Preview

INTERNATIONAL LABOUR OFFICE

THE QUALITY OF LABOUR
AND ECONOMIC DEVELOPMENT
IN CERTAIN COUNTRIES
A Preliminary Study
by
Walter GALENSON and Graham PYATT

GENEVA
1964

STUDIES AND REPORTS
New Series, No. 68

First printed : October 1964
Second impression : May 1966

PRINTEE! BY H. STUDER S.A., GENEVA

FOREWORD
For the preparation of this study the International Labour Office
was fortunate in securing the services of Professor W. Galenson of the
University of California, Berkeley, and Dr. G. Pyatt of the Department
of Applied Economics, University of Cambridge, and Fellow of Gonville
and Caius College, Cambridge. Professor Galenson spent a sabbatical
year in the Economic Division of the International Labour Office and
Dr. Pyatt worked on the study for several months, partly in Geneva
and partly in Cambridge.1 The views and opinions expressed are those
of the authors and do not necessarily represent those of the I.L.O.
An introductory summary was prepared by Mr. K. Taira of the
I.L.O.'s Economic Division as a non-technical résumé of the method
and findings of the study. In order to make this outline sufficiently
clear many important cautions and qualifications had to be omitted.
This is a pioneering study in an area that is of great importance for
the work of the I.L.O. and other international organisations. A great
deal of work has been done over the last 20 years by national statistical
services and international organisations in providing more and better
economic and social statistics. An enormous amount remains to be
done both in improving the range and quality of the statistics and in
refining concepts and methods of analysis. Notwithstanding the many
major gaps and deficiencies in the statistics, the authors felt that the
time had come for an attempt to see what relationships might be discovered between rates of economic growth on the one hand and, on the
other, certain indicators of the quality of labour, which is no doubt
affected by such factors as calorie intake and expenditure on education,
health and social security. Their aim was to provide an answer to the
question: can any systematic tendencies be detected which might help
to guide decisions on the use to be made of scarce resources by throwing
light on the question of how rates of economic growth are affected by
(or at least what rates of growth have been found by experience to be
1
The authors wish to acknowledge the assistance received from Professor Carl
Stevens of Reed College and Dr. Malcolm Fisher of Cambridge University, who
were in Geneva at the time of the study. The I.L.O. wishes to thank the Director
of the Mathematical Laboratory of the University of Cambridge for permission to
use the EDSAC II computer in connection with the study.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

compatible with) différent levels of expenditure on various social
objectives?
In the present state of knowledge only a beginning can be made in
answering this question, tut :.n methodology the present study breaks
new ground, and its substantive findings are at least suggestive. In
Chapter VI the authors present some observations on desirable next
steps. Working in co-operation with others in this field, the I.L.O.
hopes to develop its work along lines which the authors of this study
have helped to chart. Botti for what they have themselves accomplished
and for their suggestions as to how the work they have begun might
be developed and extended, the I.L.O. is glad to acknowledge its
indebtedness to the authors.
The reasons determiniig the selection of countries and statistical
series included in this study are discussed by the authors on page 52
and following pages. One; essential criterion was that the data had to
be comparable internationally. The authors explain that it is difficult
to compare market and centrally-planned economies because of differences in the definition of some major economic variables—national
product, for example—as well as differences in the way in which output
is valued. Since they wers not in a position to conduct two separate
studies, which would have meant developing two models varying considerably in concept, the present study is confined to the market economies. A study of the relationship between economic growth and
certain types of social expenditure in centrally-planned economies
would be of great interest and value to developing countries. It is
hoped that the I.L.O. will later be able to undertake or participate in
such a study.

and findings of the study. In order to make this outline sufficiently
clear many important cautions and qualifications had to be omitted.
This is a pioneering study in an area that is of great importance for
the work of the I.L.O. and other international organisations. A great
deal of work has been done over the last 20 years by national statistical
services and international organisations in providing more and better
economic and social statistics. An enormous amount remains to be
done both in improving the range and quality of the statistics and in
refining concepts and methods of analysis. Notwithstanding the many
major gaps and deficiencies in the statistics, the authors felt that the
time had come for an attempt to see what relationships might be discovered between rates of economic growth on the one hand and, on the
other, certain indicators of the quality of labour, which is no doubt
affected by such factors as calorie intake and expenditure on education,
health and social security. Their aim was to provide an answer to the
question: can any systematic tendencies be detected which might help
to guide decisions on the use to be made of scarce resources by throwing
light on the question of how rates of economic growth are affected by
(or at least what rates of growth have been found by experience to be
1
The authors wish to acknowledge the assistance received from Professor Carl
Stevens of Reed College and Dr. Malcolm Fisher of Cambridge University, who
were in Geneva at the time of the study. The I.L.O. wishes to thank the Director
of the Mathematical Laboratory of the University of Cambridge for permission to
use the EDSAC II computer in connection with the study.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

compatible with) different levels of expenditure on various social
objectives?
In the present state of knowledge only a beginning can be made in
answering this question, but in methodology the present study breaks
new ground, and its substantive findings are at least suggestive. In
Chapter VI the authors present some observations on desirable next
steps. Working in co-operation with others in this field, the I.L.O.
hopes to develop its work along lines which the authors of this study
have helped to chart. Both for what they have themselves accomplished
and for their suggestions as to how the work they have begun might
be developed and extended, the I.L.O. is glad to acknowledge its
indebtedness to the authors.
The reasons determining the selection of countries and statistical
series included in this study are discussed by the authors on page 52
and following pages. One essential criterion was that the data had to
be comparable internationally. The authors explain that it is difficult
to compare market and centrally-planned economies because of differences in the definition of some major economic variables—national
product, for example—as well as differences in the way in which output
is valued. Since they were not in a position to conduct two separate
studies, which would have meant developing two models varying considerably in concept, the present study is confined to the market economies. A study of the relationship between economic growth and
certain types of social expenditure in centrally-planned economies
would be of great interest and value to developing countries. It is
hoped that the I.L.O. will later be able to undertake or participate in
such a study.

CONTENTS
Page
INTRODUCTION

SUMMARY OF AIMS, METHODS AND RESULTS OF THE STUDY (by K. TAIRA) . .

CHAPTER I: Previous Relevant Work

CHAPTER II: The Method of the Present Study

CHAPTER III:

The Model

Summary of the Arguments

CHAPTER IV: Indicators of Economic Growth and Labour Quality
General Considerations
The Indicators
Economic Growth
Investment
The Labour Force
The Wage Share
Education
Health
Housing
Social Security
Subdivision of Countries by Income Level

52
52
54
54
56
57
60
60
63
65
66
67

CHAPTER V: Statistical Methods and Results
The Statistical Analysis
Comment on the Results

69
69
76

CHAPTER VI: Possible Extensions of the Study
The Data Problem
Extensions of the Model

79
79
82

CHAPTER VII:

Conclusions

The Problem of Data
The Model
The Empirical Results

85
86
86

APPENDICES
APPENDIX I.

Mathematical Appendix

APPENDIX IL Table 1. Ratesof Growth of Gross Domestic Product, 1950-60
Table 2. Rates of Growth of the Economically Active Population, 1950-60
Table 3. Investment Ratios and Wage Shares, 1950-60 . . .
Table 4. Indicators of the Development of Education, 1950-60
Table 5. Indicators of Health, 1950-60

92
94
96
98
100
105

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT
Page

Table 6. Indicators of Housing and Social Security, 1950-60
Table 7. Values of Z Variables Used in Regression Analysis
Table 8. Least Squares Estimates of Various Regression Equations—First Version of the Model
Table 9. Least Squares Estimates of Various Regression Equations—Second Version of the Model

110
112
114
116

APPENDIX III. Figures

1. Annual Rate of Growth of Gross Domestic Product versus Annual Rate of
Growth of Labour Force, 1950-60
2. Annual Rate of Growth of Gross Domestic Product versus the Average
Investment Ratio, 1950-60
3. Contribution of Capital to Annual Rate of Growth of Output versus the
Investment Ratio, 1950-60
4. The Dimensions of Economic Growth, 1950-60
5. Key to Figure 4
6. The Explanation Achieved by the Regression 16
7. Rate of Growth of Calories per Head versus Rate of Growth of Real Wages
8. Rate of Growth of Higher Education Enrolment Ratio versus Rate of
Growth of Productivity

LIST OF TABLES IN THE TEXT
I. Extent to Which International Differences in Adjusted Rate of Growth
of Labour Productivity [Z0] Are Explained by International Differences
in the Modified Investment Ratio and Labour Quality Improvement
Factors

II. Rates of Growth of Gross Domestic Product, Unadjusted and Adjusted
Labour Productivity, and the Investment Ratio for 52 Countries, 1950-60

III. Results of the Regression of the Rate of Growth of Output on the Rate
of Growth of Labour and the Investment Ratio

INTRODUCTION

Despite the considerable volume of literature on economic development that has emerged during the past decade, relatively little systematic work has been undertaken on the quality of the labour force as a
factor in the promotion of growth. This has been due, in part, to preoccupation with the role of capital investment in the development
process. There has been a fairly widespread belief that, given a sufficient
volume of investment, a respectable tempo of economic growth was
virtually assured. The seeming abundance of labour characteristic of
many underdeveloped countries also served to detract attention from
the importance of this factor, the assumption being made that labour
would be forthcoming in the requisite supply where it was needed.
The puzzling failure of countries well endowed with natural resources
to achieve a satisfactory rate of growth during the past decade has given
rise to new interest in those aspects of human organisation and ability
that may be essential to progress. The Director-General of the I.L.O.,
in his Report to the 1963 International Labour Conference, put the matter
this way—
Much greater emphasis is now being given to the concept of human resources
development. [The economically underdeveloped countries] are rich in
people, but for the most part in people whose skill potential is inadequately
developed, who have insufficient opportunities for productive employment,
who lack forms of organisation which would enable them to produce more,
and whose poor health and living conditions severely limit their productivity.
They are poor in the physical equipment of modern production. Yet it is
now beginning to be realised that skills and the effective utilisation of the
labour force may, in addition to physical capital and natural resources, be
a more decisive factor for economic expansion than was hitherto assumed.1
There are many aspects of labour force quality which may have a
considerable impact on productivity. Among those which have been
most often cited are the age and sex composition of the labour force;
nutrition and health; and education and training, embodied in specific
skills. Less commonly thought of in this connection, but nonetheless
1
I.L.O.: Report of the Director-General: Programme and Structure of the I.L.O.,
Report I, International Labour Conference, 47th Session, Geneva, 1963 (Geneva,
1963), p. 33.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

possibly of considerable significance, are the sense of well-being and
personal satisfaction that accrue from adequate housing and economic
security; successful acclimatisation to the tempo of urban and industrial
life; the degree to which work provides scope for personal initiative
and freedom from compulsion; and the organisational means of promoting individual interests and redressing grievances. This is far from a
complete catalogue, but it does serve to delineate the complexity of the
problem.
We have attempted to determine whether the presence, or the absence,
of some of these factors has influenced the economic growth of nations
during the decade of the 1950s. In selecting particular factors for analysis we were guided largely by two considerations: the extent to which
they could be quantified, and the availability of internationally comparable statistics. This is clearly not an ideal basis for selection, but we
were motivated by the strong conviction that quantitative analysis offers
the most promising path to a fuller understanding of these qualitative
phenomena at the present time, and were limited in our choice of factors
by the time element and the resources at our disposal—hence the subtitle " A Preliminary Study ".
Whatever the merits of our work, the importance of the subject is
beyond question. Nations with extremely limited resources at their
disposal—and this includes the nations embracing the bulk of the
world's population—are currently obliged to make crucial decisions of
allocation on the basis of guesswork, with consequent wastage that can
ill be afforded. Criteria for resource allocation are urgently needed
and priorities must be established in the light of the requisites for sustained growth. Such criteria must be precise enough to provide administrative guidelines, and this means essentially that they must be quantitative. It is no great help to a planner or budget-maker to learn that
his country needs more education, better health measures, more and
better housing and food, and many other things. What he must know
is the specific proportion of his total resources available for all of these
factors which should be spent on each, if he is to achieve his goal of
satisfactory growth, however he may have defined it.
There is plenty of room for scepticism on the possibilities of
attaining this objective. The Economic Commission for Asia and the
Far East has argued—
It is not possible to determine appropriate priorities and balance among
broad economic and social sectors through analysis of complementarities
and the use of projective techniques. The general interactions of health,
education, social welfare services, housing, etc., with industrialisation, agricultural production, and other broad economic factors, cannot be specified

INTRODUCTION

in quantitative terms; relations that undoubtedly exist are too complex,
variable and indirect to permit simple equations.1
Perhaps so; but one can never know whether this is true until one has
tried the equations. We have taken our cue instead from a pioneering
study, the 1961 United Nations Report on the World Social Situation,
which adopts a much more hopeful position.
There are at present no quantitative criteria derivable from theoretical,
logical or mathematical analysis by which the amount of attention to be
devoted to a particular field of social development can be indicated. Ideally,
one should be able to take a given field, such as education, health, housing,
labour or family welfare, and analyse the benefits for the total developmental
effort of a given allocation of expenditure in this field at a given time...
Balanced development could then mean the combination of economic and
social factors yielding the greatest sustained increase in total development...
In spite of these theoretical difficulties, decisions on balanced development
have to be made and are made as a practical necessity all the time. Each
allocation of resources in the normal budget or in a developmental budget is
justified on the assumption that it contributes to the economic and social
pattern that is optimal for the country—although, in practice, for the very
reason of lack of a systematic framework, interests other than the welfare
of the nation come into play...
While it is theoretically not possible to state what levels of development
in the various social components should go with given levels of economic
development, it is quite possible to state what social levels do go with given
economic levels—that is, to examine the patterns of development from a
purely empirical point of view. It is conceivable that, in the light of some
ideal model, the majority of the countries of the world would turn out to be
unbalanced in the emphasis they give to the different social and economic
fields. Certainly there are regional differences and differences along political
lines. What is appropriate for one country will not necessarily be appropriate
for another. But after these cautions have been expressed and emphasised,
the judgment can still be maintained that knowledge of the experiences and
practices of other countries in regard to the inter-relationship of economic
and social development can be a useful type of information, particularly
for those who must make decisions in countries that lack experience in
development.2
It was clear that there would be many obstacles in the path to a fuller
understanding of the inter-relationships among the factors with which
we are concerned. -Deficiencies in the quality of the available statistics,
the hazards of international comparison, the necessity of resorting to
rough-and-ready estimation methods for bridging gaps in the data,
the discouragement of having to abandon promising lines of analysis
1
United Nations, Economic Commission for Asia and the Far East: Notes on
Policies and Methods of Co-ordinating and Integrating Economic and Social Development Programmes (E/CN.ll/DPWP, 5/L.8, 17 August 1959), p. 31.
a
United Nations: Report on the World Social Situation, with Special Reference
to the Problem of Balanced Social and Economic Development (New York, 1961),
pp. 38-39.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

because even these methods proved unavailing when total ignorance of
key economic magnitudes existed—these are some of the practical
difficulties that any investigator in this field who does not have the means
of large-scale data fabrication must face. Then there is the crucial
problem of causation; the establishment of a statistical relationship
among variables is, at best, the beginning of wisdom. We do not pretend
to have arrived at a satisfactory solution of these problems, but we do
have a better understanding than we did at the start of the lacunae in
the data and the limitations of the conceptual framework that has been
developed for tackling this range of questions.
It should be emphasised that a study of the character undertaken
here can provide only a partial answer to the relevant policy questions.
If, for example, it should appear that investment in a particular social
programme, through its action on the quality and productivity of labour,
tends to further development, it does not necessarily follow that this
programme is to be preferred to alternative uses of investment funds.
Other considerations of a political or social nature, or economic considerations involving a time dimension other than the one used in the inquiry,
may dictate a different decision. But at least decisions may be taken
in the light of alternative benefits and costs, and the economic sacrifices
entailed assessed more precisely.
The International Labour Office for some years now has been engaged
in technical assistance and other programmes designed to improve the
quality of the labour force of the developing nations. Vocational and
technical training, improved health and safety conditions on the job,
adequate nutrition, and more and better designed social security programmes are some of the means employed in the pursuit of this objective.
However, the Director-General has recently stated that—
.. .little has been done towards a systematic evaluation of the contribution of
operational programmes towards the broad goals of economic and social
development... Research is needed to clarify the relationships between
different factors in economic and social development and thus to enable
goals to be realistically defined and projects prepared to contribute toward
the attainment of these goals.1
It is the hope of the authors that the present study will make some contribution, modest though it may be, to the research objective defined by
the Director-General. They are the first to acknowledge that much
more will have to be done in the gathering and collating of statistics,
and in the refinement of concept and methodology, before conclusive
results can be anticipated.
1

Report of the Director-General, op. cit., p. 204.

SUMMARY OF AIMS, METHODS AND RESULTS
OF THE STUDY 1
Production is the process of organising and applying the inputs of
the factors of production in order to obtain output. The process of
production may for some purposes be likened to a box with two openings ; the factors of production are fed into this box from one end and the
product flows out of the other. An observer who looks "at this box
from a distance may be able to see simultaneously what factors, in what
amounts and in what proportions, enter it and what kind and amount
of product comes out of it. Even without analysing what is happening
inside the box, much can be ascertained about the relationships between
what enters it and what comes out of it.
In this study, the box is the national economy. The factors of
production are, broadly, labour and capital. The product that comes
out of the box is the total flow of goods and services expressed in money
units (gross domestic product or G.D.P.). For the purposes of observation and accounting it is necessary to cut the continuous process of
production into convenient, uniform units of time. Usually this unit is
a year.
An observer of the process of production in an economy as a whole
cannot fail to note that the amounts of capital and labour which annually
enter the process, and the amount of output which annually comes out
of it, are changing, becoming sometimes larger and sometimes smaller.
He wonders if the changes in the output are due to changes in the capital
input, or in the labour input, or to something else.
The experience of a single national economy, over a sufficiently
long period, may be expected to yield meaningful answers to this question.
Alternatively, as is done in this study, one can observe a number of
different national economies for a shorter period of time in order to
derive some useful conclusions from a comparative analysis of their
characteristics.
A striking feature of comparisons of national economies is the enormous variety in the amounts of national output. In this study 52 countries
are divided into six groups according to a common measure, namely,
1

Prepared by Mr. Koji Taira of the International Labour Office.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

the value of their national incomes per head of population in terms of
a common currency unit, the U.S. dollar. The first group contains
countries with a national income per head of $1,000 or more in 1956-58;
the other groups range downwards to group VI, containing countries
with a national income per head of under $100.
In addition to the wide differences in national income per head,
economies vary considerably in the extent to which increased inputs
lead to increased outputs. This is apparent from a study of the relationships between the growth of output and the increase in the input of
the " capital " or " labour " factor of production, or in the joint inputs
of these two factors.
As regards the factor capital, the amount of input (investment)
is measured by calculating what proportion of the gross domestic
product has been spent on capital goods. This is the " investment
ratio ". A comparison of the rates of growth of gross domestic product
and the investment ratio in the various countries studied yields a graph
in which the points are widely scattered. If the correlation between
investment ratio and growth of gross domestic product were perfect
all the points on the graph would lie on a straight Une. The more
scattered the points the smaller is the degree of correlation between the
two.
The degree of correlation between the rate of growth of output
and the investment ratio for all 52 countries together is found to be
low (0.20). The group IV countries (per caput incomes from $200
to $350) show the highest coefficient of correlation (0.70). The most
advanced countries (group I) show a low coefficient (0.36). The least
developed countries (group VI) show a somewhat higher coefficient
(0.42).
This generally low degree of association between the rate of growth
of output and the investment ratio has an important practical significance.
A decade ago, when the idea of the investment requirements for a given
rate of economic growth began to be widely discussed, there was a
tendency to assume that a given investment ratio would give rise to the
same rate of growth everywhere, through the operation of a constant
known as the " capital-output ratio ". Thus, if the capital-output ratio
were, say, 3, a country desiring to realise a rate of growth of 5 per cent.
per year would need an investment ratio equal to 15 per cent, of the
national income. The weakness of the relationship between the rate
of growth and the investment ratio observed in the study cautions against
the practical utility of simple arithmetic with regard to the rates of
growth, saving and investment.

SUMMARY OF AIMS, METHODS AND RESULTS

THE LABOUR FACTOR IN GROWTH OF OUTPUT

The absence of a straightforward relationship between output growth
and the investment ratio leads to this question: why is it that some countries realise higher (or lower) rates of growth than others investing the
same proportion of their respective domestic products? There are
many ways of attempting an explanation.
In the first place, in some countries those who undertake capital
investment may be interested only in quick returns; in other countries
they may be willing to invest in more slowly yielding projects; and
some investors may be willing to assume greater risks than others.
As a result the areas of investment activity may be chosen differently in
various nations, with consequent dhTerences in the rate of increase in
output relative to the same rate of investment.
Secondly, the labour force may be increasing faster in the more rapidly
developing countries than in less rapidly developing countries, so that
additions to the productive capital of the economy are more intensively
utilised in the former countries than in the latter.
Thirdly, labour input may not be measured accurately; whether the
labour force is measured by the number of persons or by the number of
man-days worked, the efficiency of work performance that accompanies
the same quantitative indicator of the labour force varies among countries and over time within a single country. Thus, between two countries
the rate of growth of output may differ even though the investment
ratio and the rate of increase in the labour force (in terms of the number
of occupied persons or of the number of man-days or man-hours worked)
are the same. The difference in the rate of growth between the two
countries may be due to changes in the " quality " of the labour factor.
The present study is concerned with the extent to which such qualitative
differences influence the rate of growth.
The only generally available statistical measure of the labour force
is quantitative, i.e. the number of gainfully occupied persons. If this
number increases from one year to the next (whether capital increases
or not) output must increase unless the production of the existing workers
falls simultaneously, or the additional workers produce nothing.
In the countries studied, capital and labour inputs tended to rise
together. At the same time the rate of growth of output was in all
cases higher than the rate of growth of the economically active population
in terms of numbers of persons. The difference between the rate of
growth of output and the rate of growth of the economically active
population represents the rate of growth of productivity per person.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

The problem is to identify the specific factors responsible for this increase
in labour productivity.
This task is approached by constructing a simplified working model
of an economy—i.e. by postulating that the economy works according
to certain simplifying assumptions. The model is of the type known
as a " vintage " model and its characteristics are described as follows :
At a given moment in time a range of alternative techniques is assumed to
exist from which a choice must be made for current investment. Each
technique is characterised by the output it produces, its labour requirements,
and the cost of the capital goods associated with it. Once a technique has
been chosen it cannot be altered: its output and labour requirements do not
change with its age. Its use is assumed to continue until such time as it
ceases to earn a profit. When this happens the capital goods associated with
the technique are scrapped and the labour that was employed to work with
these goods is released for employment in a new plant.1
It follows from these characteristics that any change in output in
an economy from one year to another must be equal to the output of
newly installed plant, minus the output of scrapped plant. 2 Similarly,
a change in employment must be equal to the employment provided
in newly installed plant, less the loss of employment resulting from
the scrapping of old plant. 3
The total return from the new investment undertaken in any year
is equal to the price obtained for the additional output sold.4 This
falls into two parts, of which one p a r t 6 goes to pay the wages of the
additional labour employed, and what is left is the remuneration of
all other factors of production, which in this " two factor " model are
lumped together as capital. This residual amount 6 may be called the
" immediate profit " on the new output produced during the year.
If no plant had been scrapped during the year it would be easy to
see that this immediate profit would be equal to the price obtained for
the sale of the new output, less the wages paid to the workers producing
it.7 At first sight it may seem that we should deduct the loss of profits
1
2

See p. 39.
Or AY=X-XS
If Y stands for total output, and A for " the change in ",
Xsstands for the output of newly installed plant, and
X stands for the output of scrapped plant.
3
Or AL = N-Ns
If AL stands for the change in employment,
N stands for employment in newly installed plant, and
Ns stands for employment in scrapped plant.
4
Denoted by pA Y.
5
Denoted by wAL.
6
pAY-wAL.
7
That is, would be equal to pX—wN.

[1]

[21

SUMMARY OF AIMS, METHODS AND RESULTS

from plants scrapped during the year, but it will be recalled that it is
one of the assumptions of the model that plant continues in use until
such time as it ceases to make a profit. Thus no profits would have been
earned on scrapped plant and we do not have to modify the statement
in the first sentence of this paragraph.1
This immediate profit may also be expressed as a rate of return on
the capital invested. The cost of the plant installed this year is this
year's investment in physical capital measured in current prices.2 The
amount of this year's profit on new plant 3, divided by the cost of the
new plant, gives us a rate of return 4 on the capital invested.5
Expressing immediate profit as a rate of return on capital in this
way, it can be shown that the rate of growth of output has two components, one depending on labour and the other on capital. The labour
component is the rate of growth of employment multiplied by the
proportion of the national product which is received as wages. The
capital component is similarly the product of two terms—the investment
ratio, i.e. the proportion of the national product that is spent on capital
goods, and the immediate rate of profit on current investment.6
Labour inputs, as noted above, are usually measured in terms of
numbers of men or numbers of man-hours. But there is nothing in
this model that makes it necessary to measure labour input in this way.
If the quality of labour is gradually improving—if, for example, workers
1
2
3
4

That is, pAY-wAL=pX-wN
This is denoted by /.
pX-wN.
Denoted by r.

[3]

pX-wN

That is, r =

This can be shown in symbols as follows. From equation 4 it follows that:
pX-wN

[41

= ri

[5]

From equations 3 and 5 it follows that
pAY-wAL

= ri

. . .

[6]

By manipulating equation 6 we get the rate of growth on the left-hand side and the
factors on which it depends on the right-hand side. The manipulations necessary
are to divide throughout by p Y and to rearrange the terms. Dividing equation 6
by p Y throughout gives us :

AY wAL
y

= —

[71

This can also be written :
AY
(wL)AL
I
— =
-— + r
Y
{pY) L
pY

[8]

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

are becoming better educated, better nourished, more highly skilled or
otherwise more productive—one way of describing this is to say that
the effective increase in the input of labour will over a period of time
exceed the rate of growth in the number of heads in the labour force.
The rate of growth of labour input will be the rate of growth in the quality
of labour, plus the rate of growth in the number of workers.1
It is then shown that the " adjusted " rate of growth of labour productivity (i.e. that part of the rate of growth of output which is not
attributable to growth in the labour force, measured in numbers of
men, divided by the proportion of the national product received in
wages) is equal to the sum of two terms. The first is the product of the
immediate rate of profit on current investment and the ratio of investment
expenditure to the wage bill. This is termed the " modified " investment
ratio. The second term is the rate of growth in the quality of labour.2
It is further assumed that improvements in the quality of labour come
about for a number of different reasons—better education, better health,
1

This can be written

AQ AL
1

provided it is recognised that L refers to the

number of workers and not to labour input in a broader sense.
Equation 8 can be expanded to take account of qualitative improvements as well
as quantitative increases in labour inputs by writing it in the following form:

AY_ {wLQ)VAL, AQl
AQ,
+—
8

\ + r—

[9]

Since this study is concerned to investigate the effects of qualitative changes

in labour inputs, denoted by — , it is convenient to rearrange equation 9 so as to

have this term standing by itself. This is done in the authors' equations 3.13 and
(in a form in which w stands for wage per man instead of wage per unit of labour)
3.15 on pp. 45-46. Their equation 3.15 is as follows:

(wL)AL

(pY) L
(wL)

(I) AQ
(wL) G

(PY)
The whole of the left-hand side of this equation can be represented for convenience
by the symbol Zo • This is the variable that the statistical exercises are designed to
explain.
The modified investment ratio, in the sense defined above, is denoted, also for
convenience, by Z i . Equation 3.15, in the shorter form:
AQ
Z 0 = rZi+—

[10]

is the first of two alternative versions of the authors' model. The second is derived
from it in a manner explained by them.

SUMMARY OF AIMS, METHODS AND RESULTS

and so on—the effects of which can simply be added together.1 Thus,
a model is obtained by means of which it is possible to explore the quantitative relationships between an index of economic growth on the one
hand (the adjusted rate of growth of labour productivity as defined above)
and on the other hand the modified investment ratio and a series of
labour quality factors identified below.
The next step was to seek to determine statistically, on the basis
of actual data, to what extent the modified investment ratio and the
rates of growth of the labour quality factors account for the adjusted
rate of growth of labour productivity. The procedure followed was,
first, to see how far, in terms of the equation to be tested, the modified
investment ratio alone explains the adjusted rate of growth of labour
productivity, and then to add the rates of growth of a number of labour
quality factors to see whether the degree of explanation improves.
It should be noted that what is undertaken in the present study is
a " statistical explanation ", which is something quite different from an
attempt to establish causal relationships. Statistical analysis is basically
concerned with finding associations among phenomena. " Good explanation " in the statistical sense means, so to speak, a " remarkable coincidence " among the phenomena compared. However, coincidence
can sometimes be so remarkable that one is compelled to doubt that it
was due to mere chance. Improvement of statistical " explanation "
of relationships among phenomena is indeed no more than an increase
in the degree of coincidence, which may strengthen the feeling that there
must be more than mere chance behind these relationships.2 The
statistical measure of the extent to which the variability of one phenomenon is " explained " or accounted for by that of several others is
called the " coefficient of determination ", which varies from zero to
one (or 100 per cent.).
According to the coefficients of determination found in the present
study the inter-country variations in the investment ratio account for
a small proportion of inter-country differences in the adjusted rate of
growth of labour productivity.
1

In mathematical terms, the model is completed by substituting for the variable

— , whether in equation 10 or in the alternative version of it, an expression in

ß
the form a«

AQ,

AQ2
ha,

AQn
h...a.

AQ,
where

AQ2
,

.
etc. stand for improve-

ments in education, health or other aspects of the quality of labour, and the values
of <*i, a 2 , etc. have to be determined from statistical data.
2
That statistical " explanation " can arouse considerable " feeling " of this
nature is demonstrated by the strong public response to the statistical relationship
between smoking and lung cancer.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

The best, but not very encouraging, relationship is found for the
countries in the two highest national income groups, taken together;
for these the coefficient of determination is 0.20, indicating that variations
in the rate of investment account for 20 per cent, of those in the adjusted
rate of growth of labour productivity.
Thus, once again, it seems that only to a limited extent can labour
productivity be said to increase with increased capital per worker.
This conclusion leads to the hypothesis that the portion of the increase
in labour productivity unexplained by investment must, to some extent
at least, be due to improvements in the " quality " of the working population as a whole. There is, unfortunately, no direct measure of this
quality comparable to the simple number by which persons are counted
or the currency units in which capital is valued. The degree of qualitative
improvement can be measured only indirectly by the changes in factors
which are considered closely related to such quality. The availability
and international comparability of statistical materials severely limit
the selection of the labour quality indicators. Subject to this limitation,
the study presents and examines statistical materials relating to four
major groups of social factors that probably have an impact on labour
quality.
(1) Education. The production potential of education has come in
for much discussion during the last few years. However, there are
many conceptual difficulties in the way of measuring this factor. The
most obvious statistics are those for school enrolment, which are available
for different levels and types of education. In addition to the primary,
secondary and higher education categories there are, for many countries,
separate data on vocational schools and adult education but not, unfortunately, on those directly productive forms of skill formation that
occur outside formal education systems.
The effects on economic growth of the different categories of
education are not uniform. Time lags constitute one problem. An
increased expenditure on primary education in a given year will not
become an economic asset until some years later. But the lag may
be smaller for such other forms of education as short-term vocational
training.
Another problem is that of the intrinsic value of a particular type
of education as a development stimulus. The case for vocational training
is clear. Adult education, on the other hand, varies greatly in its
purpose. Of conventional primary, secondary, and higher education
there can be little doubt in terms of ultimate contribution to economic
efficiency, though results may vary with the specific type.

SUMMARY OF AIMS, METHODS AND RESULTS

Despite these difficulties some choice of statistical indicators must
be made. Four indicators have been chosen to represent education
as a labour quality improvement factor. They are—
(a) primary school enrolment as a percentage of population aged
5 to 14 years;
(b) secondary school enrolment as a percentage of population
aged 15 to 19 years;
(c) vocational school enrolment as a percentage of population
15 to 19 years; and
(d) higher educational enrolment as a percentage of population
aged 20 to 24 years.
In each case the rate of increase (or decrease) of the ratio has been
calculated over the period 1950 to 1960 or for as large a fraction of
the period as the availability of data allows, using terminal years rather
than averages of annual data to determine the rates.
(2) Health. The indicators chosen to indicate progress in health
are—
(a)
(b)
(c)
(d)

infant mortality;
number of inhabitants per physician ;
number of hospital beds per 1,000 inhabitants;
calories available per head.

(3) Housing. The housing indicators are—
(a) dwelling units completed per head ;
(b) the ratio of fixed capital formation in dwellings to the gross
national product.
(4) Social Security. Social security is represented by the following
indicators :
(a) social security benefits paid as a percentage of national income;
(b) average annual social security expenditures per head of population between 15 and 64 years of age, in constant prices.
There are thus 12 indices of labour quality improvement factors
(four indices each for health and education, and two each for housing
and social security). However, all 12 indices are not available for all
countries. In order to show to what extent (in terms of percentages)
the international variabihty of the adjusted rate of growth of labour

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

productivity is accounted for by investment alone or by the combination
of investment with one or more of the labour quality improvement
factors, coefficients of determination (R2) have been computed. Because
of differences in data availability for various groups of countries, the
number of countries for which the coefficients in table I have been
computed varies from column to column in each row.
Nevertheless, the column-by-column comparisons of the ability of
international differences in various factors to account for differences
in the adjusted rate of growth of labour productivity are of interest.
A full array of the capacity of the modified investment ratio to
explain differences in this rate of growth is shown in column 1 of table I.
The modified investment ratio appears to explain this growth more
effectively for the developed countries than for the developing ones,
although even for the former, 80 per cent, of the variations between
countries are not accounted for by differences in the modified investment
ratio alone.
The next four columns (2-5) of the table show how the explanation
of inter-country differences in the adjusted rate of growth of labour
productivity is improved by combining with the modified investment
ratio each of the four major groups of labour quality factors. For
" all groups " of countries, the addition to the modified investment
ratio of the labour quality factors increases the explained proportion
of the inter-country differences except in the case of the education
factors.1 For the individual groupings of countries the addition of the
education factors to the explanatory variables substantially improves the
explanation. The failure of the education factors to improve the explanation for all groups of countries is due in part to the great inter-group
variability of education factors, while these factors are more nearly
related to the rate of growth of labour productivity within each grouping
of countries. For all groups of countries, social security improves the
explanation of this rate of growth better than the other labour quality
improvement factors.
For the groups of less developed countries (groups III-VI) the most
impressive improvement in the explanation of the rate of growth of
labour productivity is observed when the health factors are combined
with the modified investment ratio. For these countries the modified
investment ratio and health factors together account for 72 per cent.
of the inter-country differences. This percentage is quite substantial.
1
The fact that R' can fall when explanatory variables are added is due to the
fact that some countries for which data on the additional variables are not available
must now be excluded from the sample. R* is not corrected for degrees of freedom.

TABÍ .E J. FXTENT TO WHICH INTERNATIONAL DIFFERENCES IN ADJUSTED RATE OF GROWTH OF LABOUR PRODUCTIVITY
[Z0] Akü EXPLAINED BY INTERNATIONAL DIFFERENCES IN THE MODIFIED INVESTMENT RATIO AND LABOUR QUALITY
IMPROVEMENT FACTORS
C
S

Percentage of differences explained by

So
Groups of countries 1

Inv. ratio
and four
health
factors

Inv. ratio
and two
housing
factors

Inv. ratio
and two
social security
factors

Inv. ratio
and four
education
factors

Inv. ratio
and calories
per head

Inv. ratio,
calories per
head and
higher
education

Inv. ratio,
calories,
higher ed.,
and inv. in
dwellings

Inv. ratio,
calories,
higher ed.,
dwellings
and social
security

Investment
ratio
alone

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

20
11
5
9

39
33
55
7

33
58
5
37

43
| 60
47

•il

o
a
I and II
Ill and IV
V and VI
All groups

I 72
32

1 Not
> com) puted
21

}"
34
a

Source: Appendix II, table 8. The figures are the coefficients of determination (Ä ) in column 3 of that table.
1

I:
II:
III:

The groups of countries, ranked in descending order of average annual per caput national income in 1956-58, estimated in U.S. dollars, are$1,000 and over (6 countries);
IV: S2OO-S350 (10 countries);
S575-S1,000 (12 countries);
V: S100-S200 (11 countries);
J350-S575 (8 countries);
Under SlOO (5 countries).

i Not
\ comJ puted
49

1 Not
'> comJ puted
65

v¡

r
H

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

Next to health, social security is the most important factor in improving
the explanation for less developed countries. For the developed countries
(groups I and II) social security makes a greater contribution than any
other labour quality improvement factor toward the improved explanation (48 per cent, in contrast to 20 per cent, explained by the modified
investment ratio alone). The higher figure for education factors in
the least developed countries (column 5) is also of interest.
The remaining four columns in table I are experiments in the explanation of inter-country differences in the rate of growth of labour productivity by successive additions to the modified investment ratio of representative single indicators of the four groups of labour quality improvement factors (calories per head from the health group, higher education
from the education group, investment in dwellings from the housing
group, and benefits paid per head from the social security group).
Concentrating on the row of all groups of countries, it is clear that more
variables explain the rate of growth of labour productivity more fully.1
For all groups, the modified investment ratio and calories per head
explain 37 per cent, of the inter-country differences in this rate of growth.
By adding higher education, the proportion explained rises to 47 per
cent.; by the addition of still another factor, investment in dwellings,
it rises to 49 per cent. ; and finally, all the five variables together account
for 65 per cent, of the inter-country differences. It should be noted that
the per caput calorie consumption improves the explanation more
substantially than the other factors.
There are other interesting aspects of this statistical exercise. A discussion in detail of these points will be found in the relevant chapters
of the study. However, one point among the findings of the experiment
may be stressed here. It is the sensitivity of the adjusted rate of growth
of labour productivity to the rate of increase in the average calorie
intake of the population. According to the background data related
to column 9 of table 1 2 , for example, a 1 per cent, increase in the rates
of change of selected labour quality improvement factors is accompanied
by the following responses of the adjusted rate of growth of labour
productivity:
Calories per head
Investment in dwellings
Higher education
Social security benefits

1
2

See footnote, p. 14.
See Appendix II, table 8, row 16.

2.27
0.13
0.11
0.04

per cent.
„
„
„
„
„
„

SUMMARY OF AIMS, METHODS AND RESULTS

There is a statistical method of testing whether and to what extent
the observed numerical relationship between two variables is reliable.
Even if the relationship between two variables is highly sensitive as
represented by the large coefficient that relates the two (like the number
that relates calorie intake to the rate of growth of labour productivity
in the previous paragraph) it is possible that this coefficient cannot
be depended upon because of large elements of chance involved in the
relationship. When a statistical test of reliability is applied to the
coefficients that relate the labour quality improvement factors to the
rate of growth of labour productivity it appears that the order of reliability is (1) calories, (2) education, (3) dwellings and (4) social security.
The importance of calories per head as a determinant of labour
quality having been suggested, one might now regroup the factors into
(1) investment, (2) calories and (3) all other factors, to find out how
each of these major variables influences the adjusted rate of growth
of labour productivity and hence the growth of output.1
1
The values found for the different groups are combined according to equations
given in Appendix II, table 8, rows 29-33, the general form being—

Z 0 = a + bZi + c - ^

[11]

where Zo is the adjusted rate of growth of labour productivity, Z i , the investment
ratio divided by the share of wages in the gross domestic product,

, the rate

of increase in the calorie intake per head and a, b, and c are coefficients that are
to be estimated from the data.
In this equation, all the variables are in percentages. The equation merely says,
for example, that Zo grows at the rate of c per cent, per annum if the average calorie
intake (Ô3) grows at 1 per cent, per annum. The equation also says that even if
there is no growth in investment and in the average calorie intake, Zo grows at a
per cent, per annum, because of the aggregate effects of all factors other than
investment and calorie intake.
Numerical values can be given to the coefficients in this equation in a set of three;
the first refers to the most advanced countries (groups I and II), the second to all
other countries (groups III-VI), and the third to all the countries.

Z 0 = 1.66+0.0854Z1 + 1.24dß3/Ö3
Z 0 = 0.98+0.1064Z1 + 1.65JQ3/Ô3
Z 0 = 1.32 + 0 . 0 9 5 ^ + 1.53^103/03

UKOl
[1100]
[11 (Hi)]

In each equation the degree of reliability of the coefficients (as mentioned in a
previous paragraph) is in descending order: (1) Q3 (increase in calorie intake),
(2) Z\ (investment ratio divided by the share of wages in value added), and (3) all
other factors.
It is possible to translate these estimates into estimates of how the rate of growth
of output is affected by investment and calorie intake. To do this it is necessary

AY
to replace equation 11 by an equation that has on the left-hand side not Zo, but >

This means going back to the form of equation 8 in footnote 6 on page 9 but expanding
it as follows:
(footnote continued overleqf)

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

In concluding this section, it again seems appropriate to sound a
warning. Statistical relationships among variables, however carefully
computed, do not by themselves establish relationships. For example,
the relationship between the rate of growth of labour productivity and
calorie intake may be interpreted in at least two ways: (1) people who
are more productive earn more and thus eat more, and (2) people who
are better fed work more and thus are more productive. In judging
in what direction the line of cause and effect runs, the ultimate test is
a pragmatic one. The services to production rendered by a man depend
on his ability and the energy he puts into his work. Ability is nurtured
by education, and energy expended is conditional on the availability
AY
Y

(wL)f
AL bAQ3
I
\a-ì
+
++ rr —
—

[12]

wL
For — , the share of wages in gross national product, we may write s, and for
/
PY
, the investment ratio, we may write K. The equation cannot be used to find out
PY
AY
how the rate of growth of output,
, is affected by the other variables unless a
particular value of s is assumed. Estimating s is not easy, given the quality of currently
available data. However, the data which are available suggest that for purposes of
exposition a value of s of 70 per cent, is worth considering. Assuming this value for
s, the estimates given in equations 11 (i) to 11 (iii) can be expressed in the form of
equation 12 with the following results:

AY
AL
AQ3
— = 1.16+0.70— + 0.87—+0.085K
Y
L
Q3

(12(01

AY
AL
AQ3
— = 0.69+0.70—+ 1.16—+0.106K
Y
L
03

[12(ii)J

AY
AL
AQ3
— = 0.91+0.70— + 1.07—+ 0.095K
Y
L
Q3

[12 (iii)l

The above set of equations has many interesting implications. Suppose, for
example, that all the countries are uniformly investing 10 per cent, of their gross
domestic product and increasing per caput calorie consumption by 1 per cent, per
annum: then, if the employed labour force grows at 2 per cent, per annum, output
will grow at 4.3 per cent. This calculation is made as follows:
(a) Developed countries: output grows at 4.28 per cent, per annum:

4.28 = 1.16+ (0.70x2)+ (0.87x1)+ (0.085x10)
(b) Underdeveloped countries: output grows at 4.31 per cent, per annum:

4.31 = 0.69 + (0.70x2) + (1.16xl)+(0.106xl0)
However, whilst the growth rates for these two groups of countries are the same,
the sources of this growth differ. In underdeveloped countries the effects of investment and calorie consumption are greater than in the developed countries, whilst
the effects of all other labour quality improvement factors are less.

SUMMARY OF AIMS, METHODS AND RESULTS

of its raw material, namely calories. There seems to be at least a
preliminary case for interpreting the results of this study as indicating
that calorie intake is an important determinant of labour quality, with
other factors contributing somewhat less.
THE NEXT STEPS

The results achieved in this study are far from conclusive. They
represent merely an essay in economic and statistical insight into a
major problem of our time. Much more work will have to be done
before firm policy lines emerge. This work should not be postponed,
since policy decisions cannot be deferred but must be made even in the
absence of sufficient knowledge.
There are at least four lines of attack which, in the opinion of the
authors, should prove fruitful:
(1) The present sample of 52 countries should be expanded to include
additional nations, particularly the less developed ones. Each year
the statistical yearbooks of the United Nations family are being expanded
in coverage, and the sample will automatically increase on this account
alone. Special efforts can be made to secure data from countries which
have not yet submitted their statistics to the United Nations. At the
same time, there is great need for refinement of the existing data. Many
of the estimates on which this study was based are quite crude, but they
were essential if any progress at all was to be made.
(2) There is need for a less aggregative approach than the one adopted
here. This means, for example, that the unit of investigation might be
the non-agricultural sector of the economy, or the manufacturing sector,
or even individual industries. Particularly in the less developed nations
the large subsistence sector of agriculture tends to be little affected by
the various measures which are analysed here, and presumably a less
global analysis would reveal much greater sensitivity in the relationship
between, say, manufacturing output and inputs of education, health, etc.
(3) A useful supplement to broad statistical analysis would be intensive study of a smaller number of countries. For example, six countries
might be chosen for this purpose, three which had developed at a satisfactory rate and three which had not. Careful comparison could be
made of the respective policies regarding the kinds of indicators which
have been used here, to see whether any patterns which might be
described either as favourable or unfavourable emerge.
(4) Even more intensive study of the relationships between production and the various economic and social inputs, at the level of the

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

individual firm, might prove very revealing. For example, the production records of a single enterprise which had installed housing, health,
and educational facilities for its employees might be examined over a
period of time to determine whether there were any specific quantitative
links between these investments in human beings and the productivity
and quality of the labour force. Several such studies in different countries might, in thefinalanalysis, tell a more interesting story than broader
statistical investigation at the national level.

CHAPTER I
PREVIOUS RELEVANT WORK
In this chapter we review briefly some previous work that we found
particularly helpful. This is by no means intended to be an exhaustive
catalogue of the relevant literature on growth or the qualitative aspects
of labour; such a list would be a long one and take us too far afield.
It is our purpose, rather, to sketch the evolution of the concepts and the
procedures which underlie the present essay. It might be added that,
in addition to the considerable bulk of printed material available, there
are some valuable documents buried in the files of international agencies
which deserve to be made available to a wider audience.1
One of the most important studies that has thus far appeared on the
subject under consideration is the Report on the World Social Situation 2,
prepared by the Department of Economic and Social Affairs of the
United Nations. This study is thefirst,to our knowledge, which attempts
to analyse on an international basis and in a comprehensive fashion
the relationship between economic growth and such factors as health,
medical care, nutrition, housing, education, wages and conditions of
labour, and social security. The methodology employed was to rank
some 74 countries on the basis of the level of economic development,
on the one hand using per caput national income and per caput energy
consumption as indicators, and on the other according to appropriate
indicators of such facets of social progress as health, education, and
nutrition. Through rank correlation techniques it was found that
indicators such as infant mortality, school enrolment ratios, and calorie
consumption were fairly closely related to national income per head.
Inter-country comparisons brought to light some discrepancies between
the level of development and the social indicators, leading to the conclusion that—
1

One might mention in particular a series of country case studies commissioned
by the United Nations Economic and Social Council. In this series the reports
on Uganda by Professor David Walker (E/CN. 5/346/Add.9, 18 Apr. 1962) and on
the Netherlands by Professor W. Brand (E/CN.5/346/Add.6, 27 Sep. 1961) are of
special interest in the present context.
2
Report on the World Social Situation, op. cit.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

those countries where considerable discrepancies exist between the economic
and social indicators.. .are usually countries where the existence of social or
economic strains reflecting this disparity are widely recognised; especially
where the economic indicators are much higher than the social, political strain
and instability are also apt to be quite marked.1
On the question of causation, the Report is understandably cautious.
The important point is made that many social expenditures which have
been regarded as primarily in the nature cf consumption are, in fact,
investments as well: a view that is gaining wide acceptance. However,
it is stated that social programmes " have no simple and consistent
relationship to economic development. They are of wide variety,
with varying economic implications. It is wishful thinking to assume
that each of them will contribute substantially to economic growth." 2
For example, a public health programme can, on the one hand, raise the
productivity of labour and, on the other, lead to lower mortality and
increased population pressure on available resources. Minimumwage legislation and housing can lead to an improvement of labour
quality, but it may also slow down industrial capital formation. An
important strategy of economic development is " to examine economic
implications and select, as far as possible, from among specific alternative
social programmes directed toward the same goal, those programmes
that can be shown to be economically most advantageous." 3
The problem of quantification, the Report points out, is complicated
by the fact that " there is no common mathematical measure of economic
and social development, no way of equating economic and social values
in order to add them up on a common scale." 4 Nevertheless, the Report
takes a significant step forward by demonstrating that it is possible
to find a substantial number of quantitative social indicators that can
be arrayed in a meaningful fashion to yield provocative questions, if
not definitive answers.
A second study which we found to be of very great interest was of
a quite different character. This is the growth study prepared by the
United Nations Economic Commission for Europe (E.C.E.) as part of
an economic survey.5 One of the main purposes of the study was to
determine whether, for the countries of Europe plus the United States
and Canada, there was a systematic relationship between the rate of
1

Report on the World Social Situation, p. 61.
Ibid., p. 33.
8
Ibid., pp. 34-35.
1
Ibid., p. 38.
6
United Nations, Economic Commission for Europe: Economic Survey of Europe
in 1961, Part 2: Some Factors in Economic Growth in Europe during the 1950s
(Geneva, 1964).
2

PREVIOUS RELEVANT WORK

economic growth and inputs of labour and capital. The following
conclusions emerged from correlating the appropriate indicators.
1. There was a fairly strong degree of correlation between the rate
of growth of the labour force and of gross domestic product.
2. The association of rates of growth of domestic product with
rates of fixed capital formation was less strong than that of growth
rates of labour force and domestic product.
3. The multiple regression equation linking the three variables
left a large residual (about half) of the total variation in rates of growth
among the countries " unexplained " by inputs of labour and capital
taken in conjunction. The indicator of labour input used, it is important
to note, was simply the increase in the economically active population
over the period covered (the 1950s), no attempt being made to correct
for quality factors.
4. These conclusions are not greatly altered when the comparisons
are restricted to the manufacturing sector, rather than embracing the
entire economy.
These summary statements do not begin to do justice to the wealth
of statistical data assembled and the careful methodology employed
by the authors of the study, which is certainly the most ambitious
international comparison of its kind thus far undertaken. One of its
major findings, from our point of view, was the negative one that economic growth could not be " explained " satisfactorily on the basis of
the quantity of capital and labour inputs. This was hardly a new
finding, for most economic growth studies in recent years have come up
with the same conclusion. But, because it relied on cross-national
data rendered comparable with unusual thoroughness, the demonstration
was more convincing than in the case of studies confined to single
countries.

CHAPTER II
THE METHOD OF THE PRESENT STUDY
As a starting point for our analysis, data comparable to those collected
by the E.C.E. were gathered for 36 countries in addition to the 16
countries covered in its study, thus making a total of 52 countries in
all. These data are shown in table II and presented in alternative
forms in Appendix III, figures 1 to 4.
Using these data, we reproduced for the entire sample of 52 countries
some of the regressions which the E.C.E. study presented for its sample
of 16 countries.1 The basic estimating equation was
AY
Y

AL
= a^T

I
r -

[2.11

where — is the annual rate of growth of the gross domestic product,

is the annual rate of growth of the economically active population,
I
and — is the average annual ratio of gross fixed investment to the gross
pY
domestic product.2 Our estimates of the coefficients ß and y and the
constant term a are shown in table III.
In order to determine whether the relationships were influenced
by the degree of economic development, the sample of 52 countries
was subdivided into per caput national income groups, ranging from
the highest, group I, to the lowest, group VI.3 Since some of these
groups contained relatively few observations, they were combined into
three broader groups, for which coefficients were also estimated.
Table III contains four sets of estimates. The first refers to the
regression of the rate of growth of gross domestic product (G.D.P.) on
the rate of growth of the labour force ; the second is analogous for the
investment ratio and the rate of growth of G.D.P. ; the third shows the
L

We have not undertaken any analysis for individual sectors of the economy.
Such analysis was one of the most significant parts of the E.C.E. study.
2
" / " represents investment expenditure in current prices, " Y " total output,
and " p " the current factor cost per unit of output.
8
The rationale of this grouping is discussed below, p. 67.

THE METHOD OF THE PRESENT STUDY

results of the multiple regression of G.D.P. growth on labour force growth
and the investment ratio ; the fourth represents a similar multiple regression with the constant term a excluded.
The results tend to confirm the E.C.E. study conclusions, but contain a warning that they cannot be generalised. The simple relationship between the growth of G.D.P. and labour force is strong only
for group II income countries, most of which are in the E.C.E. sample.
For lower income groups (except for the very lowest, which includes
only six countries) there is virtually no correlation between labour force
growth and economic progress as exemplified by the growth in G.D.P.
The observation of the E.C.E. that " it might reasonably be supposed
that the correlation of growth rates of output and of labour force would
be stronger in the industrialised than in the economically less developed
countries—where the active labour force is likely to be a particularly
poor indicator of effective employment "* is borne out by our data.
The results on the capital side are less satisfactory. The investment
ratio has almost no correlation with the growth of G.D.P. when all
52 countries are considered together. When the sample is broken
down by income group, it is for the more advanced countries that one
finds the better association, R2 declining from 0.30 for the two highest
income groups to 0.02 for the two lowest. The negative coefficient
of investment for countries in group V cannot be regarded, of course,
as implying an inverse relationship between investment and growth.
It must be read in conjunction with the high constant term for this group
which results from the manner in which the regression line was fitted
to the observations. When the constant is eliminated, the results become
more reasonable.
The multiple regression between labour and capital and growth
reveals a moderate degree of correlation (for all the countries, R2 is
0.31). In detail, however, the results are not always meaningful. For
example, there is a low degree of association for the two lowest income
groups (R2 = 0.08) about the line
AY
AL
I
— = 2.1+0.3
2.9—
Y
L
pY
This implies, for these countries, a rate of growth of 2 per cent, unrelated
to changes in inputs of labour and capital, and a net negative relationship
1
Economic Survey of Europe in 1961, Part 2, op. cit., Ch. II, p. 14. The low
coefficient of correlation for the highest income group is based upon a sample of only
five countries, and cannot be regarded as disproving the general hypothesis. It will
be noted that there is a steady decline in the correlation ratio when the six income
groups are combined into three of almost equal size.

TABLE II. RATES OF GROWTH OF GROSS DOMESTIC PRODUCT, UNADJUSTED AND ADJUSTED LABOUR PRODUCTIVITY,
AND THE INVESTMENT RATIO FOR 52 COUNTRIES, 1950-60 a
(In percentages)
ON

Country

Algeria
Argentina
Australia
Austria
Belgium
Brazil
Canada
Ceylon
Chile
China (Taiwan)
Colombia
Costa Rica
Cyprus
Denmark
Ecuador
Finland
France
Germany (Federal Republic)
Greece
Guatemala
Honduras
Iceland
Ireland
Israel
Italy

Annual increase
of gross domestic
product at
constant prices

Average investment
ratio

Annual increase
in labour
productivity

Annual increase
in adjusted labour
productivity

Annual increase
in economically
active population

(2)

(3)

(4)

(5)

4.90
1.66
4.11
5.83
3.13
5.47
3.81
3.57
2.35
6.55
4.57
5.43
2.56

2.60
0.12
2.06
4.74
2.94
2.37
1.29
1.03
1.42
3.39
2.04
2.78
2.11

4.76

13.3

4.55
4.01
7.17
5.48
4.85
3.68
6.78
1.74
7.90
5.83

29.6
20.2
24.8
18.7
12.6
14.5
30.4
17.9
28.7
22.7

3.11
0.45
2.69
4.96
3.02
3.35
2.18
1.71
1.67
4.59
2.84
3.43
2.25
2.42
2.93
4.00
3.56
5.99
3.75
2.93
2.25
5.39
2.54
4.81
5.10

2.30
1.54
2.05
1.09
0.19
3.10
2.52
2.54
0.93
3.16
2.53
2.65
0.45

3.13

26.0
21.2
28.8
24.2
17.6
16.7
26.1
11.4
9.9
15.9
18.0
19.9
18.5
19.3

2.16
1.57
3.82
3.37

5.57
2.93
1.86
1.69
4.80
2.80
3.71
4.75

0.97
3.19
0.73
0.64
1.60
2.55
2.99
1.99.
1.98
-1.06
4.19
1.08

o
Ti

r
>
CO

o
>
o

tn
O
O

o
g
o
o
ra

Jamaica
Japan
Korea (South) . . . . ,
Luxembourg
Malaya (Federation of)
Malta
Mauritius
Mexico
Netherlands
New Zealand
Nigeria
Norway
Panama
Peru
Philippines
Puerto Rico
Portugal
South Africa
Spain
Sweden
Switzerland
Thailand
Tunisia
Turkey
United Kingdom. . . .
United States
Venezuela

7.51
8.83
4.53
3.86
6.00
4.41
3.41
5.05
4.18
5.71
4.71
3.45
4.42
3.36
6.35
5.78
4.50
4.36
5.25
3.16
4.41
4.42
2.43
6.22
2.57
3.20
7.17

18.4
25.3
12.4
23.2
8.6
22.8
15.2
14.5
24.8
23.7
10.2
31.3

12.5
24.3
8.1
21.1
16.0
23.3
16.3
22.1
24.9
15.0
12.8
12.5
16.3
18.3
26.2

4.89
6.28
1.17
2.78
4.04
3.04
-0.18
1.71
3.00
3.64
3.00
3.12
2.05
0.82
3.90
5.22
3.88
1.93
4.40
2.65
2.87
2.32
0.83
3.89
2.00
1.23
4.30

5.79
7.04
2.17
3.19
4.67
3.31
1.12
2.78
3.33
4.30
3.51
3.26
2.59
2.31
4.63
5.32
4.10
2.71
4.61
2.79
3.34
2.95
1.31
4.64
2.15
1.80
5.15

2.62
2.55
3.36
1.08
1.96
1.37
3.59
3.34
1.18
2.07
1.71
0.33
2.37
2.54
2.45
0.56
0.62
2.43
0.85
0.51
1.54
2.10
1.60
2.33
0.57
1.97
2.87

Sources:
Gross domestic product: Appendix II, table 1.
Investment ratio: Appendix II, table 3. Calculated as the average ratio of gross fixed capital formation to gross domestic product over the period.»
Economically active population: Appendix II, table 2.
Labour productivity: Calculated as the difference between the rates of growth of gross domestic product and labour force (column 1 minus column 5).
Adjusted labour productivity : Calculated as the difference between (a J the rate of growth of gross domestic product at constant prices and (b) the rate of growth of the
labour force multiplied by the wage share. The latter figure is derived from Appendix II, table 3.
° For some countries, the period covered is less than the full decade 1950-60. The precise period covered for each series is given in Appendix II, tables 1, 2, and 3.

TABLE III. RESULTS OF THE REGRESSION OF THE RATE OF GROWTH OF OUTPUT ON THE RATE OF GROWTH OF
LABOUR AND THE INVESTMENT RATIO

Constant
term a

Standard
error of a

Labour
variable ß

Standard
error of 3

Investment
variable Y

Standard
error of Y

Number of
countries in
the sample

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Relation of G.D.P. and labour force:
Group I
Group II
Group III
Group IV
Group V
Group VI
Groups I and II
Groups III and IV
Groups V and VI
All groups

3.2
2.9
3.2
5.4
4.5
1.8
2.8
3.3
3.7
3.3

1.2
0.4
0.8
2.0
1.1
1.9
0.5
0.7
0.9
0.4

0.5
1.4
0.7
0.3
0.0
1.2
1.2
1.0
0.3
0.7

0.6
0.3
0.5
0.8
0.4
0.8
0.3
0.4
0.4
0.2

0.13
0.75
0.17
0.02
0.00
0.42
0.53
0.30
0.06
0.24

6
12
9
9
11
5
18
18
16
52

Relation of G.D.P. and investment:
t:
Group I
Group II
Group III
Group IV
Group V
Group VI
Groups I and II
Groups III and IV
Groups V and VI
All groups
.

1.8
—0.4
—0.9
2.8
5.3
0.9
—0.2
3.3
4.9
3.7

3.0
2.3
2.5
1.4
0.9
4.5
1.9
1.8
1.0
0.7

0.13
0.34
0.33
0.49
0.09
0.18
0.30
0.05
0.02
0.04

6
12
9
9
11
5
18
18
16

Relation of
gross domestic
product

9.5
21.4
24.3
20.4
—5.3
27.4
19.5
8.7
—2.9
5.1

12.3
9.4
12.2
8.5
5.6
33.9
7.5
9.6
6.1
3.7

3. Relation of G.D.P. to labour force
and investment:
Group I
Group II
Group III
Group IV
Group V
Group VI
Groups I and II
Groups III and IV
Groups V and VI
All groups . . .

1.7
0.8
—0.7
2.7
5.5
0.6
0.5
1.4
4.1
1.9

4. Relation of G.D.P. to labour force
and investment (without a constant
term):
Group I
Group II
Group III
Group IV
Group V
Group VI
.Groups I and II
Groups III and IV
Groups V and VI
All groups . . .
Source: Calculated from the data in table H.
Group I
(6 countries) S1,000 and over (average annual per
Group II (12 countries) $575-1,000
"
"
Group III (8 countries) $350-575
»
»
Group rv (10 countries) $200-350
»
Group V (11 countries) $100-200
»
»
Group VI (5 countries) under $100
"
"

3.3
1.4
2.6
2.0
1.5
4.6
1.4
1.7
1.3
0.8

0.4
1.2
0.4
0.1
—0.1
1.0
1.0
1.0
0.3
0.8

0.7
0.3
0.5
0.7
0.4
1.0
0.3
0.4
0.4
0.2

7.2
9.6
20.8
20.2
—5.4
11.6
11.0
10.1
—2.9
7.0

0.1
5.9
13.1
9.4
6.0
37.6
6.3
8.1
6.1
3.2

0.20
0.81
0.39
0.49
0.09
0.45
0.61
0.37
0.08
0.31

6
12
9
9
11
5
18
18
16
52

33
ta

8
0.4
1.2
0.5
0.6
1.1
1.0
1.0
1.1
1.1

0.7
0.2
0.5
0.5
0.4
8.4
0.3
0.3
0.3

13.8
13.0
17.5
28.1
10.1
15.5
12.9
16.4
10.1

5.2
1.7
3.5
7.7
6.5
15.7
2.0
3.4
5.6

0.14
0.80
0.39
0.31
0.01
0.44
0.61
0.37
0.06

1.0

0.2

13.7

1.7

0.22

caput national income of 1956-58).
"
"
"
"
»
»
»
»
»
»
»
»
»
»
"
"
"

6
12
9
9
11
5
18
18
16
52

a
M

H
c/î

to
v©

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

between the increase in the investment ratio and in G.D.P. The results
are more reasonable for the other income groups: for the two highest,
the autonomous rate of growth given by the constant term is small,
and there is a one-to-one relationship between growth of G.D.P. and of
the labour force. For every per cent, increase in the investment ratio
the rate of growth increases by 0.11 percentage points. The relationship
for the two intermediate income groups is quite similar, except for the
somewhat greater autonomous growth rate.
To complete this part of the study, the constant term was dropped
from the regression equation, throwing the entire burden of explanation
on the labour and capital coefficients. The effect of this was to produce
a more plausible set of relationships, particularly in those cases in which
the constant term had implied a high rate of autonomous growth. It is
interesting to note that the labour coefficients are now almost identical
for all income groups when combined into three sub-groupings, a 1 per
cent, increase in the labour force implying a 1 per cent, increase in the
growth rate of G.D.P. The investment contribution has increased
substantially from the previous case, taking on most of the explanatory
contributions of the omitted constants. For the entire sample, it
appears that increasing the investment ratio by 7 percentage points has
about the same impact on growth as a 1 per cent, increase in the labour
force. However, the value of R2 is not high enough to permit one to
attach too much significance to this relationship.
This part of our study led us to the preliminary conclusion that a
simple model, restricted to the two independent variables of the quantity
of labour inputs and the investment ratio, is not likely to yield a satisfactory explanation of economic growth, even when applied to the
experience of a substantial number of nations. This may be too pessimistic; perhaps an improvement in the quality of the data, a longer time
period, and a larger sample of countries, particularly at the lower end
of the income scale, would produce more stable results. But as far as
we have gone, our results tend to substantiate those obtained by other
investigators, and particularly the conclusions reached by the Economic
Commission for Europe.
In pursuit of a more satisfactory explanation of the data of table II,
we proceeded to an analysis along fines similar to those followed by the
E.C.E. This was conducted largely by graphical methods, and the
relevant information is shown in figure 4. A key to the interpretation
and construction to be put on the data when expressed in this way is
given in figure 5 (see Appendix III).
Figures 4 and 5 show the rates of growth of various parameters of
economic development plotted against the investment ratio, i.e. the

THE METHOD OF THE PRESENT STUDY

ratio of gross investment to the gross domestic product.

31
This ratio is

denoted by — where / is current gross investment expenditure, Y is
the quantity produced, andp is the primary factor cost per unit of output.
Thus, p Y is the domestic product in current prices.
Three different growth rates are plotted against the investment ratio
AY
for each country. The first is the rate of growth of output, — . Thus
the construction of figure 4 begins with the detail of figure 2. The
points in figure 2 appear as the tops of the vertical lines in figure 4.
The length of these lines are the rates of growth of the economically
AL
active population, — .x Hence the ordinates of the bottom of the
L
AY AL
lines are the rates of growth of productivity,
, i.e. the rate of
i

growth of the gross domestic product at constant prices minus the rate
of growth of the labour force.
A ray drawn from the origin through the top point of a line determines the incremental capital-output ratio (ICOR). 2 This can be
read off for each country from the scales on the upper and right-hand
borders of figure 4. This is the standard ratio familiar from the literature
of economic development.
A ray drawn through the bottom point of the line determines what
the E.C.E. has termed the ICOR(L), obtained by dividing the investment
ratio by the growth rate of labour productivity.3 However, the E.C.E.
points out that this measure has the following limitations:
To attach significance to this coefficient, and to treat capital formation
simply as a determinant of the increase in labour productivity, implies acceptance of the extremely questionable assumption that the " true " marginal
productivity of labour... is equal to its average productivity. In other terms,
it is assumed that the average level of output per head of a growing labour
force can be disregarded.
1
The rate of growth of the labour force is plotted against the growth of G.D.P.
infigure1.
J ¡AY
I
2
Symbolically, this ratio is simply
/
or
, i.e. the ratio of current
pYl Y
pAY
investment expenditure to that part of the current increase in G.D.P. which is
independent of price changes.
3
In terms of symbols infigure5, this can be represented by the expression

/
pY

(AY
\Y

AL\
L

The rate of growth of labour productivity is here defined as the difference between
the rate of growth of output and the rate of growth of the labour force.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

In fact, however, in almost any circumstances a growing labour force
needs to be equipped with productive capital—probably at a rate at least
sufficient to maintain the average volume of equipment per worker—if labour
productivity is not to fall.... Thus, the ICOR(L), regarded as an indicator
of " true " capital productivity (or of " total productivity " in the sense of the
effect of all influences determining the return to inputs of additional labour
and capital), is biased in favour of countries experiencing a slow growth of
labour force. The ICOR (regarded as such an indicator) is, conversely,
biased in favour of countries with a rapid growth of labour force since it
implies that the " true " marginal productivity of labour unequipped with
additional capital is zero—an hypothesis fully as improbable as that underlying
the ICOR(L). Naturally, the two coefficients for a particular country are
equal if there is no change in the labour force.1
To overcome this objection, the E.C.E. calculated (and we did the
same for 36 additional countries) an alternative ICOR(L) based on the
assumption that the " true " marginal productivity of labour is represented by the real wage.2 This yields a somewhat more justifiable
estimate of the contribution of " productivity ", or " the contribution of
all influences on the growth of output other than the increment of labour
supply and the rate of fixed capital formation ". 3
This magnitude has been the subject of much recent interest among
economists, and it is our purpose in the present work to explore its
composition. The alternative ICOR(L) can be determined from a line
1

Economic Survey of Europe in 1961, Part 2, op. cit., Ch. II, p. 32.
wL
3
Symbolically, if — represents the share of wages in G.D.P., the alternative
PY
ICOR(L) can be expressed as follows :
J
AY

~P~Y
/wL\ AL

~T~{P~Y)'T
If one compares the expressions for the ICOR, the ICOR(L) and the alternative
ICOR(L), it is clear that the last must lie between the former two, except in the
wL
unlikely case that — is equal to one or zero.
pY
8
Economic Survey of Europe in 1961, Part 2, op. cit., Ch. II, p. 33. The interpretation of this ratio might be made somewhat clearer by imagining two countries which
( l
\
were investing at the same rate I — equal J, and where the wage share and growth
of the labour force were also identical

(wL AL
\
— equal 1. If the growth rates of
\PY L
)

output differed, the country with the higher growth rate would have a smaller ICOR(L)
(alternative), indicating the greater influence on productivity of all elements other
than the inputs of labour and capital.

THE METHOD OF THE PRESENT STUDY

drawn through the intermediate point of each vertical line in figure 4,
by reference to the scale at the right-hand side. These intermediate
points are shown plotted against the investment ratio in figure 3.
Figure 4 is useful in helping to illustrate the degree to which we are
ignorant of the factors contributing to economic growth and the magnitude of the area which remains to be explored. If a percentage point on
the investment ratio had the same effect in all countries, all of the intermediate points should lie on a single ray drawn from the origin. This is
very clearly not the case, as figure 3 shows. Part of the discrepancy, of
course, may lie in deficiencies and incomparabilities in the underlying
statistical data ; but, even allowing for that, the data undoubtedly reveal
considerable variations in the effectiveness of influences other than the
inputs of labour and capital.
A few examples may serve to clarify the foregoing points. At one
extreme is Norway with a high rate of investment and a very high capitaloutput ratio. The ray through the intermediate point measuring the
Norwegian alternative ICOR(L) approximates to the lower limit of the
alternative ICOR(L) band (only Argentina and Mauritius, countries
which were average or below average investors, were lower). At the
other end of the spectrum is the Philippines, with the lowest investment
ratio and the lowest capital-output ratio. In between are a number
of other bands : on one of them, indicating a capital-output ratio between
7 and 8, are some of the most developed economies—Belgium, Netherlands, Sweden, Switzerland, United Kingdom and United States—but
also some less developed ones: Ceylon, Cyprus, Malta and Tunisia.
Another band, embracing countries with higher growth rates during
the period, includes such developed nations as Austria, the Federal
Republic of Germany, Italy and Japan, as well as such less developed
ones as Brazil, China (Taiwan) and Turkey.
What is one to make of the differences in the experience of countries
illustrated by figure 4? In particular, can it be argued that there are
factors at work which would tend to lengthen the true labour force
growth lines in some countries in the sense that the increase in effective
labour input is greater than the increase in employment, thus tending to
align the alternative ICOR(L) points more closely? Or must it be
conceded that there are other factors, which can perhaps be classified
as capital efficiency factors, which would tend to explain the variations
among the alternative ICOR(L) points ?
The present study is addressed to the first of these queries : the standardisation of labour inputs for quality as well as quantity. Japan,
for example, had an extremely high growth rate, associated with a
fairly (but not unusually) high rate of increase in the labour force, and

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

showed a very low capital-output ratio. Could this have been due,
in some measure, to improvements in the quality of the labour force
through advances in education, health, or nutrition? High growth
rates associated with increases in the labour force are not as remarkable,
from the point of view of our inquiry, as those which stem from labour
inputs of a superior quality.
Several other recent studies, while not entering into our work directly,
illustrate the current preoccupation with the range of problems to which
we have addressed ourselves. One writer, in attempting to explain
economic growth in the United States, has taken into account not only
changes in employment and hours of work, but also a number of aspects
of changes in labour quality: education, increased experience and better
utilisation of women workers, and changes in the age-sex composition
of the labour force.1 He estimates that education has made a major
contribution to economic growth in the United States; from 1929 to
1957 it accounted for 23 per cent, of the growth of national income.
Improvement in the quality of women workers yielded an additional 4 per
cent, of the over-all growth, by his calculation. His estimates are
rough, and there are a number of questionable assumptions, but of the
pioneering character of the work and of its great importance there can
be no doubt.2
There is a still more recent attempt to determine the relationship
between various levels of economic development and human resource
development on the basis of international comparisons. In a recent
study3, levels of national income were correlated with such indicators
of educational level, and presumably labour quality, as teachers, scientists, engineers, and physicians, as a proportion of the population;
primary, secondary, and higher school enrolment ratios; and the specialisation of students enrolled in higher educational institutions. Some
interesting statistical relationships emerge, though these do not by
themselves provide a basis for distinguishing cause and effect.
During the years in which the empirical studies described above were
being prepared, a considerable evolution took place in the economic
1

Edward F. DENISON: The Sources of Economic Growth in the United States
and the Alternatives before Us, Supplementary Paper No. 13 (New York, Committee
for Economic Development, 1962). He also takes into account changes in the
structure of capital and the effect of such factors as the economies of scale, the
advance of knowledge, and reduced waste in agriculture.
2
For an excellent appraisal of Denison's work see Moses ABRAMOVITZ : " Economic
Growth in the United States ", in American Economic Review (Menasha, Wis.,
American Economic Association), Vol. LII, No. 4, Sep. 1962, p. 762.
3
Frederick HARBISON and Charles A. MYERS: Education, Manpower and Economic
Growth: Strategies of Human Resource Development (New York, McGraw-Hill,
1964), pp. 23 if.

THE METHOD OF THE PRESENT STUDY

theory of the relationship between the primary factors of production
and output. It is not our intention to describe the whole of the debate
that has surrounded this question, but it is essential that some of the
main points at issue should be discussed in order that our empirical
results can be interpreted in an appropriate context.
Ten years ago economists interested in measuring the relative contributions of labour and capital goods to economic growth had no
hesitation in postulating the existence of an aggregate production function according to which the rate of output, Y, is related to the rates of
input of the primary factors, labour and capital. Thus, denoting these
rates of input by L and K respectively, some production function, F, is
assumed to exist such that:
Y = F(L, K)

[2.2]

Further, assuming that this function is subject to constant returns to
scale, i.e. that a given percentage increase in both primary factors
would result in the same percentage increase in output, the rate of output,
Y, is a linear function of the rates of input, L and K, in which the coefficients of these variables are their respective marginal productivities.
Thus:
dY
8Y
Y
L
K
M

-ÏL +m

ÔY
dY
where — is the marginal productivity of labour and — is the marginal
ÔL
dK
productivity of capital. The interpretation of these concepts is that
the marginal productivity of labour shows the extra output that would
accrue from increasing the labour input by one unit if the capital input
remained constant. The marginal productivity of capital may be interpreted in similar fashion.
The need to attempt direct empirical investigation of marginal
productivities can be avoided at the expense of making a further assumption. This is that labour receives the value of its marginal product
or, in other words, that the money wage is equal to the extra revenue
that would accrue from selling, at current prices, the extra output that
an increase of one unit in the labour input would yield. Thus if p
is the price level and w is the money wage, then the assumption is that :
dY
w = p—

[2.4]

Its validity depends on profit maximisation in a context of perfect product and factor markets.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

From equations 2.2 to 2.4 it may be shown that a necessary consequence of the assumptions is that the rate of growth of output is a
weighted average of the rates of growth of the labour and capital inputs.
The weights associated with these variables are the shares of wages and
profits, respectively, in the domestic product.1 Since w is the wage
rate and L is the labour input, the wage bill is given by wL. Further,
since p is the price level and Y is output, the gross domestic product is
wL
p Y. Thus the share of wages in the domestic product is — . Profits
for our purpose are to be understood as that part of the domestic product
which is not received as wages. So, denoting profits by n, we have:
n = PY-wL

[2.5]

and the share of profits in the domestic product is — . If A Y is the
AY
PY
change in output, then — is the rate of growth of output. Thus the
result that follows from the assumptions made above is that:
AY=(wL)AL
Y
(pY) L

VVAK
(pY) K

i.e. that the rate of growth of output is equal to the share of wages
multiplied by the rate of growth of the labour input, plus the share of
profits multiplied by the rate of growth of capital.
It has often been observed that historical data show that the shares
of wages and profits in the domestic product change little over time.
If the share of wages is in fact a constant, a say, then the share of profits
is one minus <x and equation 2.6 can be written as:
AY
_ =

AL .
_+(l_

a )

AK
_

[2.71

This equation is well known to economists as a derivative of the celebrated
" Cobb-Douglas " production function. This function is a particular
version of our equation 2.2 in which it is assumed that:
Y^AUK1-*
1

[2.81

With very little loss of generality, equation 2.2 can be written as

ÔY
ÔY
AY=—AL+—AK
dL
dK
Combining this relationship with equations 2.3 and 2.4 yields the result given as
equation 2.6.

THE METHOD OF THE PRESENT STUDY

where A is a constant. If we had started with this equation instead of
the more general equation 2.2, then by following the same process by
which equation 2.6 was reached, we would have finished up with equation 2.7 instead. The fact that historical evidence indicates that equation 2.7 is a reasonable approximation to equation 2.6 suggests that it is
appropriate to start with equation 2.8 rather than with equation 2.2.
This view is strengthened by empirical results collected in pioneering
studies by Cobb and Douglas and their associates. Using their equation 2.8 they estimated the coefficient a for a variety of different sets
of data and found in each case that the estimate was very close to the
observed share of wages. This is not in the least surprising when one
compares equations 2.6 and 2.7. If the Cobb-Douglas equation is valid
and the wage is equal to the value of the marginal product of labour,
then the share of labour in domestic product will be constant over time
and equation 2.7 will hold good if a is set equal to this share.
The observation that these implications are borne out in practice
does not establish the validity of the assumptions upon which they are
based, however. More recent work using the Cobb-Douglas formulation suggests that the assumption of constant returns to scale is suspect.
If, instead of assuming that the coefficient of capital in equation 2.8
is 1 —a, this coefficient is called ß and allowed to be independent of a,
then the estimates of a and ß that are obtained add up to more than one,
i.e. there are increasing returns to scale. One of the consequences of
conceding that increasing returns to scale may prevail is that equation 2.4
is no longer valid. The real wage cannot be claimed to be in a one-to-one
relationship with the marginal product of labour.
The phenomenon of increasing returns to scale can be accommodated
in two different ways in the Cobb-Douglas equation. The first has
already been described above. This method, which allows the exponents
of labour and capital to sum to more than unity, implies that there are
increasing returns to scale at a moment in time and that the production
function does not change over time. An alternative procedure, which
has been favoured in two important papers \ is to postulate that there
are constant returns to scale at each moment in time, but that the production function shifts over time in one or other of two different ways.
Allowing for shifts in the production function over time is synonymous in the literature with allowing for technical progress. The reason
1
See R. M. SOLOW: " Technical Change and the Aggregate Production Function ",
in Review of Economics and Statistics (Cambridge, Mass. Harvard University), Vol. 39,
1957, pp. 312-320, and "Investment and Technical Progress", in K. J. ARROW,
S. KARLIN and P. SUPPES (editors): Mathematical Methods in the Social Sciences,
1959 (Stanford University Press, 1960), pp. 89-104.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

the function shifts is that new ways of using labour and capital are
discovered.
In the context of the equation 2.8, the first method of allowing for
technical progress is to postulate that the coefficient A in equation 2.8
is not a constant but an increasing function of time. Thus a given
increase in either labour or capital will result in a greater increase in
output the later the date at which the increase in the factor input takes
place. Such shifts in the function are known either as " neutral technical
progress " or as " disembodied technical progress " : " neutral " because
they shift the whole function symmetrically, " disembodied " because
they are not conditional on capital formation having taken place.
This type of technical progress was assumed by Solow in the first
of the two papers referred to above. It is equivalent to assuming that
the units in which labour should be measured change over time because
the quality of labour changes over time. Putting it the other way round,
if labour is measured over time as numbers of men and constant returns
to scale are assumed, then if a in equation 2.8 is set equal to the share of
labour in domestic product, the growth of output as estimated by
equation 2.8 is less than that observed. The difference is the infamous
" residual " factor in economic growth. In the paper of Solow referred
to above, this residual is attributable to changes in the quality of labour.
In the second of these Solow papers, the residual is interpreted as
being due to changes in the quality of capital goods. Thus embodied,
rather than disembodied, technical progress is postulated: technical
progress takes the form of making available more efficient machines at
the same real cost (whatever that may mean) as the cost of those that
were available previously. The production function shifts asymmetrically over time and technical progress is non-neutral. Mathematically
there is no difference between accepting Solow's improvement factor
for the quality of capital and assuming that the price index he uses for
deflating each year's investment grows too quickly.
Unless one assumes constant returns to scale at a moment in time
and adopts a measure of capital which involves cumulating investment
expenditures measured at constant prices in some way, the problem of
the residual factor does not arise. These assumptions are necessary
if one wants to be able to say that labour receives the value of its marginal
product, and to persist with measures of capital of the type specified.
But they have no other virtue. In particular, they are not necessary
either for explaining the past or for designing policies for the future.
Indeed, in this latter context they can be grossly misleading.

CHAPTER III
THE MODEL
Economists have been increasingly attracted to what are known as
" vintage " models of economic growth. The model we use in our study
is of this type.
In vintage models the factors of production, labour and capital,
can be made complements rather than substitutes. At a given moment
in time a range of alternative techniques is assumed to exist from which
a choice must be made for current investment. Each technique is
characterised by the output it produces, its labour requirements and
the cost of the capital goods associated with it. Once a technique has
been chosen it cannot be altered: its output and labour requirements
do not change with its age. Its use is assumed to continue until such
time as it ceases to earn a profit. When this happens the capital goods
associated with the technique are scrapped and the labour that was
employed to work with these goods is released for employment in a
new plant.
It follows from this statement of the vintage model that the change
in capacity output in any one year, A Y, is equal to the output of plant
newly installed, X, minus the output scrapped, Xs. Thus:
AY=X-X'

[3.1]

Similarly, the change in capacity employment, AL, has two components :
the increase in employment resulting from the new jobs made available
by new plant installations, N, and the decline in employment resulting
from the loss of jobs caused by the scrapping of old plant, Ns. Hence:
AL=N-NS

[3.2]

It follows directly from equations 3.1 and 3.2 that:
pAY-wAL=(pX-wN)-(pXs-wNs)

[3.3]

The left-hand side of equation 3.3 is that part of the change in profits
which is independent of changes in the prices of labour and the product.
On the right-hand side of equation 3.3 there are two expressions in
brackets. Both are of interest. We consider the second one first.
Since Xs is the output that would have been produced by plant that is

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

in fact scrapped, pX* is the revenue that such plant could have earned.
Similarly, wNs is the total amount of wages that would have had to be
paid in such plant. Hence pX"—wNs is the profit that would have been
earned by plant that was in fact scrapped. So, if plant is scrapped
when it ceases to earn profit, this profit, pXs—wNs, is zero and equation
3.3 reduces to:
pA Y-wáL= pX- wN
[3.4]
The expression on the right-hand side of equation 3.4 has two components. The first is this year's revenue of plants installed this year.
Its second component is this year's wage bill of plants installed this
year. The right-hand side of 3.4 is therefore this year's profit of plants
installed this year or, in other words, the first year's profit of newly
installed plant.
The cost of the capital goods installed this year is simply this year's
investment measured in current prices. We denote this variable by /.
Since pX—wN is this year's profit on this year's investment and / is
the cost of this year's investment, the variable r defined as :
r =

PX WN

13.5)

J
is the rate of return this year on this year's investment. Hence, the
variable r may be referred to as the immediate rate of profit on investment
expenditure.
It follows from equation 3.5 that we can substitute ri for the righthand side of equation 3.4 to give:
pAY-wAL=rI

[3.6]

or, by dividing through this relationship by p Y and rearranging terms :
AY
(wL)AL
I
— =
\- r—
Y
(pY) L
pY

[3.7J

This equation is fundamental to our model. According to it the
rate of growth of output, —çr, has two components, one depending
on labour and the other on capital.

The labour component is familiar

from equation 2.6.1 It is the rate of growth of employment, — ,
Li

weighted by the proportion of the domestic product which is received
as wages. The contribution of capital to growth given by equation 3.7
1

See p. 36.

THE MODEL

is different from that shown in equation 2.6. According to equation 3.7,
the contribution of capital to growth is the product of two terms. The
first is the immediate rate of profit on current investment, r, which is
defined in equation 3.5. The second is the investment ratio: that is, the
proportion of the domestic product which is spent on capital goods.
According to the aggregate neo-classical production function studies
referred to in the previous chapter, the contribution of capital to changes
in output is given by:
ÔY
AY—-AL
[3.81
oh
Further, if we multiply this expression by p and make the assumption
w
that the marginal product of labour is equal to the real wage, —, the
expression 3.8 becomes:
pAY-wAL
If this expression is set equal to -j-, where I is investment expenditure
measured in current prices, we have:
pAY-wAL=—

[3.9]
k

It follows that k is the ratio of investment to the change in output attributable to changes in capital, both measured in current prices. Thus k
is the alternative ICOR(L) referred to in the previous chapter. It is
the variable given by the slope of the lines drawn from the origin through
the mid-points of the lines on figure 4 and to be read off on the outer
right-hand and upper scales of the chart. But as can be seen from a
comparison of equations 3.6 and 3.9, the variable k is open to another
interpretation in the context of vintage models. This comparison shows
that:
1
k = [3.101
r
i.e. that the alternative ICOR(L) is the reciprocal of the immediate
rate of profit.
Since the immediate rate of profit is the ratio of the first year's
profit to the cost of plant, it is equal to the proportion of the cost of the
plant that is recouped in its first year of operation. Thus its reciprocal
is the number of years that the plant would have to operate in order
that its total capital cost be recouped, without any allowance for time
discounts or changes in the price level and money wages. In this simple
sense, it is the pay-off period on current investment. And so we see

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

that, when we move from the aggregate production function approach
to the vintage model, k, the alternative ICOR(L) becomes the pay-off
period on current investment as defined above, and its reciprocal is the
immediate rate of profit which this investment earns.
From this analysis it follows that the outer right-hand and upper
scales of figure 4 have the dimension " years " and refer to the pay-off
period. The inner scale is its reciprocal and hence observations on it
refer to immediate rates of profit.
As shown by figure 4, the country experiencing the longest pay-off
period was Argentina. That with the shortest was the Philippines,
closely followed by the Federation of Malaya. In general, there was a
tendency for the more developed economies to have the longer pay-off
periods or smaller immediate profit rates. The countries with the shorter
pay-off periods or higher immediate profit rates are among the least
developed in our sample.
Why do immediate profit rates, or pay-off periods, differ between
countries ? One of the main factors, in our view, is the different evolution of the quality of their labour forces. This point is taken up later.
Here some of the alternative explanations are briefly considered.
An obvious reason why average pay-off periods vary among countries is that the infrastructures of the latter may be different. A country
in which the length of life of capital goods is above average may be
expected to have an above-average pay-off period. This, it might be
argued, is the reason why Norway, which must maintain transportation
and communication facilities over large and sparsely populated areas,
has such a low immediate profit rate. But there is some evidence to
show that this is not a complete explanation. The evidence is contained
in the report of the Economic Commission for Europe discussed in
Chapter I and indicates that, taking the economy sector by sector, the
alternative ICOR(L) (and hence our pay-off period) is typically greater
for Norway than for other countries in their sample. However, while
infrastructure may not present a complete explanation, it is nonetheless
true that the evidence collected by the E.C.E. shows that it should not be
ignored.
Other factors which might tend to generate differences in immediate
profit rates between countries can best be discussed in terms of a mathematical result the proof of which, by virtue of its complexity, has been
relegated to Appendix I. The result is as follows. Let A be the average
rate of return which it is expected will be yielded by investment in some
w
particular technique. Further, let it be assumed that real wages, — ,
P
are going to grow at a constant rate, co, over time. Let X * be the maxi-

THE MODEL

mum value of X which is obtainable given currently available techniques
and prices, and let r * be the immediate rate of profit yielded by the
technique for which A is a maximum. The result that follows from these
assumptions is that:
r* = X* + coe

[3.11]

where e is the percentage increase in investment expenditure required
to offset a fall of 1 per cent, in the labour force allocated to new plant.
From this equation, a number of alternative explanations of differences
in immediate profit rates, r, are apparent. The first, which should not
be overlooked, is that countries might differ in the extent to which they
try to maximise X, and those that do try may differ in the extent to which
they are successful. Given these qualifications, differences in immediate
profit rates between countries can be explained in terms of differences
in X *, co and e.
The variable X * introduced in equation 3.11 is the average rate of
return expected on current investment in an economy. We may anticipate, therefore, that it will be at least equal to the rate of interest, and
consequently may well differ between countries because of peculiarities
in their capital markets, government policy, and the like. In particular,
high risk premiums will be expected in countries which are economically
and politically unstable.
Differences in the expected rate of growth of real wages, co, may have
their origin in either psychological or institutional behaviour, but it
would be surprising to find that such expectations differed considerably
from recent past experience. In countries operating an active wages
policy, the annual settlement could well have a strong influence on the
variable co.
The third term, e, in equation 3.11 is determined by the shape of
the technology opportunity frontier at a moment in time. This frontier
shows the various combinations of labour and capital which can be
employed as between the alternative techniques which are currently
available. Thus the number e specifies the percentage increase in
capital cost involved in changing from that technique which has the
highest expected rate of return to one in which labour productivity
is 1 per cent, greater. It is important to note that e is a number and,
therefore, unit free. In a country which imports all its plant, a change
in the terms of trade would not affect e except in so far as there was an
associated shift in the technique chosen. Similarly, and more important,
e can be the same for two countries even though labour is measured
in different units for each. In general, a high value of e indicates that

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

plants with a higher level of productivity than those chosen have substantially greater capital-output ratios: e is the elasticity as between
techniques of the capital-output ratio with respect to the average product
of labour.
We have already indicated above that there is considerable doubt
as to whether or not countries choose investment projects in order to
maximise the expected average rate of return on their capital expenditure.
The question may also reasonably be asked whether or not they should
try to do this. 1 Clearly some form of optimalising procedure should
be followed, and an alternative which readily suggests itself is that the
technique chosen should be that which maximises the immediate rate of
profit.2 If this criterion is followed, then the pay-off period is a minimum
and hence, subject to differential movements in the price level and the
wage rate, the cost of investment is recouped as profit at the earliest
possible date.
If the immediate rate of profit is maximised, then as shown in
theorem II of Appendix I, this maximum immediate rate of profit is
the marginal revenue product of expenditure on capital goods, i.e. the
value of the additional output ensuing from an addition of one unit
to investment expenditure.
The effects of changes in the quality of labour can be derived from
equation 3.7, which states that:

AY_(wL)AL

I
+r

T ~ {pY)T ~¡¡Y
The variable L which appears in this equation is the labour input into
the production process. The role of the quality of labour is essentially
to specify the units in which this variable should be measured. Nothing
has been said in this chapter so far which requires that labour should
be measured in numbers of workers. What we have said is that the
labour input into plant of any particular vintage is constant over time.
If we concede that there is a difference between the input of productive
labour services and the number of men working, then it follows that the
assumption of constant labour inputs in particular plants does not
1
For a discussion of different choice criteria see A. K. SEN: Choice of Techniques:
an Aspect of the Theory of Planned Economic Development (Oxford, Blackwell, 1962).
2
This is the criterion proposed by Galenson and Leibenstein in the context of
an aggregate production function of the type discussed in Chapter II. Operationally,
however, the criterion is the same as maximising the immediate rate of profit in the

context of vintage models.

See W. GALENSON and H. LEIBENSTEIN: " Investment

Criteria, Productivity and Economic Development ", in Quarterly Journal of Economics (Cambridge, Mass., Harvard University), Vol. LXIX, No. 3, Aug. 1955.

THE MODEL

imply that the term — in equation 3.7 is the same as the rate of growth
of the number of persons in the labour force. The former can, in fact,
exceed the latter by the rate of growth of the quality of the labour force.
Thus suppose that Q is an index of the quality of labour such that a
1 per cent, increase in Q implies that all plants can now manage with
1 per cent, fewer men. The labour variable L in equation 3.7 can now
be replaced by the product of two variables, one measuring the quality
of workers in the sense defined above and the other measuring the number.
The rate of growth of the labour input is now the rate of growth of the
quality of labour, —^, plus the rate of growth of the number of men.
AL
This latter can be represented symbolically as — , provided that it is
now recognised that L refers to numbers of persons and not to the
labour input. Hence, equation 3.7 can be rewritten as:
AY^i^L&VALAQl
Y
(pY)lL

or, by rearrangement of terms, as :
AY (wLQ)AL
~Y~ (pY) T
(wLQ)
(pY)

(I) AQ
\wLQ) Q

I 3 ' 13 '

The expression wLQ which appears in equations 3.12 and 3.13 is the
wage bill. Consequently, if the symbol w is redefined so that now it is
the wage per worker rather than the wage per unit of labour1, these two
equations can be simplified to:
AY

(wL)\~AL AQl

—
Y = —^\
(pY)lL—+~ Q] +r—pY

[3.14]

1
The following numerical example might clarify these manipulations.
At a moment in time there are 100 employees of quality 1 who receive a wage
of 10 units each. At a later moment in time Aere are only 95 men employed, each
of quality 1.07 and each receiving a wage of 11 units. Thus, the number of men
has decreased by 5 per cent., but the labour input has increased by 2 per cent., because
the quality of men has increased by 7 per cent. The wage bill has increased from
1,000 to 1,045, i.e. by 4.5 per cent. Thus the wage per unit of labour input has
increased by 2.5 per cent., i.e. by 4.5 per cent, minus 2 per cent.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

and:

AY (wL)AL
Y (pY) L
(wL)

r «

(wL)

+—
Ö

[3.15]

(pY)
The left-hand side of equation 3.15 is the ratio of two terms. In the
numerator we have that part of the rate of growth of output which is not
attributable to growth in the labour force measured in numbers of
workers. The expression in the denominator is the proportion of
domestic product received as wages. This variable, which for convenience we shall refer to as Z 0 , is the variable that our statistical exercises
are designed to explain. As can be seen from equation 3.15, Z 0 is the
sum of two terms. One is the rate of growth of the quality of labour,
—^.

The second is the product of the immediate rate of profit on

current investment and the ratio of investment expenditure to the wage
bill. The latter ratio is given the symbolic representation Z t . Hence,
equation 3.15 becomes:
AQ
Z0 = rZ1 + -^

[3.16]

This is the first version of our model. The second is derived from it
by substituting for r the expression given in equation 3.11. According
to this equation:
/•* = X* + cos
and, therefore, if we assume that r is equal to /•*, we have:
Z 0 = 1*Z1+coeZ1 + —

13.171

Z 0 = A*Z 1 +6Z 2 + ^

[3.18]

or:

where the variable Z 2 is defined by:
Z 2 = coZ1

[3.19]

The equation 3.18 is the second version of our model. It is derived
from the first by assuming that r = r*, i.e. that techniques are chosen
in such a way that the expected average rate of return on investment is a
maximum.

THE MODEL

The treatment of the variable Q in our model is simplistic. This is a
preliminary study and, as discussed in the next chapter, the various
indicators of the quality of labour which were available did not merit
more sophisticated treatment. What we have done is to assume that
the rate of growth of the quality of labour is simply a linear combination
of the rates of growth of various indices. Thus, if we have a set of
indices Qu Q2,
, Q„, the assumption is that:
AQ
AQ,
AQ2
AQ„
— = a,
Ha,
+ ...a„
ß
'Öl
Ql
On

J3.20J

The model is completed by substituting the expression on the righthand side of equation 3.20 for the variable — in either of the two
versions given in equations 3.16 and 3.17. The exact specification of
the indicators of the quality of labour, Qu Q2,
Q„, which we
have used and the combinations in which they have been analysed are
discussed in the next two chapters.
SUMMARY OF THE ARGUMENTS

It may be helpful at this stage to summarise the arguments of this
chapter and to compare the two versions of our model with one another
and with the aggregate production function studies discussed in
Chapter II.
The most important relationship obtained in this chapter is that
shown as equation 3.7, i.e.

AY_(wL)AL

According to this relationship the rate of growth of capacity output,
-TTJT , can be split into two parts, one depending on the rate of growth
of the capacity labour force, — , the other on the proportion of the
domestic product which is currently being spent on capital goods, — .
The derivation of the relationship follows directly from the assumptions
made in the first part of this chapter which define the type of vintage
model being considered. In particular, it is important to be clear that
the coefficient, r, of the investment ratio in equation 3.7 is the immediate
rate of profit on current investment irrespective of how investment pro-

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

jects are selected. The only form of profit maximisation involved in the
derivation of equation 3.7 is that plants must be scrapped when they
cease to earn a profit. Thus the relationship 3.7 is consistent with,
and independent of, whether current investment decisions are in fact
based on the maximisation of absolute initial profits, initial or average
profit rates, or even the foibles of an irrational entrepreneurial élite.
Within the assumptions made, 3.7 is simply an accounting identity.
If r is the rate of profit earned in the first year by new plant, then it is
again a matter of accountancy that its reciprocal is the number of years
for which the plant must operate if its initial cost is to be recouped,
assuming that its profit earnings remain constant and without allowance
for interest charges. With the pay-off period defined in this way, to
say that the pay-off period is the reciprocal of the immediate rate of
profit is tautological. Nor does the immediate rate of profit need to
be maximised or the pay-off period minimised (which is the same thing
anyway) for this reciprocal relationship to hold.
Because it is true that no theory about how investment decisions
are made is involved in the derivation of equation 3.7, we are at liberty
to consider how the interpretation of the equation, and in particular of
r, would vary as between different theories. The only restraint is that
the theories must be compatible with the vintage model assumptions
made in deriving 3.7.
There are two obvious candidates for the role of investment decision
theory within our vintage model. They are referred to in this study
and developed formally in Appendix I, the mathematical appendix.
The first is that investment projects are chosen so that, given the capacity
which the new investment must generate, the average rate of return earned
by the investment over its lifetime is a maximum. This lifetime is the
length of time from now to the date when the use of the plant ceases to
be profitable. This must obviously depend on how the real wage is
expected to move, and it has been assumed here for simplicity that the
real wage is expected to grow at some constant rate, a>, indefinitely into
the future. The lifetime of a plant will also depend on how labourintensive it is. Profits will vanish most quickly in those plants in which
the average product of labour is lowest. However, such plants will
cost less than plants with higher labour productivity and therefore
longer expected lifetimes. Thus there is a genuine decision problem.
If it is resolved by selecting that plant for which the average rate of return
is a maximum, then equation 3.11 is the result. This equation states
that:
r* = A* + cos

THE MODEL

where r* and A* are respectively the immediate and the average rates
of return in the plant thus selected, co is the constant rate at which real
wages are expected to grow in future, and e is the percentage increase
(decrease) in investment expenditure that would be involved in obtaining
the required new capacity by buying plant in which the average product
of labour was 1 per cent, more (less) than it in fact is in the plant which
is in fact selected. It is important to be clear that equation 3.11 exists
only when it is assumed that the maximisation of A governs the choice of
investment projects. Assuming that the vintage model yielding the
relationship 3.7 is true, the difference between immediate and average
rates of return, (r—A), is given by toa only if it is further assumed that
A is maximised, i.e. only if A is A* and r is the immediate rate of profit,
r*, which results.
The second obvious contender for the role of investment decision
theory within the context of our vintage model is that the choice between
alternative investment projects yielding the same increment to total
capacity is made in favour ofthat project which has the highest immediate
rate of profit or, which is the same thing, the shortest pay-off period.
As explained in the text and proved in theorem II of Appendix I, if this
alternative type of behaviour governs investment decisions, then the
immediate rate of profit is the marginal revenue product of investment
expenditure, i.e. the value of the addition to the domestic product that
would have resulted if investment expenditure had been slightly larger
than it in fact was per unit of additional investment expenditure.
If we leave aside the question of whether labour is measured in men
or quality units, the correct interpretation of the two versions of our
model can be seen. The first version follows directly from equation 3.7.
It depends, therefore, only on the assumptions of the vintage model;
it implies nothing about whether or not anything is maximised: it is
an accountancy model.
The second version of the model implies that the first version is
correct, and more. It implies that 3.7, and therefore the vintage assumptions, are valid. It further implies that the observed value of r is r*,
i.e. that those investment projects are chosen which have the highest
expected average rates of return, A*. Thus, equations 3.7 and 3.11
combine to give the second version of the model.
It would have been possible to have a third version of the model.
This would imply that the first version was correct and that pay-off
periods were minimised in investment decisions. Thus, if /•** is the
maximised immediate rate of profit, then in this version of the model
the equation 3.7 would be combined with the equation:

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

It follows from all this that the first and second versions of the model
and the first and third versions are consistent. The second and third
versions cannot be consistent. The third version of our model has not
been pursued here because it adds nothing, apart from two asterisks,
to the form of the first version of the model. Thus, if we had wanted
to estimate the third version of the model, our procedure would have
been the same as in estimating the first version. The only difference
is one of interpretation. On the one hand, we can state from the first
version, " Here is an estimate of the immediate rate of profit on new
investment ". On the other hand, the third version permits statements
of the form, " Here is an estimate of the immediate rate of profit on
new investment which, because of the way in which investment projects
are chosen, is also an estimate of the marginal revenue product per unit
of additional investment expenditure " ; but this expanded statement
can be justified only in terms of the validity of the assumptions by which
the third version of the model is obtained from the first.
So far we have not ventured outside the realms in which the vintage
model assumptions underlying the relationship 3.7 are valid. However
the variables involved in equation 3.7 other than the variable r are very
familiar in economics, and it is not surprising therefore that, on the
basis of assumptions entirely different from our vintage model assumptions, conclusions can be reached which suggest that they are related
to each other in a way similar to that reached using the vintage model
assumptions in 3.7.
Forget all about vintage models. Assume that there is an aggregate
production function as in equation 2.1. Assume further that this is
differentiable and subject to constant returns to scale (as in equation 2.2) and that the real wage is equal to the marginal product of
labour (as in equation 2.3).1 Hence, one arrives at equation 2.6, i.e. :
A Y _ (wL) AL

(J7) AK

The second term on the right-hand side is the contribution of capital
to growth. Its magnitude is:

ÍI_Wií
Y

See p. 24 and pp. 35-36.

(plO L

,3. 2 1 |

THE MODEL

i.e. the same as the magnitude of the contribution of capital to growth
nV

given by equation 3.7. Now multiply this magnitude by — to obtain:
pAY-wAL
-

[3.221

In terms of the vintage model, this magnitude is r, the immediate rate
of profit. Its reciprocal is, therefore, the pay-off period. In terms of
the aggregate production function model favoured by the E.C.E., the
reciprocal of the magnitude 3.22 is their alternative ICOR(L).1 In
terms of the Galenson-Leibenstein formulation already referred to 2,
the magnitude 3.22 is the thing planners should try to maximise. In
each case the magnitude is the same, but in each case also the method
by which the magnitude was obtained and the interpretation to be
put on it are entirely different.

1
s

See footnote 2, p. 32.
See footnote 2, p. 44.

CHAPTER IV
INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY
GENERAL CONSIDERATIONS

The first problem in approaching a choice of indicators was to determine which were the most suitable for an exploration of the factors
contributing to economic growth. A great many of these have been
suggested from time to time, ranging from capital investment to various
aspects of political and social structure. Since the methodology which
we employ is quantitative in nature, indicators which could not be
quantified had to be ruled out. Such institutional factors as family and
caste systems, forms of government and traditional work habits, which
impinge directly on developmental possibilities, could not be formally
taken into account in our model.
Secondly, a choice had to be made among those indicators which
could be expressed numerically. And, since our interest lay mainly
in the factors influencing the quality of labour, indicators of phenomena
not directly germane to this problem were not considered.
In the process of narrowing the scope of our inquiry, we came to
a consideration which in the last analysis dictated the precise indicators
to be employed : the availability of data. Here, there were three major
criteria which determined the final solution.
(a) The data had to be comparable internationally. This meant,
in effect, that we were restricted to statistics appearing in the yearbooks of the various international agencies. Over the last decade,
these agencies have done yeoman work in collecting and collating
statistics from their member nations and in encouraging the nations to
present their statistics in standard form. The process has by no means
been completed, but the year-books already contain a great quantity
of invaluable information. It would be foolhardy to attempt to go
back to the statistical annuals of individual countries, unless the investigator had enormous resources and time at his disposal, which was
not our case.
(b) The data had to cover a sufficiently large sample of countries
to make the exercise interesting. One quickly discovers, in dealing

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

with international statistics, that the quality and quantity of the available
data decline rapidly as one leaves the highly developed countries. Even
for the latter there are surprising gaps. For example, national income
statistics in constant prices for Australia and New Zealand are not
available from the Yearbook of National Accounts Statistics.1 Only
by some fairly desperate expedients were we able finally to secure a
sample of 52 countries, and even for these not all the relevant indicators
are available.
(c) The data had to cover a sufficiently long time period for purposes
of analysis. It is extremely difficult, for all but a handful of countries, to
go back beyond the year 1950, and so it was decided to make this the
initial year whenever possible. Since we are concerned with rates of
growth, it was desirable to have data for a period long enough to minimise the impact of unusual events. Moreover, effects of the kind with
which we deal do not normally make themselves felt within one or two
years. Indeed, it might be argued that some of the labour quality
factors—primary education, for example—require considerably more
than the decade to which we are limited for their full impact to be felt.
But at the present juncture a decade of relevant statistics is all that is
available, and this situation will improve only with the passage of time.
The result of this process of selection is the series of indicators
described below. It is obvious that they are not ideal. Many of them
are at best only imperfect indicators of the phenomena which they
purport to describe; for example, ratios of physicians and hospital
beds to population were assumed to measure the relative level of health
attained by a country.2 More refined measures are clearly in order,
for health and other factors as well, but without a great deal of further
processing of new data they are not available at the present time.
The question may legitimately be raised whether data which are
selected in this manner and are subject to so many drawbacks can be
expected to yield useful results. Even if statistical relationships do
appear, can any meaning be attributed to them ? In the final analysis,
each investigator must fall back on his own judgment in answering this
question. Having lived with the data for some time, we have come to
feel that they are meaningful in relation to real phenomena, with all
their imperfections. In a sense, similar strictures can be levied against
any aggregate data, and the purist has simply to eschew work in macro1

United Nations: Yearbook of National Accounts Statistics (New York).
It might be argued that beyond a certain point in development, physicians and
hospital beds per caput would begin to decline. However, the data give no evidence
of this, and it seems to be true that the demand for medical services increases with
rising income, at least to the level of the most developed nations today.
2

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

economics. This is not to say that we are happy either with the data
or with the results. We are strongly of the belief that more and better
statistics will yield firmer conclusions, and in the final section of this
report we record some observations on desirable next steps. On the
other hand, we do not consider the present study an exercise in abstract
methodology but rather a preliminary essay into substantive matters.
Our sample of countries was unfortunately restricted by the necessity
of omitting communist nations with centrally planned economies. Here
we were guided by the growth study of the Economic Commission for
Europe, which dealt separately with market and with centrally planned
economies. The Commission pointed out that it is difficult to compare
the two because of differences in the definition of some of the major
economic variables—national product, for example—as well as differences in the systems of valuing output.1 Since we were not in a position to conduct two separate studies, which would have meant developing two models varying considerably in concept, we have confined
ourselves to the market economies. This is not at all to say, however,
that we regard the experience of the communist nations as irrelevant to
the problem at hand. We would hope that, either through the reconciliation of the statistics or through an independent inquiry, the lessons of
their recent development could be made available to the developing
countries.
T H E INDICATORS

Economic

Growth

We selected the rate of increase of the gross domestic product at
constant prices, valued at factor cost, as the basic index of economic
growth.2 This rate of increase was calculated as the trend rate of growth
indicated by the difference in output between the initial and the terminal
years, which in most cases were 1950 and 1960, rather than by averaging
the annual rates. For 21 of the 52 countries, G.D.P. data were not
available for the full decade, and in such cases the widest spread of years
available was used. In all but ten cases this involved the loss of a single
year at the beginning or end of the decade.
1
The centrally planned economies use the concept of net material product in
their national accounting. This differs from the western national product measure
in that all services are omitted. It cannot be assumed that net material product
and gross national product move in parallel fashion. Until these two concepts
are reconciled, it will be extremely difficult to compare the two types of economies
on a global basis.
2
The E.C.E. study referred to above concluded that the rate of growth of G.D.P.
at factor cost did not differ significantly from that at market prices for the countries
of Western Europe. Inspection of the data for non-European countries indicates
that this is generally true.

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

The data were obtained mainly from the Yearbook of National
Accounts Statistics. For all but a few of the countries in our sample,
G.D.P. is given there in constant prices. In these few cases, independent
deflation was accomplished by the use of alternative price indices, where
this seemed feasible.
In our model, —— is defined as the rate of growth of capacity output.
To the extent that in any year the G.D.P. is less than capacity by reason
of unemployment of men or machines, our measure fails to constitute
an appropriate indicator. But as the Economic Commission for Europe
has pointed out, the measurement of capacity output is a complicated
matter.
Unfortunately there is no simple means of statistical measurement either
of " capacity " G.D.P. or of the degree of utilisation of capacity. The most
logically satisfactory concept of capacity G.D.P. in any year is that volume
of production which would result from optimal use of available labour and
capital equipment, taking into account the limited short-term possibilities
both of factor mobility and substitution and of increasing the efficiency of
use of domestic resources through international trade. " Full capacity
production " in this sense may be compatible with under-employment of some
labour or capital equipment.. .any attempt to adjust data on actual growth
of domestic product to allow for the effects of under-employment of resources
. . . would imply rather important assumptions about possibilities of substitution of production factors and about the effects of variations in economic
activity on the efficiency of use of labour and capital employed.1
For most of the countries in our sample, it was impossible to determine the extent to which there were variations in the degree of labour
utilisation, let alone capital utilisation. The only correction we attempted
was by scanning the plotted G.D.P. growth rates, and where either
the beginning or the end seemed manifestly off-trend, a shift in terminal
year was made in the determination of the final rate employed. This
proved necessary in only two or three cases. This does not constitute
any guarantee, however, that part of the growth, or lack of growth,
indicated for individual countries was not due to changes in capacity
utilisation between the terminal years, either in the form of increasing
or decreasing unemployment, or increasing or decreasing machine or
land utilisation. In looking at the record of any particular country,
this fact should be kept in mind.
The basic G.D.P. statistics are shown in Appendix II, table 1. In
some cases where G.D.P. at factor cost was not given in the Yearbook,
we were obliged to use some other measure: G.D.P. at market prices,
gross national product, or net domestic product. These are unlikely,
1

Economic Survey of Europe in 1961, Part 2, op. cit., Ch. II, p. 5.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

however, to have affected the growth rates by more than a fraction of a
percentage point. These and other departures from the general rule
are indicated in the footnotes to the table.
Investment
So much has been written about the investment factor that there is
very little to be added here. In general, we have followed the methodology employed by the Economic Commission for Europe 1 in calculating
the investment ratio. This was done by expressing fixed capital formation, usually at current market prices, as a percentage of gross domestic
product at current factor cost.2 The annual ratios were averaged to
secure the investment ratio for the period as a whole. The few cases
in which other expedients were employed in determining the investment
ratio, for example using G.N.P. rather than G.D.P. as the base, are
indicated in the footnotes to Appendix II, table 3.
Some of the investment ratios seem suspiciously small, in terms of the
resultant capital-output ratios, but it was not possible for us to attempt
to go behind the data appearing in the Yearbook. For about a dozen
countries, the average ratio masks substantial changes in the level of
investment during the period. Nigeria, for example, doubled its investment ratio in seven years, while in Venezuela there was a decline from
31 per cent, to 22 per cent, in a similar period of time. Japan, with an
investment ratio of 18 per cent, in 1950, reached 34 per cent, in 1960,
and Iceland experienced a similar leap from 24 per cent. (1952) to 39
per cent. (I960).3 No correction was made for such cases, although
this might have seemed desirable in view of the fact that steady investment
obviously has different implications for economic growth from an
investment ratio subject to wide fluctuations or substantial trends.
This is a matter that merits further investigation.
There does not appear to be any simple relationship between the
rate of growth of output and the investment ratio. This is apparent
at a glance from figure 2. Some of the highest growth rates were
1

See Economic Survey of Europe in 1961, Part 2, op. cit., Ch. II, pp. 16-23.
The E.C.E. found that " it makes little difference for most of the western countries during the fifties whether gross investment ratios are calculated in current or in
constant prices...". Ibid., Ch. II, p. 17, note 25.
Only to the extent that indirect taxes on investment goods varied greatly from
one country to another—and this is not believed to be the case—would the resultant
ratios tend to be distorted by failure to use the identical price regimen, a choice
dictated by the data.
a
Other countries in which the investment ratio varied by more than a few percentage points during the decade were China (Taiwan), Cyprus, Greece, Jamaica,
Malta, Mauritius and Panama—all in the category of relatively underdeveloped
countries.
2

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

associated with low average investment ratios, e.g. China (Taiwan),
Malaya, the Philippines and Turkey and conversely some high investment
ratios yielded low growth, e.g. Australia, Finland and Norway. In
general, the underdeveloped countries were at the lower end of the
investment spectrum, and the developed countries at the upper end.
The Labour Force
This is perhaps the weakest statistical link in the entire study. The
deficiency lies not only in the availability of data, but in the entire concept of employment, particularly in the underdeveloped nations.
The labour force data shown in Appendix II, table 2 represent the
so-called economically active population. The rate of change used in
our calculations was determined simply by comparing the figures at
the beginning and the end of the period, 1950 and 1960 wherever possible,
and for shorter periods when data for these years were lacking. In all
cases but one, that of Ireland, the labour force measured in this way
increased during the 1950s. However, the varying length of the lines
on figure 4 indicates the lack of uniformity in the rates of growth. All
one can say is that to a large extent this rate was a function of population
change.
The economically active population is defined as " the total of employed persons (including employers, persons working on their own
account, salaried employees and wage earners, and, so far as data are
available, unpaid family workers) and of persons unemployed at the
time of the census ".x There are several problems connected with the
use of this definition to indicate labour inputs.
(a) The unemployed, as well as the employed, are included in the
labour input data. If unemployment increases during the period, the
rate of growth of the working labour force (thus measured) is overstated;
if unemployment declines, it is understated. Where unemployment is
measured through labour force sample surveys, an adequate index of
changes in the level of unemployment may exist. But in most countries
there is nothing between the population census on the one hand and measures which are restricted to industrial wage earners on the other (unemployment insurance recipients, labour exchange registrants, trade
union statistics). Census data may not refer to the years desired for
analytical purposes, while industrial unemployment may be a very poor
indicator of the general level of unemployment in the heavily agrarian economies typical of so many underdeveloped nations.
1

I.L.O.: Year Book of Labour Statistics 1962 (Geneva, 1962), p. 1.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

(b) The concept and substance of employment, unemployment,
and underemployment vary not only between industry and agriculture
but even more between countries at different stages of development.
There are variations in the duration of working hours, in the intensity
of work and in the continuity of work, all of which affect the real labour
input.
(c) Among the groups excluded from the economically active
population are " students, women occupied solely in domestic duties,
retired persons, persons living entirely on their own means, and persons
wholly dependent upon others ".1 Different interpretations of several
of these categories may give rise to considerable variation in the reported
size of the labour force.
Since we are concerned not with international comparisons of the
ratio of economically active population to total population but rather
with the rate of growth of the labour force within each country, it might
be felt that the foregoing factors, with the exception of unemployment,
are irrelevant to our purposes. This feeling is based upon the assumption
that there is stability among those other factors, i.e. that the relative
levels of underemployment, working hours, and so on within each
country do not change. One would not be too bold in making this
assumption for developed nations, or for underdeveloped ones with
a slow growth rate, but it may be far off the mark for countries which
are undergoing rapid economic transformation. Among the events
that could greatly affect real labour inputs without a corresponding
reflection in economically active population data are large migrations
of farm and rural workers to cities, the elimination of petty commerce
and handicrafts, and the inculcation of greater discipline in the labour
force. These are effects which it would be extremely difficult to measure
even if abundant data were available, and impossible under present
circumstances. Yet it must be recognised that increases in labour
productivity which are ascribed to improvements in the quality of the
labour force may stem from structural shifts in the economy and from
the breakdown of traditional attitudes to work, unrelated to the specific
qualitative factors which we attempt to measure.
The labour force estimates themselves were derived from several
different sources, but all are meant to reflect identical definitions of the
economically active population. For 21 countries, mainly in western
Europe, these were available from the E.C.E. study, while for an addi1

Year Book of Labour Statistics 1962, op. cit., p. 1.

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

tional five countries, estimates were derived from a recent United Nations
publication.1 In the remaining 26 cases, estimates were based upon
data appearing in the Demographic Yearbook.2 For 12 of these, the
ratio of economically active population was available for only a single
year during the decade, so that the estimate had to be made by applying
this ratio to total population data, on the assumption that " the proportion of total population economically active is not normally subject
to rapid change ". 3 While this assumption is undoubtedly justified
for most countries, it can be vitiated by rapid changes in population
growth or other influences. In Chile, for example, the ratio fell from
37 per cent, in 1952 to 32 per cent, in 1960, while in Greece it rose from
37 per cent, in 1952 to 44 per cent, in 1961. Of the 12 countries for
which this assumption was used in the estimate of the labour force, nine
had average annual net population growth rates of 2 per cent, or more
during the 1950s, and two of these were over 2.5 per cent.4 There
appeared to be no way of adjusting the data which did not involve further
arbitrary assumptions, so that we considered it advisable to use the data
derived in the manner indicated without correction.
One of the conclusions which quickly emerged in the process of
gathering data was that further comparative analytical work, particularly
if greater refinement is desired, will be difficult to carry out unless labour
force data are increased in quantity and improved in quality. There is
an urgent need, in most of the underdeveloped nations, and in many
developed ones as well, to secure more basic information on existing
labour resources. Such information is an obvious prerequisite to
realistic planning. Labour force censuses, carried out on the basis
of uniform international definitions, would yield information of the
greatest value to governmental authorities, to say nothing of students
of economic development. We are the first to admit that the present
study rests heavily on labour input data of dubious quality; our only
excuse for using them is that there is simply no alternative if one wants to
go beyond the remanipulation of data for a few of the most advanced
nations, a procedure which is not likely to yield results of great interest
to the developing nations.

1
United Nations, Economic Commission for Latin America: Human Resources
of Central America, Panama and Mexico, 1950-1980, in Relaiion to Some Aspects
of Economic Development (1960).
2
United Nations: Demographic Yearbook (New York).
3
Year Book of Labour Statistics 1962, op. cit., p. 1.
4
The annual net rate of population increase for Venezuela was 4.3 per cent.,
which makes the labour force estimate for that country particularly dubious.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

The Wage Share
Our model calls for a labour force weight which is the proportion
of value added paid out in wages. Two alternative estimates of this
magnitude were derived. The first, which we have termed the " minimum " wage share, is the ratio of compensation of employees to the
gross domestic product, both in current prices. The " maximum "
share is the " minimum " plus the ratio of income from entrepreneurship
to G.D.P. 1
For countries for which income from unincorporated enterprises
(entrepreneurship) was not available, the maximum was taken to be
the largest of the maxima available for the countries in the same per
caput income group 2 , subject to the qualification that if this figure
exceeded the ratio of income from property and entrepreneurship to
G.D.P. plus the minimum share for the country concerned, the latter
figure was used. Where the minimum was not available, the lowest
of the available minima for countries in the same income group was used.
Fortunately, this expedient did not prove necessary in many cases. The
final wage share weight was computed by averaging the minimum and
maximum shares by the following formula 3 :
Minimum
(1 + Minimum) —Maximum
There is great variation among countries in both the minimum and
the maximum wage shares. The minimum is typically lower for underdeveloped countries, though not consistently so. The maximum, on
the other hand, seems to have no relationship to level of development :
it is roughly similar for China (Taiwan), Jamaica, Sweden and the United
States on the one hand and for Algeria, Argentina, Austria and the
Philippines on the other. The relevant data are shown in Appendix II,
table 3.
Education
Until fairly recently, expenditures on education were looked upon
as consumption expenditures. The production potential of education
has come in for so much discussion during the last few years that the
case for it is sometimes overstated. Denison, as already noted, attributed
1

Income from entrepreneurship is partly compensation for services and partly
profit. It is the former alone which should be added, but the data do not permit
this adjustment.
2
This term is defined below, pp. 67-68.
3
Since the difference between the maximum and minimum wage shares is the
proportion of G.D.P. which is income from entrepreneurship, the formula assumes
that this difference is split between wages and profits in the same proportion as
the rest of G.D.P.

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

almost one-quarter of United States growth from 1929 to 1957 to this
factor, and since the marginal returns to education are usually assumed
to be higher in less developed countries, it is small wonder that so much
interest has been aroused.
A great many conceptual difficulties lie in the way of measuring effective educational inputs. The most obvious statistics are those for school
enrolment, and are available for different levels and types of education.
In addition to the primary, secondary and higher education categories,
there are for many countries separate data on vocational schools and
adult education.
The growth-producing effect of the different categories is not at all
uniform. Time-lags constitute one problem. An increased expenditure
on primary education at year t will not become an economic asset until
year t+n, the n varying with the years of schooling. The lag will be
smaller for other forms of education, particularly short-term vocational
training, but it exists. One way of handling the problem would be to
introduce appropriate lags into the analysis, e.g. to correlate growth
in year t with primary school enrolment in year t—6. We decided
that in this preliminary study there was a case for avoiding complications
of the kind that would be entailed in the use of lagged variables, but certainly this is a procedure that should be considered carefully in further
work.
Another problem is that of the intrinsic value of a particular type
of education as a development stimulus. The case for vocational
training is clear. Students in vocational schools are being prepared
directly for working life, and such training can be looked upon as an
immediate input into a nation's productive fund. Adult education, on
the other hand, varies greatly in its purpose. Some of it may be purely
cultural—evening classes in art, music, dancing, pottery-making. Much
of it however, and particularly in underdeveloped countries, is quite
utilitarian in purpose, including literacy courses and evening technical
training. However, since so little is known about the composition of
adult education, it was felt that this category had better be omitted.
Of conventional primary, secondary, and higher education there can be
little doubt in terms of ultimate contribution to economic efficiency,
though immediate pay-offs may vary with the specific type.
The four indicators that were chosen to represent the educational
factor are:
(a) primary school enrolment as a percentage of the population
aged 5 to 14 years;
(b) secondary school enrolment as a percentage of the population
aged 15 to 19 years;

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

(c) vocational school enrolment as a percentage of the population
aged 15 to 19 years;
(d) higher educational enrolment as a percentage of the population
aged 20 to 24 years.
In each case, the summary figures shown in Appendix II, table 4
represent the rate of increase (or decrease) of the ratio over the period
1950-60, or for as large a fraction of the period as the availability of
data allows, using terminal years rather than averages of annual data
to determine the rates.
Harbison and Myers1 use primary and secondary school enrolment
ratios, among others, with an adjustment to correct for differences in
the duration of schooling among the countries. The school data on
which this adjustment was based were not available to us, so we have
contented ourselves with the unadjusted enrolment ratios. The ratios
adjusted for duration of schooling offer some drawbacks of their own
in situations in which the length of schooling is shifting. The number
of students completing each level of education would be a better measure,
but there are no comprehensive data of this character.
There are other possible indicators of the educational effort a nation
is making. Harbison and Myers use teachers, engineers and scientists,
and physicians and dentists in proportion to the population, as well
as the specialities of students enrolled in higher educational institutions,
as measures of this. The physician-population ratio was used by us as
an indicator of health, but the teacher-population ratio suffers from the
defect of great variation in the quality of teachers. The others, while
germane to the problem of specialised manpower with which they are
concerned, are of less interest in the present context.
Ratios of expenditures on education to national product were our
first choice for educational indicators. Unfortunately, such data are
available for relatively few countries prior to the closing years of the
1950s, so that rates of change of such expenditures could not have been
secured. Moreover, even the data available are often incomplete,
usually omitting privately financed education and often the contribution
of government below the level of the national administration.
The objection may be raised that in any event school enrolment is
not an adequate indicator of educational effort. Ideally, it would be
desirable to have an educational production function which would
translate inputs of education into outputs of trained manpower. Fixed
combinations of iron and energy will yield determinable quantities of
Education, Manpower and Economic Growth, op. cit.

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

steel of known quality, but resources allocated to education may lead
to greatly different outputs depending on curriculum, teaching ability
and, most of all, the quality of the student. We are obliged perforce
to fall back on the crude measures of input available to us, and assume
a transformation process through which, in the long run and in the mass,
equality in the distribution of human endowment among nations and
the spread of advanced educational methods will yield a product that
does not vary radically from one country to another. In the short run,
unfortunately, this assumption may not be justified, and it may well be
that, as a result, our ratios overstate the output of the educational systems
in the less developed nations.
Health
As in the case of other social expenditures, the primary purpose of
expenditures on health is not to increase labour productivity. Yet
there can be little question that raising the general level of health of a
population contributes in important measure to the effectiveness of its
manpower input through promoting greater mental and physical effort,
reduced loss of working time, and fewer accidents on the job. The
question is whether the indicators available to measure improvement
in health are sufficiently significant and sensitive to reveal its contribution
to productivity.
Statistics on expenditures for personal care and health are available
from the Yearbook of National Accounts Statistics. The deficiency
of this measure is that it excludes current operating expenditures by
government on medical care and health services, as well as capital
construction for health purposes. Government health expenditures
vary greatly from country to country, and they can be large.1 Particularly in underdeveloped countries, they can far exceed private health
expenditures in magnitude. The Yearbook data are thus inadequate
for our purposes.
The International Labour Office, in two publications 2, has gathered
data on medical care expenditures under social security schemes for
a number of countries. However, retrospective data are available for
a relatively small sample of countries.
1
A recent study indicates that government expenditure on medical care and health
services ranged from 12 to 37 per cent, of total government consumption expenditures
for the sample of countries covered. See World Health Organization: The Cost
and Means of Financing Medical Care Services, A Study of Health Costs (Geneva,
1962) (W.H.O./PA/77.62).
2
I.L.O. : The Cost of Medical Care, Studies and Reports, New Series, No. 51
(Geneva, 1959) and The Cost of Social Security, 1949-1957 (Geneva, 1961). A new
edition of the latter is in preparation.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

For want of a more direct measure, we have fallen back on the following indirect indicators of health (see Appendix II, table 5) :
(a) number of inhabitants per physician ;
(b) number of hospital beds per 1,000 inhabitants;
(c) calories available per caput;
(d) infant mortality.
These indicators have the advantage, from our point of view, of
being available for most of the countries in our sample. The question
is whether they constitute à suitable index of changes in health inputs
over time.
A recent United Nations report 1 recommended as health indicators
expectation of life at birth, infant mortality rates and crude annual death
rates. These measures reflect the end result of many influences other
than health expenditures, such as urbanisation, the spread of medical
knowledge, sanitation, better housing and international disease control,
but nevertheless they do bear some relation to the medical effort. We
selected the infant mortality indicator as the one from which rates of
change during the decade 1950-60 could most readily be calculated.
The United Nations report has the following to say of the number of
hospital beds and of physicians :
Although these indicators are available for a larger number of countries than
other indicators that have been recommended and although they may be
useful for national purposes, they are not satisfactory as measures of levels
of health, since the effectiveness of these services depends to a considerable
extent on the way in which they are organised, on their distribution and on

the professional qualifications of the medical personnel.2

We are using the health indicators to obtain national rates of change
of health inputs rather than to compare absolute levels of health among
countries, so that the objection cited above is not critical, from our point
of view. There may be qualification in the fact that, over time, better
health organisation results in a greater yield per unit of input (a situation
analogous to what happens in the production of goods), but we have not
allowed for this possibility.
The nutrition variable, which is based upon estimates by the Food
and Agriculture Organization, represents the number of calories per head
available daily for human consumption within a country at the retail
level. We have little doubt of the relationship, at low income levels,
1

United Nations : International Definition and Measurement of Levels of Living:
an Interim Guide fNew York, 1961), p. 5.
a
Ibid., p. 6.

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

between food intake and the productive efficiency of labour. Thus,
for example, an increase above the 1,900 daily calories available to the
Japanese worker in 1949 or the 1,990 calories available in Ceylon in
1952 probably led to greater mental and physical vigour. However,
as the daily level approaches 3,000 calories, the marginal contribution,
via health and energy, to production may decline rapidly and even
become negative. During the 1950s the availability of calories in some
advanced countries in which per caput income was rising tended to decline :
from 3,110 to 2,930 in Sweden, from 3,170 to 2,980 in Switzerland,
and from 3,180 to 3,120 in the United States. This was by no means
universally true—there was an increase from 3,220 to 3,260 in Australia,
from 3,050 to 3,150 in Canada, and from 3,130 to 3,290 in the United
Kingdom—but the possibility of a change in the direction of the nutrition-health-productivity relationship is not to be ruled out. A correction
factor in the form of the percentage of calories of animal origin or the
protein content of the diet would have been possible, but since the
possibility of an inverse relationship appears to he only at the top of the
income scale, it seemed preferable to handle the matter by appropriate
sub-groupings of countries based on levels of income.
Housing
When one leaves education and health and approaches the other
social variables, the link between expenditures and contribution to
development via increased labour productivity becomes more tenuous.
Yet it is probably true that decent housing satisfies one of the most
urgent needs of workers in developing areas, particularly in the cities,
and helps inculcate orderly habits of living and personal care with a
carry-over to the factory floor. Men who live in shacks, or under
conditions which offer no opportunity for privacy or personal hygiene,
can hardly be expected to put in a good day's work. Once minimum
food supplies are assured, housing seems to come next in the order of
priority demand. The question is whether, and to what extent, through
investment in housing, the labour force is being endowed with quality
attributes leading to increased efficiency.
We have selected two measures of improvements in housing:
(a) dwelling units completed per head;
(b) 'the ratio of fixed capital formation in dwellings to the gross
national product.
In the first of these measures, which is available from the Statistical
Yearbook1, a dwelling unit is defined as a building or part of a building
1

United Nations: Statistical

Yearbook (New York).

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

suitable for occupancy by one family. Luxury villas and one-room
flats are equated in this measure, but at least there is some indication
of how many families are newly housed. The second measure, which is
drawn from the Yearbook of National Accounts Statistics, suffers from
the defect that it gives no notion of the composition of the new housing
stock. Two countries can have the same gross investment in housing
with completely different social effects, depending upon the distribution
of the investment among different types and qualities of dwellings.
The physical measure seems the better one for our purposes, but since
it is available for only 20 countries in the sample, the investment data
were added as a supplementary variable.1 Summary ratios for the
decade were calculated by averaging the annual ratios.
Social Security
Social security is a term covering a great diversity of separate programmes. For some of these, such as old-age pensions, unemployment
benefits and family allowances, any direct significant influence on economic growth-inducing propensities may be difficult to discern except
in so far as they raise the level of work efficiency by allaying fears of
economic catastrophe. Such other programmes as sickness insurance,
workmen's compensation and public health services may contribute

more directly to the quality of the labour force.
Other things being equal, we would not have been disposed, at this
preliminary stage of the inquiry, to include social security among labour
quality variables to be tested. However, the I.L.O. has very heavy
responsibilities in this area. Its advice is constantly being sought by
member nations, particularly the less developed ones, and some clarification of the contribution of social security to economic growth, either
positive or negative, is urgently needed. It is for this reason that we
have experimented with social security variables, in full knowledge
of the causal complexities.
The indicators that have been employed for social security are:
1. Social security benefits paid as a percentage of national income ;
2. Average annual social security expenditures per head of population between 15 and 64 years of age, in constant prices.

1
G.N.P. rather than G.D.P. was used as a base for the housing investment ratio,
since series for a substantial number of European countries had already been calculated
in this way. See United Nations, Economic Commission for Europe : Annual Bulletin
of Housing and Building Statistics for Europe (Geneva). The substitution of G.D.P.
would not substantially affect the comparative ratios.

INDICATORS OF ECONOMIC GROWTH AND LABOUR QUALITY

These figures were calculated by the I.L.O.1 The summary figures
for the ratio of benefits paid to national income were calculated as the
averages of the annual ratios. Benefits paid per head have been expressed
in terms of their rates of increase.
Less aggregated indicators of social security would have been
desirable, in view of the varying impact of the several programmes on
growth. An effort was made to use the data on persons receiving payment under various programmes as shown in the Year Book of Labour
Statistics, but the coverage by country and by year proved inadequate
for our purposes. Use of the I.L.O. expenditure break-downs by programme did not seem warranted at this stage because of limitations with
respect to international comparability and the amount of calculation
that would have been involved. However, we are convinced that these
data merit further analysis, particularly when they are carried beyond
1957. It is also to be hoped that data for further countries beyond the
30 odd (out of 52) in our sample will become available.2
The more relevant of the two indicators chosen for the purpose of
this study is that showing the rate of increase of benefits paid per head.
This shows the relative improvement in the lot of benefit recipients,
over the decade, from one country to another. It is perhaps not surprising that these rates should be quite high for some underdeveloped
countries which started the decade with little or nothing in the way
of social security, and it is interesting to note that, whether through
cause or effect, three of the more advanced countries with high rates
of growth, the Federal Republic of Germany, Italy and Japan, showed
relatively large increases in benefits paid.
Subdivision of Countries by Income Level
The impact of investment in social programmes on the quality of
labour is likely to vary greatly with the level of national income.
Marginal returns to such investment would ordinarily be greater at
lower income levels and tend to decline as the level of income rose. In
the case of some of the social variables there may even be zero returns
at very high income levels. The programmes may nonetheless be
pushed, despite the fact that they do not contribute to national economic
growth.
In order to test the behaviour of the social variables at different
income levels, we have divided our sample of countries into six per caput
1
Unfortunately, the I.L.O. series in the 1961 edition of the Cost of Social Security
did not go beyond 1957, and we did not make independent estimates for later years.
2
Expansion of the sample is limited by the fact that social security programmes
are not widespread in the low income countries.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

income groups. Here we followed the Report on the World Social
Situation, which found that the following income groups, based on the
average annual per caput national income of 1956-58 estimated in U.S.
dollars, provided a meaningful basis for the analysis of social factors
in economic development:
Group
Group
Group
Group
Group
Group

I
II
III
IV
V
VI

$1,000 and over
$575-1,000
$350-575
$200-350
$100-200
under $100

(6
(12
(8
(10
(11
(5

countries)
countries)
countries)
countries)
countries)
countries)

The distribution of countries under this classification is unfortunately
skewed in the direction of the developed countries because of the
availability of statistics. The great majority of the underdeveloped
nations had per caput incomes of less than $200 during the years indicated,
but the data necessary for their inclusion in our sample were not available.
Even to get as many as 16 required a good deal of improvisation. This
process of selection must remain a weakness of any comparative study
until the data problem is solved in a more satisfactory manner.
A final word on the behaviour of the social variables at different
income levels given in the Report is worth repeating:
.. .the rate of economic development is proportionately greater at the higher
levels, while the rate of social development—particularly health—is greater
at the lower levels. Thus, it is easier for the high-income countries to expand
their industry than to lower their mortality ratio, whereas, comparatively
speaking, the opposite is true of the low-income countries... There is a
suggestion in the data of a break somewhere between the top three and bottom
three groups, around the $300-35350 level—a natural watershed above which
the economic indicators advance rapidly, and the health and education indicators start to move more slowly towards their ceiling.1

Report on the World Social Situation, op. cit., p. 43.

CHAPTER V
STATISTICAL METHODS AND RESULTS
THE STATISTICAL ANALYSIS

The whole of our statistical analysis of the effects of the quality
of labour on economic growth in the 1950s is in terms of multiple regression estimation of the inter-relationships among 15 variables.
The 15 variables in our analysis can be divided into two groups:
economic variables, of which there are three, and indicators of labour
quality, of which we have 12.
The three economic variables we have considered are the variables
Z 0 , Z\, and Z 2 , introduced toward the end of Chapter III. These
are defined as follows.
AY
(wL)AL
~r~(pY)~L
Z0 =

i.e. that part of the growth in output which is
(wL)
(pY)
not accounted for by growth in the labour input measured in numbers
of men, divided by the proportion of domestic product which is received
as wages. This is the dependent variable in all our regression analyses,
i.e. it is the variable which we try to explain by various combinations
of the other 14 variables we have studied. It is tabulated for each
of the 52 countries in the sample in the first column of Appendix II,
table 7.
Zl = — i.e. the ratio of investment expenditure to the wage bill.
wL
This variable is an independent variable in all of our regression equations;
it is always one of the variables used to " explain " the variable Zo.
Zi is tabulated in column 2 of Appendix II, table 7.
Z 2 = co— i.e. the ratio of investment expenditure to the wage bill
wL
multiplied by the average annual rate of growth of the real wage experienced in the decade 1950 to 1960. The absence or presence of this
variable among the collection of explanatory variables used distinguishes
the two versions of the model. In the first version it is not used at all.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

In the second version it is always used. This variable is tabulated for
each of the 52 countries in the third column of Appendix II, table 7.
The first part of the statistical analysis is concerned solely with the
first version of the model. At this stage we are concerned with ascertaining the extent to which the variable Z 0 can be explained by the variable
Zi and different collections of indicators of labour quality.
The first relationship to be investigated was of the form
Z0 = 6 Z 1 + c

[5.11

i.e. Zo is some linear function of Z\ . Estimates of the coefficients of
this relationship were obtained by least-squares fitting of the equation
to the observed values of the variables Z 0 and Z\ . The results of this
analysis are shown in the first four rows of Appendix II, table 8. Those
in the first row of the table relate to the estimates obtained when equation
5.1 was fitted to the data for all countries in the first and second income
groups; those in the second row pertain to countries in income groups III
and IV; and those in the third row to countries in income groups V and VI.
In obtaining the results given in the fourth row of the table, all 52 countries in the sample were considered simultaneously. The number of
countries in the sample to which the results in any row relate is shown
in column 2 of the table. The income groups within which these countries he are shown in the left-hand column. Each regression is given a
number, which is recorded in column 1 of the table. Thus the regression 1 involves the least-squares fitting of equation 5.1 to the data for the
18 countries in income groups I and II.
Estimates of the parameters b and c of equation 5.1 are given in
columns 4 and 6 respectively. Estimates of the standard errors of these
estimates are given in columns 5 and 7. The coefficient of determination
is recorded in column 3. For example, for the regression 1, an estimate
of b of 9.86 with an estimated standard error of 4.87 was recorded.
The constant term and its standard error were estimated to be 1.60 and
1.75 respectively. R2, which is the proportion of the variance of the
dependent variable Z 0 which is accounted for by the regression is, 0.20,
or 20 per cent., as shown in column 3.
The interpretation of the results of the regressions 1 to 4 follows from
a comparison of equation 3.16 and the regression equation 5.1. According to these two equations
Zo = r Z i + —
and
Z 0 = bZy + c

STATISTICAL METHODS AND RESULTS

Thus, r in equation 3.16 is replaced by b in equation 5.1 and ——- in 3.16
is replaced by the constant term c. The immediate interpretation of the
estimate of b obtained by fitting equation 5.1 to data for a group of
countries is, therefore, that b is an estimate of the average value of r
for that group of countries and, similarly, that c is an estimate of the
average value of —r-.
Following the interpretation outlined above, the average immediate
rate of profit in countries of groups I and II is just under 10 per cent.
For group III and IV countries, in which the level of income per head
is lower than in groups I and II, the estimated average immediate rate
of profit is a little under 12 per cent., indicating that the average pay-off
period for new investment projects in these countries is between eight
and nine years. For the least developed countries in the sample, those
in income groups V and VI, the average immediate rate of profit is
very low—only 3.4 per cent.—indicating a pay-off period of nearly
30 years. If the estimates of the constant term c can be interpreted as
average rates of growth of the quality of labour, then the results for
the regressions 1 to 3 indicate that, on average, the lower the level of
economic development the faster has the quality of labour grown in
the decade 1950 to 1960. The average figure for the countries in the
two highest income groups is 1.6 per cent, per annum. For countries
in the income groups III and IV the quality of labour has grown at an
average rate of 2.2 per cent, per annum, while for countries in the lowest
income groups an even greater rate of growth of 3.5 per cent, per annum
is estimated.
These results sound interesting but perhaps exceed the limits of
plausibility. They should be viewed with caution. On statistical
grounds the estimates cannot be-regarded as justifying the above picture
of results at any substantial level of significance. The estimates of the
standard errors of the coefficients given in columns 5 and 7 of Appendix II,
table 8 are calculated in the usual text-book manner. According to
them only one of the six coefficient estimates discussed in the previous
paragraph is significantly different from zero at the 5 per cent, level and,
at the same level, neither the estimated coefficients of Z\ nor the
estimated constant terms differ significantly as between income groups.
But this mechanistic interpretation of sample statistics is inappropriate to the type of analysis in which we are here engaged. The relative
magnitudes of estimates of coefficients and their standard errors is an
index of the relative reliability of the coefficients. It is a long way

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

from being an absolute index, however. We would not have estimated
equation 5.1 in the first place had we thought that it might not be a
meaningful formulation. There is little justification in assuming that
the error term appropriate to equation 5.1 has a normal distribution
with zero mean and the same variance for all countries, the assumptions
required if the conventional significance level construction is to be put on
the regression results.
Quite apart from the relevance of significance level constructions is
the question of whether it is appropriate to interpret the equations
which have been estimated in the context of the theoretically derived
relationship 3.16. This is essentially a question of identification and one
which is taken up in the next section of this chapter. For the present,
let it suffice to say that we hope that the analogy is appropriate.
The constant terms of regression equations 1 to 4 give the impression
that growth in the quality of labour has taken place at the rate of 1 to 3
per cent, per annum on average during the 1950s in the countries in
the sample. Thus there is some evidence that this factor has a substantial
effect on economic development, and particularly so in those countries
in which development is most needed. It would seem reasonable,
therefore, to pursue the matter further by trying to discover what it
is that this constant is an average of. The constant term is, in itself,
of very little interest beyond indicating the order of magnitude of the
contribution of growth in the quality of labour to the variable Z 0 .
It does not indicate what accounts for this growth. The rest of the
statistical analysis is designed to ascertain the implications of replacing
the constant term by indices of labour's quality. By so doing, it becomes
possible to obtain estimates of the growth of labour quality which differ
between countries in the same income range, and hence perhaps to
improve on the very low proportion of the variance of Zo which is
explained by the simple relationship 5.1.
The 12 indices of labour quality which were used in the analysis have
been described in the previous chapter. They can be classified under
four headings : health and education (for each of which there are four
different indices) and housing and social security (for each of which
there are two indicators). The observed values of the rates of growth
of the indicators under each heading are recorded in Appendix II,
tables 4 to 6.
As is clear from the tables, one of the main difficulties to be overcome
in analysing this information is that the data are incomplete. It is
not possible to include every country in regressions involving a particular
collection of quality indicators. The criterion adopted was to use the

STATISTICAL METHODS AND RESULTS

data for all the countries in the sample for which we have a complete
set of information on the quality indicators involved in the regression.
Thus, for example, regression 5 in table 8 involves the quality indicators
<2i to 04 . (The definitions of the indicators Qx to Qu are given in the
notes to Appendix II, tables 8 and 9.) As shown in the left-hand
column, the estimates are based on observations of the variables for
countries in income groups I and II only. The significance of the number
16 in column 2 is that for only 16 of the 18 countries in the income
groups I and II were there complete sets of information for the variables
Qx to Qa,. The results shown in row 5 of the table are based on the
observations for these 16 countries.
In general, the equation estimated in the regressions 5 to 33 is of the
form:
¿ßi
AQ,
Z 0 = bZt + a f — + a,-=^+...
Gi

[5.2]

AQi

where

is the rate of growth of the indicator Q\ . As shown in

Appendix II, table 8, various collections of quality indicators were used
for these regressions, but never more than four for any particular
regression.
The interpretation of the results of the regressions 5 to 33 follows
from a comparison of equation 5.2 and equation 3.16, and is similar
to that appropriate to the regressions 1 to 4. The estimate of the coefficient b may still be interpreted as an estimate of the average immediate
rate of profit for the group of countries involved. The only difference
is that now, instead of assuming that the rate of growth of the quality
of labour is the same for all countries, it is allowed to vary according
to, and to be identified with, the growth of the quality indicators involved
in each regression.
It is not our intention to recount here the detail of the story that the
regressions 5 to 33 have to tell. The interested reader can reconstruct
it for himself from the numerical results given in Appendix II, table 8.
Some general comment is, however, appropriate.
The first step in the analysis after running the regressions 1 to 4
was to identify the quality of labour first with health, then with housing,
then with social security and, finally, with education. The results
obtained are given in rows 5 to 15 of Appendix II, table 8. When
quality was equated to health, for example, all four of the health indicators were considered simultaneously, the object being to ascertain which

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

one(s) of them showed sensible association with the dependent variable.
Of course, what is sense is a subjective judgment, but one which we were
prepared to make. For example, in the health analysis recorded as
the results of regressions 5 to 7, the variable g 3 , calories per head,
was considered to show the most sensible relationship; growth in Qi,
number of inhabitants per physician, was found to be positively associated with the dependent variable; and the directions of the association
of the other two variables, Qi and Q4 , varied as between income groupings. This is not in itself necessarily implausible, but consideration
of the values of the coefficients of these indicators relative to their
standard errors when compared with the analogous ratios for Ô3 indicated
that of the four variables ßi to Q4, the variable for calories per head,
Qi, was outstandingly the most successful. Similarly, as between the
two housing indicators, the variable Qe , which is the proportion of the
domestic product spent on housing, was the superior. Of the two social
security variables, benefits paid per head was considered to be the better.
From the results obtained by using the four education indicators it
was clear that higher education showed the most substantial association,
but we were perhaps a little hasty in not regarding the results for secondary and vocational education a little more sympathetically than we in
fact did. However that may be, the next step was to discover what
happened when the four leading indicators, one from each of our four
main groups selected as just described, were considered simultaneously.
The results are set out as regression 16, from which it can be seen that,
first, all four indicators have the " right " signs and, secondly, that
calories per head and then higher education appear as the most important
explanatory values. Unfortunately, only 21 countries could be included
in this analysis, and so experiments were made in which some of these
four variables were excluded to allow more countries to be considered.
The results are set out as regressions 17 to 23. Finally, we investigated
what happens when calories per head is alone considered as a determinant
of labour quality. These results are recorded as regressions 24 to 28.
They show surprising uniformity in the estimated coefficients of the
average immediate rate of profit (13.1 to 13.7) and the coefficient of the
rate of growth of calories per head (1.3 to 1.9) as between income groups.
Two further steps were taken to complete the analysis of the first
version of the model. One was to plot the residuals from the regression
equations 24 and 28 against the observed values of the growth rates of
all the indicators other than calories per head and the enrolment ratio
for higher education. No associations were apparent. The concluding
investigation was to put a constant term back in the ranks of indicators
along with calories per head. The results are shown as regressions 29

STATISTICAL METHODS AND RESULTS

to 33. They show no stability and little improvement on the comparable
values of R2 recorded for the regressions 24 to 28. 1
The second version of the model differs from the first in that both
Z\ and Z 2 are used throughout as explanatory variables. The regression
equation for this version of the model is of the general form:
Z0 = bíZí + b2Z2 + -^

15-31

and the results obtained from using it are given in Appendix II, table 9.
Comparison of equation 5.3 with equation 3.18, which is:
Z 0 = X*Zl + eZ1 + -2
indicates an immediate interpretation of the coefficients b\ and Z>2 in
the regression equation 5.3. The coefficient X* of the variable Z\
in equation 3.18 is the expected average rate of profit on current investment, provided that investment projects are chosen so as to maximise
this rate. Under this same condition, the coefficient e of the variable Z 2
is the elasticity of the capital-output ratio with respect to the average
product of labour as between alternative techniques currently available.
There is therefore some basis for interpreting the coefficient b\ as an
average rate of profit and è 2 as a measure of the difficulty of substituting
capital for labour.
Some of the results obtained by using this second version of the
model are set out in Appendix II, table 9. A few comments on them are
appropriate. First, it is clear from a comparison of the results for
comparable regressions as between the first and the second versions of
the model that the introduction of the variable Z 2 results in a substantial
increase in the explanatory power of the regression equation as measured
by R2. The coefficient of Z 2 is always positive, as it should be, and there
is some evidence in the results to suggest that it decreases with the level
of income per head, i.e. the countries in the higher income groups have
more difficulty in substituting capital for labour than those in the lower
groups.
The estimates of the coefficient of Z\ do not look altogether like
estimates of average rates of profits. The numbers are too variable,
and some greater degree of control over them is clearly desirable.
1
The reason why inclusion of a constant term increases the value of R2 is that,
for all regressions, R2 is calculated as one minus the ratio of the sum of squares of
the residuals from the fitted regression line to the sum of squares of the deviations of
the dependent variable, Zo, from its mean. When there is no constant term in the
regression, the mean of these residuals is not necessarily zero.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

A priori there was no reason why the best quality indicator in each
of our four groups should turn out to be the same in the second version
of the model as in the first, but this is what in fact happened. Indeed,
the behaviour of the quality indicators was found to be very much the
same as in the first version of the model, and for this reason the results
are given in a more abbreviated form. The regression 38 is the analogue
of the regression 16 as between the two versions of the model, and again
we find that, of the indicators considered in this study, calories per head
and then higher education enrolment ratios have claims to being the
most directly associated with the quality of labour in the context of the
model.
COMMENT ON THE RESULTS

One of the stated objectives of this inquiry was to ascertain the extent
to which the heterogeneity in the development of the countries in the
sample could be explained by differences in the quality of their labour
forces. The answer, in statistical terms, is given by the values of R2
recorded in Appendix II, tables 8 and 9. However, while these values
indicate that the degree of explanation achieved, particularly in the case
of the first version of the model, is not very large, it is only fair to point
out that had high values of R2 been our prime objective, it would have
been possible, by simultaneous consideration of more quality indicators,
to produce larger values than those shown.
It is assumed in the model that there are no differences among
countries in the efficiency of new capital as measured by its immediate
rate of profit. In the first version of the model, the contribution of
capital to growth is given by the product of the investment ratio (for
each country) and the assumed common immediate rate of profit for
all the countries in the sample. For regression equation 16, which
has the highest explanatory power of all regressions using this version,
the common profit rate was estimated at 9.7 per cent. This rate is
represented by the straight line in figure 6, which is to be interpreted in
a fashion similar to figure 1. The tails of the arrows in figure 6 show the
rates of growth of output not attributable to increases in the size of the
labour force for each country; these are the same points as those shown
in figure 3. The vertical distance between the tail of each arrow and
the line marking off the profit rate 9.7 per cent, is, therefore, the growth
rate to be explained by changes in the quality of labour. The arrows
show how far regression 16 takes us in this respect for all countries for
which data are available. Not all the variation has been removed,
but there is a much greater uniformity among the countries than in
the original diagram.

STATISTICAL METHODS AND RESULTS

A problem which arises in interpreting the results of all the regressions
is that of distinguishing between the cause and effect relationships
between variables which appear to be statistically correlated. In the
present context it may legitimately be asked whether the increase in the
available number of calories during the 1950s or the increase in higher
educational enrolment or other quality indicators made a contribution
to operating efficiency through improving the quality of labour, or
whether the association with the quality indicators is simply a reflection
of greater demand for the goods and services which they represent
consequent upon the growth of national product and income.
There are two reasons for believing that the direction of causality
runs from calories and higher education to economic growth, although
it cannot be denied that there must be something of the opposite effect
as well. The first stems from our use of cross-section data in which
increments in the variables rather than their absolute levels are compared.
Thus, suppose that for each country in the sample Engel's law 1 holds
in the form :
- = ag 3 3
P

[5.41

w
where — is the real wage and 0 3 is calorie consumption per head. Taking
P
logarithms of both sides of this equation and differentiating yields:

co =

ffi

[5.5]

where co is the rate of growth of the real wage, and — ^ is the rate of
increase of calories per head. It is this latter variable and not Q3
itself which enters into the regression analysis.
In figure 7 the rate of increase of calories per head is compared with
the rate of growth of real wages for the countries in our sample. In
order to make some allowance for the possible interdependence of ß,
the elasticity of demand for food, as measured by calorific intake, and
the level of income per head, the sample is broken down into two broad
income groups. Figure 7 does not reveal any clear relationship between
the variables. Figure 8 shows a similar comparison for higher education
1
The basic notion of Engel's law is that as family income rises the proportion
spent on food and other necessities declines, while the absolute level of food expenditure
continues to increase. This is consistent with equation 5.4 provided '()<#< 1, i.e.
the income elasticity of demand for food is positive but less than one.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

enrolment and the rate of growth of productivity (output per head).1
Again, there does not appear to be any evidence of a homogeneous
demand relationship among countries in either of the two broad income
groups.
The fact that the scatters in figures 7 and 8 are random is not a
sufficient condition for regarding the regression coefficients in
Appendix II, tables 8 and 9 as being identifiable with the influence of
the indicators on the quality of labour, although it is true to say that
had the scatter diagrams revealed strong associations, then the regression
results could not legitimately be interpreted as measures of supply
effects. This situation maintains despite the fact that, in terms of the
formal criteria for identifiability, the regression equations are identified
against the most obvious demand relationships taken one at a time.
This identifiability follows from the fact that these demand relationships
typically involve the variables income per head or real wages, which do
not enter as such into the regression equations.2
It does not require much ingenuity for those versed in econometrics
to see how consideration of additional plausible relationships between
the variables would break this argument. But there is little point in
pursuing this line of thought, since the formal criterion cannot be
strictly applied in situations, like the present one, in which the specification of the regression equation cannot pretend to be other than a crude
approximation to the true relationship between the variables. A
pragmatic approach is more rewarding. The services to production
rendered by a worker depend on his ability and the energy he puts into
his work. Ability is nurtured by education, and energy expended is
conditional on the availability of its raw material, namely calories.
We claim, therefore, that there is a prima facie case for interpreting
our main results, that calorific intake and, perhaps, education received,
are statistically important determinants of labour quality, as having
causal significance. But we would not go further at this juncture and
say that the case was proven.

1
Productivity rather than real wages was used in this comparison on the assumption that the demand for higher education in most countries is induced by social
rather than personal considerations.
2
That is to say that the order condition for identifiability is satisfied. This is
a necessary but not a sufficient condition, however.

CHAPTER VI
POSSIBLE EXTENSIONS OF THE STUDY
The present study was necessarily highly experimental in character.
When we embarked upon it, we had little knowledge of data availability,
and one of the chief results is a keener appreciation of some major gaps.
Our path to the final theoretical formulation underlying the study was
by no means a straight one, and promising avenues had to be abandoned,
on occasion, for practical reasons. Our observations on some of these
points may be of some interest both to the collectors and processors
of data and to those who, like ourselves, will continue to be data consumers.
THE DATA PROBLEM

If further work of the kind we have attempted is to be fruitful,
it is imperative that the sample of countries included in the analysis
be extended. The requisite data are more or less available for the
developed nations—although even here there are surprising lacunae—
but the less developed countries for which adequate data are carried
in the international yearbooks are few and far between. It may be
necessary, in the first instance, for investigators to delve into the individual country yearbooks, and even into unpublished material, for
key nations. This will be time consuming, but there may be no alternative unless there is a rapid expansion of international yearbook coverage.
As has already been indicated, the existing labour force data are
well below the minimum standard necessary to the achievement of
conclusive statistical results. The economically active population as
a whole is too broad and vague for analytical purposes. Data should
be broken down by age and sex and, above all, distributed according
to economic sector and skill. The number of countries for which one
can readily secure sectoral employment data is surprisingly small,
and even for these the different definitions of employment complicate
international comparison.
Statistics on investment and the distribution of the domestic product
in terms of factor payments are not much more satisfactory. Indeed,
the lack of data for investment in manufacturing was the major cause

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

of our failure to investigate possible relationships at that level. The
variables representing output, employment, investment and the distribution of income are minimal ingredients for any macro-economic model
of growth. Such models are hardly worth pursuing beyond the point
we have reached until the basic data are improved in both quality and
quantity.
We have explained how the procedure adopted in the numerical
analysis minimises the effects of differences in the profitability of investment among countries. This is obviously a biased procedure, and one
of the more important respects in which the analysis could be improved
is to remove this bias. This could be done by direct observation of
immediate profit rates (or pay-off periods) and average profit rates
(or long-run interest rates). Much of the relevant information could
be collected from international financial institutions. The resulting
improvement in the reliability of the model should be considerable,
since from thefirstversion of the model one could derive unique estimates
for each country of the rate of growth of the quality of labour. At the
same time, it would give more explicit recognition to the relevance of
monetary policy to development.
Many of the concepts involved in our model are more immediately
relevant when applied to particular sectors rather than to the entire
economy. The model is perhaps more appropriate for studying the
industrial rather than the agricultural and service sectors, and even for
the former it would be desirable to disaggregate still further. It is
not unlikely that, if the necessary information could be obtained, one
would learn more from a comparison of the world's textile industries
than from a study on a grossly aggregate basis.
There is no need to go into great detail on what would have been
desirable in the way of labour quality indicators. Their deficiencies are
well known to those who compile them, and are immediately brought
to the attention of those who use them. In the sphere of education,
financial information would be useful as a supplement to the physical
indicators we employed. Our health indicators corresponded only
very indirectly to the phenomena sought to be measured. The detailed
housing indicators currently available for a few countries would be
quite satisfactory if the country samples could be increased. The contribution of nutrition could perhaps be made more precise by qualifying
caloric intake with such measures as relative fat and protein levels in
the diet, and reliance on particular types of food.
There may well be indicators of labour quality other than those used
in the present study which will throw light on the growth process.

POSSIBLE EXTENSIONS OF THE STUDY

Denison, in his study of development in the United States, states that
" few studies offer more promise of adding to welfare and contributing
to wise decisions in a matter that may greatly affect the future growth
rate than a really thorough investigation of the present relationship
between hours [of work] and output ".1 Levels of unemployment
and real wages are other possibilities. Any factor which can be quantified and has a potential impact on the rate of economic growth should
be brought into the picture in the interest of reducing the area of uncertainty.
One of the most frustrating aspects of work based on international
comparisons of social statistics is that there are only a few countries
for which complete sets of data on a given collection of subjects exist.2
Thus, when we wanted to consider simultaneously calories available,
higher education, social security benefits paid, and expenditures on
housing, complete sets of the relevant data could be found for only
21 countries, predominantly among the most developed. If we had had
more complete information on the 12 labour quality factors, there could
have been a more sophisticated treatment of them in two respects.
Firstly, it might have been possible to assume more subtle relationships
between these factors and labour quality than the constant elasticity
relationships which our formulation implies. Secondly, we might
have investigated the complementary nature of many of these indicators
rather than assume, as we did, that different aspects of labour quality
are completely substitutable. For example, a large proportion of the
benefits of improved housing stem from improvements in health through
the provision of good water and sanitary facilities. This complementarity
could be recognised by replacing individual indicators by groups of
them expressed as index numbers. The weights of such indices might
be derived either directly by observation or indirectly from a principal
components analysis.
In sum, the possibility of extending the type of analysis exemplified
by this study is dependent in large measure on the production of more
data. The most hopeful portent for the future is the recent publication
of a Compendium of Social Statistics, 1963.3 This volume reached us
when we had already completed our statistical work, so that we were
not able to benefit from the labours that went into it. The Compendium
1
Edward F. DENISON: 7%e Sources of Economic Growth in the United States,
op. cit., p. 39.
2
There is also the problem, referred to above, of the lack of comparable national
product measures for the market and centrally planned economies.
3
United Nations : Compendium of Social Statistics: 1963 (Data available as of
1 November 1962) (New York, 1963).

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

should make it possible to experiment with indicators that we had to
eschew. However, since it was compiled largely from previously
published yearbooks, the problem of expanding the sample of countries
and providing greater disaggregation remains.
EXTENSIONS OF THE MODEL

The model we have used as a basis for our analysis is a very simple
one. It depends on a number of assumptions which inevitably make
it an abstraction from reality. But this is true of any model, and the
one we have used was chosen largely because the assumptions upon
which it is based represent a less serious abstraction from reality than
those underlying many of the alternative models which might have
been used.
The strongest assumption of the model is that techniques of production cannot be scrapped until they cease to earn a profit. This might
not appear to be a stringent assumption, but in fact it is critical in
defining the extent to which labour and capital are complements rather
than substitutes. For suppose that there were no restriction on the
date at which techniques could be scrapped : it would then follow
that at any moment in time the capital goods being used in conjunction
with labour, as specified by some particular technique, could be supplemented or replaced by new capital goods and hence a new technique
would come into operation and an old one would be scrapped. The
assumption of scrapping at zero profits restricts the choice of date at
which such realignments of capital goods and labour can take place.
It does not imply that capital goods are literally thrown away when they
cease to earn a profit in the technological context in which they have
been used so far. This is seen most clearly in the case of buildings.
When a particular industrial activity ceases to be profitable, it will
cease and the machinery associated with it may well be broken up.
But the building is unlikely to be pulled down. Much more likely, it
will be sold to house some new form of industrial activity, although in
these days of radical alteration in factory design serviceable buildings
are often destroyed.
It was implicitly assumed in the derivation of the model given in
Chapter III that the wage rate and price level do not differ between
plants of different vintages. The latter part of this assumption is probably not seriously at odds with reality, but it is common experience that
wage rates in an industry tend to be higher in new plants in which labour
productivity is presumably greater. This implies that wages in each
plant increase less rapidly than in industries as a whole, and the extent

POSSIBLE EXTENSIONS OF THE STUDY

of this phenomenon must be a reflection both of the rate of growth
of the industry and of the institutional framework within which the
labour market operates. One possible modification of the model
is to take account of this aspect of reality, if only to ascertain the degree
to which it is important.
We were at great pains in formulating our model to avoid the need
to specify the properties of the range of alternative techniques which
exist at a moment in time. The relationships which make up the model
are financial in character; at no point in the analysis has it been necessary
to specify the physical relationships which exist between output, labour
and capital. It is nevertheless true that the existence of such physical
relationships is assumed and underlies the model. A possible line of
development is to make these relationships explicit and to investigate
them directly.
In vintage models, because primary factors are complements and
not substitutes, there are two physical relationships implied—one between
output and employment, the other between output and capital. Either
may be investigated directly, but for practical reasons of data availability
the former is the more attractive. For each technique which is chosen,
a physical relationship exists between the output and the labour requirements of that technique. The precise form of this relationship is not
known, but it is deducible if one is prepared to specify the physical
characteristics of the alternative techniques which are available and the

criterion by which a choice between them is made. By aggregating
these physical relationships for all the techniques which are currently
being used, a macro-relationship between output and employment can
be derived. The quality of labour enters into this picture in terms of
the specification of the appropriate units in which labour should be
measured. In this way a theoretical context can be provided for studies
such as that of Harbison and Myers, discussed earlier.1
At a slightly more sophisticated level, if it is held that at a moment
in time there is a limit to the extent to which the average product of
labour can be increased by switching to more capital-intensive techniques,
then an upper limit to labour productivity can be specified. The gap
between this upper limit and actual productivity can be closed by increased
investment. Hence the growth of productivity in a country can be
analysed in terms of the growth of the upper limit due to improvements
in the quality of the labour force and the invention of new tephniques,
and the closing of the gap between current productivity and what it
could be, given the present state of knowledge, by capital formation.
1

For a discussion of this work, see Chapter II.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

In conclusion, it can be said that these and other elaborations would
make our model a much more effective analytic tool. One of the reasons
for undertaking the present study was to see how far it would take us
without further elaboration and to form some impression of whether
it represented a useful way of approaching the analysis of economic
growth. It is our opinion that the model performed fairly well, and the
possibilities for its development described in this chapter are not the
least of the reasons for favouring its continued use.

CHAPTER VII
CONCLUSIONS
The traditional approach to the analysis of economic growth has
been largely in terms of the investment factor. A sufficient volume of
investment, provided minimum standards of sagacity in allocating capital
among alternative lines of endeavour were observed, has often been
regarded as the key to progress. To the extent that labour was taken
into account, it was largely in terms of numbers employed, without
much effort being made to distinguish one man from another in terms
of skills and ability.
In this study we have gone to the opposite extreme of assuming that
labour is the crucial variable and have attempted to explain differences
in growth performance by concentrating on some of the components
of labour quality. This extreme position is implied in our assumption
of international equality in the immediate rate of profit (the variable r
in equation 3.16) or the pay-off period on new investment. We have
no illusions about the superiority of our approach to the earlier one;
quite clearly, there must be a marriage of the two before a viable theory
of growth can be produced. The only justification for our procedure
is that we thought it desirable to make a beginning in the relatively
neglected analysis of the labour quality factors, and that this task more
than exhausted our resources of manpower and time. To remove the
assumption of a standard rate of return on capital, and to evaluate
properly the real differences that undoubtedly exist in this factor from
country to country, based upon variations in infrastructure and in the
risk element, as well as other things that we have considered briefly,
would have involved a research enterprise as big as the one we have
undertaken. We proceeded as we did because we felt that despite this
limitation a partial approach on the labour quality side might nonetheless
throw new light on the economics of growth and pave the way for a
more balanced attack upon the subject.
THE PROBLEM OF DATA

Even within this limited framework, we were further circumscribed
by the necessity of adjusting our sights to the availability of data. The

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

variables which we were obliged to use to represent the quality of labour
are not ideal from any theoretical point of view; they are simply those
which were readily available. A certain scepticism must be maintained
toward many of the series which had to be used in order to take some
first steps. Not only is there some doubt about their quality in terms
of the mechanics of collection and compilation but, even more fundamental, there are basic conceptual discrepancies between what has been
and what we would have liked to see measured. Superimposed upon
this is the difficulty inherent in all international studies of ensuring that
economic and social factors which go by the same name in different
countries are really comparable, i.e. that a school or a dwelling or a
specified type of social benefit has roughly the same meaning from one
country to another.
THE MODEL

The model which we used for estimating the relationships in which
we were interested appears, on the whole, to be quite suitable for the
purpose. It involves a strong assumption of factor complementarity
arising out of the freezing of technological relationships for each separate
vintage of capital. Most production function formulations go to
the opposite extreme of assuming perfect substitutability. While
substitution of factors is undoubtedly possible to some degree, the line
we have chosen is probably a better approximation to reality than the
opposite one.
One of the outstanding virtues of the model, from the point of view
of practicality, is that it avoids the necessity of measuring the capital
stock, which otherwise would be an immediate and, at the present time,
insurmountable obstacle to the use of empirical analysis. Among
other advantages (and anticipating future research work along similar
lines), the model can be applied to individual sectors as well as to the
entire economy, and the number of labour quality variables to be studied
can be expanded indefinitely. One of the principal restrictions, the
assumption of international equality in the immediate rate of profit,
can be removed on the basis of additional information. At the same
time, however, we recognise that there may be other approaches to the
problem which are superior theoretically or in practice, or both, but
until they emerge we are not unhappy to pursue the lines followed here.
THE EMPIRICAL RESULTS

It is necessary to preface discussion of the relationships which we
have observed with a strong warning to the unwary: this is a preliminary

CONCLUSIONS

study, the data are rough, and the results must be taken with more than
the customary grain of salt. It is never safe to crystalise ideas into a
hard mould on the basis of correlation coefficients. If certain relationships which emerge fly in the face of common sense, they should be
examined once again. If hypotheses that might have appeared valid
from practical experience fail to emerge, they should not be considered
disproved at first blush. Above all, reasonable hypotheses which do
seem to be supported by the data should not be accepted automatically.
With this cautionary note, we may proceed to an examination of
the results shown in Appendix II, table 8. Neither the housing nor the
educational variables, taken as separate groups, appear to afford an
adequate explanation of the growth variable, judging from the value
of R2.1 The health and social security groups do somewhat better,
with the former in particular appearing to offer a fair degree of explanation. However, apart from calories per head, the coefficients of the
individual health variables do not seem to be significant. The other
individual quality variables that appear at all promising are investment
in dwellings, social security benefits paid per head, and higher educational
enrolment, and it was on this basis that these four indicators were singled
out for further analysis (see Appendix II, table 8, equations 16-28).
When all four are taken in combination, a fair degree of explanation
is secured (R2 = 0.65), but the explanatory power of the calories variable
overshadows the rest.
One might have expected, prima facie, that the lower the income
groups the greater would be the impact of better nutrition on the growth
rate. This seems to be borne out in equations 24 to 28, when the value
of the calorie coefficient is greater for the income groups III to VI
than for income groups I and II.2 This does not obtain, however, in the
case of the other labour quality variables, some of which even turn out
to be negatively associated with growth.
1
It will be recalled that the variable we are attempting to explain, Zo, is not
simply the rate of growth of G.D.P. It is that part of the growth of G.D.P. not
attributable to an increase in the labour force (unadjusted for quality), divided by the
proportion of G.D.P. paid out in wages. If the latter ratio were constant among
countries, then best estimates of the growth of G.D.P. " unexplained " by increases
in employment (the numerator of equation 3.16) could be secured simply by multiplying the coefficients of the independent variables by the constant term, and in this
case Zo could be considered the growth rate. In fact, the wage ratio varies among
countries from about 50 to 80 per cent, (see Appendix II, table 3). However, one
may still think of the dependent variable as representing the rate of growth to be
explained, but with the understanding that in any particular case the coefficients of
the independent variables would have to be multiplied by the wage ratio to obtain
the best estimate of their effects on the growth rate.
a
There is a small decline from groups III and IV to groups V and VI, however.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

The same picture emerges when the variable Z^ is added to the model.
The correlation coefficients are generally higher, but the calories indicator
emerges once more as the outstanding explanatory variable. In this
case, the value of its coefficient rises without exception as the level of
income declines.
The regression equations which have been fitted to the data can be
used for prediction in the customary manner. Thus, for example,
if one wanted to estimate the effect upon economic growth of increased
inputs of labour quality factors for a particular country, it would only
be necessary to insert the specific values of Z\ and Z 2 determined for
that country, and then to vary the ß factors along the lines of the experiment being undertaken. It need hardly be added that an assumption
of invariance in social and technological relationships among countries
and from past to future periods is fundamental to any such essay in
prediction.
The foregoing observations may be summarised more generally as
follows:
(1) In attempting to determine the factors contributing to economic
growth, a model which defines labour in terms of its quality provides
a better explanation than one in which labour input is measured in

numbers of persons.
(2) Of the labour quality indicators tested, the level of nutrition, as
measured by daily calories available per head, seemed to yield the closest
relationship with economic growth. Moreover, there are prima facie
grounds for believing that a causal relationship runs from the provision
of additional food to increased labour efficiency. However, the coefficients appearing in the regression equations should not be taken literally.
The reader is referred to our earlier remarks on the tentative nature of
the empirical results.
(3) The increase in higher educational enrolment showed some
promise as an explanatory variable, particularly among the low income
countries. This suggests that particular attention might be paid to the
role of this factor in these countries. However, the relationship was
not sufficiently strong to warrant the flat assertion that an expansion
of higher education is essential to growth.
(4) We do not feel that on the basis of our estimates one can conclude that other aspects of health, education, housing and social welfare
do not contribute to labour efficiency and, through it, to development.
To be in a position to reach such a conclusion, far more detailed work
will have to be done. In aiming toward this goal, it is likely that if the

CONCLUSIONS

model were applied to the non-agricultural sector of the economy,
and to manufacturing in particular, more significant results would be
obtained. This belief stems from the fact that, while most of the labour
quality indicators which were employed in the present study affect the
non-agricultural labour force almost exclusively (housing, social security,
most of education and health), the growth, capital and labour force
indicators cover the entire economy. Particularly in underdeveloped
countries, where the agricultural sector is apt to be large, this discrepancy may have resulted in masking the effects of the labour quality
inputs.
(5) There is the further consideration that the indicators chosen
to represent labour quality may not be appropriate for testing the hypothesis. Different combinations of the indicators we used, or of others
that were not considered for lack of data or imagination, may yield
more substantial results.
(6) We believe that our general approach to solving the riddle of
economic growth is a fruitful one. The solution is not likely to be
revealed in one swoop. On the contrary, there will have to be a good
deal of additional painstaking analysis of data, as well as some rethinking
of the theoretical problems, before hard and fast policy rules begin to
appear. But in the process of the analysis, and of the thought, ideas
about the proper path to growth will undoubtedly emerge.

APPENDICES

APPENDIX I
MATHEMATICAL APPENDIX

X:
N:
/:
p:
w:

The following notation is used:
the output of new plant.
the labour requirements of new plant.
the cost of new plant.
the price level.
the wage rate.
The average rate of return on an investment is given by A in the expression
/ = f (ptX - wxN)e~x-dx
o
pt

[A.1]

where 6 is the age of plant when it is scrapped.
The expected rate of growth of real wages is given by co in the expression
[A.2]
Px

The immediate rate of profit on investment is given by r where
r =

pX-wN

IA.3]

THEOREM 1

If X is a maximum then
dbg I
A = r+a>—~
d log N
Proof: From equations A.l and A.2 it follows that
/ = ¡"(pX-wNe^y-^dx
o

[A.4]

whence

Hence, given that dX = 0,
di =

wdN
co—X

(l-eia-X)e)

+ (pXe-X9-wNe(a-xy,)dO

[A.61

APPENDICES

A necessary condition for dX = 0 is that the coefficient of dB in equation A.6
be zero, i.e.
pX = wJVe*0*
[A.71
Substituting this result into equation A.6 we obtain
wNJ

<D-X\

From equations A.5 andI A.7 it is easily shown
sh
that
gPX\_

(0-¿fwN-pX+XI\

co V wN

wNJ

substituting this result in equation A.8 we obtain

)dN

cadi =UI-(pX-

wN) \—

IA.10]

or
dbg

d log N
Q.E.D.
THEOREM II

If r is a maximum and X is a homogeneous function of the first degree
of / and N, then
ÔX
r = p—
di
Proof: From the definition of r given in equation A.3 it follows that the condition dr = 0 is satisfied only if
pXdl = wNdl-wIdN

[A.121

Further, since if X is held constant,
dX

ÔX
dN +

° =™
equation A. 12 can be written as
dX

(

-ÖIdI

,A 131

dX\

Consequently, if A' is a homogeneous function of degree one of N and /,
then equation A. 14 reduces to
ÔX
p— = w
[A.151
oN
Under this condition, therefore, the expression for r given by equation A.3
reduces to

r = p—
ol

[A.161
Q.E.D.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

Sgq&
g.so <

Tf^^i/ScW'nco<^c4so^>nc4c<iTtT! : Tj : p^iri-^ : fnv©^H[-^»n

o -o o CA m en

. .-.

•o
a

M _
Tf - H j<

._ . „ o os

ino¡

o\ S

v > o \ s o c N o \ * o ^ * o o o o0r0- 3 o o « n ^ ^ - f n
< s © - *T s ov so o>no o s o <1 s c *V>
0 M

oo vo « « Tt P I o\ — f ^ . * * ^
loirfTtt^ m
—i oo vo
<MC4
vo
r4
-'—

"i.**!. *
vf©
--n ~*

J«

S. t.
Os©
or-

"t^
^

•o S

o *o o ON OS •—' 00
«SCNVSOsCsOsinvoOSVifn

osvocnvovrio<Scor-r-os
rHrtTt

~*co*t

«*1 S,
f^«o

m N t—^«-j^t—^ "^."T.»•H

O\^0>n^

r t

r- en" m ^ ^3"

t*^^t^ntnp*
<S
— oo

£88S8888S88S!88S88888S!8888

I
OSOSOSOSOVONONOSOSOSOSCSOSOVCSOSOSOVONOSOSONOSONOS

.J..JJ.IJ J„.
0i«>ij«í'«fefe25fafc!
O O. O O O, O. O & &
O „ „ Ü
IH « U
O V

5
8 2 2 - 2 2 ¿5-*.*
|H l , ^

"fia
11

ÖS

o &a
8858 Mil

Hilllili

OS OS OS

&*S5 8J ! as:i-l
jg.5;
9
z ^

U d R

t o »

IEE-SEE&E

ä a l o3"û~
i 11-8ñ l "i o í

V V

•SS
<2 E E

Ü «Ö « ü . . .. . r i oo O O »n

Bill
UH

•filili'

ää^lPl^h

fifia

Jamaica . . . .
Japan
Korea (South) . .
Luxembourg . .
Malaya (Fed. of)
Malta
Mauritius . . . .
Mexico
Netherlands . . . .
New Zealand . .
Nigeria
Norway . . . .
Panama . . . .
Peru
Philippines . . .
Portugal . . . .
Puerto Rico . . .
South Africa . .
Spain
Sweden
Switzerland . . .
Thailand . . . .
Tunisia
Turkey
United Kingdom
United States . .
Venezuela . . . .

Pounds
Yen a
Hwan a
Francs
U.S. dollars
Pounds
U.S. dollars
U.S. dollars
Guilders
Pounds
U.S. dollars
Kroner
U.S. dollars
Soles
Pesos
Escudos
Dollars
U.S. dollars
Pesetas a
Kroner
Francs a
Baht
U.S. dollars
Liras
Pounds
Dollars
Bolivars

1956 market prices
1955 market prices
1955 market prices
1959 factor cost
1950 prices
1953 market prices
1950 prices
1950 prices
1958 market prices
1953 factor cost
1950 prices
1954 factor cost
1950 prices
1954 market prices
1955 market prices
1954 factor prices
1958 market prices
1950 prices
1953 factor cost
1954 market prices
1958 market prices
1956 market prices
1950 prices
1948 factor cost
1954 factor cost
1958 market prices
1957 market prices

1954
1950
1953
1950
1950
1954
1950
1950
1950
1952
1950
1950
1950
1950
1950
1950
1950
1952
1949
1950
1950
1950
1950
1950
1950
1950
1950

1959
1960
1960
1959
1959
1960
1959
1959
1960
1960
1959
1960
1958
1958
1960
1960
1960
1959
1959
1960
1959
1960
1959
1960
1960
1960
1960

141.0
5,310.6°
856.3
13,670
907 c
35.3°
89 c
5,414 c
26,748
707 e
1,468 e
17,491
259 e
22,504 P
5,722 P
38,054
972.6
3,361 e
197,570
37,340
22.6
32,744

457 e
9,477.8
14,322
353,522
12,907

205.3
12,846.3 n
1,175.5
19,327 e
1,557
46.0°
121 ee
8,531
40,630 e
1,116
2,254 e
24,697
369 e
29,435 P
10,797 P
56,985
1,734.1
5,195 e
334,100
51,180
33.6
48,756
569 e
17,646.0
18,525
486,860
26,433

7.51
8.83
4.53
3.86
6.00
4.41
3.41
5.05
4.18
5.71
4.71
3.45
4.42
3.36
6.35
4.50
5.78
4.36
5.25
3.16
4.41
4.42
2.43
6.22
2.57
3.20
7.17

a
Thousand million, ° Unless otherwise noted, all data are from United Nations: Yearbook of National Accounts Statistics, 1957 to 1961. c Based on unpublished estimates
of the Division of General Economic Research and Policy of the United Nations, d Estimated by linking a series on G.D.P. at market prices to an index of G-D.P. at factor
cost, e Original data in current prices deflated by an index of wholesale prices of domestic goods, f Gross national product. 8 Estimated by linking G.N.P. for 1950-54 to N.P.P.
for 1954-60. h 1950 figure estimated on basis of G.N.P. for that year, i Estimated by linking indices of N.D.P. and G.D.P. J Estimated by linking G.D.P. 1950-53 in 1950 U.S.
dollars to G.D.P. 1953-60 in 1958 pesos, k From United Nations, Economic Commission for Europe: Economic Survey of Europe in 1961, Part 2, Appendix A. 1 Estimated by
linking G.N.P. 1950-53 to nG.D.P. 1953-60. m Estimated by linking G.N.P. 1950-54 in 1952 prices to G.N.P. 1955-60 in 1955 prices. The series were linked by use of the wholesale
price index for 1954-55.
Estimated from an index of national income 1950-54 from United Nations: Statistics of National Income and Expenditure, Series A, No. 9, and G.N.P.
1954-60. Final figures represent G.N.P. for fiscal year beginning April 1. o Estimated by deflating G.N.P. given in the source in current prices by the I.L.O. consumer price index.
P Estimated by linking G.D.P. 1950-52 in 1950 U.S. dollars to G.D.P. index for later years.

>
ta

S
C»

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

TABLE 2. RATES OF GROWTH OF THE ECONOMICALLY
ACTIVE POPULATION, 1950-60

Country

Algeria
Argentina
Australia
Austria
Belgium
Brazil
Canada
Ceylon
Chile
China (Taiwan} . . .
Colombia
Costa Rica
Cyprus
Denmark
Ecuador
Finland
France
Germany (Fed. Rep.)
Greece
Guatemala
Honduras
Iceland
Ireland
Israel
Italy
Jamaica
Japan
Korea (South) . . .
Luxembourg . . . .
Malaya (Fed. of) . .
Malta
Mauritius
Mexico
Netherlands . . . .
New Zealand . . . .
Nigeria
Norway

Initial
year

End
year

Economically active
population a
(in thousands)
Initial
year

End
year

(1)

(2)

(3)

(4)

1950
1950
1950
1951
1950
1950
1950
1950
1952
1950
1951
1950
1956
1950
1950
1950
1950
1950
1951
1950
1950
1950
1951
1951
1950
1953
1950
1955
1950
1950
1954
1950
1950
1949
1951
1950
1949

1960
1960
1960
1959
1960
1960
1960
1960
1960
1960
1960
1960
1961
1960
1961
1960
1960
1959
1961
1960
1960
1960
1959
1960
1959
1959
1960
1960
1959
1960
1960
1960
1960
1959
1960
1960
1959

3,326
6,979
3,435
3,347
3,545
17,117
5,086
2,841
2,188
2,438
3,756
278 b
265
1,920
1,237
1,984
19,032
21,580
2,839
968
509 b
64
1,272
505
18,455
615
36,347
8,053
137
1,920
79 d

4,188
8,144
4,215
3,649
3,615
23,364
6,391
3,662
2,356
3,344
4,720
362 b
271
2,115
1,701
2,135
20,265
24,940
3,663
1,306
621 b
78
1,169
736
20,340
719
46,945 c
9,527
151
2,384
85 d
219
11,645
4,340
892
16,809
1,538

153
8,345
3,855
740
14,913
1,489

Annual
rate
of increase
(per cent.)

(5)

2.30
1.54
2.05
1.09
0.19
3.10
2.52
2.54
0.93
3.16
2.53
2.65
0.45
0.97
3.19
0.73
0.64
1.60
2.55
2.99
1.99
1.98
-1.06
4.19
1.08
2.62
2.55
3.36
1.08
1.96
1.37
3.59
3.34
1.31
2.07
1.71
0.33

APPENDICES
TABLE 2 (conci.)

Country

Panama . . . .
Peru
Philippines . . .
Portugal . . . .
Puerto Rico . .
South Africa . .
Spain
Sweden . . . .
Switzerland . .
Thailand . . . .
Tunisia
. . . .
Turkey
. . . .
United Kingdom
United States
Venezuela . . .

Initial
year

End
year

Economically active
population a
(in thousands)

Initial
year

End
year

Annual
rate
of increase
(per cent.)

(1)

(2)

(3)

(4)

(3)

1950
1950
1948
1950
1950
1951
1950
1950
1950
1950
1950
1950
1949
1950
1950

1960
1959
1959
1958
1960
1960
1958
1960
1960
1959
1959
1960
1959
1960
1960

266
3,072
7,416
3,005
597
4,592
11,838
3,105
2,156
9,540
1,244
10,725
23,339
60,037
1,706

337
3,894
9,708
3,159
632
5,712
12,666
3,266
2,514
11,531

2.37
2.54
2.45
0.62
0.56
2.43
0.85
0.51
1.54
2.10
1.60
2.33
0.57
1.97
2.87

1,437

13,550
24,714
73,126
2,274

a Unless otherwise indicated, data are from IX.O. : Year Book of Labour Statistics, and United
Nations: Demographic Yearbook. Where the data were not available for the desired years, estimates
were made as follows: to the total population statistics appearing in the Demographic Yearbook for the
desired years were applied the percentage of the total population economically active for the nearest
available year, b Source: United Nations: Human Resources of Central America, Panama and Mexico,
¡950-1980 (1960). c Estimated from 1959 by using the index of the civilian labour force employed for
1959-60. d Civilian labour force employed.

THE QUALITY OF LABOUR AND ECONOMIC DEVELOPMENT

»1N»-iino«1t»O*-"00l;NNrt>Cvej'l»
•

•

ÇH

•

• m GO

* CO

jMjj

^oot~oo\ooovooooot-ooooBoo»ooot~ooBCooCooC"C

• CO v u

II
000\NCIO\

ca a
B C t ~ B o o fi C

•£B

o o o o o o o o r - o o s o o o o o r - o o o o » ON \o oo M »

3 IS

so^v^v^^^woTtrn^cnincoviTtviinvNTfr

SIS

! ! *

j O

^
go

JrtOO

G T Í G v*> »n G

i n rn q o^ oo Tf « N q ^ ^ H » h; q rt ^ N N j
N r i wi « r í o rtV' ^ <n "o ^ V ' oí i n o\ NO r i vo . ;

j o ON
-• m' m

ON
ON

t^ooONaN©oo^enfncioo"vioocnr^vÍ©<^r^^«©vÍfnoo*
r*t^vOh>voNosòr"t**Nuvor^NDh'Vìr»fsh'\o«nt^ts'h'r^No

m N ( S O O O O O N ' - i n O
ONONONONONONONONONONONON

¡s
.§•8

s ï23lsaiSSSSS3g3Sgi335s:S
. . _ _> so \ o v> NÖ so v i S S S S ?î S S S S S
i
t
i
j
i
i
i
i
i
t
i
j
i
i
i
j
t
i
i
m íS (N O O O O O O <S rn O ^ O O O O *-i <*
ini««niovìJnvì»nio'Avi'o<n'nvì«o"/i<nifì

S?
S• vï!
j
O
m

O «n
<o«o

ONONONONOSONONONONONONOSONONONOSONONON

ON ON

J3
^1 " ì " Ì **î **:
oó © v i ^r v i —¡
- ^ vo v i Q" ^t " i <n oí r i v i oó ^ r*Í oo j m J r i ^ «;
m v> v i <n -it ^
-<$•• v > T f ^ - ^ - f n v ì < s v ì T r i n v ì i o m G •<$• C i n m C

,.^ S V Ç S S I O..N O S V O V Ç S S V O V O S V Ç S
V
i

íSSSS
i

* - ( ^ H r J O O ^ O O O ^ f O O O O O O O O O O O r 4 0 ' -| « 0
»nvì«rt«o<oiri«nvìinvìVìiniriwì<nin«oiovìv><n'ovi 0'o
ON ON ON ON OS ON OS ON ON ON ONOs ONOsOsONON.OSON; Os; O N O s O N ON ON
OMOo<Nvor~«-<fCT\0;OCT\«rtmrovOfsoot--vo<riTto\t~:r-vo « oo • * t-' vo vo ^H o\ v i od CJÑ ori o\ m' oÑ o •< ori r i T Í Ö t—' ori r i
<s fs <s <s » i —i ts n
n « !-> ^H H « fs <s fs 1 1 —i «•> *i t s n
<§!

) « < * K

C J P f i l S O ß

O S ' S fi D . .

Jamaica . .
Japan . . .
Korea (South)
Luxembourg
Malaya (Fed. o 0
Malta . . .
Mauritius . .
Mexico . . .
Netherlands .
New Zealand
Nigeria . . .
Norway . .
Panama . .
Peru . . . .
Philippines .
Portugal . .
Puerto Rico .
South Africa
Spain . . .
Sweden . . .
Switzerland .
Thailand . .
Tunisia . . .
Turkey . . .
United Kingdom
United States . . .

18.4
25.3
12.4
23.2
8.6
22.8
15.2
14.5 b
24.8
23.7
10.2
31.3
12.5
24.3
8.1
16.0
21.1
23.3
16.3
22.1
24.9
15.0
12.8
14.5
16.3
18.3
26.2

1950-60
1950-60
1953-60
1950-59
1955-59
1954-60
1950-60
1952-60
1950-60
1952-60
1952-57
1950-60
1950-58
1950-58
1950-60
1950-60
1950-60
1952-59
1949-58
1950-60
1954-59
1952-59
1950-59 c
1950-60
1950-60
1950-60
1950-59

52.7
43.8
32.8
48.1
n.a.
55.8
53.5
n.a.
50.2
52.1
n.a.
52.1
61.1
29.9
39.4
n.a.
65.6
n.a.
58.5
63.1
57.5
n.a.
n.a.
n.a.
66.9
62.4
n.a.

1953-60
1950-60
1953-60
1950-59
1954-60
1950-59
1950-60
1952-60
1950-60
1952-58
1950-58
1950-60
1950-60
1954-57
1950-60
1954-59

1950-60
1950-60

72.4
81.5
86.0
70.2
n.a.
86.2
69.5
n.a.
81.4
75.5
n.a.
62.8
84.7
57.3
89.5
n.a.
85.5
n.a.
80.7
75.4
74.8
n.a.
n.a.
n.a.
75.7
74.4
n.a.

1953-60
1950-60
1953-60
1950-59
1954-60
1950-59
1950-60
1952-60
1950-60
1950-58
1950-58
1950-60
1950-60
1954-57
1952-60
1954-59

1950-60
1950-60

65.6
70.3
70.1
61.7
67.9
80.2
63.7
67.9
72.2
68.0
70.1
58.3
77.1
41.2
70.2
64.2
81.9
68.0
75.2
71.9
69.5
70.1
70.1
67.9
73.4
71.2
70.3

54.1
39.4
28.1
50.5
n.a.
56.2
50.6
n.a.
49.8
50.3
n.a.
49.5
59.9
30.1
40.5
n.a.
63.5
n.a.
58.3
58.3
57.0
n.a.
n.a.
n.a.
66.9
59.2
n.a.

52.0
44.5
36.8
49.6
n.a.
57.3
56.3
n.a.
51.8
52.9
n.a.
55.8
62.0
31.0
38.6
n.a.
61.1
n.a.
53.3
66.4
58.0
n.a.
n.a.
n.a.
68.4
64.2
n.a.

77.9
82.3
87.2
75.5
n.a.
84.6
69.6
n.a.
80.0
76.2
n.a.
61.3
86.9
62.1
92.1
n.a.
87.5
n.a.
81.0
74.8
75.1
n.a.
n.a.
n.a.
77.3
73.0
n.a.

68.4
77.6
84.6
70.5
n.a.
87.0
69.6
n.a.
82.4
75.3
n.a.
66.3
84.4
55.7
87.7
n.a.
77.5
n.a.
76.3
75.9
75.0
n.a.
n.a.
n.a.
76.4
74.7
n.a.

a Except where otherwise noted, this is the ratio of gross fixed capital formation to gross domestic product at factor cost, both in current prices. The ratios were computed
for years between 1950 and 1960 for which data were available, and averaged for the period. All data are from the United Nations: Yearbook of National Accounts Statistics, 1957
to 196Ì. b Based on G.N.P. rather than G.D.P. c Source: unpublished estimates of the United Nations Secretariat, d Except where otherwise noted, this is the ratio of compensation of employees to gross domestic product at factor cost, both in current prices. The ratios were computed for years between 1950 and 1960 for which data were available, and
averaged for the period. All data are from the United Nations: Yearbook of National Accounts Statistics, 1957 to 1961. e Except where otherwise noted, this is the ratio of compensation of employees plus income from unincorporated enterprises to gross domestic product at factor cost, both in current prices. Where data on income from unincorporated enterprises were unavailable, income from property and entrepreneurship was used. Same source as note <*. f See text for an explanation of the manner in which the average was derived.

>
hi
tu
Z

a
O

SO
SO

TABLE 4. INDICATORS OF THE DEVELOPMENT OF EDUCATION, 1950-60
Primary school
enrolment
Country

Years

Secondary school
enrolment

Vocational school
enrolment

Higher school
enrolment

Per 1,000
population
aged 5 to
14 years

Annual rate
of increase f

Per 1,000
population
aged IS to
19 years

Annual rate
of increase f

Per 1,000
population
aged 15 to
19 years

Annual rate
of increase f

Per 1,000
population
aged 20 to
24 years

Annual rate
of increase t"

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

1950
1954

158.6
185.8

3.9

54.4
64.9

4.4

13.0
23.6

1950
1960

128.7 a
a

140.3

0.9 a

0.8 a

11.1aa
13.5

2.0 a

Australia . . .

1950
1960

784.7
780.9

0.0

486.4
720.5

3.9

80.6
94.9 '

2.3

65.0
105.8

4.9

Austria

1951
1959

799.4
728.7

-1.2

139.8
140.6

0.1

256.5
385.5

5.1

46.7
76.1

6.1

Belgium....

1949
1959

691.5
696.4

0.1

202.1
405.7

7.0

366.2 i
532.1 k

4.7

Brazil

. . . .

1950
1960

72.9 a
108.8 a

4.0 a

Canada

. . . .

1950
1959

855.3
883.6

0.4

Algeria

. . . .

Argentina . . .

Ceylon

. . . .

Chile
China (Taiwan)

b
b

8.9

7.9'
12.3'
376.0
608.4

1950
1954

535.3
567.0

1950
1960

134.3 aa
155.0

1.4 a

1950
1960

479.3
658.3

3.2

1.5

8.2 aa

13.4 a
24.3 a
96.9
274.7

4.4'

2.0 a
3.1 a

5.3

32.9
86.0

4.5
2.6

5.9'
10.4

6.3 a
6.5 a
41.8
88.0

14.9

4.3'

8.4
6.5
4.8
7.7

a
a

g
4.7

30.8 J

71.2

1.0 a
1.3 a

9.3
2.6 a

67.0
72.9

0.9

-13.7

3.6
5.4

10.1

7.4

•<

-6.6

10.7

0.3'

o
C

1.6 a
2.6 a
8.0
39.7

4.9 a
16.0

r
>
w
O

g
>

§
ta
o
§
S
O

'5
z
H

Colombia

1950
1960

74.1 a
119.6 a

4.8'

6.6

Costa Rica . . .' . .

1950
1960

494.8
664.1

3.0

12.0

Cyprus

1951
1958

630.4
662.0

0.7

6.7

Denmark

1950
1959

584.7
666.3

1.5

2.9

Ecuador

1950
1960

106.7 a
a

Finland

1950
1959

703.9
680.9

-0.4

6.9

1950
1959

711.1
761.5

0.8

France

131.3

1.8 a
5.7 a

11.5

1.0 aa
1.6

4.7

n.a.

20.0
36.7

6.1

5.0
8.0

6.7

4.8
8.6

8.3

339.9
374.6

1.1

46.0
42.0

n.a.
n.a.

-1.0
a
a

4.4 a

1.3
1.8

92.5
163.8

6.4

43.9
62.4

3.9

8.5

83.2
200.4

9.8

40.8
76.4

7.0

-1.2

1.6

438.1
478.2

1.0

31.1
51.8

5.7

0.5

3.9

16.3

16.3
29.8

8.6

2.4'

6.7 a

2.1'

4.5 a

2.9
4.5 a

3.3

Germany (Fed. Rep.)

1950
1959

767.5
699.0

Greece

1951
1958

646.6
670.5

Guatemala

1950
1959

59.1 a
73.0

Honduras

1950
1960

71.0 a
a

112.0

Iceland

1950
1958

622.0
608.4

-0.3

5.9

328.6
227.6

-4.6

50.4
60.6

2.3

Ireland

1950
1955

843.7
884.1

0.9

3.1

64.3
93.4

7.5

37.3
43.9

3.3

(For footnotes see p. 104.)

4.6'

18.9 a

32.1
73.1 •
3.2 a
0.7 a>c
2.1a
a
2.1

-19.0

0.0 a

0.8 a
0.9 a
0.6
0.7

a
a

1.3
1.5

(Table continued overleaf)

TABLE 4 (cont.)
Primary school
enrolment
Country

Years

Israel
Italy
Jamaica

Japan
Korea (South)
Luxembourg

Malaya (Federation of) .
Malta
Mauritius
Mexico

Vocational school
enrolment

Higher school
enrolment

Per 1,000
population
aged 5 to
14 years

Annual rate
of increase f

Per 1,000
population
aged IS to
19 years

Annual rate
of increase f

Per 1,000
population
aged IS to
19 years

Annual rate
of increase f

Per 1,000
population
aged 20 to
24 years

Annual rate
of increase *

(1)

(2)

(3)

(4)

(S)

(6)

(7)

(8)

1950
1959

855.0
804.3

-0.6

140.7
243.1

6.1

90.4
99.2

1.0

29.0
41.0

3.8

1950
1958

576.2
565.1

-0.2

132.5
219.7

6.3

125.6
245.6

8.4

37.8
39.4

0.5

1950
1955

664.7 d
669.1

0.1

55.1
58.2

1.1

12.3
11.5

-1.3

1.2
3.3

20.2

1950
1960

614.1
623.9

0.2

792.5
985.8

2.2

55.6
145.3

9.6

50.6
78.6

4.4

2.1 a

182.7
261.3

4.0

23.3
42.7

6.7

20.1
42.2

8.2

0.6

126.6
228.3

6.6

147.5
192.2

2.9

4.3
4.0

-0.8

3.4 a

5.1
22.0

14.6

0.4
1.1

10.1

0.04
0.3

20.2

130.9
158.2

a
a

730.0
773.1

1950
1960

113.9
159.9

a
a

1950
1959

605.9
770.8

2.7

72.1
282.7

15.2

12.9
73.9

19.4

10.0
16.5

5.6

1950
1960

n.a.
n.a.

n.a.

n.a.
n.a.

n.a.

4.0
4.2

0.0

0.7
1.5

7.6

1950
1960

103.2 aa

1950
1959
. . . .

Secondary school
enrolment

1950
1959

139.9

3.0 a

3.0
7.5

a
a

9.2 a

0.9
1.4 a

4.4 a

1.1aa

2.5

8.2 a

Netherlands .
New Zealand
Nigeria . . .
Norway
Panama
Peru

. .
. .

. . . .

Philippines .
Portugal . .
Puerto Rico .

1950
1960

706.4
638.0

-1.0

5.8

1951
1960

903.3
861.7

-0.5

3.6«

Sweden . . .
Switzerland .

(For footnotes see p. 104.)

6.5

g
g

g
a

36.7
50.3

3.2

81.9
105.4

2.5

0.01 a
a

1950
1960

79.1

1951
1959

681.9
773.9

1.6

1.0

201.6
199.3

-0.1

26.6
40.9

1950
1960

573.5
593.7

0.4

7.7

138.3
88.1

-5.0

25.3
38.1

1951
1959

123.5 aa

140.7

1.6

1950
1959

702.1
611.0

-1.7

0.6

17.0
40.5

10.9

126.1
115.4

—1.1

1951
1959

396.4
514.9

3.3

6.9

49.2
106.3

9.6

20.9
27.7

4.0

1950
1960

603.6
813.3

3.0

g
g

69.3
167.0 h

8.8 h

n.a.
n.a.

n.a.

n.a.
n.a.

n.a.

n.a.
n.a.

n.a.

South Africa
Spain . . . .

31.4 a
a

166.3
317.2

1950
1959

99.8
129.9

1950
1959
1948
1960

a
a

9.2 a

15.5 a

0.04
0.2 a

16.1 a

7.7

- 1 . 9 g, h

n.a.

2.9 a

6.7 a

632.4 e
723.0

1.3

1.6

675.4
692.6

0.2

1.7

2.0
3.8 "»"

5.5
6.3

a
a

309.6
307.8
64.4
83.3 *

8.0 m

1.5

0.06

1.6
2.5

2.7
2.5

17.9 a
5.4
4.6

a
a

5.6 a

-0.9

0.0

36.8
72.2

7.5

3.2

47.2
57.7

2.0

(Table concluded overleaf.)

TABLE 4 (conci.)
Primary school
enrolment
Country

Years

Per 1,000
population
aged 5 to
14 years
(1)

Thailand

United Kingdom
United States

1951
1959

. . .

579.2
560.2

Secondary school
enrolment
Per 1,000
population
aged 15 to
19 years

Annual rate
of increase f

Per 1,000
population
aged 15 to
19 years

Annual rate
of increase f

Per 1,000
population
aged 20 to
24 years

Annual rate
of increase f

(2)

(3)

(4)

(5)

(6)

(7)

(8)

74.1
192.4

11.9
a
a

7.8 a

6.2
11.0

328.0
352.8

1.5

37.5
70.8

1950
1959

690.0
640.2

-0.8

732.4 8
1023.2 «

1949
1960

852.5
855.7

0.0

1950
1960

100.3 aa
171.9

1950
1959

44.2
89.5

1950
1955

Higher school
enrolment

Annual rate
of increase f

-0.4
a
a

Vocational school
enrolment

5.4

6.4 a
12.7

12.1
36.0
3.1a
a
3.5

15.4
30.6

3.7 8

g
g

600.2«
775.3 «

2.3 8

g
g

3.7 aa
14.1

13.4 a

1.2 aa
5.7

19.2
28.8

13.6
1.3 a

0.5
0.6

a
a

2.0 a

r
=
o

13.7

4.8

24.6
38.6

5.0

177.2
298.5

4.7

15.6

1.3 a
3.5 a

5.1

12.6
16.0

9.9 a

Source: Population data are from United Nations: Demographic Yearbook. School enrolment data are from United Nations: Statistical Yearbook; and U.N.E.S.C.O.: World
Survey of Education. Detailed footnotes to the data in these sources are not reproduced here. International comparison of the absolute ratios is hazardous because of differences
in definition and coverage among countries.
a For these countries the appropriate age distribution of the populations was not available. The ratios represent school enrolment to total population. These ratios are not
comparable with the ratios not so noted, b Primary and secondary enrolment taken together, c 1951. d 1959., © 1949. f Annual compounded rate between the years indicated.
g Secondary and vocational enrolment combined, h The reliability of this estimate is questionable, i 1957. J 1950. k 1958. 1 1956. m This figure may include groups not
included in the earlier years and thus overstate the rate of increase.

'§
o
s
a
ö
M
r

'S
H

TABLE 5. INDICATORS OF HEALTH, 1950-60
Inhabitants per physician

Year

Number

Rate of
change
during
period

Year

Number
per 1,000
inhabitants

Rate of
change
during
period

Year

Daily
per head

Rate of
change
during
period

Year

Deaths per
1,000 live
births

Rate of
change
during
period

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

1953
1959

5,300
5,800

1.5

1950
1959

2.6
3.1

2.1

n.a.
n.a.

n.a.

1950
1959

86.2
117.9

3.5

1952
1960

1,300
660

-8.1

1952
1960

6.4
6.3

-0.2

1952
1958

2,980
3,090

0.1

1950
1959

68.2
59.1

-1.6

1950
1960

1,100
860

-2.4

1952
1960

11.2
11.7

0.6

1952
1959-60

3,220
3,260

0.2

1950
1960

24.5
20.2

-1.9

1949
1960

650
620

-0.4

1949
1960

8.2
10.5

2.2

1952
1959-60

2,700
2,950

1.2

1950
1960

66.1
37.5

-5.7

1950
1959

1,060
800

-3.1

1950
1958

3.4
7.7

10.2

1952
1959-60

2,950
2,930

-0.1

1950
1960

53.4
30.6

-5.6

1949
1958

2,700
2,100

-2.8

1949
1958

3.2
3.7

1.6

1951-52
1957

2,410
2,640

1.6

1950
1960

109.1
70.1

-4.4

1950
1960

900
900

0.0

1948
1959

10.1
11.4

1.1

1952
1959-60

3,050
3,150

0.4

1950
1960

41.5
27.3

-4.2

1951
1959

6,000
4,500

-3.5

1951
1959

2.2
3.5

5.8

1952-53
1960

1,990
2,150

1.0

1950
1959

81.6
57.5

-3.9

1951
1960

1,800
1,700

-0.6

n.a.
n.a.

n.a.

1951-52
1957

2,430
2,570

1.0

1950
1960

139.4
127.9

-0.8

Country

Algeria

. . . .

Argentina . . .
Australia
Austria
Belgium

Canada
Ceylon
Chile

. .
. . . .
. . . .

. . . .
. . . .

Infant mortality

Calories available

Hospital beds

(Table continued overleaf.)

TABLE 5 (coni.)
Hospital beds

Inhabitants per physician
Country

China (Taiwan) .
Colombia . . .
Costa Rica
Cyprus
Denmark
Ecuador
Finland
France

. .

. . . .
. . .
. . . .
. . . .
. . . .

Germany
(Fed. Rep.) . .

Calories available

Infant mortality

Year

Number

Rate of
change
during
period

Year

Number
per 1,000
inhabitants

Rate of
change
during
period

Year

Daily
per head

Rate of
change
during
period

Year

Deaths per
1,000 live
births

Rate of
change
during
period

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

01)

(12)

1951
1960

2,400
1,500

-5.2

n.a.
n.a.

n.a.

1952
1959

2,140
2,310

1.1

1950
1959

35.3
30.5

-1.6

1952
1960

2,800
2,400

-1.9

1952
1959

2.6
3.1

2.5

1948-49
1957

2,370
2,170

-1.0

1950
1960

123.9
99.8

-2.2

1951
1960

3,200
2,600

-2.3

1951
1959

5.0
5.1

0.2

n.a.
n.a.

n.a.

1950
1960

91.3
80.3

-1.3

1949
1960

1,300
1,400

0.7

1949
1958

2.6
4.8

6.8

1948-49
1955

2,500
2,590

0.6

1950
1960

63.4
29.9

-7.5

1949
1959

1,000
830

-1.9

1949
1958

10.6
10.4

-0.2

1949
1959-60

3,240
3,340

0.3

1950
1959

30.7
,22.5

-3.4

1946
1960

4,000
2,600

-3.1

1953
1959

2.3
2.1

-0.2

1955
1958

2,170
2,230

0.9

1950
1958

109.7
105.8

-0.4

1950
1959

2,000
1,600

-2.4

1950
1959

7.5
9.0

2.0

1949-50
1958-59

2,980
3,120

0.5

1950
1960

43.5
21.0

-7.3

1951
1958

1,100
930

-2.4

1951
1959

14.6
14.6

0.0

1949
1959-60

2,800
2,940

0.5

1950
1960

52.0
27.4

-6.4

1952
1959

750
730

-0.4

1951
1959

10.6
10.7

0.1

1949
1960-61

2,730
2,940

0.7

1950
1960

55.6
33.8

-5.0

Greece

. . . .

Guatemala . . .
Honduras
Iceland
Ireland

. . .
. . . .

1,100
800

-3.5

1951
1959

8.0

1951
1957

5,800
6,400

1.6

1951
1958

1.7

1951-52
1957

6,500
4,800

-5.5

1948
1959

1.9

800
840

0.5

1950
1958

0.9

n.a.

1951
1959

4.0

1950
1958

. . . .

Israel
Italy
Jamaica

1951
1960

. . . .

Japan
Korea (South) .
Luxembourg . .
Malaya (Fed. of)

n.a.
n.a.

2,490
2,900

1.5

1951
1960

43.6
40.1

-0.9

n.a.
n.a.

n.a.

1950
1960

106.8
91.9

-1.5

2,200
2,200

0.0

1950
1960

85.6
52.0

-5.0

n.a.
n.a.

n.a.

1950
1960

21.7
13.3

-4.9

1949
1959

3,430
3,570

0.4

1950
1960

46.2
29.3

-4.6

1949
1959

1949
1954-55

1950
1960

380
400

0.5

1951
1959

3.0

1950-51
1958

2,680
2,780

0.3

1950
1960

47.3
30.8

-4.3

1951
1960

820
610

-3.3

1951
1958

3.3

1949
1960-61

2,350
2,740

1.3

1950
1960

63.8
43.8

-3.7

1949
1959

4,000
4,300

0.7

1949
1959

-0.7

n.a.
n.a.

n.a.

1950
1960

78.3
51.0

-4.3

1952
1959

1,000
930

-1.0

1951
1959

8.2

1,900
2,210

1.7

1950
1960

60.1
30.7

-6.7

1954
1958

4,200
2,400

-14.0

n.a.

n.a.
n.a.

n.a.

1951
1960

1,200
910

-3.1

n.a.

n.a.
n.a.

n.a.

1950
1960

45.7
31.5

-3.7

1952
1960

10,000
6,400

-5.6

-2.2

n.a.
n.a.

n.a.

1950
1960

101.6
68.?

-3.9

1950
1959

1949
1959

n.a.
n.a.

>
•tí

n.a.

o
(Table continued overleaf.)

TABLE 5 (conci.)
Hospital beds

Inhabitants per physician

Mauritius . . .
Mexico

. . . .

Netherlands . .

New Zealand
Nigeria
Norway
Panama
Peru

. . . .
. . . .
. . . .

O
oo

Infant mortality

Year

Number

Rate of
change
during
period

Year

Number
per 1,000
inhabitants

Rate of
change
during
period

Year

Daily
per head

Rate of
change
during
period

Year

Deaths per
1,000 live
births

Rate of
change
during
period

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

1950
1960

1,100
980

-1.1

1950
1959

10.5
9.4

n.a.
n.a.

n.a.

1950
1960

88.5
38.3

-8.4

1952
1960

5,500
4,500

-2.5

1951
1959

4.5
4.7

0.5

1955-56
1960

2,290
2,350

0.6

1950
1960

76.3
69.5

-0.9

1947
1960

2,200
1,700

-2.0

1947
1956

1.2
1.5

n.a.

1955
1958

2,390
2,330

-0.2

1950
1960

96.2
75.1

-2.4

1949
1959

1,250
900

-3.3

1949
1959

5.0
7.9

4.6

1949
1959-60

2,930
2,970

0.1

1950
1960

25.2
16.5

-4.2

1951
1960

800
700

-1.5

1951
1959

13.0
11.7

-1.3

1949
1959

3,360
3,450

0.3

1950
1960

27.6
22.6

-2.0

1949
1960

88,000
32,000

-9.2

1951
1959

0.4
0.5

2.5

n.a.
n.a.

n.a.

1950
1960

86.3
62.9

-3.2

1949
1959

1,000
900

-1.0

1949
1959

9.0
10.4

1.4

-0.4

1950
1959

28.2
18.7

-4.6

1950
1960

3,300
3,200

-0.3

1950
1959

4.1
3.9

-0.6

n.a.

1950
1960

68.4
57.0

-1.8

1952
1960

4,500
2,100

-9.5

1952
1959

2.0
2.2

1.4

—0.6

1950
1960

103.7
103.4

0.0

Country

Malta

Calories available

-1.2

1949
1959-60

3,100
2,980
n.a.
n.a.

1952
1959

2,070
1,980

n.a.
n.a.

Philippines . . .
Portugal

. . . .

Puerto Rico

. .

South Africa . .
Spain
Sweden

. . . .

Switzerland. . .
Thailand
Tunisia
Turkey

. . .
. . . .
. . . .

United Kingdom
United States

Venezuela . . .

n.a.

n.a.
n.a.

n.a.

1952-53
1958

1,940
2,100

1.5

1950
1960

101.7
73.1

-3.3

1949
1960

2,320
2,420

0.4

1950
1960

94.1
77.5

-1.9

n.a.
n.a.

n.a.

1950
1960

67.5
44.4

-4.2

1950
1960

1,500
1,300

-1.4

1951
1959

4.0
5.2

3.3

1950
1959

2,600
2,200

-1.9

1950
1959

5.2
5.4

0.4

1950
1960

2,200
2,000

-1.0

1950
1959

3.5
6.2

6.4

1951
1960

1,000
1,000

0.0

1949
1959

4.2
3.2

1950
1959

1,400
1,100

-2.7

1950
1959

1950
1960

700
740

0.6

1954
1960

6,800
7,500

1950
1959

1949
1959

2,640
2,580

-0.3

1950
1960

134.3
125.5

-0.7

-2.7

1952-53
1959-60

2,490
2,750

1.4

1950
1960

69.8
43.5

-4.7

11.3
15.2

3.2

1949
1960-61

3,110
2,930

-0.6

1950
1960

21.0
16.6

-2.4

1950
1956

14.5
13.6

-1.1

1949
1959-60

3,170
2,980

-0.6

1950
1960

31.2
21.1

—3.9

1.6

1954
1959

0.6
1.0

8.5

n.a.
n.a.

n.a.

1950
1959

62.4
47.1

-3.1

6,000
8,200

3.4

1950
1959

1.7
2.7

5.1

n.a.
n.a.

n.a.

n.a.
n.a.

n.a.

1951
1960

3,200
2,800

-1.4

1950
1959

0.9
1.7

7.5

1949
1958-59

2,510
2,850

1.3

n.a.
n.a.

n.a.

1951
1960

1,200
960

-2.4

1951
1959

10.1
10.8

0.8

1949
1959-60

3,130
3,290

0.5

1950
1960

30.0
21.8

-3.2

1950
1961

750
780

0.4

1950
1959

9.5
9.1

-0.5

1949
1960

3,180
3,120

-0.2

1950
1960

29.2
25.6

-1.3

1950
1960

2,200
1,300

-5.3

1950
1960

3.6
3.5

-0.3

1952-53
1959

2,010
2,300

2.0

1950
1960

80.6
45.1

-5.8

Sources: All data are from World Health Organization: Annual Epidemiological and Vital Statistics; and United Nations : Statistical Yearbook.
1960 were the starting and terminal years. When data wsre not available, the nearest available years were substituted.

Wherever possible, 1950 and

TABLE 6. INDICATORS OF HOUSING AND SOCIAL SECURITY, 1950-60

Dwellings completed
per 1,000 inhabitants

Ratio of investment in
dwellings to G.N.P.

Social security benefits paid
per head population0 aged
15 to 64 years

Ratio of social security benefits
paid to national income

Country
Years
covered

Average for
period

Years
covered

Average for
period
(per cent.)

Years
covered

Rate of growth
for period
(per cent.)

Years
covered

Average for
period
(per cent.)

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Australia
Austria
Belgium

1951-60
1950-60
1950-61

Canada
Ceylon
Chile
China (Taiwan)
Colombia

1950-60
. . .

Cyprus
Denmark
Finland
France
Germany (Fed. Rep.) .
Greece
Guatemala
Honduras

1950-61
1950-61
1950-61
1950-61
1950-59.
1950-60

n.a.
n.a.
8.5
5.6
5.0
n.a.
7.1
n.a.
n.a.
n.a.
n.a.
n.a.
n.a.
5.2
n.a.
7.3
4.7
10.0
6.6
n.a.
n.a.
7.3

1950-60
1950-60
1950-60
1957-60
1953-60
1950-60
1950-60
1950-60
1950-60
1954-60
1950-61
1950-58
1952-60

n.a.
n.a.
n.a.
4.3
4.3
n.a.
4.6
n.a.
n.a.
1.9
n.a.
n.a.
5.6
2.9
1.8
6.3
4.0
5.1
4.8
n.a.
3.4
9.0

1949-57
1949-57
1949-57
1949-57
1950-57
1950-57
1950-57

1949-57
1949-57
1949-57
1949-57
1950-57
1950-57

n.a.
n.a.
4.2
7.8
4.6
n.a.
5.5
6.4
3.4
25.2
n.a.
n.a.
n.a.
5.3
n.a.
8.4
10.1
10.9
n.a.
11.6
n.a.
5.4

1949-57
1949-57
1950-57
1949-57
1951-57
1950-57
1955-57

1949-57
1955-57
1949-57
1949-57
1949-57
1951-57
1952-57

n.a.
n.a.
7.9
16.0
14.5
n.a.
7.8
3.3
7.6
0.8
n.a.
n.a.
n.a.
10.2
1.4
9.5
16.4
18.3
n.a.
2.5
n.a.
7.6

APPENDICES

oo <S i-¡ c¿ O) cd Os cd ed ed ed' so Os ed «S t - ed C d O ed «
CO WS <S £ T f
tj os <s d ° ó ^ B C « d - *

d^; « e e

f-> 00 ("«n<n«n

00
in

^ t ^ •<*•
0\OM>

wON

i-H i-H i—(

r^ r-. O

i-H

6
m

r~ f i n »n
OS OS

t ~ so
»n i n
os >o

r~
in

r")
m
Os

•<t
Os

—
i<

i-i

*—(

*-l

cd o> ed **" os Cd ed ed «N " 1 ed • * ed* ed e d M

i—i

f - oo r >n<n<n
Os o \ os
-*t -st TJOs OS Os
*—< »-* *-H

—
i

oc
w
*—i
tr
O
*—

ed N

t ~ r~
m in

rin

• *

•*•*
i—1 i—1

•*
1—1

o\
1—*

1—
m
r*
TdOs

1-H

t~ f*
i n »n

»0 r - oo r „OS i-i OS
f^-rt m -^
v^ Os Os Os

•*•*
Os Os

i—i

e r- e e e d t > ~ * ¿ t - ß e O » ¿ m

tn
os

cd oo Os cd cN en <s Os cd
d o s o e i n «-i i-i - ^ a

1—11—1

ed —" o ce r - ^ p
B so t~
^(sH

—I

<n ed*

ví

cin

t~- r i n »n

r - t^- oo r^
m i n »n «n

rf
Os

•*•*
Os Os

• n - ^ i n <o
Os Os Os Os

•-H 1-H t—( 1-H

.O
o o s o o o i - H O s O O N r ^ s o t - * (9 cs - < c i so ed ed c i O so ed ed' »-! ~ ! cd ed 1 ^ O m m
Q m f<í r i r n
( S o o ^ - n * - N P 1 * H co m t j Tf m co T f q c

« " • * d o«"«'» e

o
Q Os Os
so »o in so
ri*
O M ¿ v ¿ O f O M > o
in»n»nin»n<n»n>ntnv-ì
O O ^ H O O

.-< Os t ^ —1

OSOSOSONOSOSONOSONOS
»-H i—1 ^ H r-4 y— 1 — 1 * - < 1 - 1 i - H 1—1

O O O
so so so

so m in so

m m m
Os Os Os
i—«

0 \ H O Q

m so so so

so so

<¿ r¡< d>

¿ci

<i
d> ò oò
i n m <o «n

m i n
Os Os

Os Os Os Os

1—1

«
^ cd « cd o cd cd «S Cd Cd eä I-; cd cd «N cd Ci cdOs_ pj 1-1 00 Os »-i cd ed ed ^ ^*¡ ed
M f l ' ^ d c e o e e e e so e e«» a c d " e* i—i o r-* t-" e

a ri^°° ri

soo—i.m so so so

so i n

d> ÒÓÓÓ
sn * n i n i n

ó
¿
in in

i—i

ó
•n

ó
in

1—4

i-H

Os Os Os Os

*—1

1—1 1—1 1—1 1—1

OS OS
1—1

• -Q
. . o
. . e
o

s ••
•S'a

1
E

.& B

Si
Ml PH

Puerto Rico
South Africa
Snain . . .

1 1g ì

Norwav .

Korea (South
Luxembourg
Malaya (Fede
Malta . . .
Mauritius .
Mexico . .
Netherlands
New Zealand
Nigeria . .

~ :!

1-*

•

•o

c
ed

e13

•o

31 1 '5
x/ixn

United Kingd
United States
Venezuela .

*-t

•2 h^

TABLE 7. VALUES OF Z VARIABLES USED IN THE REGRESSION ANALYSIS a
Country and group

Income group I
Australia
Canada
New Zealand . . . .
Sweden
Switzerland
United States . . . .
Income group II
Belgium
Denmark
Finland
France
Germany (Fed. Rep.)
Iceland
Israel
Luxembourg . . . .
Netherlands . . . .
Norway
United Kingdom . .
Venezuela
Income group III
Argentina
Austria
Chile
Cyprus
Ireland
Italy
Malta
Puerto Rico
. . . .
South Africa . . . .

Variable

(1)

(2)

(3)

3.89
3.36
6.32
3.88
4.81
2.53

0.416
0.403
0.349
0.307
0.358
0.257

0.96
1.09
1.47
1.23
1.04
0.51

5.00
3.32
5.33
5.03
8.14
7.67
6.52
5.17
4.61
5.59
2.93
7.33

0.291
0.264
0.394
0.285
0.337
0.432
0.389
0.376
0.343
0.537
0.222
0.373

1.05
0.77
1.42
1.45
2.09
2.07
1.44
1.13
1.17
2.31
0.49
1.60

0.57
6.22
2.28
3.31
3.36
7.50
4.13
6.50
3.99

0.270
0.304
0.135
0.272
0.236
0.334
0.284
0.258
0.343

-0.68
1.67
0.04
0.87
0.71
1.57
0.94
1.44
0.65

Country and group

Variable
Z.

Variable
Z,

(2)

(3)

4.54
5.52
8.83
10.01
6.88
4.09
3.36
6.13
6.83

0.264
0.275
0.280
0.360
0.127
0.214
0.162
0.217
0.214

0.79
1.15
1.54
2.70
0.51
0.37
0.42
0.95
0.83

Income group V
Algeria . . . .
Brazil . . . .
Ceylon . . . .
Colombia . .
Ecuador . . .
Guatemala . .
Honduras . .
Mauritius . .
Peru
Philippines . .
Portugal . . .

4.00
4.89
2.34
4.16
5.10
4.56
3.13
1.76
5.61
6.60
6.39

0.334
0.244
0.156
0.264
0.232
0.196
0.202
0.239
0.599
0.115
0.249

2.20
0.83
0.05
0.66
0.58
0.37
0.65
0.22
0.69
0.51
0.97

Income group VI
China (Taiwan)
Korea (South)
Nigeria . . .
Thailand . . .
Tunisia . . .

7.40
3.10
5.01
4.21
1.87

0.256
0.177
0.146
0.214
0.183

0.72
0.87
0.44
0.50
0.15

Income group IV
Costa Rica . . .
Greece
Jamaica
. . . .
Japan
Malaya (Fed. of)
Mexico
. . . .
Panama
. . . .
Spain
Turkey

» For definitions of these variables see the text.
Source: Appendix n , tables 1-3 for Z„ and Z t . Z, is obtained as the product of Z, and an estimate of the rate of growth of real wages. This latter is estimated as the rate
of growth of labour productivity plus the rate of growth of the minimum wage share as implied by the data of columns 8 and 9 of Appendix II, table 3. If these wage share data are
not available, then the growth rate of real wages is estimated under the assumption that the wage share is constant.

TABLE 8
LEAST SQUARES ESTIMATES O F VARIOUS
REGRESSION EQUATIONS — FIRST VERSION
OF THE MODEL
AND
TABLE 9
LEAST SQUARES ESTIMATES OF VARIOUS
REGRESSION EQUATIONS — SECOND VERSION
OF THE MODEL

TABLE 8. LEAST SQUARES ESTIMATES OF VARIOUS REGRESSION EQUATIONS—FIRST VERSION OF THE MODEL

Income group

Regression Number Coemcient
of deter- Variable
equation
of
•Zi
number countries mination
R>
(1)

I and II
Ill and IV . . . .
V and VI
. . . .
All groups . . . .

I and II
Ill to VI
All groups

All groups

. . . .

1
2
3
4

5
6
7

(2)

18
18
16
52

16
16
32

(3)

0.20
0.11
0.05
0.09

0.42
0.72
0.57

0.21

(4)

9.86
11.77
3.38
5.97

10.25
16.60
13.25

8.63

Standard
error

(5)

(4.87)
(8.48)
(4.06)
(2.69)

(2.91)
(3.06)
(1.75)

LABOUR QUALITY VARIABLES

(7)

Constant

Standard
error

1.60
2.24
3.51
3.20

(1.75)
(2.21)
(1.06)
(0.80)

ßi

Standard
error

0.12
0.35
0.21

(0.29)
(0.17)
(0.11)

ßs

Standard
error

ß6

Standard
error

(0.24)

0.70

(0.35)

Standard
error

Q<¡

Standard
error

(5.50) —0.16
Qi

I and II
Ill to VI
All groups

. . . .

9
10
11

17
11
28

0.48
0.59
0.34

7.44
20.30
9.41

(2.38)
(5.61)
(2.60)

0.14
0.10
0.22
ß»

I and II
Ill and TV

. . . .

12
13

15
15

0.39
0.33

9.60
15.91

(1.98) -0.15
(4.55) -0.37

(8)

(9)

(6)

(10)

(12)

(13)

'S>
r

«;

E
ß2

0.04
-0.10
0.01

0.14
(0.11)
(0.10) -0.07
(0.06)
0.06
Standard
error

ßio

(0.28)
(0.77)

0.33
0.13

Standard
error

(0.17)
(0.12)
(0.08)

1.16
1.90
1.71

Standard
error

0.29
(0.88)
(0.59)
0.03
(0.38) -0.07

Standard
error

(0.27)
(0.18)
(0.13)

n
o

o•z
o
o
w

(0.06)
(0.13)
(0.07)
Standard
error

o
e
>
z

§
Qu

(0.16) -0.17
0.00
(0.22)

Standard
error

ßi2

Standard
error

(0.11)
(0.12)

0.18
0.14

(0.12)
(0.11)

V and VI .
All groups . . . .

All groups
All groups
I and II .
m to VI .
All groups
I and II .
Ill to VI .
All groups
I and II .
Ill and IV
V and VI
Ill to VI .
All groups

I and II .
niandIV
V and VI
I l l to VI .
All groups

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

14
15

16
17
18
19
20
21
22
23
24
25
26
27
28

29
30
31
32
33

14
44

21
28
17
19
36
16
20
36
16
12
10
22
38

16
12
10
22
38

0.55
0.07

0.65
0.49
0.45
0.30
0.09
0.43
0.60
0.47
0.33
0.58
0.05
0.39
0.37

0.38
0.59
0.16
0.41
0.39

12.30
12.11

9.66
13.36
7.80
28.27
15.87
11.09
11.87
12.50
13.09
13.35
13.67
13.57
13.23

8.54
16.29
6.20
10.64
9.51

(3.96)
(1.50)

(3.43)
(3.43)
(3.34)
(5.78)
(3.65)
(1.67)
(2.21)
(1.32)
(1.03)
(2.66)
(2.38)
(1.62)
(0.92)

(4.55)
(8.47)
(6.11)
(4.24)
(2.79)

-0.08
-0.15

(0.25)
(0.16)

0.05
0.08

(0.15)
(0.08)

0.06
0.04

(0.08)
(0.05)

0.14
0.17

(0.12)
(0.06)

QÌ

Standard
error

Ö12

Standard
error

2.27
2.11

(0.48)
(0.42)

0.04

(0.08)

1.08
2.61
2.07
1.31
1.91
1.81
1.85
1.76

(0.56)
(0.48)
(0.35)
(0.56)
(0.67)
(0.68)
(0.43)
(0.32)

Constant

Standard
error

(1.62)
(2.30)
(1.20)
(1.31)
(0.93)

1.24
1.95
0.96
1.65
1.53

(0.57)
(0.70)
(0.92)
(0.51)
(0.36)

1.66
-0.85
2.63
0.98
1.32

0.13
-0.03
0.36
-0.75
-0.13

(0.22)
(0.23)
(0.22)
(0.41)
(0.25)

0.11
0.02
0.16
0.14
0.19
0.17
-0.08
-0.02

(0.10)
(0.08)
(0.10)
(0.07)
(0.05)
(0.11)
(0.09)
(0.07)

>
m
O

Source: Appendix II, tables 1-6.
For an explanation of variable Zx, see the text. The remaining variables are as follows :
Qt, inhabitants per physician.
Qi , social security benefits paid per head.
Qi, hospital beds per 1,000 inhabitants.
ß s , social security benefits as a proportion of G.N.P.
0 3 > calories per head.
Q9 , primary educational enrolment.
QA , infant mortality.
Qio secondary educational enrolment.
Qs • dwellings completed per head.
O n , vocational educational enrolment.
06 » investment in dwellings.
612 , higher educational enrolment.

TABLE 9. LEAST SQUARES ESTIMATES OF VARIOUS REGRESSION EQUATIONS—SECOND VERSION OF THE MODEL

Income group

I and II .
Ill and IV
V and V I .
All groups

All groups
All groups
I and II .
Ill to VI .
All groups
I and II .
Ill to VI .
All groups
I and II .
Ill and IV
V and V I .
Ill to VI .
All groups

RegresCoeffision Number cient of
StandStandequa- of coun- determi- Variable ard
Variable ard
tion
nation
error
error
tries
Zx
2,
number
(l)

(2)

(3)

34
35
36
37

18
18
16
52

0.61
0.58

38
39
40
41
42
43
44
45
46
47
48
49
50

21
28
17
19
36
16
20
36
16
12
10
22
38

0.15

0.76
0.63
0.69
0.45
0.40
0.70
0.74
0.68
0.67
0.84
0.57
0.66
0.58

(4)

(5)

(6)

(7)

3.12

(2.66)
(2.74)
(3.40)
(1.84)

3.00
2.79
1.40
2.20

(0.69)
(0.63)
(1.14)
(0.48)

9.67
12.36
8.63

1.12
3.26
0.08
13.93
3.07
3.33
8.20
6.61
3.67
8.75
6.03
9.18
7.92

(4.35)
(4.44)
(3.52)
(8.87)
(4.40)
(2.70)
(2.20)
(1.64)
(2.68)
(2.15)
(2.91)
(1.66)
(1.45)

1.77
2.02
2.50
2.21
2.79
2.44
1.76
1.81
2.64
2.05
4.61
2.15
1.80

(0.67)
(0.67)
(0.77)
(1.10)
(0.70)
(0.75)
(0.59)
(0.39)
(0.72)
(0.55)
(1.45)
(0.55)
(0.42)

LABOUR QUALITY VARIABLES

a
(8)

(9)

(10)

G»

Standard
error

ß.

1.73
1.78

(0.46)
(0.38)

0.82
1.68
1.59
0.92
0.96
1.02
1.09
1.34

(0.43)
(0.50)
(0.29)
(0.42)
(0.51)
(0.53)
(0.38)
(0.28)

0.26
0.14
0.29
-0.38
0.01

(ID

(12)

(13)

(14)

(15)

Standard
error

Q»

Standard
error

0.04

(0.07)

0.13
0.06
0.08
0.15
0.18
0.09
-0.01
0.03

(0.09)
(0.07)
(0.08)
(0.06)
(0.04)
(0.09)
(0.08)
(0.06)

(0.19)
(0.20)
(0.17)
(0.41)
(0.21)

Z
Source: Appendix II, tables 1-6.
For an explanation of variables Zi and Z, see the text. Variable Q, represents calories available per head; Q, , investment in dwellings as a percentage of gross national
product; Q,, social security benefits paid per head; and Qlt, higher educational enrolment.

APPENDIX III
FIGURES

Key to the countries appearing in the figures
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

Algeria.
Argentina.
Australia.
Austria.
Belgium.
Brazil.
Canada.
Ceylon.
Chile.
China (Taiwan).
Colombia.
Costa Rica.
Cyprus.
Denmark.
Ecuador.
Federation of Malaya.
Finland.
France.
Germany (Fed. Rep.).
Greece.
Guatemala.
Honduras.
Iceland.
Ireland.
Israel.
Italy.

27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52

Jamaica.
Japan.
Korea (South).
Luxembourg.
Malta.
Mauritius.
Mexico.
Netherlands.
New Zealand.
Nigeria.
Norway.
Panama.
Peru.
Philippines.
Puerto Rico.
Portugal.
South Africa.
Spain.
Sweden.
Switzerland.
Thailand.
Tunisia.
Turkey.
United Kingdom.
United States.
Venezuela.

FIGURE 1. ANNUAL RATE OF GROWTH OF GROSS DOMESTIC PRODUCT
VERSUS ANNUAL RATE OF GROWTH OF LABOUR FORCE, 1950-60

FIGURE 2. ANNUAL RATE OF GROWTH OF GROSS DOMESTIC PRODUCT
VERSUS THE AVERAGE INVESTMENT RATIO, 1950-60

(Percentages)

(Percentages)
9-

•

•
27

•

52
2i

•

•
1

•

•
•

••
4-

•

••

t B

•

•
4'

•

•
7

•
•

•

17
#

«e

•

224

•

-»

growth per annum

•é°

0
01

0- 1 .5

-1D

-Q5

0Ì5

1. 0

Labour force rate of growth per annum

2.5

3j0

315

Investment ratio

FIGURE 3. CONTRIBUTION OF CAPITAL TO ANNUAL RATE OF GROWTH OF
OUTPUT VERSUS THE INVESTMENT RATIO, 1950-60
(Percentages)

•
•
<qjs

•

• £.<

«2

è
Â

•

-21_J5-

ür-h

• •

1 3 *

è «
•

Investment ratio

10 " ¡ 2

ÎÎT" 20

FIGURE 4. THE DIMENSIONS OF ECONOMIC GROWTH, 1950-60
(Percentages)

W /

¿
/ r '
3T
E

<io y

Incremental f,
25'•r o
Immediate rate %
-7-

-yÇf -?-

l'y

21^
JUL
19^

17-^6

@©' r

15'

13-

1 ©

®
11-

®
D-

9--12.5

7- ,r

--25
3-

110

Investment ratio

Rate of growth of gross domestic product.

Rate of growth of labour productivity.

Rate of growth of G.D.P. attributable
to capital.

Rate of growth of economically active
labour force.

FIGURE 5. KEY TO FIGURE 4

Rate of growth
( % per annum )

I , _L_ (ICOR)

AY
Y
(rate of growth
of output )
AY/wlN AL
Y "IPY; L
AY - _AL_
Y
L
(rate of growth of
productivity or
output per head )
ICOR : Incremental Capital Output Ratio.

_I
(alternative ICOR(U)
PAY-wAL

r/pY
(ICOR(U)
AY-AL
Y
L

* Investment Ratio (7o)

FIGURE 6.

THE EXPLANATION ACHIEVED BY THE REGRESSION 16
_ 30

NOTE.
to capital
the same
estimated

The tail of each arrow refers to the rate of growth of output attributable
when labour is measured as numbers of men. The tip of each arrow refers to
variable after an allowance Is made for the growth in the quality of labour as
In regression 16.

10
37

Investment ratio

FIGURE 7. RATE OF GROWTH OF CALORIES PER HEAD VERSUS RATE OF GROWTH OF REAL WAGES

-*-

1.5

1.0
*
0.5-

*
*
*

•0.5-

>s -1.0-

Rate of growth of real wages

*
Countries in income groups I and II

Countries in income groups III to VI

FIGURE 8. RATE OF GROWTH OF HIGHER EDUCATION ENROLMENT RATIO VERSUS
RATE OF GROWTH OF PRODUCTIVITY
22.5-

o
o
15

-*-

*
;:

«
ft

* !; * 0

o
o

O
-7.5
-0.5

Rate of growth of productivity
Countries in income groups III and IV
Countries in income groups I and II

Countries in income groups V and VI