange for
industrial goods, eating into the capitalist surplus. If the terms of
trade move against the industrial sector immediately the process of
labour transference commences on a large scale, unlimited supplies
oflabour at a constant real wage may be very limited indeed. Labour
will be needed in agriculture to meet the demand for food. There is
no problem if productivity in agriculture is expanding rapidly, but
in practice Lewis is of the view that the failure of peasant agriculture
to increase its productivity has probably been the chieffactor holding
1 G. Ranis and J. Fei, 'A Theory of Economic Development', American
Economic Review (Sep 1961).
down the expansion of the industrial sector in most of the less de-
veloped countries. If this is so, the growth of non-farm employment
depends not so much on the reinvestment of the capitalist surplus as
on the growth of the agricultural surplus. This, in fact, is the starting
point of neo-classical models of development.! In practice, both a
lack offood and lack of capital can limit the growth of the industrial
sector, and it is impossible to say a priori which will be the limiting
factor.
The question is not so much whether the marginal product of
labour is positive or zero, because unlimited supplies oflabour in the
Lewis sense is consistent with both, but whether there are unlimited
supplies of labour at a constant real wage rate in the agricultural
sector. This is undoubtedly the most important assumption under-
lying the classical approach to the theory of development, and Jor-
genson argues that the classical approach stands or falls by this
hypothesis.2 Historically, of course, the agricultural real wage has
risen but the capitalist sector has also expanded rapidly, which lends
support to a middle view between the classical and neo-classical
approaches. Lewis himself recognises the importance of both capital
and food, and it is this consideration which forms the basis of his
argument for the balanced growth of the agricultural and industrial
sectors.
What has happened in recent years, however, does not fit neatly
into either model. The rural-urban wage differential has widened
considerably, the agricultural surplus has increased, and there has
been rural-urban migration on an unprecedented scale, but the
expansion of the capitalist sector has generated very little extra
employment. Instead, migration has simply transferred unemploy-
ment from rural to urban areas. The only conclusion to be drawn is
that the expected value of the urban wage, even in conditions of
unemployment, still exceeds the wage in the rural sector, and given
an institutionally determined urban minimum wage which is above
the free market rate a high-level unemployment equilibrium is quite
possible. Development theory must now concentrate on urban
unemployment as well as rural unemployment.
The higher the marginal product of labour in agriculture, the
1 See D. Jorgenson, 'Testing Alternative Theories of the Development of
a Dual Economy', in I. Adelman and E. Thorbecke (eds), The Theory and

Design of Economic Development (Baltimore: Johns Hopkins Press, 1966).
z Ibid., p. 54.
greater the force of the neo-classical argument that it is the growth
of the agricultural surplus that determines the growth of non-farm
employment. We must now examine more critically the classical
assumption of unlimited supplies of labour, defined as marginal
product below the subsistence wage.
If the marginal product of labour in the rural sector is positive
(which is not precluded in Lewis's model as long as it is below the
subsistence wage), the withdrawal of labour from the subsistence
sector will obviously reduce total output. To argue that develop-
ment via unlimited supplies oflabour is feasible and relatively pain-
less must implicitly assume that the marginal product of labour is
virtually zero. The term 'disguised unemployment' is usually defined
loosely in this way. But the question arises of how workers can
survive on the land if their marginal product is zero, or even positive
but below subsistence. Who would employ such labour? Would
output in the subsistence sector really remain unaffected if substan-
tial quantities oflabour migrated? In short, what precisely is meant
by the term 'disguised unemployment'? Can it be quantified, and
what are we to make of the argument that industrial development in
surplus-labour economies is a relatively painless exercise?
Disguised Unemployment
Let us redraw Fig. 3.1 from the point of diminishing returns and
describe more formally three possible interpretations of the concept
of disguised unemployment which are commonly found in the
literature. Let OA in Fig. 3.5 below be the actual number of workers
employable. One possible measure of disguised unemployment is the
difference between OA and OS, or the gap between the number of
workers available for work and the amount of employment which
equates the marginal product of labour and the subsistence wage.
This is the definition of unlimited supplies of labour in Lewis's
model where, if the marginal product of labour is below the insti-
tutional or subsistence wage, landowners have no interest in retain-
ing these workers and therefore do not compete for them with
the industrial sector. The extent of disguised unemployment on this
definition will differ from a second possible measure of disguised
unemployment which is the difference between OA and OD, or the
gap between the actual number of workers available for employ-
ment and the level of employment at which the marginal product of
labour is zero, which is sometimes referred to as the static surplus.

This surplus is obviously less than if disguised unemployment is
defined as labour with a marginal product below the institutional
or subsistence wage. A third measure of disguised unemployment
is the difference between the actual number of workers available
....
"
0
.c
.2
-"e
M.P.
0
()
"0
l
0.
Wt---------'""'------Subsistence wage
0
:e
FIG. 3.5
and the level of employment at which the marginal product of

labour would be zero if some change occurred which enabled the
same level of output to be produced with fewer men. This is repre-
sented by a pivoting of the marginal product curve to M.P. 1 . Dis-
guised unemployment is now measured by the difference between
OA and OU, which is sometimes referred to as the dynamic surplus.
The dynamic surplus clearly embraces many 'types' of disguised
unemployment because there are many reasons, particularly in
less developed countries, why labour may not be fulfilling its poten-
tial and why small changes may release substantial quantities of
labour.
There is no way of measuring the extent of the static surplus
except by examination of instances where substantial numbers of the
agricultural labour force have been withdrawn from the land for
short periods of time, either through an industrialisation project in
a neighbourhood or through illness (as in the case of the influenza
epidemic in India in 1918-19 which Schultz reportsl). In most
cases agricultural output seems to have fallen, suggesting that the
static surplus is a fiction. The only alternative measure of the
surplus is the difference between actual and required labour to
produce a given output, which, in effect, becomes a measure of the
1 Schultz, Transforming Traditional Agriculture.
dynamic surplus because some change would have to be postulated
for required labour to be less than actual labour. Defenders of the
classical model of development argue that this is quite legitimate.
Marglinl claims that no one has ever argued that the withdrawal
of labour under all circumstances will not affect output. The pur-
pose of the classical model is to draw attention to the fact that the
industrial product far exceeds the opportunity cost of labour in
agriculture, and those who use the classical model normally stipu-
late some dynamic change as migration takes place.
Unfortunately, investigators who have measured the dynamic
surplus have not generally distinguished between the causes of the
surplus, or made explicit the assumptions on which their estimates
of labour requirements are based, and this is a major reason why
estimates and opinions differ on the extent and existence of dis-
guised unemployment. If the surplus is measured simply by the
difference between the amount of labour that, in the investigator's
opinion, should be necessary to produce a given output and the
amount of labour that there actually is, this does not distinguish
between low productivity due to such factors as poor health, lack
of incentive, a preference for leisure, primitive technology or the
seasonal nature of production. There may be genuine differences
in the extent of disguised unemployment within and between
countries; on the other hand, different investigators may have been
estimating different things. Some have argued that the major part
of the observed labour surplus results from the seasonal nature
of production. Studies which exclude this possibility will certainly
overestimate the existence of disguised unemployment. What this
also means, however, is that the introduction of small amounts of
capital to substitute for labour at times of peak demand could release
substantial quantities of workers for the industrial sector without
agricultural output falling. If disguised unemployment is seasonal
unemployment, the dynamic surplus may be very large indeed.
An attempt has been made to define a set of logically distinct
types of disguised unemployment to avoid this confusion and
conflict.2 Five 'types' of disguised unemployment are distinguished:
firstly, unrealised potential output per worker due to low nutritional
1 S. Marglin, 'Comment' on Jorgenson, op. cit., in Theory and Design qf
Economic Development.
2 W. Robinson, 'Types of Disguised Rural Unemployment and Some
Policy Implications', Oxford Economic Papers (Nov 1969).

and health levels of the labour force; secondly, low levels of output
per labour input due to inadequate motivation for the cultivators to
pursue maximisation; thirdly, low average product due to low
aspirations for material income as compared to leisure; fourthly,
unemployment due to the lack of co-operating factors ('tech-
nological' unemployment); and lastly, seasonal unemployment.
Even with a logically distinct classification of the causes of dis-
guised unemployment, however, measurement will still be arbitrary
depending on the investigator's judgement of the potential level of
output if the causes of low output were to be removed. Ideally,
studies of disguised unemployment must be micro-oriented, based
on detailed information of actual and potential labour utilisation,
and with the assumptions of the analysis explicitly stated.
A reconciliation between those who argue that there is such a
phenomenon of disguised unemployment, in the sense of a very
low marginal product of labour in agriculture, and those who dis-
agree, is provided by the distinction between the amount of labour
time employed and the number of men employed. In a wage-pay-
ment system it is, indeed, extremely unlikely that labour would be
used up to the point where its marginal product is zero. If the wage is
positive, so will be the marginal product. But profit-maximising
behaviour is quite consistent with redundant labour. Labour is
employed up to the point where the marginal product of a unit of
labour time is equal to the wage, and disguised unemployment takes
the form of a small number of hours worked per person. It is not that
there is too much labour time but too many labourers spending it.
Total output would fall if men were drawn from the land unless
those remaining worked longer hours to compensate. How much
disguised unemployment is estimated to exist depends on what is
regarded as a normal working day. Estimates may be subjective, as
with the dynamic surplus, but unlimited supplies of labour exist in
the classical sense provided those remaining on the land work
harder or longer. Let us illustrate these points diagrammatically.
In Fig. 3.6 total output is measured on the vertical axis above the
origin, and the amount of labour time on the horizontal axis. Let
£ 1 be the point where the marginal product of labour time is equal
to the subsistence wage corresponding to total output OQ.. The
number of workers is measured on the vertical axis below the origin,
so that the tangent of the angle OT£1 (tan a) gives the average
number of hours worked by each unit oflabour. If the tangent of the
angle OXL1 is regarded as the normal length of a working day so
that the same output, OQ, could be produced by OX labour instead
of 0 r, the amount of disguised unemployment would be equal to
XT. It can easily be seen that if there was a reduction in the labour
force from 0 r to Ot and the number of hours worked per man
remained the same (i.e. tan OtS =tan OTLt), total output would
0
Total
.... p
output
"'
.E-
"'0
0
;e Subsistence wage
Labour time
(hours)
.,l.'!
""'
~
....0
...., f
.c
E
:::1
z
Fr.G. 3.6
fall from OQ to OP. If the normal working day is considered to
be greater or less than the hours given by tan b, the amount of
disguised unemployment will be greater or less than XT. Let us now
give a practical example. Suppose a producer employs 10 men
cor= 10), each doing five hours' work a day (tan a= 5), and that
the marginal product of the 50th hour is equal to the subsistence
wage (L1 = 50). If one man leaves (say Tt), total output will fall
from OQ to OP unless the 9 workers now do the 50 hours' work
previously done by 10 workers, i.e. the working day must be
increased by I of an hour. The amount of disguised unemployment
depends on what is considered to be a full day's work. Ifeight hours is
considered normal, then only 6t workers would be required to do
50 hours' work and 3f could be regarded as disguisedly unemployed.
The amount ofunderutilised labour is likely to be greater, the less
capitalistic is the organisation of agriculture. In fact, in the extreme
situation of no wage-payment system, no competitive pressure, and
little desire to maximise, the distinction between a unit of labour

and a unit of labour time becomes largely redundant in support of
the classical model. It is perhaps this type of environment which
originators of the classical model primarily had in mind. In an
extended family-type system, for example, the marginal product of
both workers and labour time may be below the subsistence wage.
Marginal
and
overage
product
p Subsistence wage
Labour time
0
Number of
workers
FIG. 3.7
It is the average product that matters for the group as a whole,
not the product of the last man or hour, and the average product
may still be above the subsistence level when the marginal product
of labour and labour time is below. It is difficult to represent both
cases on the same diagram, but if the marginal product of labour is
zero, the marginal product oflabour time is bound to be zero (and
probably negative), so we may continue to illustrate the argument
in terms oflabour time, as in Fig. 3.7.
The basis of Fig. 3.7 is the same as Fig. 3.6. When the marginal
product of labour time is zero at £2 the average product of labour
time is OP1. or PP1 in excess of the subsistence wage OP. The amount
oflabour time could be extended to OLs without the average product
of labour time falling below subsistence, and the amount of labour
time could be made up of any combination of workers and hours
worked. If the number of workers was 0 Y1 they could work hours
equal to the tangent of 0 Y1L3 without the average product of
labour time falling below subsistence. Even though the marginal
product of labour time, L 2Ls, is negative, all workers can subsist
if the total product is equally shared. A zero or negative marginal
product of labour time is not inconsistent with rational worker be-
haviour if positive utility is attached to work regardless of the effect
on output. Suppose, as in Fig. 3.8 below,l that the marginal product
w Subsistence level
Marginal utility
of leisure
24 hours -"7:::-t-~:7"::~-----~0
leisure
Fro. 3.8
of a unit oflabour time is zero after four hours' work but the mar-
ginal disutility of leisure is still negative at this point. The worker
may substitute work for leisure, working, say, six hours, despite the
fact that the marginal product of labour time is negative after the
fourth hour. If such behaviour is observed, the presumption must be
that the marginal utility attached to working exceeds the loss of
utility resulting from a lower average product. The fact that people
receive positive utility from work may partly explain why work
habits in poor countries seem more leisurely than in advanced coun-
tries. What could be done in X hours is done in X + Y hours.
1 Adapted from J. Uppal, 'Work Habits and Disguised 'Unemployment
in Underdeveloped Countries: A Theoretical Analysis', Oxford Economic
Papers (Nov 1969).
As long as the marginal product oflabour is zero, labour could be

drawn from the land without total output falling. In fact, in our
example with 0 Y1 workers, working tan a hours, labour drawn from
the land would increase total output because the marginal product
oflabour time is negative. Y1Ylabour could be released without the
labour that remains working any longer or harder. If negative
marginal product can exist in practice, this would be disguised
unemployment of an extreme form. If labour time does not extend
beyond the point of zero marginal product, however, we should
reach our earlier conclusion that disguised unemployment takes the
form of a small number of hours worked per person, which is per-
haps more accurately described as underemployment.
Incentives and the Costs of Labour Transfers

Whether workers are willing to work more intensively to compen-
sate for lost production as labour migrates, or whether capital is
substituted for labour to raise productivity, requires some discussion
of worker motivation, and attitudes to industrialisation in general,
in a predominantly agricultural economy.
Some economic incentive will almost certainly be required to
induce agricultural labour to work extra hours. At least there will
need to be goods with which to exchange their surplus production.
It is sometimes argued, however, that peasant producers, accustomed
to a traditional way of life, may not respond to such incentives -
that their horizons are so limited that they have no desire to increase
their surplus either by installing capital or working longer hours.
The corollary of this argument is that as labour productivity in-
creases, workers will reduce their hours after a certain traditional
income level is attained. This is the notion of the backward-bending
supply curve of effort, illustrated in Fig. 3.9. SS is the supply curve
of effort relating hours worked to the wage, determined by pro-
ductivity. Total income is equal to the product of hours worked and
the wage. Up to income level SWZX, supply responds positively
to the wage. Beyond the wage SW, however, fewer hours will be
offered because SWZX is the maximum total income desired. This is
the point where the positive substitution effect of work for leisure
(leisure is more 'expensive' the higher the wage) is offset by a
negative income effect because of low aspirations.
It can be said straight away, however, that there is no .firm evidence
for the proposition of a backward-bending supply curve of effort
in backward economies. The evidence suggests that, provided the
incentives are there, producers will respond.! But the need for in-
centives implies a claim on the community's real resources, which
creates added difficulties for the argument that a pool of disguised
unemployment can be used to build up 'productive' goods, and to
G> Wl-------f
"'
~
FIG. 3.9
expand the industrial sector, in a 'costless' way. Not only may the
opportunity cost of agricultural labour not be zero but resource
costs will also be involved in providing incentives to increased effort
and productivity in order for resources to be released from agri-
culture in the first place. The resource costs will include the pro-
vision of investment goods in agriculture, consumer goods for
peasant producers to buy and social capital in the industrial sector
to cater for migrants.
All this has an important bearing on the question of the valuation
of labour in surplus-labour economies, when planning the social
optimum allocation of resources and deciding on the degree to
which activities in the industrial sector should be labour-intensive.
Even if labour's opportunity cost is negligible, the resource costs
oflabour transference must be considered as a cost to the community
in expanding the industrial sector.
There is also the question of increased consumption to consider.
If the objective of a surplus-labour economy is to maximise growth,
as opposed to the level of immediate output, the transference
of labour will also involve a further ' cost' in terms of increased
consumption because there will be a reduction in the size of the
1 The evidence largely relates to the goods market, but since rising
productivity (at constant prices) has the same effect on total income as rising
prices, it may be taken as indirect evidence for the labour market. See R.
Krishna, 'Agricultural Price Policy and Economic Development', in H.
Southworth and B. Johnston (eds), Agricultural Development and Economic
Growth (Ithaca, N.Y.: Cornell University Press, 1967).
capitalist surplus iflabour is valued at its opportunity cost. Consider

Fig. 3.10 below.
The diagram represents the capitalist sector of the economy. On
the normal assumption of profit maximisation, labour will be em-
ployed up to the point where the marginal product is equal to the
industrial wage. The capitalist surplus is equal to WNX. But
M.arginal
·revenue
product
FIG. 3.10
suppose the supply of labour to the industrial sector is assumed

'costless' to society and is given, by the planners, a notional (or
shadow) wage of zero. In this planned system labour would be
employed up to the point 0£1. Given the industrial wage, OW,
and assuming the propensity to consume out of wages is unity, each
additional unit of labour employed beyond OL will involve con-
sumption in excess of production. If LL1 additional labour is em-
ployed, the size of the investible surplus would be reduced by
XX1L1. It follows that in an economy geared to growth the relation-
ship between the consumption and production of migrant labour
must also be taken into account in estimating the costs to society of
industrial expansion with surplus labour from agriculture. If at the
margin additional saving is valued more than an additional unit of
consumption, the cost of a unit oflabour transferred from agriculture
to industry must include an allowance for increased consumption.
These matters are taken up more fully in the appendix to Chapter 7
on the planning wage (pp. 212-23 below).
4
Capital and Technical Progress
The Role of Capital in Development
The capital stock of a country increases through the process of net
investment (I), which is the difference between a country's net income
in an accounting period (i.e. gross income minus depreciation) and
how much it consumes out of that income in the same period. Capital
accumulation enlarges a country's capacity to produce goods. As
we saw in Chapter 2, however, production function studies, at least
for advanced countries, cast some doubt on whether capital accumu-
lation by itself is central to the development process. A lot depends
on how capital is defined and whether it is inclusive of technological
progress. Capital is certainly a wider concept than capital goods
traditionally defined, i.e. goods which yield no immediate utility
but produce goods which do. If capital is defined as any asset which
generates an additional future stream of measurable income to
society, many goods and services commonly regarded as consump-
tion goods ought strictly to be included as part of a country's
capital stock. Expenditure on education, for instance, which may
permanently enhance the earning capacity of individuals, as well
as giving immediate satisfaction, must be regarded partly as invest-
ment expenditure. Similarly, if certain types of 'consumption'
goods, e.g. clothes, durable consumer goods, etc., are necessary to
induce peasant producers in the agricultural sector to increase
their productivity, they, too, ought to be considered as part of the
capital stock. If it is agreed, therefore, that the only way to build
up a country's productive potential, and to raise per capita income,
is to expand the capacity for producing goods, this need not refer
simply to the production of physical capital but to the production of
other types of capital such as 'incentive' consumer goods and the
expansion of facilities for investment in human capital, all of which
can contribute to increased productivity and higher living standards.
In addition, there is the very important type of capital referred to as

social capital which again need not produce other goods directly,
but none the less expands the capacity to produce by facilitating the
smoother operation of directly productive activities. Housing and
transport facilities are obvious examples. If the production function
approach to the analysis of growth is to be employed, it is of the
utmost importance to define investment and the capital stock as
meaningfully as possible if the relation between capital and growth
is not to be misconstrued. This is in addition to the fact, which we
dwelt on at some length in Chapter 2, that capital accumulation
may be the main vehicle for the introduction of technical progress
in the productive system. In this respect the rate of technical pro-
gress in the capital-goods sector itselfis vitally important because this
will determine the price of capital relative to labour.
Developing economies themselves lay great emphasis on the im-
portance of capital accumulation, and stress the need to raise the
level of investment in relation to output. A glance at any national
development plan will testify to this. Development is associated with
industrialisation and industrialisation with capital accumulation.
Many development economists also see investment as the most
important single factor in the growth process. Professor Rostowl
defines the process of' take-off' into sustained growth in terms of a
critical ratio of investment to national product, and Professor Arthur
Lewis has described the process of development as one of trans-
forming a country from being a 5 per cent saver and investor to a
12 per cent saver and investor. 2 It is common, in fact, for countries
to calculate fairly precise ratios of investment to national income that
will be required either to achieve a particular rate of growth or to
prevent capital per head or income per head from falling. These
calculations involve assumptions about the normal relation between
capital and output, a relation which is formally expressed in the
concept of the capital-output ratio (which we shall consider later).
Professor Johnson, too, sees capital accumulation in its widest
sense as the distinguishing characteristic of development, and the
structural transformation of economies as a generalised process
of capital accumulation: 'The condition of being "developed"
consists of having accumulated, and having established, efficient
social and economic mechanisms for maintaining and increasing
1 Rostow, The Stages of Economic Growth.
2 A. Lewis, The TheoryofEconomicGrowth (London: Allen & Unwin, 1955).
CAPITAL AND TECHNICAL PROGRESS 99
large stocks of capital per head in the various forms. Similarly the
condition of being "underdeveloped" is characterised by the pos-
session of relatively small stocks of the various kinds of capital.' 1
The emphasis on investment in less developed countries may not be
without foundation despite the findings for advanced countries that
capital accumulation seems to have made only a relatively minor
contribution to measured growth in the present century. For one
thing, capital is a broader concept than some of these studies admit.
Secondly, even if capital is relatively unimportant in developed
countries, there may be a fundamental difference between countries
which already have large quantities of capital per head and those
which do not. In countries where specialisation and the division of
labour is minimal, the scope for capital to permit more roundabout
methods of production, and to increase productivity, may be greater
than where specialisation has already reached a high level of sophis-
tication. Moreover, in technologically backward countries the rate
of growth of capital required to absorb new technology is likely to
be greater than in advanced countries. By definition, technologically
backward countries also have a backlog of technology to make up.
Furthermore, in a labour-abundant economy, with a low capital
to labour ratio, the very act of capital deepening - giving each
worker a little more capital to work with - may make a substantial
difference to total product - much more so than in countries where
the process of capital deepening has been a continuing process for
some length of time. All these factors represent important contri-
butions that capital can make to economic progress, which may be
relatively more important the smaller is the initial capital stock of a
country relative to its population. It is a familiar economic proposi-
tion that the scarcer is one factor of production in relation to another,
the higher is its productivity, other things being equal.
Capital accumulation is also seen as the escape from the so-called
'vicious circle of poverty'- a circle oflow productivity, leading to
low per capita income, leading to a low level of saving per head,
leading to a low level of capital accumulation per head, leading to
low productivity. Low productivity is seen as the source of the
'vicious circle of poverty', and the point where the circle must be
broken by capital accumulation.
1 H. G. Johnson, 'Comparative Cost and Commercial Policy Theory for
a Developing World Economy', Pakistan Development Review, supplement
(spring 1969) p. 9.
Since productivity can be raised by other means as well, there is

implicit in the argument the pessimism that no further reorganisa-
tion of the existing factors of production could have much impact on
output, and that technical progress is either relatively static or mainly
endogenous, requiring net additions to the capital stock. Whether
this pessimism is justified is an empirical question in the final analysis,
but the emphasis placed by less developed countries on capital
accumulation is undeniable.
The Capital-Output Ratio

The link between capital and output is embodied in the concept of
the capital-output ratio which can measure in physical units, or in
value terms, either the average or the marginal relation between
capital and output. The average capital-output ratio of an economy
is the stock of capital divided by the annual flow of output (K/0),
while the marginal or incremental capital-output ratio (I.C.O.R.)
measures the relation between increments to the capital stock and
increases in output ('8Kf'80 or 1/'80). If the gross value of a country's
stock of capital is £100 million and its gross annual income is £25
million, the average capital-output ratio would be 4. If the economy
grows smoothly at full capacity, the incremental capital-output
ratio will approximate to the average, so that in this example four
extra units of capital would be required to produce an additional
unit of output (in £s).
With fluctuations in income, however, the average capital-
output ratio may differ substantially from the I.C.O.R. In periods
of recession the average capital-output ratio will tend to be higher
than its 'normal' value since output will be depressed in relation to
the size of the capital stock which is relatively fixed. But while the
average ratio is still high at the start of the upturn of the business
cycle, the incremental ratio will appear low. If the economy is
working below capacity, very little extra capital will be needed to
increase output and substantial growth may be associated with a
relatively small I.C.O.R. In attempting to calculate I.C.O.R.s,
the stage of the business cycle must always be borne in mind.
The dependence of the I.C.O.R. on the stage of the business
cycle raises the question of whether it is legitimate to treat the
capital-output ratio as an independent variable in the economic
system, and to use the ratio as a parameter in investment planning.
Is not the ratio the dependent variable in the system determined by
the level of investment and the rate of growth? This is a fairly
crucial question to consider, because it is common for countries to
use an aggregate I.C.O.R., and also incremental capital-output
ratios for sectors of the economy, for deciding on the rate of invest-
ment necessary to achieve a particular target rate of growth of
income. Yet, if the ratio itself is determined by the rate of growth,
is this permissible? As we shall come to see, a lot depends on the
length of the time period over which investment is being planned.
Taking both time-series and cross-section data, there is substantial
empirical evidence of a strong inverse relation between growth and
the value of the I.C.O.R.,1 and good theoretical reasons for believing
that the I.C.O.R. is the dependent variable in the relation. Two main
reasons can be advanced. Firstly, the investment rate tends to be
more stable than other factors so that growth may increase or
decrease while the stock of capital remains virtually unchanged.
The inverse relation would be less marked if capital-output ratios
were computed by taking a measure of capital services, rather than
the stock of capital, but this is not normally the case. Secondly, if the
significance of non-capital inputs in the growth process is greater
than that of capital, there will be an inverse relation between in-
vestment and the capital-output ratio. This means that if the rela-
tion between growth and investment is positive (which no one
disputes), the relation between growth and the capital-output ratio
will be negative.
Let us illustrate these points by making use ofthe Cobb-Douglas
production function. From the equation in its estimating form,
0 = T + ~L + rxK, the I.C.O.R. is given by ~ = T + ~f+ rxK
If L or T varies more than K, 0 and KfO will vary in opposite
directions. Likewise, if L contributes more to growth than K, 0 and
K/0 will vary in opposite directions. In the short run, therefore, a
low measured I.C.O.R. more probably implies that growth has been
rapid, and not that relatively rapid growth will be associated with a
low ratio of investment to output.
The use of the incremental capital-output ratio for estimating the
1 E.g. see H. Leibenstein, 'Incremental Capital-Output Ratios and

Growth Rates in the Short-Run', Review qfEconomics and Statistics (Feb 1966);
andJ. Vanek and A. Studenmund, 'Towards a Better Understanding of the
Incremental Capital-Output Ratio', Quarterly Journal of Economics (Aug
1968).
amount of investment required to increase output by a given

amount was inspired by the work of Harrod and Domar.l Although
their work was originally designed to establish the conditions for
equilibrium growth, their fundamental equations have been put to
various alternative uses. Harrod's basic growth equation is GC = S,
where G is the rate of growth (80/0), Cis the I.C.O.R. (IJW) and
Sis the savings ratio (S/0). By substituting terms it can be seen that
the equation is merely the tautology that saving equals investment
ex post, which follows from the accounting identities that income
equals consumption plus investment and income equals consumption
plus saving. None the less, it is this equation that has been endlessly
manipulated to give the savings (or investment) ratio (S) necessary
to achieve a particular target rate of growth (G), given the value of
C, and to calculate the value of C, given G and S. This is in spite of
the fact that no causality is implied by the equation.
The Harrod equation certainly predicts an inverse relation
between growth and the I.C.O.R., but since Cis assumed constant
in Harrod's original model, it is the I.C.O.R. that determines
growth and not the other way round.
It is not difficult to demonstrate that the exercise of calculating
the savings and investment required to achieve a target rate of
growth, given C, and the exercise of calculating C, given S and G,
are fraught with danger. If the economy has been operating below
capacity, there is no telling what is the 'true' value of C, and the
investment needed in the short run to produce a given G cannot be
calculated with the precision that the equation suggests. Similarly,
it is highly misleading to calculate C by dividing the savings or
investment ratio by the rate of growth. Apart from the fact that
the economy may not be working at full capacity, one is assuming
that all increases in output are attributable to increases in capital,
which is patently untrue. The productivity of capital (WJI) is
overstated, and, since the I.C.O.R. (l/80) is the reciprocal of the
productivity of capital, the I.C.O.R. is understated.
This latter problem can be overcome by making the distinction
between the actual I.C.O.R., as measured above, and the adjusted
I.C.O.R., which is the I.C.O.R. adjusted for increases in the supply
of other factors (e.g. a 5 per cent increase in the labour force). The
concept of the adjusted capital-output ratio takes care of what
otherwise appears to be a conflict between the observed productivity
1 See above, p. 4, notes 1 and 2.
of capital and that implied by the value of the actual I.C.O.R.
For example, suppose that a country's investment ratio is 15 per cent
and its growth rate is 5 per cent, giving an actual capital-output
ratio of 3. This would suggest a productivity of capital of 33·3 per
cent. But typically, the rate of return on capital is much less than
this. The reason is that other factors contribute to growth. Suppose
that the growth of the labour force is such that it contributes 3·5
percentage points to the 5 per cent growth rate, leaving 1·5 per-
centage points to be 'explained' by capital (assuming no technical
progress). In this case, with a 15 per cent investment ratio, the ad-
justed capital-output ratio is 10, giving a rate of return on capital
of 10 per cent. The actual capital-output ratio is 3, but if there was
no growth of the labour force, the adjusted capital-output ratio
tells us that an investment ratio of 50 per cent would be required to
grow at 5 per cent. In short, if other factors remain unchanged, much
more investment is required to achieve a target rate of growth than
is implied by calculations of C from the Harrod equation. The ad-
justed I.C.O.R. is sometimes referred to as the net incremental
capital-output ratio; that is, the I.C.O.R. on the assumption that
other factors do not change. Net or adjusted I.C.O.R.s are difficult
to estimate, but clearly the variability of the factors that co-operate
with capital in the productive system can play havoc with calcula-
tions of the actual I.C.O.R. from data on past growth and the in-
vestment ratio alone.
When calculations are made of saving and investment require-
ments to achieve target rates of growth for, say, a five- to ten-year
planning period, the normal assumption is that the incremental
capital-output ratio will approximate to the average and that over
the long term the factors to co-operate with capital will be forth-
coming as in the past. For a planning period in excess of five years
this assumption may not be unreasonable. Criticisms of the use of the
incremental capital-output ratio in investment planning are more
confined to its use in the short run.
On the other hand, as a result of technical progress, and shifts
in the pattern of demand and the distribution of resources between
sectors with difference productivities, the I.C.O.R. may also be
subject to secular change. In principle, this should be easier to allow
for in investment planning. In practice, in the absence of extensive
time series, there may be some dispute as to the direction of the
trend. Some economists argue that it is likely to fall as per capita
incomes rise owing to greater economies of scale, external econo-

mies, improvements in the quality of labour and shifts in demand
towards service activities requiring less capital. Others argue,
however, that it can be expected to rise owing to diminishing re-
turns if capital grows faster than labour, which is normally the case
as countries get richer. The only reliable evidence we have relates
to advanced countries and suggests that, in practice, long-run
changes in the capital-output ratio are quite small, with the factors
leading to increases and decreases in the ratio presumably off-
setting one another.
In the short run, though, as we have said, the ratio may be
subject to serious variability either as a result of variations in the
supply of co-operating factors or variations in capacity utilisation.
In planning the level of investment the issue that must be faced is
whethez: the planning period is long enough for the short-run
variability of the ratio not to matter. If the ratio does stay fairly
constant over long periods, capital requirements can perhaps be
estimated with some accuracy for a ten- to fifteen-year period, but
planning investment requirements for anything less than a five-year
period on the assumption of a stable capital-output ratio may be
asking for trouble.
Finally and briefly, other more obvious dangers may be men-
tioned of using the capital-output ratio as a basis for investment
planning. Firstly, exhaustive attention to the capital-output ratio
may exaggerate the need for investment when output may be in-
creased by other, simpler means. On the basis oflndian experience,
Professor Reddaway suggests that before calculating the additional
capital required to produce a target output, it would be wiser for a
less developed economy to think first of the output that might reason-
ably be expected from increased utilisation of existing factors and
better methods applied to old capital.l Secondly, the accuracy of
any calculations of the capital-output ratio may be called into
question in backward economies lacking comprehensive and
reliable statistics. Thirdly, it must be remembered that the aggregate
I.C.O.R. is bound to disguise sectoral differences, and in calculating
overall investment requirements some allowance should be made
for the changing sectoral distribution of resources.
1 W. Reddaway, The Development of the Indian Economy (London: Allen &
Unwin, 1962).
Technical Progress
The term 'technical progress' is used in several different senses to
describe a variety of phenomena; but three, especially, can be
singled out. Firstly, it is a term used by economists to refer to the
effects of changes in technology, or more specifically the role of
technical change in the growth process. It is in this sense that we
used the term in Chapter 2; that is, as an umbrella head to cover all
those factors which contribute to the growth of' total' productivity.
Secondly, technical progress is used by economists in a narrow
specialist sense to describe the character of technical improvements,
and is often prefaced for this purpose by the adjectives 'labour-
saving', 'capital-saving' or 'neutral'. Thirdly, technical progress
is used more literally to refer to changes in technology itself, defining
technology as useful knowledge pertaining to the art of production.
Used in this sense, the emphasis is on describing improvements in
the design, sophistication and performance of plant and machinery,
and the economic activities through which improvements come
about - research, invention, development and innovation.
Having already discussed technical progress in the first sense, we
shall concentrate here on the narrow specialist descriptions of tech-
nical progress, and on how societies progress technologically.
Capital- and Labour-saving Technical Progress

The classification of technical progress as to whether it is capital-
saving, labour-saving or neutral owes its origins primarily to the
work of Harrodl and Professor Hicks.2 Their criterion of classifi-
cation differs, however. Harrod's classification of technical progress
employs the concept of the capital-output ratio. Given the rate of
profit, technical change is said to be capital-saving if it lowers the
capital-output ratio, labour-saving if it raises the capital-output
ratio, and neutral if it leaves the capital-output ratio unchanged.
This criterion of technical progress is an amalgam of the effect of
'pure' technical change on factor combinations and the effect of
the substitution of capital for labour (as, for example, relative factor
prices change). As such, Harrod neutrality at the aggregate level
is quite consistent with capital-saving technical progress at the
1 R. Harrod, Towards a Dynamic Economics (London: Macmillan, 1948).
2 The Theory of Wages (London: Macmillan, 1932).
J. Hicks,
industry level. In fact, most of the evidence for advanced countries

suggests that if technical progress is neutral in the aggregate in the
Harrod sense, this must be due to substitution because 'pure'
technical advance has saved capital.
But there is some dispute as to what the aggregate capital-output
ratio does actually show, because of alleged biases in the estimates.
This is in addition to differences of opinion over the direction of the
ratio in the course of development. Some argue that the ratio has
fallen and may be expected to fall; others maintain that the ratio
has risen and may be expected to rise as development proceeds. The
difficulty stems from the perennial problem of defining capital.
If certain expenditures, which ought properly to be regarded as
adding to the capital stock, are excluded from the measurement of
capital, and these expenditures have grown faster than measured
capital, the capital-output ratio will be biased downwards. While
technical progress may appear neutral or capital-saving because of
a stable or declining capital-output ratio, it may nevertheless be
capital-using and would appear so if the capital stock were more
appropriately measured. One notable item not conventionally
included in the measurement of the capital stock is investment in
human capital. In most advanced countries during the twentieth
century investment in human beings has grown faster than tan-
gible capital. To the extent that this is true, estimates of the capital-
output ratio are biased downwards and would probably show a slight
upward trend during the twentieth century if investment in human
capital was included in the measurement of capital. We merely stress
again the need for defining capital as meaningfully as possible if
the relation between factor supplies and growth is to be properly
understood.
One important economic reason why technical progress may
appear to be capital-using in the Harrod sense is that as countries
become richer the price of labour relative to capital tends to rise,
which may induce a substitution of capital for labour. As countries
mature, capital tends to grow faster than labour and the resulting
change in relative prices may not only induce 'pure' substitution
but also induce inventive effort towards saving labour which is
becoming relatively scarce.
Hicks's classification of technical progress takes the concept of
the marginal rate of substitution between factors, which is the rate
at which one factor must be substituted for another leaving output
unchanged. The marginal rate of substitution is given by the ratio
of the marginal products of factors. Holding constant the ratio
oflabour to capital, technical progress is said to be capital-saving if it
raises the marginal product oflabour in greater proportion than the
marginal product of capital; labour-saving if its raises the marginal
product of capital in greater proportion than the marginal product of
labour; and neutral if it leaves unchanged the ratio of marginal
products. These definitions are illustrated in Figs 4.1, 4.2 and 4.3
respectively.
Kl
2
·a.
0
<.) K2 CAPITAL-SAVING
K3
L2 L
Labour
FIG. 4.1
It will be recalled from Chapter 2 that technical progress on a
production function map is represented by shifts in the function
towards the origin showing that the same output can be produced
with fewer inputs, or that the same volume of inputs can produce
a greater output. According to the shape of the new production
function, either fewer of one or both factors will be required to pro-
duce the same output. In the case of neutral technical progress some
of both factors can be dispensed with. In the case of non-neutral
technical progress, if only one factor is saved technical progress is
said to be absolutely labour- or capital-saving. Iffewer of both factors
are required, technical progress is said to be relatively labour- or
capital-saving.
Consider first neutral technical progress (Fig. 4.3). The ray from
the origin, or expansion path, OZ, goes through the minimum-cost
point of tangency between the production function rr and the
factor-price ratio line, KL. With neutral technical progress the pro-
duction function shifts such that the new point of tangency at the
same factor-price ratio lies on the same expansion path. This means
that the ratio of marginal products is the same at the same capital
to labour ratio, and equal proportionate amounts of the two factors
are saved. The condition for neutral technical progress is simply

that the new production function is parallel to the old.
With labour-saving technical change (Fig. 4.2) the ratio of the
marginal product of capital to the marginal product of labour rises
LABOUR- SAVING
y
such as to shift the minimum-cost point of tangency from the old

expansion path OZ to a new expansion path, OZ1· At P1, where
the new production function cuts the old expansion path, the ratio
of the marginal product of labour to capital is lower than at P.
P1 is not an equilibrium point and it will pay producers to move to
K
NEUTRAL
OL-------~~~----~----~---
L3 L 2 L1 L
Labour
FIG. 4.3
point Q, substituting capital for labour. The ratio of marginal pro-
ducts has not remained unchanged at a constant labour to capital
ratio, and L2La labour is saved. The isoquants have been so
dra'Yn as to keep the volume of capital the same, but this is purely
incidental.
Capital-saving technical progress (Fig. 4.1) may be described
in ~ctly analogous fashion. In this case the ratio of the marginal
product of labour to the marginal product of capital rises and the
shift in the production function is such that the minimum-cost point
of tangency now lies to the right of the old expansion path. At P1,
where the new production function cuts the old expansion path,
the ratio of the marginal product of labour to capital is higher than
at P. Again, P1 is not an equilibrium point and it will pay producers
to move to point Q, substituting labour for capital. The ratio of
marginal products has not remained unchanged at a constant
labour to capital ratio, and in this case K2Ks capital is saved.
As with Harrod technical progress, it is difficult to know what form
Hicks technical progress takes in practice, largely because of identi-
fication problems. While the classification is analytically distinct,
how does one distinguish empirically between a change in factor
proportions due to a shift in the production function and a change in
factor proportions due to a change in relative prices? Hicks himself
seemed to be of the view that technical progress is relatively labour-
saving, but the indirect evidence we have for this is slight. For
example, given the magnitude of the rise in the price of labour
relative to capital and an elasticity of substitution of close to unity,
labour could not have maintained or increased its share of the
national income (as it has done slightly in some advanced countries)
if technical progress was markedly biased in the labour-saving
direction. If technical progress is biased in one direction or another,
its major impact will probably be on factor utilisation. The type of
technology employed, and the factor proportions it entails, must bear
a major responsibility for the present level of unemployment in less
developed countries.
How Societies Progress Technological{y

Improvements in the art of production, which is the most literal
interpretation of technical progress, result from a combination of
research, invention, development and innovation. Research and
invention are the activities which 'create' knowledge, and develop-
ment and innovation are the activities which apply new knowledge
to the task of production. These are all basically economic activities.
But the study of the way in which societies progress technologically,
and the speed of progress, is not only the preserve of the economist.
The economist can identify the mainsprings of progress, but their
pervasiveness and acceptance in societies is not a purely economic
matter. The spread of new knowledge, for example, depends on its

rate of adoption and diffusion and this raises questions of individual
motivation, the willingness to assimilate new ideas and to break with
custom and tradition, which impinge heavily on territory occupied
by development sociologists. The relative importance of different
factors contributing to progress, and the speed of progress itself,
will vary from country to country according to its stage of develop-
ment and a whole complex of social and economic forces. More-
over, many of the mainsprings of technological progress are not
mutually exclusive. At the risk of excessive simplification, attention
here will be confined to four main sources of progress which are of
potential significance to any society.
One major source of improvement in technology, and progress, is
the inventive and innovative activity of the native population. All
societies are endowed to some degree with a potential supply of
inventors, innovators and risk-takers and, in the absence of imported
technology and personnel, it is on the emergence of this class of
person that technological progress will primarily depend in the
early stages of development. Economic backwardness in many
countries may quite legitimately be traced back to a relative shortage
of inventors, innovators and risk-takers. It is fairly well established
that some cultures and some environments are more amenable to
change than others, and in the past have produced a greater supply
of entrepreneurs. The view is frequently expressed that the major
source of growth during Britain's industrial revolution was primarily
technological progress fostered by an abundant supply of inventors,
innovators, entrepreneurs and risk-takers, with the accumulation of
capital playing an essentially secondary role. Schumpeter has laid
great stress on the role of the entrepreneur and innovation in the
development process.l Ultimately, however, it is the lag between the
creation of knowledge and its adoption, and the rate of dissemination
of new knowledge, that most directly affects the rate of measured
technical progress between countries; and these two facets of inno-
vation are intimately connected with the attitudes of society at
large. For Schumpeter, progress results from what he calls the 'pro-
cess of creative destruction', which is bound up with innovation,
and instigated by competition. Innovation, in turn, is the driving
1 E.g. in J. Schumpeter, Capitalism, Socialism and Democracy (London:
Allen & Unwin, 1943) and The Theory of Economic Development (Cambridge,
Mass.: Harvard University Press, 1934).
force behind competition. But innovation requires decision-takers
and hence his complementary stress on the role of the entrepreneur.
A characteristic of many poor countries is a shortage of decision-
takers, a relative lack of competitive spirit and a general aversion to
risk-taking. These may be partly cultural traits and also partly, if
not mainly, a function of the stage of development itself. The charac-
teristics commonly associated with business dynamism are themselves
a function of business, and more particularly the form of organisa-
tion we call capitalism.
In a closed economy, it might be said that invention is a necessary
but not a sufficient condition for progress. In an open economy, it is
not even a necessary condition because technology embodying new
knowledge can be imported from abroad, from societies more
technically advanced. If backward economies lack inventors they
can still innovate by applying technology developed abroad. Tech-
nical assistance programmes in less developed countries, organised
by the developed countries, are designed to foster innovation,
particularly the adoption of new techniques and new goods. Societies'
exposure to new goods may be a powerful factor in economic develop-
ment by acting as an incentive to producers and workers to increase
their surplus for exchange. The spread of technology and ideas may
also be expected to come about naturally in the general process
of commercial intercourse, and the exchange ofinformation through
trade. The speed with which advanced technology is absorbed by
economically backward countries will depend on the same class of
factors as the diffusion of knowledge within countries - which in the
final analysis amounts to the receptiveness of all sections of the
community to change. The absorption of technology from abroad is
referred to as the cultural diffusion process, and again its study in-
volves sociological analysis as well as economics.
Learning
A third means by which societies progress technologically, gradually
raising their efficiency and productivity, is through the process of
'learning by doing', which refers to the accumulation of experience
by workers, managers and owners of capital in the course of pro-
duction which enables productive efficiency to be improved in the
future. It is a learning process that Adam Smith referred to in
discussing the division of labour. Smith stressed the importance of
the division of labour for three main reasons: firstly, as a means of

improving the dexterity of workers; secondly, to save time which is
lost in the absence of specialisation; and thirdly, to encourage the
invention ofmachines which facilitate and abridge labour to improve
the productivity of labour. All these advantages of the division of
labour are part of a learning process. Labour improves its skill
through specialisation and work experience, and becomes more
adept at the job in hand. Managers see deficiencies in organisation,
which can subsequently be remedied; and on the basis of accumu-
lated knowledge they are also able to embody more productive
techniques in the capital stock.
Learning may be regarded as either endogenous or exogenous,
or both, depending on the factor of production considered. Ifexisting
labour and existing capital is subject to a learning process, then
learning by doing can be regarded as exogenous and a part of dis-
embodied technical progress. If, however, it is assumed that learning
enters the productive system only through the addition of new factors,
then learning by doing must be regarded as endogenous. This is
the basis of Arrow's capital model, from which the term 'learning
by doing' originates.l His hypothesis with respect to capital is that
at any moment of time new capital goods incorporate all the know-
ledge then available based on accumulated experience, but once
built their productive efficiencies cannot be altered by subsequent
learning.
The endogenous model may be more appropriate in the case of
capital but is much less relevant in the case oflabour. It is in relation
to labour that most research into the learning process has been
conducted. The notion of the learning curve, or progress function,
which has been found in many industries, relates direct labour
input per unit of output to cumulative output as the measure of
experience. Typically, labour input per unit of output is found to
decline by between 10 and 20 per cent for each doubling of cumula-
tive output, with a corresponding rise in the productivity oflabour.
For any one product, of course, learning cannot go on at the same
rate for ever, but since product types are constantly changing it is
probably safe to conclude that in the aggregate there is no limit to
the learning process. For economies as a whole the residual factor in
the growth equation must partly reflect the process of learning by
1 K. Arrow, 'The Economic Implications of Learning by Doing', Review
of Economic Studies (June 1962).
doing. There is no easy way of adjusting the labour input series for
learning to include it as part of the contribution oflabour to growth.
Education
Finally, we come to the relation between technological progress
and improvements in the health, education and skills of the labour
force, or what is commonly called investment in human capital.
Investment in human capital takes many different forms, including
expenditure on health facilities, on-the-job and institutional training
and re-training, formally organised education, study programmes
and adult education, etc. Investment in human capital can over-
come many of the characteristics of the labour force that act as
impediments to greater productivity, such as poor health, illiteracy,
unreceptiveness to new knowledge, fear of change, a lack of in-
centive and immobility. Improvements in the health, education and
skill of labour can increase considerably the productivity and
earnings oflabour and may be preconditions for the introduction of
more sophisticated, advanced technology applied to production.
The capacity to absorb physical capital may be limited, among
other things, by investment in human capital. It is in this respect
that there is likely to be a close interrelationship between the main-
springs of technological progress.
Detailed empirical work on the relation between education and
growth relates largely to advanced countries. We gave some of the
results of such studies in Chapter 2. Now let us elaborate in a little
more detail. Professor Schultz, in his Presidential Address to the
American Economic Association in 1961,1 was one of the first to
suggest that growth in America 'unexplained' by conventional
factor inputs might be due to the rapid increase in the quality of
labour through education. Similarly, it may be the increase in
education in the twentieth century that accounts for the substantial
rise in earnings per worker. According to Schultz the stock of edu-
cation in America rose by approximately 850 per cent between
1900 and 1956 compared with an increase in reproducible capital
of450 per cent (at constant 1956 prices). Schultz acknowledged the
difficulties in estimating the return to education, but argued that
even when every conceivable cost is considered, and all expenditure
1 T. Schultz, 'Investment in Human Capital', American Economic Review
(Mar 1961).
is treated as investment and none as consumption,l the return to

investment in education is at least as high as, if not higher than,
that to non-human capital. Becker's2 lower estimate of the social
return on male college graduates in America, excluding 'spill-
overs', is 12·5 per cent, which is the same as Blaug's3 estimate for
Britain for the last three years of secondary schooling. Taking 60 per
cent as the measure of 'unexplained' growth between 1929 and
1956, Schultz concluded that between 30 and 50 per cent of this
residual can be taken as representing a return to the increased
education of the labour force. Bowman4 reaches similar con-
clusions. Using Schultz's data on the stock of education, and
applying rates of return of 9, 11 and 17·3 per cent, Bowman has
calculated corresponding percentage contributions of the growth of
education to national income of21, 26 and 40 per cent respectively.
Denison5 also reaches roughly the same conclusion, using a more
direct approach which has now become fairly standard and is
perhaps worth outlining. The method involves two steps. The first
involves gathering information on the distribution of the labour
force by amounts of schooling at different dates. The second step
involves collecting information on income differences between
educational cohorts with different amounts of schooling embodied
in them, which are then used as weights to derive an index of the
improvement in the quality of labour due to education on the
assumption that a certain percentage of differences in earnings are
due to differences in the amount of education. Suppose, for instance,
that the earnings differential between those with eight years'
schooling and those with ten years' schooling is 20 per cent, that
one-half of the difference is assumed to be due to the extra two years'
schooling, and that a person with eight years' schooling is treated as
one unit; then the person with ten years' education is counted as
I+ (0·5 X 0·2) = 1·1 units. On this basis, but assuming that 60 per
cent of income differences are due to education, Denison estimates
that the expansion of education in America over the period 1929-5 7
1 The greater the proportion of expenditure treated as pure consumption,
the higher the rate of return on the investment component.
2 G. Becker, Human Capital (New York: Columbia University Press, 1964).
s M. Blaug, 'The Rate of Return on Investment in Education in Great
Britain', Manchester School (Sep 1965).
4 M. Bowman, 'Schultz, Denison and the Contribution of Education
to National Income Growth', Journal of Political Economy (Oct 1964).
5 Denison, The Sources of Economic Growth in the U.S.
raised the quality of labour by the quantity equivalent of 29·6 per
cent or 0·93 per cent per annum. Taking the annual growth rate
of 2·93 per cent and assuming an elasticity of output with respect to
labour of 0·73, this gives a contribution of education to measured
growth of 23 per cent (i.e. (0·73 X 0·93)/2·93 = 0·23). The effect of
education on the growth of per capita income is even more striking.
Denison gives the contribution as 42 per cent, and Schultz puts the
contribution at between 25 and 45 per cent for the same period.
These results are very impressive, and furthermore there are good
reasons why the calculations may be underestimates. For one thing,
the methodology employed forgets the role of education in maintain-
ing the average quality of the labour force; secondly, no allowance is
made for improvements in the quality of education; and thirdly,
there are all the 'spillovers' from education to consider, such as the
contribution of education to knowledge and its diffusion throughout
society. Becker wants to attribute the whole of Denison's residual
to these 'spillovers' and to double his own estimate of the social
rate of return to college education as a result.
It is the apparent importance of education in the growth process
in developed countries that has invoked the response that investment
in human capital may be as important as investment in physical
capital in less developed countries. Assuming the analyses of
developed countries to be sound, the question is whether it is legiti-
mate to draw policy conclusions for less developed countries from
the experience of developed countries. Can predominantly Western
experience be used to forecast the returns to education in the poorer
nations of the world? The potential dangers of doing so have been
forcefully set out by Balogh and Streeten.l How convenient, they
say, to elevate a statistical residual to the engine of development and
by doing so convert ignorance into 'knowledge'!
Apart from the problem of deciding how much of earnings
differences between people of the same age are due to differences in
the amounts of education embodied in them, the first point is that
education is not a homogeneous commodity. The returns to dif-
ferent types of education may differ radically between backward and
advanced countries owing to differences in supply (cost) and demand
conditions. A high return to training in physics in America, say,
does not imply the same high return in Zambia. The returns to
1 T. Balogh and P. Streeten, 'The Coefficient oflgnorance', Bulletin of the
Oxford Institute of Statistics (May 1963).
different types of education and training for different age groups

and types of institution must be worked out in individual countries
before policy conclusions are reached. Edw;ation for education's
sake, except perhaps at the primary level, would involve inexcus-
able resource waste. The greatest need is to decide on the type of
educated manpower required and to plan accordingly with the aid
of manpower budgets. In this way skill bottlenecks can be eased and
unemployment among the educated avoided. An unemployed in-
telligentsia, which is already a reality in some poor countries, is
not only an avoidable waste but may even be inimical to develop-
ment.
One of the strongest associations in less developed countries is
between levels of per capita income and the proportion of the popu-
lation in primary education, and the evidence is that educational
expansion at the elementary level has led growth. Sustained growth
seems to be more strongly associated with the percentage of children
enrolled in primary schools than any other factor connected with
human capital. Lewis has remarked that 'allowing for some ab-
sorption into farming and for the expansion of non-agricultural
employment, a developing economy needs to have at least 50 per
cent of its children in primary schools' .1 The hypothesis is that tradi-
tional customs and attitudes cannot be changed significantly until
a large section of the community at a fairly young age is exposed to
new ideas and ways of doing things. The relevance of primary
education in this regard may partly explain why traditional agri-
culture has been transformed in some countries but not others.
The fact that the capacity of a country to absorb physical capital
and technological progress may be constrained by the availability
of human capital, however, is not a sufficient reason for giving it
preferential treatment. As a general rule, both types of capital
formation need to be considered together in the planning process
and carried on simultaneously. Ultimately, the amount of resources
devoted to investment in human capital is an allocative problem
that each country must solve itself, on which no hard-and-fast
rules can be laid down a priori. All that can be done in general is to
point to the potential benefits of education, and investment in human
beings in the widest sense.
1 A. Lewis, Development Planning (London: Allen & Unwin, 1966) p. 109.
5
The Persistence
of Underdevelopment
It is easy to argue that poverty and backwardness are due to a
general shortage and inefficient use of the key factors of production;
it is much harder to determine precisely why there should be a
dearth of some factors and an abundance of others, and why de-
velopment may be a slow and lengthy process. It is certainly im-
possible to explain present-day international discrepancies in the
level of development with reference to initial differences in factor
endowments. The present development gap has arisen largely
through industrial development in certain selected areas of the
world which has, in turn, generated its own factor endowments.
The purpose here, however, is not to consider why some countries
were able to industrialise sooner than others, but rather to consider
some of the potential obstacles to growth in the present less developed
countries, from which the developed countries do not suffer.
There is no shortage of hypotheses to explain why per capita
income growth may be slow in the early stages of a country's transi-
tion from feudalism to take-off, and why countries which are slow
to develop may find themselves at a permanent comparative dis-
advantage compared to those countries which develop quickly. In
this chapter, three popularly advocated hypotheses will be con-
sidered: firstly, the thesis of dualism; secondly, Myrdal's hypothesis
of the process of cumulative causation;! and thirdly, the so-called
population 'problem'. It has been argued that dualism, the process
of cumulative causation, and rapid population growth are all fac-
tors which may depress per capita income growth and contribute to
the persistence of poverty in relatively backward economies.
While it is virtually impossible to test these hypotheses empirically,
1 G. Myrdal, Economic Theory and Underdeveloped Regions (London: Duck-
worth, 1957; paperback ed., Methuen, 1963).

a general discussion and assessment of them will facilitate greater

understanding of the analysis in Chapter 1 of the growing relative
and absolute per capita income gap between the present developed
and less developed countries.
Dualism
Dualism is a description of a state of affairs in which less developed
countries may find themselves in the early stages of development,
and which may have significant implications for their future course
of development. There are a number of possible definitions and
interpretations of the term 'dualism', but in the main it refers to
economic and social divisions in an economy, such as differences in
the level of technology between sectors or regions, differences in the
degree of geographic development, and differences in social customs
and attitudes between an indigenous and imported social system.
Dualism in all its aspects is a concomitant ofthe growth of a money
economy which, as we saw in Chapter 3, may either arise naturally
as a result of specialisation or be imposed from outside by the im-
portation of an alien economic system - typically capitalism. Basic-
ally, therefore, a dual economy is one characterised by a difference
in social customs between the subsistence and exchange sectors of an
economy, a gap in the level of technology between the rural sub-
sistence sector and the industrial monetised sector, and possibly a
gap in the level of per capita income between regions of a country
if the money economy and industrial development are geographically
concentrated. In fact, it is not unusual for geographic, social and
technological dualism to occur together, with each type of dualism
tending to reinforce the other. Also, the more 'progressive' sectors
typically have favourable access to scarce factors of production,
which is a major cause of the persistence of dualism.
If the basic origin of dualism is the introduction of money into a
subsistence barter economy, and development depends on the ex-
tension of the money economy, development must contend with the
existence of dualism in all its aspects. We shall consider here social
and technological dualism, leaving geographic dualism until
later when we consider Myrdal's hypothesis of cumulative causation.
The first question is, what development problems does the exis-
tence of dualism pose for an economy, and how can dualism im-
pede and retard development? As far as social dualism is concerned,
THE PERSISTENCE OF UNDERDEVELOPMENT 119
the obstacles appear to be similar to those presented by a traditional
society with no modern exchange sector at all. The task is one of
providing incentives in the subsistence sector and drawing the
subsistence sector into the money economy. The fact that the
indigenous subsistence sector may be reluctant to alter its traditional
way of life and to respond to incentives is not peculiar to a dual
economy. It is true, therefore, that underdevelopment tends to be
associated with social dualism, but it is perhaps a little misleading to
regard social dualism as an underlying cause of backwardness and
poverty. It would be difficult to argue that development would be
more rapid in the absence of a monetary sector, from which the
existence of dualism stems. Even if the growth of the exchange
sector makes little impact on attitudes in the indigenous sector, it is
bound to make some contribution to development by employing
labour from the subsistence sector and disseminating knowledge.
Without the growth of the money economy it is difficult to en-
visage much progress at all. In short, it seems more realistic to
regard social dualism as an inevitable consequence of development
rather than as a basic cause of underdevelopment.
This is not to say that social dualism does not create problems
of its own. For example, different development strategies will be
required to cope with dissimilar conditions in the two sectors, and
this may involve real resource costs not encountered by the developed
economy. It is in this sense that the dual economy is at a compara-
tive disadvantage.
Similar reservations can be raised over whether it is accurate to
describe technological dualism as a cause of underdevelopment. As
with social dualism, it is probably more realistic to regard it as an
inevitable feature of the development process. Again, though, the
difficulties that may ensue from gaps in technology between the
rural and industrial sectors of the economy must be recognised. Two
disadvantages are commonly associated with technological dualism.
The first we mentioned earlier: where technological dualism
is the result of a foreign enclave a proportion of the profits
generated in the industrial sector will be remitted to the home
country, reducing the level of saving and investment below what it
might have been. The second disadvantage is more fundamental,
but difficult to avoid. If in the rural, or non-monetised, sector of the
economy production processes are characterised by labour-intensive
techniques and variable technical coefficients of production, while
production processes in the industrial, technologically advanced

sector are capital-intensive and possess relatively fixed technical
coefficients, it is possible that the technology of the industrial sector
may impede progress in the agricultural sector on which the rapid
development of the total economy, in part, depends. Firstly, rela-
tively fixed technical coefficients (i.e. a low elasticity of substitution
between factors) means that labour can only be absorbed from agri-
culture into industry as fast as the growth of capital, and secondly,
capital intensity itself will restrict employment opportunities in the
industrial sector, possibly leading to urban unemployment and per-
petuating underdevelopment in the rural sector. Hence, productivity
growth in the agricultural sector, which is recognised as being neces-
sary to establish a secure basis for take-off into sustained growth,
may be slowed down.
It is true that if the technology of the modern sector (imported
or otherwise) does embody fixed technological coefficients, it may be
difficult for an economy to use the socially optimum combination of
factors, but this short-run disadvantage must be weighed against
the favourable repercussions on productivity stemming from the
advanced technology. If capital accumulation and technical pro-
gress, and the development of an industrial sector - in addition to
agricultural development- are essential prerequisites to raising the
level of per capita income, it is difficult to see how technological
dualism can be avoided, at least in the early stages of development.
The best that can be done is firstly to encourage the widespread
application and rapid assimilation of technical progress throughout
all sectors of the economy, and secondly to ensure the 'proper'
pricing offactors of production to prevent in the first place the intro-
duction of a technology which may be profitable to private indi-
viduals but which does not maximise the returns to society at large
because factor prices do not adequately reflect relative factor endow-
ments. But even a technology which is socially optimal in this sense
may not be the technology which provides the soundest basis for
sustained growth in the long run. This raises the big question of
development strategy which is the subject of Chapters 6 and 7.
The Process of Cumulative Causation

The hypothesis of cumulative causation as an explanation of the
lagging growth ofless developed nations is associated with the name
of Gunnar Myrdal.l Basically, it is a hypothesis of geographic
dualism, applicable to nations and regions within nations, which
can be advanced to account for the persistence of spatial differences
in a wide variety of development indices including per capita in-
come, rates of growth of industrialisation and trade, employment
growth rates and levels of unemployment. As such, the process of
cumulative causation is a direct challenge to static equilibrium
theory which predicts that the working of economic forces will cause
spatial differences to narrow. Myrdal contends that in the context
of development both economic and social forces produce tendencies
towards disequilibrium, and that the assumption in economic theory
that disequilibrium situations tend towards equilibrium is false. If
this were not so, how could the tendency for international differences
in living standards to widen be explained? Thus Myrdal replaces
the assumption of stable equilibrium with what he calls the hypo-
thesis of circular cumulative causation, arguing that the use of this
hypothesis can go a long way towards explaining why international
differences in levels of development, and inter-regional differences
in development within nations, may persist and even widen over
time.
He first considers the hypothesis in the context of a geographically
dual economy, describing how, through the medium oflabour migra-
tion, capital movements and trade, the existence of dualism not only
retards the development of the backward regions but can also slow
up the development of the whole economy. To describe the process
of circular cumulative causation, let us start off with a country in
which all regions have attained the same stage of development as
measured by the same level of per capita income, or by similar
levels of productivity and wages in the same occupations. Then
assume that an exogenous or endogenous stimulant produces a
disequilibrium situation with development proceeding more rapidly
in one region than another. The proposition is that economic and
social forces will tend to strengthen the disequilibrium situation by
leading to cumulative expansion in the favoured region at the
expense of other regions, which then become comparatively worse
off, retarding their future development. What Myrdal has in mind
is a type of multiplier-accelerator mechanism producing increasing
returns in the favoured region. Instead of leading to equality, the
1 G. Myrdal, Economic Theory and Underdeveloped Regions (London: Duck-
forces of supply and demand interact with each other to produce

cumulative movements away from spatial equilibrium. Since the
wage level is the basic determinant of per capita income, let us
take the example of wages and wage differences to illustrate the kind
of process that Myrdal is suggesting. Take two regions, A and B
(e.g. north and south Italy), and assume that wages are determined
by supply and demand, as in Figs 5.1 and 5.2.
REGION A
._..::..__ _ _ _ _ _ _ _ _ _ _ Supply and

0
demand for
labour
FIG. 5.1
0
s
:fl w~
~"' Ws REGION B
s;
Supply and
0 demand for
labour
FIG. 5.2
Suppose to start with that wage levels are identical in the two
regions, WA = WB. Then assume that a stimulus of some sort
causes the demand for labour, and therefore wages, to rise in region
A relative to region B; that is, the demand curve for labour in region
A shifts to D1D1, causing wages to rise to OWA 1 • Since labour tends
to respond to differences in economic opportunities of this sort,
the wage discrepancy may be assumed to induce labour migration
from region B to region A. Equilibrium theory then predicts that
there will be a tendency for wage levels to be equalised once more
through a reduction in labour supply in region B from SS to S1S1 and
an increase in labour supply in region A from SS to S1S1, giving a wage
in region A of WAs equal to a wage in region B of WB 1 • According
to the hypothesis of cumulative causation, however, changes in supply
may be expected to repercuss on demand in such a way as to counter-
act the tendency towards equilibrium. Migration from region B
denudes the area of human capital and entrepreneurs, and de-
presses the demand for goods and services and factors of production,
while movements into region A, on the other hand, will tend to
stimulate enterprise and the demand for products, adding to the
demand for factors of production. In short, migration from region B
will cause the demand curve for labour to shift to the left, say to
D1D1, and migration into region A will cause the demand curve for
labour to shift further to the right, say to D2D2, causing the initial
wage discrepancy at least to persist, if not widen (if the shifts in
demand are greater than those assumed). Thus once development
differences appear, there is set in motion a chain of cumulative
expansion in the favoured region which has what Myrdal calls
'backwash' effects on other regions, causing development differences
in general to persist or even diverge.
Capital movements and trade also play a part in the process of
cumulative causation. In a free market, capital, like labour, will
tend to move where the return is highest and this will be to the
region where demand is buoyant. Capital, labour and entre-
preneurship will tend to migrate together. The benefits of trade will
also accrue to the host region. Regions within a nation using a com-
mon currency cannot have balance of payments difficulties, but the
ability to import depends on the ability to export. If production is
subject to increasing returns, the region experiencing the rapid
growth of factor supplies will be able to increase its competitive
advantage over the relatively lagging regions containing smaller-
scale industries, and increase its real income accordingly. In the
same way the general freeing and widening of international markets
and the expansion of world trade will tend to favour the more rapidly
growing regions within nation•states.
The impact of immigration into the expanding region is also
likely to induce improvements in transport and communications,
education and health facilities, etc., improving efficiency and
productivity and widening still further the competitive advantage

of the growing region over the lagging regions experiencing out-
migration of the factors of production.
Such is the potential strength of the 'backwash' effects of the
process of circular cumulative causation, that Professor Hirschman!
has suggested that the lagging regions may possibly be better off if
they were made into sovereign political states. If a lagging area was
an independent' country', the mobility offactors of production could
be more easily controlled, competition between the leading and
lagging regions could be lessened, each region could more easily
concentrate on producing goods in which it possessed the compara-
tive cost advantage, separate exchange rates could be fixed for the
two regions, and resort could be more easily made to protection.
Despite these potential advantages of nationhood for a backward
region, however, Hirschman argues against sovereignty because he
believes that the forces making for the inter-regional transmission
of growth are likely to be more powerful than those making for
'international' transmission. This presupposes, however, that the
forces making for the inter-regional transmission of growth are lost,
and begs the question of whether the differential advantages ofbeing
a region within a nation, as distinct from a separate 'nation', offset
the 'backwash' effects that still remain. Hirschman recognises the
continued existence of' backwash' effects and argues that, to offset
them, a nation which is concerned with developing its backward
regions should provide certain equivalents of sovereignty, such as a
separate tax system and the right to protect certain activities.
Policies must be designed to reduce what he calls the 'polarisation'
effects of inter-regional differences in development and to strengthen
the 'trickling down' effects. The 'trickling down' effects are the
favourable repercussions on backward regions emanating from
expanding regions, which M yrdal calls 'spread' effects. These
'trickling down' or 'spread' effects consist mainly of an increased
demand for the backward areas' products and the diffusion of tech-
nology and knowledge. In Myrdal's view, the 'spread' effects are
weaker than the 'backwash' effects, and if inter-regional differences
are to be narrowed, nations must rely on state intervention. The
alternative is to wait for a natural end to the process of cumulative
causation, which may be a long time coming.
1 A. Hirschman, Strategy of Economic Development (New Haven: Yale
But a time must eventually come when increasing costs in the
expanding region will halt expansion. The higher costs ofliving, and
the external diseconomies produced by congestion, will ultimately
outweigh the benefits of greater efficiency and higher money returns
to the factors of production. The process of migration will then be
halted, and possibly reversed. In some developed countries this
stage is now beginning to be reached. The question for governments
with certain growth and welfare objectives is whether they can
afford to let the process take its natural course, and to tolerate the
inequalities that may arise before the process ends. In practice,
governments in many advanced countries have taken active steps
for many years to redress regional imbalances, and this is one reason
why regional disparities tend to be less in advanced countries
than in less developed countries. In the less developed countries,
however, Myrdal is of the view that, far from lessening regional in-
equalities, the state has been a positive force making for their per-
sistence: 'In many of the poorer countries the natural drift towards
inequalities has been supported and magnified by built-in feudal
and other inegalitarian institutions and power structures which aid
the rich in exploiting the poor.' I
According to calculations by Williamson,2 the average population-
weighted coefficient of variation of regional income for high-income
countries is approximately 0·2, compared with values in excess of
0·5 for some of the poorer countries undergoing rapid development
such as Brazil, Puerto Rico, the Philippines, etc. Whatever the
activities of the state, there are good economic reasons for a pattern
of this sort, given that the biggest geographical income differences
are likely to be urban-rural ones and that development is associated
with the spread of urbanisation and a reduction in the size of the
rural sector. Not surprisingly, very poor countries tend to be uni-
formly poor.
The process of circular cumulative causation is also used by
Myrdal in an attempt to explain widening international differences
in the level of development from similar initial conditions. Through
the media of labour migration, capital movements and trade, inter-
national inequalities are perpetuated in exactly the same way as
1Myrdal, EconrJmic Theory and Underdeveloped Regions (1963 ed.) p. 40.
2 J.
G. Williamson, 'Regional Inequality and the Process of National
Development: A Description of Patterns', Economic Development and Cultural
Change (July 1965).
regional inequalities within nations. Myrdal argues that through

trade the less developed countries have been forced into the produc-
tion of goods, notably primary products, with inelastic demand
with respect to both price and income. This has put the less developed
countries at a grave disadvantage compared with the developed
countries with respect to the balance of payments and the availability
of foreign exchange. Moreover, with the tendency for the efficiency
wage (i.e. the money wage in relation to productivity) to fall in
faster-growing areas relative to other areas, the developed countries
have gained a cumulative competitive trading advantage, especially
in manufactured commodities. Myrdal, of course, is not alone in
this view. Many economists have been arguing the case for some
time for a complete restructuring of the pattern of world trade if
the unfavourable international position of less developed countries
is not to be a permanent obstacle to their growth. We must leave
the basis of these arguments to Chapter 9, however, where the
relation between trade and comparative rates of development will
be examined more fully.
Myrdal argues in the same vein in the case of capital movements.
If it were not for exchange controls and generous incentives for
indigenous and foreign capitalists to earn high profits, the natural
tendency would be for the less developed countries to be net ex-
porters of capital. Risks associated with investment tend to be much
higher in less developed countries, and safety is as much a factor
determining capital movements as rates of return. As it happens,
owing to the favourable tax treatment of profits, market guarantees,
and the large volume of capital from international lending organisa-
tions, the less developed countries are net importers of long-term
capital, but the short-term account is still adverse. Despite capital
controls, it has been estimated that between 1957 and 1965 approxi-
mately $400 million of private short-term funds flowed from less
developed to developed countries annually.!
The potential weakness ofMyrdal's hypothesis at the international
level concerns the effects of labour migration. The international
migration of labour from less developed to developed countries can
have beneficial as well as harmful effects on backward economies.
The greatest deleterious effect on backward economies is the obvious
one of possible loss of human capital, although even here, if the
1 M. Diamond, 'Trends in the Flow of International Private Capital,
1957-1965', I.M.F. Staff Papers (Mar 1967).

human capital is unemployed, migration will not be a disastrous
loss. But it is not only the skilled and educated that may be induced
to leave their native lands. Unskilled labour may also respond to
the existence of better employment opportunities elsewhere. If it is
argued that less developed countries suffer from underemployment,
and that productivity is low owing to 'overpopulation', the out-
migration of unskilled labour could be a substantial benefit to less
developed countries. It is possible, for example, that emigration has
helped to raise per capita income in some of the poorer European
countries such as Greece, Turkey and Spain, and improved the
balance of payments at the same time through remittances by
emigrants to their home countries. In this important respect,
generous immigration policies in advanced countries can provide a
valuable means of development assistance. In the event of un-
restricted emigration from the less developed countries, and un-
restricted immigration into the developed countries, it is difficult
to know where the balance of advantage lies for the less developed
countries. In a world which restricts the immigration of unskilled
labour, however, but permits the immigration of skilled labour,
the less developed countries undoubtedly suffer. While in theory,
therefore, certain types of international labour migration could help
to narrow international differences in levels of development, in
practice the assumptions of such a model are rarely fulfilled.
Even so, any potential gain from unrestricted labour mobility is
unlikely to offset the international 'backwash' effects arising from
trade and international capital movements. Even with unrestricted
migration, therefore, there would still be a tendency for inter-
national differences in the level of development to widen through
trade and the free movement of capital. And the existence of inter-
national 'spread' effects gives no cause for modifYing this conclusion.
International 'spread' effects are fairly weak- certainly weaker than
'spread' effects within nations.
What, then, should be our verdict on the hypothesis of cumulative
causation? Given that the hypothesis assumes free trade and free
mobility of the factors of production, it clearly contains more force
with respect to inter-regional differences in development than inter-
national differences. On the other hand, it cannot be dismissed
lightly in discussing the development gap. In view of the fact that
there has been no tendency in the recent past for international
per capita income levels to converge, the hypothesis is not refuted by
the evidence. In particular, the present international trading and

payments position of less developed countries does not inspire con-
fidence that the total gains from trade between the developed and
less developed countries are distributed equally.
The contribution of the hypothesis of cumulative causation to an
understanding of development and underdevelopment is its emphasis
on development as a cumulative phenomenon and, more important
still, its challenge to static equilibrium theory that regions or nations
which gain an initial advantage may maintain that advantage in
perpetuity to the detriment of development elsewhere. At its root
is the phenomenon of increasing returns defined broadly as the
accumulation of productive advantages of the type discussed m
Chapter 2, relating to how societies progress technologically.
The Population 'Problem'

Population growth plays a conflicting role in the development
process. It can act both as a stimulus and an impediment to growth
and development. The question, to which there is no easy answer,
is, at what point do the economic disadvantages begin to outweigh
the advantages; what is the 'optimum' rate of population growth?
The positive role of population growth in the development process
is, firstly, that population increases add to the supply of factors of
production and at the same time add to the demand for finished
products. Provided labour's marginal product is positive, additional
labour supplies will increase the national product. Secondly, if
production is subject to increasing returns, at least in the industrial
sector, an expansion oflabour supply will be a stimulus to growth by
extending the market and enabling an increase in output per unit
of input. Thirdly, if population growth depresses per capita income
in the short run, and people resist the reduction in living standards,
forces may be released in the economy to exploit the opportunities
for economic growth that hitherto were left unutilised. The greater
the pressure of population on resources and the more living standards
are threatened, the greater will be the stimulus of population to
growth provided the potential for further growth is there to exploit.
There is no doubt that the pressure of population on food supplies
has been a major encouragement to technical progress in the agri-
cultural sector.
On the debit side, however, if there are abundant supplies of
labour, and labour is cheap relative to capital, the capital to labour
ratio may remain low in industry and agriculture, retarding the
rate of growth oflabour productivity. Unfortunately, because of the
interrelation between physical increases in the quantity of factors of
production and their productivity, it is difficult to say what the rate
of growth of a country would be if population or labour supply
grew at a different rate. This is a well-known drawback of the pro-
duction function approach to an analysis of the sources of growth.
The potentially conflicting role of population growth in the
development process is highlighted by the difference in attitude
towards population increase in developed and less developed
countries. In most advanced countries population growth is not
condemned for growth reasons. In some countries it is actively
encouraged by generous family allowance provisions (e.g. France)
and liberal immigration policies (e.g. Canada). In the less developed
countries, however, the overwhelming view seems to be that
population growth and abundant supplies of labour are serious
obstacles to development. The difference in attitude between rich
and poor countries stems basically from the fact that whereas in-
creased supplies of labour in developed countries have abundant
capital to work with, this is not so in the less developed countries.
If the rate of population increase approaches the rate of growth of
national income, there will be no chance for per capita incomes to
rise. If per capita incomes remain depressed, savings will remain
low and, in the absence of savings for capital accumulation, in-
creases in the labour supply will depress productivity on which
rising living standards primarily depend. In short, one advantage
of slower population growth is that a higher proportion of resources
can be invested to raise the future growth of per capita income. As it
is, approximately 60 per cent of investment in less developed countries
is devoted to maintaining per capita income at a constant level.
Thus, while population increase can contribute to an expansion
of markets, permitting specialisation, economies of scale and the
establishment of previously uneconomic activities, it is seen as im-
peding development not only by making it difficult for capital to
be accumulated but also by rendering unattractive the intro-
duction of advanced labour-saving technology. It is broadly for these
reasons that population growth is regarded as a development ob-
stacle.
What may be called the 'paradox of labour' can be illustrated
with reference to the identity 0 = L X OL or ro = TL +

roL, where
0 is output, Lis labour, OL is labour productivity and r stands for
the rate of growth of the variables. If TL and roL are positively cor-
related, the contribution of the growth of labour to the growth of
output will be unambiguously positive. If there is an inverse rela-
tion, however, labour-force growth may depress the rate of growth
of productivity such that output growth is less than it could have
been had the labour force not grown so fast. To answer the question
of whether growth would be faster iflabour-force growth was slower
requires a knowledge of the interrelation between variables in the
aggregate production function. If the rate of labour-force growth
was reduced, for instance, we should have to know how much capital
and 'technical progress' would have to be substituted in order to
prevent the growth of output falling, and whether there was an
automatic relation between the growth of the labour force falling
and a growth of capital and 'technical progress' necessary to com-
pensate. Given the parameters of the system (i.e. the partial elas-
ticities of output with respect to capital and labour), it is easy to
work out the extent to which increases in capital and/or 'total'
productivity would have to compensate for a decrease in the growth
of the labour force in order to maintain the growth of output at a
constant rate. But it is difficult to know in practice whether the
interrelationship between factor inputs is such that a reduction in
labour-force growth would automatically bring about the necessary
increases elsewhere. The presumption of those who argue that
population growth is too rapid in less developed countries must be
that if the growth was reduced, increased productivity growth and
the growth of capital would more than compensate.
Let us take the Cobb-Douglas production function in its esti-
mating form and give the degree of compensation required through
increases in other variables when labour-force growth is reduced.l
As before (see equation (2.6)),
ro = rp + rJ.TK + ~r£. (5.1)
Totally differentiating equation (5.1), and putting dro = 0, gives
dro = drp + r1.drK + ~drL = 0. (5.2)
1 We are assuming throughout the discussion, for the sake of convenience,
a constant relation between the current rate of growth of population and the
work-force, even though in practice the relation is a lagged one and may
not be constant.
THE PERSISTENCE OF UNDERDEVE.LOPMENT 131
Holding 'total' productivity growth constant (drp = 0), the re-
quired change in the rate of growth of capital to compensate for a
reduction in the rate of growth of labour would be
(5.3)
Holding the rate of growth of capital constant (drK = 0), the re-
quired change in the rate of growth of' total' productivity to com-
pensate for a reduction in the rate of growth of labour would be
drp
or -d = -~. (5.4)
T£
Allowing both 'total' productivity and capital to compensate for a

reduction in labour-force growth, we get
drp + a.drK = -~drL, or

drp + a.drK
drL = - ....
f.l
(5.5)
Suppose, for illustration, that we assume that production is
subject to constant returns (a.+ ~ = 1), and that a.= 0·2 and
~ = 0·8. This would imply from equation (5.3) that a drop in the
growth rate of the labour force of one percentage point would have
to be compensated by a rise in the rate of growth of capital of four
percentage points. From equation (5.4), a drop in labour-force
growth of one percentage point would have to be compensated by
a 0·8 percentage point rise in the rate of growth of' total' produc-
tivity. The question is, to what extent have increases in the labour
force been depressing the rate of growth of capital and the rate of
growth of 'total' productivity by these amounts, and would a re-
duction in the rate of growth of the labour force by one percentage
point immediately be compensated for by a rise in 'total' produc-
tivity or capital, or a combination of both, by the required amount?
The evidence we have on the interrelationship between variables in
the growth process is virtually non-existent. We do possess cross-
section data, however, which suggest that there is very little syste-
matic relation at all between inter-country rates of population
growth and rates of growth of per capita income. Taking figures
presented by Colin Clark,l which are produced in Table 5.1 below,
it can be seen that some countries with high population growth rates
1 C. Clark, 'The "Population Explosion" Myth', Bulletin of the Institute of
Development Studies (University of Sussex, May 1969).
TABLE 5.1
Population Growth and Increases in Real Pn(Jduct per Head, 1950-65,
in Asia, Africa and Latin America
(all countries included for which information available)
Population growth: Population growth: Population growth:

under 2% per year 2-2·9% per year 3% per year or over
Increase Increase Increase

in real in real in real
product per product per product per
head head head
(%per year) (%per year) (%per year)
Algeria 1·8 Morocco -0·3 Kenya 1·9

Tunisia 2·2 Egypt 3·1 Rhodesia 0·9
Nigeria 1·2 Ghana 2·4 Costa Rica 0·7
Tanzania 1·2 Sudan 3·0 Dominican
Jamaica 5·1 Uganda 0·8 Republic 2·3
Antilles -1·4 Haiti -0·7 El Salvador 2·2
Argentina 1·2 Panama 3·0 Guatemala 1·7
Bolivia 0·1 Trinidad 5·3 Honduras 0·5
Chile 1·2 Mexico 2·7
Colombia 1·7 Nicaragua 2·7
Paraguay 0·8 Brazil 2·3
Peru 2·8 Ecuador 1·5
Iraq 6·9 Venezuela 2·7
Burma 2·9 Israel 5·8
Ceylon 0·1 Syria 0·7
India 1·5 Cambodia 2·7
Pakistan 1·3 Taiwan 3·9
South Korea 4·2 Philippines 2·1
Thailand 3·0
Average 1-4 Average 2·3 Average 2·2
have progressed very quickly while other countries with low popu-
lation growth rates have stagnated, relatively speaking. Clark goes
as far as to call the so-called population 'problem' a development
'myth'.
This is not to deny, of course, that population growth may present
a problem in other respects- in particular the ability to feed itself.
And presumably no one would dispute a population 'problem' if
population growth actually exceeded income growth. In this case
living standards would fall to subsistence level; the economy would
regress and be forced to live off its capital. This state of affairs is
referred to by development economists as a 'low-level equilibrium
trap' situation. We shall later consider a model formalising this
possibility, but first let us lay out briefly some of the facts about
world population.
Facts about World Population

At the present time the world's population is approximately 3,500
million, of which two-thirds is desperately poor, and over one-
third of which resides in Asia. The current rate of growth of world
population is about 2 per cent per annum, which has no precedent
~istorically. From A.D. 1 to 1750 the rate of growth was no more than
0·05 per cent per annum; from 1750 to 1850, 0·5 per cent per annum;
and even between 1900 and 1950 it was only 0·8 per cent per annum.
The present rate of increase will double world population every
thirty-five years. It can be said with some confidence that by the
year 2000, barring a major catastrophe, the world's population will
be in the region of 7,000 million. Food production in the world
as a whole is growing slightly faster than population, but food
production per head in some parts of the world, e.g. Asia and Africa,
is fairly static.
The rates of population growth in less developed countries in the
recent past have been substantially in excess of the rate of growth for
the world as a whole, and look like continuing so in the foreseeable
future. Over the two decades 1950-70, for example, Latin America's
population grew at approximately 2·8 per cent per annum, and
can be. expected to grow at the same annual average rate at least
until 1980. Africa's predicted population growth rate to 1980 is
2·5 per cent per annum, and Asia's expected rate of increase is 1·9
per cent per annum. These rates compare with predicted rates of
less than 1·0 per cent per annum for Europe and 1·4 per cent for
Russia and North America.
The country with the largest population is China, which in 1970
had an estimated 800 million persons and which is expected to add
150 million to the world's population by 1980. India has the second
largest population with 550 million, and is adding to the world's
population at the rate of 10 million a year.l
1 As a matter of interest, on average two babies are born per second,
which means that in the last minute the world's population has increased by
approxinlately 120 persons!
The rate of growth of population is the difference between the

number oflive births per thousand of the population and the number
of deaths per thousand. In a country where the birth rate is 40 per
1,000 and the death rate is 20 per 1,000, the rate ofpopulation growth
40-20
will therefore be 1 OOO = 20 per thousand or 2 per cent per
annum. If (in normal' circumstances) a birth rate of60 per 1,000 is
considered to be a biological maximum, and a death rate of 10 per
1,000 is considered a 'medical' minimum, the maximum possible
growth rate of population, ignoring immigration, would be about
5 per cent per annum. These birth and death rates are, in fact,
extremes, and are rarely found in practice. Guinea has a birth rate
of60 per 1,000, but this represents exceptional fertility! For the less
developed countries as a whole the average birth rate is about 38
per thousand and the average death rate is now about 15 per 1,000,
giving an average rate of population increase of approximately 2·3
per cent per annum. This rapid rate of population growth, com-
pared with advanced countries (and also in relation to the growth of
national income), can be seen to be the result of relatively high
birth rates coupled with death rates almost as low as in advanced
countries. If population growth is a 'problem' in less developed
countries, this is the simple source of the difficulty and the remedy
is plain.
The vital questions are: can high birth rates be expected to fall
naturally with development, and, if so, what is the crucial level of
development and per capita income at which the adjustment takes
place, and how long does the process take? There is a good deal of
evidence from the history of the present advanced countries that
birth rates do fall as a natural consequence of rising living standards.
Japan provides a dramatic post-war case study where the birth
rate has fallen from 34 per 1,000 in 1947 to 15 per 1,000 at the
present day. It would not be legitimate to generalise from the
experience of Japan, but even this success story would suggest a
minimum of another twenty years before birth rates are brought into
a closer relation with death rates.
Given that there may be a lag between the death rate falling and
a subsequent decline in the birth rate, rapid population growth may
be considered a transitional or more enduring 'problem' for a coun-
try depending on the currently prevailing levels of the rate of births
and deaths. This proposition is best illustrated by means of a simple
diagram (Fig. 5.3). The curves RB and RD represent the time
paths of the birth rate and death rate respectively. Population
growth is determined by the gap between the two curves. To save
drawing more diagrams, let us suppose that points X and T in
Fig. 5.3 represent two countries with the same current rate of popu-
lation growth (Pq = St). In the case of country Y, population growth
will soon slow down since the death rate has reached its minimum
.
J!?
...
c
.,c
~
"C
"C
c
c
m
~ 0
X y
Time
FIG. 5.3
and the birth rate is falling. In the case of country X, however,
which has the same current population growth, the population
growth rate can be expected to accelerate in the future as the gap
between the birth rate and the death rate widens. The death rate is
falling but the birth rate remains constant to the point V; only
after this point will population growth decelerate. Here, then, are
two countries with the same observed population growth in the
present but with radically different future prospects. In comparing
countries, and their population 'problems', the time profile of
countries must be borne in mind. But the crucial question, as we
suggested earlier, is, what is the length of the time lag between
the death rate falling and the downturn of the birth rate? It is the
length of this lag which determines the short-run prospects of
countries emerging from a transitional state and attempting to
'take off' into self-sustaining growth.
The experience of the less developed countries today has no
historical parallel, at least in Western Europe. In nineteenth-
century Europe birth and death rates tended to fall together and a
situation never arose of population growth in excess of 2 per cent
per annum. It could almost be argued that the 'balance of nature'
has been upset in the present less developed countries. The intro-
duction of public health measures and medical attention has
reduced death rates suddenly and dramatically, but the means and
know-how to effect an equally dramatic fall in the birth rate have
not been provided. Modern science and methods of public ad-
ministration have contributed to the ending of premature death,
but have yet to exert a significant impact on births.
The ' Optimum' Population

What is the' optimum' population? The term' optimum' population
is used in several different senses, but four, particularly, are com-
monly employed. Firstly, it is sometimes used to refer to the size of
population which maximises the average product or income per
head. It is in this situation that a society's savings ratio is likely to be
maximised. Thus if the total product curve for an economy is
drawn as in Fig. 5.4-, the optimum population is OP, where a ray
FIG. 5.4
from the origin is tangential to the total product curve. At OP,

total product (OY) divided by population (OP), or average product
per head of population, is at a maximum. The condition for maxi-
mum average product per head is that the marginal product per
head should be equal to the average. If the marginal product of an
addition to the population is above the average, average product
could be increased by an expansion of the population. Contrariwise,
if the marginal product of an addition to the population is below the
average, a further increase in population will reduce the average
product and the population would have exceeded its optimum level
in the way defined. If there was no saving, the maximisation of
product per head would maximise welfare per head because con-
sumption per head would then be at a maximum.
On the surface this concept of optimum population seems art
attractive one on which to base a population policy. It provides the
greatest scope for maximising savings per head if desired or, in the
absence of forced or voluntary saving, it will lead to the maximisa-
tion of welfare per head. The policy implications are frightening,
however. If the marginal product of an addition to the working
population was less than the average in the short run, there may be
a call for an immediate halt to population growth. Secondly, in the
absence of any offsetting benefits, the logic of pursuing the goal of
maximising product per head would require the extermination of
any communities whose contribution to total product was lower than
the average.
1 - - - - - ' l r - - - ' T - - - - - - Welfare subsistence

level
Average product
Marginal product
FIG. 5.5
A more feasible, and less contentious, objective would be the

maximisation of total welfare, in which case the condition for an
optimum population would be when the additional satisfaction
arising from an increase in the population was just offset by the
reduction in the welfare of the existing population arising from the
addition to the population. To illustrate, suppose we argue that
there is such a thing as a minimum welfare subsistence level at
which man can enjoy a full and contented life. If the marginal
product of labour is above this welfare subsistence level, the popu-
lation can be said to be below the optimum, since with additions
to the population no one else need be affected and total welfare can
increase. If the distribution of income of the existing population is
unaltered the optimum population is reached when the marginal
product of labour has reached the welfare subsistence level (as in
Fig. 5.5). OP would be the optimum population.
It can be seen, however, that at population OP some of the popu-

lation will be enjoying welfare in excess of the welfare subsistence
level; that is, the average product exceeds the marginal product. If
the distribution of income is not assumed unalterable, and some of
those enjoying welfare in excess of the welfare subsistence level are
taxed, a larger population than OP could be supported at the wel-
fare subsistence level. On the crucial assumption that the marginal utiliry
of income declines as income rises, so that the rich receive (lose) less
marginal satisfaction from any marginal change in income than the
poor, total welfare could be increased by raising the population and
redistributing income simultaneously. The population would then
be at an optimum when all incomes are equalised at the welfare
subsistence level.l
The notion of a pure subsistence level of output can also be
introduced, giving a third concept of optimum population, i.e. that
level of population beyond which the average product in an economy
falls below the level of production necessary for subsistence on the
assumption that the total product is equally shared. In this case, the
term 'optimum' simply refers to the maximum population that can
be supported with existing resources, and is the point of Malthusian
equilibrium. In Fig. 5.6 a population beyond OP1 could not be
supported because the average product of the population would be
below the level of subsistence.
Of course, if total product was not equally shared, a total popu-
lation of OP1 would not be supportable, for some would have more
income than necessary for subsistence and others less. But note that
if the product is equally shared, a much larger population can be
maintained than the population at which the marginal product
falls below subsistence, i.e. OP. In fact, the optimum population,
OP1, is consistent with a negative marginal product.
This last point leads us to a fourth sense in which the term 'op-
timum' population is sometimes used, which is to describe a state of
affairs where a country's population is so large that increases in it
1 Some may object to this concept of 'optimum' because it makes inter-
personal comparisons of utility which are considered to be normative
judgements. Pareto optimality in welfare economics excludes the possibility
of income redistribution to increase welfare by defining 'optimum' as a state
of affairs in which it is impossible to increase the utility of one person without
decreasing the utility of another. Pareto optimality itself, however, cannot
escape normative criticism; but in any case to say that the marginal utility of
income declines as income rises is not in itself a normative statement.
can only be detrimental to growth- not only detrimental to a coun-
try's long-run growth prospects but also to growth in the short run,
implying zero or negative marginal product. The population is
optimal in this sense when total product is maximised, at OP2 in
Fig. 5.6. This definition of optimum population is closely linked with
the notion of population density and attempts to define under-
population and overpopulation in terms of the relation between
-
g
l
g
-~
0
E
Total product
1-------,~~--\--+!o,:,...----- Subsistence lev...
product
FIG. 5.6
population and resources, and, in particular, land. Since resources,
such as land, vary considerably in quality, however, inter-country
comparisons of ratios of population to resources are virtually mean-
ingless. A country may be regarded as 'underpopulated' in relation
to another even though it has a higher population-resource ratio,
simply because its technology to exploit resources is superior. Tech-
nology will influence the position and shape of the total product
curve, and hence the optimum population, for any given ratio of
population to resources. In view of the variety of interpretations of
the concept of optimum population, the claim that country X, Y
or Z is 'overpopulated' or 'underpopulated' needs to be viewed
with some scepticism unless a precise definition of terms is given.
A Model of the Low-level Equilibrium Trap

To repeat, there are two main interrelated reasons why rapid
population growth may be regarded as a retarding influence on
development. Firstly, rapid population growth may not permit a
rise in per capita incomes sufficient to provide savings necessary for
the required amount of capital formation for growth. Secondly, if

population growth outstrips the capacity of industry to absorb
new labour, either urban unemployment will develop or rural under-
employment will be exacerbated, depressing productivity in the
agricultural sector. It is not inconceivable, moreover, that rises in
per capita income in the early stages of development may be accom-
panied by, or even induce, population growth in excess of income
growth, holding down per capita income to a subsistence level.
Today we witness falling death rates (associated with development)
eroding gains in living standards; and presumably for centuries
past the population of most countries has been oscillating around the
subsistence level, with small gains in living standards (due to 'tech-
nical progress') being wiped out either by higher birth rates or such
factors as disease, famine and war. One vital question that a theory
of economic development must answer is, at what level of per capita
income can income growth be expected to exceed population growth
permanently, thus sustaining further rises in per capita income in-
definitely? This question cannot be answered at a point in time by
casual observation that income growth exceeds population growth,
because the very rise in per capita income may induce population
growth greater than income growth in the future. A full behavioural
model is required, and a theory of population which is able to explain
the pattern of population growth during the course of development.
What happens to population as per capita income rises, and what
happens to national income, and therefore per capita income, as
population rises ?
Models of the low-level equilibrium trap attempt to integrate
population and development theory by recognising the inter-
dependence between population growth, per capita income, and
national income growth. This type of model, which originated in
the 1950s, is designed firstly to demonstrate the difficulties that less
developed countries may face in achieving a self-sustaining rise in
living standards, and secondly to provide pointers to policy action.
One such model by Nelson,l on which we shall concentrate, con-
tains three basic equations dealing with the determination of net
capital formation, population growth and income growth. Let us
consider these equations separately before examining the simple
workings of the model.
1 R. Nelson, 'A Theory of the Low Level Equilibrium Trap in Under-
developed Economies', American Economic Review (Dec 1956).
Capital formation. Capital formation takes place through saving
and new land brought into cultivation, i.e. dk = dk' + dR, where
k is capital, k' is savings-created capital and R is land. We shall
ignore new land as a part of capital and concentrate on savings-
created capital. It is assumed that all saving is invested.
.
Th e rate of savmgs . (dk')
per capita p IS. reIate d to mcome
. per
capita (~) in the way shown in Fig. 5.7. In equation form:
p = b
dk' p - X J, where p0 > (0)'
[(0) p (5.6)
= -C, where p0 (0)'

~p .
Per capita income is so low up to the level (~)' that there is dis-
investment (dissaving), assumed to be at a constant rate, C. Beyond
(~)' the rate of savings per capita rises linearly with per capita
Income.
dk'
p
FIG. 5.7
Population growth. With nsmg per capita income, population
growth ( ~) is first assumed to accelerate owing to falling death
rates. Then at a critical level of per capita income(~)" population

growth reaches a maximum p (dP*) where the death rate has fallen
to a minimum. Since Nelson's model is short-run, the effect of per
capita income on the birth rate is ignored, but the relation can
easily be accommodated if desired. The relation between popula-

tion growth and per capita income can be represented as in Fig. 5.8.
In equation form:
~= P[ (~) - SJ. where (~) < (~)" (5.7)
=~*,where(~)~ (~r
S is the subsistence level of per capita income. At per capita income
levels below S, ~ is negative since the death rate would exceed the
birth rate.
dP
p
FIG. 5.8
Growth of national income. Income (or output) is assumed to be a

linear homogeneous function of factor inputs, i.e. 0 = Tj(K, L),
where K is capital, L is labour (as a constant proportion of the
population) and Tis an index of 'total' productivity. Technology
and the social structure are assumed constant.
From the assumptions of the model, the income growth curve can
be drawn as in Fig. 5.9. Where S =X (taken from Figs 5.7 and 5.8),
population is stationary; the rate of savings-created capital per head
is zero, and therefore income will be stationary ( dg = 0). With
rising per capita income beyond this stationary equilibrium, growth
accelerates owing to increases in the labour force and capital per
head. As population growth reaches a maximum, however, and
savings as a percentage of national income approach a constant,
income growth will level off. In the absence of technical progress,
growth will ultimately decelerate owing to the law of variable pro-
portions, in this case a full in the capital-labour ratio.
Combining Figs 5.8 and 5.9, we have a diagrammatic representa-
tion of the possibility of a low-level equilibrium trap situation in
dO
0
FIG. 5.9
which per capita income is permanently depressed. Fig. 5.10 shows

this. Any level of per capita income between the subsistence point
(S = X) and Oa will be accompanied by a growth of population
faster than the growth of income, forcing per capita income down.
dO dP
o'?
FIG. 5.10
The equilibrium level of per capita income will be where the popu-
lation growth curve cuts the income curve from below. One such
point is to the left of Oa where S = X, and this point represents the
low-level equilibrium trap. Any level of per capita income below
Oa will force per capita income down to this subsistence level. Con-
trariwise, any per capita income level beyond Oa will mean a sus-
tained rise in per capita income until the two curves cross again at
Oq. This would be a new stable equilibrium with the population
growth curve again cutting the income growth curve from below.
To escape from the low-level equilibrium trap, per capita income
must either be raised to Oa, or the dg and f; curves must be shifted
favourably. The origin of 'big push' theories of development (see
Chapter 6 and below), and the concept of a 'critical minimum
effort', was the belief that to escape from the 'trap' it would be
necessary to raise per capita income to Oa in one go. If countries are
in a trap situation, however, much greater hope probably lies in the
dO curve d n·11·
tmg upward s over time,
. . to teeh mea
owmg . l progress,
0
or in a sudden drop in the f; curve from a reduction in the birth rate.
Capital from abroad, raising the dg curve, and emigration, lowering
dP
the p curve, could also free an economy from such a trap.
To take account of factors other than population growth which
may depress per capita income, and factors other than increases
in capital per head that may raise per capita income, the low-level
equilibrium trap model can be extended and generalised by adopting
Leibenstein'sl terminology of income-depressing forces and income-
raising forces. Leibenstein's approach is illustrated in Fig. 5.11.
The curve representing income-depressing forces, Zt, is measured
horizontally from the 45° line, and the curve representing income-
raising forces, Xt, is measured vertically from the 45° line. Per capita
income level Oa is the only point of stable equilibrium. Between Oa
and Oq income-depressing forces are greater than income-raising
forces and per capita income will slip back to Oa. Only beyond Oq
are income-raising forces greater than income-depressing forces
such that a sustained increase in per capita income becomes pos-
sible. Oq is the critical per capita income level necessary to escape
the low-level equilibrium trap.
1 H. Leibenstein, Economic Backwardness and Economic Growth (New York:
Wiley, 1957).
X,
IC....---------------- Per capita

income, and
induced income
growth
FIG. 5.11
The Critical Minimum Effort Thesis

If a low-level equilibrium trap exists, it has been argued that a
critical minimum effort will be required to escape from it. This is
the so-called critical minimum effort thesis which refers to the effort
needed, normally measured by investment requirements, to raise
per capita income to that level beyond which the further growth
of per capita income will not be associated with income-depressing
forces exceeding income-raising forces. But as soon as one begins to
calculate the amount of extra investment required to raise per capita
income in one go, by even a modest amount, it becomes all too
obvious that a critical minimum effort, or a 'big push' as it is some-
times called, is not feasible. But, in any case, it is doubtful whether
a critical minimum effort or 'big push' is necessary in practice.
While it is important to recognise the interdependence between
population growth, per capita income and income growth, models
of the low-level equilibrium trap tend to be unduly pessimistic
and restrictive in their assumptions. Furthermore, the prediction
of a low-level trap in the absence of a critical minimum effort
does not wholly accord with historical experience. Hagen has
drawn attention to the fact that in Western Europe income-induced
population growth did not prevent substantial rises in per capita
income from taking place.! On the contrary, Hagen claims that it
1 E. Hagen, 'Population and Economic Growth', American Economic
Review (June 1959).
was not until population started to grow rapidly that per capita
income started to rise, and that population growth preceded income
growth rather than the other way round. We hinted earlier that
one of the most likely escapes from a Malthusian situation is a per-
manent fall in the birth rate due to a standard of living effect. It is
to this effect that Hagen attributes the prevention of a Malthusian
situation in nineteenth-century Europe. Birth rates followed death
rates downwards long before the maximum possible population
growth rate had been reached.
What happened in Europe has not been the experience of the
present less developed countries over the last three decades, but
there are other potent forces at work which offer a similar escape
from Malthusianism. The most significant of these forces are tech-
nical progress and 'irreversible' additions to the capital stock (es-
pecially in the form of social and human capital), the consequence of
which is to shift the income growth curve upwards over time so that
the level of per capita income representing stable equilibrium is
continually rising over time. Technical progress was an important
source of economic progress in nineteenth-century Europe; it is,
potentially, an even greater force in the present developing nations,
given their ready access to modern technology from countries already
developed.
The way in which rising living standards may be regarded as a
natural phenomenon, obviating the need for a critical minimum
effort, is shown diagrammatically in Fig. 5.12. Consider a small rise
in the per capita income level from OX (the subsistence level) to
OX1. The traditional argument is that this will be accompanied by
population growth in excess of income growth, causing per capita
income to fall back to the subsistence level. If time is not abstracted
from the analysis, however, it is possible to argue that the increase
in per capita income from ox to oxl will be accompanied by per-
manent changes in the quality of the capital stock and skills, etc.,
such that per capita income will not fall back to OX but to some
higher level, say OX2. Income growth in the range of per capita
income ox to ox2 is now permanently higher, represented by the
curve dg'. If ox2 becomes the new stable equilibrium, and the
sequence of events is repeated, per capita income will reach the
level Oa in a series of steps. No critical minimum effort will be needed.
With continuous productivity growth due to all forms of technical
progress, a 'ratchet' mechanism of the type described is quite
feasible, and it is probably by this mechanism that, in practice,
countries typically 'take off'.
Most developing countries in the world today are experiencing
dO,JlE.
7) p
f1Q
0
FIG. 5.12
income growth faster than the rate of growth of population. Whether

income growth would be faster if population growth was reduced
is an open question. It is possible to conceive of a low-level equi-
librium trap, but its level almost certainly rises over time owing
largely to technical progress before a reduction in birth rates sets in.
6
The Allocation of Resources
The Broad Policy Choices

Given the scarcity of resources in less developed countries in relation
to development needs, one of the central issues in development
economics is the allocation of resources among competing ends. For
most less developed countries the two major constraints on the growth
of output are the ability to invest and import, and most theories of
resource allocation and most investment criteria reflect this fact.
A common starting-point in the consideration of resource allocation
is how to maximise the level or growth of output from the investible
resources available, and how to minimise the use offoreign exchange.
Apart from the decision of how much to invest, three broad types
of allocation decision may be distinguished: firstly, there is the
question of which sectors to invest in; secondly, there is the question
of which projects should receive priority given the factor endowment
of a country and its development goals; and thirdly, there is the
question of the combination offactors that should be used to produce
a given vector of goods and services which will determine the tech-
nology of production. While these decisions may look independent,
in fact they are not. In practice, interdependence between decisions
on priorities and decisions on technology is inevitable. The decision
on which goods to produce will, to a certain extent, dictate factor
proportions if technical coefficients are relatively fixed, and decisions
about technology are bound to influence the types of goods and
services that are produced in so far as factor proportions cannot be
varied. Some goods and services are obviously more labour-
intensive than others. The choice of technology, in turn, will be
particularly influenced by the way in which resources are valued,
and the relative valuation given to present versus future consump-
tion and welfare.
THE ALLOCATION OF RESOURCES 149
Investment decisions of the micro-type outlined above will also
be influenced to a certain degree by the nature of the development
strategy intended; that is, by the broader policy issues such as
whether emphasis is to be given to agriculture or industry, whether
resources are to be used to build up complementary activities or
whether imbalances are to be deliberately created in order to induce
investment and decision-making, and whether emphasis is to be on
static short-term efficiency in the allocation of resources or whether
emphasis is to be on laying the foundations for faster growth in the
future. And in an open economy, the potential clash between
efficiency and growth also requires a consideration of the impli-
cations of adherence to different versions of the comparative cost
doctrine. In short, the question of resource allocation between
projects cannot be divorced from consideration of the wider policy
issues of industry versus agriculture, balanced versus unbalanced
growth, foreign trade strategy, etc. And influencing all these de-
cisions will be the underlying objectives of the development strategy:
whether the aim is to maximise output in the present regardless
of the balance between investment and consumption goods, or
whether the objective is to maximise growth and output at some
future point in time.
The choice of development strategy itself will be subject to
political, social and economic constraints. A particular strategy, for
example, may conflict with the desired income distribution or other
social objectives. Other strategies may involve political reper-
cussions inimical to development. One factor that cannot be ignored
is the regional distribution of political power. Spatial considerations
of this sort add a further dimension to the allocation problem. The
pursuit of balanced growth or massive investment in social overhead
capital may imply a large public sector in the economy which may
not be politically tolerable. Certain development plans may an-
tagonise foreign investors or multilateral aid-giving agencies such
that if the plans are carried out foreign capital or 'agency' capital
dries up. Bearing in mind these constraints, it may be helpful to
consider first some of the broader aspects of development strategy,
and to discuss briefly development goals, before examining a number
of specific criteria that have been recommended for application to
the investment decision.
Industry versus Agriculture

The issue of the choice between industry and agriculture, and where
the emphasis should lie, can be dismissed very quickly. It is in-
creasingly recognisedthat the industry versus agriculture debate is a
sterile one, which diverts attention from more important matters.
In practice the fortunes of agriculture and industry are closely inter-
woven in that the expansion of industry depends to a large extent
on improvements in agricultural productivity, and improvements
in agricultural productivity depend on adequate supplies of in-
dustrial 'inputs', especially the provision of consumer goods to
act as incentives to peasant farmers to increase the agricultural
surplus. It is worth mentioning, however, that the emphasis on
balance between industry and agriculture is of fairly recent origin.
On the one hand it represents a shift of emphasis away from the
'modern' view of an all-out drive for industrialisation by less
developed countries, and at the same time it represents a reaction
against the traditional doctrine of comparative cost advantage which
when applied to less developed countries almost certainly dictates
the production of primary commodities, and a pattern of trade which
puts these countries at a relative development disadvantage.
The Comparative Cost Doctrine

Whether the comparative cost doctrine should be adhered to is
itself a question of development strategy which is intimately bound
up with the goals of less developed countries (i.e. what they are
trying to maximise), and with the controversy over whether de-
velopment should be looked at more from the point of view of the
balance of payments than from that of the allocation of real re-
sources. Assuming the full employment of resources, and that the
price of a commodity reflects its opportunity cost (admittedly bold
assumptions in any country), adherence to the comparative cost
doctrine will produce the optimum pattern of production and trade
for a country. Efficiency will be maximised when no commodity
is produced which could be imported at a lower cost measured by
the resources that would have to be sacrificed to produce it at home.
In a free-trade world this would almost certainly rule out the pro-
duction of a wide range of industrial commodities in less developed
countries. If the objective is growth, however, as opposed to static
efficiency, the theory of growth suggests investment criteria that are
quite contradictory to those derived from the theory of comparative
advantage. If growth depends on increases in per capita investment,
for example, it may be unwise to channel resources into activities
which are labour-intensive, where the income generated is all
consumed and none is saved, or where there is no scope for in-
creasing returns. Similarly, if growth depends on a healthy balance
of payments, it may be equally misplaced to develop activities pro-
ducing goods with a low price and income elasticity of demand in
world markets. A low price elasticity of demand can cause fluctua-
tions in export earnings with shifts in supply and cause the terms of
trade to move adversely. A low income elasticity of demand will
mean that for any given growth of world income countries producing
these commodities will be put at a permanent balance of payments
disadvantage compared with other countries producing goods with
a higher income elasticity of demand.
The question ultimately boils down to one of the relative valuation
of present versus future output and consumption (or welfare)- be-
tween consumption today and consumption tomorrow. Efficiency in
resource allocation will maximise present output and consumption
from a given amount of resources, but may impair future growth.
Striving for growth will lower present consumption but will provide
greater output in the future.
Labour-intensive versus Capital-intensive Technology

The valuation of present versus future welfare is also the central
issue regarding the choice of technology- whether techniques should
be capital- or labour-intensive. On the surface it would seem
sensible, in a labour-abundant economy, to use labour-intensive
methods of production, and to encourage activities that use factors
which are in abundance. Indeed, more recent versionsl of the
comparative cost doctrine maintain that countries will benefit by
producing and exporting commodities which use more of their
relatively abundant factors, and importing commodities using more
of their relatively scarce factors of production.2 Again, however,
1 E.g. the Heckscher-Ohlin version of the comparative cost doctrine.
2 It should be stressed, however, that although labour is abundant it is
not necessarily 'cheap'. For one thing it may be extremely inefficient, and
for another imperfections in the market may cause its money price to be
there is a potential conflict between efficiency and growth: a clash

between the maximisation of present output and consumption and
growth and output in the future. The more labour-intensive the
technology, the less the likelihood of an income distribution favour-
able to saving and a higher level of investment per man in the future.
Total
output
FIG. 6.1
If the propensity to save ofworkers is lower than that of the owners

of capital, the surplus per unit of capital left for reinvestment will
be smaller than if the technology was more capital-intensive. On the
other hand, the more capital-intensive the technology, the lower the
level of employment and consumption in the present. Consider Fig.
6.1 above in which output is measured on the vertical axis above the
origin, labour on the horizontal axis, and a fixed amount of capital
(OK) on the vertical axis below the origin. lflabour is abundant, and
the wage is given in the industrial sector, the wage bill will rise
linearly with the amount of labour employed, given by 0 W. (And
consumption will rise linearly also if all wages are consumed.)
With the fixed amount of capital, OK, output is maximised with the
employment of labour, 0 T, but the surplus of production over
consumption ( = HQ) is very small. The investible surplus is
high. Ideally, a measure of' efficiency wages' is required. The Heckscher-

Ohlin theory assumes that the relative efficiencies oflabour and capital are
the same in each country.
maximised with the employment oflabour, OR, where a tangent to
the total product curve is parallel at S to the wage line 0 W. The
more labour-intensive techniques maximise output in the short run;
but the more capital-intensive techniques provide a greater surplus
for reinvestment for future growth.
In general, we reach the conclusion that the higher the valuation
placed on raising the present level of employment and consumption
as compared to future growth, the more labour-intensive techniques
should be favoured. On the other hand, the greater the valuation
placed on future growth in relation to present welfare, the more
capital-intensive methods of production should be favoured. It is
capital-intensive techniques that are capable of yielding the largest
surplus of income over wage costs for a given capital outlay, making
possible a higher rate of reinvestment for the future. The choice
between projects of different degrees of capital intensity, therefore-
in so far as there is a choice- boils down to the relative weights to be
attached to an additional increment of investment compared to an
additional increment of consumption. If saving is regarded as
'suboptimal', then at the margin a unit of investment must be
regarded as more valuable than a unit of consumption and there is
a case for capital-intensive techniques which will generate this
extra investment potential at the margin. As we said earlier, in
Chapter 3, the fact that the social cost, or opportunity cost, oflabour
is zero makes no difference to the argument because even 'costless'
labour will consume if employed in industry, and the object is to
minimise consumption. This has important implications for the
valuation of labour in resource allocation, which we take up in
Appendix 7.3.
Balanced versus Unbalanced Growth

Another broad choice of development strategy is between so-called
balanced and unbalanced growth. Whether or not there exists a
low-level equilibrium trap, which needs a 'critical minimum effort'
to overcome it, sound economic reasons can be advanced in support
of a 'big push', taking the form of a planned large-scale expansion
of a wide range of economic activities. The economic rationale for a
'big push' forms part of what has come to be known as the doctrine
of balanced growth. Opposed to this thesis, however, is a school of
thought which argues that a 'big push' is not feasible, and that in
any case development is best stimulated in less developed countries

by the deliberate creation of imbalance. In this respect balanced
versus unbalanced growth is an issue of development strategy which
may constrain the application of investment criteria. But before we
evaluate the balanced-unbalanced growth controversy, we must
first outline in more detail precisely what is meant by the terms
'balanced' and' unbalanced', and examine the potential advantages
and disadvantages of the alternative strategies - assuming for
the moment that they are alternatives, and not complements, to
one another.
The term 'balanced growth' is used in many different senses, but
the original exponents of the balanced growth doctrine had in
mind the scale of investment necessary to overcome indivisibilities
on both the supply and demand side of the development process.l
Indivisibilities on the supply side refer to the 'lumpiness' of capital
(especially social overhead capital), and the fact that only invest-
ment in a large number of activities simultaneously can take ad-
vantage of various external economies of scale. lndivisibilities
on the demand side refer to the limitations imposed by the size of
the market on the profitability, and hence feasibility, of economic
activities. This was the original interpretation of the doctrine of
balanced growth: the large-scale expansion of activities to overcome
divergences between private and social benefit.
The doctrine has since been extended, however, to refer to the path
of economic development, and the pattern of investment, necessary to
keep the different sectors of the economy in balance so that lack of
development in one sector does not impede development in others.
This does not mean, of course, that output in all sectors should grow
at the same rate, but according to the income elasticity of demand
for products so that supply equals demand. The notion of equili-
brium is implied, and an absence of shortages and bottlenecks. A
definition of balanced growth embracing the several different in-
terpretations and emphases that have been placed on the term has
been attempted by Paul Streeten: 'Wherever several non-infinitesi-
mal investment decisions (or decisions generally) depend for their
success on each other, simultaneous investment in a series of in-
dustries (or firms or plants), in conformity with the pattern of
consumers' demand and of different industries' demand for each
1 E.g. P. Rosenstein-Rodan, 'Problems of Industrialisation of East and
South-East Europe', Economic Journal (June-Sep 1943).
other's products, is required.'l Balanced growth, therefore, has a
horizontal and vertical aspect. On the one hand it recognises in-
divisibilities in supply and complementarities of demand, and on the
other it stresses the importance of achieving balance between such
sectors as agriculture and industry, between capital-goods and
consumer-goods industries, and between social capital and directly
productive activities.
Thus there are two fairly distinct versions of the balanced growth
doctrine that need considering: one referring to the path of develop-
ment and the pattern of investment necessary for the smooth
functioning of the economy, and the other referring to the scale of
investment necessary to overcome indivisibilities in the productive
process on both sides of the market. Nurkse's2 views on balanced
growth tend to embrace both versions of the theory of balanced
growth, while Rosenstein-Rodan3 concentrates on the necessity for
a 'big push' to overcome the existence of indivisibilities. We shall
first consider the economic arguments for a large-scale investment
programme and then discuss the desirability of achieving balance
between different sectors of the economy.
On the demand side, the argument amounts to little more than
Adam Smith's famous dictum that specialisation, or the division
of labour, is limited by the extent of the market, and that if the
market is limited certain activities may not be economically viable.
If, however, several activities are established simultaneously, each
could provide a market for each other's products, so that activities
that are not profitable considered in isolation would become profit-
able considered in the context of a large-scale development pro-
gramme. The onus would presumably be on the government to
organise such a programme, and planning would supersede the
market system for the programme to be initiated. It is true that
industrial enterprises may have to be of a certain minimum size to
be profitable, if only to compete in international markets, but the
argument begs certain questions and leaves others unanswered. For
example, there may be latent demand for certain commodities
which would obviate the need to establish other activities to generate
the income and purchasing power necessary for those commodities
1P. Streeten, 'Unbalanced GrQwth', Oxford &onomic Papers (June 1959).
2R. Nurkse, Problems of Capital Formation in Underdeveloped Countries
(Oxford University Press, 1953).
a Op. cit.
to be bought. 1 Moreover, there are several alternative ways in

which markets can be widened. The improvement of transport
facilities in less developed countries, and of communications in
general, offers tremendous scope for the expansion of markets; and
the doctrine of balanced growth for demand reasons must also lose
much of its force in the context of an open economy. Restrictions on
imported commodities could expand overnight the market for home-
produced goods, and export promotion can also be resorted to unless
commodities are very price and income inelastic in world markets.
No one would deny that there are development difficulties connected
with a limited home market for commodities, but a 'big push' is not
the only way, and not necessarily the most expeditious or economi-
cal, in which these difficulties can be overcome.
On the supply side, the argument for a' big push' is bound up with
the assumed existence of external economies of scale. The external
economies referred to in this context go beyond the external econo-
mies of the traditional theory of the firm. Whereas in traditional
equilibrium theory external economies refer to the fact that the
nature of the production function in one activity may be altered
by the existence of other activities (e.g. in close proximity), in the
context of development theory they refer mainly to the impact of a
large investment programme on the profit functions of participating
firms. If there exist external economies in either sense, the social
cost of an activity will be less than the private cost. It is argued
that the way to eliminate this divergence is to make each activity
part of an overall programme of investment expansion. Enterprises
which are not, or do not appear to be, profitable in isolation become
profitable when considered as part of a comprehensive plan for
industrial expansion embracing several activities. As several com-
mentators2 have pointed out, however, while the expansion of
several activities may improve the profitability of each activity -
1 It is for this reason that the example of the shoe factory has always
seemed to me to be singularly inappropriate for illustrating the problem
posed by a limited market; the proposition being that a shoe factory would
never be viable unless it was part of a larger investmsnt package since shoe
workers will want to spend their income on commodities other than shoes.
Unless an economy is totally devoid of other activities, there should be
enough people working elsewhere to justify at least one isolated shoe
factory!
2 E.g. M. Fleming, 'External Economies and the Doctrine of Balanced
Growth', Economic Journal (June 1955).

assuming there are inter-industry linkages - the cost offactor inputs
may rise, offsetting the benefit of these external economies. Under
the impact of a 'big push' the price of capital goods and the cost of
capital may rise substantially, and so too may the wage rate. If fac-
tors of production are not in elastic supply, the case for a 'big push'
to eliminate discrepancies between the private and social costs of
an activity is weakened - unless rising supply prices of factors are
anticipated and included in the calculations as an additional cost.
It would then be a question of weighing the benefits of the
external economies against the increased cost of factor inputs. One
advantage of gradual expansion as opposed to a 'big push' is that
the supply offactors would have some time to adjust to the increased
demand.
It was in the light of these demand and supply considerations,
however, that Rosenstein-Rodan in his pioneer article recommended
a massive investment programme for the industrialisation of East
and South-east Europe extending over a ten-year period. His idea
was for an Eastern European Industrial Trust to be established to
instigate a 'big push' sufficient to absorb ten million workers re-
garded as surplus in the agricultural sector. Calculating on the basis
that each worker would require between £300 and £350 of capital,
he estimated that the investment ratio would have to be raised to
18 per cent of national product for the ten-year period, or double its
prevailing level. Rosenstein-Rodan was not optimistic that this
could be achieved, and it is true to say that in the case of the present
less developed countries there are few today who believe that a
'big push' is feasible, even if desirable. The prevailing view would
seem to be that if less developed countries do possess the resources
for a 'big push', they ought hardly to be described as less developed
in the first place.
The fact remains, however, that certain investments must be of a
minimum size to be economically worth while. It is very uneconomi-
cal to build roads, railways and power stations merely to meet
current demand. Ideally, social overhead capital of this type must be
planned and built on a large scale to achieve long-run economy in
the use of resources. But this argument hardly qualifies for the special
label 'balanced growth'. The case for a 'big push' for supply
reasons seems to be quite weak in the absence of resources to attempt
such a strategy and without detailed knowledge of the precise magni-
tude of the net economies that are likely to accrue.
We turn very briefly, therefore, to the second version of the

balanced growth doctrine which lays emphasis on the path of
development and the pattern of investment. This version of the
balanced growth doctrine stresses the necessity of balance between
sectors of the economy to prevent bottlenecks developing in some
sectors, which may be a hindrance to development, and excess
capacity in others which would be wasteful. Among the foremost
proponents of this version of the balanced growth doctrine are
Nurksel and Professor Arthur Lewis.2 Particular emphasis is
placed by these two economists on achieving a balance between the
agricultural and industrial sectors of less developed economies.
This is for two main reasons, both of which recognise the inter-
dependence between the two sectors and the mutual assistance and
stimulus that each can give the other. The first is that if agricultural
productivity is to improve, there must be incentives for farmers to
expand their marketable surplus, and this requires a balance
between the agricultural and consumer-goods sectors of the economy.
The second is that agriculture requires capital inputs and this
requires a balance between agriculture and the production of
capital goods and the provision of social overhead capital. This is
in addition to the fact that agricultural output can provide a basis
for the development of local industries and that the industrial
sector relies on the agricultural sector for food. Furthermore, in
the absence of increasing exports the agricultural sector must rely
on the industrial sector for a substantial proportion of the increased
demand for its products. The doctrine of balanced growth in this
form, especially the stress on the balance between agriculture and
industry, steers a middle course between the traditional comparative
cost doctrine that less developed countries should confine themselves
to the production of primary commodities, and the modern view,
which gained momentum in the 1950s, that these countries ought to
embark on an all-out drive for industrialisation.
Unbalanced Growth
The major overriding general criticism of the balanced growth
doctrine is that it fails to come to grips with the fundamental ob-
stacle to development in less developed countries, namely a shortage
1 Nurkse, Problems of Capital Formation.
1 A. Lewis, The Theory ofEconomic Growth (London: Allen & Unwin, 1955).
of resources of all kinds. Critics of balanced growth do not deny the
importance of a large-scale investment programme and the ex-
pansion of complementary activities. Their argument is simply that
in the absence of sufficient resources, especially capital, entre-
preneurs and decision-makers, the striving for balanced growth may
not provide a sufficient stimulus to the spontaneous mobilisation of
resources or the inducement to invest, and will certainly not econo-
mise on decision-taking if planning is required.
One of the most provocative books on development strategy in
recent years is by Professor Hirschman,! whose argument is along
these lines. Hirschman is probably the foremost exponent of the
doctrine of unbalanced growth and we must briefly consider his
views. The question he attempts to answer is this: given a limited
amount of investment resources, and a series of proposed investment
projects whose total cost exceeds the available resources, how do we
pick out the projects that will make the greatest contribution to
development relative to their cost? And how should 'contribution'
be measured? First of all, he distinguishes two types of investment
choices - substitution choices and postponement choices. Substi-
tution choices are those which involve a decision as to whether
project A or B should be undertaken. Postponement choices are
those which involve a decision as to the sequence of projects A and
B, i.e. which should precede the other. Hirschman is mainly con-
cerned with postponement choices and how they are made. His
fundamental thesis is that the question of priority must be resolved
on the basis of a comparative appraisal of the strength with which
progress in one area will induce progress in another. The efficient
sequence of projects will necessarily vary from region to region and
from country to country depending on the nature of the obstacles to
development, but the basic case for the approach remains the same;
that is, to economise on decision-making. In Hirschman's view, the
real scarcity in less developed countries is not the resources them-
selves but the means and ability to bring them into play. Preference
should be given to that sequence of projects which maximises 'in-
duced' decision-making.
He illustrates his argument by considering the relation between
social capital (S.C.) and directly productive activities (D.P.A.).
The case in which S.C. precedes D.P.A. he calls 'development via
1 A. Hirschman, Strategy of Economic Development (New Haven: Yale

160 GROWTH Al'i"D DEVELOPMENT
excess capacity', and the case in which D.P.A. precedes S.C. he

labels 'development via shortages'. Both sequences create induce-
ments and pressures conducive to development; but which sequence
should be adopted if it is not possible to pursue a 'balanced'
growth path, to produce D.P.A. output at minimum cost in terms of
inputs into both D.P.A. and S.C.? The question can be made clearer
with the aid of a diagram (Fig. 6.2).
-: :J
a.
!:i
0 X
<i
a:
-
c::i
0
t;
0
(.)
FIG. 6.2
If the total cost ofD.P.A. output is measured on the vertical axis,
and the availability and cost of S.C. is measured on the horizontal
axis, curves can be drawn (a, b, c) showing the cost of producing a
given full-capacity output ofD.P.A., from a given amount of invest-
ment in D.P.A., as a function of the availability of S.C. Successive
curves, a, b, c, represent different levels of D.P.A. output from
successively higher investment in D.P.A. The curves are negatively
sloped and convex to the origin because D.P.A. costs will decrease
the greater the availability of social capital, but there is a minimum
amount of S.C. necessary for any level of D.P.A. output (e.g. OS1
corresponding to curve a), and as S.C. increases its impact on the
costs of D.P.A. output becomes less and less.
Now assume that the objective of the economy is to obtain
increasing outputs of D.P.A. with the minimum use of resources
devoted to both D.P.A. and S.C. On each curve, a, b, c, the point
where the sum of the co-ordinates is smallest will represent the most
desirable combination ofD.P.A. and S.C. on this criterion. The line
OX connects the optimal points on the different curves and this
represents the most 'efficient' expansion path, or 'balanced' growth
path, between S.C. and D.P.A.
But suppose that 'optimal' amounts of S.C. and D.P.A. cannot
be expanded simultaneously to keep in balance with one another.
On what criteria is the postponement choice made? One possibility
is the sequence AA1BB2C where the initial expansionary step is
always taken by social capital. This is the sequence that is called
'development via excess capacity'. The other (opposite) possibility
is the sequence AB1BC1C where the initial expansionary step is
taken by D.P.A. This is the sequence that is called 'development via
shortages'. According to Hirschman, the preference should go to
that sequence of expansion that maximises 'induced' decision-
making. It is difficult to tell a priori which sequence this is likely to
be. If S.C. is expanded, existing D.P.A. becomes less costly, en-
couraging further D.P.A. If D.P.A. are expanded first, costs will
rise but pressures will arise for S.C. facilities to be provided. Both
sequences set up incentives and pressures and ultimately, in Hirsch-
man's view, the sequence chosen must depend on the relative
strength of entrepreneurial motivations on the one hand and on the
response to public pressure of the authorities responsible for social
capital on the other.
In general, however, Hirschman has some harsh things to say
about the traditional view that S.C. must precede, or even be kept
in balance with, D.P.A. if development is to progress smoothly.
While he admits that a certain minimum of social capital is a pre-
requisite. to the establishment of D.P.A., he argues that develop-
ment via excess capacity is purely permissive, and that to strive for
balance is equally dangerous because there will be no incentive to
induced investment (or induced decision-making). On the other
hand, development via shortages will compel further investment, and
hence the most 'efficient' sequence as far as 'induced' decision-
making is concerned is likely to be that where D.P.A. precedes
S.C.
It is true that where there is strong social and economic resistance
to change, 'permissive' acts, such as the construction of social over-
head capital, are not likely to provide much impetus to develop-
ment. Italy has recently discovered this in its attempt to develop the
Mezzogiorno, and policy has been changed as a result. On the other
hand, Hirschman's analysis leaves several questions unanswered.
He concedes that the objective must be to obtain increasing out-
puts of directly productive activities at minimum cost in terms of
resources devoted to both D.P.A. and S.C., and that the cost of
producing any given output from D.P.A. will be higher the more
inadequate is S.C., but what is the minimum amount of S.C. required
in a less developed economy? Is this minimum so high as to contra-

dict the argument that D.P.A. should precede S.C., at least in the
earliest stages of development? Furthermore, what is the guarantee
that S.C. will subsequently be provided once D.P.A. have been
established? Indivisibilities with respect to social capital may be
so large that private investors are not induced to supply at any
price. Reliance would then be on the government. Hirschman
neglects to pay adequate attention to the government's crucial
responsibility in this connection. It looks suspiciously as if un-
balanced growth via shortages, like the 'big push', may not be
economically feasible (albeit for different reasons).
Hirschman would apply the same criterion of' induced decision-
making' to the choice and sequence of projects within the directly
productive sector. Here inducements stem from interdependences
between activities, or what Hirschman calls backward and forward
linkage effects. Backward linkages measure the proportion of an
activity's output that represents purchases from other domestic
activities. Forward linkages measure the proportion of an activity's
output that does not go to meet final demand but is used as inputs
into other activities. With knowledge of inter-industry flows in an
economy, from an input-output table (see Appendix 7.1), it should
be possible to rank activities according to the magnitude of their
combined linkage effects. Hirschman is suggesting that within the
directly productive sector a useful development strategy would be
to encourage those activities with the potentially highest combined
linkages, because this will provide the greatest inducement and in-
centive to other activities to develop.
Unfortunately, one of the typical characteristics ofless developed
countries is a lack of interdependence between activities. Primary
product production has very limited backward linkages with other
activities, and forward linkages, although potentially greater, also
tend to be small in practice. Agriculture's demands on other
sectors are minimal, and only a comparatively small fraction of total
agricultural output in less developed countries is processed domesti-
cally; most is exported. The fact that manufacturing activities possess
greater backward and forward linkages, strengthening the cumula-
tive nature of development, is another powerful reason for industrial-
isation. Hirschman advocates the expansion of industry through the
transformation of semi-manufactures into goods required for final
demand, or what he calls 'enclave import' industries. In general he
lays great stress on the role of imports in the development process,
seeing imports as part of the· inducement mechanism. For not only
can semi-manufactured imports be processed into goods for final
demand but final-demand imports themselves can then be readily
produced at home once the market has attained a certain size (or
production threshold). If one of the major obstacles to development
is a shortage of decision-makers, coupled with uncertainty and a
limited market, the existence of imports provides conclusive proof
that the market is there. As imports increase, so too do the chances
that domestic production will one day be profitable. Hirschman
criticises less developed countries for restricting imports prematurely,
and claims that infant industry protection should not be given prior
to the establishment of industries but only after imports have
reached such a level as to guarantee domestic producers a market
for their goods.
Rather like the operation of the price mechanism, Hirschman
seems to be arguing for development through a process of'successive
adaptation', with the system itself pointing to the shortages and
bottlenecks and overcoming them accordingly. In fact, one of the
major appeals of unbalanced growth is that it does not imply the
abandonment of the market mechanism which is regarded as so
sacred by some economists, whatever the problem at stake and
whatever the context. Like the market mechanism, unbalanced
growth also decentralises decision-making (as well as econoxnising its
use), which may be an advantage in some less developed countries
which lack an honest and competent administration. While un-
balanced growth econoxnises on decision-making, however, decisions
must still be taken and the need for managerial ability and adminis-
trative expertise is in no way diminished. A further advantage of
unbalanced growth, as opposed to balance between sectors, is that
there may also be a certain degree of economy in the use of physical
resources. If capital is scarce, and indivisibilities do exist, it is
probably more economical to produce a few products in optimum-
size plants rather than a whole range of products in suboptimal
plants. There may be a trade-off, in other words, between balance
and technical efficiency.
The inherent danger of unbalanced growth is that it leaves too
much to chance. With the unbalanced growth model one has the
uncoxnfortable feeling that the pace of development rests solely on
the whims of entrepreneurs and whether individuals are induced to
invest or not. There is little discussion of how to overcome dis-

crepancies between the private and social profitability of develop-
ment projects, and the possibility of waste and excess capacity
through actions being abortive must also be faced. Supposing im-
balances are deliberately created, or arise naturally, but supply is
totally inelastic, what then? Is the simultaneous existence of excess
capacity in some sectors and shortages in others allowed to persist,
or is the onus on the public authorities to produce some semblance
of co-ordination? The role of shortages in promoting development
is probably overrated. At least a distinction needs to be made
between types of shortage: those that can be filled by response from
the market mechanism and those which cannot, and those which
induce expansion of capacity and those which do not. Unbalanced
growth may further lead to the concentration of production on one
or two commodities, with possible harmful effects on the balance of
payments if the goods are price inelastic and income inelastic in
demand. Lastly, imbalances can be a powerful source of inflationary
pressure, with cost and price increases emanating from the bottle-
neck sectors spreading throughout the economy. Methods of con-
trolling this type of inflation, other than by restoring balance, have
severe limitations: firstly, if price controls are imposed, the supply
situation may be made worse in the shortage sectors by discouraging
production; secondly, general demand deflation will be wasteful
by creating unemployed resources in the non-bottleneck sectors; and
thirdly, currency depreciation which may be necessary to maintain
a balance of payments is itself inflationary by raising the domestic
price of imports. There is evidence that in some South American
countries one of the major sources of rising prices is sector inflation
coupled with exchange depreciation. We shall have more to say
about sector inflation in Chapter 8.
There can be no end to the debate as to whether balanced or
unbalanced growth is the greater stimulus to development. For one
thing, the theories cannot easily be tested empirically, and for an-
other the strategies are politically contentious. From the purely
economic point of view, however, there is really no reason at all
why balanced and unbalanced growth should be presented as
alternatives in the first place. Several attempts have been made,
in fact, to present a reconciliatory view, notably by Mathur.! One
1 A. Mathur, 'Balanced v. Unbalanced Growth: A Reconciliatory View',
Oxford Economic Papers (July 1966).

step towards this reconciliatory approach, which both Hirschman
and Nurkse mention themselves, is to treat unbalanced growth as a
means of achieving the ultimate objective of balanced growth. A
further step is to look for compatibilities in the two approaches.
Since balanced growth is consistent with sectors growing at different
rates, provided supply and demand are in balance, the distinctive
feature of the alternative strategies must be the imbalance between
supply and demand. Unbalanced growth would concentrate
resources in a few selected areas, creating shortages elsewhere. But
this is quite consistent with the 'big push' version of the balanced
growth doctrine provided the scale of investment is sufficient to
overcome indivisibilities and complementarities. If unbalanced
growth is defined not so much in terms of shortages as in terms of
concentration on certain activities, according to comparative ad-
vantage or the existence of increasing returns, balanced and un-
balanced growth can be complementary strategies. There is no
reason why development strategy should not draw on the strong
points of both doctrines so that an optimum strategy of develop-
ment combines some elements of balance as well as imbalance. The
gain of real resources from concentration and specialisation must be
weighed against the potential disadvantages, and perhaps a saving
of resources through maintaining balance between sectors. His-
torically, growth has been unbalanced, and this is implicit in classical
trade theory, with balance being restored through imports. By and
large, the present less developed countries are following the same
route and pursuing a combination of strategies. Resources tend to be
concentrated in a limited number of sectors, but within these sectors
the investment programme attempts to maintain vertical balance.
Concentration of effort is feasible in an open economy with the
capacity to buy imports, and at the same time investment is on a
large enough scale for complementary activities to be established
recognising interdependences in production and consumption.
It is the association of balanced growth with planning which has
tended to polarise opinion on the two strategies; but unbalanced
growth need not preclude planning. With a mixed strategy of
concentration on particular sectors but balance within them, a deci-
sion has first to be made on which sectors to concentrate and then
resources allocated accordingly.
If unbalanced growth precludes planning- indeed, if that is its
appeal - the issue is largely one of the distribution of gains from
development. With balanced growth there is the opportunity to

distribute the benefits of development more evenly throughout
society, to remedy inequity, to control inflation and to avoid the
unemployment of resources. With unplanned unbalanced growth
there is no guarantee against unemployment, inflation and a nega-
tively skewed income distribution. Left to itself, unbalanced growth
is essentially a doctine of laissez-jaire, with no safeguards against the
socially divisive repercussions of change. At the purely economic
level the two doctrines can be complementary rather than alternative
strategies. But if unbalanced growth excludes planning and the co-
ordination of decision-making, the implications of the two strategies
may be very different. In this case the debate becomes a discussion
of the relative merits and demerits of planning which we take up in
the next chapter.
Investment Criteria
Traditional micro-theory teaches that under perfect competition
resources will be optimally allocated when each factor of production
is employed up to the point where its marginal product is equal to its
price, and that society's output (welfare) will be maximised when the
marginal products offactors are equated in all their uses. This is the
so-called 'marginal rule' for resource allocation, and implies
'efficiency' in the sense that a society's total output of goods and
services could not be increased by any redistribution of resources
between activities because each factor of production is equally
productive in existing activities. In static analysis, therefore, 'effi-
ciency' in resource allocation implies maximising the national
product, and this is achieved when the marginal products of factors
are equated in their different uses.
If the application of the marginal rule leads to an efficient alloca-
tion of resources, what is the allocation 'problem' in less developed
countries? Why seek for other criteria to decide on the allocation
of resources? One good reason is that the assumptions of traditional
micro-theory accord neither with the realities nor with the aspirations
ofless developed countries. Two major drawbacks of the application
of the marginal rule may be cited. One is that the marginal rule is a
static criterion, and as we have said before it is by no means certain
that the aim of less developed countries is, or ought to be, the
maximisation of the present level of output, consumption or welfare.
Secondly, traditional static theory ignores a host of factors which
may have a bearing on the social optimum allocation of resources. In
countries characterised by fundamental structural disequilibria and
extreme imperfections in the market, it cannot be assumed that the
market prices of goods and factors of production reflect the social
costs and benefits of production. The application of the marginal
rule will only lead to a socially optimal allocation of resources in the
absence of divergences between market prices and social costs
and benefits, or if market prices are corrected to reflect social
values.
Several factors may lead to divergences between market prices
and the social valuation of goods and factors of production. If
external economies and increasing returns are attached to some
projects their social value will exceed their private value, and the
application of the marginal rule must make allowances for this if
output is to be maximised from a given endowment of factors.
Secondly, if perfect competition does not prevail in the product
market, product prices will not reflect society's valuation of those
products, and market prices must somehow be adjusted to achieve
a social optimum. Similarly, if perfect competition does not prevail
in the factor market, the price of factors will not reflect their oppor-
tunity cost to society so that employing factors up to the point where
their marginal product equals price will not produce a social opti-
mum. Idle resources, such as labour, may be overvalued, and scarce
resources, such as capital and foreign exchange, may be undervalued,
and market prices must therefore be corrected to reflect the value
of these resources to society. Thirdly, static analysis ignores the
future structure of product and factor prices arising from the choice
of projects in the present. An optimum resource allocation in the
present may produce a non-optimal allocation in the future and
vice versa. The only way of coping with this difficulty is through
what is called the programming approach to resource allocation,
by which the repercussions of one activity on others is explicitly
considered and by making due allowance for time. Lastly, the appli-
cation of the marginal rule can only lead to optimal resource
allocation on the assumption that the income distribution is
'optimal', and remains unaffected by whatever programme is
decided on. If a new pattern ofresource allocation alters the income
distribution, output may be maximised but welfare diminished
because of' undesirable' changes in the distribution of income gains.
To say anything concrete on this score requires an explicit statement

of societal objectives if interpersonal comparisons of utility are to be
avoided. Presumably there might be a fair degree of consensus that
an income distribution which leaves half the population unemployed
and starving is 'inferior' to one that does not. Only the conditions
for Pareto optimality would deny it!
For all the above reasons, there has been a prolonged debate in
recent years over the most appropriate criterion for resource allo-
cation in the light of the development obstacles of less developed
countries and their aspirations. The different criteria that have been
suggested reflect, by and large, differences of opinion as to what less
developed countries ought to attempt to maximise, the broad choice
being between present and future levels of output and consumption.
Most of the criteria discussed by early writers in this field refer to the
allocation of capital, reflecting the view of capital as the primary
scarce resource. Increasingly, however, attention is being paid to the
effects of resource allocation decisions on the balance of payments,
in recognition of foreign exchange as an equally scarce resource.
Let us examine some of the more important criteria that have been
suggested and comment on their strengths and weaknesses.
The Minimum Capital-Output Ratio Criterion

With the long tradition, from Adam Smith on, of emphasising
capital's strategic role in the growth process, it is not surprising that
early formulations of investment criteria concentrated on the need
to minimise the use of capital per unit of output; hence the so-called
minimum capital-output ratio criterion (or highest rate of turnover
criterion as it is sometimes called). Under this criterion the object
. . .... I d .. W h
ofht e a ll ocatwn exercise Is to mimmise oO an maximise T, w ere
I IS
oO . th e mcrementa
. 1 capita
. l-output ratio oO . t h e prod uctlvity
. an d TIS . .
of capital. Appealing as this criterion may seem, its use is open to a

number of objections and suffers from all the deficiencies of static
criteria just discussed. Apart from the very real difficulty of measuring
capital-output ratios, the criterion ignores the factor of time and
the relation between the annual output of an investment and the
life of an investment as determined by its rate of depreciation. A
project with a high annual capital-output ratio may produce a
total output over its lifetime far greater than the same amount of
investment in a project with a lower capital-output ratio but with a
shorter life. Moreover, it is difficult to override this weakness by
choosing the project with the lowest ratio of capital to lifetime output
because the present value of future output is a function of the time
pattern of that output. The total output of a project lasting thirty
years may be much greater than that from a project with only half
the life, but the present value of the future output of the project
with the shorter life may still be higher. But even if project lives
were, or were assumed to be, roughly equal, differences in the annual
capital-output ratio may still not provide a rational basis for policy
action. Some projects may have a higher productivity of capital
than others (and a lower capital-output ratio) owing to the greater
efficiency of co-operating factors, but there may be no guarantee
that these co-operating factors will be in elastic supply in the future.
Thus projects with the lowest capital-output ratios in the present
may turn out to have much higher capital-output ratios in the
future owing to rigidities in the supply of other factors required to
work with capital. Similarly, if technology is changing very rapidly,
projects with the lowest capital-output ratios in the short run may
not have the lowest capital-output ratios in the long run. Where
capital is indivhible, the use of this criterion may also imply capital-
intensive technology which means that scarce capital may have to
be concentrated on a narrow range of activities, conflicting with the
desire for balance in the economy and the diversification of exports.
Capital-intensive technology may also produce undesirable econo-
mic and social repercussions in the form of more urban and rural
unemployment. But perhaps the most serious objection to the use of
the minimum capital-output ratio criterion, at least in its basic
form and especially in developing countries, is that it is not explicitly
concerned with the social returns or benefits of a particular pattern
of resource allocation. As we have already argued, in a society where
market prices do not reflect the social costs and benefits of using
resources, there will be divergences between the social and private
returns from a particular pattern of resource allocation, and
the greater the imperfections in the market and the greater the
supplementary benefits of projects, the wider the divergences are
likely to be. Opinion has shifted, therefore, from the view that
resources ought to be allocated simply to minimise the use of capital
per unit of output to the more complex, ambitious aim of maximising
social product; hence the social marginal product criterion for the
allocation of investment resources.
The Social Product Criterion

The social marginal productivity of an investment may be defined as
the return to the private investor plus the net contribution of the
investment to the national product, which may be positive or
negative depending on whether the extra costs of the project to
society are greater or less than the supplementary benefits. Social
marginal productivity (S.M.P.) is thus the difference between the
annual value of output to society (V) (which will be greater than
the value of output to private individuals if the project yields
external benefits) and social costs (C) (which may be greater or less
than private costs depending largely on whether factor prices
overvalue or undervalue the use of resources from society's point
of view), expressed as a ratio of the capital invested (K), i.e.
S.M.P. = V K c. The most likely cases in which the S.M.P. will
exceed the private product are where external economies exist or
where the opportunity cost of a particular factor of production is
less than private investors must pay for it. It is often argued, for
example, that labour's opportunity cost to society is zero in less
developed countries because of disguised unemployment, yet
employers must pay a positive wage. In this case the private cost of
labour is positive but its cost to society is zero. Whereas private
costs refer to market prices, social costs refer to opportunity costs;
that is, worth in alternative uses.
If it is assumed that the social cost of labour is zero, the social
marginal product criterion becomes equivalent, in fact, to the
minimum capital-output ratio criterion. This is easily demonstrated.
Let 8 1= 8(V -;;, C). Ignoring the valuation of V, the capital-
8
. IS
output ratio · mmimise 80 maximise
· · · d , or T . . d , wh en C IS
. zero. There
are, of course, production costs other than labour costs, but these
would have to be incurred whether labour was employed or not.
On the extreme assumption that the social cost of labour is zero,
therefore, the application of the capital-output ratio criterion and
the S.M.P. criterion amount to the same thing. The general con-
elusion is also reached that if the opportunity cost of labour is zero
and labour can do the job of capital equally well, certain investments
may be highly profitable to capitalists but very unprofitable to the
community - the opportunity cost of capital being very high.
In reality, the social cost of employing labour is not likely to be
zero. As we have already seen, it is only legitimate to regard the
marginal product of labour as zero in surplus-labour economies
under special assumptions concerning the organisation of agriculture
and the length of the working day, and whether workers on pro-
jects from which others are drawn away work harder. But even if
the opportunity cost in terms of lost production was zero, there are
bound to be other social costs involved in the case oflabour migrating
from the agricultural to the industrial sector. Social capital in the
towns will have to be provided, and the migrants may impose
various social costs on the community by adding to congestion.
Also, if consumption increases in the process of transference, the
size of the investible surplus will fall and this can be regarded as a
social cost if the objective is to accumulate capital to maximise
output, consumption and welfare in the future.
But supposing for the sake of argument that C is zero, the equiv-
alence of the capital-output criterion and the S.M.P. criterion would
still require that V, the annual value of output to society, was the
same as the private product, implying no supplementary benefits
attached to projects. This state of affairs is hard to envisage in an
underdeveloped country. Investment in the infrastructure and social
capital of a backward economy can be expected to have wide-
ranging effects on the profitability of other activities. In countries
lacking infrastructure and social capital, the gap between the private
return on these types of projects and the social return is likely to be
enormous. These sorts of divergence remain the classic justification
for state provision of transport and communications, health and
educational facilities in any society. In less developed countries
there is likely to be a wide range of projects where the social return
far exceeds the private return, and if these projects are not to be
neglected in an investment programme the social product criterion
must be applied. The difficulty is measuring social return over the
life of the investments. Apart from the fact that many investments
of the social capital type do not yield marketable outputs, it is
virtually impossible in many cases to estimate the impact that one
project has on the functioning and efficiency of others. Apart from
social capital and infrastructure investment, supplementary benefits

are most likely to arise in the case of projects which possess a high
degree of interdependence, but again the difficulty arises of assigning
money values to the externalities. If there is extreme interdepen-
dence between projects it is probably not wise to calculate the social
product of each individual project anyway, or to plan on a project-
by-project basis, but to programme an 'optimal' sequence of
projects. This is one of the objects of the programming ap-
proach to economic development which is discussed in Appendix
7.2.
Assuming the S.M.P. of projects can be calculated reasonably
accurately, adjusting V for supplementary benefits, and factor prices
for opportunity costs, the application of the S.M.P. criterion re-
quires that the S.M.P. of investments be equated within the con-
straint of the total capital available for investment purposes. Like
the capital-output ratio criterion, and the application of the
marginal rule in general, however, the S.M.P. criterion can be
criticised on the grounds that it ignores dynamic considerations
and assumes that the goal of less developed countries should
be to maximise current social welfare with little regard to the
future.
The opposite view is that less developed countries ought to strive
for growth and maximisation of future welfare. Galenson and
Leibensteinl were among the first to argue this way and to examine
the implications for resource allocation. Within a growth frame-
work the maximand becomes the accumulation of capital over time,
and the Galenson-Leibenstein criterion for resource allocation is the
maximisation of what they call the marginal per capita reinvestment
quotient. Let us consider this view, which is the antithesis of micro-
static analysis, in more detail.
The Marginal Per Capita Reinvestment Quotient Criterion

The basis of the Galenson-Leibenstein argument is that resources
ought to be allocated to maximise output and consumption in the
future. This requires maximising the level of per capita output of the
employed population in the present so that the maximum amount of
1 W. Galenson and H. Leibenstein, 'Investment Criteria, Produc-
tivity and Economic Development', Quarterly Journal of Economics (Aug

1955).
savings is available for future reinvestment. This is nicely illustrated
in our earlier Fig. 6.1 (p. 152 above) if S was the point at which
average product is maximised. The criterion also corresponds with
the first concept of optimum population discussed in Chapter 5 if
labour and the population are regarded as synonymous. Resources
should be allocated, therefore, so that the marginal per capita
reinvestment quotient of capital is equalised between projects. The
Galenson-Leibenstein approach is designed to take account of the
influence of the choice of projects on the rate of capital accumu-
lation; or, to put it another way, to take account of the choice of
technology on the distribution of income between savers and spen..
ders. The economic basis of the approach is that the level of per
capita output of the employed population (labour productivity) is
largely determined by the capital to labour ratio, and that the
amount of capital per worker, in turn, must largely depend on the
level of profits (or surplus over wages) stemming from the original
investment. This view is very much in line with Lewis's argument
that the savings and investment ratio in less developed countries is
dependent on the size of the capitalist surplus. Without question,
the implication of the application of the marginal per capita re-
investment quotient criterion would be capital-intensive projects
which minimised the use of labour and the payment of wages,
leaving the maximum possible surplus per head of employment. The
present level of consumption and employment would be sacrificed
for the sake of a high investment ratio and growth rate to produce
a greater volume of goods and services than otherwise in the
future.
The rate of reinvestment per unit of capital invested in the present
is given by
p-ew (6.1)
r=--k-
where p is the net output per machine

e is the number of workers per machine
w is the real wage rate
and k is the cost per machine.
On the assumption that all profits are reinvested, and all wages are
consumed, the reinvestment quotient is equal to the rate of profit.
r is also the growth rate because on the assumption that all wages are
consumed and all profits invested, the reinvestment formula is equal

to the growth formula ~' where S is the savings ratio and C is the
incremental capital-output ratio, i.e.
r
=p-ew_
k -
(P)(l
k
_pew) -c
_ §_ (6.2)
where S =p-ew
p
and
The current growth rate is maximised but at the expense of em-

ployment and the production of goods for consumption. Hence the
marginal per capita reinvestment quotient criterion goes to the
other extreme from the static marginal criteria by neglecting present
consumption and welfare, concentrating on the building up of the
means of production for a greater volume of consumption in the
future.
Galenson and Leibenstein's analysis is useful in stressing the
relationship between technology and the distribution of income and
the rate of saving, but their precise conclusions rest on the assumption
that saving and investment is solely a function of the distribution of
income and nothing else. But suppose that saving depends on the
total level of income as well (which will not be maximised iflabour
as the abundant factor is not employed up to the point where its mar-
ginal product is equal to its opportunity cost). If so, it may not be
wise simply to aim at maximising output per head of the employed
population. Output per head of the total population must also be
considered. There are a number of reasons why total income may
be important, not the least of which is that the less the amount of
employment offered, the greater the level of unemployment and the
greater the probability of consumption in excess of production and
dissaving. In surplus-labour economies the maximisation of total
output (and total output per head) requires combining capital and
labour until both their marginal products are virtually zero, which
means spreading as little capital as possible over as much labour as
possible (capital widening). If total saving is a function of total
income, as well as its distribution, a compromise must be reached
between maximising output per head of the employed population,
using capital-intensive techniques, and maximising output per head
of the total population using less capital-intensive techniques.
Moreover, should not some allowance be made for the fact that
society is bound to value present consumption more than future
consumption by some margin, however small, if only because if
present consumption is less than future consumption its marginal
utility is greater? If so, an investment criterion is required which
reflects both the immediate contribution of projects to consumer
welfare and the future welfare made possible by additions to the
capital stock from saving generated by the current investment. A
partial solution to the problem of the relative valuation of present
and future consumption, and hence the choice of projects, is to
fix a time horizon, acceptable to society, within which future
welfare gains must offset welfare losses in the present.l If this time
horizon exceeds the actual length of time it takes for welfare gains
to match welfare losses, relatively capital-intensive techniques may
be favoured with some sacrifice of present consumption. If, however,
the period over which future welfare gains just compensate for
welfare losses in the present exceeds the acceptable time horizon,
relatively labour-intensive projects should be favoured. Needless to
say, this approach is very arbitrary depending on the time horizon
chosen. If, however, some democratic consensus can be reached, this
would appear to be a better solution to the inevitable clash between
present and future consumption than a straight choice between one
or other of the investment criteria discussed above which in their
basic form ignore the trade-off between present and future welfare.
The problem is one of intergenerational equity which is usually
solved by a fairly arbitrary valuation of investment now, relative
to future consumption, discounted by the rate of time preference
or consumption rate of interest (see Appendix 7.3).
The Labour Absorption Criterion

The investment criterion most directly in conflict with the view that
capital-intensive projects should be chosen is the labour absorption
criterion, which goes even further than ordinary static criteria in its
stress on present employment. It has been argued that if labour is
plentiful and capital scarce, projects should naturally be chosen
1 This is suggested by Professor A. K. Sen in Choice of Techniques, 3rd ed.
(Oxford: Basil Blackwell, 1968).
which absorb the maximum amount of labour per unit of capital,

irrespective of differences in the output generated by projects. The
use of this criterion would maximise the labour to capital ratio
but could easily conflict with the S.M.P. criterion, and any other
criterion aimed at maximising production, since output is a secon-
dary consideration to the amount of labour employed. Imagine two
projects X and r with production functions X', r', as shown in Fig.
6.3. Both projects use the same amount of capital. Project X yields
-
::>
0.
'5
0
0
;§ y'
8:--- Labour employed

o-c;;;..------A~-----=
FIG. 6.3
the highest output for any amount of labour employed, but can
employ only OA labour up to the point where its marginal product
becomes negative. Project r yields a lower output but uses more
labour, OB. Using the employment absorption criterion, project
r should be chosen in preference to project X. Unless the mini-
misation of unemployment at all costs is the primary goal of
development policy, there can be little justification for the use of this
criterion. The more rational policy would be to choose the project
with the higher output per head and compensate any resulting
unemployment through the fiscal process.
Balance of Payments Considerations

So far in discussing investment criteria we have implicitly assumed
a closed economy and have paid no attention to the balance of
payments effects of resource allocation. Increasing attention is now
being paid to foreign exchange as a scarce resource in growing
recognition of the part that imports can play in the development
process, and the fact that over the last decade the balance of pay-
ments of less developed countries has deteriorated considerably in
their effort to grow. If foreign exchange is treated as a scarce
resource it is necessary to incorporate into investment criteria the
balance of payments effects of allocation decisions. To illustrate how
this may be done, let us take the S.M.P. criterion, which most
development economists seem to accept in practice as a basis for
resource allocation.
V-C
S.M.P. = - k - + TwB (6.3)
where B is the total effect of a project on the balance of payments

and w is the premium to be attached to foreign exchange to
reflect its true opportunity cost.
The value of w must be decided on so that foreign exchange and

domestic capital are exhausted simultaneously. If foreign exchange
is exhausted before domestic capital, w must be up-valued. When
both capital and foreign exchange are exhausted simultaneously,
w will fully reflect the relative scarcity of the two scarce factors.
w can take on a value between zero and unity - zero when foreign
exchange is plentiful and unity when foreign exchange is very scarce.
An alternative approach to the above is to add the opportunity
cost of imports to the social cost of a project, that is the calculation
of C, valuing imports at their export equivalent. This would involve
calculating how much more exports would have to be shipped
abroad to meet additional imports, then adding this to the total
cost. For example, if £500 worth of additional imports means a
sacrifice of £600 worth of output for export (measured at home
prices), the cost of imports should be regarded as £600 (i.e. their
opportunity cost), not £500.
If foreign exchange is the dominant scarcity, which it may well
be in the take-offstage of development, and the balance of payments
is a constraint on growth, there may be a strong case for diverting
resources to export- or import-substitute activities when foreign
exchange considerations are incorporated into investment criteria.
Few would deny the potential importance of exports for growth, or
dispute the historical instances of export-led growth, but perhaps it
is timely here to express certain reservations concerning import-
competing activities. Experience suggests that import-substitution
projects rarely save much foreign exchange because the import
substitutes themselves require imports. When the foreign exchange
used by the import substitutes is taken into account it is quite

possible that these projects, although well intentioned, will yield
rates of return lower than other alternatives. In this case it would be
more productive to borrow money from abroad to finance the
balance of payments deficit, provided the alternative projects
yielded a rate of return in excess of the rate of interest on foreign
borrowing. This policy will give a higher level of output than con-
tinuing with import-substitute activities.
On the other hand this policy will give rise to the need for debt
servicing and the commitment offuture foreign exchange. The debt-
servicing problems of foreign borrowing are considered in Chapter 8.
All that we need to stress here is that even though foreign exchange
may be scarce, import-substitute activities may not be the best way
of overcoming balance of payments difficulties and foreign exchange
shortages. It may be better to borrow from abroad, and to pay back
the loan with interest at a future date out of savings generated by
the projects financed from foreign borrowing.
The Social Welfare Function

To take account of the multiplicity of development objectives, and
of the scarcity of resources other than capital, the concept of the
social welfare function has come into vogue. Economists (normally
outside the less developed countries) talk of the less developed coun-
tries maximising a social welfare function where all objectives dis-
counted to the present are translated into a common unit of account,
and all resources are valued at their social (opportunity) cost. The
maximisation of the social welfare function requires the maximisation
of the present value of the future stream of factor services valued
at their opportunity cost. The appeal of the social welfare function is
that every goal and aspiration can be included in it, with prices
adjusted to reflect social values and with discounting to equate
present and future returns. Whether it is feasible to draw up a
social welfare function, however, and calculate it, is another matter.
As far as application is concerned, it is a concept very much in its
embryonic stage. In conception it is similar to the notion of social
product, and a hypothetical example of how a function might be
constructed is as follows. First, translate all objectives into a figure
for income, e.g. the employment of n extra men= £A, extra
savings of S = £B, etc. This would produce a series of values for a
project, the sum of which was V = A + + ... +
B Z· This
valuation of projects, with all objectives considered and expressed
numerically, could then be expressed in terms of one or more scarce
factors, e.g. capital or foreign exchange. Finally, projects could be
ranked on this basis and those with the highest ratios chosen within
the constraints imposed by the availability of scarce resources.
The ultimate valuation of resources would be on a trial and error
basis according to which were exhausted first. If one resource was
exhausted before another, optimisation would require that it be
revalued so that all resources are exhausted simultaneously.
7
Planning Economic
Development
Arguments for and against Planning
Planning in a variety offorms is frequently advocated as an alterna-
tive to the market mechanism, and the use of market prices, for
the allocation of resources in less developed countries. Reliance on
the market mechanism and market prices for resource allocation is
attacked for several reasons. Firstly, given the natural preference of
people for present rather than future satisfaction, resources in a free
market will tend to be allocated for the production of goods for
immediate consumption rather than for building up the means of
production, i.e. for the production of capital goods. Left to itself,
the operation of the market is likely to lead to a much slower pace of
development and a much lower level of future welfare than if
resources, via some form ofinterference with the market mechanism,
can be diverted to the production of capital goods. Secondly, as we
have already seen, market prices may provide a very imperfect
guide to the social optimum allocation of resources because they do
not reflect the opportunity cost or value to society of the use of
factors of production or the production of certain commodities.
A perpetual shortage of capital and foreign exchange, and a surplus
oflabour, at existing market prices is prima facie evidence of struc-
tural disequilibrium and a very imperfect market system which
may not operate to the benefit of society at large. Thirdly, because
of externalities many projects that less developed countries need, and
which would be profitable to society, may not appear profitable
under a pure market system in which all investment decisions are
left to private individuals. The level ofinvestment may fall below the
social optimum firstly because private investors ignore the external
PLANNING ECONOMIC DEVELOPMENT 181
economies and supplementary benefits of projects in calculating
prospective returns, and secondly, because the element of risk will
be higher for a series of uncoordinated individual projects than for a
co-ordinated investment programme systematically undertaken
through some central direction. A further disadvantage of the
market mechanism is that by itself it is unlikely to produce the rapid
structural changes which development requires. There is no guaran-
tee in a free market setting that the supply of' development' goods
and factors of production will be forthcoming in the quantities
required. Supply may be completely price inelastic, in which case
there is little alternative to the public provision of the goods or
factors in question. The market mechanism as an efficient allocative
device assumes that producers and owners of factors of production
respond to incentives and strive to maximise gains. Part of the
reason for bottlenecks and supply inelasticities may be that this
assumption does not always hold good in the context of backward
economies.
For a variety of reasons, therefore, interference with the market
mechanism is seen by some as a necessary prerequisite of a more
rapid pace of development. The belief of those that advocate some
form of planning or central direction is that positive action of this
type will help to achieve more expeditiously and more reliably the
goals that less developed countries set themselves. The overriding
object of a planned development programme is to incorporate
society's choice of goals and to allocate resources to meet these
goals. The vital question, to which there is no easy answer, is, what
form should planning take? Should the market mechanism be
superseded altogether, and the state assume total responsibility for
resource allocation? Or should planning take place within the
framework of the market mechanism with market prices adjusted
to reflect social values, and with taxes and subsidies to eliminate
divergences between private and social costs and benefits? In short,
should a command economy be established, or should the govern-
ment merely give incentives, and tax to provide certain goods and
services itself? Moreover, who should decide on the weight to be
given to the choice between present and future consumption,
between labour- and capital-intensive technology, and between
goods and services and leisure? Should the state be authoritarian
in the interests of future generations, or should the government
submit to the will of the present generation? No objective economic
answer exists to these questions. In the final analysis they must be

decided politically. It is the political system and the ideology of those
in power that ultimately determines the type of economic organisa-
tion.
Whatever form interference with the market mechanism takes,
however, it will inevitably involve some degree of state intervention
and control over the means of production, distribution and ex-
change, and the partial replacement of the market mechanism.
But it should be remembered that planning is not incompatible
with the use of the price mechanism. The abandonment of the market
mechanism need not necessarily involve dispensing with the use of
prices for resource allocation. The abandonment of the market
mechanism merely involves replacing one set of decision-makers
with another. But planners can, and do, use prices (or profits
determined by prices) as an aid to decision-making. Price move-
ments can still be used as signals for resource allocation even in a
totally planned economy in which private investment and owner-
ship is disallowed. The success of planning depends crucially on the
existence of an ample supply of able and technically qualified ad-
ministrators who are committed to development and not to the
furtherance of a political ideology at all costs or to their own
aggrandisement. The dangers of planning are that unqualified
and corrupt administrators will assume responsibility for resource
allocation and perform 'worse' than the market mechanism, and
that goals will be set which far exceed the country's capacity to
achieve them, leading to widespread disillusion with the planning
process. One attraction of the market mechanism, which classical
economists continually emphasised, is that it decentralises decision-
making and, if it functions smoothly, can stimulate efficiency and
growth without any elaborate administrative apparatus. Those who
see dangers in assigning decision-making to possibly inexperienced
and inept administrators may well accord with Adam Smith's view
that 'governments are rarely more effective than when they are
negative'.
In response to the clamour for planning in less developed coun-
tries, there are some who defend the market mechanism for resource
allocation on these traditional 'classical' grounds. It is argued that
if the market does not function properly in the interests of society,
there is a stronger case for making the market more perfect, and for
improving its functioning, than there is for planning. Market
imperfections and price distortions, in particular, are not thexnselves
arguments for planning, but rather arguments for ensuring that the
price mechanism functions better. There is, perhaps, a stronger case
for planning in the event of market failure, due to divergences
between private and social costs and benefits, but even here market
prices can be brought more into line with opportunity costs through
policies to promote competition, and subsidies could be given to
private producers in cases where the private return from socially
desirable projects falls short of the social return. The problem is
more intractable if producers and factors of production do not
respond to incentives, but defenders of the market mechanism would
question whether producers and workers in less developed countries
do have the aversion to risk-taking and effort that is sometimes
supposed.
But in reality, of course, the choice for the vast majority of
countries is not between complete laissez-jaire and total state plan-
ning of the means of production, but rather: what combination of
private and public enterprise; to what extent should the public
sector be extended; to what extent should there be interference with
private decision-making; to what degree should the private sector
be integrated into a national development plan? These are the
questions on which most less developed countries must reach
decisions.
Development Plans
Whatever a country's political ideology, a development plan is an
ideal way for a government to set out its development objectives and
to demonstrate initiative in tackling the country's development
problems. A development plan can serve as a stimulant to effort
throughout the country, and also act as a catalyst for foreign
investment and agency capital from international institutions.
Depending on the politics of a country, and its available expertise,
a development plan will vary in its ambitiousness from a mere
statement of aixns and pious hopes to detailed calculations (and
proposals for action) of the resources needed, and the amount of
output that each sector of the economy must generate, in order to
achieve a stipulated rate of growth of output or per capita income.
Anything more than a statement of aixns inevitably involves some
form of model-building, if only to delineate the relationships between
sectors of the economy and between the key variables in the growth
process.
Three basic types of model are typically used in development
planning. Firstly, there are macro- or aggregate models of the economy
which may either be of the simple Harrod-Damar type, or of a more
econometric nature, consisting of a series of n equations inn variables
which represent the basic structural relations in an economy between,
say, factor inputs and product outputs, saving and income, imports
and expenditure, etc. Secondly, there are sector models which isolate
the major sectors of an economy and give the structural relations
within each sector, and perhaps also specify the interrelationships
between sectors, e.g. between agriculture and industry, between capi-
tal- and consumer-goods industries, and between the government and
the rest of the economy. Thirdly, there are inter-industry models
which show transactions and interrelationships between producing
sectors of an economy normally in the form of an input-output
matrix.
Models of these types serve a twofold purpose. In the first place
they enable planners to reach decisions on how to achieve specified
goals. They highlight the strategic choices open to the policy-maker
in the knowledge that not all desirable goals are achievable simul-
taneously. Only with an understanding of the interrelationships
between different parts of the economy, and a knowledge of the
parameters of the economic system, is it possible for meaningful
and consistent policy decisions to be reached. Without detailed
information on which to base planning (or what has been called
'planning without facts'), the case for decentralised decision-making
becomes overwhelming. Secondly, models of the type described
above can perform an equally valuable function from the point of
view of enabling the future to be projected with a greater degree of
certainty than would otherwise be possible, thereby providing some
knowledge of what resources are likely to be available in relation to
requirements within a stipulated planning period. Various types of
model may be classified, therefore, according to whether they are
required for policy or decision purposes or for the purpose of pro-
jection and forecasting. The necessary constituents of a plan con-
taining both types of model are a statement of economic goals; the
specification of policy instruments or instrumental variables; the
estimation of structural relationships; the availability of historical
data; the recognition of exogenous variables; and last, but not least,
a set of national accounts for national income and expenditure,
foreign trade and even manpower to ensure consistency between
demand and the supply of resources available.
Policy Models
The essence of a policy model is that a certain set of objectives is
specified and the model is then used to determine the most appro-
priate measures to achieve those objectives within certain constraints.
In most development plans the primary objective is the achievement
of a target rate of growth of output or per capita income within
the planning period, or some terminal level of consumption, subject
to constraints on the composition of output, the distribution of
income and the availability of factor supplies. Given the time
horizon of the plan, and the objective function, the optimum strategy
can be worked out from the initial conditions. One of the big dangers
of planning in less developed countries is that planners and policy-
makers are prone to choose targets based on needs and aspirations
rather than on the basis of available resources, with the inevitable
consequence that the targets are not achieved. It is incumbent on
the planner to specify the constraints as accurately and honestly as
possible to avoid disillusion with the planning process.
The solution to a policy model of this kind is somewhat different
from one designed to find the value of certain instrumental variables
that will achieve a particular target rate of growth. The difference
can be illustrated with reference to the simple Harrod model of
growth which gives the level of savings necessary to achieve a par-
ticular growth rate assuming some value of the capital-output
ratio. The level of savings required to achieve a target rate of growth
may be very different from the level of savings available when the
problem is posed as one of maximising growth subject to constraints.
The level of savings necessary to achieve the target rate of growth
may conflict with society's wish for present rather than future con-
sumption; it may be incompatible with the supplies of skilled labour
necessary to work with capital, and the level of savings may not
even be achievable because of a limit to the capacity to tax or to
'force' saving through inflation.
It is clear from what we have said so far that the first step in the
formulation of a policy model must be the construction of an aggre-
gate model of the economy (incorporating, if need be, a sector
model and an inter-industry model) to highlight and check the

implications of the target rate of growth. If it is possible to formulate
a reasonably sound set of structural equations for the economy, this
exercise should not be too difficult.
The structural relationships given by the structural equations
are the restrictions on the instrumental variables which the policy-
makers must consider as datum. They represent, as we have said
already, such things as technological relationships in industry,
income-consumption relationships, foreign trade relationships, and
other behavioural and institutional relationships which, in the short
run at any rate, are outside the sphere of the policy-maker to
influence. In the long run, certain structural relations may, of
course, change with changes in the economy induced by the plan-
ning process itself. The capital-output ratio, for example, may
change under the influence of educational expansion and the de-
velopment of a country's infrastructure. Other, more basic, struc-
tural relationships, however, must be contended with in the planning
process even in the long run. Some of the important questions that
need to be answered with the help of the structural equations are:
Can capital be guaranteed in the quantities required? Will exports
and foreign assistance keep pace with the imports required? Will
the future demand for consumption goods, out of increases in per
capita income, exceed the supply and cause inflation? Can the
required interrelationships between industries be maintained so
that bottlenecks do not arise? If the objective function or goals of a
plan require a certain degree of balance between supply and demand
in various markets and sectors of the economy, then clearly the
objective function must be maximised subject to this general
requirement of avoiding shortages of capital, labour and foreign
exchange. The basic structural relationships required can be found
or determined by drawing on economic theory and econometric
research, or by intuitive 'feel' in the case of behavioural and insti-
tutional relationships.
Next, the instrumental variables must be specified. These are the
variables that the planners intend to influence in some way in order
to achieve the objectives specified within the constraints laid down.
If the basic objective is a higher rate of growth, one obvious instru-
mental variable is the level of savings and investment in relation to
national income. This can be influenced by tax policy, inflation and
assistance from abroad. If the structural relationships are given, and
the constraints are outlined, a solution to a policy model will give
a set of values for thejnstrumental variables that satisfies all the struc-
tural (and behavioural) equations in the model consistent with the
constraints imposed. In the case of savings, for example, the solu-
tion to a policy model would give the level of saving possible
compatible with balance between the agricultural and industrial
sectors of the economy; foreign borrowing of no more than a certain
amount; consumers' desires for present rather than future con-
sumption, and so on. This sort of policy model obviously lends
itself to programming, which we discuss briefly in Appendix
7.2.
It should be emphasised at this point that the planners ought not
to lay down rigid policy prescriptions for more than a very short
length of time. Planning itself may alter circumstances in the future
in an unforeseen way and planners must be ready to adapt to the
new situation. The means by which to achieve plan goals must be
flexible, and so too must the planning period. The term 'rolling
plan' is used to describe planning of a flexible nature where plans
are continually being reviewed and revised in the light of new
developments. If plans need to be revised, however, it is important
that they are modified cautiously to avoid erosion of confidence.
One of the purposes of planning is to imbue confidence, especially
among the private decision-making class who are most susceptible
to uncertainty, and it is vital that this confidence is not upset and
uncertainty increased by sudden changes in policy. There are three
typical planning horizons: one-year plans corresponding to a nor-
mal budget and accounting period; medium-term plans stretching
over four to seven years; and ten- to twenty-year plans (which is
the maximum length of time for which it is thought possible to make
meaningful projections).
Because of computational limitations, separate decision models
will probably have to be constructed for distinct policy purposes,
e.g. the choice of techniques, the optimisation of trade, the pattern of
final demand, the allocation of resources, etc. The application of
aggregate macro-models based on interconnections between income,
consumption, investment, employment and foreign trade, etc., is
also likely to run into the problem of sectoral differences in struc-
tural relationships such that if the structure or pattern of production
changes over time within the plan period, the use of average struc-
tural coefficients may lead to misleading results. This is where
sectoral models become important. If the economy can be con-

veniently divided into two or three basic sectors, the average co-
efficients can be adjusted for expected changes in the expansion or
contraction of the different sectors over the period of the plan.
Ultimately, any macro-model must be consistent with the models
and predictions for various sectors of the total economy.
Despite the potential use of decision models, they are not so
widely used as might be expected in less developed countries, es-
pecially decision models which attempt to optimise one thing or
another. In practice, planning tends to be much more ad hoc. The
reason is very often the difficulty in arriving at agreement over the
objective function. It is extremely difficult to balance the conflicting
aims of different parts of the community and to reach a balance over
conflicting economic objectives.
Projection Models
Any one, or all three, of the basic types of model mentioned at the
outset may be used for projection or forecasting purposes. Typically,
aggregate models for projection are of the Harrod-Domar type and
are used to indicate the amount of capital required in the future,
either to achieve a particular target rate of growth or to keep the
labour force fully employed (or both), assuming the labour force is
exogenously determined. Sector models, on the other hand, are
designed to project the output of various sectors and, if balanced
growth is required, to allocate resources accordingly. Lastly, inter-
industry models are designed to estimate the intermediate demands
for products as a result of a projection of expansion of the final
demand for goods and services over a future period of time (see
Appendix 7.1).
It would be wrong, of course, to think of forecasting models as
entirely distinct and divorced from decision models. Forecasting
models, especially of the econometric type, are often used, and
designed to be used, as decision models, with current choices and de-
cisions depending on the future the models portray. Indeed, there is
little use for pure forecasting models as such, except for esoteric
purposes.
Appendix 7.1
Input-Output Analysis
Input-output analysis is a particular planning and forecasting tech-
nique with a wide variety of applications. The purpose here is
simply to present the basic elements of the technique without going
into its refinements. If the reader's appetite is whetted, references
for further reading are given in the bibliography.
An input-output table provides a descriptive set of social ac-
counts, recording purchases by, and sales from, the different sectors
of the economy distinguished. Tables may be constructed for a whole
economy, for a region within an economy, or to show flows between
regions. They may vary in size and ambitiousness according to the
number of sectors identified and the purposes for which they are
constructed. The most common type of input-output table at the
national and regional level is that recording inter-industry trans-
actions showing, in Hirschman's terms, the backward and forward
linkages in an economy.
Input-output analysis has two major uses which must be clearly
distinguished. Firstly, input-output analysis may be used for pro-
jection and forecasting purposes. After some manipulation (which
we shall consider shortly), an inter-industry transactions matrix
can provide information to the planner on how much of commodi-
ties x~, x2, x3, etc., will be required at some future date assuming a
certain growth rate of national income or final demand. Information
of this nature is important if planning is to achieve consistency and
if future bottlenecks in the productive process are to be avoided.
Secondly, input-output analysis can be used for simulation pur-
poses. The simulation of development is concerned with what is
economically feasible, as opposed to forecasting which is concerned
with what one expects to happen on the basis of a certain set of
assumptions. In the case of simulation there is no presumption that
the simulated changes in the economy are actually going to occur.
If they are economically feasible the changes could occur, but
whether they do or do not may depend on a variety of prior changes

of an economic or institutional nature which may be outside the
planner's control. Using an input-output table for simulation
purposes requires first that the feasible changes (e.g. new activities
such as import substitutes) be identified, and then that the table
be used to estimate the impact of the changes on the rest of the
economy. Again, with some manipulation the inter-industry trans-
actions matrix can then be used for providing answers to such
questions as: how will the structure of the economy be altered if a
new steel mill is introduced; or what will be the repercussions on
the economy if a series of import-saving schemes are introduced?
Three major stages must be gone through before knowledge of
input-output relations can be put to practical uses in the planning
field in the manner described above. The first step must obviously
be the construction of the input-output table itself recording the
relevant transactions. The second stage involves the derivation of
what are called input-output coefficients (or input coefficients, for
short). The third task is the inversion of the matrix of direct input
coefficients to obtain a general solution. The content of these stages,
and the underlying assumptions of input-output analysis, will
become clear as we examine the three stages in turn. We shall wait
until the end, however, to spell out the assumptions explicitly, and
to discuss the shortcomings of input-output analysis for planning
purposes.
The Input-Output Table

The layout of a typical input-output table, and the sectors and
transactions distinguished, is shown in Table A7.1. The table is
divided horizontally into a processing and payments sector, and
vertically to distinguish between intermediate users of goods and
factors of production, and final users. Thus the upper left-hand
quadrant of the table records inter-industry transactions; the lower
left-hand quadrant gives the payments by industries to the factors of
production; the upper right-hand quadrant gives the final demand
for goods and services produced (including exports but minus
imports since the table is only concerned with domestic production)
and the lower right-hand quadrant gives the direct sales of factors of
production to final users, e.g. the private consumption sector buys
domestic help, and the government sector' consumes' civil servants I
Purchases Intermediate users ~ Final demand
by -
~ Consumption Investment Exports Imports Total
-
== (I) (E) (M) output
== Private Government
Sales Industries = (C) (G)
by -
I 2 3 j n f=
1--
I Xu X12 Xts Xti Xtn f= Xtc Xw x11 XtE XtM Xt
1--
2 x21 X22 X2s X21 X2n r= x2C X2G x21 X2E X2M x2
1--
PROCESSING 3 xa1 Xa2 Xaa XaJ X an F= Xsc Xac: Xar XaE XaM Xa
t--
i Xu x,2 x,a x,J x,n I==
t-- X;c Xw Xu x,E XIM x,
n Xnt Xn2 Xna XnJ Xnn F=
t--
Xnc XnG Xnr XnE Xnlll Xn
p Wages
Wt w2 Wa WJ Wn ~ We We: w
A 1--
y 1--
M
Rent Rt. R2 Ra R, R" Rc Ra R
~
E
N Interest Dt D2 Da -!>J Dn F=
1---
De De: D
T p
s Profits Pt p2 Pa P, P" ~ Pc Pa
t--
Total output Xt x2 Xa x, Xn ~ c G I E lvJ
t-- ~-------- - - - - - - ----
~------
TABLE A7.1
Input-Output Table
The notation used in the table to denote the various transactions

should be easy to follow. Take first the processing sector. The Xs
stand for the value of output, and the two subscripts denote the
origin of the output and its destination respectively. Generalising,
X11 denotes sales by industry i to industry j or inputs into industry
jfromindustryi,wherei = 1, ... , nandj =I, .. .,n. The disposal
of outputs to final users can be represented in exactly the same
way.
In the payments sector, the subscript simply refers to the industry
or final demand sector making the factor payment.
We see, then, that in the processing half of the table each row
shows how the output of each industry is disposed of, and each
column shows the origins of inputs into each sector. The sum of the
rows (minus imports) gives the total value of domestic produ~tion
of each industry.
In the payments half of the table, the sum of each row gives the
value of the various factor payments, and the sum of each column
in the bottom left-hand quadrant gives the total value added (in
the form of factor payments) to the inputs bought by the different
industries.
In the intermediate sector, the sum of the columns gives the
value of the total output of each industry, i.e. the sum of the value
of inputs in the processing half plus the sum of value added in the
payments half. Thus the columns in the intermediate sector and the
rows in the processing sector both add up to the total value of
domestic production. Let us demonstrate this proposition using the
notation of our table. Call industry I the coal industry. The (domes-
tic) output of coal must equal all its intermediate uses (Xu, ... , X1n)
plus all its final uses (minus imports). We thus have a row for the
coal industry of the form
n
X1 = ~ X11 + Xw + Xw +Xu+ Xu:- XIM. (A7.1)
i=l
But the (domestic) output of coal must also equal the value of
inputs into the coal industry plus the value added to those inputs
by employing factors of production to work on them. We thus have
a column for the coal industry of the form
n
X1 = ~ Xu
i=l
+ W1 + R1 + D1 + P1. (A7.2)
Input Coeificients
The second stage in the practical use of input-output analysis is
to derive input coefficients from the inter-industry section of the
transactions table. This is done very simply by dividing each column
entry in the matrix by the sum of the column. If we denote the input
coefficient atJ, then ati = ~;1 . The input coefficient a11 thus gives
the amount of purchases from each industry to support one unit of
output of industry j. If output was measured in pounds sterling
and aii equalled 0·1, this would mean that every pound's worth of
output of industry j would require lOp worth of input from in-
dustry i.
Xti = aii XJ, and our equation (A7.1) in general form can now
be written
X~, = zn
j= I
ati X1 + Xw + Xw + Xu + XtE - XtM· (A7.3)
With a series of equations of this nature for each sector, represent-

ing the rows in the processing sector of the transactions matrix,
it is possible through the use of the input coefficients, and with the
aid of matrix algebra, to work out what the effect will be of changes
in demand for the product of one sector on all other sectors. Equa-
tion (A7.3) for all industries i = 1, ... , n becomes a system of n
simultaneous equations in n unknown variables which can be solved
if certain conditions are satisfied. The technique is to specify the
final demands for each industry, i, and then to determine the Xis
from the new values of the input requirements, which give both the
direct and indirect requirements from each industry.
In specifying the final demands for each industry, one obtains
from equation (A7.3) the following set of relations for each industry:
Yt =Xi- z ati X1 + mtXt

n
j= I
(i = 1, ... , n) (A7.4)
where ri is total final demand

Xt is total output
z atiXi is the sum of intermediate demands

n
j= I
and mtXt is imports.
The solution to the problem of finding the values of all Xts which
satisfy all the final demands may be written in the form
X, = bu1i + bi2 1"2 + ... + btn Tn (i = 1, ... , n). (A7.5)
The coefficients bu, ... , n give, per pound of delivery to final
demand made by the industries listed along the top of the inter-
industry matrix, the total input directly and indirectly required
from industries listed at the left of the matrix. Equation (A7.5),
therefore, gives the total output (X) of industry (i) consistent with
the final demands for all other industries' products (1"1, ... , Tn)·
These equations for each industry represent a transformation of
the original equation (A7.4) in which a new set of constants, biJs,
is derived from the original parameters (aiJ and mi)· The new equa-
tions are referred to as the general solution and can be found, in diffi-
cult cases involving more than two or three equations with two or
three unknowns, by the technique of inversion of the matrix, i.e. the
inversion of equation (A7.4) in matrix form.
Let us now express, therefore, our input-output problem in
matrix form and, using a little elementary matrix algebra, show
how a general solution is arrived at by matrix inversion. For ease
of exposition, and also to compare our results using a simple iterative
procedure, we shall take two sectors and two outputs only.l
With two industries, and ignoring imports, the equations of type
(A7.4) become
(A7.6)
or in matrix form:
(A7.7)2
1 More complicated, real-life, situations require a greater knowledge of

matrix algebra than can be given here, but good introductions can be
found in W. Miernyk, The Elements of Input-Output Anarysis (New York:
Random House, 1965); M. H. Peston, Elementary Matrices for Economics
(London: Routledge & Kegan Paul, 1969); and G. Mills, Introduction to
Linear Algebra (London: Allen & Unwin, 1969).
2 (A7.7) is equivalent to (A7.6) bymultiplyingeachelementin the rows of
the 2 x 2 matrix by the equivalent element in the column vector of Xs.
Let A be the matrix of input-output coefficients, i.e.
i.l __ [an a1J (A7.81

ll21 a2
The expression on the left-hand side of equation system (A7.7) is
thus equal to the identiry matrixl minus the matrix of input-output
coefficients in (A7.8).
In matrix notation, therefore, our input-output system may be
written as (I - A) X = r. ((I - A) is often called the Leontief
matrix, named after the 'father' of input-output analysis, Wassily
Leontief.) It follows that X= (I- A)-1 Y; that is, to solve for X,
given Y, we need to find the inverse of the Leontief matrix.
The inverse of a matrix, A, is another matrix, B, which when multi-
plied by A gives the identity matrix I. This corresponds to the
inversion of real numbers where b is the inverse of a if ab = ba = 1.
Similarly, B is the inverse of the A matrix if AB = BA = 1, or the
identity matrix I. (I= 1 since AI= A.)
Hence B is the inverse of A if
[
an a12l X [bn b12] = [l ~]
a21 a2J b21 b22 0
This gives four equations in four unknowns, and it is then possible
to solve for the elements in the inverse matrix bu, b12, b21, b22· By
simple arithmetic:
bt2 = ____a=12=---
an a22 - a12 a21
-a21
au a22 - a12 a21
au
1 An identity matrix (I) is a square matrix with 1's in its principal
diagonal and zeros everywhere else. In the 2 X 2 case I= [~ ~l As
we see later in the text, the importance of this matrix is that it plays a
similar role to the number 1 in ordinary algebra, i.e. 1 X a = a X 1 = a.
Similarly, in matrix algebra IA = AI= A. In short, an identity matrix
is a matrix which when multiplied by A yields A as the product. In the
_[au a12] X [1 0] -_[au a12] _

2x2case
M- -~
a21 a22 0 1 a21 azz
In the simple2 X 2 case these are the formulae for working out the
elements of the inverse from the elements of the original matrix.
Notice that the denominator is the same in all cases, namely the
product of the elements of the principal diagonal in the original
matrix minus the product of the other diagonal (or what is called
the determinant of the matrix).l
(/- A)-1 is therefore some matrix which we can call B, which
when multiplied by (/-A) gives the identity matrix I. We now
haveX=BY.
In other words, each industry's output is equal to the sum of the
elements of the final demand matrix (Y) (which is a column vector
n X 1) times coefficients of the inverse matrix of I - A.
[ xl] = [bu b12] x

x2 b21 b22 rJ
[rq
This is exactly what is stated in equation form in (A7.5). By matrix
multiplication:
X1 = bu r1 + b12 r2
X2 = b21 r1 + b22 r2.
The output of industry X will be consistent with the final demand for
the products of industries 1, 2, etc.
The btjS give the direct and indirect requirements from industries
1, ... , n per pound of final demand for the products of industries
1, ... , n.
Now let us consider a simple arithmetic example. Suppose that
the direct input-output coefficients (the atJS) for the steel and
coal industries of a country are as given in Table A7.2. We wish to
show what the output of coal and steel must be to satisfy certain
final demands for steel and coal.
TABLE A7.2
Input-Output Coefficients
~~ s
Coal
Coal
0·1
Steel
0·4
Steel 0·3 0·2
1 A knowledge of determinants is necessary for the inversion of larger
matrices unless iterative techniques are used (see references cited earlier).
Thus A= [0·1 0·4]

0·3 0·2
I- A = G] - [::: :::J = [ _:::

-0·4]
0·8
From our formulae to derive the elements of the inverse matrix
bn = 1-!
ht2 = i
b21 =-!
b22 = 1-!
Thus (I- A)-1 =[I: 1-!i] =B.
If final demands are

Y= c:J
then total industry outputs consistent with this final demand for
both products would be
X= [1: ~:J c:J [1: : ::: ~: : ::J G~:J

X = =
or coal output = 1!- X 15 +i X 10 = 26-J

steel output= -! X 15 +
1~- X 10 = 22!.
We can also solve for the change in each sector's output following
a change in final demand. Suppose, for instance, the final demand
for steel is projected to rise from 10 to 15. From our inverse matrix we
can work out by how much coal and steel output will have to in-
crease in order to meet this increase in final demand. The new final
demand vector is
Y= c:J
and the output of the two industries consistent with this new demand
vector is
X= [
1-!
t 1!
i] X
[15] [30]
15 = 30
coal output must rise by 3-l from 26f to 30
and steel output must rise by 7! from 22! to 30.
Given the inverse matrix, any manipulation of the final demand

vector is possible and a general solution can be reached.
Our result using matrix methods can now be checked by going
back to the original 2 X 2 matrix of direct input coefficients given
in Table A7.2 and calculating step by step the direct and indirect
demands made upon the steel and coal industries as a result of the
increase in final demand for steel. The task is laborious, but let us
proceed if only to underline the beauty and usefulness of matrix
methods.
We use an iterative procedure analogous to that used in solving a
single-equation national income model with a given propensity to
consume and a postulated change in autonomous demand. In inter-
industry analysis, however, there is more than one 'multiplier'
coefficient to contend with and there may be several changes in
final demands.
The ultimate change in output must equal the sum of the direct
and indirect effects of the change in final demand. We assume that
the first increment of production of commodity Xt is equal to the
increase in final demand for Xt. The first-round effects on production
in each sector are then given by the input-output coefficients. The
second-round effects are given by derived demands in the first
round multiplied by the input-output coefficients, and so on.
In our two-sector example, the primary effect of an increase in
demand for steel of 5 is an increment of steel production of the same
amount, i.e. the primary effect on steel is 5.
The first-round effect on steel is 0·2 X 5 = I.
The first-round effect on coal is 0-4 X 5 = 2.
These derived demands, following the initial increase in demand,
are then multiplied by their respective input coefficients.
Thus, the second-round effect on steel is 0·2 X I = 0·2 } _ 0 .8
and 0·3 X 2 = 0·6 -
and the second-round effect on coal is 0-4 X I . 0·4 }
0·1 X 2 = 0·2 = 0 "6·
The second-round derived demand figures are further multiplied
by their respective input coefficients.
Thus, the third-round effect on steel is 0·2 X 0·8 = 0·16} _ 0 .34
and 0·3 X 0·6 = 0·18 -
and the third-round effect on coal is 0-4 X 0·8 = 0·32} 0
0·1 X 0·6 = 0·06 = "38
and so on.
Summing after three rounds, we have:
Change in steel output is = 5 + 1 + 0·8 + 0·34 = 7·14.
Change in coal output is = 2 + 0·6 + 0·38 = 2·98.
These compare with our final results by matrix methods of 7! and
3l for steel and coal, respectively.
Calculation by this iterative procedure is clearly tedious and
becomes impossibly complicated with more than five or six sectors
to cope with. Similarly, matrix inversion by hand is laborious and
complicated, but fortunately there is recourse to the computer.
When it comes to input-output work for practical policy-making
purposes, a computer is absolutely indispensable. To be of much
practical use, the degree of disaggregation required in input-output
analysis is such that work would be impossible without mechanical
aids. As a matter of interest, a regular computer will invert a
30 X 30 matrix in a matter of seconds.
The Assumptions of Input-Output Ana!Jsis

Now let us briefly turn to the assumptions underlying input-output
analysis and consider their validity. The basic properties of tradi-
tional (Leontief-type) input-output models are fourfold: firstly,
that there are no joint products, which means that each commodity
is produced by only one industry and each industry produces only
one product; secondly, that technical coefficients are fixed, i.e. no
substitutability between inputs, so that input functions are linear
and the marginal input coefficient is equal to the average; thirdly,
that the total effect of carrying on several activities is the sum of the
separate effects, ruling out external economies and diseconomies;
and lastly, that technical progress is static and production is subject
to constant returns. All these assumptions are necessary if the input-
output coefficients are to be constant and the estimated inputs and
outputs are to balance at the aggregate level.
The validity of these assumptions is, of course, open to question,
especially over the long run. Indeed, we know that changes do occur
of a type which input-output analysis denies. The question is
essentially an empirical one: how large are the errors associated with
sacrificing accuracy for convenience? In the short run they may

not be substantial, and the use of input-output relations for fore-
casting and simulation within periods of five years may not yield
results that are too far wide of the mark. There may not be much
substitutability between inputs in the short run owing to the nature
of technology or the comparative stability of relative factor prices.
Over the long run, however, changes in the ratio of factor prices,
and technical progress, can alter input-output coefficients substan-
tially. Similarly, the composition of demand may be assumed to be
fairly static in the short run, but is liable to considerable change in
the longer run. If commodities with different input coefficients are
grouped together in the same sectors, a changing pattern of demand
will alter a sector's input-output coefficients automatically. To
overcome this problem it is common to use the similarity of input
structures as a criterion for the aggregation of activities in the original
input-output table.
The most serious criticism of input-output analysis for pro-
jection, forecasting and simulation purposes is the fact that the
marginal or incremental input coefficient may differ substantially
from the average relationship embodied in the matrix. This is an
analogous problem to the use of the average capital-output ratio
in investment planning, and assuming that the incremental capital-
output ratio is equal to the average. With technical change, in-
creasing returns to scale, and the changing structure of international
trade, marginal input coefficients may diverge substantially from
average coefficients over time. Needless to say, this possibility could
lead to inaccurate forecasts of the need for certain commodities on
the basis of projections of final demand. But there are ways of
overcoming this inherent weakness of traditional input-output
economics. The problem is lessened if the transactions table from
which the input coefficients are derived can be produced at regular
and frequent intervals. With better methods of data collection, this
itself is more feasible. The problem can also be overcome to a
certain extent if some satisfactory adjustment can be made to 'out-of
date' coefficients to allow for such factors as changes in technology
and the pattern of trade. It is fairly common now, for instance, to
adjust out-of-date national coefficients for a sector by the coefficients
prevailing in the best-practice techniques in that sector. And for less
developed countries, or backward regions within a nation, it can be
assumed that within a certain time period they will have the
coefficients prevailing at the present in other, more advanced,
countries or regions.
In growth and planning exercises, knowledge of input-output
relationships is of greatest use when growth is based on the expan-
sion of domestic demand and where there is considerable inter-
dependence between activities. We noted earlier, however, that
typically in less developed countries the backward and forward
linkages between sectors are not very great. Many of the cells in the
inter-industry transactions matrix are likely to be empty. The
reason is that a great deal of output goes directly to meet final
demand, or else is exported. If growth is primarily based on ex-
ports, the traditional input-output table is of limited use, although
there are a number of ways in which it may be modified for the
general purpose of assisting the achievement of balance within the
economy.l
Appendix 7. 2
The Programming Approach
to Development
The main task of development strategy is to ensure that resources
will be forthcoming to meet the goals of a development pro-
gramme, and that the resources are allocated efficiently subject to
certain constraints. In our earlier discussion of resource allocation
we were more concerned with the investment criteria that should be
applied in the light of particular goals than with the efficiency with
which resources were to be used, or whether the application of the
1 For Zambia, for example, Seers has used the inter-industry matrix in a
social accounting framework for the wider purpose of balance in sectors of
the economy other than the production sector. See D. Seers, 'The Use of a
Modified Input-Output System for an Economic Program in Zambia', in
I. Adelman and E. Thorbecke (eds), The Theory and Design of Economic
Development (Baltimore: Johns Hopkins Press, 1966).
criteria violated certain of the broader constraints mentioned. This

was, in fact, a matter of necessity because ordinary marginal analysis
is not appropriate to situations in which the aim is to obtain an
'optimum' solution subject to constraints which are not precisely
specified (or what are called inequalities, e.g. that no more than R
resources should be used). Programming, however, can provide a
simultaneous solution to the three basic purposes of development
planning, which are: the 'optimum' allocation of resources; effi-
ciency in the use of resources (through the proper valuation of
resources, and the avoidance of 'social' waste) ; and the balance
between different branches of the national economy.
The essential feature of the programming approach is that it
deals with a whole complex of interrelated projects rather than with
a marginal project, and is concerned with much wider considera-
tions than marginal analysis. The partial equilibrium approach in
development planning is only justified if the number of projects is
small or social considerations are taken into account in the appli-
cation of investment criteria, as the social marginal product cri-
terion tries to do. The programming approach also tends to be more
dynamic, looking at the development of the economy and the
interrelationship between projects over a long period of time.
Programming in general is best described as the mathematical
method for the analysis and derivation of optimum decisions which
do not violate non-specific constraints. Theoretically, it offers the
best solution to rational decision-making in economic planning.
Indeed, it is virtually true to say that the optimal solution to a
given problem within constraints can onry be given by the program-
ming approach. The methods of computation normally involve
trial and error procedures, but the methods are so constructed that
each trial yields values which are closer to the correct answer than
the preceding values. We shall not here go into the computational
methods. We are more concerned with the underlying principles.
The programming approach to development planning was ini-
tially used for the purpose of achieving balance between the supply
of and demand for commodities and factors of production. As
Cheneryl says, it can be considered, historically speaking, the opera-
tional counterpart of the theory of balanced growth. With the ad-
vent of mathematical programming techniques, however, it has
1 H. Chenery, 'Comparative Advantage and Development Policy',
also become possible to test the efficiency with which resources are
used, and to reconcile the consistency of plans (i.e. balance) with
efficiency (which means comparative advantage in an open
economy). One particular programming technique which can assist
in providing a simultaneous solution to the three main tasks of
development planning is the technique of linear programming.
The prefix 'linear ' will become clear below. The purpose here is to
describe the features of a simple linear programming problem, and
to show how the solution to the problem of the optimum allocation
of resources also implies a solution to the proper valuation of re-
sources and efficiency in their use. In other words we shall show,
given certain constraints, how a solution to the problem of resource
valuation emerges as a simultaneous by-product of the solution to
the problem of resource allocation and which commodities to
produce. This property of linear programming is called the duality
of price determination and allocation.
LINEAR PROGRAMMING
Linear programming is a very versatile technique. Four important

uses to which it can be put in the context of developing countries
(or any country concerned with allocative efficiency) are as follows:
firstly, it can be used for choosing between different techniques for
making the same commodity; secondly, it can be used for deciding
the best combination of outputs with given techniques and factor
endowments; thirdly, it can be used for deciding whether it is more
efficient to produce commodities at home or buy them from abroad;
and fourthly, it can be used to determine the most efficient spatial
location of activities. All the above-mentioned issues are optimising
decisions of one form or another and can be incorporated into the
standard form of a programming problem consisting of three parts:
(1) the objective function to be maximised; (2) the constraints which
must not be violated; and (3) the non-negativity conditions, e.g.
that outputs or exports should not be negative.
Given this standard form of a programming problem, let us illus-
trate the principle oflinear programming, and its important use for
the valuation of resources, by taking the case of the choice of outputs
xl and x2 that can be pro.duced with given techniques subject to
constraints over the availability of resources Y1 and Y2 .
Consider, then, the case of two commodities X1 and X 2 whosr
price per unit is P1 and Pz. The objective function to be maximised

is P1X1 + PzXz, which will maximise national income. The problem
is what combination of X1 and Xz to produce, subject to the non-
negativity conditions, X1 ~ 0, X 2 ~ 0, and the resource con-
straints. Suppose that there are two resources, capital and labour,
and that no more than Y1 labour is available and no more than
Yz capital. Suppose further that commodity X1 requires a11 labour
and a21 capital, and commodity Xz requires a12 labour and a 22
capital. The objective function must therefore be maximised subject
to the two constraints or inequalities
auX1 + a1zXz :'( Y1 (labour)
az1X1 + azzXz :'( Yz (capital).
The problem is linear since in both the expressions to be maxi-
mised, and in the constraints, all the variables are multiplied by
constants and added together. Commodity prices remain the same
regardless of the level of output; similarly, the input coefficients
remain constant, implying constant returns to scale.
Our programming problem can now be considered diagram-
matically. In Fig. B7.1 the quantities of the two goods, X1 and X 2,
are measured on the two axes and the resource constraints are
represented by the straight lines Y1Y1 and YzYz.
+
Consider first the constraint or inequality auX1 a12X2 :'( Yt,
which is represented by the line Y1Y1. This gives the maximum
combination of the two commodities that can be produced given
the amount of labour available. Any combination of X1 and Xz on
the line, or within the area bounded by the line, is feasible. But now
consider the second constraint or inequality a21X1 +azzXz :'( Y2 ,
which is represented by the line YzYz. The combination of the two
goods X1 and Xz that can be produced must satisfy both inequalities,
and thus only combinations on, or within, the boundary Y 2PY1
are now feasible. Outside this boundary one of the constraints would
be violated. The area bounded by the line Y 2PY1 represents a kind
of production possibility curve and is called in linear programming
'the feasible region'. Any point within the feasible region, represent-
ing combinations of X1 and Xz which do not violate the capacity
constraints, is called a 'feasible solution'.
It is clear that not every feasible solution will involve the full
employment of resources. Any point within the area YzPY10 will
involve unemployment of both resources, Points on the line Y2PY1,
but off one of the constraint lines, will involve underutilisation of
one of the resources. Point P is the only combination of X1 and X 2
that would fully employ both resources. Underutilised resources
are called 'slack variables' and represent 'costless' resources. As we
indicated at the outset, we are interested in the optimal value of
these 'slack variables' since they will be indicative of shortages and
Fw. B7.1
surpluses. Let us label the slack variables S1 and S2, and now refer
to the variables X1 and X 2 as structural, or ordinary, variables. With
the use of slack variables, the constraint equations become equalities
and the programming problem becomes
Maximise P1X1 + P2X2
subject to anX1 + a12X2 + S1 = Y1
a2lxl + a22x2 + s2 = r2
and the non-negativity conditions
x1 ~ o; x2 ~ o; s1 ~ o; s2 ~ o.
As before, any values of the variables (X1, X2, S1 and S2) which
satisfy both the constraints and the non-negativity conditions are
said to represent a feasible solution. Any set of values of the variables
in which the number of non-zero-valued variables (either ordinary
or slack) is equal to the number of constraints is called a 'basic
solution'. All basic solutions are feasible, and what we want to show
is that in any linear programming problem an optimal solution can
be found by considering only the basic solutions. As we shall come
to see, this reduces the task of the allocation problem enormously.
In our example, an optimal solution is that which maximises the
national income, i.e. the sum of the quantities of xl and x2 each
multiplied by their respective prices. The respective prices of X1
and X 2 can be represented by parallel price lines as in Fig. B7.2.
Given that prices are fixed and do not vary with output, it follows
that the optimal point of a linear programme must always lie on the
boundary of the feasible region since, if a commodity has a positive
price, income can always be increased by expanding production
to capacity until one or both capacity limits are reached. Given the
FIG. B7.2
feasible region T2PT1 in Fig. B7.1, there will always be at least one
optimal solution which occurs at one of the corners r2, P or r1 of
the feasible region. The optimal point will either be a tangency
point, P, or, if the price lines are parallel to one of the segments of
the feasible region, say Prb then the entire segment, including P
and rb will be optimal.
We can now show that all basic feasible solutions are represented
by corners of the feasible region. Take for example the labour
constraint T12Tn, as in Fig. B7.3.
X
l
FIG. B7.3
At corner T12, X1 = 0, S1 (our slack variable) = 0, since all the

resource is used, and X2 > 0.
At Tn, X2 = 0, S1 = 0, and X1 > 0.
At 0, X2 = 0, X1 = O, and S1 > 0.
Therefore, at all these corner points the number of non-zero
values of the variables is equal to the number of constraints (i.e.
one), and this is the condition for a basic solution. It can easily be
seen that no other point represents a basic solution since on the
line xl > 0, x2 > 0 and sl = 0, and inside the line all variables
are positive; that is, in all other cases the number of non-zero
values of the variables will be greater than the number of con-
straints.
We conclude, therefore, that in any linear programming problem
an optimal solution can be found by considering only the basic
solutions. There will always exist an optimal solution in which the
number of non-zero-valued variables (both ordinary and slack) is
exactly equal to the number of constraints in the problem. The
problem of resource allocation would seem to be greatly simplified
by this conclusion, for the implication is that if an economy has
only n constraints but n + m activities to choose from (m > 0),
national income can be maximised by cutting its range of activities
to n (i.e. to the number of constraints). The logic of this implication
is easy to understand. Take our previous example of one constraint.
If there is linearity (i.e. constant returns to scale), it would pay to
produce the commodity with the highest price up to capacity. It
will only pay to introduce additional commodities if the number of
constraints is increased. It is equally clear, however, that the validity
of the conclusion reached hinges crucially on the assumption of
linearity.
The Dual
The solution to a linear programming problem yields, in fact, two
sets of solutions. One solution, which we have been considering up to
now, is the 'best' combination of outputs, i.e. the values for X1 and
x2 which maximise the national income given the prices of products.
This is called the 'primal'. The second solution is known as the
'dual', or the prices of the resources that minimise the costs of
producing a combination of outputs. The prices are those that would
prevail if the investment pattern to maximise national income was
undertaken. Let us first write out the dual of the maximising prob-
lem (the primal), and then outline the symmetry between the
maximising and minimising problems, followed by an economic
interpretation of the dual. We may write the dual of our maximising

probl~m as follows:
Primal
subject to auX1 + a12X2 ~ Y1
a21x1 + a22x2 ~ r2
and non-negativity conditions x1 ~ 0; x2 ~ 0.

Dual
Minimise 71 Y1 + 72 Y2
subject to au71 + a2172 ~ P1
a1271 + a2272 ~ P2
and non-negativity conditions 71 ~ 0; 72 ~ 0

where 71 and 72 are the shadow (or accounting) prices of the factors
Y1 and r2. The requirement is that the cost of producing commodi-
ties x1 and x2 should not be less than the price because this would
violate the condition of maximising income, i.e. additional income
could be generated by producing more.
In the dual, the inequality signs in the constraints are the reverse
of the signs in the primal. If slack variables are introduced, their
sign remains positive in the primal, but in the dual they must be
subtracted from the left-hand side of the constraint equations. Let us
label the slack variables in the dual as Z1 and Z2· The constraints
now become equalities and the primal and dual may be written as
subject to auX1 + a12X2 + S1 = Y1
a21x1 + a22x2 + s2 = r2.
Minimise 71Y1 + 72Y2

subject to au71 + a2172 - Z1 = P1
a1271 + a2272 - Z2 = P2.
If we now put the primal and the dual in matrix form, the symmetry
between the two problems becomes apparent immediately.
Primal matrix Dual matrix
x1 x2 71 72
p1 p2 r1 r2
s1 r1 au a12 Z1 p1 au a21
s2 r2 a21 a22 Z2 p2 a12 a22
If we look at the primal and dual in equation and matrix form we
observe the following: if the primal involves ?= signs the dual in-
volves :::::;; signs; the price constants, P1 and P 2 , in the primal,
become constraints in the dual- replacing the capacity constraints
Y1 and Y2; the coefficients in the constraint inequalities, au, a12,
which go from left to right in the primal, go from top to bottom in
the dual. And as far as the two matrices are concerned, the dual
matrix is merely the primal matrix with the positions of the bottom
left-hand corner and the top right-hand corner reversed.
Let us now give a numerical example of the primal and dual and
use this example to give an economic interpretation of the dual.
Maximise 2X1 + 6X2
subject to 4XI + X2 ::S; 5
3X1 + 2X2 ~ 7.
Minimise 571 + 772
subject to 471 + 372 ?= 2
71 + 272 ?= 6.
To produce one unit of commodity X 1 the total value of inputs is
471 + 372. This may exceed or equal, but not be less than, 2, which
is the price per unit of output X1 . The inequality sign tells us that we
must assign values to each of the inputs sufficient to account for the
price of the commodity in question. An analogy with a business
enterprise will be useful here. A firm must choose values for its
scarce inputs (Y1 and Y2 ) that will account for all its profits. Values
of 71 and 72 must be chosen so that the net profit from production is
accounted for. If the accounting values are such that profits are not
completely imputed to the factors, then the accounting values must
be raised until the unimputed profit has been eliminated. Thus the
variables 71 and 7 2 in the dual problem can be interpreted as the
required accounting values of the scarce inputs. If, of course, factors
are not scarce, their accounting values must be zero because if their
supply increased profits could not be increased because of other
constraints, and if their supply was reduced profits would not fall
because output would not be affected. Thus the aim of the dual is to
minimise the total cost of providing the inputs in question. At the
minimum the total cost of providing the inputs will be exactly equal
to the total profit derived for their use. If an input is in excess
supply, an additional unit will add nothing to profit and therefore
its opportunity cost is zero.
The valuation of the factor inputs affects the optimum com-

bination of outputs through the slack variables, Z1 and ..(2, in the
dual.
We previously wrote the dual equation for product X1 as
aur1 + a21r2 - Z1 = P1.
This can be written
Z1 = aur1 + a21r2 - P1.
If aur1 + a21r2 represents the value of resources going into X1, and
Z1 is positive, this implies that the resources used in producing X1
are worth more than the commodity and thus production of X1 would
involve a relative loss to society. The valuation of factors, therefore,
affects the opportunity cost of producing products. Products with
non-zero slack variables should not be produced. If, however,
resources were valued such that there was no difference between
the total value of resources and the price of the commodity, then
Z1 would be zero and it would be 'profitable' to produce X1. (In a
nutshell, this is why valuing surplus labour at zero would lead to the
production of commodities using labour-intensive techniques.)
We are now in a position to state an important duality theorem.
An optimal solution to the allocation and valuation problems
requires that
X1Z1 = 0 for each commodity j
and reSt = 0 for each input i.
X1Z1 = 0 requires either that X1 = 0 or that Z1 = 0 (or both).
If the output of X1 is positive, then we must have Z1 = o (i.e. its
accounting loss must be zero with all profits imputed to the scarce
factors). If Zl > 0, then we must have xl = 0 (i.e. xl must not be
produced). The logic of not producing commodities with a non-
zero accounting loss is simply that the inputs used in producing the
good would be more valuable elsewhere. The accounting losses may
be thought of as opportunity costs. Thus if two commodities required
the same amount of scarce inputs, it would pay to produce the
commodity with the least opportunity cost.
The interpretation of the other equation is similar. reSt = 0
requires that either rc= 0 or St = 0 (or both). If Se > 0 (i.e there
are surplus inputs), then we must have re = 0 (i.e. the input must be
given a zero accounting price). The logic of valuing surplus inputs
at zero is simply that if an extra unit of the input was forthcoming,
it would add nothing to profit because it is already in surplus;
similarly, if a unit were taken away it would not subtract from
profit.
When all profits have been eliminated, there will be a unique set
of factor prices corresponding to each factor's marginal contribution
to output (as under perfect competition) and to the combination of
outputs. In the language of programming, these derived or shadow
prices are the Lagrange multipliers of a constrained optimisation
problem. They correspond with the marginal products of factors of
production in an optimal situation when all alternative uses have
been taken into account. That combination of products which
maximises income and minimises costs by obtaining the shadow
prices of factors is called 'efficient'. The properties of the set of prices
associated with the point of efficiency are:
1. That no activity has a positive profitability.
2. That every activity carried out has zero profitability.
3. That the prices of all factors which are not exhausted are
zero.
Linear programming highlights the importance of using shadow
prices for factors in achieving 'efficient' resource allocation. The
results are identical with those that would prevail under perfect
competition, given the same technological assumptions. The
shadow prices themselves may therefore be considered 'efficiency'
prices since they lead to the most efficient allocation of resources.
Two main drawbacks oflinear programming may be cited. Firstly,
the programming problem may not be linear, in which case other
techniques must be resorted to. Secondly, it is a weakness of linear
programming that product prices are assumed to reflect society's
marginal valuation of them. On the one hand there is the problem
of imperfect product markets with prices in excess of marginal
cost. The other related, but more serious, problem is that the
relative prices of capital and consumption goods may under-
value the importance that a society attaches to growth. If so,
either shadow prices must be derived for products, or some allowance
must be made in valuing the factors of production in recognition
of the growth objective. The latter solution in the case of labour is
considered in Appendix 7.3, where a more general formulation is
presented of the valuation of labour in surplus-labour economies
in which emphasis is on growth as well as efficiency.
212 CROWTH AND DEVELOPMENT
Appendix 7. 3
The Planning W agel
The two extreme views on the valuation oflabour in surplus-labour
economies have already been touched on previously. On the one
hand there is the 'traditional' view that if the marginal product of
labour in agriculture is zero, labour ought to be considered 'cost-
less' in the planning process in order for current total output to be
maximised (see Chapter 3). This is also the programming solution
where the objective function is to maximise current income subject
to resource constraints, but where labour is not a scarce resource and
capital- and consumption-good prices are not adjusted to reflect
the social valuation of them (see Appendix 7.2). On the other hand,
there is the opposite 'modern' view that if the wage in industry is
higher than the wage in agriculture, and the propensity to consume
is unity, labour transference will involve an increase in consumption,
so that to maximise growth and output in the future labour ought to
be valued at the industrial wage to maximise the investible surplus
(see Chapter 3).
Both views, representing distinct development o~jectives, can be
summed up in one diagram depicting the industrial sector of the
economy. If, in Fig. C7.1, the marginal product of labour in agri-
culture is zero, maximisation of the total product, L1P, requires that
labour be given a shadow wage of zero so that OL1 workers are
employed. If the propensity to consume out of the industrial wage,
OW, is unity, however, maximisation of the investible surplus,
WRS, requires that labour should be given a shadow wage equal to
the industrial wage. To value labour less than 0 W would encourage
the employment of labour beyond OL and involve consumption
in excess of production which would reduce the size of the investible
surplus and impair future growth. But if there is a trade-off between
present and future welfare, neither solution can be optimal.
1 This appendix is a revised and condensed version of my paper 'The
Valuation of Labour in Surplus Labour Economies: A Synoptic View',
Scottish Journal of Political Economy (Nov. 1971).
PLANNING :&CONOMIC DEVELOPMENT 213
Apart from this, there are also other reasons for taking an inter-
mediate position between these two extreme views on the valuation
of labour. These, too, need consideration in formulating the plan-
ning wage. Some of these considerations have already been men-
tioned earlier in the book. For one thing it is by no means certain
Marginal revenue
product
FIG. C7.1
that the opportunity cost of labour in the agricultural sector is

zero. The process of labour transference from agriculture to in-
dustry may well have production effects in the agricultural sector
which then require explicit consideration in valuing labour.
Secondly, there are likely to be changes in consumption and saving
in the agricultural sector, as well as in the industrial sector, as labour
migrates, which must also be taken into account. Then, considering
the valuation of labour in a planning framework, it cannot be
assumed that the size of the investible surplus for growth is deter-
mined solely by how labour is valued when governments have the
power to tax and to redistribute income. Finally, domestic saving
may not be the dominant restraint on growth in an open developing
economy. The valuation of domestic resources must have regard
to the value of resources obtainable from abroad.
To incorporate these considerations into a more general formu-
lation of the valuation of labour, let us start with the basic growth-
maximising model, assuming zero marginal product in agriculture,
and then proceed by relaxing some of its restrictive assumptions.
A basic model of this type has been presented by Professor Sen,l

who was one of the first economists to react against the traditional
view that labour should be regarded as 'costless' in surplus-labour
economies even though its marginal product may be zero. The
objective is to value labour to maximise the investible surplus.
Taking the industrial and agricultural sectors together, the change
in consumption from labour transference is given by
X= W.c-d(l-c') (07.1)
where X is the change in consumption
W is the industrial wage
c is the propensity to consume of workers in the industrial
sector
d is the consumption of workers disguisedly unemployed in
agriculture
and c' is the propensity to consume of the former hosts of the
unemployed workers in agriculture.
lfaccountis taken ofthe consumption of capitalists, equation (07.1)
becomes
X= W. c- d (1- c') + (Pr- W) c* (07.2)
where Pr is the marginal product of workers in industry
and c* is the propensity to consume of capitalists.
On classical assumptions that c and c' are equal to unity and c* is
zero, the change in consumption is exactly equal to the industrial
wage and this is the real cost oflabour (W*).2 From society's point
of view, therefore, the cost oflabour is no different from the cost to
the private entrepreneur. The point at which saving and invest-
ment are maximised (see Fig. 07.1) is identical to the free market
solution if firms strive to maximise profits (i.e. where W = P1 ).
Now let us relax the assumption that the marginal product of
labour in the agricultural sector is zero. This assumption is un-
1 A. K. Sen, Clwice of Techniques, 1st ed. (Oxford: Basil Blackwell, 1960).
2 If c < I or c' < I, the real cost of labour can be less than W, but it is
generally assumed that W* can never be greater than W. W* could be
greater than W if c' > I due to an income elasticity of demand for food in
excess of unity as income per head rose in agriculture as a result of labour
transference, or if c > I to make allowance for the provision of more social
capital in the industrial sector. For present purposes, however, we shall
ignore these possibilities. We shall also assume that the terms of trade
between the two sectors remain unchanged.
necessarily restrictive and hard to reconcile with the various con-
cepts and definitions of disguised unemployment that we gave in
Chapter 3. Disguised unemployment can obtain with either positive
or negative marginal product depending on its definition, and in
practice marginal product may be positive or negative, depending
on how agriculture is organised. On these alternative assumptions,
the production effects, as well as the consumption effects, of labour
transference from agriculture to industry are important in esti-
mating the real cost oflabour, if c' < 1. This is shown in equation
(C7.3) below which gives the change in the aggregate surplus (S')
with respect to the transference of labour from agriculture to
industry:
S' = (PI- W. c) - c* (PI- W) - (PA-d) (1 - c') {C7.3)
where PA is the marginal product of workers in agriculture.
Compare the result from equation (C7.3) with that of equation
(C7.2). On the assumption that c and c' are unity and c* and PA
are zero, the aggregate surplus (S) is maxixnised when PI= W,
which is the earlier result. But if PAis not zero, and 0 < c', c* < I,
which is the general case to consider, Swill be maximised when
(PI- W. c) - c* (PI- W) =(PA-d) (1 - c') (C7.4)
and the real cost of labour is given by
W* = W. c + c* (PI- W) +(PA-d) (I - c').l (C7.5)
The real cost oflabour is greater or less than X in equation (C7.2)
according to whether PA (I - c') ~ 0.2
In short, if the marginal product of labour is not zero, the
production effects of labour transference must be considered (in
conjunction with c') in arriving at the shadow price oflabour.
Now let us relax the assumption of growth maximisation. To
assume maximisation of the investible surplus is to assume that society
attaches no value to extra consumption at the margin. This, in
effect, implies a social discount rate of zero which not even the most
1 Note that if P.t < d (which embraces all definitions of disguised un-
employment), it will always pay to induce labour into the industrial sector
by reducing its shadow price below the value of extra consumption generated
in the industrial sector.
2 If c' = 1, it obviously does not matter whether PA is zero, positive or
negative; the same result is arrived at as in equation (C7.2).
austere planned economies have ever adopted. Some adjustment is

required, therefore, to take account of society's trade-off between
investment and consumption, or consumption today and con-
sumption tomorrow. A simple way of valuing investment relative
to consumption, which fits in with Sen's time horizon approach to
the choice of techniques (see p. 175), is suggested by Little and
Mirrlees. 1 Suppose within a certain time period acceptable to
society n million pounds worth of consumption arises from q million
pounds worth ofinvestment today, where n is the sum of the annual
flows of consumption over the time period discounted by the rate
of time preference or consumption rate of interest (r). In this case
the benefit arising from the consumption enjoyed as a result of
employing one more man is qjn of the benefit that would result from
investing an equal sum q. The man's consumption is worth 1/So of
the same amount of investment, where So is the value of investment
relative to consumption, i.e. njq.2 This relative valuation term can
now be incorporated into the previous analysis. Let C be the total
reduction in saving from employing one extra man. We know from
equation (C7.5) that employing the man reduces savings by the
increase in consumption plus the reduction in agricultural output
(PA) so that consumption increases by (C - P A). Hence the real
cost of labour, as modified by the valuation placed on investment
relative to present consumption, becomes
I
W* = C- So (C-PA)· (C7.6)
The higher the valuation given to saving and investment compared

to consumption (i.e. the higher S0), the higher will be the real cost
of labour and the closer the shadow wage will approximate to the
industrial wage if all wages are consumed. In the extreme case of no
value attached to consumption at the margin, then from (C7.6) the
real cost of labour is equal to C, i.e. the increase in consumption
plus PA, which is the result derived in equation (C7.5) assuming
growth maximisation. If at the other extreme saving and consump-
1 I. M. D. Little and J. Mirrlees, Manual of Industrial Project Anarysis in
Developing Countries, vol. II: Social Cost-Benefit Anarysis (Paris: O.E.C.D., 1969).
2 The higher the consumption rate of interest (r), the lower is n. The
faster consumption is growing, the higher r will be. n will also depend on the
time horizon over which the flows of consumption are calculated. The lower
is n, the lower is the shadow wage because the less the value of investment in
terms of consumption sacrificed.
tion are valued equally at the margin so that So = 1, the real cost of
labour is simply equal to PA - the marginal product of labour -
which can be zero, positive or negative. This is the static efficiency,
or linear programming, result with relative prices of capital and
consumption goods correctly reflecting social utilities. In general,
the poorer the country, the lower its investment ratio, and the slower
consumption is expanding, the greater the weight that will be
attached by planners to investment and the higher the shadow wage,
discouraging labour-intensive techniques. If the shadow wage lies
below the market wage because some value is attached to consump-
tion per se, an optimal solution to the choice of technology requires
some form of wage subsidy. In the limit, with no value placed on
investment at the margin, the entire wage would consist of subsidy.
Planning in a free market would entail payment of the subsidy to
private enterprise. Planning in a command economy could actually
use the shadow wage derived for choosing the range of enterprises
established.
We mentioned at the beginning that one of the implicit assump·
tions underlying the emphasis on obtaining the 'correct' shadow
wage is that the investible surplus is a function of the choice of
technique, which presupposes that the planning authorities can
regulate the choice of technique but lack adequate policy instru-
ments for varying the savings rate by other means. In reality, of
course, this is far from the truth. Governments can and do tax and
redistribute income for reasons of equity, efficiency and growth so
that in practice tax policy can be a substitute for, or used in con-
junction with, shadow wages to achieve specific development goals.
Wage subsidies themselves would require financing. With policy
instruments in the form of subsidies and taxes we must henceforth
refer explicitly to the shadow wage as the planning wage, recognising
that the shadow wage is not simply something determined by 'the
system' (i.e. by the societal objective to be maximised), but is an
instrumental variable capable of manipulation for the reconciliation
of conflicting objectives.
To illustrate the point, the relaxation of the no-tax assumption
clearly lessens the conflict between those who put a premium on
maximising present employment and consumption and those who
emphasise maximisation of the investible surplus for future welfare.
The conflict is reduced to the extent that if the planning wage is
fixed below the level that would maximise the investible surplus
for the sake of increasing employment and consumption (e.g.

through wage subsidies), real saving can be prevented from falling
(and possibly increased) by a redistribution of income from low
savers to high savers.
This raises the broader question of how the planning wage might
be adjusted by the scope for government to tax to reduce real
consumption out of the real wage, and to save and invest on society's
behalf, without the uncertainty involved in redistributing income.
The government has two main weapons at its disposal. It can
either tax consumption directly by the use of commodity taxation, or
indirectly by taxes on income which reduce disposable income. To
make the analysis simple for illustrative purposes, we shall confine
ourselves to taxation in the industrial sector which is where the major
part of increased consumption occurs with labour transference.
The scope for taxation to reduce real consumption depends on two
interrelated factors: firstly, on the size of the differential required
between the industrial wage and the subsistence wage to ensure an
elastic supply oflabour to the industrial sector; and secondly, on the
reaction of workers to a cut in their real disposable income. We noted
earlier in Chapter 3 that in recent years there has been a marked
tendency for the differential between earnings in the subsistence
agricultural sector and the modern industrial sector to widen. Owing
to the growing income differential, many formerly in disguised
unemployment in the rural sector have transferred into visible un-
employment in the modern urban sector. The increasing income
differential may be due to genuine productivity differences between
the two sectors or partly the result oflabour unions in the industrial
sector pushing up money wages in excess of productivity increases.
In certain Latin American countries such as Chile, Brazil and
Colombia, Meierl suggests that there is considerable evidence that
union pressun~ and minimum wage legislation have raised the
wage above the opportunity cost of labour, which has also induced
the introduction of more capital-intensive techniques. Government
policies have generally supported urban wage increases, and wages
have risen very rapidly in the government sector itself.
Prima facie evidence of the power of monopolistic elements in
the industrial labour market would be an increase in earnings
differentials between the modern and subsistence sectors, coupled
1 G. Meier, 'Development without Employment', Banta Nazionale del
Lavoro Quarter(y Review (Sep 1969).
with a movement in the terms of trade in favour of industrial com-
modities. Secondary evidence of the power of labour unions would
be a rise in the differential between the two sectors coupled with
growing unemployment in the industrial sector itself. There is
evidence of both of these tendencies. In many less developed
countries in recent years, wages have risen by up to 5 per cent
per annum in excess of productivity, and the price of industrial
goods has risen faster than the price of agricultural goods within
less developed countries (as well as internationally). Furthermore,
many countries have witnessed for some time growing unemploy-
ment in urban centres. In short, there may be scope for the real
wage in the industrial sector to be reduced by taxation without the
supply of labour drying up, given some control over monopoly
elements in the labour market.
If the existing differential between the industrial and the sub-
sistence wage is more than sufficient to ensure an elastic supply of
labour to manufacturers in the industrial sector and there is no
reaction among workers to a cut in their real disposable income, it
would be advantageous to reduce the planning wage by the product
of the amount of the excess differential and the propensity to con-
sume, and to tax equivalently. On the assumption that there is no
reaction to the cut in real disposable income, the real cost of labour is
I
W* = (C- cT)- So [C-PA] (C7.7)
where Tis total tax revenue.
Suppose, however, that there is no excess differential between the
industrial and subsistence wage. In this event, whether the planning
wage can be lowered by a policy of taxation, without sacrificing
growth, depends on the reaction of workers to a change in their real
disposable income. In the absence of money illusion, workers will
resist both a reduction in their money wage through direct taxation
and also a reduction in their real wage through commodity taxation.
If money illusion exists, commodity taxation is a possibility until
workers become aware of the cut in their real consumption. Given
the valuation of investment relative to consumption, the change in
the planning wage is equal to
6.W* = (6.W- c6.T). (C7.8)
For a reduction in the planning wage the term (6. W - c6. T) must
be negative, which will depend on whether workers react completely
to a reduction in their disposable income through an increase in T

by bidding for a higher money wage.
The Valuation of Labour in an Open Econom;vl

In an open economy a shadow wage which is calculated with refer-
ence to the reduction in saving caused by the employment of an
additional unit of labour may not be 'optimal' having regard to a
country's foreign exchange position. In an open economy dual-gap
analysis, and the theory of the dominant restraint, become relevant,
and the scope must be considered for using wages as a policy instru-
ment for equating the two gaps ex ante. The two gaps referred to
are the savings-investment gap and the export-import, or foreign
exchange, gap. If domestic saving is calculated to be less than the
level necessary to achieve the target rate of growth, there is said to
exist a savings-investment gap. It is essentially the minimisation of
this gap that we have been concerned with so far. In an open
economy, however, there may also be a foreign exchange gap if
minimum import requirements to achieve the growth target are
calculated to be greater than the maximum feasible level of exports.
In the absence of foreign borrowing, growth will proceed at the
highest rate permitted by the most limiting factor. If the largest gap
is the foreign exchange gap, growth is said to be trade-limited and
domestic saving may go unused despite a 'shortage'.
Whether or not the foreign exchange gap is the dominant re-
straint, a shortage of foreign exchange obviously has implications
for the valuation of resources used in investment projects. If there is
excess demand for foreign exchange at the ruling price, foreign goods
must be worth more to the economy than their domestic price
suggests. A shortage of foreign exchange implies that the foreign
exchange rate is overvalued and foreign goods are too cheap relative
to domestic goods. Foreign resources used in investment projects,
if valued in the home currency, will understate their cost to the
economy relative to the use of domestic resources. One solution is
to adopt a shadow price of foreign exchange. An alternative solu-
tion, adopted by Little and Mirrlees,2 is to make domestic and
1 Before reading this section, readers are advised to look first at the section
on dual-gap analysis in Chapter 8.
2 Little and Mirrlees, Manual of Industrial Project Anarysis, vol. u.
foreign resources comparable by valuing all inputs and outputs at
world prices.
A further way to economise on the use of foreign exchange would
be to encourage the use of'cheaper' domestic resources by reducing
the planning wage. Before recommending this as a policy measure,
however, let us follow through its implications. The first reper-
cussion will be to cheapen domestic resources relative to resources
from abroad, thereby releasing foreign exchange. On the other hand,
a lower planning wage will increase the level of employment, which
will increase consumption and reduce the level of domestic saving
if the planning wage was previously fixed to maximise the investible
surplus consistent with time preference. The case for lowering the
planning wage when foreign exchange is the dominant restraint
must therefore rest on the assumption that additional increments of
foreign exchange are more useful than the domestic saving sacrificed;
that is, that either some domestic saving would be redundant if the
foreign exchange bottleneck is not overcome, or that more resources
are released for investment purposes through the release of foreign
exchange than are used up in consumption through reducing the
planning wage. A further repercussion of reducing the planning
wage and increasing the labour intensity of investment projects
is that imports of consumption goods may be expected to rise, which
will absorb foreign exchange and offset some of the savings offoreign
exchange owing to the switch in demand from foreign to domestic
resources.
In considering a policy of reducing the planning wage to equate,
ex ante, the savings-investment gap and the foreign exchange gap,
two considerations must therefore be borne in mind. The first is
whether the foreign exchange gap is truly the dominant restraint,
and the second is the net saving of foreign exchange that can be
expected from the policy.
As we show in Chapter 8, the theory of the dominant restraint
has recently been called into question by many economists. On the
one hand it has been argued that if capital and imports are sub-
stitutable for one another there can only be one gap, not two, and
on the other hand it has been argued that if the capital inflow
required to fill the foreign exchange gap is larger than the savings-
investment gap, this need not imply, as traditional theory argues,
that savings will fall below the saving potential of the community
or that less productive investment will take place. On the second
point, which is the one more pertinent to the present discussion,

there are plenty of types of productive investment, e.g. investment
in human resources, that do not require imported capital goods.
So long as this type of investment can be expanded, the existence of
a foreign exchange gap should not prevent the translation of poten-
tial savings into investment, and the possibility of excess saving
disappears. But if this is the case, it is also difficult to argue that
foreign exchange is genuinely more valuable than the sacrifice of
domestic saving that a reduction in the planning wage would
entail. In short, the savings-investment gap, as traditionally cal-
culated from the Harrod growth equation, may be understated
owing to scope for types of investment with large spillovers to the
community.
If it can be demonstrated that domestic saving is just as valuable
as foreign exchange, and that there are alternative outlets for domes-
tic saving which do not require capital inputs from abroad which
may yield returns equal to the returns from releasing foreign ex-
change by using cheaper domestic resources in investment projects,
the foreign exchange problem must be tackled in ways which do
not reduce domestic savings. This is tantamount to arguing that
there may be only one gap, not two, not in the sense that imports
and domestic capital are perfect substitutes for one another but in
the sense of foreign exchange and domestic saving being equally
valuable, such that a policy measure which reduces one gap at
the expense of another is pointless.
But suppose foreign exchange is genuinely more valuable than
domestic saving. The extent of the reduction in the planning wage
must depend on the net release of foreign exchange brought about.
If the foreign exchange gap is the dominant restraint, the general
proposition can be advanced that a planning wage should be set
which equates the two gaps ex ante, provided the loss of saving is
compensated by a release of foreign exchange equal to the net
release of foreign exchange times the valuation of foreign exchange
relative to saving. Ignoring the valuation of investment relative to
consumption, the change in the planning wage will be equal to
aW* = ( C- F.:~) (07.9)
where C is the total reduction in saving

F is the net release of foreign exchange
and Wp is the ratio of weights attached to foreign exchange and
ws
domestic saving.
The term ( C - F . ::) gives the scope for changing the planning
wage to equate the savings-investment and foreign exchange gaps
ex ante. For a reduction in the planning wage the term must be
F.::) =
negative, and it would pay to reduce the planning wage until
( C- 0. The expression will approach zero as Crises
and Wp falls. If no foreign exchange is released because increased

ws
consumption-good imports match the decrease in foreign resources
required for development, the expression will be positive and there
is no scope for reducing the planning wage since saving would
merely fall with no offsetting benefits. In this situation, measures
such as import controls on consumption goods and devaluation may
have to be implemented to remedy the foreign exchange shortage.
8
Financing Development
Real capital formation from domestic resources involves two basic
steps. Firstly, it requires an increase in the volume of real saving, and
secondly, the act of investment itsel£ In the absence of international
trade and foreign borrowing, capital formation is only possible
through abstinence from present consumption and when society
produces a surplus of consumer goods sufficient to meet the needs of
labour engaged in producing capital. In a purely subsistence eco-
nomy, saving and investment are simultaneous acts in the sense that
if a producer is to develop the means of production he must sacrifice
time and resources which would otherwise be used for consumption
purposes. In a money economy, savings and investment may be
undertaken by different groups, and the process of capital formation
is likely to require some form of finance and credit mechanism to
channel savings from willing lenders to willing borrowers. In fact,
in the early stages of development savings may not be the major
barrier to capital formation but rather the unwillingness or inability
to invest. The unwillingness to invest may stem from cultural atti-
tudes or simply a realistic assessment of the risks involved. The in-
ability to invest, on the other hand, may result from shortages of
factors of production and supplies necessary for particular types of
capital formation that would be profitable. In short, in the transition
from the subsistence to the money economy, the main obstacle to
capital formation may not be a shortage of saving but bottlenecks in
the productive system which do not permit investment or make it
too risky. One of the ways of overcoming these bottlenecks is through
international trade and foreign borrowing. Owing to these domestic
bottlenecks it is often claimed (and found to be the case empirically)
that the import-export gap (i.e. the need for imports in relation
to the capacity to export) is likely to be greater than the savings-
investment gap in the early stages of development.
FINANCING DEVELOPMENT 225
For the moment, however, let us take for granted that the willing-
ness and ability to invest exists, and consider the supply of savings to
less developed countries from both internal and external sources. In
this respect we must not only consider the sectors from which saving
originates but also how savings may be stimulated, especially
through the deliberate policy actions of government. This will in-
volve a brief consideration of fiscal and monetary policy in less
developed countries, including the role of deficit finance and infla-
tion in providing real resources for capital formation.
As far as domestic saving is concerned, three broad 'types' may
be distinguished: voluntary; involuntary; and savings generated
through policies to increase output in situations of unemployed or
underemployed resources. The origin of these 'types' of saving
should be fairly self-explanatory. Voluntary savings are savings that
arise through voluntary reductions in consumption out of disposable
income. Both the household and the business sectors may be a source
of voluntary savings. By contrast, involuntary savings (or what is
euphemistically called 'collective thrift') are savings brought about
through involuntary reductions in consumption. All forms of taxa-
tion, and schemes for compulsory lending to governments, are
traditional measures involving involuntary reductions in consump-
tion. In addition, consumers may have their capacity to consume
reduced by the process of inflation (i.e. rising prices), and this is
often referred to as 'forced' savings. For a variety of reasons, which
will be considered later, inflation is likely to be a natural concomitant
of development, but it can be deliberately induced or exacerbated by
policy measures such as monetary expansion and deficit finance at
full employment. Thirdly, deficit finance at less than full employ-
ment can generate saving by activating unemployed or under-
employed resources. An increase in the volume of real saving can
always be brought about by increases in real output, provided the
propensity to consume out of the extra income generated is less
than unity.
Domestic savings for investment can also be supplemented from
abroad. Private foreign investment is a direct source of capital
formation and provides a direct addition to domestic investment. It
can also be a source of savings by stimulating income and employing
previously underutilised resources. Secondly, borrowing from abroad
provides resources for investment by enabling imports to exceed
exports, which in the national accounts shows up as investment in
excess of domestic saving. Foreign assistance may be from multi-

lateral or bilateral sources, and may take a variety of forms ranging
from loans at commercial rates of interest to outright gifts of goods
and services and technical assistance. Lastly, a country's commercial
policy can stimulate savings and release resources for investment
purposes. An improvement in a country's terms of trade can provide
additional resources for investment if the increase in real income
made possible is not fully consumed. Likewise, policies to restrict
imports of consumption goods can release additional resources for
investment, provided that domestic saving is not reduced by the pur-
chasing power released being switched to home consumption goods.
Now let us examine these internal and external sources of saving in
more detail.
A. DOMESTIC RESOURCES FOR INVESTMENT
Voluntary Saving
The level of voluntary saving, and the ratio ofvoluntary saving to
national income, will depend on a variety of economic and non-
economic factors. Economic factors largely determine the ability to
save, but the willingness to save may depend on non-economic
factors as well. The main determinants of the ability to save will be
the average level of per capita disposable income, the distribution of
per capita disposable income and the size of the capitalist surplus.
The willingness to save, in turn, will depend on such monetary
factors as the existence of acceptable and reliable institutions in
which to deposit savings; the interest rate in relation to risk and time
preference; and, in addition, societal attitudes towards the accumu-
lation of capital. Saving in a form which contributes to productivity
growth is not simply a function of the level of income. Countries at
the same stage of development, measured by per capita income, or
some alternative index, may have very different levels of saving
relative to national income because of important differences in these
other determinants of voluntary saving. In Lewis's model of develop-
ment with unlimited supplies oflabour, it is not the absolute level of
per capita income in a country that determines the savings ratio but
the size of the capitalist surplus, and the distribution of income
between entrepreneurial profits and other incomes. This view is
typical of classical models of development and underlies those invest-
ment criteria which imply the choice of capital-intensive technology
in less developed countries. Saving depends on the size of the capital-
ist class, and the share of profits in national income. Lewis himself
believes that in practice no nation is so poor that it could not save
and invest at least 12 per cent of its national income if it so wished.
Investment as a percentage of national income is not small because
of an incapacity to save but because the surplus generated in less
developed countries is used to maintain 'unproductive hoards of
retainers' and to build pyramids, temples, etc. As long as a significant
share of income accrues to landlords, traders and other conspicuous
consumers, there is little chance of providing significant amounts of
voluntary saving. Lewis's view is very much in line with our earlier
remark that in the early stages of development there may be no
savings-investment gap in the sense of an incapacity to save; merely
an unwillingness or inability to use savings for 'productive'l capital
formation.
Other economists have also taken this view by pointing to the
extreme income inequalities within less developed countries. It is a
well-documented fact that for most less developed countries, coupled
with overall poverty, the distribution of income tends to be more
unequal than in developed countries. In the case of India, for
example, where approximately 20 per cent of the population is in
receipt of 60 per cent of the national income, it should not be
difficult to raise saving and investment to 20 per cent of the national
income. The problem is to get those with the capacity to save to
channel their savings into 'productive' investment. If the rich do not
save and invest voluntarily, this adds an economic justification to the
egalitarian argument for heavy taxation of the upper income groups,
forcibly extracting the surplus and achieving a more equal distribu-
tion of income simultaneously.
The fact remains, however, that the ability to save of the vast
majority of people in less developed countries is limited by the
desperately low level of per capita income. The task of financing
capital formation through domestic voluntary means is essentially
twofold, therefore. The first task is to encourage the production of a
1 By 'productive' is meant activities which yield a high stream of measur-
able income in the future. On this definition, temples, pyramids, etc., are
excluded as well as real estate and many types of service activities where the
scope for productivity growth is low compared with investment in manu-
facturing industry.
surplus in those sectors of the economy where it is very small. This

brings us back to the overriding importance of encouraging produc-
tivity growth in the dominant agricultural sector through simple
means requiring very little in the way of investment resources.
Measures of land reform and a slight reorganisation of productive
factors may be all that is required. The second task is to encourage
and exhort those with the ability to save to cut down on extravagant
consumption and to invest productively the surplus of income over
whateverlevel of consumption is decided upon. A narrowing of the
gap between the potential and actual surplus available for invest-
ment depends on discouraging conspicuous consumption. The en-
couragement to save and invest must come from the government
or some organ of government such as the planning agency. If the
government fails in its attempt to generate an increase in the surplus,
and to encourage its utilisation voluntarily, it can (and must in the
interests ofdevelopment) extract the surplus involuntarily, and utilise
the surplus itself for productive purposes. This is where fiscal and
monetary policy becomes important. Taxes are versatile policy
instruments and a tax system can be designed both to provide an
incentive to saving and investment and to extract a surplus forcibly.
The primary aim of fiscal policy in less developed countries must be
to raise the savings ratio by keeping the marginal propensity to save
above the average, without simply substituting public savings and
investment for personal savings and investment.
Monetary measures can also be taken to foster savings and invest-
ment, such as floating government securities at attractive interest
rates (with possible tax exemption on the full value of the bond), and
establishing institutions to encourage the deposit of savings, reducing
at the same time the dependence on the extortionate money-lender
for borrowing.
Involuntary Saving
Involuntary saving is saving undertaken by the government on
society's behalf, largely through taxation, deficit finance, and
induced inflation. Let us consider these three policy instruments in
turn.
Taxation
The ratio of total tax receipts to national income is typically low in
less developed countries, ranging from 5 to 20 per cent - compared
with 25 per cent and over in most advanced countries. On the sur-
face there would seem to be plenty of scope for increasing taxation to
raise the level ofsaving. How much additional real investment can be
financed by taxation, however, depends on the type of taxation im-
posed and on whom. If additional taxation taps funds which would
either have been consumed or gone into 'non-productive' channels,
then government taxation can certainly finance a greater volume of
real investment. If the additional tax burden is imposed on the rich,
however, they may decide to cut their voluntary saving to meet the tax
demands rather than cut down conspicuous consumption or specula-
tive activities, in which case the increase in real investment may be
minimal. In these circumstances the finance of investment through
taxation is likely to be inflationary through the redistribution of in-
come between those with very low propensities to consume and those
with a very high propensity to consume for the nation as a whole.
The most effective tax policy to finance real investment would be to
impose all taxes on those with high marginal propensities to con-
sume, i.e. the poor. But there are obvious difficulties and objections
in carrying this policy to the extreme, not the least of which might be
a reduction in work effort and output due to disincentive and mal-
nutrition. In the last resort, a balance must be struck between the
efficiency and equity of a tax system in achieving its objectives.
Having said this, however, it is in practice the poorer sections of the
community in less developed countries that do bear the brunt of the
tax burden owing to the heavy emphasis on indirect taxation and
taxes on the agricultural sector. Indirect taxation, as opposed to
direct taxes on income, is preferred in less developed countries for
many reasons. For one thing it is virtually impossible to assess the
incomes of the vast majority of people in a typical underdeveloped
country; but even if it were possible, a degree of literacy would be
required for simple form-filling which cannot yet be guaranteed.
Moreover, in countries lacking good communications, and honest
and efficient local administrators, the scope for evading income
taxation is so enormous as to make it impracticable, and certainly
unfair, to collect. The beauty of indirect taxation, such as taxes on
commodities and on exports and imports, is that they are difficult to
evade and relatively easy to collect. On the other hand most indirect
taxes are regressive, hitting hardest those who can least afford to pay.
Taxes impinge predominantly on the agricultural sector simply
because this sector happens to be the largest in less developed
countries, and more often than not contributes the most to export
earnings. The ideal tax system for the agricultural sector is one that
will not only extract any surplus that is available but which will also
act as an incentive for producers to produce and to export more. But
again the clash between equity and efficiency arises. One obvious
tax, for instance, is a poll tax or land tax which would compel in-
creases in output to meet the tax bill. Lump-sum taxes are extremely
inequitable, however, ignoring the capacity to pay and such welfare
principles as equimarginal sacrifice. Another interesting possibility
(which has been experimented with in some parts of Latin America)
is a tax on the productive potential ofland to deter the waste of good
land and to encourage improved methods ofcultivation. Since the tax
would have to be paid whether production took place or not, or
whatever the level of efficiency, it must pay to cultivate the land as
productively as possible. The major drawback of the tax is the
difficulty of estimating productive potential.
The alternative to taxing the land, or those who work on it, is to
tax the produce of the land With a system of marketing boards for
the distribution of agricultural produce, this type of taxation is
extremely common on goods marketed both domestically and for
export. As far as exports are concerned two main systems are nor-
mally adopted. Either the state-controlled marketing board pays
the producer a price less than the international price received, or
alternatively the government requires that all foreign exchange
receipts be surrendered, with compensation given in local currency
at a rate which overvalues the local currency. The two methods are
virtually identical in effect, differing only in administrative detail.
They both provide an easy and convenient means of taxing the
agricultural sector.
In formulating the tax system, however, the policy-maker must
bear in mind the income and substitution effects of taxation. The
production of goods for export will be curtailed if the substitution
effects of the taxes outweigh the income effects. The income effects
of taxation are to encourage production, in the same way that a poll
tax or any tax which reduces income encourages production and
effort. But the substitution effect operates to discourage production,
or at least to switch production to the home market unless produc-
tion for the home market is taxed at the same rate.
Types of taxation, other than commodity taxation, are largely in-
appropriate for the agricultural sector in the early stages of develop-
ment. There seems no reason, however, why experiments should not
be made with a whole range of tax instruments in the industrial,
modern sector, e.g. income taxes, consumption taxes, wealth taxes,
etc., where the level ofliteracy and administrative efficiency is higher,
and where the assessment of income, expenditure and wealth is
relatively easier, given the existence of wage-payment systems and
the rudiments of a banking system. And, of course, in most less
developed countries there do exist taxes on the incomes of certain
groups: taxes on business enterprises including foreign companies,
and also a wide range of consumption taxes sometimes steeply and
increasingly progressive on luxury items.
With the desire of the rich for conspicuous consumption and the
government's desire to increase its tax revenue as a percentage of
national income, progressive consumption taxes can perform the
valuable dual role of discouraging the consumption ofluxury goods
(often with a high import content) and increasing the elasticity of
tax revenue with respect to national income. A tax on any com-
modity with an income elasticity of demand greater than unity will
ensure that tax revenue from the commodity as a percentage of
national income will rise whether the tax is progressive or not. This
is highly desirable if a government wants to undertake an increasing
proportion of total capital formation itself. Even if the imposition of
consumption taxes leads only to a small increase in real resources for
investment owing to the low propensity to consume of those who
consume the goods that are taxed, the government can at least gain
greater control over the total amount of investible resources and
perhaps divert resources into more socially useful channels than
would otherwise be the case. This must be the main justification for
mass commodity taxation in less developed countries - not so much
to raise the rate of investment in the short run as to prevent con-
sumption rising as fast as income over the long run. This is a par-
ticularly important consideration if there is migration from the rural
to the industrial sector, when consumption might be expected to rise
proportionately with increases in income. In general, direct taxes
should be levied on commodities that have at least one of the
following characteristics: a low price elasticity of demand; a high
income elasticity of demand; that absorb a large amount offoreign

exchange or scarce resources needed in the public sector; and that
are consumed predominantly by the rich. Taxes on commodities
with these characteristics will raise revenue; increase revenue faster
than income; save foreign exchange; release scarce resources for the
public sector; and be in accordance with ability to pay.
No universal tax policy can be prescribed to suit all countries.
The efficiency of different taxes to raise the level of domestic savings
without offending equity and discouraging growth will vary from
country to country according to a host of social and economic cir-
cumstances. What seems vitally important, however, is that if a tax
system is to be accepted by a poor community, it must be seen to
be administered honestly and efficiently, which means that every
attempt must be made to minimise the scope for evasion. On this
point attention may be drawn to Kaldor's tax plan for India and to
Higgins's outline for a self-enforcing tax system which builds on
Kaldor's framework.l The objective is to formulate the tax system in
such a way that efforts to evade one tax will automatically involve
the evader in even heavier tax liabilities so that evasion is not worth
while. Kaldor's scheme included a personal income tax, a wealth
tax, an expenditure tax and a gift tax, but falls short of being a
closed system since it does not prevent evasion in the case where both
parties to a transaction would benefit by concealing or understating
the amount ofthe transaction, e.g. in the sale of goods. Higgins adds
to Kaldor's system a penal tax on excess inventories, and a turnover
tax, so that anyone failing to report one taxable transaction either
finds himself confronted with a higher tax bill or having the trans-
action reported by the other party because the other party can reduce
his tax liability by doing so. Higgins's scheme also possesses built-in
incentive features so that for any taxpayer with a given income the
total tax burden on that income is minimised by maximising the
share of income spent on productive equipment. Enough has been
said, however, to indicate that the practical difficulties of designing
and administering a tax system which strikes a balance between the
different branches of the budget are immense; and it is doubly
difficult to extract savings involuntarily in the most efficacious way
without conflicting with distributive and allocative goals. Deficit
1 N. Kaldor, Indian Tax Refqrm (Delhi: Ministry of Finance, 1956), and
discussion by B. Higgins in his text, Ecqnomic Develqpment (New York:
Norton, 1959).
finance and inflation appear on the surface attractive alternatives
for achieving the same end, and we now turn to the role of deficit
finance and inflation in the development process.
Deficit Finance
The main justification for deficit finance in less developed countries,
as in any other country, must be the existence of unemployed re-
sources due to deficient demand. If resources are unemployed or
underutilised, real output and real savings can be increased by
governments running budget deficits financed either by printing
money or issuing government bonds to the banking system and the
public. Although deficit finance is likely to be inflationary in the
short run, there is an important analytical distinction between the
means by which additional resources are made available for invest-
ment through deficit finance and the means by which savings are
generated by inflation. In the former case savings are generated from
the increase in real output; in the latter case, by a reduction in real
consumption which may result from a combination of three factors:
money illusion, the inability to maintain real expenditure, and in-
come redistribution from low savers to high savers.
In a situation of genuine 'Keynesian' unemployment, any ten-
dency towards inflation, whatever method of deficit finance is used,
should burn itself out as the supply of goods rises to meet the addi-
tional purchasing power created. Some economists have questioned,
however, whether the observed unemployment of labour in less
developed countries is strictly of the Keynesian variety, and whether
the supply of output would respond very much to increased demand.
An early exponent of this view was Professor Rao, who disputes the
secondary repercussions on output as visualised by the multiplier
process:' ... the secondary, tertiary and other increases in income,
output and employment visualised by the multiplier principle do not
follow, even though the marginal propensity to consume is very high.
This is because the consumption goods industries to which the in-
creased demand is directed are not in a position to expand output
and offer effective additional employment.'! Professor Reddaway
1 V. K. Rao, 'Investment, Income and the Multiplier in an Under-
developed Economy', in A. N. Agarwala and S. P. Singh (eds), The Economics
of Underdevelopment (Oxford University Press, 1958) p. 208, first published in
the Indian Economic Review (Feb 1952).
reiterates this view by questioning whether putting unlimited supplies

of labour to work will have the significant real income multiplier
repercussions that Keynesian theory suggests. Referring to road
building he says:
The secondary consequences of this, in the shape of additional
demand for consumer goods, mainly serve to increase the demand
for food; and the output of food is not limited by the lack of
demand, even though there is plenty of underemployed labour
attached to agriculture. The extra demand consequently falls on
the balance of payments [and/or] it serves to drive up prices.
There is growth in the national income corresponding with the
work done on the road building, but there is little or no multiplier
effect.l
While there is some element of truth in these statements, there are
arguments to be put on the other side. Some deficit-financed projects
may have enormous secondary repercussions on output if they
eliminate production bottlenecks at the same time. To take Redda-
way's example of road building: if transport facilities open up
previously uncultivable areas, the capacity to produce more food is
immediately increased. Secondly, the capital-output ratio may fall
both in response to a change in relative factor prices and the
possibility of increasing the utilisation of capital. Reddaway himself
talks of the underutilisation of capital, as well as of labour, and
admits that not all of the extra demand generated by deficit finance
will necessarily be spent on food. Furthermore, neither Reddaway
nor Rao denies a primary impact on real income from deficit
finance.
But there are other objections to deficit finance than the presumed
absence of secondary multiplier repercussions. Policies of deficit
finance are usually urged in connection with unemployed resources
in the agricultural sector. Suppose labour's marginal product in
agriculture is not zero, however, and the propensity to consume is
very high; what hope is there ofincreasing the volume of real saving
through budget deficits to transfer labour from agriculture to other
activities? If anything, output will fall in the short term and con-
sumption rise - a point we made earlier in connection with the
valuation oflabour in surplus-labour economies. Added to this is the
1 W. Reddaway, 'The Economics of Underdeveloped Countries',
Economic Journal (Mar 1963) pp. 8-9.
fact that people transferred from the land will need to be provided
with a minimum of social capital, putting additional strain on the
investible resources available. Unless resources are genuinely un-
employed, therefore, and the propensity to consume is less than
unity, the prospects of generating much additional saving through
deficit finance appear remote.
Secondly, deficit finance may well become inflationary whether or
not there are unemployed resources and secondary multiplier reper-
cussions. There is bound to be some inflation in the short run, but
there is a very real danger that deficit finance will not proceed with
caution and that inflation will get out of hand. If the government
does not deliberately divert part of the increased output itself (by
taxation), and private investment absorbs all the new savings
created, the amount of deficit-financed expenditure will probably
continue to grow. There is plenty of empirical evidence, in developed
and less developed countries alike, for the 'law of ever-increasing
state activity', and tax revenue is notoriously inelastic in most less
developed countries.l When underemployed resources have been
eliminated, the government may continue to exert an inflationary
impact through a desire to expand its own activities. Unless the
expenditure is completely financed by reductions in private ex-
penditure, deficit finance becomes akin to inflationary finance.
Aggregate policies to cure essentially sectoral difficulties suffer
from the further disadvantage that it is difficult to prevent their
impact from spreading into sectors that are in equilibrium or ex-
periencing opposite tendencies to those which the policies are
designed to remedy. In countries characterised by structural dis-
equilibrium, and imbalances between markets, macro-policies to
cure sectoral unemployment are likely to exacerbate bottlenecks in
other sectors, leading to balance of payments difficulties and inflation
of the 'structural' type.
In short, policies of deficit finance to increase the supply of real
savings in a typical less developed country, and policies that permit
or deliberately induce inflation, may not be so distinct after all,
especially if deficit finance cannot be relied upon to increase output
and short-term inflation does not dissipate itself. If this is so, the
major hope of increasing the supply of real savings lies in a cut in real
consumption through rising prices.
1 A. Prest, Public Finance in Underdeveloped Countries (London: Weidenfeld &
Nicolson, 1962).
Inflation
An increase in the volume of real saving brought about by rising
prices is often referred to as 'forced' saving. The significance of the
term 'forced' is that if consumers cannot maintain their level of real
expenditure because of rising prices, real consumption must go down
and resources are 'forcibly' released for investment purposes. In
practice, though, it may be possible for consumers to draw on savings
to maintain their real standard of living and at the same time to
bargain for higher money wages to match price increases. There
is little evidence that people in less developed countries, in their
capacity as consumers or workers, suffer from money illusion. Thus
whether rising prices can and do increase the volume of real saving
significantly, and release resources for investment, in the way
suggested by the term 'forced', is very much an open, empirical
question.
In the absence of unemployed resources and money illusion, the
only way that savings can be increased through rising prices is by re-
distributing income from groups with high propensities to consume
(low propensities to save) to groups with lower propensities to con-
sume (higher propensities to save). In short, if increases in real output
and income are impossible because the economy is operating at full
employment, and people are able to compensate fully for reductions
in living standards by drawing on voluntary savings, the importance
of inflation for capital formation rests on its ability to shift real in-
come from the poor to the rich or from wage-earners to profit-
earners. Capital formation may be at a high social cost.
We can illustrate these propositions using Maynard's! model of
the 'efficiency' of inflation- the 'efficiency' of inflation being the
amount of inflation necessary to raise saving as a proportion of
national income by a given amount. The efficiency of inflation is
.
g1ven by (e) = 3PJP where p.1s pnces,
'aSJO' . S.1s savmgs
. an d 0 1s
. mcome,
.
which can be calculated given the relevant data.

Let us assume that real income cannot increase in the short run so
that rises in money income equal rises in the price level, and that the
government wishes to raise the savings-investment ratio from its
initial level (a) to a higher level (x). Letting b be the average
1 G. Maynard, Economic Development and the Price Level (London: Macmillan,
1962).
propensity to consume (so that a = l - b), and b' be the marginal
propensity to consume, a formula for the rise in money national in-
come (or prices) may then be derived:
w = u + ac (8.1)
W = O(x- a) + i30x + i30b' (8.2)1
W _ O(x- a)
- I - b'- x· (8.3)
ao has a finite value if b' + X < I. In conditions of full employ-

ment this condition is not likely to hold unless changes in the
distribution of income produce a fall in the average propensity to
consume. This follows from the fact that since a +b = I, and x > a,
b' must be less than b if b' +xis to be less than unity. In short, the
marginal propensity to consume must be lower than the average
which, in conditions offull employment and in the absence of money
illusion, can only come about through a redistribution of income
from low savers to high savers.
The efficiency of inflation is calculated as follows: from equation
(8.3), the proportionate rise in money income and prices is
i30 X- a
0 = l - b'- X
(8.4)
and from our earlier formula the efficiency of inflation is therefore
e-
_(t ~ {; ~ x) -_ I
' . (8.5)
(x- a) I - b - x
The greater l - b' - x, the greater the efficiency of inflation;

that is, the less the proportionate rise in prices for any given target
increase in the savings-investment ratio. The maximum efficiency
will be when the marginal propensity to consume (b') is zero, and the
minimum efficiency when b' +
x = I. As we have seen, however, in
conditions of static real income b' is hardly likely to fall below b. It is
likely that b'+ x will approach unity, which means that there can
be no increase in real saving (except in the unlikely case of money
illusion) unless the propensity to consume falls through a redistribu-
tion of income from high spenders to low spenders. In the full
1 The term i30x is the consequential change in spending to keep invest-
ment a constant proportion of income following the initial increase in

investment, O(x - a).
employment case, inflation cannot be successful in 'forcing' saving

in the absence of redistribution.
Now let us consider, therefore, the efficiency of inflation in a two-
sector model of wage-earners and profit-earners, in which it is
assumed that wage-earners have a higher propensity to consume
than profit-earners. Again full employment is assumed. The rate of
change of money income and prices equivalent to equation (8.4) is
30 X- a
0 = I - [ry(u - v) + v + x]' (8.6)
Hence the efficiency of inflation is

(x - a)
I - [ry(u - v) + v + x]
e= (8.7)
(x - a)
1
= 1- [ry(u - v) + v + x] (8.8)
where u is the marginal (=average) propensity to consume of

wage-earners
v is the marginal ( = average) propensity to consume of
profit-earners
r is the wage-price coefficient (i.e. the proportionate change
in wages divided by the proportionate change in prices)
and y is the initial ratio of wages to national income.
The size of r determines the success with which inflation brings about
the necessary redistribution of income. Clearly, if r = 1 there can be
no redistribution. (u - v) determines the success of income re-
distribution in bringing about a rise in savings. If r and v are both
low, a substantial rise in real investment can be brought about with
very little rise in prices. For example, ifr = 0·1 and (u- v) = 0-4,
then increasing investment from 8 to 9 per cent of national income
would entail price rises of as little as 4 per cent. On the other hand,
the value of the wage-price coefficient may be such that a new
equilibrium of prices and incomes is not attainable (given values of
the other variables); or a new equilibrium is attainable but prices
would have to rise by a considerable amount before sufficient real
saving had been 'forced' to finance the desired level of investment.
The efficiency of inflation depends crucially on the values of r, u and
v. In practice, all the parameters are probably fairly high.
We see, then, that the relation between inflation and saving is
highly complex. Many factors will influence the efficiency of an
inflation, some of which cannot easily be incorporated into the
model, and others which it is difficult to know the importance of a
priori. There is the possibility of money illusion to contend with, as
well as the fact that only a proportion of the shift in real income from
consumers to profit-earners can be expected to be saved. Some of
the extra profits will almost certainly be taxed and possibly used for
consumption purposes; some will be privately consumed, and per-
haps a further portion transferred abroad. What value v takes will
partly depend on inflation itsel£
Ignoring for the moment the effect of rising prices on the willing-
ness to save in the household sector, it is the proportion of the increase
in profits that remains as saving that determines how large a shift in
real income via inflation is necessary to permit a given increase in
investment. For example, if only one-third of the increased profits is
saved, the shift in real income will have to be three times greater than
the increase in real investment required. The smaller the shift in
income relative to the increase in saving and investment required is
another way of saying the greater the efficiency of inflation. For the
smaller the shift in income required, the less inflation will be neces-
sary, and the less will be the social hardship entailed by creating
savings in this way by reducing the real consumption of wage-earners.
But suppose rising prices do shift income from groups with low
propensities to save to groups with high propensities to save; there is
still no guarantee that the total volume of community savings will
rise. It has already been noted that voluntary saving in the household
sector may fall under the impact of inflation. The short-run pro-
pensity to consume tends to be much lower than the long-term
average, and if consumers do not suffer from money illusion a portion
of voluntary saving may be expected to be withdrawn to maintain
real expenditure. Furthermore, anticipated price rises tend to create
a preference for present over future consumption, affecting the
future level of voluntary saving. If business saves no more out of
increased profits than the reduction in savings by other sectors, there
will be no net increase in the total volume of saving.
It should also be clear from the discussion so far that the re-
distribution of income from consumers to profit-earners is only per-
missive as far as investment is concerned. There is no guarantee that
the additional savings will be used for socially productive investment.
It is true that rising prices can act as an investment incentive by
raising the rate of return on capital and reducing the real rate of
interest on borrowed money; but inflation, especially rapid and
persistent inflation, can also lead to less socially desirable types of
investment than in periods of relative price stability, or if savings are
in the hands of the government. Much depends on the degree of
inflation and whether it is expected to last. For the purposes of the
argument, socially desirable investment is defined as those types of
investment which raise productivity, increase the demand for labour
and increase the supply of goods to the community at large. Con-
tinuous inflation may instead induce investment in inventories, real
estate, foreign assets and speculative activities in general, where
capital gains and profits are high but from which the social benefits
are minimal because productivity is not increased, real wages remain
depressed, and the supply of goods to the mass of consumers is in-
creased very little, if at all. In contrast, investment in physical capital
becomes relatively unattractive under severe inflationary conditions
compared with these other outlets for saving because there is little
prospect of short-term gains and the replacement cost of plant and
equipment becomes higher and higher.
In conclusion, there is plenty of room for disagreement over
whether inflation is a help or a hindrance to development. Some
argue that it can help to raise the level of real saving and encourage
investment, while others maintain that it is liable to stimulate the
'wrong' type of investment and that inflation may become endemic
and impair development, especially through its adverse repercussions
on the balance of payments. What is badly needed is a substantial
body of empirical evidence (both cross-section and time-series) of the
relation between inflation and some of the key variables in the
development process, especially savings, as well as between inflation
and development itself. Until more positive evidence is forthcoming,
any conclusions about the role of inflation in the development pro-
cess must necessarily be tentative and speculative.!
The Causes of Inflation in Less Developed Countries
Whether or not inflation provides an additional source of real sav-
ings, it is a fact oflife in most less developed countries, especially in
Latin America. A certain amount of inflation is perhaps inevitable
in any country, if only for institutional reasons such as the existence
1 For some preliminary results, see A. P. Thirlwall and C. Barton,
'Inflation and Growth: The International Evidence', Banca Nazionale del
Lavoro Quarterly Review, (Sep 1971).
of national or local monopolistic organisations in the factor and
product markets, but more especially in less developed countries
undergoing rapid structural change. The fact that growth tends to be
unbalanced (which is how the price mechanism works), and supply
inelastic, means that bottlenecks are bound to arise in the process of
rapid transformation, causing prices to rise in the sectors affected.
As price increases in the shortage sectors transmit themselves to other
sectors, the economy becomes caught in a price-price spiral. And the
faster the less developed country attempts to grow to achieve a target
rate of growth of per capita income, the more severe the inflation is
likely to become. Apart from the existence ofinternal bottlenecks, and
supply inelasticities, import bottlenecks are also likely to arise owing
to the incapacity to expand exports fast enough to pay for imports.
The consequence will be one of two things, both of which are in-
flationary. Either the country will be compelled to develop high-cost
import-substitute activities, or the value of its currency will be
forced down, adding a twist to the price-price spiral by raising im-
port prices. But are these factors the main initiating causes of infla-
tion in rapidly developing economies?
In Latin America, which has experienced the most serious infla-
tion in recent decades, a heated debate developed in the early post-
war years, which still smoulders today, over the primary impetus
behind rapid price increases. Participants in the debate polarised into
two schools, frequently referred to as the 'structuralists' and the
'monetarists'. Although the debate is set in the Latin American
context, it is none the less of general interest and in many ways has
been analogous to the cost-push/demand-pull debate over the causes
of inflation in developed countries. We should also add that the two
debates have been equally sterile and inconclusive.
The essence of the structuralist argument is that price stability can
only be attained through selective and managed policies for economic
growth. It is claimed that the basic forces of inflation are structural
in nature; that inflation is a supply phenomenon and, 3.l! such, can
only be remedied by monetary and fiscal means at the expense of
intolerable underutilisation of resources. Structuralists do not deny
that inflation could not persist for long without monetary expansion,
but they regard this as irrelevant because price stability could only
be achieved by monetary means at the cost of stagnation and under-
employed resources. Thus the role of financial factors in propagating
inflations is not denied; what is disputed is that inflation has its
origins in monetary factors. In the structuralists' view, monetary policy

would only attack the symptoms of inflation, not its root causes.
They argue further that, owing to the instability of export proceeds,
and the slower growth of exports relative to the demand for imports,
the control of inflation in the short run, for balance of payments
reasons, is virtually impossible without external assistance, unless
growth is to dry up completely.
In support of their case that inflation emanates from the supply
side, the structuralists point to the characteristic features of develop-
ing countries that we briefly touched on earlier, e.g. the rapid
structural changes taking place in the economy, the supply in-
elasticities leading to bottlenecks, etc., and refer back to the pre-
industrialisation era in Latin America when inflation was much less
severe than it has been in the recent past. The main plank of the
argument, and on which blame for inflation is placed, seems to be
the lack of preparedness for industrialisation. Prior to 1930 there was
relative price stability due to fairly elastic supplies of agricultural
output and low population growth. It is claimed, however, that
Latin America entered the industrialisation era burdened with a
capitalist class unwilling to invest and a growing pressure of popula-
tion on food supplies, which together contributed to bottlenecks and
the beginnings of inflation - subsequently exacerbated by a wage-
price spiral.
There is some dispute, though, whether this is an accurate picture
for the whole of Latin America. According to some observers the
sequence of events described by the structuralist school is more a
description of a particular country, Chile. Indeed, Campos I has gone
as far as to say that any visitor to the Economic Commission for Latin
America in Santiago cannot help feeling that the thinking of the
structuralist school has been affected by the peculiarities of the
Chilean inflation. But Campos is a confessed monetarist! For even
in the Chilean case he claims that the bottlenecks observed were in-
duced by inflation itself and not the causal elements in the process.
This is a more general claim of the monetarist school. They argue
that supply bottlenecks are created by policies which discourage
investment, e.g. price controls. Thus they maintain that the process
of repressing inflation (which is fairly typical in Latin America), in-
stead of tackling the root causes of inflation, creates bottlenecks
1 R. Campos, 'Two Views on Inflation in Latin America', in A. Hirsch-
man (ed.), Latin American Issues (New Haven: Yale University Press, 1961).
which subsequently feed the inflation. But in the first instance in-
flation is caused by excess demand due to monetary expansion. And
balance of payments difficulties are also ascribed to monetary
irresponsibility. In summary, the monetarists hold that the only way
to curb inflation is through monetary and fiscal policy to dampen
demand, and that most of the alleged supply inelasticities and bottle-
necks are not autonomous, or structural, but result from price and
exchange-rate distor:ions generated in the inflationary process. And
on the broader question of whether or not inflation is desirable in the
interests of development, the answer of the monetarists is an unequi-
vocal 'no'.
It is true that it is hard to find a systematic relation between the
rate of inflation and the rate of development. Argentina, Chile and
Bolivia have stagnated despite rapid increases in the price level,
while Mexico, Venezuela and Ecuador, which stand lower in the
inflation league, have developed comparatively quickly. If anything,
the relation may be negative rather than positive. It also seems to be
the case that in countries where prices have risen the fastest the
money supply has also grown most rapidly, at rates of between 20
and 30 per cent per annum. Neither of these observations, however,
answers the question of whether monetary expansion initiates infla-
tion or whether it simply stokes inflationary tendencies already
present on the supply side. Moreover, if monetary expansion was cur-
tailed, would it be output or prices that would fall the most? There
is no consensus, but a majority of observers as far as Brazil, Argentina
and Chile are concerned seem to pin-point supply factors as the
main contributors to rising prices. In the case of Brazil, especially,
there seems to be a fair degree of unanimity that inflation has resulted
from a combination of policies to foster high-priced import-substitute
activities and exchange depreciation due to balance of payments
difficulties. It is, of course, possible to argue that all balance of pay-
ments deficits are a monetary phenomenon, but the more relevant
question is: what is the cost of balance of payments stability in terms
of growth? If the cost is high, inflation can be regarded as the price
paid for the worthy goal of attempting to maintain growth in the face
of balance of payments deficits. There is a case, at least for policy-
making purposes, for regarding inflation as primarily a supply
phenomenon if a reasonable rate of growth is incompatible with
a balance of international account, or if price stability is only com-
patible with massive underutilisation of resources. A case in point is
Chile. Some recent evidence for the modern sector of Chile suggests,
for example, that while wage-rate increases and prices are responsive
to variations in the pressure of demand, complete wage and price
stability would require unemployment rates of between 15 and 20
per cent. 1 This is prima facie evidence of extreme structural dis-
equilibrium in the economy, and although demand management
could be used to control inflation, the extent of unemployment im-
plied would apparently be colossal. In situations like this, what is
clearly required are policies on the supply side to bring about a
greater degree of balance between sub-markets so as to shift inwards
the 'trade-off' curve (as conventionally drawn) between inflation
and the pressure of demand for the total economy.
What, then, are we to make of these two schools of thought on
inflation in Latin America, and the relatively scanty evidence on
which to make a judgement? The best solution is undoubtedly to
play safe and steer a middle course. Indeed, it would not be difficult
to defend the view, even without much knowledge of individual
situations, that inflation has probably been a combination of both
demand and supply factors; all inflations, other than hyper-infla-
tions, usually are!
B. FOREIGN RESOURCES FOR INVESTMENT
Dual-gap Ana(ysis and Foreign Borrowing

In an open economy domestic savings can be supplemented by many
kinds of external assistance. In this section the role of foreign borrow-
ing in the development process will be considered, together with the
debt-servicing problems associated with it. Later, we shall consider
types of foreign assistance such as private foreign investment and
international assistance, both bilateral and multilateral.
In national income accounting an excess of investment over
domestic saving is equivalent to a surplus of imports over exports.
If we write
Income =Consumption + Imports + Saving
Output =Consumption + Exports + Investment
and Income =Output
then Investment - Savings =
Imports - Exports.
1 R. Coot, 'Wage Changes, Unemployment and Inflation in Chile',
Industrial and Labour Relations Review (July 1969).
An import surplus financed by foreign borrowing can supplement
domestic savings directly, or indirectly by providing foreign ex-
change to buy imports which could be capital goods or substitutes
for domestically produced consumer goods.
Notice that in accounting terms the amount of foreign borrowing
required to supplement domestic savings is the same whether the
need is just for more resources for capital formation or for imports as
well. The identity between the two gaps, the savings-investment
(S-J) gap and the export-import (X-M) gap, follows from the nature
of the accounting procedures. It is a matter of arithmetic that if a
country tries to invest more than it saves, a balance of payments
deficit will result. Or to put it a different way, an excess of imports
over exports necessarily implies an excess of resources used by an
economy over resources supplied by it, or an excess of investment
over saving. There is no reason in principle, however, why the two
gaps should be equal ex ante. This i, the starting-point of dual-gap
analysis.
Consider a country with a particular target rate of growth. To
achieve that target rate of growth, savings and imports will be re-
quired. In the Harrod model of growth it will be remembered that
the relation between growth and saving is given by the incremental
capital-output ratio (c), which is the reciprocal of the productivity
of capital (p), i.e. g = sfc or g = sp, whereg is the rate of growth and
s is the savings ratio. Likewise, the relation between growth and
imports is given by the incremental output-import ratio (m'),
i.e. g = im', where i is the import ratio. If p and m' are given, an
increase in g requires an increase in sand i. Let r be the target rate of
growth. The required saving ratio (s*) to achieve that target is then
s* = rfp, and the required import ratio (i*) is i* = rfm'. If domestic
saving is calculated to be less than the level necessary to achieve
the target rate of growth, there is said to exist a savings-investment
gap equal to s* - s. Similarly, if minimum import requirements to
achieve the growth target are calculated to be greater than the
maximum feasible level of exports, there is said to exist an export-
import, or foreign exchange, gap equal to i* - i. In the absence of
foreign borrowing, growth will proceed at the highest rate permitted
by the most limiting factor. If the biggest gap is the savings-invest-
ment gap, growth is limited by the availability of domestic savings
and is said to be investment-limited. If the biggest gap is the foreign
exchange gap, growth is limited by the availability of foreign
exchange and is said to be trade-limited. Traditionally, the role of

foreign borrowing was to supplement deficient domestic saving. The
distinctive contribution of dual-gap analysis to development theory
is that if foreign exchange is the dominant constraint it points to
the additional role of foreign borrowing in supplementing foreign
exchange, without which a fraction of domestic savings might go
unutilised because actual growth would be constrained by the
inability to import necessary inputs. That is, if the foreign exchange
gap is the larger, (i* - i)m' > (s* - s)p, growth cannot proceed at
the rate sp but must proceed at the lower rate im'. If p is given, a
fraction of s must go unused.
Dual-gap analysis also performs the valuable service of emphasis-
ing the role of imports and foreign exchange in the development
process. Dual-gap analysis synthesises traditional and more modern
views concerning aid, trade and development. On the one hand it
embraces the traditional view of foreign assistance as merely a boost
to domestic savings; on the other hand, it takes in the more modern
view that many goods necessary for growth cannot be produced by
the less developed countries themselves, and must therefore be
imported with the aid of foreign assistance. Indeed, if foreign ex-
change is truly the dominant constraint, and consistently in short
supply, some would say that dual-gap analysis also presents a more
relevant theory of trade for less developed countries which justifies
protection and import substitution.! If growth is constrained by a
lack of foreign exchange, free trade cannot guarantee simultaneous
internal and external equilibrium, and the gains from trade may be
offset by the underutilisation of domestic resources. We shall take up
this matter in Chapter 9. Before going on to the assumptions of dual-
gap analysis, and its practical policy implications, however, let us
first analyse the two gaps in a little more detail and consider the role
of foreign borrowing in relation to them. It should be stressed from
the outset, of course, that since growth is limited by the larger of the
two gaps, foreign borrowing is only required to meet the larger of the
two gaps. If the export-import gap is the larger, the role of foreign
borrowing in supplementing domestic saving is, in effect, superfluous
and vice versa. The two gaps are not additive.
1 S. B. Linder, Trade and Trade Policy for Development (New York: Praeger,
1967).
The Savings-Investment Gap

Suppose to start with that we assume that the S-/ gap is the larger of
the two gaps, so that foreign borrowing must be sufficient to meet the
shortfall of domestic saving below the level necessary to achieve the
target rate of growth. What we wish to consider is the size of the
initial gap that must be filled by foreign borrowing and the deter-
minants of the size of the gap to be filled in future years by foreign
assistance. If the gap is to narrow, and foreign borrowing be ter-
minated, the presumption must be that additional increments to
saving out of the increases in national income generated are greater
than the increments of investment. For any target rate of growth r,
the required foreign assistance in the base year (Fo) is
Fo = lo - So = roer - Tosa = To(cr - sa) (8.9)
where Io is investment in the base period
So is savings in the base period
rois income in the base period
c is the incremental capital-output ratio
sa is the average savings ratio
and r is the target rate of growth.
If the rate of saving is expected to rise over time, the savings function
may now be written as
St = saTo + s'(Tt- To) = (sa- s')To + s'Tt (8.10)
where s' is the marginal savings ratio.
Rewriting investment requirements at time t as
It = TtCT (8.11)
and combining (8.10) and (8.11), we get the net inflow of capital
required in time t:
Ft = Ttcr- [(sa- s')To + s'Tt] (8.12)
or Ft = (cr- s')Tt + (s'- sa) To. (8.13)
The difference between borrowing requirements in the base year and
borrowing in period t is the difference between equations (8.13)
and (8.9), which reduces to
cr(Tt- To) - s'(Tt- To),
or Ft - Fo = Ill- !:i.S (8.14)
i.e. increments in external capital finance the difference between

investment requirements to sustain the target rate of growth and
increases in saving generated by rising income. If foreign assistance
is to decline (i.e. Ft < F 0 ), llS must be greater than Ill. The savings-
investment gap will disappear, and the phase of investment-limited
growth come to an end, when domestic saving reaches a level
adequate to sustain the target rate of growth.
From (8.13) the rate of growth that can be achieved with a given
inflow of foreign capital is
r= c1((Sa - , Yo
s ) Yt + s + Ft)
Yt . (8.15)
The derivative of r with respect to F is positive, so that a larger

inflow offoreign capital can achieve a higher growth rate, provided
there is no rise in c - the capital-output ratio.
From (8.13) we can also calculate the number of years after which
a country can generate enough domestic saving to finance its target
rate of growth and dispense with foreign borrowing. Equation
(8.13) becomes at the nth year, when Fn = 0,
(cr- s')Yn + (s'- sa) Yo= 0 (8.16)

s ' - Sa
Yn =-,--Yo. (8.17)
s - cr
Since Yn = Yo(l + r)n (8.18)
then (I + r)n -_ s' - Sa

s'- cr (8.19)
from which n can be calculated. We stress again that the essential

condition for a country to reduce its external borrowing requirements
is that the marginal rate of saving should exceed the required rate of
investment, i.e. s' > cr in (8.19) above. n is obviously highly sensitive
to s' and c.
The Export-Import, or Foreign Exchange, Gap

Now let us suppose that the X-M gap is the larger of the two gaps.
In the base year the foreign assistance required to cover the foreign
exchange gap is
Fo = Mo- Xo = Yoma- Yoxa = Yo(ma- xa) (8.20)

where Mo is imports in the base period
Xo is exports in the base period
To is income in the base period
ma is the average import coefficient
and Xa is the average export coefficient.
If imports in year t equal

Mt =maYo+ m'(Tt- To), (8.21)
and exports in year t equal
Xt = xaTo + x'(Tt- To), (8.22)
where m' is the marginal import coefficient

x' is the marginal export coefficient
and Tt = ltfcr,
then the foreign borrowing requirement in time t to achieve the

target rate of growth is
Ft =maYo+ m'(Tt- To) - xaTo- x'(Tt- To). (8.23)
Subtracting (8.20) from (8.23) gives
Ft - Fo = tlM - tlX
i.e. increments in external assistance finance the difference between

imports to sustain the target rate of growth and the increment of
exports. If the level of foreign assistance is to decline over time (i.e.
Ft < Fo), x' must be greater than m'.
From (8.23) the rate of growth that can be achieved with a given
inflow of capital is
r= :~(m' - x'). (8.24)
The condition for reduced dependence on foreign assistance is that

x' should increase relative to m'.
An alternative approach is to express the minimum import re-
quirement for growth as a proportion of investment requirements, in
which case, from (8.11), we have
(8.25)
where m-1 is the investment import coefficient.

The condition that Ft = Mt - Xt gives

'Ytmtcr = Xt + Ft (8.26)
and the trade-limited growth rate is then
r = _1 (Xt
m,c 'Yt
+ Ft)
'Yt ·
(8.27)
The export-import gap will become less and less restrictive as time
goes on, provided exports increase at a faster rate than national in-
come (i.e. if:;:> 0 or x' > xa), and/or if mt falls, which is likely as
more and more capital goods are produced domestically.
With respect to policy, the analysis highlights the importance of a
greater proportion of resources devoted to exports. The X-M gap
will only disappear, and trade-limited growth come to an end, when
exports rise to a level sufficient to meet the import requirements of
the target rate of growth set. One of the preconditions of an end to
reliance on foreign borrowing to finance development, and to pay
off past debts, is that a balance of payments deficit be turned into a
healthy continuing surplus through price and income adjustment
mechanisms internally.
Two Gaps in Practice

The first task is to reconcile the existence of two gaps ex ante with
the fact that, ex post, the two gaps are identical by definition. This
reconciliation is easily brought about by treating the main para-
meters of the system, i.e. the capital-output ratio, the savings ratio
and the import coefficient, not as constants but as upper and lower
limits. The second question is whether in practice there is the lack of
substitutability between imports and domestic resources which dual-
gap analysis assumes. The assumption of fixed factor proportions,
an inflexible composition of imports, and rigid relationships between
imports and investment and imports and output is somewhat strong.
In the first place, capital and imports are synonymous if capital
goods can be hired from abroad. Secondly, imports allow the domes-
tic product mix to be varied and thus can act as a substitute for
capital by allowing countries to produce goods with lower capital-
output ratios. Thirdly, many imports are ofsuch strategic importance
that they can lower the capital-output ratio over a wide range of
activities. Imports can be a source of new or improved technology
which has wide-ranging repercussions on productivity growth in
the country at large. If there is complete substitutability between
imports and domestic resources, theoretically there is only one gap-
ex ante as well as ex post.
On the other hand, factor proportions may be slow to adjust, and
substitution between foreign and domestic resources may be a
long-drawn-out process, which makes it quite legitimate to under-
take empirical analysis of shortages of foreign exchange and domestic
savings at particular points in time, and over time as well. A con-
siderable amount of work in this field has been undertaken by
Professor Chenery with collaborators.! Of the countries that have
been studied, the typical sequence seems to be a savings-investment
gap followed by an export-import gap, sometimes of considerable
stubbornness.
As long as the adjustment between capital and imports is slow,
the traditional policy conclusions of dual-gap analysis remain valid.
Firstly, to prevent resource 'waste', a co-ordinated development
policy must attempt to speed up the process of adjustment to equate
the two gaps ex ante. If the capital inflow required by the S-1 gap is
greater than the foreign exchange gap, the two gaps can be equated
by having imports in excess of the necessary level required for growth,
or by reducing exports. If foreign exchange is the dominant restraint,
'excess' savings should be used either for promoting export- or
import-competing activities, or used for types of productive invest-
ment which do not require imported capital goods, e.g. investment
in human resources. So long as this type of investment can be ex-
panded with positive productivity, the existence ofa foreign exchange
gap should not prevent the translation of potential saving into in-
vestment, and the possibility of excess saving disappears. If the
export-import gap is so dominant, however, it is probably wiser to
devote more investment to import saving and export promotion even
though this may involve some cost in terms of domestic output.
1 See, e.g., H. Chenery and M. Bruno, 'Development Alternatives in an
Open Economy: The Case of Israel', Economic Journal (Mar 1962); H.
Chenery and I. Adelman, 'Foreign Aid and Economic Development: The
Case of Greece', Review of Economics and Statistics (Feb 1966); H. Chenery
and A. Macewan, 'Optimal Patterns of Growth and Aid: The Case of
Pakistan', Pakistan Development Review (summer 1966); H. Chenery and
A. Strout, 'Foreign Assistance and Economic Development', American
The second point is that since the two gaps are not additive, the
criticisms of dual-gap analysis do not seriously affect calculations of
international assistance necessary for growth targets to be achieved,
except where the foreign exchange gap substantially exceeds the
savings gap and no allowance has been made for the possibility of
substituting domestic resources for imports. Taking the less developed
countries as a whole there is, in fact, no tendency for one gap to pre-
dominate as the major constraint; but, in any case, certain types of
foreign assistance can themselves help the substitution of domestic
resources for imports. The fact that domestic resources can be sub-
stituted for imports should not be allowed to demote the role of
imports and foreign exchange in the development process and the
possibility that imports may increase the level of domestic saving
without any specific act of thrift on the part of the country con-
cerned.
A recent study of forty developing market economies by
the United Nations Commission for Trade and Development
(UNCTAD) found no significant difference between the savings-
investment gap and the export-import gap for countries overall, and
puts the foreign exchange gap at $24,000 million in 1975 on the
assumption of 6 per cent growth.l This compares with the present
flow of assistance to less developed countries of approximately
$12,000 million, with no allowance for the outflow of interest and
profits.
Foreign Borrowing and the Debt-servicing Problem

The role and need of foreign assistance has been established. But
what form should assistance take, and what are the implications of
the different types of foreign assistance from the recipient's point of
view? Generally, the less developed countries do not have much
choice in the matter, but the implications may be profound. For
example, borrowing in the form of interest-bearing loans from inter-
national institutions or from the world capital market involves debt-
servicing problems which do not arise with pure aid, or private
foreign investment for that matter. How much is it prudent for less
developed countries to borrow, given that interest and amortisation
payments commit future foreign exchange?
1 UNCTAD, Trade Prospects and Capital Needs of Developing Countries
(Geneva: United Nations, 1968).
In recent years a good deal of concern has been expressed over
the amount of borrowing indulged in by the less developed countries
and the future strain on the balance of payments that this implies.
Already some countries have reached the stage where outflows of
funds to service old debts exceed 50 per cent of new foreign assistance
coming in and offset a substantial proportion of total foreign exchange
earnings. The total of debt-service payments of developing countries
in various continents for the period 1961-8 is given in Table 8.1, and
payments as a percentage of gross assistance are shown in Table 8.2.
TABLE 8.1
Debt Service Payments, 1961-8
($U.S. million)
South East Middle South Latin

Total Africa Europe Asia East Asia America
1961 2,314 172 252 224 170 246 1,250

1962 2,585 225 222 264 210 227 1,437
1963 2,749 494 265 165 188 269 1,368
1964 3,177 433 330 171 212 359 1,672
1965 3,279 445 407 206 182 347 1,692
1966 3,781 463 444 341 200 417 1,916
1967 3,969 535 461 280 168 486 2,039
1968 4,018 443 506 369 162 565 1,973
Source: Partners in Development: Report of the Commission on International

Development (Pearson Report) (London: Pall Mall Press, 1969) p. 372.
In several countries the ratio of debt-service payments to export

earnings is also high and growing. One cannot fix a limit to the debt-
service/exchange earnings ratio, but we can direct attention to the
general question of whether it is possible for a country to borrow 'too
much'. Is it possible that after a point, even though foreign borrow-.
ing has not filled the largest of the two gaps restraining growth, the
disadvantages of further borrowing outweigh the advantages? Un-
fortunately there are no precise criteria on which to make a firm
judgement, but two factors would seem to warrant consideration.
The first is whether or not increases in the interest burden exceed
the growth of national income, which will determine whether or not
domestic saving continues to increase as a result of foreign borrow-
ing. The second is the ratio of payments out on old loans to new
investment coming in.
As far as domestic saving is concerned, this will continue to in-

crease as long as income increases (assuming the propensity to con-
sume is less than unity); and income will increase as long as the rate
of interest of foreign loans is less than the reciprocal of the capital-
output ratio, i.e. the productivity of capital. Ifinterest charges exceed
the contribution offurther investment to output, income and saving
will fall. A useful concept in this connection is the concept of the
TABLE 8.2
Debt service as % Debt service as %

of gross assistance, of gross assistance
1965-7* in 1977 if gross
flow of assistance
remains unchanged t
Africa 73 121
Europe 92 109
East Asia 52 134
South Asia/Middle East 40 97
Latin America 87 130
* Including suppliers' credits, private and government loans, and inter-

national loans, but excluding grants and direct private investment.
t Assuming composition of assistance remains unchanged.
Source: Pearson Report, p. 74.
critical rate of interest which is the interest rate on foreign borrowing

which keeps interest payments from rising faster than national in-
come on certain assumptions with respect to the savings and capital-
output ratios. Assuming, for simplicity, no external debt at the
beginning of the initial year, the critical rate of interest can be
expressed as
z. = _,r(....::so~_s..t..')
- (8.28)1
so- cr
where i is the critical rate of interest
r is the growth of national income
so is the initial savings ratio
s' is the marginal savings ratio
and c is the incremental capital-output ratio.
1 SeeJ. P. Hayes, 'Long Run Growth and Debt Servicing Problems', in
D. Avramovic and Associates, Economic Growth and External Debt (Baltimore:
Johns Hopkins Press, 1964).
To give an example, if r = 4 per cent, so = 12 per cent, s' = 15
per cent and c = 4, then i = 3 per cent. If the interest charge on
foreign loans was greater than 3 per cent it would not pay to borrow.
The higher the marginal savings rate, other things remaining the
same, the higher is the critical rate of interest. The concept of the
critical rate of interest is important because if interest payments do
start to rise faster than national income, further borrowing must take
place simply to maintain the present level of saving and cannot fulfil
its true role as a supplement to domestic saving.
As far as the ratio of outflows to new investment is concerned, pay-
ments out on old loans will not exceed new investment coming in as
long as the growth rate of new investment is greater than the rate of
interest. If the interest rate does exceed the growth rate, the ratio
will rise but not indefinitely.! To illustrate, let R be the ratio of pay-
ments out to new investment coming in, let A be the annual amortisa-
tion charge, let I be the annual interest charge, and G be the annual
new investment. We then have
R =A+l
G . (8.29)
This expression will approach the limit

R _A+h_a+i (8.30)
L- G -a+r
where i is the interest rate
r is the rate of growth of new investment
and a is the annual rate of amortisation.2
lfr > i RL < 1
Ifr = i RL = 1
Ifr<i RL-+ limit as given by expression (8.30).
The policy implication is that the way to help developing coun-

tries to reduce the debt-servicing burden is not to cut investment but
to reduce interest charges, or alternatively to allow the repayment of
debts in domestic currency. Another solution, of course, would be
1 See E. Domar, 'The Effect of Foreign Investment on the Balance of
Payments', American EconomU; Review (Dec 1950).

2 This term appears in the denominator and the numerator because on
the one hand a larger rate of amortisation increases A and therefore RL; on
the other hand, it reduces the outstanding debt and hence both A and I.
for donors to dispense with loans and substitute aid, in which case
the debt-servicing obligation could be dispensed with entirely in a
matter ofyears.
The debt-servicing implications of borrowing can be integrated
nicely with dual-gap analysis by means of a simple diagram showing
the time sequence of the savings-investment, or export-import, gap
against net resource flows, net borrowing and net indebtedness. An
understanding of the interrelationships can be obtained from Fig.
8.1, without a numerical example. We shall illustrate with reference
Investment (j)
Savings (S)
Net resource
flow (BR) I
Net borrowing (BB)
Net debt (80)
0
FIG. 8.1
to the S-1 gap, but the analysis is exactly analogous in the case of
the X-M gap. The savings-investment gap is given by the relation
between the curves SS and II. The net resource flow (BR) required
to bridge the savings-investment gap to maintain a target rate of
growth declines steadily, becoming a net resource outflow after OX
years. Net borrowing goes on a little longer in order to cover interest
charges on accumulated indebtedness. Net borrowing (BB) is zero
after Or years and at this point net indebtedness (BD) starts to
decline. Debt repayments, in theory, take place by converting the
excess of savings over investment into a balance of payments surplus
until all indebtedness is repaid by the year ,Z, after whic;:h the country
becomes a net creditor.
This is a model sequence of events which has been approximated
to by many of the now developed countries. In the case of many less
developed countries today, however, there is little evidence that they
have the ability to pay off their past indebtedness and cut down on
net resource inflows. The need for resources is as acute as ever and
their indebtedness is mounting because an export-import gap has
replaced the savings-investment gap. The countries find it difficult
to translate domestic savings into activities which ease the shortage
of foreign exchange because exports are price and income inelastic
and imports are income elastic. The cry 'Trade, not aid' is a plea for
more liberal trading policies to solve at one and the same time the
foreign exchange gap and the debt-servicing problem. Unless some-
thing is done, the debt-servicing problem arising from mounting
resource flows may well become unmanageable in the not too distant
future. It will certainly be a long time before these countries become
net exporters of capital even in the absence of a savings-investment
gap. In the case of India, Little and Clifford remark that
India's foreign exchange earnings are so low as to be barely
sufficient to pay for the materials needed to keep in full operation
the already established capacity to produce plus a very small
element of consumption good imports .... Thus she could buy
virtually no foreign machinery, nor pay for foreign exports, if it
were not for aid. At the same time her own capital goods industries
are quite inadequate to make the equipment needed.l
For countries like India there can be no question of cutting down on
the annual resource inflow, provided the critical rate of interest is
not exceeded.
An ingenious proposal for mitigating the debt-servicing problem
has recently been put forward by Khatkchate.2 The proposal in
brief is that the less developed countries should be allowed to dis-
charge their repayment commitments to the creditor-developed
countries not through export surpluses with themselves, which
implies the relinquishing of valuable foreign exchange, but through
1 I. Little and J. Clifford, International Aid (London: Allen & Unwin,
1965) p. 144.
2 D. R. Khatkchate, 'Debt-Servicing as an Aid to Promotion of Trade
of Developing Countries', Oxford Economic Papers (July 1966).

additional exports to other less developed countries. The developing

countries would service their outstanding debt by making interest
and amortisation payments in local currencies to Regional Develop-
ment Banks which would then 're-lend' these local currencies to
other less developed countries for the purchase of exports from the
original debtor countries. The developing countries which receive
these exports would increase their liabilities to the creditor-developed
countries to the same extent as the debt liabilities are reduced of the
developing countries which export these goods. As far as the creditor-
developed countries are concerned, it would mean that the
repayments due to them from one debtor-developing country would
be relent immediately to another developing country, and so on.
This process could continue until all the developing countries had
built up their economies to such an extent that they are able to raise
their exports to the developed countries and eventually to discharge
their debt obligations to them. The proposal amounts to postpone-
ment of the retirement of debt to some indefinite date.
The scheme appears admirable on the surface, but are there dis-
advantages, and is it feasible? Firstly, it must be recognised that the
proposal, while relieving pressure on foreign exchange resources,
amounts to a form of' aid tying' between the less developed coun-
tries themselves, and the question must be asked: how many develop-
ing countries are there that can provide the sort of goods that other
less developed countries require for development purposes? On the
other hand, trade between the less developed countries is something
to be encouraged, and this kind of scheme would provide a useful
stimulus, and need not preclude the expansion of trade with the
developed nations as well.
Whether such a scheme is feasible or not depends on the attitude
of the creditor-developed nations and of the multilateral aid-giving
institutions such as the World Bank. The balance of payments of
creditor countries would be affected adversely in at least two ways.
Firstly, the inflow of interest and amortisation payments from
developing countries would cease (although interest would continue
to accumulate to be paid at some future date). Secondly, exports to
developing countries, financed by foreign assistance, would probably
fall. The scheme would involve a balance of payments loss to
developed countries unless gross aid is reduced. If so desired, how-
ever, the scheme could be a disguised way for governments to in-
crease net aid. Since people in general appear to attach more
importance to gross figures than net, a government could achieve a
net increase in the transfer of resources, without arousing public
hostility, simply by maintaining its flow of gross aid.
Multilateral institutions would also suffer under the scheme
through the non-repayment of loans. The proposal does not imply
default, however, or writing-off interest payments. It merely involves
a transfer of the debt liabilities of the developing countries to the
World Bank to Regional Development Banks. The Bank can roll
over its debt by new bond flotations, in the same way that a national
government might fund its national debt.
Short of an ambitious scheme of this nature, the only other solu-
tion to the debt-service problem is for the developed countries to
make the terms of assistance more generous by reducing or scrapping
interest charges on loans and granting pure aid.
International Assistance
The main forms of international assistance to less developed coun-
tries consist of private foreign investment, bilateral grants, loans and
technical assistance, and multilateral assistance of various types
channelled through international institutions such as the United
Nations, and the World Bank plus its two affiliates, the International
Development Association (I.D.A.) and the International Finance
Corporation (I.F.C.).
Before going on to consider the magnitude of different types of
assistance, and the criteria by which it is allocated, a clear distinction
needs to be made between the nominal and real value of assistance
from the donor's point of view, and between assistance and aid from
the recipient's point of view. Unconditional grants, where the
nominal cost equals the real cost, are the purest form of aid, but only
a very small percentage of total resource flows consists of assistance
of this type. The major part of international assistance consists of
loan facilities with a small subsidy element, and what is required is
a standard means of converting loan figures into a figure for aid and
to distinguish the nominal cost of assistance from the real cost.
Pincusl suggests expressing the value of aid as the combined nominal
(or market) value of all forms of assistance less the present value of
loan repayments discounted at a rate of interest reflecting the
1 J. A. Pincus, 'The Cost of Foreign Aid', Review of Economics and Statistics
(Nov 1963).
alternative employment of long-term capital. By this definition, all

forms of assistance are reduced to their value as grant or subsidy.
What this, in effect, means is that countries receive aid if the rate of
interest on assistance is lower than the productivity of capital. If
the same discount rate is used in the donor country, the amount of
aid to the recipient country represents the difference between the
nominal cost and real cost of assistance to the donor country.
When countries' nominal assistance is converted into aid there is
a wide discrepancy between the nominal and real cost. The impres-
sion commonly given that international .development institutions
exist to dispense charity, and that bilateral assistance from rich to
poor countries is a form of poor relief, is far from the truth. Less than
one-half of nominal assistance to less developed countries represents
aid. Most assistance represents the creation of assets for the donor,
involving a much smaller transfer of real resources than the nominal
figure suggests, and representing an even smaller cost to the balance
of payments if the assistance is tied to the purchase of the donor's
products. The World Bank itself is a commercial institution lending
(and borrowing in the world's capital markets) at commercial rates
of interest. The I.D.A. does dispense 'soft' loans -loans at low in-
terest rates with long repayment payments -but the assistance is not
pure aid. The I.F.C., the other of the World Bank's main affiliates,
concentrates on encouraging private enterprise in less developed
countries by making equity investments. Very little aid is involved.
Since these three institutions together are the major source of multi-
lateral development finance, the situation is far removed from world
charity on a grand scale.
The total net flow of official and private assistance from industrial-
ised countries of the world (comprising the Development Assistance
Committee of the O.E.E.C.) to developing countries and multi-
lateral agencies in 1968 was as in the table below. 1
Of the total, approximately 60 per cent consisted of official (non-
private) disbursements, and 40 per cent constituted private overseas
investment. Nearly $13,000 million- almost double the figure for
1 The total net flow of financial assistance is the gross flow of private and
public assistance minus amortisation but with no allowance for interest and
profits. The 1 per cent aid target established by the Pearson Commission
refers to this net flow. Official development assistance is similarly defined as
net of amortisation but not of interest repayments. Given the magnitude of
interest repayments and profit remittances, the net flow of resources is
clearly much less than net assistance.
%of
Amount national
Country ($m.) income
United States 5,675 0·65
Canada 306 0·49
Sweden 127 0·50
Denmark 74 0·55
Australia 187 0·63
Norway 58 0·65
France 1,483 1·24
United Kingdom 846 0·83
Germany 1,635 1·24
Belgium 243 1·15
Netherlands 276 1-10
Austria 74 0·66
Italy 505 D-70
Japan 1,049 0·74
Switzerland 214 1·26
Total 12,752 0·77

1958- is undoubtedly a substantial sum absolutely, but trivial com-

pared to the development needs of the countries to which it flows,
and looks even more derisory when divided by the population ofless
developed countries. In the latter half of the 1960s the average net
flow of assistance per head of the population of the less developed
countries was $4 per annum. By continents the distribution was as
follows:
Numbers of countries receiving:
over $6 $3-$6 under $3
per head per head per head
Mrica (average $6·1) 9 8 3
Latin America (average $4·4) 8 9 5
Asia (average $3·2) 5 5 15
Southern Europe (average $4·7) I 2 I
To achieve per capita income growth in the less developed countries

of between 4 and 5 per cent per annum, assuming population growth
of 1·5 per cent and a capital-output ratio of between 3 and 4, would
require rates of investment of about 20 per cent. With domestic
saving of 15 per cent ofG.N.P., this implies foreign assistance equal
to 5 per cent of the current national income of less developed
countries. At present, foreign assistance is half this figure.
Official Bilateral and Multilateral Assistance

The net flow of non-private international assistance from developed
countries to less developed countries and multilateral agencies in
1968 was approximately $7,000 million, ofwhich one-half was from
the United States, $900 million from France, $600 million from
Germany and $450 million from the United Kingdom. These
countries are, and have been for some time, the principal sources of
official assistance to less developed countries. In absolute terms the
total net flow sounds impressive, but spread over the whole less
developed world amounts to less than $3 per head per annum.
Bilateral assistance predominates. Only some 10 per cent of total
official grants and loans are channelled through international in-
stitutions, and this proportion has not increased over the last ten
years despite lip-service paid to the desirability of increasing the
proportion of assistance given on a multilateral basis. Of the total
official disbursements of approximately $7,000 million in 1968,
approximately $6,300 million was bilateral and $700 million
multilateral. This latter figure should not be confused, of course,
with the total of gross disbursements by multilateral agencies to
less developed countries, which consists not only of the contri-
butions of developed countries but also funds raised on the
capital market and repayments on previous loans. Gross disburse-
ments to less developed countries in 1968,1 by various multilateral
agencies, were as follows:
Agency $m.
World Bank 605
I.D.A. 215
I.F.C. 31
Inter-American Development Bank 233
Asian Development Bank 20
Mrican Development Bank 2
European Development Fund 121
U.N. Institutions 300
Total 1,527
1 The figures for 1969 show an increase of nearly 100 per cent, and there
is every hope that the contribution of the World Bank will increase further
in the coming years.
Private Foreign Investment

A supplement or substitute for official assistance, although not with-
out its own difficulties, is private foreign investment. Given the world
shortage of capital, individual countries must compete with each
other for the favours of foreign investors. Less developed countries,
desirous of attracting private foreign investors, must create , an
environment as favourable as possible. For the investor this means
political stability, security, facilities for the remittance of profits,
dividends and interest, and guarantees of full compensation in the
event of expropriation or nationalisation. From studies that have
been undertaken on attitudes to private investment overseas, the
major deterrent factors appear to be the threat ofinterference in the
operation of business, labour problems, foreign exchange controls
and political and legal uncertainties. And the crucial economic
limitation on investment in less developed countries is the low
prospective rate of return on capital due to the small size of the
effective market and the difficulties of exporting.
The main disadvantage of private foreign investment compared
with other forms of foreign assistance, and especially loan finance, is
that the recipient country has little control over the direction of the
resources and that a large proportion of profits earned from the
investment may leave the country. On the other hand, there is bound
to be some net benefit. The demand for labour will be increased; tax
revenue will rise; external economies may be generated; and the
foreign investment may set up backward and forward linkages and
act as a stimulus to domestic investment. Direct investment from
abroad also possesses the distinct advantage that it is often ac-
companied by advanced technology and technical expertise. The
potential is there for a profound impact on indigenous industry, on
attitudes, and the state of competition. In all, the national economic
benefit from private investment will almost certainly exceed the
private benefit. As long as the total increase in productivity is not
appropriated by the investor and remitted abroad, the less developed
country will gain from private foreign investment. Moreover, direct
investment in less developed countries does not involve the debt-ser-
vicing obligations that are the inevitable consequence ofloan finance.
On the other hand, as we mentioned earlier, private foreign investment
usually involves a continual outflow of profits, which will typically
continue for longer than outflows of interest and amortisation on
a loan of equivalent amount. The relative advantages and dis-

advantages of private investment versus loans for development can be
listed, but it is difficult to give an objective answer to the question of
which is the superior means of development finance.
The total flow of private foreign investment from developed to
less developed countries in 1968 was as follows :
$m.
Australia 30
Austria 46
Belgium 150
Canada 94
Denmark 45
Finland 3
France 628
Germany 1,040
Italy 356
Japan 542
Netherlands 142
Norway 35
Switzerland 223
Sweden 54
United Kingdom 417
United States 2,071
Total 5,876
Source: \Vorld Bank, Annual Report, p. 47.
The total for 1968 compares with figures for previous years of:
$m.
1961 3,097
1962 2,497
1963 2,512
1964 3,192
1965 4,172
1966 3,849
1967 4,182
It can be seen that there is some tendency for this type of assistance
to fluctuate from year to year. This is mainly due to the 'lumpiness'
of certain types of investment, especially in oil. Much of the recorded
increase between 1963 and 1965 is attributable to a sharp expansion
of United States direct investment in the oil sheikhdoms of the
Middle East.
Of the total flow of private investment to less developed countries
in 1968, over one-third came from the United States. The other
major sources were Germany, France, Japan and the United King-
dom. As to the type of investment, nearly $3,000 million was direct
investment including reinvested earnings, nearly $1,000 million was
bilateral portfolio investment, and the remainder export credits.
Private foreign investment in less developed countries represents,
of course, only a small proportion of private foreign investment in the
world as a whole. The major part of overseas investment takes place
within the developed world itself. Canada and Europe, for example,
absorb between 50 and 60 per cent of American private overseas
investment, and private overseas investment from Britain to less
developed countries is less than 50 per cent of the total. With the
increased political instability in less developed countries in the early
1960s, from coups d'etat to civil strife, private foreign investment as a
source of capital formation in less developed countries declined in
relative importance compared with official assistance. The level of
private investment picked up again in the latter half of the 1960s, but
the threat of political turmoil in less developed countries still remains
the most formidable barrier to an increased flow of private capital
which, economically speaking, the less developed countries can ill
afford to forgo.
Criteria for the Allocation of Assistance

The criteria for the distribution of bilateral assistance are largely
a matter for the donors concerned. In practice the criteria employed
tend to be more non-economic than economic, reflecting historical
relationships between countries and military and political objectives.
To a certain extent many of the less developed countries have become
a battleground in the cold war, and the distribution of bilateral
assistance reflects this. Most donors refrain from making explicit the
criteria on which they distribute assistance, but Ohlinl suggests that
four distinct systems of development assistance can be discerned.
Firstly, there is American involvement in world-wide political
development. Secondly, there is the assistance of colonial powers to
former territories. Thirdly, there is the emergence of countries, like
Germany and Japan, making extensive grants and loans aimed at
1 G. Ohlin, 'The Evolution of Aid Doctrine', in Foreign Aid Policies
Reconsidered (Paris: O.E.C.D., 1956).

promoting economic relationships between donor and recipient. And

lastly, there is the existence of smaller donor countries which give
assistance multilaterally.
The criteria for the distribution of multilateral assistance through
international agencies, to which many countries contribute, are of
wider concern. Although the international aid doctrine is essentially
a moral one, the fact remains that if the supply of resources is limited
in relation to demand, the issue must be discussed of what principles,
if any, should govern the distribution of assistance between countries.
Furthermore, the humanitarian motive is very inadequate as an
explanation of why and how assistance has actually been given.
Most assistance provided by multilateral agencies is in the form of
loans. We saw in the section on foreign borrowing that if the interest
rate on loans is in excess of the productivity of capital, or the rate of
inflow of new investment, certain debt-servicing problems will arise.
One possible criterion to adopt as a guide to distribution, therefore,
might be a productivity criterion, as an indication of a country's
capacity to pay back the loan with interest. Before the World Bank
lends, it must be satisfied that the borrowing country will be in a
position to repay the loan. To this end, its foreign exchange position
is scrutinised carefully because loans are made, and must usually be
repaid, in currencies other than that of the borrowing country. How
a productivity criterion would be applied in practice, however, is not
so clear-cut. It could be argued that assistance should not be given
to a country where the productivity of capital is below the rate of
interest; but should assistance necessarily be distributed to those
countries where its productivity is highest? The return from
assistance would be maximised, but distribution would be unrelated
to needs, and also unrelated to each country's long-term prospects of
development unless they could somehow be incorporated into the
measure of productivity. The reciprocal of a country's capital-output
ratio may be used as the measure of productivity, but we have al-
ready discussed the dangers of using the capital-output ratio as a
measure of the productivity of capital, and as a criterion for the
allocation of investment resources when different projects have very
different life-spans and dissimilar external and secondary reper-
cussions. The use of a productivity criterion also takes the terms of
the assistance as datum, whereas one might well argue for a change
in the terms on which assistance is given, in which case the capacity
to repay assistance in foreign exchange becomes less relevant.
The productivity criterion for distribution is closely linked with
the notion of absorptive capacity and the view that assistance ought
to be distributed according to absorptive capacity. Unfortunately
the absorptive capacity criterion suffers from a lack of an adequate
and acceptable definition of absorptive capacity. One possible
measure is a country's past rate of increase in investment expendi-
ture,! but this ignores the varying levels of productivity on past
investment and neglects future productivity. Absorptive capacity
clearly implies some acceptable rate of return on investment, but
what rate of return should be regarded as acceptable, and how
should factors that contribute to the productivity of capital be
treated? If a workable definition of absorptive capacity could
be agreed on, one might well argue that assistance should not be
distributed to countries which lack the capacity to absorb it; on the
other hand, it could equally be argued that one of the main purposes
of assistance should be to remove obstacles, such as domestic or
import bottlenecks, which limit absorptive capacity. In this case
the fact that absorptive capacity, however measured, is found to be
low, is not very helpful. The role of international assistance itself
ought to be to shift upwards the whole marginal efficiency of capital
schedule. A much more embracing criterion is required which
includes the ability of external resources to increase absorptive
capacity. Rosenstein-Rodan's suggestion is that assistance should be
distributed according to the degree of induced domestic effort in the
recipient countries which accompanies aid, which he suggests could
be judged on the country's own investment record, trends in the
marginal savings ratio, and its organisational and administrative
capacity. This is very much in line with America's self-help philoso-
phy, or what is called the criterion of potential performance, which
insists that before countries receive assistance for non-military
purposes they must present self-help development strategies as
evidence of their capacity to develop. Crude reliance on savings rates
and the rate of growth of productivity, however, can be very mis-
leading indicators of self-help if the initial conditions and natural
endowments of countries vary.
Another possible criterion is the availability of foreign exchange.
Until recently it was the rule of the World Bank only to meet the
foreign exchange costs of particular projects. If foreign exchange is
1 P. Rosenstein-Rodan, 'International Aid for Underdeveloped Coun-
tries', Review of Economics and Statistics (May 1961).
short, and the export-import gap is the dominant restraint, this

seems an attractive criterion on the surface. On the other hand, it has
no regard to the capacity to repay, or absorptive capacity. It would
be distribution mainly on the basis of need.
In the absence of an economically objective and value-free
criterion for the allocation of assistance, some would argue that the
criterion of need is as good as any. One possibility in this connection
would be to gear the allocation of assistance to raising per capita
income in different countries by a specified amount. Assuming that
assistance is productive, one type of scheme would be to distribute
assistance on a per capita income basis according to some target per
capita income level which ensured a tolerable standard of living in
all poor countries. Certain graduated rates of per capita income
assistance could be applied to the gap between the actual level of per
capita income and the target level. A country which fell way below
the target would receive a greater amount of assistance per head of
the population than a country which approached the target or
exceeded it. The rates of per capita assistance would need to be
worked out actuarially so that assistance was not solely distributed
on the 'worst-first' principle. Given knowledge of the total amount
of resources available, rates could be fixed which ensured a wide
spread of assistance across countries, but which did not make
demands on resources in excess of supply.
As it happens, there seems to be no relation at all between the
allocation of international assistance and need. In a model which
tries to explain the distribution of assistance across thirty-four
countries in terms of per capita income, absorptive capacity and
the foreign exchange position, Davenport found a positive relation
between assistance per head and income per head. Multilateral
assistance was as unevenly distributed as bilateral assistance.!
Given the present paucity of international assistance, one wonders
whether very precise criteria for distribution are worth formulating.
Annual assistance provided by multilateral institutions amounts to
less than $1 per head of the population in less developed countries,
and the total of non-private resource flows amounts to not more than
$3 per head. If bilateral assistance continues to be the major form of
official assistance to less developed countries, the allocation of re-
sources will continue to be determined largely by political, military
1 M. Davenport, 'The Allocation of Foreign Aid', Yorkshire Bulletin (May
1970).
and historical considerations. The only hope for a more just and
rational allocation of resources available is greater generosity towards
multilateral institutions on the part of individual nations. The criteria
to be applied will then take on more importance.
Technical Assistance
In the case of most developing countries, with a limited capacity
to absorb financial assistance because of other constraints, the real
need is to combine assistance with advice, expertise and technology
to help remove the constraints at the same time that the assistance is
given. In this way financial assistance can be fully exploited. Of
course, advice, expertise and technology are themselves forms of
assistance, and they are playing an increasingly important part in
multilateral and bilateral aid programmes. In 1968, total technical
assistance amounted to $1,481 million, or 21 per cent of official
development assistance, compared to 13 per cent in 1962. The
World Bank links the distribution of assistance with forms of techni-
cal advice, including the efforts of countries to curb their population
growth so that the benefits of assistance are not eroded by there
being more people to support. The World Bank, and individual
donor countries as well, also have a preference for supporting projects
as part of a development programme rather than distributing assist-
ance for unspecified purposes. Technical assistance can be more
easily lent for specific projects, and the benefits and return from the
assistance can be more clearly discerned. From both the donor's and
the recipient's point of view, project assistance has certain advan-
tages, I although the disbursement of assistance tends to be a more
lengthy process than in the case of non-project assistance.
Tied Assistance
Another characteristic of international assistance is that a great
deal of it is tied to the purchase of donors' goods (single tying) or
specific donors' commodities (double tying). Thus, even the aid
element of international assistance may not be without some costs to
the recipient country which reduce the value of aid. In the case of
assistance on commercial terms, these costs of tying are in addition
1 A. Carlin, 'Project v. Programme Aid', Economic Journal (Mar 1967).
to interest and amortisation payments. The greatest potential cost is

the opportunity cost involved if the donor's goods are not exactly of
the type required, or if there are cheaper markets in which to buy the
goods in question. These particular costs of tying are difficult to
estimate, but some idea of excess costs can be obtained from bids for
competitive tendering for World Bank loans and I.D.A. credits.
Bhagwatil has analysed the spread of bids on tenders from twenty
World Bank loans and three I.D.A. credits and found that the ratio
of the difference between high bids and successful bids to successful
bids was on average 49·3 per cent, which is a measure of the potential
excess cost. Haq2 has estimated for Pakistan a potential excess cost of
12 per cent, arid a figure of 15 per cent has been estimated by
Eshag3 on disbursements of approximately £20 million for transport
and harbour equipment in Iran. The average opportunity cost of
tying, just from the inability to buy in the cheapest market, would
seem to be of the order of 20 per cent, reducing the effectiveness of
assistance equivalently.
But there are many other costs of tying, apart from the inability of
the recipient to buy in the cheapest market. Projects to which aid is
tied may have been largely determined by the donor in the first place
and not square perfectly with the recipient's development plan. The
donor may attempt to raise unnecessarily the import content of
assistance, and suppliers may indulge in monopolistic exploitation.
The servicing of capital goods financed by donors' assistance may
also be high over the lifetime of the assets and involve a further
opportunity cost.
The only advantage of tied assistance from the recipient's point
of view is that the assistance might not be forthcoming at all if it
were not tied. Tied assistance is the only form of assistance which can
be guaranteed not to strain the donor's balance of payments in the
short run. With the international monetary instability which charac-
terised the decade of the 1960s, assistance from such countries as
1 J. Bhagwati, The Tying of Aid (Geneva: UNCTAD Secretariat, United

Nations, 1967), reprinted in J. Bhagwati and R. Eckaus (eds), Foreign Aid
(Harmondsworth: Penguin Books, 1970).
2 M. Haq, 'Tied Credits: A Quantitative Analysis', in J. Adler (ed.),
Capital Movements and Economic Development, International Economic Associa-

tion Conference (London: Macmillan, 1967).
3 E. Eshag, Study of the Excess Cosf of Tied Economic Aid Given to Iran
1966-67 (Geneva: UNCTAD, 1967).
America and Britain might have been substantially less had the less
developed countries not accepted assistance with strings attached.
But the losses to the donor from not tying are probably much less
than supposed. For one thing, donors may not gain much from tying
if recipient countries which already trade with donor countries in-
dulge in substitution and switching between goods tied to assistance
and goods already purchased from donor countries. Also, non-
recipient countries may benefit from assistance and increase their
demand for donors' exports.
Several proposals have been put forward in recent years for
mitigating the disadvantages of aid tying to less developed countries
without straining the balance of payments of donor countries. One
possibility is multilateral purchasing arrangements which allow
recipients to buy in the cheapest markets within a group of countries
so chosen that the losses and gains from untying aid would cancel out.
Another possibility is to institute bilateral arrangements between
donor countries whereby assistance from one donor to less developed
countries could be swapped for assistance from another donor
country, leaving tying levels intact but increasing the competitive
possibilities open to the recipient countries. Short of schemes of this
nature, bids from donor countries under tied aid could perhaps be
scrutinised by some international body, and recipients themselves
could perhaps improve their procurement practices by the collection
of more price information from alternative sources of supply.
While tied assistance involves costs, and militates against the most
effective use of assistance, the relevant question for the less developed
countries at the present time is whether tied assistance is better than
no assistance. The practical issue is as stark as this. Ninety per cent
of American bilateral assistance is tied, and roughly 60 per cent
of British assistance is either fully or partially tied. There seems little
prospect of these proportions being reduced in the foreseeable future.
There seems little hope of America or Britain taking a unilateral
lead in the hope that other countries will follow, even though their
foreign exchange earnings would probably benefit from an untying
of assistance in the world at large owing to their prominent position
as traders with less developed countries. It does seem particularly
pernicious, however, for donor countries both to tie assistance and
also to insist on the repayment ofloans in scarce currencies. If tying
is necessary to protect the balance of payments of donors, would it
not be possible for them to accept repayment ofloans in the recipient's
currency, easing the debt-service burden and stimulating trade

simultaneously? Alternatively, the excess costs imposed by donors on
recipients through aid tying could be shown as export subsidies rather
than aid.l This would not only bring aid figures closer to their
'true' value, but could also save interest payments on a part of the
tied assistance.
British Assistance to Less Developed Countries

Over the last ten years the amount of financial assistance given by
Britain to less developed countries has been consistently in the
region of £200 million per annum, but with a gradual upward trend
discernible. In 1965 the figure for gross official assistance was £196
million; in 1966, £209 million; and in 1968, £210 million. Mter
capital and interest payments the net transfer of resources amounted
to £145 million in 1965; £151 million in 1966; and £178 million in
1968. The figure for the net transfer of resources does not, of course,
represent the foreign exchange cost of assistance. Approximately 60
per cent of capital assistance is fully or partially tied to the purchase
of British goods, so that the foreign exchange cost ofBritish efforts to
help the less developed countries is probably not more than about
£80 million (ignoring inflows from past loans) -and probably sub-
stantially less, since all types of assistance also benefit our exports.
The greater part of assistance is bilateral. Of the £210 million dis-
bursed in 1968, only about £20 million represented donations to
multilateral organisations. About one-half of bilateral assistance
(£96 million) in 1968 consisted ofloans, one-quarter (£47 million)
grants and another quarter (£41 million) technical assistance. Of the
£96 million given in loans, approximately £60 million was granted
free of interest and one-half of the remaining loans at commercial
rates carried a waiver of interest for various periods. The effective
interest rate on all loans amounted to no more than 1 per cent in
1968, in contrast to the early 1960s when the average interest charge
was 6 per cent. The change is accounted for by the introduction of
interest-free loans in 1965. From the figures it can be seen that
technical assistance is an integral and growing part of the British
programme. In 1965 over 20,000 people, recruited or financed by the
United Kingdom, were serving in developing countries.
1 Bhagwati, The Tying of Aid.
The relative importance of different 'types' of assistance, and
detailed calculations of their net impact on our exports in the short
term, are shown in Table 8.3 below.l
TABLE 8.3
Types of aid % of bilateral Prima facie return (%)

aid, (i.e. direct impact on
1964--6 exports)
l. Capital aid:
(a) fully tied 36·9 100·0
(b) partially tied 18·8 60·0
(c) untied:
budgetary 13·8 3·5
other 14·4 61·0
2. Technical assistance:
(a) expenditure in the U.K. 5·1 100·0
(b) expenditure overseas ll·O 90·2
3. Total direct return 72·5
(weighted average of above)
Considering simply the direct effect of assistance on the balance

of payments, the above calculations suggest that for every £100 of
assistance, £72·5 'returns' in the form of increased exports. The
direct effect of assistance on exports, however, is not an adequate
measure of the true return to the balance of payments because cer-
tain indirect effects are ignored. Some of the assistance may be tied to
exports which would have been bought by the recipient country any-
way. In this case, assistance merely releases foreign exchange, some
of which may be spent on British exports but some of which may be
spent elsewhere. It is difficult to obtain a direct measure of the
extent of what may be called 'switching', but Hopkin puts the loss
at £18·9 million, thus lowering the true return below the prima facie
return by the same amount. Offsetting this, however, there are
bound to be induced imports from the spending of assistance in the
recipient country, some of which are likely to be bought from the
donor country. In the case of Britain this is estimated by Hopkin to
be £4·1 per £100 of assistance. Lastly, there will be reflection effects
as a result of assistance being spent in third countries to which the
1 From B. Hopkin and Associates, 'Aid and the Balance of Payments',
Economic Journal (Mar 1970).
donor is a supplier. Britain is estimated to benefit from this effect to

the extent of £4·9 per £100 of assistance. The effect of these three
sorts of indirect effects on the estimated return to Britain of £100 of
bilateral assistance may be summarised as follows, and a calculation
of the 'true' return made:
Prima facie return +£72-5
Effect of switching -£18·9
Induced imports + £4·1
Reflection effects + £4·9
= +£62·6
Ignoring repayments of loans and interest, the immediate foreign
exchange cost of bilateral assistance is only 37-4 per cent of the total
disbursements.
As far as geographical distribution is concerned, 70 per cent of
bilateral assistance goes to Africa and Asia and 90 per cent to colonial
or ex-colonial territories. In all, 116 countries received assistance in
1968, ranging from amounts ofless than £1,000 to £40 million in the
case of India - the largest recipient. To put the Indian sum in
perspective, however, it represents no more than 5p per head of
India's population.
Britain's balance of payments probably gains from international
assistance in general. Its exports benefit from assistance provided by
other countries. Britain provides approximately 8 per cent of total
official assistance from developed to less developed countries but
supplies 13 per cent of the exports from developed to less developed
countries. The World Bank has estimated that for every £1 Britain
gives the I.D.A. it receives £1·5 of export orders from projects
financed by I.D.A. credits. This, of course, is a statistical associa-
tion, not a causal connection.
Britain made an effort to maintain its assistance programme in the
1960s despite international monetary difficulties, culminating in the
devaluation of sterling in 1967, but its total contribution was no-
where near the target of I per cent of national income set for the
Development Decade 1960-70. In real terms the proportion of
national income devoted to assistance has probably declined slightly
over the period 1965-8, owing to rising prices. In 1968 the propor-
tion was approximately 0·45 per cent. To reach the I per cent target
would require a diversion of resources of some £160 million. In a
country with a per capita income of approximately £700 per annum
and a population of 53 million, this hardly represents an insurmount-
able challenge compared to the challenge of development.
Recent Trends in Savings and Investment in Less Developed Countriesl

The evidence of trends in savings and investment in less developed
countries indicates that there has been a widespread improvement in
the savings ratio in the post-war period, but with virtually all the gain
occurring in the 1950s. Between 1954 and 1967, 75 per cent of
countries classed as less developed increased their savings ratio,
mainly as a result of external assistance to finance widening trade
gaps. During the 1950s the rate of increase in gross capital formation
was 6 per cent per annum compared with 4 per cent per annum
between 1960 and 1963. This slackening of the rate of growth of
investment in the early 1960s was not due to a falling off in the rate
of growth of domestic savings but to a slackening of the rate of inflow
of capital from abroad- especially private capital- and a rise in the
rate of outflow of interest and profits. In 1964 the net outflow of
interest and profits from less developed countries was equivalent to
over one-half of the net inflow of grants and loans. The average
savings and gross investment ratios for less developed continents over
the period 1960-7 were as follows :
Savings Gross
investment
(% ofG.N.P.)
Mrica 13·1 16·7
South Asia 11·3 13·7
East Asia 11·0 15·6
South Europe 21·5 24·8
Latin America 16·3 17·7
Middle East 14·8 19·8
Weighted average 15·0 17·8
Industrialised countries 21·7 21·2

The gap between gross investment and domestic saving is made up

by foreign borrowing which, as can be seen, accounted on average
for 15 per cent of gross investment. The savings and investment
1 From United Nations, World Economic Survey, 1966 (New York, 1967).
ratios for less developed countries together still fall some way behind
the levels for industrialised countries, although the degree of diver-
gence varies considerably from country to country.
While higher savings rates are required to achieve' tolerable' rates
of growth of per capita incomes, recent experience suggests that such
an increase can only be effective if it is complemented by an adequate
supply of external purchasing power. If export structures cannot
guarantee this, there will be a continual dependence on foreign
borrowing, and the debt-servicing burdens that this implies.
Experience also suggests that the effectiveness of savings mobilisa-
tion depends on three main factors: (a) fiscal management; (b) the
growth of the business sector; and (c) the elaboration of mechanisms
enabling individuals to contribute to the functioning of a capital
market. We discussed earlier the role of the budget and taxation in
the financing of development, but the United Nations points to the
fact that a number of governments have been net dissavers in the
1960s. Moreover, attempts to widen the tax base have met with
resistance. Taxation through marketing boards has affected incen-
tives, and public enterprises' attempts to transfer resources from
consumption to investment have in fact supported consumption by
running at a loss. The experience of fiscal policy has not been
uniformly successful.
On the monetary front, the primary need is for institutions to
channel savings from the small saver to the potential entrepreneur.
Where institutions have developed, there seems to have been a
positive response among small savers to changes in interest rates.
The view of the United Nations on inflation, however, is that it has
been inimical to savings. The capital market is still rudimentary in
most less developed countries. Banks give mainly short-term credit,
and many of the newer institutions to finance activities in specific
sectors seem to be more concerned with lending money provided by
the government than with attracting savings. Experience points to
the need for at least one investment or lending agency from which
small and medium-sized enterprises can seek finance. The establish-
ment of such institutions would seem to be an essential step in the
evolution of local capital markets and could play a key role in
implementing development plans.
9
Trade and Development
In the previous chapter we attempted to establish the role of foreign
borrowing in the development process. Using dual-gap analysis, it
was shown that foreign borrowing can be used to bridge either a
domestic savings-investment gap or a foreign exchange gap, which-
ever is the larger. We saw that the policy issue is the decision on how
far foreign borrowing should go. How large can the import surplus be
without leading to severe future balance of payments difficulties in
the form of large outflows of debt repayments, and without generat-
ing inflation due to currency depreciation and the necessity to
develop high-cost import-substitute activities? The empirical evi-
dence is overwhelming that the conflict is very real between main-
taining an adequate growth rate and preserving a reasonable balance
on international payments. The ultimate solution must lie in improv-
ing the balance of payments through trade. The growth rates of
individual developing countries since 1950 correlate better with their
export performance than with almost any other single economic
indicator. The export performance of the less developed countries,
however, has continued to lag behind that of developed, industrial-
ised countries. Throughout the 1950s and 1960s the volume of
exports from less developed countries grew at a rate ofapproximately
5 per cent per annum compared with 8 per cent for developed
countries. The discrepancy in rates of growth was even wider in
value terms, causing the developing countries' share of the total
value of world trade to fall from 28 per cent in 1950 to 18 per cent
in 1968.
Although the export trade ofless developed countries is dominated
by primary products, the conception must be dispelled that the
world is neatly polarised into two camps: the underdeveloped world
producing and exporting solely primary products in exchange for
manufactures from developed countries, and the developed world
producing and exporting solely manufactures in exchange for primary

commodities from less developed countries. In practice, a good deal
of trade in both manufactures and primary products goes on among
the developed and less developed countries alike, with the developed
countries exporting substantial quantities of primary commodities
(especially temperate-zone foodstuffs) and the less developed coun-
tries exporting a few manufactured goods. Developed countries, in
fact, account for about 50 per cent of the world's supply of primary
products. In short, the distinction between less developed and de-
veloped countries is not wholly synonymous with the distinction
between primary producers and producers of manufactured goods.
This must be borne in mind in discussing the terms of trade - the
ratio of export to import prices. There is a distinction to be made
between the terms of trade for less developed and developed coun-
tries and the terms of trade for primary and manufactured goods.
The fact remains, however, that primary products do dominate
the balance of payments of most less developed countries, and that
the less developed countries' share of world trade in manufactures is
very small. Exports of primary products account for approximately
85 per cent of the export earnings of the less developed countries
taken together, and manufactured goods from less developed coun-
tries account for only about 5 per cent of world trade in manu-
factures. Moreover, the range of these traded goods is narrow, about
80 per cent consisting of textiles and light manufactures in which
world competition is intense.
Historically, trade has been an important mainspring of growth
for countries at different stages of development. In the nineteenth
century the countries that were industrialising had access to food
and raw materials in primary producing countries which allowed
the more developed countries to reap the gains from international
specialisation. The less developed countries, in turn, were assisted in
their development by the demand for raw materials and the inter-
national investment that followed in its train. The situation today is
very different. Most world trade takes place in industrial commodities
from which the less developed countries are largely excluded, and
the demand for less developed countries' traditional exports is slack
relative to the demand for industrial goods. Trade does not work to
the equal advantage of both sets of countries.
Three distinct factors have been at work in the developed coun-
tries retarding the growth of the exports of the less developed
TRADE AND DEVELOPMENT 279
countries. Firstly, the pattern of demand has shifted to goods with a
relatively low import content of primary commodities. Secondly,
technological change has led to the development of synthetic sub-
stitutes for raw materials. Thirdly, developed countries have pursued
protectionist policies which have retarded the growth of their im-
ports of both primary commodities and manufactured goods.
In view of these trading developments, and the emergence of a
foreign exchange gap as the constraint on growth in less developed
countries, there has been a complete rethinking by some economists
in recent years as to the lines on which trade should take place. The
balance of payments difficulties and foreign exchange shortage ofless
developed countries has led to a switch from viewing trade from the
traditional classical standpoint of resource allocation to viewing the
effects of trade on the balance of payments. It is balance of payments
difficulties, necessitating foreign borrowing if growth is to be sus-
tained, that has led to the cry in recent years of 'Trade, not aid '. 1
The problem facing less developed countries is not so much whether
to trade but in what commodities to trade and to ensure that the
terms on which they trade with the developed countries are favour-
able. There is no dispute that there are both static and dynamic
gains from trade. What is in dispute is whether the overall gains
could not be greater, and the distribution of gains between countries
fairer, if the pattern of trade was different from its present structure
and the developed countries were to modifY their trading policies
towards the developing world.
If the price elasticity of the exports of less developed countries is
low, and demand is slow in expanding, it would seem useless to push
factors of production into existing export activities. The effect of
increased output would be to worsen the terms of trade. This is the
notion of immiserising growth, with adverse movements in the terms
of trade from further specialisation offsetting the real income gains
from specialisation.2 So what are the less developed countries to do?
The answer would appear to be: industrialisation- the production of
1 Trade to improve the foreign exchange situation obviously involves no
future commitments of foreign exchange, while borrowing does. If foreign
assistance was tmly aid, of course, there would be no problem either. In this
sense the slogan 'Trade, not aid' is misleading; it must be accepted as a
cliche, but its meaning is clear.
2 This is the basis of J. Bhagwati's model of immiserising growth. See
'Immiserising Growth: A Geometrical Note', Review of Economic Studies
(June 1958).
industrial goods with a higher price and income elasticity of demand,

either for export or the home market, the latter implying import
substitution. Dynamic considerations about the demand for existing
export products and the utilisation of increases in factor supplies
do not diminish the case for international specialisation. What is
involved is a reconsideration of what lines of activity to pursue in
the face of changed circumstances, and a recognition of the distinc-
tion between established comparative advantage and incremental
comparative advantage. Before going on to consider trade strategy
for less developed countries in the light of these 'new' trade theories,
however, let us first establish more firmly the static and dynamic
gains from trade stressed by traditional theory.
The Gains from Trade

While it is quite legitimate to look at trade from the point of view of
the balance of payments, and to regard the balance of payments as
a development problem that can only be solved by new trade
policies, the benefit from trade in traditional trade theory is not
measured by the foreign exchange earned but by the increase in the
value of output and real income from domestic resources that trade
permits. Optimal trade policy, measured by the output gains from
trade, must be clearly distinguished from the maximisation of
foreign exchange earnings.
The gains from trade can be divided into static and dynamic. The
static gains are those which accrue from international specialisation
according to the doctrine of comparative advantage. The dynamic
gains are those which result from the impact of trade on production
possibilities at large. Economies of scale, international investment
and the transmission of technical knowledge would be examples of
dynamic gains.
We have established, however, that trade has balance of payments
effects as well as real resource (or output) effects, which can be con-
sidered as equally important for growth ifforeign exchange is a scarce
resource. In this section, therefore, we must do two things. Firstly,
we must establish the precise nature of the static and dynamic gains
from trade, and secondly, we must examine critically the underlying
assumptions of the comparative advantage doctrine and the classical
advocacy of international specialisation and free trade. We can then
go on to consider the argument that the balance of payments
implications of international specialisation and free trade may
seriously offset the allocative gains from trade, and whether this
establishes a case for protection.
The static gains from trade are based on the law of comparative
advantage. Consider the case of two countries A and B both with the
capacity to produce commodities X and r. The simple proposition of
classical trade theory is that if country A has the comparative
advantage in the production of commodity X, and country B has the
comparative advantage in the production of commodity r, it will be
mutual!J profitable for country A to specialise in the production of X
and for country B to specialise in the production of r, and for sur-
pluses of X and r in excess of domestic needs to be freely traded,
provided that the international rate of exchange between the two
commodities lies between the domestic rates of exchange. Compara-
tive advantage is an opportunity-cost concept measured by the
marginal rate of transformation between one commodity and
another as given by the slope of the production possibility curve.
Given perfect competition, which the above analysis assumes, the
domestic price ratio between two commodities will equal their
marginal rate of transformation. If this were not so, it would pay
producers to switch from one commodity to another to take advantage
of the relatively favourable price ratio.
Now let us give a practical example of the static gains from trade.
In Fig. 9.1 below, the production possibility curves for countries A
and B are drawn showing the different combinations of the two
goods X and Tthat can be produced with each country's given factor
endowments. We assume for simplicity that factors of production in
both countries are sufficiently versatile as to be able to produce either
commodity equally efficiently so that the production possibility
curves are linear; that is, there is a constant marginal rate of
transformation.
Now suppose that the marginal rate of transformation between X
and r in country A is 10/8, and in country B, 10/2. Commodity X is
relatively cheaper in country B than A measured by the amount of
commodity r that must be forgone, and r is relatively cheaper in
country A. We say that country A has the comparative advantage in
the production of r, and country B in the production of X, and that
it will be mutually advantageous for A to concentrate on Y, and B
r
on X, and for A to swop for X and for B to swop X for Y.
If the countries do start trading freely with each other, there can
only be one price ratio between X and r, which will be determined

by the interaction of demand and supply in both countries together,
and which will lie somewhere between each country's transformation
(price) ratio. Suppose that the international rate of exchange settles
X
lj \
\
\
\
\
Q \
\
\
\
"'
at 10/5, shown by the broken lines in Fig. 9.1. Country A, specialis-

ing in the production of r, can exchange r for more X internation-
ally than it could domestically, and country B, specialising in the
production of X, can exchange X for more r internationally than it
could domestically. Before trade, country A's transformation ratio
was the line qq2; after trade it can be seen that country A will receive
more units of X for every unit of r sacrificed domestically, as given
by the international price ratio r1q2 (i.e. h~ > :;2 ). Likewise,
country B's transformation ratio before trade was qq; after trade,
for every unit of X sacrificed domestically country B will receive
more units of r as given by the international price ratio qr
. PJIJ
( 1.e. PJS!)
qPJ > qPJ •
World production should certainly increase as a result of the
international division of labour. Specialisation on the basis of com-
parative advantage enables the maximum to be produced from a
given amount of factor resources. The increase in the value of output
that trade permits results from the opportunity to obtain foreign
products more cheaply, in terms of real resources forgone, than the
alternative of import substitution which means domestic production.
In Hicks's beautifully concise phrase, 'the gain from trade is the
difference between the value of things that are got and the value of
things that are given up'.l Through the international division of
labour one is supposed to get more than one gives up! If comparative
advantage was exactly the same in the two countries there would,
of course, be no point in specialisation and trade, except to reap
economies of scale and other dynamic gains.
Whether the consumption of both commodities in both countries
rises depends on the international rate of exchange. At some rates of
exchange, even between the two domestic price ratios, the consump-
tion of one of the commodities in one of the countries after trade may
be less than before trade. Even so, it can still be maintained that the
post-trade position is better than the pre-trade position, if those
individuals whose welfare is increased can fully compensate those
who suffer and still be better off.
The static gains from trade are the same gains as from trade
creation that accrue with the establishment of customs unions, when
high-cost suppliers are replaced by lower-cost suppliers. It is the
doctrine of comparative costs that provides the rationale for the
increasing advocacy of the formation of customs unions, or trading
areas, between less developed countries. It should be emphasised,
though, that there is nothing in the comparative cost doctrine that
ensures equality in the distribution of gains from trade creation
(which explains, in part,- why attempts in East Mrica and the West
Indies to establish free-trade areas have met with difficulties).
Now let us turn to the dynamic gains from trade. The major
dynamic benefit of trade is that export markets widen the total
market for a country's producers. If production is subject to in-
creasing returns, the total gains from trade will exceed the static
gains from a more efficient allocation of resources. With increasing
returns to scale, any country may benefit from trade irrespective of
the terms of trade. Hicks2 argues that it is impossible to make sense
of the phenomenon of international trade unless one lays great
1 J. Hicks, Essays in World Economics (Oxford: Clarendon Press, 1959)
p. 181.
2 Ibid.
stress upon increasing returns because of the close connection

between increasing returns and the accumulation of capital. For a
small country with no trade there is very limited scope for large-scale
investment in advanced capital equipment; specialisation is limited
by the extent of the market. But if the poor less developed country
can develop trade, there is some prospect of industrialisation and of
dispensing with traditional methods of production. The larger the
market, the easier capital accumulation becomes if there are increas-
ing returns to scale. In this respect large countries, like India, are in
a more favourable position than smaller countries such as Ceylon or
Jamaica. India's large population offers a promising basis for the
establishment of capital-goods industries and the production of
manufactured goods, since production can take place on a viable
basis before trade. The smaller country, however, may need sub-
stantial protection for a commodity before it can be produced
economically and compete in world markets.
Another dynamic benefit oftrade is that it provides underdeveloped
countries with capital goods which can alter and increase production
possibilities. We saw in the discussion of dual-gap analysis that
imports can be regarded as substitutes for domestic capital goods,
with the capacity to lower the overall capital-output ratio through
their superior efficiency and impact on the economy at large. Also,
trade in capital goods may bring with it the dissemination and
spread of technical knowledge. Finally, trade can promote efficiency
through competition, which is a dynamic benefit hard to quantify
but likely to be important in a community starting to develop
rapidly.
Now let us consider the potential disadvantages of adherence to
the comparative cost doctrine. Like most micro-welfare theory, the
comparative advantage/free-trade argument is a static one based
on restrictive, and very often unrealistic, assumptions. As a criterion
for the international allocation of resources it suffers from the same
static defects as the investment criteria of traditional micro-economic
theory that were discussed in Chapter 6. There may, in short, exist a
conflict between short-run allocative efficiency and long-run growth.
The doctrine assumes, for example, the existence offull employment
in each country (otherwise there would be no opportunity cost in-
volved in expanding the production of commodities) ; it assumes that
the prices of resources and goods reflect their opportunity cost (i.e.
that perfect competition exists), and that factor endowments are
given and unalterable. Moreover, the doctrine ignores the effect of
free trade on the terms of trade (movements in which affect real in-
come), and the dynamic feedback effects that trade itself might have
on comparative advantage. As a result, it can be argued that the
principles of comparative advantage and free trade are not very
useful concepts to employ in the context of less developed countries
which are in the throes of rapid structural change and which are
concerned more with long-term development than with short-term
efficiency. As many economists have commented, the doctrine of
comparative advantage is more useful in explaining the past pattern
of trade than for providing a guide as to what the future pattern of
trade should be as a stimulus to development.
The question is not whether there should be trade but whether
there should be free trade, as the doctrine of comparative costs
implies. Might the long-run needs of the less developed countries not
be better served by protection? Those who question the assumptions
of the comparative cost model, and who stress the relation between
development and the balance of payments, express the view that the
efficiency gains from free trade are unlikely to offset the tendency in
a free market for the comparative position ofless developed countries
to deteriorate vis-a-vis the developed countries. Free trade is
claimed to work to the disadvantage of the less developed countries
largely because of the nature of the products which these countries
seem destined to produce and trade under such a system. The
answer is said to lie in a change in the structure of production and
exports of the less developed countries, which can only be fostered by
the protection or subsidisation of new industries in the early stages of
their development.
The development considerations that the doctrine of free trade
overlooks are numerous. Firstly, it ignores the balance of payments
effects of free trade and the effect of free trade on the terms of trade.
If the demands for different commodities grow at different rates
owing to differences in their price and income elasticity of demand,
free trade will work to the benefit of some countries and to the
relative detriment of others. In short, free trade cannot be discussed
independently of the balance of payments and the terms of trade. In
classical theory, Viner 1 quotes Torrens, J. S. Mill, Marshall, Edge-
worth and Taussig as conceding that unilateral substitution by a
1 J. Viner, International Trade and Economic Development (Oxford: Clarendon
Press, 1953) p. 39.

country of free trade for protection would move the terms of trade
against the country. But most 'free-traders' ignored the subject. In
general, the implicit assumption was that movement from protection
to free trade would not alter the commodity terms of trade, or if it
did that the gains from trade would more than offset any unfavour-
able terms of trade effect. If the terms of trade effect does offset the
gains from trade, this would appear to provide a valid argument for
protection. This is one of the lines that modern protectionists take.
The case for protection of manufactured goods produced by less
developed countries is greater: the less the demand for existing
primary products is expected to grow and the lower the price elas-
ticity, the higher the internal elasticity of demand for manufactured
products from the outside world, and the less the likelihood of
retaliation by other countries.
A second factor which the free-trade doctrine tends to overlook is
that some activities are subject to increasing returns while others are
subject to diminishing returns. The commodities most susceptible
to diminishing returns are primary products, where the scope for
technical progress is also probably less than in the case of manu-
factured goods. This being so, one might expect a rise in the ratio of
primary to manufactured good prices, and diminishing returns
would not matter so much if the goods were price inelastic. In
practice, however, there has been a substitution of synthetic sub-
stitutes for primary products, and the terms of trade have deteriorated
partly because of substitution and partly because of the fact that
demand for primary commodities in general, in relation to supply,
has expanded much less than in the case of manufactured com-
modities. But whatever the movement in the terms of trade, it is
somehow perverse to recommend a trade and development policy
based on activities subject to diminishing returns, in the light of the
theory of cumulative causation that we discussed in Chapter 5.
A third disadvantage of adherence to the comparative advantage
doctrine is that it could lead to excessive specialisation on a narrow
range of products, putting the economy at the mercy of outside
influences. The possibility of severe balance of payments instability
resulting from specialisation may be damaging to development.
Fourthly, comparative cost analysis glosses over the fact that
comparative advantage may change over time, or that it could be
changed by deliberate policies for the protection of certain activities.
There is no reason why countries should be condemned to the pro-
duction and export of the same commodities for ever. If comparative
advantage is not given by nature, as the doctrine of free trade
seems to suggest, but can be altered, the case for protection is
strengthened.
It should also be remembered that the concept of comparative
advantage is based on calculations of private cost. But we have seen
that in less developed countries social costs may diverge markedly
from private costs, and that social benefit may exceed the private
benefit because of externalities. If private costs exceed social costs in
industry (because wage rates are artificially high, for example), and
social benefits from industrial projects exceed private benefits, there
is a strong argument for protecting industry in order to encourage
the transfer oflabour from agriculture into industry to equate private
and social cost and private and social benefit.
Lastly, it may be mentioned that the export growth of some
activities has relatively little seconda'ry impact on other activities.
Primary commodities fall into this category. The evidence is abund-
ant that the export growth of primary commodities has not had the
development impact that might be expected from the expansion of
industrial exports. The reasons are not hard to seek. Primary produc-
tion has very few backward or forward linkages; it is labour-intensive
and, historically, has tended to be undertaken by colonial powers
with a consequent outflow of profits. The secondary repercussions of
trade also tend to be overlooked by the free-trade doctrine.
New Trade Theories for Developing Countries: The Prebisch Doctrinel

Raul Prebisch was one of the first development economists toques-
tion the mutual profitability of the international division of labour
for developing countries on existing lines. Prebisch is one of those
who looks at the relation between trade and development from the
standpoint of the balance of payments rather than real resources. His
major claim is that the unfavourable impact of unrestricted trade on
the terms of trade and balance of payments ofless developed countries
1 The Economic Development of Latin America and its Principal Problems (New
York: Economic Commission for Latin America, U.N. Dept of Economic

Affairs, 1950). See also his later work, 'Commercial Policy in the Under-
developed Countries', American Economic Review, Papers and Proceedings (May
1959). For an excellent evaluation of the Prebisch thesis, see M.J. Flanders,
'Prebisch on Protectionism: An Evaluation', Economic Journal (June 1964).
far outweighs any advantages with respect to a more efficient

allocation of resources. His concern is with two distinct, but not
unrelated, phenomena. One is the transference of the benefits of
technical progress from the less developed to the developed coun-
tries. The second is the balance of payments effects of differences in
the income elasticity of demand for different types of products. He
divides the world into industrial 'centres' and 'peripheral' countries
and then conducts his analysis within the framework of the tradi-
tional two-country, two-commodity case of international trade
theory - equating the less developed countries with primary pro-
ducers (the 'periphery') and the developed countries with secondary
producers (the 'centre').
Technical Progress and the Terms of Trade

As we said earlier, in theory the barter terms of trade might be
expected to move in favour of the less developed countries. For one
thing, primary product production tends to be subject to diminishing
returns, and for another technical progress tends to be more rapid in
manufacturing industry than in agriculture. If prices are related to
costs, one would expect iri theory the ratio of primary product prices
to industrial-good prices to rise. According to Prebisch, however, the
ratio has fallen. He advances two explanations for this and hence why
the benefits of technical progress tend to flow from the less developed
to developed countries and not the other way round. His first ex-
planation concerns the relation between incomes and productivity.
He suggests that whereas factor incomes tend to rise with productivity
increases in developed countries, they rise more slowly than pro-
ductivity in the less developed countries owing to population pressure
and surplus manpower. Thus there is a greater upward pressure on
final goods' prices in developed than in less developed countries,
causing the ratio of prices to move in the opposite direction to that
suggested by the pace of technical progress. The second explanation
advanced is that there is some form of ratchet effect at work affecting
the relationship between primary product and manufactured-good
prices. Primary product prices tend to rise more than manufactured-
good prices during the upswing of the trade cycle, but fall more than
industrial prices during the downswing. As Flandersl points out, this
by itself, of course, would not cause a divergence between industrial
and primary product prices. What he must mean is that primary
1 Flanders, op. cit.
product prices fall more than industrial prices during the downswing
and by more than they rose during the upswing. This would produce
a ratchet effect with the trend of primary product prices diverging
from the trend of industrial-good prices as shown in Fig. 9.2 below.
Industrial prices
(I. P.)
Primary product
prices ( P. P.)
Time
Fw. 9.2
Furthermore, according to Prebisch, not only do the terms of trade
deteriorate, but if prices must be reduced to clear the market, export
earnings will fall if the demand for primary commodities is price
inelastic. This is a related sense in which technical progress is
'transferred' from the less developed countries to the industrial
'centres'.
The Income Elasticity of Demand for Products and the Balance of

Payments
The second phenomenon mentioned by Prebisch is the balance of
payments effects of differences in the income elasticity of demand for
different types of products. It is generally recognised and agreed
that the income elasticity of demand for most primary commodities
is lower than that for manufactured products. On average, the
elasticity is probably less than unity, resulting in a decreasing pro-
portion of income spent on those commodities (commonly known as
Engel's Law). In the two-country, two-commodity case the lower
income elasticity of demand for primary commodities will mean
that for a given growth of world income the balance of payments
of primary producing, less developed countries will automatically
deteriorate vis-a-vis the balance of payments of developed countries
producing and exporting industrial goods. A simple example will
illustrate the point.
Suppose that the income elasticity of demand for the exports of

less developed countries is 0·8 and that the growth of world income is
3-Q per cent; exports will then grow at 2-4 per cent. Now suppose
that the income elasticity of demand for the exports of developed
countries is 1·3 and the growth of world income is 3·0 per cent;
exports of developed countries will then grow at 3·9 per cent. Since
there are only two sets of countries, the less developed countries'
exports are the imports of developed countries and the exports of
developed countries are the imports of the less developed countries.
Thus less developed countries' exports grow at 2·4 per cent but
imports grow at 3·9 per cent; developed countries' exports grow at
3·9 per cent and imports at 2·4 per cent. Starting from equilibrium,
the balance of payments of the less developed countries automatically
worsens while the developed countries show a surplus. This has
further repercussions on the terms of trade. With imports growing
faster than exports in less developed countries, and the balance of
payments deteriorating, the terms of trade will also deteriorate
through depreciation of the currency, which may cause the balance
of payments to deteriorate even more if imports and exports are price
inelastic.
Moreover, this is not the end of the story if we take the per capita
income growth between developed and less developed countries. If
population growth is faster in less developed countries, the growth of
income must also be faster than in the developed countries if per
capita income growth rates are to remain the same. This will mean
an even faster growth rate of imports into less developed countries
and a more serious deterioration in the balance of payments. And if
the goal is to narrow relative or absolute differences in per capita
income between developed and less developed countries, the balance
of payments implications will be even more severe. In the example
previously given, which ignores differences in population growth, it
is easily seen that the price of balance of payments equilibrium is
slower growth for the less developed countries. If their exports are
growing at 2·4 per cent, with a 3 per cent income growth in the
developed countries, income growth in the less developed countries
must be restrained to 2·4/1·3 = 1·85 per cent for balance of payments
equilibrium (where 1·3 is the income elasticity of demand for
imports, i.e. manufactures). In the absence of foreign borrowing to
bridge the foreign exchange gap, or a change in the structure of
exports, the result of different income elasticities of demand for
primary and manufactured products is slower growth in the primary
producing countries- perpetuating the development 'gap'. In the
absence of protection, the only other alternative is deliberate
depreciation of the currency. This has several disadvantages. For one
thing the price elasticities of exports and imports may not be right
for foreign exchange earnings to be increased, and secondly, depre-
ciation will only tend to encourage production in existing activities,
the concentration on which may have contributed to the balance of
payments difficulties in the first place.
There are certain equilibrating mechanisms in existence which
may reverse the tendencies referred to, but they are likely to be weak
and fairly slow in operating. First of all, it cannot be assumed that
the industrial structure of the less developed countries will remain
static. Over time it is natural that the proportion of total resources
employed in the production of manufactured goods should increase,
decreasing the rate of increase of imports of manufactured goods.
Secondly, assuming that money income and population grow at the
same rate in both sets of countries, if the terms of trade are deteriorat-
ing for the less developed countries, then real per capita income
cannot be growing so rapidly in the less developed countries as in the
developed countries. Thus even with a high income elasticity of
demand for imports in less developed countries, the absolute incre-
ments in imports should eventually equal exports through a terms of
trade effect. Prebisch recognises this latter equilibrating mechanism,
but rejects reliance upon it because of the sacrifice of real growth that
it clearly involves.
For balance of payments and terms of trade reasons (which are not
unconnected), Prebisch therefore argues for the protection of certain
domestically produced goods and the virtual establishment of mono-
polist export pricing by less developed countries. Prebisch's balance
of payments argument reinforces the classical infant industry argu-
ment for protection to safeguard the terms of trade. Scitovskyl
showed long ago that in the two-country, two-commodity case a
country can always gain by levying certain tariffs to improve its
terms of trade which offset any losses from resource 'misallocation'
as a result of the tariff- provided, of course, the other country does
not retaliate. This is the notion of the 'optimum' tariff.
There are several benefits that Prebisch expects from protection.
1 T. Scitovsky, 'A Reconsideration of the Theory of Tariffs', Review of
Economic Studies (summer 1942).
Protection would enable scarce foreign exchange to be rationed

between different categories of imports, and could help to correct
balance of payments disequilibrium resulting from a high income
elasticity of demand for certain types of imports. Secondly, it could
help to arrest the deterioration in the terms of trade by damping
down the demand for imports; and thirdly, it provides the oppor-
tunity to diversify exports and to start producing and exporting
goods with a much higher income elasticity of demand in world
markets.l
Recent Trends in the Terms of Trade

Whether the terms of trade have moved unfavourably against the
less developed countries is an empirical question. The evidence
suggests that a decline set in in 1954, which continued well into the
1960s. The experience of individual countries obviously varies, but
taking the less developed countries as a group and comparing the
average ofl965-7 with that ofl953-5, we find that whereas the price
of manufactures purchased by the less developed countries rose 14
per cent, the price offood and raw materials sold by them fell 10 per
cent.2 For most of the post-war period there has been a deterioration
in the terms of trade of the less developed countries.
At this stage, however, we need to make the distinction between
the barter (or commodity) terms of trade (which is the usual mean-
ing of the terms of trade) and the income terms of trade, which is the
ratio of export prices to import prices times the quantity of exports
. Pm
( I.e. Px • 'n.
><-x ) , not JUSt
· t h e ratio
. of export to Import
. .
pnces. The m-
.
come terms of trade is thus a measure of the total purchasing power

of exports over imports.
From the point of view of development, measured by per capita
income, the income terms of trade is perhaps the more relevant con-
cept to consider than the barter terms of_trade. It may well be, for
1 One argument for protection which must be dismissed is that a cut in the
demand for imports through protection will reduce their price. If the sup-
plies of imports come from many industries in many countries, the demand
for industrial imports from any one country is unlikely to be greatly affected
by tariffs or import quotas and prices will remain unchanged. But even if
demand was affected significantly, prices could well rise if prices are cost-
determined, and productivity falls with wages sticky downwards.
2 C. Clark, 'Too Much Food?', Lloyds Bank Review (Jan 1970).
instance, that the price of exports falls relative to imports owing to
increased efficiency in the exporting country, and this releases
resources for further exports which subsequently expand more than
proportionately to the fall in price. The barter terms of trade would
have worsened, but development would be stimulated. It is also
worth remembering that when a country devalues its currency it
deliberately worsens its barter terms of trade in the hope that the
balance of payments will improve, providing scope for a faster
growth of real income through a rapidly improving income terms of
trade. On the other hand, if the demand for a country's exports is
inelastic, then a decline in the barter terms of trade will also mean
a deterioration in the income terms of trade, other things being
equal.
In the long run, if world trade is buoyant, every country can
experience an improvement in its income terms of trade. The ques-
tion is not who are the gainers and who are the losers, as in the case of
the barter terms of trade, but what are the relative rates of improve-
ment in the income terms of trade? The evidence for the period
1957-65 1 is that the income terms of trade for the developed
countries improved at the rate of 8·1 per cent per annum, while in
the less developed countries the rate of improvement was only 5·1 per
cent per annum. Since the quantity of exports from developed
countries grew at 7·2 per cent, and from less developed
countries at 5·6 per cent, this also confirms our earlier conclusion of a
decline in the barter terms of trade of the less developed countries.
When the income terms of trade for both sets of countries are
adjusted for differences in rates of growth of population, the less
developed countries appear even worse off. A decline in the per
capita income terms of trade is an added force contributing to the
maintenance of the development 'gap'.
Trade Theory and Dual-gap Ana!Jsis

Another dimension to the protectionist argument has recently been
added by Linder,2 drawing explicitly on dual-gap analysis. The
crucial question, argues Linder, is whether a less developed country's
1 T. Wilson, R. P. Sinha andJ. R. Castree, 'The Income Terms of Trade
of Developed and Developing Countries', Economic Journal (Dec 1969).

2 S. Linder, Trade and Trade Policy for Development (New York: Praeger,
1967).
economic potential can be fully utilised at the same time as equili-

brium externally is maintained. Free trade may not lead to the full
employment of resources for two reasons: firstly, because of factor
immobility; and secondly, because certain imports may be required
to achieve full utilisation of resources and these imports may exceed
exports. In our earlier discussion of dual-gap analysis in Chapter 8,
we saw that if the export-import gap is dominant and foreign ex-
change is scarce, domestic resources may go unutilised in the absence
of development policies to equate the export-import gap and the
savings-investment gap ex ante. Potential domestic saving will fall
either through a fall in the potential level of output or through a fall
in the propensity to save through a redirection of expenditure. One
obvious policy, however, is to devote more domestic resources to
import substitution or export promotion.
But Linder argues that in developing countries it may not be
possible to solve the problem of the dome.stic underutilisation of
resources through trade because of an export maximum. Classical
trade theory, however, does not admit this. The notion of an export
maximum is related to what Linder calls the theory of' representative
demand', which determines the relative price structure for goods.
The theory of representative demand states that the production
function for a commodity will be the more advantageous in a country
the more the demand for a commodity is typical of the economic
structure of a country compared with other countries. The chief
determinant of demand structure is per capita income, so that goods
in demand in advanced countries have unfavourable production
functions in less developed countries and vice versa. The less
developed countries are therefore faced with severe marketing prob-
lems if they are to develop by trade. The goods they are best at
producing are not demanded in developed countries, and they are
very inefficient at producing the goods that are demanded in
developed countries. Productivity may not be high enough to
support resources in the production of these goods, and input
imports for export production might absorb more foreign exchange
than the exports eventually yield.
In Linder's view these circumstances provide the case for protec-
tion to save foreign exchange and to enable the full utilisation of
domestic resources to take place. This contrasts with conventional
theory which does not allow for protection for balance of payments
reasons or to increase the effective demand for domestic products to
eliminate underemployment. Moreover, protection in this model
involves no allocation losses because if foreign exchange is required
to utilise domestic resources fully, the opportunity cost of resources
is zero. The only qualification Linder makes to his argument for
development based on import substitution and the expansion of
domestic demand is if value-added is negative; that is, if input
imports for domestic production involve a higher foreign exchange
cost than the importation of the end-products themselves. Except in
these circumstances, there is no conflict between allocation and
capacity considerations or allocative efficiency and import sub-
stitution as long as a foreign exchange gap exists. Import substitution
frees foreign exchange for input imports which allow the full
utilisation of domestic resources.
Trade Policies
Prebisch and Linder both provide powerful critiques of neo-classical
trade theory, but attack from different angles. While Linder argues
the case for protection for the full utilisation of domestic resources,
Prebisch argues the case for protection on the more 'orthodox'
grounds of improving the terms of trade, and as a substitute for
exchange depreciation to preserve simultaneous internal and
external equilibrium. Moreover, while Linder argues explicitly for
import substitution because of the existence of an export maximum,
Prebisch seems to be more optimistic about· growth through trade
and against 'inward-looking' development policies. New export
activities may, of course, require protection or subsidisation in the
early stages of their establish~ent, but there is a distinct difference
between identifying lines of activity in which to promote exports and
identifying lines of activity in which to develop import substitutes,
and we must briefly examine these different strategies. In the former
case, one is seeking out lines of comparative advantage; in the latter
case, one is attempting to reverse the pattern of trade by altering
comparative advantage.
If we accept the possibility of an export-import gap, because of
a lack of substitutability in the short run between imports and
domestic resources, Linder's argument for import substitution
basically reduces to pessimism over the chances of promoting exports
as an alternative means of fully utilising domestic resources. He
admits that import-substitute activities will themselves require
imports and presumably thinks, therefore, that, per unit of expansion

of domestic output, import substitution would save more foreign
exchange than could be earned by export promotion. The traditional
case for export promotion is that it allows growth to proceed without
foreign exchange difficulties and without suffering the allocative
losses from violating comparative advantage.
What trade policy should be adopted by any particular country
clearly cannot be decided a priori. Certainly not all less developed
countries fit the Linder model. Many of the Asian and South Ameri-
can countries do suffer from severe foreign exchange shortages, but
the Middle Eastern countries do not, and in many parts of Africa the
savings-investment gap would seem to be dominant. One must also
be a little suspicious of the concept of the export maximum. It is true
that the demand for less developed countries' exports tends to lag
behind that of the developed countries, but as our earlier figures on
the income terms of trade showed, the export record ofless developed
countries as a whole over the last two decades is not unrespectable.
This, in some measure, is due to the greater diversification of exports
than in the past, and there must be room for considerable improve-
ment. The great danger of import-substitution policies is that by
violating comparative advantage, increasing costs, and reducing
competitiveness, they may impair the long-run efficiency and export-
growth prospects of a country in the future.
The more lasting solution to balance of payments difficulties
would seem to be the promotion of manufactured-goods exports,
preferably with high price and income elasticities of demand, coupled
with policies to substitute domestic resources for imports. If the prices
of domestic and foreign resources reflected more accurately their
opportunity costs, the import content of many activities might be
reduced substantially. At the same time new areas of comparative
advantage must be sought. If trade is not the engine of growth, it is
probably a more desirable handmaiden than development by import
substitution.
The policy recommendation of export-led growth leads on to the
question of the trading relation between developed and less developed
countries. There are several ways in which the developed countries
can contribute to a solution of the balance of payments problems of
less developed countries, especially by helping export earnings. The
means of help fall into two main categories which we shall consider
in turn: firstly, the granting of trade preferences; and secondly,
international commodity agreements to stabilise or increase earnings
from exports.
Trade Preferences and Effective Protection

The main pressure group for trade preferences for less developed
countries' exports is the United Nations Conference on Trade and
Development (UNCTAD), which was first convened in Geneva in
1964. It was in this year that the Conference was converted into a
permanent organisation consisting of 132 member countries, with a
55-member Directing Board and a permanent Secretariat under
Prebisch as Secretary-General (who previously headed the U.N.
Commission for Latin America). There was a further meeting of
UNCTAD in New Delhi and another in 1971. Apart from these
periodic meetings, the organisation exists as a continuous pressure
group with the aim of assisting less developed countries through
trade and aid - primarily trade. Included among its objectives are:
greater access to the markets of the developed countries through the
reduction of trade restrictions; more stable commodity prices;
assistance from the developed countries equivalent to 1 per cent of
their national income; and compensation for less developed countries
whose foreign exchange earnings fall below 'expectations' owing to
deterioration of the terms of trade. Its main platform, however, is for
a system of tariff preferences for the less developed countries' exports
of manufactured and semi-manufactured goods on the lines of the
Commonwealth preference agreement, or in the form of quotas of
duty-free imports. In this respect there is an important difference
between UNCTAD and other bodies merely concerned with non-
discriminatory reductions in barriers to world trade, e.g. the
General Agreement on Tariffs and Trade (GATT).
In arguing for lower tariffs and tariff preferences, a distinction
needs to be made between nominal tariffs and effective rates of
protection. It is now widely recognised, in theory and in practice,
that nominal tariffs on commodities are not the appropriate basis for
assessing the restrictive effect of a tariff structure on trade. What is
relevant is the effective rate of protection on productive processes;
that is, on the value-added of those processes. Suppose, for example,
that a developed country allows the import of a primary product
duty free but imposes a high tariff on goods for which the primary
product is an input, i.e. on the process of these products. The free
import of the primary product increases the effective rate of protection

of the processed product, above what it would be if the raw product
were subject to the same tariff rate, by allowing cheaper domestic
production of the processed product. The effective rate of protection
may be extremely high if the cost of processing is small compared
with the cost of the duty-free raw materials. The effective rate of
protection is greater than the nominal protection the lower the tariff
on the inputs of raw materials. Suppose, for example, that the value-
added of the value of a final product is 50 per cent. If there was no
tariff on inputs but a 30 per cent tariff on the final product, this
would mean a 60 per cent tariff on value-added. What this amounts
to is that the more the developed countries get their inputs duty free
from the less developed countries, the greater the protection the
developed countries receive on their manufactured goods against
imports from less developed countries. Less developed countries, by
pressing for a lowering of tariff barriers on primary products, are, in
effect, increasing the effective rate of protection against their manu-
factured goods. And ifless developed countries want to protect their
own industries they must import inputs duty free.
The argument will become clearer if we give a little formal
analysis. The effective rate of protection of industry X is defined as
P _ V'x- Vv
x- Vx
where V' x is domestic value-added under protection and Vx is
value-added under free market conditions. Domestic value-added is
equal to the sales of the industry's product minus the sum of inter-
mediate inputs valued at domestic market prices. The free market
value-added can be defined identically, but with final product and
input prices valued at world prices. The less the degree of protection
of inputs, the higher the domestic value-added (V'x) and the higher
the effective rate of protection. In short, tariffs, like taxes, lower
domestic value-added. At the other extreme, if a country taxes
raw material imports but imposes no tariff to protect finished
products, the effective rate of protection is negative because the
domestic value-added will be less than the free market value-
added.
Let us now give a practical example. Suppose Indian textiles have
a world price of $5, of which $3 represents raw material costs and
$2 represents value-added. Now let us suppose that imports of
Indian textiles into a developed country are subject to a tariff of 20
per cent while domestic producers must pay a tariff of I 0 per cent on
textile raw materials. To remain competitive, the domestic producer
must produce the commodity for not more than $6. The value-
added can be $6 minus the cost of raw materials plus the tariff on
raw materials, i.e. $6- $3 - $0.30 = $2.70. The effective rate of
protection is the difference between the domestic value~added and
value-added in the less developed country expressed as a percentage
ofvalue-added in the less developed country, i.e. $ 2 · 7 ~;- $2 = 35
per cent. This is the effective rate of protection equal to the differ-
ence between the gross subsidy on value-added provided by the
tariff on the final product (;~ = 50 per cent) and the implicit
tax on value-added as a result of the tariff on raw materials

($~: 0 = 15 per cent).
The theory of effective rates of protection suggests that the same
nominal tariff cuts mean different degrees of change in effective rates
of protection, and it may thus be unwise for the less developed
countries to press for across-the-board tariff cuts on all commodities.
Reductions in tariffs against their primary products will increase the
effective rate of protection against their manufactures which, we have
argued, are more important exports as far as long-run development
prospects are concerned. The average nominal level of protection
in developed countries is about 12 per cent, but effective protection
against the goods of less developed countries may well be in the
region of 30 per cent.
The argument for reductions in tariffs against final manufactured
goods is unexceptional, but it is doubtful whether there is much
scope for preferences to contribute substantially to an increase in
export earnings. Export prices would have to be less than the domes-
tic price in the preference-giving country and must not exceed the
competitor's price by more than the amount of the preference. On
the assumption that the tariff rate on manufactured goods in the
advanced industrial countries is now between 10 and 15 per cent,
and that the advanced countries were willing to give 50 per cent
preference, this would mean a preference margin of only 5-8 per
cent. Industries in which a preference margin of 5-8 per cent would
enable less developed countries to take markets away from domestic
producers and developed country competitors are likely to be few

and far between.l
But it is not only tariffs that restrict trade. There are many
non-tariff barriers to trade in manufactures between developed and
less developed countries, the removal of which might contribute
more to increasing export earnings than pure reductions in tariff
barriers, e.g. licensing requirements, quotas, foreign exchange
restrictions, procurement policies favouring domestic products, anti-
dumping regulations, subsidies to exports in developed countries,
and so on.
International Commodity Agreements

International commodity agreements can perform two important
tasks. Firstly, they can stabilise prices so that foreign exchange
receipts do not fluctuate widely (provided supply is fairly stable).
Secondly, they can be designed to ensure that export earnings main-
tain (or increase) their purchasing power in terms ofindustrial goods,
either through price support schemes or income compensation
schemes. In short, the basic purpose of international commodity
agreements is twofold: firstly, to even out fluctuations in foreign
exchange earnings which, it is argued, may be inimical to growth;
and secondly, to compensate for any deterioration in the terms of
trade. Since the Second World War only five international com-
modity agreements have been successfully negotiated - for coffee,
sugar, wheat, tin and olive oil. The coffee scheme was of a price
support nature; the wheat and tin agreements have been concerned
with price stability; the sugar agreement was also concerned with
price stability but broke down at the time of the Cuban crisis ofl962,
and the olive oil agreement is merely designed to encourage the
growth of consumption and fair competition. In the early 1950s the
primary role of international commodity agreements was seen simply
as price stabilisation. The United Nations Report on Measures for
International Economic Stability remarked in 1951, for instance: 'A
fundamental principle of all schemes should be that, save in excep-
tional cases, they do not attempt to make the average price over a
period of years higher or lower than it would otherwise have been.
Their objective should be merely to reduce fluctuations around the
1 See G. Patterson, 'Would Tariff Preference Help Economic Develop-
ment?', Iloyds Bank Review (Apr 1965).

long-run trend.' But circumstances have changed since the early
1950s. This was the beginning of the steep decline in the terms of
trade of primary products, and of the primary product-producing
countries, and in the late 1950s international commodity agreements
came to be seen in their other role of providing a form of international
assistance by keeping prices and export earnings above their trend
level. Today the emphasis is on the maintenance of import purchas-
ing power rather than price stability as such.
Primary product prices do tend to be more unstable than the
prices of manufactured goods, but it appears to be a myth that the
export earnings of less developed countries are any more volatile
than the export earnings of the developed countries.! Moreover,
there appears to be little evidence that the instability of export
earnings between countries has been negatively associated with
differences in rates of growth. On the other hand, small fluctuations
in export earnings are capable of offsetting the total value offoreign
assistance to less developed countries in any one year. Approximately
a 5 per cent fall in export prices would be equivalent to the entire
annual inflow of private and public capital to less developed
countries. Stability, it would appear, is at least as important as foreign
assistance.
In general, the instability of export proceeds is the joint product
of variations in price and quantity. Large fluctuations in earnings
may be causally related to four factors: (l) excessive variability of
supply and demand; (2) the low price elasticity of supply and
demand; (3) excessive specialisation on one or two commodities; and
(4) the concentration of exports in particular markets. Macbean's
study examines these causal factors for a selection of less developed
countries. He finds variations in supply as the major determinant of
the instability of export earnings. If the source of instability does
come from the side of supply, stabilising prices will not, of course,
stabilise earnings. It will positively impair them in times of scarcity
and boost them in periods of glut. If there is perpetual over-supply,
and demand is price inelastic, price stabilisation will maintain
earnings, but price stabilisation will further encourage supply and
may lead to inefficiency in production if producing countries are
allocated production quotas regardless of efficiency in the production
of the commodity in question. This is not to argue that there is not a
1 A. Mac bean, Export Instability and Economic Development (London: Allen
& Unwin, 1966).
case for compensation, but that methods should be avoided which

encourage overproduction or inefficiency. It would be better to let
prices find their market level and for the producing countries to be
compensated by the beneficiaries under long-term agreements, with
the compensation used to encourage some producers into other activi-
ties. Alternatively, income compensation schemes could be worked
out, especially in cases where export instability was due to variations
in domestic supply.
In practice, though, several alternative methods of price stabilisa-
tion have been tried or recommended, including buffer stock schemes,
export restriction schemes, and price compensation schemes, and we
must examine these briefly.
Buffer Stock Schemes

Buffer stock schemes operate by buying up the stock of a commodity
when its price is abnormally low and selling the commodity when its
price is unusually high. The success of such schemes rests on the fore-
sight of those who manage them. Purchases must be made when
prices are low relative to future prices and sold when prices are high
relative to future prices. Clearly, buffer stock schemes are only suit-
able for evening out price fluctuations. They cannot cope with per-
sistent downward trends in price without accumulating large stocks
of the commodity which must be paid for - and presumably sold
in the future at still lower prices. Storage schemes are also only
appropriate for goods that can easily be stored, and for which the cost
of storage is not excessive.
Restriction Schemes
As opposed to buffer stock schemes, which are concerned with
stabilising prices, restriction schemes are more concerned with
maintaining the purchasing power of commodity prices in relation to
industrial goods; that is, in preventing a deterioration in the terms
of trade of primary commodities. The essence of a restriction scheme
is that major producers or nations (on behalf of producers) get
together and agree to restrict the production and export of a good
whose price is falling, thus maintaining or increasing (if demand is
inelastic) revenue from a smaller volume of output. In practice, it is
very difficult to maintain and supervise a scheme of this nature,
largely because it becomes extremely attractive for any one producer
or nation to break away from, or refuse to join, the scheme. This is
something Prebisch overlooks in his recommendation for monopoly
exporting pricing. It is very convenient to conduct theoretical
analysis in terms of two countries and two commodities, but when
it comes to practical policies the reality of the existence of many
countries must be contended with. The disadvantages of restriction
schemes are firstly that it is by no means certain that demand is not
elastic in the long run, so that raising price by restricting supply may
reduce export earnings in the long run. Restriction schemes may
ultimately lead to substitution for the product, and falling sales.
Secondly, restriction schemes can lead to substantial resource alloca-
tion inefficiencies stemming from the arbitrary allocation of export
quotas between countries, and production quotas between producers
within countries, unless the quotas are revised regularly to take
account of changes in the efficiency of production between producers
and between regions of the world. There is also a danger with any
form of price support scheme of a multilateral nature, where both
developed and less developed countries produce the good in question,
that 'assistance' does not all go where it is most needed. In this event
there is a greater case for bilateral arrangements over commodities
between developed and less developed countries, rather than schemes
which embrace developed countries which subsequently reap the
benefit.
Price Compensation Agreements

Price compensation agreements lend themselves to this form of bi-
lateral arrangement, an example of which has been worked out by
Professor Meade.l He has suggested a scheme whereby if the price of
a commodity falls, two countries could agree upon a sliding scale of
compensation such that the importing country pays an increasing
sum of money to the exporter as the price falls below a 'normal' price
specified in advance. The sliding scale of compensation could be
applied to the deviations of the actual price from the 'normal' price.
Since restrictions on output and quotas are no part of the scheme,
arrangements of this kind have the beauty of divorcing the efficiency
1 J. Meade, 'International Commodity Agreements', Lloyds Bank Review
(July 1964).
S04 GROWTH AND DEVELOPMENT
aspects of pricing and commodity arrangements from the distribu-

tional aspects. The commodity would be traded at world prices, and
the lack offull compensation would ensure that if world prices were
falling some countries would decide to shift resources, so maintaining
some degree of allocative efficiency.
Agreements could not only be bilateral. It would also be possible
to draw up multilateral price compensation agreements, with the
governments of all exporting and importing countries of a com-
modity agreeing jointly on a standard price, and on a 'normal'
quantity of imports and exports of the commodity for each country
concerned. There could also be a common sliding scale of compen-
sation. For reasons mentioned earlier, however, there is perhaps a
greater case for bilateral deals so that assistance can be given where
it is most needed and where countries are not bound by the conven-
tions of an international agreement.
There is no reason why price compensation schemes should not
run concurrently with other types of international commodity agree-
ments. Indeed, if there is a continually declining price of a com-
modity, it may be necessary to couple a restriction scheme with a
price compensation scheme, otherwise importing countries will be
persistently subsidising the exporting countries, which may not be
welcomed. There is also the danger in this case, and also in the case
of price support schemes, that one form of assistance will replace
another. If developed countries continually have to pay higher than
market prices for their primary products, and argue at the same time
that the major constraint on financial assistance to less developed
countries is their balance of payments, they may use price compensa-
tion agreements as an excuse for cutting other forms of assistance.
If so, what primary producers gain in the form of higher prices or
higher export earnings than if the market was free, they lose in
other ways.
If fluctuations in price emanate from the supply side, and not from
changes in demand, price compensation will operate perversely on
the stabilisation of export earnings. This is illustrated in Fig. 9.3
below. Price in the market is determined by the intersection of the
supply and demand curves, D1D1 and S1S1, giving equilibrium price
P1. Now suppose that there is a decrease in demand to D 2Dz, causing
price to fall to Pz. Earnings before the price fall were OP1XS1; after
the price fall, OPzX1S1. Assume P1 to be the 'normal' price agreed
on under the price compensation scheme, and that PzC represents
the appropriate amount of price compensation in relation to the
deviation from 'normal' price following the decrease in demand.
Total revenue under the price compensation scheme will be
occlsl, which is not far short of total revenue before the price fall.
Dl Sz
sl I
Price 02 I
\ I
'' I
I
'' I
I
.c; '' I 'Normal' price
c ' ,cl
-------,--
p2
' \XI
Dl
0 sl s2
Quantity supplied and demanded
FIG. 9.3
Consider, however, an equivalent fall in price from P 1 to P 2 as a

result of an increase in supply from S1S1 to S 2S2 • Under the same
price compensation scheme total revenue is now occ2s2, which is
greatly in excess of the original total revenue, before the price fall,
of OP1XS1. Conversely, if the supply falls, and the price rises above
the 'normal' price, revenue will be less than before the price rise
since the exporting country would presumably be compensating the
importing country - unless the scheme only works one way! Mac-
bean's work suggests that the major cause of price variations is shifts
in supply.
Income Compensation Schemes

The only way to overcome the induced instability of price compensa-
tion schemes is to formulate income compensation schemes which
take account of both price and quantity changes. The practical
difficulty is reaching agreement on a 'normal' level of income. With
the trend rate of growth of output positive for most commodities, to
settle for a fixed level of 'normal' income would be unjust. Each
year's compensation could perhaps be based on deviations of actual

export earnings from the moving average of a series of previous years.
The I.M.F. have a scheme in existence of the income stabilisation
type which operates on these lines. The I.M.F. have made special
arrangements to enable countries whose export earnings decline as a
result of fluctuations in the trade of primary commodities to borrow
for short periods. Countries may borrow readily if export earnings
fall by a certain percentage below the average for the previous five
years. Income compensation is another type of international assist-
ance, and as a form of aid is much to be preferred to commodity
schemes.
References and Further
Reading
Chapter 1
S. ANDIC and A. PEACOCK, 'The International Distribution of Income,
1949 and 1957', Journal of the Royal Statistical Society, part 2 (1961).
P. BAUER and B. YAMEY, 'Economic Progress and Occupational Distribu-
bution ', Economic Journal (Dec 1951).
W. BECKERMAN and R. BAcON, 'International Comparisons of Income
Levels: A Suggested New Measure', Economic Journal (Sep 1966).
A. K. CAIRNCRoss, 'Essays in Bibliography and Criticism, XLV: The
Stages of Economic Growth', Economic History Review (Apr 1961).
H. CHENERY, 'Patterns of Industrial Growth', American Economic Review
(Sep 1960).
--and L. TAYLOR, 'Development Patterns: Among Countries and Over
Time', Review of Economics and Statistics (Nov 1968).
C. CLARK, The Conditions of Economic Progress (London: Macmillan, 1940).
E. DOMAR, 'Expansion and Employment', American Economic Review,
(Mar 1947).
A. G. B. FISHER, 'Capital and the Growth of Knowledge', Economic
Journal (Sep 1933).
- - , 'Production: Primary, Secondary and Tertiary', Economic Record
(June 1939).
C. FuRTADO, Development and Underdevelopment (Berkeley: University of
California Press, 1964).
E. HAGEN, 'Some Facts about Income Levels and Economic Growth',
Review of Economics and Statistics (Feb 1960).
R. HARROD, 'An Essay in Dynamic Theory', Economic Journal (Mar 1939).
- - , Towards a Dynamic Economics (London: Macmillan, 1948).
J. HICKS, 'Growth and Anti-Growth', Oxford Economic Papers (Nov 1966).
M. A. KATOUZIAN, 'The Development of the Service Sector: A New
Approach', Oxford Economic Papers (Nov 1970).
S. KuzNETS, 'Notes on the Take-off', in W. W. Rostow (ed.), The Economics
of Take-off into Sustained Growth (London: Macmillan, 1963).
--,Economic Growth and Structure (London: Heinemann, 1965).
- - , 'International Differences in Income Levels', in B. Okun and R.
Richardson (eds), Studies in Economic Development (New York: Holt, Rine-
hart & Winston, 1961).
308 REFERENCES AND FURTHER READING
F. MACHLUP, 'Disputes, Paradoxes and Dilemmas Concerning Economic

Dev~lopment', in Essays in Economic Semantics (New York: Norton, 1967).
A. MAIZELS, Industrial Growth and World Trade (Cambridge University
Press, 1963).
B. 0KUN and R. RICHARDSON, 'Economic Development: Concepts and
Meaning', in B. Okun and R. Richardson (eds), Studies in Economic
Development (New York: Holt, Rinehart & Winston, 1961).
S. J. PATEL, 'The Economic Distance between Nations: Its Origins,
Measurement and Outlook', Economic Journal (Mar 1964).
W. W. RosTow, The Stages of Economic Growth (Cambridge University Press,
1960).
H. THEIL, 'International Inequalities and General Criteria for Develop-
ment Aid', International Economic Papers, no. 10 ( 1960).
A. P. THIRLWALL, 'The Development "Gap"', National Westminster Bank
Review (Feb 1970).
D. UsHER, Rich and Poor Countries, Eaton Paper No. 9 (London: Institute
of Economic Affairs, 1966).
- - , The Price Mechanism and the Meaning of National Income Statistics
Chapter 2
M. ABRAMOVITZ, 'Resource and Output Trends in the United States
since 1870', American Economic Review, Papers and Proceedings (May 1956).
M. BROWN, On the Theory and Measurement of Technical Change (Cambridge
--and J. DE CANI, 'Technological Change in the U.S., 1950-1960',
Productivity Measurement Review (May 1962).
C. CoBB and P. DoUGLAs, 'A Theory of Production', American Economic
Review, supplement (Mar 1928).
E. DENISON, The Sources of Economic Growth in the U.S. and the Alternatives
bifore Us (New York: Committee for Economic Development, Library of
Congress, 1962).
- - , 'The Unimportance of the Embodied Question', American Economic
Review (Mar 1964).
- - , Why Growth Rates Differ: Postwar Experience in Nine Western Countries
(Washington: Brookings Institution, 1967).
P. DouGLAS, 'Are There Laws of Production?', American Economic Review
(Mar 1948).
A. GAATHON, Capital Stock, Employment and Output in Israel, 1950-1959
(Jerusalem: Bank oflsrael, 1961).
J. HicKs, Capital and Growth (Oxford University Press, 1965).
C. KENNEDY and A. P. THIRLWALL, 'Surveys in Applied Economics:
Technical Progress', Economic Journal (Mar 1972).
A. MADDISON, Economic Progress and Policy in Developing Countries (London:
B. MASSELL, 'A Disaggregated View of Technical Change', Journal oj
Political Economy (Dec 1961).
REFERENCES AND FURTHER READING 309
R. NELSON, 'Aggregate Production Functions and Medium Range Growth
Projections', American Economic Review (Sep 1964).
R. SoLow, 'Technical Change and the Aggregate Production Function',
Review of Economics and Statistics (Aug 1957).
- - , 'Investment and Technical Progress', inK. Arrow, S. Karlin and
P. Suppes (eds), Mathematical Methods in the Social Sciences (Stanford
- - , 'Technical Progress, Capital Formation and Economic Growth',
American Economic Review, Papers and Proceedings (May 1962).
A. P. THIRLWALL, 'Denison on "Why Growth Rates Differ"', Moneta e
Credito (Banca Nazionale del Lavoro, July 1969).
J. G. WILLIAMSON, 'Production Functions, Technological Change and the
Developing Economies: A Review Article', Malayan Economic Review
(Oct 1968).
Chapter 3
T. BALOGH, 'Agriculture and Economic Development', Oxford Economic
Papers (Feb 1961).
R. A. BERRY and R. SoLioo, 'Rural-Urban Migration, Agricultural
Output and the Supply Price of Labour in a Labour-Surplus Economy',
S. ENKE, 'Economic Development with Unlimited and Limited Supplies of
Labour', Oxford Economic Papers (June 1962).
J. HARRIS and M. ToDARO, 'Migration, Unemployment and Develop-
ment: A Two-Sector Analysis', American Economic Review (Mar 1970).
B. joHNSTON and J. MELLOR, 'The Role of Agriculture in Economic
Development', American Economic Review (Sep 1961).
B. F. JOHNSTON, 'Agriculture and Structural Transformation in Develop-
ing Countries: A Survey of Research', Journal of Economic Literature (June
1970).
D. JORGENSON, 'Testing Alternative Theories of the Development of a
Dual Economy', in I. Adelman and E. Thorbecke (eds), The Theory and
Design of Economic Development (Baltimore: Johns Hopkins Press, 1966).
H. LEIBENSTEIN, Economic Backwardness and Economic Growth (New York!
Wiley, 1957).
A. LEWIS, 'Economic Development with Unlimited Supplies of Labour',
Manchester School (May 1954).
- - , 'Unlimited Supplies of Labour: Further Notes', Manchester School
(Jan 1958).
W. E. MooRE, 'Labour Attitudes towards Industrialisation in Under-
developed Countries', American Economic Review, Papers and Proceedings
(May 1955).
M. PAGLIN, '"Surplus" Agricultural Labour and Development', American
G. RANIS and J. FEI, 'A Theory of Economic Development', American
W. RoBINSON, 'Types of Disguised Rural Unemployment and Some
Policy Implications', Oxford Economic Papers (Nov 1969).
T. ScHULTz, Transforming Traditional Agriculture (New Haven: Yale
--,Economic Growth and Agriculture (New York: McGraw-Hill, 1968).
A. S.HONFIELD, Attack on World Poverty (New York: Random House, 1960).
H. SouTHWORTH and B. joHNSTON (eds), Agricultural Development and
Economic Growth (Ithaca, N.Y.: Cornell University Press, 1967).
M. ToDARO, 'A Model of Labour Migration and Urban Unemployment in
Less Developed Countries', American Economic Review (Mar 1969).
J. UPPAL, 'WorkHabitsandDisguised Unemployment in Underdeveloped
Countries: A Theoretical Analysis', Oxford Economic Papers (Nov 1969).
Chapter 4
K. ARROW, 'The Economic Implications of Learning by Doing', Review of
Economic Studies (June 1962).
T. BALOGH and P. STREETEN, 'The Coefficient of Ignorance', Bulletin of
the Oxford Institute of Statistics (May 1963).
G. BECKER, Human Capital (New York: Columbia University Press, 1964).
M. BLAUG, 'The Rate of Return on Investment in Education in Great
Britain', Manchester School (Sep 1965).
M. BowMAN, 'Schultz, Denison and the Contribution of Education to
National Income Growth', Journal of Political Economy (Oct 1964).
F. HARBISON and C. MYERS, Education, Manpower and Economic Growth
(New York: McGraw-Hill, 1964).
R. HARROD, Towards a Dynamic Economics (London: Macmillan, 1948).
J. HicKs, The Theory of Wages (London: Macmillan, 1932).
H. G. JOHNSON, 'Comparative Cost and Commercial Policy Theory for a
Developing World Economy', Pakistan Development Review, supplement
(spring 1969).
H. LEIBENSTEIN, 'Incremental Capital-Output Ratios and Growth Rates
in the Short-Run', Review of Economics and Statistics (Feb 1966).
A. LEwis, The Theory of Economic Growth (London: Allen & Unwin, 1955).
--,Development Planning (London: Allen & Unwin, 1966).
W. MILLER, 'Education as a Source of Economic Growth', Journal of
Economic Issues (Dec 1967).
W. REDDAWAY, The Development of the Indian Economy (London: Allen &
Unwin, 1962).
W. W. RosTow, The Stages ofEconomic Growth (Cambridge University Press,
1960).
T. ScHULTz, 'Investment in Human Capital', American Economic Review
(Mar 1961).
J. ScHUMPETER, The Theory of Economic Development (Cambridge, Mass.:
Harvard University Press, 1934).
--,Capitalism, Socialism and Democracy (London: Allen & Unwin, 1943).
J. VANEK and A. STUDENMUND, 'Towards a Better Understanding of the
Incremental Capital-Output Ratio', Q.uarterry Journal of Economics (Aug
1968).
Chapter 5
C. CLARK, 'The "Population Explosion" Myth', Bulletin of the Institute of
Development Studies (University of Sussex, May 1969).
M. DIAMOND, 'Trends in the Flow oflnternational Private Capital, 1957-
1965', I.M.F. Staff Papers (Mar 1967).
S. ENKE, 'The Economic Aspects of Slowing Population Growth', Economic
Journal (Mar 1966).
E. HAGEN, 'Population and Economic Growth', American Economic Review
(June 1959).
B. HIGGINS, 'The "Dualistic Theory" ofUnderdeveloped Areas', Economic
Development and Cultural Change (Jan 1956).
A. HIRSCHMAN, Strategy Q/ Economic Development (New Haven: Yale
N. KALDOR, 'The Case for Regional Policies', Scottish Journal of Political
Economy (Nov 1970).
H. LEIBENSTEIN, Economic Backwardness and Economic Growth (New York:
Wiley, 1957).
J. MEADE, 'Population Explosion, Standard of Living and Social Conflict',
Economic Journal (june 1967).
G. MYRDAL, Economic Theory and Underdeveloped Regions (London: Duck-
R. NELSON, 'A Theory of the Low Level Equilibrium Trap in Under-
developed Economies', American Economic Review (Dec 1956).
G. OHLIN, Population Control and Economic Development (Paris: O.E.C.D.,
1967).
J. G. WILLIAMSON, 'Regional Inequality and the Process of National
Development: A Description of Patterns', Economic Development and
Cultural Change (July 1965).
Chapter 6
V. V. BHATT, 'Theories ofBalanced and Unbalanced Growth: A Critical
Appraisal', Kyklos ( 1964).
R. EcKAus, 'The Factor Proportions Problem in Underdeve1opedAreas',
American Economic Review (Sep 1955).
0. EcKSTEIN, 'Investment Criteria for Economic Development and the
Theory of Intertemporal Welfare Economics', Q.uarterry Journal ofEconomics
(Feb 1957).
M. FLEMING, 'External Economies and the Doctrine ofBalanced Growth',
Economic Journal (june 1955).
W. GALENSON and H. LEIBENSTEIN, 'Investment Criteria, Productivity
and Economic Development', Q.uarterry Journal of Economics (Aug 1955).
A. HIRSCHMAN, Strategy Q/ Economic Development (New Haven: Yale
A. KAHN, 'Investment Criteria in Development Programmes', Quarterly
Journal of Economics (Feb 1951).
T. KING, 'Development Strategy and Investment Criteria: Complementary
or Competitive? ', Quarterly Journal of Economics (Feb 1966).
A. LEWIS, The Theory of Economic Growth (London: Allen & Unwin, 1955).
M. LIPTON, 'Balanced and Unbalanced Growth in Underdeveloped
Countries', Economic Journal (Sep 1962).
A. MATHUR, 'Balanced v. Unbalanced Growth: A Reconciliatory View',
S. K. NATH, 'The Theory of Balanced Growth', Oxford Economic Papers
(July 1962).
R. NuRKSE, Problems of Capital Formation in Underdeveloped Countries
P. RosENSTEIN-RODAN, 'Problems oflndustrialisation ofEast and South-
East Europe', Economic Journal (June-Sep 1943).
T. SciTOVSKY, 'Two Concepts of External Economies', Journal of Political
Economy (Apr 1954).
A. K. SEN,' Some Notes on the Choice of Capital Intensity in Development
Planning', Quarterly Journal of Economics (Nov 1957).
- - , Choice of Techniques, 3rd ed. (Oxford: Basil Blackwell, 1968).
P. STREETEN, 'Unbalanced Growth', Oxford Economic Papers (June 1959).
R. SuTCLIFFE, 'Balanced and Unbalanced Growth', Quarterly Journal of
Economics (Nov 1964).
Chapter 7
K. GRIFFIN andJ. ENos, Planning Development (Reading, Mass.: Addis on-
Wesley, 1970).
E. HAGEN, Planning Economic Development (Homewood, Ill.: Irwin, 1963).
W. LEWIS, Development Planning (London: Allen & Unwin, 1966).
R. MEIER, Developmental Planning (New York: McGraw-Hill, 1965).
UNITED NATIONS, Programming Techniques for Economic Development (New
York, 1960).
--,Formulating Industrial Development Programmes (New York, 1961).
A. WATERsoN, Development Planning: Lessons ofExperience (Oxford University
Press, 1966).
Appendix 7.1
T. BARNA (ed.), Structural Interdependence and Economic Development (New
York: StMartin's Press, 1963).
H. B. CHENERY and P. G. CLARK, Interindustry Economics (New York:
Wiley, 1959),
W. LEONTIEF et al., Studies in the Structure of the American Economy (Oxford
A. LEWIS, Development Planning (London: Allen & Unwin, 1966).
W. MIERNYK, The Elements of Input-Output Analysis (New York: Random
House, 1965).
G. MILLS, Introduction to Linear Algebra (London: Allen & Unwin, 1969).
A. PEACOCK and D. DossER, 'Input-Output Analysis in an Under-
developed Country: A Case Study', Review of Economic Studies (Oct 1957).
M. PES TON, Elementary Matricesfor Economics (London: Routledge & Kegan
Paul, 1969).
D. SEERS, 'The Use of a Modified Input-Output System for an Economic
Program in Zambia', in I. Adelman and E. Thorbecke (eds), The Theory
and Design of Economic Development (Baltimore: Johns Hopkins Press, 1966).
Appendix 7.2
I. ADELMAN and E. THORBECKE (eds), The Theory and Design of Economic
Development (Baltimore: Johns Hopkins Press, 1966).
W. ]. BAUMOL, Economic Theory and Operations Anafysis (Englewood Cliffs,
N.J.: Prentice-Hall, 1961).
H. CHENERY, 'Comparative Advantage and Development Policy',
A. LEWIS, Development Planning (London: Allen & Unwin, 1966).
P. N. RosENSTEIN-RODAN (ed.), Capital Formation and Economic Develop-
ment (London: Allen & Unwin, 1964).
]. TINBERGEN, The Design of Development (Baltimore:Johns Hopkins Press,
1958).
Appendix 7.3
H. BRuTON, 'The Two Gap Approach to Aid and Development: Com-
ment', American Economic Review (June 1969).
C. KENNEDY, 'Restraints and the Allocation of Resources', Oxford
Economic Papers (July 1968).
L. LEFEBER, 'Planning in a Surplus Labour Economy', American Economic
Review (] une 1968).
I. M. D. LITTLE, 'The Real Cost of Labour and the Choice between
Consumption and Investment', Quarterfy Journal of Economics (Feb 1961).
- - and J. MIRRLEES, Manual of Industrial Project Anafysis in Developing
Countries, vol. n: Social Cost-Benefit Anafysis (Paris: O.E.C.D., 1969).
G. MEIER, 'Development without Employment', Banca Nazionale del
Lavoro Quarterfy Review (Sep 1969).
A. K. SEN, Choice of Techniques, 1st ed. (Oxford: Basil Blackwell, 1960).
A. P. THIRLWALL, 'An Extension of Sen's Model of the Valuation of
Labour in Surplus Labour Economies', Pakistan Development Review
(autumn 1970).
--,'The Valuation of Labour in Surplus Labour Economies: A Synoptic
View', Scottish Journal of Political Economy (Nov 1971).
Chapter 8
W. BAER, 'Inflation and Economic Growth', Economic Development and
Cultural Change (Oct 1962).
W. BAER and I. KERSTENETSKY (eds), Inflation and Growth n Latin

America (Homewood, Ill. : Irwin, 1964).
R. B. BANGS, Financing Economic Development: Fiscal Policy for Emerging
Countries (University of Chicago Press, 1968).
E. BERNSTEIN and I. PATEL, 'Inflation in Relation to Economic Develop-
ment', I.M.F. Staff Papers (Nov 1952).
J. BHAGWATI, The Tying of Aid (Geneva: UNCTAD Secretariat, United
Nations, 1967), reprinted inJ. Bhagwati and R. Eckaus (eds), Foreign Aid
(Harmondsworth: Penguin Books, 1970).
R. CAMPos, 'Two Views on Inflation in Latin America', in A. Hirschman
(ed.), Latin American Issues (New Haven: Yale University Press, 1961).
A. CARLIN; 'Project v. Programme Aid', Economic Journal (Mar 1967).
R. J. CHELLIAH, Fiscal Policy in Underdeveloped Countries (London: Allen &
Unwin, 1960).
H. CHENERY and M. BRuNo, 'Development Alternatives in an Open
Economy: The Case oflsrael', Economic Journal (Mar 1962).
--and I. ADELMAN, 'Foreign Aid and Economic Development: The
Case of Greece', Review of Economics and Statistics (Feb 1966).
--and A. MACEWAN, 'Optimal Patterns of Growth and Aid: The Case
of Pakistan', Pakistan Development Review (summer 1966).
--and A. STROUT, 'Foreign Assistance and Economic Development',
American Economic Review (Sep 1966).
R. CooT, 'Wage Changes, Unemployment and Inflation in Chile',
Industrial and Labour Relations Review (July 1969).
M. DAVENPORT, 'The Allocation of Foreign Aid', Yorkshire Bulletin (May
1970).
E. Do MAR, 'The Effect of Foreign Investment on the Balance ofPayments ',
American Economic Review (Dec 1950).
E. EsHAG, Study if the Excess Cost if Tied Economic Aid Given to Iran 1966-67
(Geneva: UNCTAD, 1967).
M. HAQ, 'Tied Credits: A Quantitative Analysis', inJ. Adler (ed.), Capital
Movements and Economic Development, International Economic Association
Conference (London: Macmillan, 1967).
J. P. HAYEs, 'Long Run Growth and Debt Servicing Problems', in D.
Avramovic and Associates, Economic Growth and External Debt (Baltimore:
Johns Hopkins Press, 1964).
B. HIGGINs, Economic Development (New York: Norton, 1959).
B. HoPKIN and Associates, 'Aid and the Balance of Payments', Economic
Journal (Mar 1970).
N. KALDOR, Indian Tax Riform (Delhi: Ministry of Finance, 1956).
C. KENNEDY, 'Restraints and the Allocation of Resources', Oxford
Economic Papers (July 1968).
D. R. KHATKCHATE, 'Debt-Servicing as an Aid to Promotion of Trade of
Developing Countries', Oxford Economic Papers (July 1966).
S. B. LINDER, Trade and Trade Policy for Development (New York: Praeger,
1967).
I. LITTLE andJ. CLIFFORD, International Aid (London: Allen & Unwin,
1965).
R. McKINNON, 'Foreign Exchange Constraints in Economic Development
and Efficient Aid Allocation', Economic Journal (June 1964).
R. MARRIS, 'Can We Measure the Need for Development Assistance?',
Economic Journal (Sep 1970).
G. MAYNARD, 'Inflation and Growth: Some Lessons to be Drawn from
Latin American Experience', Oxford Economic Papers (June 1961).
--,Economic Development and the Price Level (London: Macmillan, 1962).
R. MIKESELL, The Economics qf Foreign Aid (London: Weidenfeld & Nicol-
son, 1968).
G. OHLIN, 'The Evolution of Aid Doctrine', in Foreign Aid Policies Recon-
sidered (Paris: O.E.C.D., 1956).
PEARSON Report, Partners in Development: Report of the Commission on Inter-
national Development (London: Pall Mall Press, 1969).
J. A. PINcus, 'The Cost of Foreign Aid', Review of Economics and Statistics
(Nov 1963).
A. PREST, Public Finance in Underdeveloped Countries (London: Weidenfeld &
Nicolson, 1962).
V. K. RAO, 'Investment, Income and the Multiplier in an Underdeveloped
Economy', Indian Economic Review (Feb 1952); also published in A. N.
Agarwala and S. P. Singh (eds), The Economics of Underdevelopment (Oxford
W. REDDAWAY, 'The Economics ofUnderdeveloped Countries', Economic
Journal (Mar 1963).
P. RosENSTEIN-RoDAN, 'International Aid for Underdeveloped Coun-
tries', Review qf Economics and Statistics (May 1961).
D. SEERS, 'A Theory of Inflation and Growth', Oxford Economic Papers
(June 1962).
H. SINGER, 'The Distribution of Gains between Investing and Borrowing
Countries', American Economic Review (May 1950).
A. P. THIRLWALL and C. BARTON, 'Inflation and Growth: The lnter-
nati()nal Evidence', Banca Nazionale del Lavoro Quarterry Review (Sep 1971).
UNITED NATIONS, World Economic Survey, 1966 (New York, 1967).
UNCTAD, Trade Prospects and Capital Needs qf Developing Countries (Geneva,
1968).
J. VANEK, Estimating Foreign Resource Needs for Economic Development (New
York: McGraw-Hill, 1967).
Chapter 9
J. BHAGWATI, 'Immiserising Growth: A Geometrical Note', Review of
Economic Studies (June 1958).
- - , 'The Theory of Comparative Advantage in the Context of Under-
development and Growth', Pakistan Development Review (autumn 1962).
C. CLARK, 'Too Much Food?', Lloyds Bank Review (Jan 1970).
M. J. FLANDERs, 'Prebisch on Protectionism: An Evaluation', Economic
Journal (June 1964).
R. GEMMILL, 'Prebisch on Commercial Policy for Less-Developed Coun-
tries', Review of Economics and Statistics (May 1962).
J. HicKs, Essays in World Economics (Oxford: Clarendon Press, 1959).
A. HIRSCHMAN, 'The Political Economyoflmport Substituting Industriali-
sation in Latin America', Quarterry Journal of Economics (Feb 1968).
H. JoHNSON, Economic Policies towards Less Developed Countries (London:
S. LINDER, Trade and Trade Policy for Development (New York: Praeger,
1967).
A. MACBEAN, Export Instability and Economic Development (London: Allen &
Unwin, 1966).
J. MEADE, 'International Commodity Agreements', Lloyds Bank Review
(July 1964).
G. MEIER, International Trade and Development (New York: Harper & Row,
1963).
H. MYINT, 'The "Classical" Theory of International Trade and Under-
developed Countries', Economic Journal (June 1958).
G. PATTERSON, 'Would Tariff Preference Help Economic Development?',
Lloyds Bank Review (Apr 1965).
R. PREBISCH, The Economic Development of Latin America and its Principal
Problems (New York: Economic Commission for Latin America, U.N.
Dept of Economic Affairs, 1950).
- - , 'Commercial Policy in the Underdeveloped Countries', American
Economic Review, Papers and Proceedings (May 1959).
T. SciTOVSKY, 'A Reconsideration of the Theory of Tariffs', Review of
Economic Studies (summer 1942).
J. VINER, International Trade and Economic Development (Oxford: Clarendon
Press, 1953).
T. WILSON, R. P. SINHA and J. R. CASTREE, 'The Income Terms of
Trade of Developed and Developing Countries', Economic Journal (Dec
1969).
OTHER INTRODUCTORY TEXTS AND READINGS
A. N. AGARWALA and S. P. SINGH (eds), The Economics of Underdevelop-

ment (Oxford University Press, 1958).
H. BRUTON, Principles of Development Economics (Englewood Cliffs, N.J.:
Prentice-Hall, 1965).
B. HIGGINS, Economic Development, rev. ed. (New York: Norton, 1963).
C. KINDLEBERGER, Economic Development (New York: McGraw-Hill,
1965).
G. MEIER (ed.), Leading Issues in Development Economics, 2nd ed. (Oxford
--and R. BALDWIN, Economic Development (New York: Wiley, 1957).
T. MoRGAN, G. BETZ and N. K. CHOUDHRY (eds), Readings in Economic
Development (Belmont, Calif.: Wadsworth Publishing Co., 1963).
H. MYINT, The Economics of the Developing Countries (London: Hutchinson,
1964).
Index
Abramovitz, M., 49-50 Cairncross, A. K., 36
Absorptive capacity aid criterion, 267 Campos, R., 242
Advances in knowledge, 69, 71 Canada, 265
Mrica, 17, 26, 27, 133, 274 Capital: contribution to growth, 58;
Agricultural output, 158 growth of, 130, 131; human, 126-7;
Agricultural produce, taxation of, 230 measurement of, 51-2; productivity
Agricultural products, 74 of, 102-3; role in development, 97-
Agriculture, 28, 72-96; in develop- 100; social, 98, 159-62; and technical
ment, 73-6; diminishing returns and progress, 97-116
zero marginal product in, 82; land Capital accumulation, 85, 97, 98, 99
in relation to, 73; productivity in, Capital formation, 224; equation, 141
73-6, 85, 158, 228; taxes, 230; Capital input, 65
traditional, 74, 78; versus industry, Capital-intensive technology versus
150 labour-intensive technology, 151
Aid, 5, 226; tying, 258. See also Assis- Capital-output ratio, 98, 100-4, 168,
tance 250, 266; aggregate, 106; average,
America, 60,271; per capita income of, 100; incremental (I.C.O.R.), 100
11. See also United States Ceylon, 285
Andie, S., 7 Chenery, H., 30, 202, 251
Argentina, 243 Chile, 218, 242, 243, 244
Arrow, K., 112 China, 133
Asia, 17, 26, 133, 274 Church. See Roman Catholic Church
Assistance: bilateral, 262, 265, 268, Clark, C., 28, 131
274; British, 272-5; criteria for allo- Clifford, J., 257
cation of, 265-9; foreign, 244-9,252, Cobb, C., 43, 48
254,261; international, 259-62, 268; Cobb-Douglas production function,
multilateral, 262, 266, 268; per 43-51, 63, 101, 130; empirical
capita, 268; technical, 269; tied, evidence, 57; limitations of, 46;
269-72 modifications, 51 : for embodied
Australia, 28 technical progress, 51, for improve-
ments in quality of labour, 55, for
Balance of payments, 151, 164, 176, resource shifts, 56; results of apply-
240,243,245,258,277-9,287,289- ing, 48-51
92,297,304 Coffee, 300
Balogh, T., 115 Colombia, 218
Banks and banking, 258-60, 266, 267, Comparative cost doctrine, 150, 284-7
269, 270, 276 Consumption per man-hour worked,
Barton, C., 240 24
Bauer, P., 29 Costs, 288
Becker, G., 114, 115 Critical minimum effort, 153
Bhagwati,J., 270 Critical minimum effort thesis, 145-7
'Big push' theory, 144, 145, 153, 155-7 Cumulative causation, 120-8
Birth rates, 134, 135, 136, 140
Blaug, M., 114 Davenport, M., 268
Bolivia, 243 Death rates, 134, 135, 136, 140
Bowman, M., 114 Debt servicing, 252-9, 266
Brazil, 218, 243 de Cani,J., 59
Britain. See United Kingdom Decision-making, 163, 166, 182
Brown, M., 59 Decision models, 187-8
Buffer stock schemes, 302 Decision-takers, Ill
318 INDEX
Deficit finance, 225, 233-5 Europe, 133, 146, 265; growth rate, 71
Deflation, 164 European Economic Community, 11
Denison, E., 54, 57, 58, 60, 114; Why Exchange rates, 16, 26, 27. See also
Growth Rates Differ, 61-71 Foreign exchange
Development, 1-3 7; agriculture in, Export-import gap, 224, 245, 248-52,
73-6; and growth, 22-3; index of, 256, 257, 268, 294
per capita income as, 19-24; neo- Exports and imports, 76-9, 126, 163,
classical models of, 86; programming 164, 242, 245, 250, 277, 287, 290-6,
approach to, 201-11; role of capital 298, 300, 301, 303, 304
in, 97-100; simulation of, 189; stages
of, 27-31 Farmers, tenant, 75
Development Decade, 5, 11, 274 Finance, 224-76; deficit, 225, 233-5
Development economics, current Fiscal policy, 228
interest in, 1-5 Fisher-Clark prediction, 29, 30
Development gap, 9-19, 117, 127,291; Flanders, M. J., 288
assessment of, 10 Food production, 76, 133
Development plans, 183 Food supplies, 128
Development strategy, 201; choice of, Forecasting, 189; models, 188. See also
149; self-help, 267 Input-output analysis
Diminishing returns, 2, 286 Foreign aid. See Aid
Directly productive activities, 159-62 Foreign assistance. See Assistance
Disguised unemployment, 87-94 Foreign exchange and foreign exchange
Division of labour, 111-12 gap,220-3,245-6,248-52,267,279,
Domar, E., 4, 102 280, 300
Domestic value-added, 298 Foreign exchange rates, 16
Douglas, P., 43, 48 France, 265
Dual, 207-11 Free trade, 285-7, 294
Dual-gap analysis, 244-6, 252, 256, Furtado, C., 20
293
Dualism, 77, 118-20; geographic, 121 ; Gaathon, A., 58
social, 118-19; technological, 119 Galenson, W., 172, 174
Duality of price determination and Galenson-Leibenstein investment cri-
allocation, 203 terion, 172
General Agreement on Tariffs and
Earnings, fluctuations in, 301 Trade (GATT), 297
Eastern European Industrial Trust, Germany, 35, 265
157 Ghana, 17
Economic development, path of, 154; Great Britain. See United Kingdom
with unlimited supplies of labour, 'Green revolution', 75
80-7 Gross domestic product, II
Economic divisions, 5 Growth, analysis of, 38; production
Economies of scale, 68, 156. See also function approach to, 98, 129;
Increasing returns balanced, 154-8, 165, 166; balanced
Economy, models of, 184 versus unbalanced, 153; capital con-
Ecuador, 243 tribution to, 58; causes of, 38-71 ;
Education, 113; and growth, relation contribution of education, 64; and
between, 113; contribution to development, 22-3; and education,
growth, 64 relation between, 113; expression of,
Effective capital stock, 52, 53 38; Harrod model of, 4, 102, 245;
Effective protection, 297 influence of economies of scale, 68;
Elasticity of substitution, 46 labour contribution to, 58; measure-
Embodied technical progress, 51 ment of, production function ap-
Emerging countries, 1 proach to, 39; of money economy,
Emigration, 127 76-80; potential obstacles to, 117;
Empirical evidence and production process of, 39; rates of, 277; Rostow's
function, 57 stages of, 31-7; unbalanced, 153,
Engel's Law, 289 158-66
Enke, S., 83 Growth equation, 102
Entrepreneur, Ill Growth performance, comparative, 12,
Eshag, E., 270 14-15, 62
INDEX 319
Growth rates, Denison on, 61-71; International Development Associa-
Europe, 71 tion, 259, 260, 270, 274
Guinea, 134 International Finance Corp.oration,
259,260
Hagen, E., 145 International Monetary Fund, 306
Haq, M., 270 International price ratio, 282
Harrod, R., 4, 102, 105-9, 245 International transmission, 124
Harrod-Domar model, 4 Inter-regional differences, 124
Health, 113 Invention, 110-11
Heckscher-Ohlin theory, 152 Investment, 32, 33, 37, 222, 224, 225,
Hicks, J., 58, 105-9, 283 226, 240, 255; domestic resources
Higgins, B., 232 for, 226-44; foreign resources for,
Hirschman, A., 124, 159, 161, 162, 163, 244--76; pattern of, 154; private
165 foreign, 263-5; productivity of, 38;
Hopkin, B., 273 through taxation, 229; trends in,
275
Immigration, 123, 127 Investment criteria, 148, 151, 166-79
Import-export gap, 224, 245, 248-52, Investment decisions, 149
256,257,268,294 Investment-output ratio, 38
Imports and exports, 76-9, 126, 163, Investment planning, 103-4
164, 242, 245, 250, 277, 287, 290-6, Investment ratios, 18
298,300-4 Investment requirements, 2, 4
Incentives, 94 Iran, 270
Income: compensation schemes, 305; Israel, 58-9
elasticity of demand for products, Iterative procedure for input-output
289-92; equality, 5; expression of, analysis, 198-9
38; growth, 146-7: equation, 142;
national, conversion of, 25; per Jamaica, 284
capita, see Per capita income; Japan, 33, 75, 80, 134, 265
world distribution, 6-9 Johnson, H. G., 98
Increasing returns, 44, 286. See also Jorgenson, D., 86
Economies of scale
Incremental capital-output ratio Kaldor, N., 232
(I.C.O.R.), 99 Kennedy, C., 57
Index of development, per capita Keynesian unemployment, 233
income as, 19-24 Khatkchate, D. R., 257
India, 133, 227, 232, 257, 274, 285 Knowledge, advances in, 69, 71
Industrial structure, 34 Kuznets, S., 6, 21, 35-6
Industrialisation, 30-1, 77, 80, 242,
279-80 Labour, 72-96, 129; contribution to
Industry: overmanning of, 70; versus growth, 58; costless, 212, 214; divi-
agriculture, 150 sion of, 111-12; dynamic surplus, 89;
Inflation, 164, 225, 235-40, 276; economic development with un-
efficiency of, 236-9; in less developed limited supplies of, 80-7; improved
countries, 240-4 quality of, 55; international division,
Innovation, 110 282; marginal product of, 86, 87,
Input-output analysis, 2, 3, 189-201; 137; migration of, 126-7; paradox
assumptions of, 199; criticism of, 200; of, 129; real cost of, 214--20; static
iterative procedure for, 198-9; mat- surplus, 88; successive units added
rix methods, 194-8; uses, 189 to land, 81; unlimited supplies of,
Input-output coefficients, 190, 193 79, 80, 84, 85, 87; unskilled, 127;
Input-output models, 199 valuation of, 95-6, 212-13: in open
Input-output table, 189, 190-1 economy, 220-3
Inter-American Development Bank, Labour absorption, 85
5 Labour absorption investment cri-
Inter-country comparisons, 25 terion, 175
Interest rates, 254, 255, 266 Labour-force growth, 59, 62, 130,
International assistance. See Assistance 131
International commodity agreements, Labour input: index of, 63; measure-
300-2 ment of, 63
320 INDEX
Labo1,1.1'-intensive technology versus Market mechanism, 163, 164, 180-2;
capital-intensive technology, 151 replacement of, 4
Labour productivity, 94; growth of, 39 Market prices, 180
Labourtransference,94-6,212-15 Massell, B., 57
Laissez-:faire, doctrine of, 166 Mathematical programming, 3
Land, 72-96; in relation to agriculture, Mathur, A., 164
73 Matrix methods of input-output prob-
Land reform, 75-6, 228 lem, 194-8
Latin America, 5, 17, 133, 218, 240-2, Maynard, G., 236
244 Meade,J., 303
Learning, 111-13 Meier, G., 218
Leibenstein, H., 144, 172, 174 Mexico, 243
Less developed countries: British assis- Middle East, 17
tance to, 272-5; definition, 21 ; Migration, 123, 125-7
dualism, 118; emigration effect, 127; Mill, J. S., 2
finance, 80; fiscal and monetary Millikan, M. F., 26
policy, 225; foreign borrowing, 252, Minimum capital-output ratio in-
25 7; future comparative position of, vestment criterion, 168
11 ; income distribution, 227; income Mirrlees, J., 216, 220
growth, 146; industrialisation, 30, Model-building, 4
78, 80; inflation in, 240-4; inter- Models: decision, 187-8; of economy,
national trading and payments posi- 184; forecasting, 188; input-output,
tion, 128; model-building, 4; per 199; policy, 185; projection, 188
capita income, 9; politics, 5; popula- Monetary policy, 228, 242
tion growth rate, 5, 133, 134, 135; Money economy: emergence of, 80;
progress, 3; regional imbalances, growth of, 76-80
125; relative contribution of labour Money illusion, 219
and capital to growth, 60; resource Myrdal, G., 117, 121-6
allocation, 148; trade, 126, 277-8.
See also under specific subjects and National incomes, conversion of, 25
countries Nelson, R., 52, 140
Lewis, A., 80, 83, 84, 85, 98, 116, 158, Non-residential structures and equip-
226-7 ment, 66
Liberia, 17 North America, 133
Linder, S., 293-6 Nurkse, R., 155, 158, 165
Linear programming, 2, 203-11
Litde, I. M.D., 216, 220, 257 Ohlin, G., 265
Living standards, 5, 9, 10, 13, 18-19, Olive oil, 300
25, 26, 140, 146; comparison of, 16; Output: expression of, 38; per unit of
tolerable, 19 input, 67-70
Loans, 266; interest-bearing, 252; soft, Overmanning of industry, 70
260 Overpopulation, 127
Lorenz curves, 6-7, 10
Low-income countries, 21-2 Pakistan, 18, 19, 270
Low-level equilibrium trap, 133, 139- Patel, S. J ., 9
45, 153 Peacock, A., 7
Pearson Commission, 260
Macbean, A., 301 Per capita incomes, 6, 11, 12, 14-15,
Maddison, A., xii, 60 140, 143--6; of America, 11; ofE.E.C.
Maizels, A., 30 countries, 11; growth and growth
Malawi, 18 rate, 10, 12, 13, 14-15, 290; as index
Malthus, T. R., 2 of development, 19-24; of less de-
Malthusian situation, 146 veloped countries, 9; measurement
Marginal per capita reinvestment and comparability of, 24-7; spread
quotient investment criterion, 172 between countries, 8
Marginal rate of substitution between Peru, 18
labour and capital (M.R.S.), 46 Pincus, J. A., 259
Marginal rule for resource allocation, Planning, 3, 4, 165, 180-223; argu-
166 ments for and against, 180; flexibility
Marglin, s., 89 in, 187; form of, 181; horizons, 187;
INDEX 321
instrumental variables in, 186; wage, Residual factor, 67
212-23. See also Input-output analy- Residual productivity, 70, 71
sis; Linear Programming Resource allocation, 148-80, 201, 207,
Policy models, 185 303; balance of payments considera-
Political recognition, 4 tions, 176; broad policy choices,
Politics, 35, 149 148-9; choice of, 159; criteria, 168-
Poor countries, 1, 5, 6, 8, 9, 10, 13, 125. 79: labour absorption, 175, marginal
See also Poverty per capita reinvestment quotient,
Population, 11; optimum, 136-9; 172, minimum capital-output ratio,
world, 133-6 168, social product, 170-2, 177;
Population growth, 2, 16, 128-33, 135, efficiency in, 166; marginal rule for,
139, 140, 146; equation, 141; rate, 166; prices for, 182; social optimum,
134 167; social welfare function, 178;
Population problem, 128-33 and use, 68
Poverty, 5, 19,21-2, 117; vicious circle Resource shifts, 28, 46, 56
of, 99. See also Poor countries Restriction schemes, 302
Prebisch, R., 287-91, 295, 303 Ricardo, D., 2
Prebisch doctrine, 287 Rich countries, 5, 6, 8, 9, 10, 13
Prest, A., 235 Road building, 234
Price compensation agreements, 303-5 Rolling plan, 187
Price fluctuations, 304 Roman Catholic Church, 5
Price index, 24 Rosenstein-Rodan, P., 155, 157, 267
Price stability, 241, 243, 300-2 Rostow, W. W., 31-7, 98
Price weights, 68 Russia, 4, 33, 35, 80, 133
Prices, 24, 27, 239-40, 288; primary
product, 301; for resource allocation, Saving(s), 224, 225, 254; generated
182. See also Inflation through policies, 225; involuntary,
Primal, 207-11 225, 228; trends in, 275; voluntary,
Production function, 40-3, 107, 108, 225,226
109; aggregate, 40; Cobb-Douglas, Savings-investment gap, 221-4, 245,
43-51, 63, 101, 130: empirical 247-8,250-2,256,257,297
evidence, 57, limitations of, 46, Savings-investment ratio, 236
modifications, 51, for embodied tech- Schultz, T., 74-5, 113, 114, 115
nical progress, 51, for improvements Schumpeter, J ., 110
in quality of labour, 55, for resource Scitovsky, T., 291
shifts, 56, results of applying, 48-51 ; Self-help development strategies, 267
properties of, 40; vintage, 52 Sen, A. K., 214
Production function approach to analy- Service activities, 29
sis of growth, 39, 98, 129 Shonfield, A., 78
Production function diagram, 41, 81 Simulation of development, 189
Productivity, 31, 99-100; in agricul- Smith, Adam, 111, 155, 182
ture, 73-6, 85, 158, 228; as aid cri- Social marginal productivity (S.M.P.),
terion, 266, 267; of capital, 102-3; 170-2, 177
growth of total, 131; of investment, Social product investment criterion,
38; labour, 94: growth of 39; measure 170-2, 177
of, 266; residual, 70, 71 Social welfare function, 178
Productivity growth, 131 Solow, R., 49, 52, 58
Products, income elasticity of demand Sovereignty, 124
for, 289-92 Streeten, P., 115, 154
Programming: approach to develop- Subsistence level, 137, 138
ment, 201-11; linear, 203-11
Projection planning models, 188 Taiwan, 75
Protection. See Tariffs; Trade Tariffs, 291, 297-300
Taxation and taxation policy, 217,218,
Railways, 34 219, 228-33
Rao, V. K., 233 Technical assistance, 269
Reddaway, W., 104, 233-4 Technical progress, 30, 105, 130, 146;
Regional Development Banks, 258, 259 and capital, 97-116; capital- and
Representative demand, theory of, labour-saving, 105; and education,
294 113; and invention, 110; and learn-
322 INDEX
ing, Ill ; neutral, 46, 107; of a 291, 300: and technical progress,
society, 109; sources of, 110; and 288, trends in, 292-3; theory and
terms of trade, 288 dual-gap analysis, 293; world, pat-
Technological economies of scale, tern of, 126
42 Traditional societies, 32
Technology, 3; capital-intensive,
versus labour-intensive technology, Underdeveloped countries, 1
151 ; changes in, 105; level of, 41 Underdevelopment, 1-37; definition
Tenant farmers, 75 of, 20; and dualism, 119; persistence
Thailand, 27 of, 117-47
Third World, 5; development of, 1 Underemployment, 127
Thirlwall, A. P., 57, 240 Undeveloped countries, use of term, 1
Tied assistance, 269-72 Unemployment: disguised, 87-94, 215,
Tin, 300 218; Keynesian, 233
Trade, 3, 277-306; and balance of United Kingdom, 27, 28, 70,265, 271;
payments, 279; barriers to, 300; financial assistance by, 272-5; slow
barter terms of, 288, 292-3; between growth, 60
less developed countries, 258; buffer United Nations, 259, 276; Conference
stock schemes, 302; comparative on Trade and Development
cost doctrine, 150, 284-7; compari- (UNCTAD), 297; Report on Measures
son of less developed and developed for International Economic Stability, 300
countries, 277-8; dynamic gains, United States, 265
280, 283-4; free, 285-7, 294; gains
from, 280-7; income compensation Venezuela, 243
schemes, 305; income terms of, 292, Viner, J., 275
293; international commodity agree-
ments, 300-2; new theories for Wages, 83, 85, 126; planning, 212-23
developing countries, 287; policies, Welfare, 137-8, 178
295; preferences and effective pro- Wheat, 300
tection, 297; price compensation Williamson, J. G., 125
agreements, 303-5; protection, 291- World Bank, 78,258,259,260,266,267,
2, 294-5, 297-300; restriction 269,270,274
schemes, 302; static gains, 280-3;
tariffs, 291, 297-300; terms of, 290, Yamey, B., 29

(A. P. Thirlwall (Auth.) ) Growth and Development (B-Ok PDF

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

(A. P. Thirlwall (Auth.) ) Growth and Development (B-Ok PDF

Enviado por

Direitos autorais:

Formatos disponíveis

GROWTH AND DEVELOPMENT

First published 1972 by

The paperback edition of this book is sold subject to the condition

1 DEVELOPMENT AND UNDERDEVELOPMENT

Current Interest in Development Economics

2 THE PRODUCTION FuNCTION APPROACH TO THE

3 LAND, LABOUR AND AGRICULTURE 72

4 CAPITAL AND TECHNICAL PROGRESS 97

Technical Progress 105

5 THE PERSISTENCE OF UNDERDEVELOPMENT 117

6 THE ALLOCATION OF RESOURCES 148

7 PLANNING ECONOMIC DEVELOPMENT 180

B. FOREIGN RESOURCES FOR INVESTMENT 244

9 TRADE AND DEVELOPMENT 277

Restriction Schemes 302

REFERENCES AND FuRTHER READING 307

Allen & Unwin, 1970).

Current Interest in Development Economics

Academic interest in the mechanics of growth and development

the main means by which the development process may be speeded

1 R. Harrod, 'An Essay in Dynamic Theory', Economic Journal (Mar 1939),

The World Distribution of Income

% ofworld % of world income

When two Lorenz curves cross, precluding a definite conclusion

1 Reproduced in a letter to The Economist, 11 Oct 1969.

The Development Gap2

Since living standards in all countries tend to rise absolutely over

1. Supposing that the E.E.C. countries and America, i.e. the

Source: International Development Agency, Annual Report, 1968.

(I) (2) (3) (4) (5)

East and South-east Asia

necessary to prevent the present per capita income gap from

Per Capita Income as an Index of Development

California Press, 1964) pp. 141-3.

sterility and to a dangerously mechanistic approach to policy impli-

'underdevelopment', see F. Machlup, 'Disputes, Paradoxes and Dilemmas

would be consumption per man-hour worked. This index, in

The Measurement and Comparability of Per Capita Incomes

e.g. ifthe G.N.P. of country X is 100 million ducats, its population

But if living standards are to be compared by this method, it

by an index of the cost of living; that is, if $1 = 20 ducats, then

The Stages of Development

1 Statement by M. F. Millikan before the Subcommittee on Foreign

thesis of Fisherl and Clark,2 who employ the distinction between

the net value of manufacturing production per head of the total

1 This is the approach taken by A. Maizels in his book, Industrial Growth

Rostow' s Stages of Growth

growth of output; secondly, that the leading sector generates

then treats this identification as if it were a meaningful identi-

Economics of Take-off into Sustained Growth, Proceedings of the Conference of

The Production Function

By definition, slow growth is the product either of a low investment

In this formulation slow growth is attributable by definition either

The Production Function

In the opposite case of decreasing returns, the functions would be

irrespective of increases in factor supplies. These are called tech-

The Cobb-Douglas Production Function

of the function for estimating purposes, the function is said to be

dlogOt dlog Tt + dlog Kt + ~ dlog Lt (2.5)

For simplicity, equation (2.5) may be written

ro = rw + arK + ~rL (2.6)

rL is the rate of growth of labour

to capital and labour, respectively, as before.