Provisional

WEP 2-23

WORLD EMPLOYMENT PROGRAMME

Research and Action Prolamine concerning

Income Distribution and Employment

Income Distribution at Different Levels of Development

A Survey of Evidence

Pelix Paukert

International Labour Office

Geneva
March 1973

Oy,.
43602

Income Distribution at Different Levels of Development

- A Survey of Evidence
by Felix Paukert
The Changing View of Income Distribution
Throughout human histor3" the idea of equality kept appearing,
in different contexts and with different contents. The idea of
equality has not "been purely, or even predominantly., an economic
concept. The notion of economic equality appears in history
accompanied often "by notions of. equality "before law, "by the .
notion of political equality (e.g. the principle of "one man,
one vote"), and by the ideas of social equality (which, for
example, resulted in a protest against slavery).
In spite of the long standing interest in the idea of
economic equality, there has not been any long-term tradition
of systematic work on the concept of income equality, or
on size distribution of income. The theory of distribution
which has formed one of the main fields of economic theory
since Adam Smith, and even before .him, has been primarily
concerned with distribution among factors of production -r
in the form of wages, profit, rent and interest."" Only
occasionally was this analysis extended to include distribution
of income by size.
The ideal of economic equality in the late 18th and early
19th century did not find its main expression in the works of
economists (as represented ~by the line Adam Smith -.Ricardo
John S. Mill) but. rather on the borderline of economics and

Economists before Adam Smith, particularly Quesnay and

Turgot, were also interested in distribution. Their theories
were, formulated primarily in terms of classes rather than
factors of production. See H. Dalton, The Inequality/ of
Incomes, London 1320, pp. 37-33.

- 2 -

politics, in the work of people like St. Simon, Fourier and

Robert Owen.

Influenced as they were by the events of the French

Revolution they felt a need for a new organisation of human

society, for new principles of distribution of fruits of human
effort.

None of them, however, advocated a full equality of

income.

St. Simon put forward an equity, rather than equality,

criterion of distribution: to each according to his work.

Fourier, on the other hand, advocated a system of distribution
"under which a comfortable minimum income would be secured to
every person, including women, and children, five
years old, and the residue would then be divided between labour,
capital and "talent" in the ratio of 5 54:3M

Incomplete as

this rule is in its distribution between the need and the merit
criterion, and arbitrary as it is in its distribution of the
merit element, it foreshadows much of the later thinking in its
stress on minimum living standards without forgetting the
incentive effects of merit rewards.
The classical economists and the Utopian socialists, in
their separate ways, set a tone which has continued ever since,
in spite of elaboration of economic theory and increased social
awareness - on the .one hand the absence of a workable, or even
theoretically- feasible, definition of "equality of income", on
the other the acceptance of economic equality as a desirable
goal without, however, striving for an absolute equalitsr of
income. Ho economist of note, or sociologist, has advocated an
absolutely equal distribution of income among human beings. For
the nearest expression "to this ideal we have-;probably to go back
to the French Revolution, to Robespierre, who said that no one
ought to have much more or much less than 3>C00 francs a year.

Dalton, op cit., p. 51

- 3 -

This rough equality was to "be the result of public opinion, of

the natural ethics of the community and not of organised control.""
During most of the 19th century a good deal of work" was done
on classification of reasons for unequal distribution of income.
John Stuart Mill made an advance upon Adam Smith's treatment of
difference of wages, drawing attention to the absence of.
competition as an important factor for inequality/ of remuneration.Even more important was his analysis of the consequences of
inheritance, of wealth. .But there was' little advance in the
analjrsis of economic and welfare effects of income inequality.
This advance came 0113.37 towards the end of the 19th century,
with the arrival of the marginal!st school.

One.theorem of the

marginal!st theory, the so-called "law of decreasing marginal

2
utility", has clear equalitarian implications.
The idea of
decreasing marginal utility was itself not new. It was contained
already in "Bentham's Introduction to the Principle's of Morals and
1egisiatjon (1789) in the following form; "The greater the
quantity of the matter of property a man is already in possession
of, the less is the quantity of happiness he receives by the
addition of another quantity of the matter of property, to a
given amount". But the economists of the marginalist school
have managed to make out of this idea one of the cornerstones
of economic theory. Ever since, in spite of later theoretical
disputes about the possibility to make interpersonal comparisons0
it has been generally accepted that, with a given volume of
production, total economic welfare is increased with more equal
distribution.
""Crane Brinton, Equality, in Encyclopedia of Social Sciences,
New York 1931, volume 5, p 579.
2

Brinton, p. 579

JiA. .Schumpeter., ., History of Economic Analysis, London,

1954, p. 888..

Tile qualification "with a given volume of production" is

however an important one. There are some arguments, that more
equalitarian distribution has some unfavourable effects on
production because of its influence on savings. The argument that
redistribution of income from the rich (who were assumed to save
part of their income) to the poor (who were assumed to spend all
their income) will-unfavourably affect saving influenced thinking
about equality in income distribution in the 19th and well into
the 20th centur3r. Then the argument received an unexpected twist,
i n Then Creneral Theory of
ff-3ploymj3nt,_Ij^^
by
J.15. Keynes. In the words of Joseph A. Schumpeter; "It must not
be forgotten that he [Keynes] rendered a decisive service to
equalitarianism in an all-important point. Economists with an
equalitarian bent had long before learned to discount all other
aspects or function of inequality of income except one? like
J.S. Mill they had retained scruples concerning the effects of
equalitarian policies upon saving. Keynes freed them from these
scruples" .
Keynes1 argument had an important impact on economic
thinking about income equality. He turned the most telling
argument ag_ain_st income equality into the most important economic
argument .for income equality, transferring the cardinal sin into
the principal virtue. He accepted the view that richer people
save proportionately more than poor people (expressing it in an
elegant way of "marginal propensity to save being positively
correlated with income"),, but he regarded'reduced saving accompanied by increased consumption - as the principal method
of increasing production in an economy working below its capacity.
Income redistribution towards equality thus in such conditions
resulted in increased total income.
The development after the Second '.Vorld v7ar turned attention
to the situation in developing countries where Keynesian theory

Schumpeter, oj^clt^

P 1171.

-5-

is hardly applicable because the problem is not so much the lack

of effective demand as the lack of productive capacity.

The

concern with productive capacity and therefore with investment and

savings in developing countries has thus brought back the old
anti-egalitarian arguments based on the assumed lower propensity
to save of the lower income groups. A policy leading towards
more equal distribution of income was thus thought incompatible
with policy leading to economic growth.
A reaction against'this view came only in the 1960s,
primarily with the dissatisfaction with using the criterion of
growth of national income as the only indicator of economic
development. Intellectual climate changed sufficiently in the
last decade for most people concerned with development to admit
the desirability/ of more equal income distribution even if more
equal distribution means reduction in the rate: of economic.growth.
The growing emphasis on income distribution was accompanied by
increased interest in employment creation as part of the .development process as employment is regarded as the most effective. ;..
means of changing income distribution in a developing society.
The complex interrelationship between income distribution
employment, and economic growth is the subject of a research
programme, undertaken within the context of the Wor3.d' Employment
Programme which is intended to examine the determination of income
distribution by employment as well as by other factors,- and, on
the other hand, the impact of changes on income distribution on
employment. The programme also aims at examining the role of
government.in determination of primary income distribution and in
redistribution of income.
There are a number of different concepts of income distri^bution which could be examined in relation to employment and
economic development: distribution of personal income by size,
distribution of income, .between different factors of production,

distribution between different regions of a country, distribution

between rural and urban areas, etc., and most of these are to be
covered in the research programme.

In this preliminary study the

object is limited to the examination of the problem of how income

distribution by size changes at different levels of economic
development.
Distribution of Income by Size at Different Stages of
Development
Systematic work in distribution of income by size at
different stages of development is of a very recent origin - it
could even be argued that it starts really only in 1955 with the
classical article "03?- Simon Kuznets on "Economic Growth and .Income
Inequalits?"". . Until then thinking-about changes in income
2
distribution was unhappily influenced by the work of Pareto whose
famous law stated (in Samuelson's simplifying words) that "in all
places and all times, the distribution of income remains the- same.
Neither institutional change nor egalitarian taxation can alter
this fundamental constant of the social sciences" . Pareto's
contribution was the expression in a mathematical form of income
distribution (or, rather, of the upper tail of this distribution)
but his insistence on the constancy of the arithmetical value of
one of the coefficients of the formula - ''Pareto believed that he
had discovered an economic constant comparable in significance to
the gravitational constant in physics" - was a long term source
Of. American Economic Review, SLV (March 1955), pp. 1-28,
2
Vilfredo Pareto, Cours d'economie politique, -Lausanne 1897*
Paul A. Samuelson, "A Fallacy in the Introduction of Pareto's
Law of Alleged Constancy"- of Income Distribution" , in Rivist.a
Internazionale di Scienze Economiche e Commerciali, ITo. 12,. 1965,
p. 246.
A

Martin Bronfenbrenner, Income Distribution Theory, (Chicago

and Hew York, 1971) p t 44. Pareto's form states that by IT = A
- *i by Y, where Y is a level of income and IT the proportion of
income receivers with incomes equal or greater than Y. The value
of A has no economic significance, but the value ofc<is inde
pendant of the choice of units and was considered by Pareto to be
always within the range of 1.5 to 1.7,

- 7 -

of confusion until it was gradually agreed that the coefficient

does not remain constant in different societies and different
periods of history, and moreover, that the coefficient is a very
poor indicator of income distribution by size. Modern work based
mainly on indicators associated with the names of Lorenz and
Gini

began to show more substantive differences in income

distribution between countries.

Kuznet's primary contribution was in the classification of
a number of conceptual issues involved in international and
intertemporal comparisons of income distribution and in posing
and tentatively answering the question about changes in income
distribution in the process of development. The discussion of
the manyfold conceptual issues is outside the scope of this study,
but it should be stated that we are concerned herewith the
primary distribution of income, that is with distribution before
tax, and that we take as the basic income recipient the family
rather than the individual.
Empirical investigation of the relation of income
distribution by size with economic development can be conducted
in two ways. With the first method we can trace the changes' in
income distribution in a country over a long period of time;
with the second method we can compare income distribution in a
number of countries at a different level of development and try
to draw conclusions by cross country analysis. In both cases
we are hampered by lack of data.
Historical Trends in Inequality
The largest compilation of historical data was undertaken
by Kuznets whose 1963 study has indications of size distribution
for the United Kingdom, Prussia, Saxony, Germany, Netherlands,
2
Denmark, Norway, Sweden and the United States . These data are
See Appendix I for a description of these indicators .
2
S. Euznets, Substantive Aspects of the Economic Growth of
Nations: VIII. Distribution of Income by Size, in Economic
Development and Cultural Change, vol XI, No. 2, Part II, January
1963, pp. 1-80. See also S. Kuznets,-Modern Economic Growth,
New Haven, 1966, particularly pp. 206-217.

Table 1
Shares In Rational Income of Ordinal Groups. Tax Units or Consuming Unite. Selected Countries.
Lon i'e:r i o d s
J u c c e s s i v e d a t e * and
U n i t e d Kingdom
1. Dates
Income b e f o r e t a x
2 . Top 5'/
3 . Top 20;;

Rowley

entries

Clark

1913

1929

1936

1947

1933

194t9

4c
56

43
59

33
51

31
52

24
46

29
50

2 3 . ,5
4 7 . .5

- russia

Reich

Troo o p o v i t c h

4. Dates
5 . Top 5/-'
5 . Top 2 0 7
7 . L o w e s t fiC'/

Lydall

Seers

I60O

1054
21

1696
27
45

lo75
26
46
34

16
41.5

S t a t i s t i c a l Office
1926
26
49
31

1913
31
50
32

1913
30
50
33

1957

Tueller
. . Dates
9 . Top %-

1673-60

Saxony
l.cO
34
56
27

Vieich o t a t i s t i c a l

14. Dates
1 5 . Top 5>
l b . T'op 2fjjl .
1 7 . L o w e s t 60;.'

1913
31
50
32

192 6
27
49
31

Netherlands
k . Dates
1 9 . Top 5 j "
2 0 . Top 2C;:
2 1 . L o w e s t SOji'

Office

Dates
Top 'if
Top 1 0 ji
Top 20;i
L o w e s t 60j;

167C
36.5
50

1J03
2-.
.3'-

crvray
27. Dates
2... Top %-, oour; t r y d i i3 t r l c t s
2'J. Top 5>., c i t i 08

1936
2..
53
26.5

192 0
20

1936
23

193'.:
19
'.9
31

1949
17
4 5.5
34

1954
13
36.5
40
Z e u t h e n 13f
192';i
25
37
53
25

190.
30
39
55
31

1907
27
2 6-32

1936
20
22

1950
24
4u
29

Office
192 6
26
50
31

U n i t e d :U i t i o n s

1926(adj.)
21
45
34

1923
2.,
36

b t a i; i s t i c a l

1213
33
54
26

r.ueller

Zeuthen I

Doriiaark

o c h e n b e r i c h t
1955
lo
43
34

Harned i n c o m e
3 : . Dates
3 1 . Top %
3 2 . Top 20,. :
3 3 . . L o w e s t 60si

1930
30
59
19

3'-.
35.

Dates
.'op 5;-.
3 ' L Top 20>.:
3 7 . L o w e s t 60jS

1935
26
5S
23

1939
24.5
35
51
27

1949
19
2J.5
45
32

194 6
14
19

United Nations
1946
1945
23.5
20
51
47
26
29

194 5
24
52
23

1946
20
45
32

1954
17
43
34

iCuznets

States

3~. Dates
3 3 . Top I;;
\<J. Top 5;'

1935
23
56
69

1913-19
14
24(1917-19)

1919-2o
14
25

1929-3
13
25

*_

1944-46
9
17

1939 - 4 3
11
21

D e p a r t m e n t o f Commerce
41. Dates
u . Top 5;:
4 3 . Top 20?=
4 4 . l o w e s t 60?;
Uource:

1929
30
54
26

1'5 3 5 - 3 6
26.5
52
27

Kuanete , " Q u a n t i t a t i v e Aspects 1 VIII",

1941
24
49
29
op. o i t .

1944-47
21
46
32
Table 16,

..'ote: Data relating to after tax IncomeB have been omitted.

1950-54
21
45

p p . 60 f f .

1959
16
43
34

Ujerke

.ientzei

oweden

United

1912
33
55
27

162'J
36
57
2 5.5

ijermavi.y--.iest Geimany

1911-13
31

Heich

i'rooopovitoh

1 0 . Dater.
1 1 . Top 5/
1 2 . Top 26>--'
1 3 . L o w e s t 60';.

22.
23.
2:.
25.
26.

1901-10
32

1091-190C
32

1661-930

2<J

33

1955-59
20
45
32

1955
17.5
27.4
4-i
32

9 -

reproduced in Table 1.
The data are of course very heterogeneous
and in most cases give only a partial picture of personal income
distribution, as they mostly concentrate on the share of the
richest strata of population.
Nevertheless, they allow some
generalisations.
For the period through the post-Second World War years, there
is a perceptible narrowing in inequality in the size distribution
of income if judged by the declines in the share of upper ordinal
groups, less marked if judged by the rise in the shares of the
lower ordinal groups.
In most countries, the share of the top
5 per cent group in income before taxes was 20 per cent or less in
the post-Second World War years.
In the 1920s or the 1930s
the share of the top 5 per cent group in income before taxes
was about 30 per cent, in some countries above' and in others a bit
below.
Likewise the share of the top 20 per cent gr.oup in income
before taxes in p.ost-^Second World War years was b.etween 40 and
45 per.cent; whereas in the 1920s and the 1930s it was well above
50 per cent.
The evidence on the share of the lowest 60 per cent
group is much more scanty, but there is some indication that.it.
was below 30 per cent in the 1920s and the 1930s and rose to well
above' 30 per cent In the postr-Second World War years. But.
according to the evidence, the rise in the share of the lower
brackets was less conspicuous than the decline in the shares
of the upper groups.
The data in Table 1 do not give a clear answer to the
question when the trend towards equality in developed countries
started. Only in Denmark there is a clear reduction of inequality
between 1870 and the beginning of this century indicated by a
sharply reduced share of the top 5 and 10 per cent groups of the
richest income state.
In other countries - to the extent to which
we can judge from Table 1 - the reduction of inequality started

Kuznets (1963), op. cit., p. 59.

- 10
after the First World War or even during the Second World War.
ICuznets' study in the Economic Development and Cultural Change
(1353) is the only one where the historical experience of income
distribution changes in a number of countries is 'analjrsed.' There
are, however, several studies where changes in income distribution
are traced for one country for a length of time.
The study which covers the longest period is that of Lee Soltow
on Great Britain". Work on income distribution using present day
statistics is a rather daring undertaking, and large-scale international comparisons are positively hereic, but Soltow's work on
Britain surpasses other efforts. By -using various pieces of
information Soltow manages to construct Lorens curves for some years
very long ago, and by using partial information he calculates at
least Pareto coefficients'for upper income-groups going even
further back, so that he makes the first (very tentative) -estimate
for 1436. Summarised.,; his picture indicates some but not great
reduction of inequality between 1436 &nd 1688, and then no change
between 1688 and 1801. There is some evidence that in 1867 and 1880
inequality was somewhat smaller tha.n in 1801 but other evidence
"indicates no change during that period. The next period, up to 1911
and 1913 gives a similar picture with basically no change but with
a possibility of a slight reduction in income inequality. After
1913 up to 1962-3 there is a sharp reduction in inequality, and this
reduction is of course documented also in other studies.
The conclusions of Soltow's study of Great Britain, tentative
as they are, are interesting in that the;- indicate a long period of
more or less equal degree of inequality, interrupted perhaps by one
or two periods of moderate reduction of inequality, with no
indication of increased inequality. This period comes to the end
with the First World War after which there follows a strong trend

Lee Soltow, "Long-Pam Changes in British Income Inequality",

in .The E c onomic Kistorv_Keyij3W, Vol. XII, 'No. 1, April 1968,
pp. 17-29,

- 11... ,.-

towards equality, 'some short-term reversals notwithstanding.

number of authors examined the changes in British income distribution since the Second World War.

There is almost a unanimity that

income after the war was distributed more equally than in 1938
There is, however, not such a clear view of the development since
1945. 'Thus Dudley Seers, writirig'in 1956, finds that "there has
been, in recent years, a slight regression towards prewar
1
2
inequalities" . R.J. iMicolson , on the other hand, finds a trend
towards greater equality continuing until 1959 since when it has
been reversed. This finding is-more -or less- confirmed by the
latest available study, that of Thomas Stark who examined data for
1949, T954, 1959 and 1953 with the use of his own specially devised
indicators. Some of- these show constantly decreasing inequality,
but the majority of his indicators show a reduction.of inequality
between 1949 and-.195? with some increase in 1963*

:'

The picture of longr-term experience of Great Britain is not. at

variance with that of the country with the second longest period
for which indicators- of .income equality, are -known - Norway. Here
the data are of quite different character: the unique historical
records of Norway make it possible" to calculate G-ihi a6eifcienta
for eight Norwegian cities for 10 year;intervals between 1890 and

Dudley Seers, "Has the Distribution of Income become more

unequal?" in the .Bulletin- o;f _the_--fi_xfoi'd-University Institute of
StatiiLtij^s, 3IVIII, 1956, pp. '73-86.
;"!.,:.''.
2
'
R.J. Nicholson, "The Distribution of Personal Income", in
Lloyds Bank Review ,' Januar3" ,1967.
"'Thomas Stark, The Distribution, _of Personal Income ini the
United Kingdom, 1949-53, "(Caiiibridge" 1972) , ppf 5"6-7,

- 12 -d
Pi
pi
to
Pi
d

1 1 I

I 1

i n -sj- i n o~i vo m
o cr> rn oo r H
TJ-

m *+- n-) sJ" ^

H
O
cn
H

i-H

i>2
H
Pi
d
H

Pi c
^
<H i n
. \o
co
on
-d H

V>
CA
H
1
O

r-3

CM co

-d-

CH
CD
u

r<~) CM

V.O

in
CO
"3-

m ro t en

co

.^1

<d
P
H
O

-J

fc'.C
H
PI
'd

CM O

t-- H
1 .1 --+ in

00 i n H CM CM -=2H CO r n m v.O CM
"=* r ^ *=f- "tf" =*- m ]

Pi
d
co
<-i

in

CO
d

1 I

H CM =* UD i n CM
1 c cn CM o '> o^

o <-n

H
*0 H r<~)
in in in

CO
CO

co
1

=*-

1 1

CO ^ t O O t - 'O
m c - va CM H co en
r n m C\i ^ -=* CM CM

00CD

H
d
Pi
O
<H

CD
<d

*.D

V..O

C\J

O
1 CO

--=!-

1 o
m

CM [ CM * d - VO C -

en co H rn i n o co
r O r n r n -^d- " d - r n CM

CO

Pi

H
O
H
H
H
CD
O
O
d
P!
H

d
-P
co
M
d
Pi
rd
a)
Pi
P-H

1 1

co -=t-

tr- o\ i.o H ^o -J- en

CM H "=* CM m H co

in -^f-

- d - -=+- r n =}- *

rn CM

ciO
Pi

CO

CM o oo in
1 I en f - c o o-

PH

i n - d - . ^ - -5h

rQ

w*

d O
co c n
H

0)
PI
d

fciO
d
ft

So

CD
H
P=4

CO

CO

tr~
H

CD
CO
O
&
P

id
Pi

Pi
o
tH

t-- CM co ( ir- m ^!- i n c o

in oo co *o v.o H en co
rnCMCMCMm-^hCMCM

PI
a3

9d

Pi
v.O - d
t
CO rcj
1 (~1
r~i ^

H
d

IT^J

r^-H

U
d

>i

ooPi
c3
pq

CD
S
O
o
Pi
d

Pi
o
tH

l-'-Hj

-p
cjcd
H
Pi -P
o
<H >d
Pi
.co a3
d

<*

% O
o cn
Pi H
i n fd
C- Pi
co d .
H
CM
on
r Cj

CD

-P

Pi H
d O
c
d
+3
q .
Pi
d
d
o
d O
C
H O
' H cn
H
o
O rci
Pi
d d
. Pi
d
o
DO c n
Pico
d H
d
-p Pi
CO - d

CO

r\
U

d
p!

Pi

CD
r '

d
d . d
o m in in m in o in o o o o oo o o

^t- ** in <-.o b- oo on on o H CM rn <n m ^o

co co co oo co co co co on en en en en o-\ on

r-! H H H H H rH H H .H H H H r! H

^i

Ed

b
d
o
f11

Pi
d
+3
d
H
d

a
W

r"i
Pi
O
o
Pi
H
-d
Pi
d
t

&H
S

o
P1
H
o
CO

Hi

M 5
CO

Pi
o
CO
d
n3

o
Pi
p5
o
CO.

- 13 -

I960 The data are shown in Table 2 and indicate on the "basis of
data which are much more precise than those for Great Britain (hut
which at the same time indicate income distribution only in limited
areas rather than in a whole country) a very clear trend towards
equality during the period, however with a temporary "but rather
sharp reversal during the depression years of 1930 and 1938. Until
then, there was a relatively clear trend towards equality, with a
particularly pronounced reduction of inequality around the turn of
the century. After the Second World War, the degree of equality
was greater.than ever-before i.e. in six of the eight cities and in
the seventh the I960 equality was as low as ever before.
Data for other .countries do not cover anywhere an equally
long period. Thus for the United States, Irving B. Gravis2 gives
the following data for distribution of family personal income
before tax; All four indicators in the table show a considerable
reduction of inequality between 1929 (or 1935-36) and 1944, followed
however by a stable situation from then until 1958.
Later data confirm the post-war stability of income distributionThe U.S.. Bureau of the Census data of distribution of pre-tax
family incomes even indicate some further levelling, with the
decline in quintennial figures for 1950 until 1970 of the share of
the top 5 per cent of income units- from 17-.0 to 16.8 to 168 to
158 and 14.4 per cent and a similar slight decline from the G-ini

See Lee Soltow, Toward Income Equality in Eorway, MadisonMilwaukee 1955? p 17 This book has to be consulted for the
discussion of the many unique features of the data, for description
of computation, of assumptions on underreporting, etc.
2
IB. Kravis, The Structure of Income.? Some Qualitative Essays,
University of Pennsylvania, 1962. For the period before 1929,
Krayis concludes that there was a phase of narrowing inequality
extending roughly from 1890 to 1920, followed by a decade of
somewhat increasing inequality (p. 214).

- 14

co
c5
co

rc

CQ

-p

o
>

co
m

- !

O
ri

--v

-p

v+

in

i"0

d
H

_9

"=d-

G?

ro

-P
r:"i

Si

o O
o
0)

ti

Si

il
O

M!

CO
CD
H

CD

il

rO

co!

O !.n r O
e-

in

c-

un

o
co

vo

-ir
CM

O
CM

CM

co

CM

oj

c\i

ej

CD
CM

CM

O
U
H

<H
o

CO
H
CM

ko

H1C-\

^o

o
O
CU

0)
-:-' r-l
Q -H

co
3q

i
[-

o -p -->
.ci evi

-1
G

CD

ri

PS
rQ

>
o5

co
H

p
CO
CO

-P

Cj

.i

r! t
CS -1

S
CO

ri

H
un

CO

i:

CO

rn

nrS
>

CD
=}-

Ln

vo

in

in

CD H

t:-i ro

1
co o
H P4
>
C\3
H

r~

c3

5 r~
O rs
:-1 b

cr*.

co

en

co

r-

-:i-

==}-

v-f

m.

=*

-=*

-*.

ri t>

K'
IH

CM

Cn

rn

C'A
r~-\ H

LTJ

!>

H
H
CD
<
CD
ri

c
o
f-l
CD

10

ci
P

cO

en

P i -H
CQ
fc H
c. CD

r-l

H
O

CD
-p ri
(l P

'Ci
CD

'-.O

(M

=j-

<*

UA
C'A

CA
r-!

O'i

G'\
~i

CA

vo
in

en
H

co
m

Q'\

i
CO

CD!

- 15 -

ratio from 0.375 to 0.366 to 0.369 to-0.360 and 0.353

Among the developing 'countries in India there have "been
consistent attempts to measure changes in income distribution.
These attempts were possible "because of regular collection of data
through the national Sample Surveys hut at the same time the nature
of such data makes it imperative to make a number of corrections
which result in somewhat contradictory estimates. In. Table 4 we
.present some of the latest estimates of the Gini concentration
ratio for different years.
The results in changes in the distribution of income (and of
consumer expenditure, in the last two columns of Table 4) do not
support any contention about deterioration of distribution: in the
few cases where there is an increase in the Gini ratio, the change
is very small indeed. On the other hand, some case can be made for
reduction of inequality, particularly on the basis of Ahmed's data
covering 1956-7 to 1964-5* Moreover, as the Summary Report of the
Seminar on Income Distribution.pointed out there was a general
agreement among the Seminar, participants about some reduction of
inequality during the 1960s when measured by consumption
expenditure in current prices. But it is probably not possible to
go any further in generalising about distribution changes in India.
For most other developing countries, if income distribution
are available for more than one year, it is usually for a relatively
short period from which little can be deduced about trends. A good
recent study by Richard Wcisskoff can serve as an example of shortterm movements in income distribution of three Latin American
countries:
Census data quoted by Peter Menle, "Exploring the distribution
of earned income", Monthly Labour Review, December 1972, p. 22.
Such a view is however contested by Edward C. Budd, "Postwar Changes
in the Size Distribution of Income in the U.S.", American Economic
Review, May 1970, who argues that the reduction of the share of the
top 5 per oent was accompanied by a reduction of the share of the
bottom-40 or 50 per cent group and by a gain of the middle and upper
groups. Menle himself presents calculations (as well as references
to some other studies) indicating a slow but persistent trend towards
inequality in the 1958-^1970 period of earned income (i.e. the sum of
wages and salaries and earnings from self-employment). This is of
course outside the scope, of our study which is concerned only with
size distribution of total personal income.

tMt 4
3o EitlwttiofftLalBatlo. for Indi*
Ojha and
Bhatt (1964)
Personal
Income
Households
1951-52
1952-53
1953-54) 0.349
1954-55)
1955-56) 0.341
1956-57)
1957-56
1956-59
1959-60
1960-61
1961-62
1962-63
1963-64
1964-65
1965-66
1966-67
1967-68
1968-69
Sources

Ojha and
Bhatt (1971)
Paraonal
Incoile
Individuals

Ahmad (1971)
Personal
Incorna
Individuale

Banadiv (1971)
Personal
Incoa
Individuals

0.376
0.4527

0.359-O.374
0.399-0.420
0.393-0.419
0.3773.410
0.371-0.391

0.437-0.511

Banadiv (1971)
Conauaption
Kxpenditure
Individale

0.336

0.333

0.432-0.540

0.355-0.378

0.356-0.379

O.320

O.303

0.4136.
0.385

Hanadlv (1971)
Paraenal
Inooma
Household

Swamy (1964)
Consumption
Expenditure
Households

O.366
0.361
0.369
0,390
O.37O
0.407
0.398
O.383
0.385

0-3873

0.308
P. D. OJHA and V. V. BHAT, "Pattern of I.icom Distribution in an..
Underdeveloped Economy A Case Study of I dia", in American Economic
Review. September 1964, P 714.
P. D. OJHA and V. V. BHAT, "Pattern of Ineom Distribution in Indiai
1953-55 to 1961-64", p. 5. Paper presented at the Seminar on Income
Distribution organised b, the Indian Statistical Institute, Hew Delhi,
February 25-26, 1971.
Mahfoos AlttiED, "3ie Distribution of Personal Income iu India 1956-57,
1960-61 and 1964-5", '^abl 6. Paper presented at the Seminar, etc., 1971.
K. E UA1IADIV3, "Pattern of Incorna Distribution in India, 1953-54 to
1959-60", in the Bulletin of Oxford Dnlverelt - Institute of Economice and
Statistics. August 1968, p. 251 (Distribution of I,,come of individuals in
1953-54, 1954-55, 1955-56, 1957-58 and 1959-60. Hanges are due to estimtes
based on different assumptions about savings and tax evasion)
K* R. RAIADIVE, "Distribution of Income-Trends Since Planning",
pp. 16, 17 and 33. Paper presented at the Seminar, etc. 1971
SUBBAMAKIAH SWAMY, "Structural Changes and the Distribution of Income
by S I M I The Case of India", p. 173 The Bvlw of Income and Wealth.
June 1967

- 17 CM r o
^
H O ro
"=d- < - vri-

O
H -H
P
H t

ci) ft

U5 H r o
CM L a *3-

i n e n La
. . '

. .
O

,c;
-P

ft
O
La

r o CM
CM CM

CJ CO - J "
.

O CO

T- H CTv
CM r o CM

O
f-o
- t r o OJ

ai

O - P H
^ r| 50
s t H
CD M
S: O
H

S rd
o s o
fH

CD

4-' H
CO -H
O -P

o5

ce: '.o

o c--'-o

CO ---J- ' O

G'\ O
La

CA r-l CA
L 'O La

O "--- H
1.a l a L'A

H-

G)

M
d

i!rl
O CH
o O
S ,iJH
M m
CQ

CD

CD

CQ
CD
H

iI

O
o
G
M

r-

CO Lf\

CA H
H CM
P"

^.O "3- r o

CO CA

co vo cr-l r-l

ft

in r- en

H H

fr C5>

H!

P -3 d

CD
O

Ml

cai
U

1
H
,.o
t
EH

CD
P-i
<H
O

O
CO

a)
H
r-l

H a
H -H
EH

CM

H. r-l

CTi <M CM

JO

H
is
-P
CO
H

O
CD
H

o cyi
rH

r O CM r>~)
H r-l H

H
H

-P

T\ C M

^Olf\"i

-H
p'
a?.

en en

CA O

CMO^CO
OD

-0 y o

r.o

CD

CD
N
ri

CO

-d
s-1
o
o
CD
a

r O LA rH

O
H

ri

IT3

O
M

-P
CQ
CD
>
o
i-3

U
t
CD

CD
H
:r!
-P
G
-H
p
O"

S <H H
O O CD
H
-P ''"
p' S !--., P -H ,Q
H >
fH 5 r d
4 3 rrj CD
-P
00
H O t
-i, P
CD

..1 ; H r^
O T\
t
O
O
fi
M O CQ
O
- -H T I
H M -P
C
H CD t
O US H
,.y
"ce r d - H
ca d
cH t -H
CD
Ci
'S: 05

S 1
r d -H
f-H ^P
t Ci CM
,_Oq C-'DiO rHo
ri r-l
r*4

VO La

L a CA O

La

r ro
La >JD

en en

r-l

vo' =r

r o en

La in vo
CA CA CTi
H H
H

ro

!H

O O
-H L-e n 0)
r d r-: - p
S3
o o
- p - P U H
Si H O
H
l l ) P CD
P' p: ivi
O f-M CD O
r-l
O o
C S 0)
O -H R H

C-- r O

La La vo
en en en
II !\ I1

PH

o
o

0
O

rei

.3

-p

JO

H
CD
pi
^4

U
P*
o
co

o5

CD

t
PH

CD

r,

O
O
H

rH

6 o ~ t

CH

IPs -si-

?-

P-. a

P -^ -

-18,-...

The figures for the three countries do not offer a clear cut picture,
Judging by Gini ratios, income distribution in Argentina and Mexico
became more unequal during the 1950s . and then reversed towards
equality in the 1960s but remained more unequal than at the
beginning.

In Puerto Rico there was some increase in inequality.

The picture of increasing inequality is further confirmed by the

sharply declining share of the lowest 40.per cent (the first two
quintiles), particularly in the case of Mexico and Puerto Rico.
Even if the lowest SO per cent of families are taken together,
their share of income declined in all three countries

At the

same time the share of the top 5 per cent of families declined
in Puerto Rico and Mexico

and the gainers were thus the upper

middle income group - the third of families between the 60th and
95th percentile.
International Comparisons of Size Distribution
Changes in income distribution by size in the course of
economic development can be traced not only by historical statistics
of those few countries for which such data are available but also
by comparing income distribution of countries at a different level
of development. This procedure has, of course, some drawbacks as
differences in distribution cannot.be ascribed only.to differences
in development (as measured by per capita GDP) but also to numerous
other factors by which countries differ between themselves. But it
is a method generall.y accepted in other branches of economics and
rather forced on us by.the*scarcity and quality of historical data
even for countries with the best statistical information. Most of
the present day discussion of income differences is actually
concerned with international rather than with intertemporal
comparisons.

The extent of this decline in Mexico is however probably

overstated. See Weisskoff, op. cit., p. 315

- 19 -

The classical Kuznets' study of 1955 compared incomplete data

for India, Ceylon and Puerto Rico with those for the United
Kingdom and the United States and concluded that income in
developing countries was distributed more unequally than in
developed countries<> Kuznets also concluded (and here he based
his conclusions much, more on speculation than on empirical data)
that this inequality is due primarily to high concentration of
income in the top group, with the lowest group in developing
countries getting a somewhat higher share than in developed countries,
hut with the middle group getting a much smaller share than in
developed countries*
2
A study by Kravis based on a greater amount of empirical
evidence confirmed Kuznets' hypothesis of greater inequality in
developing countries0 He was able to calculate quintiles and Gini
ratios for ten developed and developing countries in the early
1950s and has found the distribution in Denmark, Netherlands and
Israel to be less unequal than in the United States, than in Great
Britain, Japan and Canada about the same, and that of Italy, Puerto
Rico, Cejrlon and El Salvador to be more unequal. He was therefore
able to conclude that the degree of equality tends to be positively
correlated with the level of per capita income but that the
correlation was not a simple one. Kravis was also able to confirm
Kuznets1 statement that the share of lowest income groups in the
poor countries tends to be higher than in the developed countries.
The greater inequality of developing countries is to be therefore
Kuznets, "Economic Growth and Income Inequality", op. cit<,
2
~
Irving Bo Kravis, "International Differences in the
Distribution of Income"', in The Review of Economics and Statistics,
November I960.,
As well as some other measures of inequality, such as the
coefficient of variation? and the standard direction of logs of
income. "'
..-

- 20 -

sought in the greater dispersion in the upper part of the distribution scale.
The next to join the discussion was Harry Oshima who expressed
hesitancy to put too much reliance on Xravis' conclusion of "greater
inequality in underdeveloped countries than in developed ones".
He suggested that if we assume that countries pass through four
stages, undeveloped, underdeveloped, semi-developed and fully
developed, we can sa.y that inequality is generally low at the
undeveloped stage and that the dispersion of incomes increases as
countries grow to the next stage. Oshima further suggested that
inequality increases during the third (semi-developed) stage "but
that it will reach its peak there and that inequality declines
during the fourth (fully developed) stage.
2
In 1963 Simon Kuznets entered the discussion again with an
analysis of data, for 18 different countries. He concluded first
of all that the shares of the upper income groups were distinctly
larger in the underdeveloped than in the developed countries0

That

was very clear in the case of the top 5 per cent of families which
in developing countries typically accounted for 30 -per cent of total
income or more, hut in developed for only 20 or 25 per cent. But
even when we concentrate on the share of the top 10 per cent or
even 20 per cent of families, the share of income is clearly
higher in developing countries,
Kuznets'
lowest income
than those in
narrower than

second finding was that, while the shares o.f the

groups are in some underdeveloped countries lower
the developed countries, the differences are much
for the shares of the upper income groups and may

Harry Oshima, "The International Comparison of Size

Distribution of Family Incomes with Special Reference to Asia",
in The Review of Economics and Statistics, November 1962, pp.
439-445.
2
S. liuznets, Qualitative Aspects of the EconomicGrowth of
ITations, VIII, Distribution of Income by Size,, <yp_ cit...

_ 21 _

not be significant.

If we observe the shares of the lowest 40 or

60 per cent of families, there is probably very little difference

between developed and developing countries.
Prom these two findings follows a third one, that with greater
concentration of income in the top group and equal share of the
lowest group, the share of the middle group is lower in developing
countries and> generalise, that in the developing countries -income
distribution is more equal below the top 5 or 20 per cent of
families.
Following Kuznets1 study of 1953 there were a number of studies
comparing income distribution in relation to the level of economic
development by the use of different measures. The data, on which
these studies were based 'were usually those of Kuznets, or
similarly limited: Kuznets had 20 observations for 18 countries
but only for 12 countries was the data sufficient to calculate the
share of all five quintiles. In most cases their findings
"corresponded to those of Kuznets' 1963 study but some studies were
hesitant about the finding of greater inequality in developing
countries.
A substantial change in the basis on which comparative studies
can be undertaken was achieved by the assembly of data on size
distribution of income by Irma Adelman and Cynthia Taft Morris .
They examined the role of 31 economic, socio-cultural and political
indicators on income distribution by .size, as measured by six'
different indicators: the Gini ratio, the share of income accruing
to the top 5 and 20 per cent and of the bottom SO and 20 per cent
of poioulation, and the share
of the middle income group (the third quintile), for 44 developing
countries. This study used a rather unusual technique based on an

Irma Adelman and Cynthia.Taft Morris, "An Anatomy of Patterns

of Income Distribution in Developing Nations", Part III of the
Final Report under AID Grant, Northwestern Univei,sity, February 1971.

- 22 -

analysis of variance and found the following 6 factors to be most

significant in explaining variations in income distribution;
(l)

the rate of improvement of human resources (as measured by the

Herhison-Myers composite index); (2)

abundance of natural

resources (as measured by fuel and non-fuel mineral resources and

agricultural land per capita)| (3) the extent of direct government
economic activity (measured by the share of government investment
in total net investment)5 (4) the extent of dualism (measured by
the relative.importance of the exchange sector)5 (5) the extent of
potential for economic development (as measured by past rates of
increase of GEP and by improvements in seven areas of economic
institutions and activities) ; and (6) the extent of political
participation.
The level of economic development,, as measured by.the level
of per capita GIMP or GDP was one of the 31 indicators .tested by
Adelman and Taft Morris. It has not emerged a-s .one of the main
six determinants but quite a way down the scale among other factors.
This would suggest, that the level of economic development is not an
important determinant of the shape of income distribution and that
the degree of equality or inequality existing in a country depends
more on quite a number of factors than on the level of national
income pel" head. This finding would then be to some extent, at .
variance with the accepted theory of income distribution changing
significantly in the course of economic development, and the
finding would receive considerable support from the fact that it is
based on. a much larger amount of information (44 countries) than any
other previous attempt.
We would argue, however, against interpreting the work of
Adelman and Taft Morris in this way. In the first place, unlike
most previous studies, the Adelman--Taft Morris study did not aim
at examining the change of equality?" as a function of development
but rather at examining the relative importance of a number of
very different factors which influence inter-country differences

- 23 -

in income distribution. Their particular method of analysis of

variance, "based on an assjnuetrical "binary branching proce'ss makes
it possible for an important factor to be overshadowed by'others
if its impact is. not linear, in other wards if there is no straight
forward relationship such as "inequality diminishes as per capita
national income increases". And such a relationship is Indeed not
suggested by previous studies.
The second reason why the Adelman-Taft Morris study should
not be interpreted against the theory of the degree of inequality
changing' in relation to changes in economic development lies in
the nature of data used by the authors. Having done a heroic job
in assembling and calculating basic distributional information for
44 countries the authors unfortunately allowed a great number of
mistakes to creep into their calculations.' Their mistakes
invalidate the data for a few countries, but what is worse,
introduce a bias by making income distribution of some of the least
developed countries appear to be less unequal than it is.
The wealth of data assembled by Irma Adelman and Cynthia Taf't
Morris provides us, however, with the" opportunity to present income
2

distribution data

'

'

for a larger number of countries than has been

For a number of countries, primarily those with very low

national, income per capita, income distribution data 'available to
Adelman and Taft Morris were aggregated at the lower end of. the .
income scale. B y incorrect disaggregation (straightforward division
of the share between the lowest two or three quintiles), artificially
high share of income was allocated to lower quintiles and low share
to higher quintiles and thus an impression was created of smaller
inequality than primary data indicate.
2
It is perhaps at this point, however, that we might quote
the warning words - well known to workers in this field - of Simon
Kuznets: "It may not be an exaggeration to say that we deal here
not with data on the distribution of income by size but with
estimates or judgments by courageous and ingenious scholars
relating to size distribution of income in the country of their
concern". Kuznets, Qualitative Aspects (1963), pp. cit., p. 12.

- 24,

done before. This has "been done by dropping data for 4 countries
where basic information was entirely inadequate, by replacing data
for 3 countries by superior data from other sources, by recalculating
data (on the basis of primary or intermediate information provided
by the authors) for 9 countries and by adding information for 16
1
otner developed and developing countries. '
The resulting information on income distribution by size for
56 countries is -presented in Table 6. For each country the table
shows the percentage cf personal pre-tax income accruing' to the
first (lowest) 20 per cent of families or households, (the first
quintile), then to the second, third and fourth quintile, then to
the next 15 per cent (8l to 95 percentile) and finally to the top
5 per cent. The table also shows two measures of Concentration
calculated from these figures; the Gini ratio and the Maximum
Equalization Percentage, which indicates what percentage of total
income would have to be shifted between quintiles in order to
achieve an equal distribution of'income.' The countries are grouped
according to their level of economic development as measured by
Gross Domestic Product per capita. For each country per .capita
GDP in 1965 is shown, expressed in US *
The grouping of cou.ntries according to the level of their
per capita GDP was made in order to investigate relation of income
equality and the level of economic development not only in the
terms of the standard question; Is income more equally distributed
in developed than in developing coim.tries?5 but also to see
whether a certain tendency (towards greater equality or inequality)
is not at some point reversed. The great number of observations
available in Table 6 makes it possible to have a number of groups
corresponding to different levels of development and.at the same
time to have in each group more than two or three observations.
Averages for the various groups are shown in Table 7 not only for
the basic information shown for each country in Table 6 but also
for some further calculations based on such data.
-1

"""See Appendix 2 for description of these changes, methods of

of calculation, sources, etc.

Table 6

S i z e D i s t r i b u t i o n of P e r s o n a l Income B e f o r e Tax i n 56 C o u n t r i e s
In_oomei S h a r e s R e c e i v e d by Q u i n t i l e s of

Recioients

P e r c e n t i l e s of R e c i p i e n t s
Below 21 - 4 1 - 6 1 - 81 - 96 2055
40$ ' 60J5 80.1
95;{'
lOOJi'

Country and l e v e l
of G.D.P. p e r head

Gini
Ratio
__

Maximum
G.D.P.
E q u i l i z a t i o n p e r head
Percentage.
i n 1965
' ( U . S . $)

Under $100.00
8.0
8.0
7.8
7.0
5.6
4.8
10.0
' 8.0
3-9.
7.0

Chad (1958)
Dahomey (1959)
Niger (I960)
i l i g e r i a (1959)
Sudan (1969)
T a n z a n i a (1964)
Burma ( 1 9 5 8 )
I n d i a (1956-57)
M a d a g a s c a r (1960)
Group Average

11.6
10.0
11.6
7.0
9.4
7.8
13.0
12.0
7.8

15.4
12.0
15.6
9.0
14.3
11.0
13.0
16.0
11.3

20.0
16.0
19.0
22.5
31.0
18.1
20.3
22.0
22.0

23.0
32.0
23.0
38.4
17.1
42.9
28.2
20.0
37.0

0.35
0.42
0.34
0.51
0.40
0.54
0.35
0.33
0.53

25.0
30.0 .
25.0
40.9
30.7 '
41.0
25.5
24.0
39.0

68
73
81
74
97
61
64
95
92

10.0

13.1 19.4 21.4

29.1

0.419

31.6

78.3

22.0
20.0
23.0
16.1
22.6
15.4
15.5
22.0
18.0

8 1 0 1 . 0 0 - S200.00

>

Morocco (1965)
Senegal (I960)
S i o r r a l e o n a (1958)
T u n i s i a (1971)
B o l i v i a (1958)-.
Ceylon (1953)
P a r i s i a n (1963-64)
Sout/. Korea "(1966)
Group Avoruge

$201.00 -

7.4
7.1
3 . 0 - 7-.06.3
3.8
5.0
5.7
6.0
3.5
9.2
4.5
11.0'
6.5
9 . 0 1 4 . 0
8.6
5.3

12.4 44.5 20.5

0.50
1 5 . 0 . 2 8 . 0 .. 3 6 . 0
0.56
16.7 30.3 33.3
0.55
14.4 42.5 22.4
0.53
15.5 25.3 35.7
0.53
2 0 . 2 33-9 1 8 . 4
0.44
22.0 25.0 20.0
0.37
2 3 . 0 2 3 . 5 1 2 . 5 - 0.-25
17.5 31.6 24.9
0.468

45.4
44.0
. .
44.1 ,
44.y
41.u
32-5
27.0
19.0.
37.2

130
192,.
142
187
132
140
101
.107
14Y.6

8300.00

i..alaya ( 1 9 5 7 - 5 8 )
F i j i (1966)
Ivor:,- Coast (1959)
Zambia (1959)
B r a z i l (I960) '
E c u a d o r (1968)
El S a l v a d o r (.1965)
Poru (1961)
I r a q (1955)
. P h i l i p p i n e s (1961)
Columbia (1964)
Group Avuragc
8301.0'J -

7.7
10.0
9.1
10.0
12.0
13.8
15.5
18.0
12.0

6.5
4.0
8.0
6.3
3.5
o.3
5.5
4.0
2.0
4.3
2.2
4.8

11.2
8.0
10.0
9.6
9.0
10.1
5.5
4.3
6.0
6.4
4.7
6.0

15.7
13.3
12.0
11.1
10.2
16.1
8.3
8.3
0.0
12.0
9.0
11.3

22.6 26.2
22.4 3 C 9
15.0 26.0
15.9 19.6
15.8 23.1
2 3 . 2 J.y.6
17.6 26.4
15.2 19.3
16.0 34.0
19.5 23.3
16.1 27.7
10.1-25.7

.6.0
6.1
5.0
11.6
4.2
D.3
9.6
6.6
9.4
7.9

7.0.14.0
11.2 15.2
10.8 19.5
14.7 20.6
15.0 16.0
14.2 21.3
12.0 20.7
11.1 19.3
13.8 15.2
12.3 16.0

17.8
21.4
29.0
37.5
38.4
24.6
33.0
48.3
34.0
27.5
40.4
32."'0

0.36
26.5
0,46
3-4.7
0.43
35-0
0.48
37.1
-0.54
.41.5.
0.3d
27.5
0.53
41.4
0.61 ' 40.2
0.60
48.0
0.43
35.J
0.62
48.0
0 . 4 9 9 3;>.5

278
295
213
207
. .207202
249
237
265
240
275
244.4

3500.00

- 2.0 .
Gabon ( I 9 6 0 ) Cooa Hica (1969) . 5 . 5
2.2
J a m a i c a (1958)
10. Y
Sorinam (1962)
3.0
Lobanon ( 1 9 5 5 - 6 0 )
3-5
Barbadoo ( 1 9 5 1 - 5 2 )
5.4
C h i l e (1968)
3-5
Mexico (1963 i
4.9
Panama (1969)
4.5
Group Average

24.0
25.0
31.3
27.0
27.0
29.3
29.7
30.7
22.2
27.4

363
360
465
424
440
366
466
44.1
.. 490
426.9

47.0.
35.0
30.2
15.4
34.0
22.3
22.6
23.6
34.5
30.0

.0.64
0.50
0.55
0.30
0.55
0.45
0.44
0.53
0.46
0.494

51.0
40.0'.
41.5
23.0
41.0
32.9
33.0
39.5
36.7
37.6

10.2 26.4 18.0

13.2 17.9 22.2

39.4
29.3

0.58
0.42

43.7
V-.O

521
782

22.5
23.2
23.0
14.3
25.4

0.44
0.42
0.38
0.39
0.433

32.9
30.0
29.5
20.y
32.9'

704
904
591
838

0.30
0.38
0.42
0.45
0.50
0.46 .
0.40
0.44
0.35
2.30
0.401

21.2
26.1
30.0
32.9
36.5
33.5.
20.0
32,1
24.9
22.2
29.0

1,243
1,590
1,400
1,667
1,732
1,568
1,011
1,101
1,717
1,823

2,076
2,406
3,233

$501.00 - 81,000.(DO
South A f r i c a (19615)
Argentina (iy61)
i r i n i d a d and i o b a g o
(1957-58)
Vonezeula (1962)
Greece (.1957)
J a p a n (1962)

1-9
1.0

. 4.2
10.4

3-4
4.4
9.0
4.7

Group Average

5.1

9.1
9.0
JO.3
10.6
3.9

14.6
16.0
13.3
15.6
13.9

24.3 26.1
22.9 23.9
17.926.5
2 2 . 9 31-2
22.1 24.7

4.7

13.4
10.2
10.0
10.1
7.6
8.7
10.5
9.2
12.1
13.4
'10.5

18.6
16.6
16.0
13.7
14.0
15.4
14.6
14.2
18.5
17.8
15.9

21.6
23.9
21.6
16.0
22.8
24.2
20.4
21.5
24.4
23.4
22.2'

5.0
4.4
5.6

10.6
9.6
12.3

18.8 24.2 26.3

17.4 24.6 26.4
. 1 7 . 6 23.4 26.3

16.9
17.6
14.8

0.37
0.39
0.34

25.4
26.
24.5

' 5.0

10.9

17.9 24.1 26.3

16.4

0.365

fl , 0 0 1 . 0 0 -

82.000.00

I s r a e l (1957)
5.8
U n i t e d Kingdom ( 1 9 6 4 ) 5 . 1
N e t h e r l a n d s (1962)
4.0
West Germany (1964)
5.3
i T u n c e (1962)
1.9
F i n l a n d (1962)
2.4
I t a l y (1948)
6.1
P u e r t o Rico (1963)
4.5
liorway (1963)
'4.5
A u s t r a l i a (1966-67)
6.6
Group Average
J?,001.00 -

28.2 11.2
25.0 19.0
24.8 23.6
i y . 2 33.7
23.7 25.0
28.3 21.0
24.3 24.1
28.6 22.0
25.1 15.4
24.4 14.4
2 5 . 7 "20.y

1,485.2

S5.000.00

Denmark (1963)
Sweden-(1963)
U.S.A. (1969)
Group Average

Sourcei

723.3

See Appendix 2.

o.2

2,572.:

3 -H fl
3 r-l O

H -H -H

I
o

in

CM

0)

X 3 P .. _

UD

no

cd D 1 d o) cd
!;:! Pq N P < - P

CM

(Ti
CM

CM

o
CM

H -rl
JCJ P
H 03

V.O
CO

r-o
cy

tn

4-J

CM

CD

CO
V.O

cy

cyrI
cy

*.o

CO

CM

CM

CM

CM

v.o

CM
O

CM
CO

CM

o
CM

<-o

PO

CM

CM

CM

H
CTi

>
CO

CM

cn

-P

CD

c.O
LTv

CM

*.

+
cr>

CO

,-(
o

CM

CM

OJ

I"CM

fH
G)

.-cy

o
o

c^

nO

CO
CM

H
cd

CM

CM

cy

CD
H
^-i

H
O
CD
fH

lf\

ICA

c y .

!C\

o
o

rH
P
r
J
,Q
H

l>ir

irN

in

4-

CM

cr>

r-'-t

-P

O
ro

CM
CM

CM

!CN
CM

a
VD

CM

ro

CD
O

IT\
CM

CM

:r\

<H
O

CM

rH

CT\

cy

CM

cy

H
CO
r-l

CO
H

r-l
r-l

ITS
CM

CM

CM

CM
CM

CM

CM
CM
CTi

cy

cr

o
*

;r\

co

in

o
-5t-

CM

cn

.-; o

J.^

H
fH P<

CD

crt
P. o

LP>

c \o
m

?
o o
H O

1
O

CD H

o
o
H CM

H
O
O
O
CM r o

cq *

<**

< ^ ^

1
H

1 O
O

o o
rn ir\
^^

o
o
iPi H
<&*$
rH O

CD
iI
rQ

CJ
CD

rH

o
CD

CD

fH
O

p,

CM

a
O
fH
'+H

CD
CO
0)
fH
Pi
CO

O
CM

CD

c3
fD

fH

ct

o
o

=H
O

fH
CD
Pi

CD
.:_3

P
CO
CD

CD

o
r-l

o
CD
O
<H

C^

P
1.1

H
H

CO

fH

1 o
p.,
g u ,d
3
-H p n
O fn -rH
O

LT\

CO

o c3a -H
)

CD

00
p

V.O

p.,
-P
CD
O

fH

<D

o
CD

03
fH
H

CD
O

CO-

rO

CT\

LCN

1
VO

v.O
CM

CM

O
O

00
'rH

CD

CD

10

CO

H
O

PH

cr
I

O
CD

CD

fH
03
CQ

i'-^H

tH
O

0)

fH
<D

CD
N
H
CO

To

CD

03
r-l

r-(
.; .:>
.-3

0
fH
03

r\

V.O
IPs

r-l

CI
H

H
O

X
CD

CD

JO

CD
r~t
-M

CM

fn

Pi

-P
CD

CM
-cy

cy

fH

<-o
cy

cy

CM - H

-o
cy

>

o3
o

fH

fn

CD

Ii

a)
fn
o

fH

ic\

0)

CD

CD

H-

o
fn
cb

tjD

.y

CO
o

CM

<-o

CM

o
o

U
o

CT\

P"

CO
H

1
H
o O
o
o o#s
.i-HICM
C^^

H
CD
O
CD
>
t) o
CM G r o

* d o3

- 27 -

The first observation based on the data in Table 7 concerns

the overall indicators of income inequality, the Gini ratio and the
Maximum Equalization Percentage. Both indicators show a similar
picture. There is a sharp increase in inequality as we move from
countries in the lowest income group to countries in the $101-$200
group. There is some further, but less pronounced, increase in
equality as we move to the $201-$300. This group, together with
the $301.-4500 group represent the peak of inequality. There is a
substantial reduction in inequalit^r in the $501-$1,000 group: the
general level of inequality in this group corresponds to that of
the lowest income group, under $100. As we move further along the
developed path, to the $l,000-$2,000 and to above $2,000 groups
there is a clear reduction in the extent of inequality.
How far is'this picture of the changes in the general
indicators of inequality confirmed by comparison of the shares of
particular income groups? Starting with the share of income
accruing to the top 5 per cent of income, we find that the highest
percentage of total income accrues to this income class in
countries in the $20l-$300 and $301-8500 groups, indicating the
greatest inequality in these countries. The picture is less clear
for countries at a lower level of development, as the group below
$100 shows only a very slightly smaller percentage, but the $101$200 group shows (surprisingly) a lower share accruing to the top
5 per cent In countries with GDP per capita above $500 the share

The relatively low share of the top 5 per cent income

recipients in the $101-$2G0 group is to some extent due to the
inclusion in this group of Morocco and Tunisia, where the data
clearly understate the share of the top 5 per cent and overstate
the share of the next 15 per cent of income recipients. But
exclusion of these two countries would not change significantly
the shares of the various quintiles, as the average share (of the
remaining six countries) are as follows: 0^ - 5?, Q 2 - 9.2$,
Q 3 - 13,1^, Q 4 - 18.9$, P 8 0 _ 9 5 - 27.7/S and P 9 6 _ 1 0 0 - 26.1/$.
The average Gini ratio for the six countries is 0.45 and the
maximum equalitarian percentage is34.2.

Diagram 1.
D i s t r i b u t i o n of income at d i f f e r e n t

levels
MSr-

100%-

Wo

i+O

M -

lot -

.. U

U%

- 3v

b07* >

W.

.. So

.29

3C%

.. U

io r. ,

10% -

II

.rt'iL

GfiP

.. lh

- 29 -

of the top 5 per cent diminishes sharply and in relation to

increasing national income per head.
If we replace the share of the top 5 per cent by the share of
the top^jlOjer cent of income.recipients, the share of the rich
shows the same picture as the two general indicators of inequality.
There is an increase in inequality as we proceed to the $20l~$300
and $30l-$500 groups and then there is a decline, with the $501$1,000 group having the same share as the below $100 group. There
is a further fall in the share of the rich at higher levels of
economic development.
Turning now our attention to the share of the poorest classes
in each country, we find a prima facia surprising fact that the
share of the lowest 20 per cent of income recipients (the first
quintale) is actually highest in the poorest countries, where the
average share of the first quintile is 7 per cent in the "below
$100 group and 5*3 per cent in the $100-$200 group. The lowest
average share is the $301-4500 group. This finding corresponds to
the conclusions of Euznets and Eravis in so far that the share of
the poorest classes in the least developed countries is higher
than elsewhere, hut cannot he generalised into saying that the
poorest share is higher in developing than in developed countries.
Once we go above the $200 level, there is no clear pattern.
An analysis of the share of the poorest 4-0 per cent population
(the first two quintiles) still confirms the highest share in the
least developed group of countries, but for the rest the general
pattern established by our data begins to appear: As inequality
increases up to the $200-$5CO level and then starts to diminish, so
doe3 the share of the first two quintiles decline up to the critical
point and then starts to rise. With the extension of our analysis
to the poorest 60 per cent of population (first three quintiles) our
general pattern predominates, although the share of the. first three
deciles in the below $100 group is still higher than in any group
up to the $1,000 level.

- 30 -

To examine the distribution of income between different classes

of the poorer part of population we have calculated the ratio of
the third quintile to the first quintile, in other words the ratio
of the average to-the lowest income classes.

This ratio shows a

remarkably straightforward'and sharp increase as we move from the

less to the more developed countries, thus indicating much smaller
dispersion of the lower incomes in developing than in developed
countries.
Finally two more indicators expressing the ratio of the share
of the top 20 per cent to the bottom 60 per cent (Q^/Q + Q ? + Q-)
and to the bottom 20 per cent (Qc/Q-,) show once again the general
pattern of changes in income distribution in relation to economic
development as was established by our study? Increasing inequality
up to the level of $200-$500 and then a rapid reduction of inequality,
Summary of Findings
The examination of the problem of the effect of the level of
economic development on size distribution of income by comparing
historical data for particular countries leads to a rather
stir-prising finding of a clear long-term trend towards equality.
This is shown primarily by long-term studies of Great Britain and
Norway where the long-term trend towards equality was sometimes
interrupted by a rather long period of stability or even by a
short-term reversal.
But there was no indication that income

"The survey of historical experience makes us rather cautious

about accepting the generally propounded view that the reduction in
income inequality in developed countries started only with the
First World War.. Kuznets already observed that Denmark's inequality
was reduced before the First World War, Soltow's data quoted in
our study also indicate a long trend of inequality reduction before
the First World War in Korway. Kravis indicates a plan of narrowing
inequality in the USA between 1890 and 1920. The British situation
was basicalljr stable but there are also some indications of
reduction in inequality prior to the war. On the other hand it h.as
to be admitted that the reduction in inequality after the War was
sharper and more general than at any time before 4

- 31 -

distribution would be worsening for any substantial ..period of time.

The data for the USA and India also indicate that the trend, as
far as discernible, is for gradual movement towards equality. The
data for Puerto Rico, Argentina and Mexico do show some increases
in inequality but they are for.too short a period to contradict
the finding of overall trend of reduction of inequality.
The scarcity of data on historical series makes it however
preferable,, at least at present, to study the changes in income
distribution in the course- of economic.development by international
comparisons, even if we take into account the poor quality of some
of the data. 3y assembling data for 56 countries and by making
the first analysis'of differences- in some of the indicators-we
can offer some comments on several points discussed in the:last
decade .or two.
First of all we .have the question of whether there is higher
equality in'developed or developing countries. Our data allow a
clear answer. 'Taking as developing' countries 43 countries with
GDP per' capita in 1955 below $1,000 and as' developed the remaining
13, we find that the average G-ini ratio for developing countries
is 0.46? and' for developed countries 0.322. Similarly, the
maximum equalisation percentage for developing countries is 35-8
and for developed only 28.4 per cent',
The higher degree of inequality in developing countries is
due primarily to the high share of income received by the richest
5 per cent of population where the contrast between the developing
and developed countries is most startling: in developing countries
this share amounts to 28.7 per cent, and in the developed countries
to 19*9 per cent. There is a similar difference, although less
pronounced in relative terns, between the share of the richest
20 per cent income recipients in developing countries where they
receive 54.8 per cent of total personal income, and in developed
countries where they receive 457 ner cent of income.

- 32 -

On the other hand, the difference-is smaller if we compare

the shares of the lowest income groups.

The shares of the poorest

20 per- cent .of population are not very different and that in the
develo-ping countries (5*3 per cent) is actually slightly higher
than that of the developed countries (4.8. per cent).

Even if we

compare the-poorest 60 per cent of population, the difference is

not great, with the share in developing countries amounting to
26*3 Ver

cent and in the developing couutrj.es 31.7 per cent.

However, the difference in income distribution between the

developed and developing countries are not so important as the
pattern of income distribution which is typical for particular
levels of development. The data presented in this study support
the hypothesis expressed but not full3- tested hj Kuznets and
Oshima, that income inequality tends to increase with economic .
development, then remain stable and then fall. Our data show
clearly that there is an increase in inequality as countries
proceed from the below $100 level to the $101 - $200 level and
further. We established that the peak of inequality was.reached
in the groups with per capita income between $200 and $500. The
-per capita income figures cannot be of course taken literally as
they relate to the level of income in 1955 Rather, one should
think about the peak of inequality when countries arrive at the
level of development and structural pattern as characterised by
the countries.(shown in our Table S) which in 1965 had GDP per
capita in the $200- $500 range. With up-to-date' figures, both on
income'distribution and in per capita GDP we would find the peak
of inequality higher, perhaps between $250 and $600.
Most of the countries with the most pronounced inequality
of income can be found within this range, although two or three
of the high inequality countries are.in the $101 - $200 group
and one (South.Africa) in the $501 - $1,000. If generalisation
is wanted, one might say that highest inequality of income is
to be found among the somewhat richer African countries (Gabon,

- 33 -

South Africa, but also Senegal and Sierra Leone) and among the
poorer Latin American countries (Columbia, Peru, Brazil, Bolivia),
together with two Middle Eastern countries, Iraq and Lebanon. Ho
other Asian countries are among those with the highest inequality
of income.
Countries in the $501 - Si,000 group (other than South Africa)
have a markedly lower inequality and even further reduction in
inequality is to be found in the group above $1,000. Somewhat
exceptional - for this level of income - is the inequality of pretax income in France, but also West Germany and Finland are well
above the general pattern of this class.
Our data also confirm Kuznets1 hypothesis about the share of
the lower income groups in total income. The share of the lowest
20 per cent can be expected to be higher in countries with the
lowest per capita income, given a subsistence minimum which of
course forms a higher percentage of average income in the poorest
countries than elsewhere. But even the share of the lower 60
per cent population is higher in the below $100 group than in other
developing countries. It is only in the developed countries where
the lowest SO per cent of population receive a higher share than in
the poorest countries.
The establishment of a pattern of changes in the size
distribution of pre-tax income in the course of economic
development leaves us still fe,cing a number of unanswered
questions. Why is it that at each level of development there are
some countries whose share of distribution is in sharp contrast
with the pattern prevalent at that level? Imperfections in
statistics, numerous as they are, cannot account for more than a
minor part of these deviations. V/hat then are the factors causing
some countries to deviate from the pattern? And, more generally,
what are the factors causing the differences in inequality and
the changes in this pattern in the course of economic development?
And what role do changes in the level and structure of employment
play in the process?

- 35 -

Appendix 1
Measures of Inequality of Income Distribution
by Size
The size distribution of personal income is in effect a
frequency distribution of income, showing income recipient units
(in our paper mostly households or families) ranked according to
the size of their personal income. It can be presented (and
usually this is done in primary statistical sources) in the form
of size brackets of income, with an indication of the number of
income recipient imits falling into a particular size bracket, and
of the total income accruing to these units. For comparative
purposes it is useful to dispense with the indication of size.
brackets and the recipient units are grouped together into groups
of equal size, with an indication of the share of total income
accruing to each group. The most usual groupings are percentiles,
deciles and quintiles.
Size distribution of income described by percentiles can be
graphically expressed by a method devised at the beginning of this
century by Lorenz , by plotting cumulative percentages of
households or other income recipient units along the household
axis and cumulative percentages of income (or wealth) along the
vertical axis, as shown in Diagram 2. With a perfect equality of
incomes, the Lorenz curve would coincide with the diagonal, with
perfect inequality of incomes (one recipient unit receiving all
income), the curve would coincide with the horizontal line and
the right hand vertical line of the diagram.
The nearer the curve is to the 45 line, the greater is the
equality of the income distribution represented by the curve. Our
M.O. Lorenz, "Methods for Measuring Concentration of Wealth",
Journal of the American Statistical Association, vol. 9, 1905

- 36 -

diagram shows a more equal distribution for Norway (1963) and a

less equal distribution for Prance (1962).

Both curves are

obtained by plotting cumulative values of quintiles.

In addition

a Lorenz curve is shown for Prance using the same data but with
a finer breakdown, by deciles and with the subdivision of the 10th
decile by the 95th percentile.

'The curve based on more detailed

data is smoother*
The most usual single indicator of size inequality of incomes
is the Gini concentration ratio, which is the ratio of the area
distributed by the Lorenz curve (area A in the diagram) and the
diagonal to the area of the triangle (area B) As with perfect
equality the Lorenz curve coincides with the diagonal, the value
of the Gini ratio would be zero. .With x^erfect inequality the
value of the ratio would be unity. Our calculations of the Gini
ratio for 56 countries shown in Table 6 were based on the method
of approximate triangles given in J. Morgan , and were based on
quantiles and the 95 percentile. The Gini ratios shown are
therefore slightly smaller than would be the case if they were
calculated from deciles or even a finer breakdown.
The other overall measure of inequalitjr used in Tables 6 and
7 is the Maximum Equalization Percentage, a measure developed in
the EGS study on Income in Postwar Europe. This measure shows
what percentage of total income would have to be shifted between
quintiles in order to achieve- an equal distribution of income.
As it is calculated as a sum of the excess >of the share of income
over share of income recipients, it is equal to the sum of the
percentages by which share of income falls short of the share of
income recipients, and the measure, is thus one-half of the value
of the Kuznets ratio which is the sum of absolute differences
between shares of income and percentage shares of income
recipients.
James Morgan, The Analogy of Income Distribution, in.The Review
of Economics and Statistics9 August 1962, pp. 281-28

DiQ, fUtM/^

C.

fiO U.M.Z. (UjSt/y*. a^sd A ^ ~ u>L.'t>

/I^UKW

(Mi)

icuvf

/icMi.(/f6i.)

*v

la Led

em

>o

//

II

Co
/

/
/

4*e* ft

to

/
//

//

fce

16

I
10

PtiUu.fa.ft

10

*f

.<kCow*4

U-Ccf^iccfn.

\
iro

- 39 -

Appendix' 2
Statistical Data On
Distribution of Income by Size
<

' ~"

' Our collection of data on income distribution by size is

based primarily on the largest compilation so far available, that
of Irroa Adelman and Cynthia Taft-Morris This compilation covers
44 countries and is presented in several papers by these authors,
such as "Who benefits from Economic Development?", paper presented
at the International Meeting of Directors of Development Research
and Training Institutes, Belgrade, August, 1972, The fullest'
statement of primary and secondary sources, and of new data and
intermediate calculations, is however given by Adelman and
Taft Morris in their mimeographed paper "An Anatomy of Patterns' of
Income Distribution in Developing Nations", Part III of the Pinal
Report under AID Grant, Northwestern University, February 1971.*
In the course of work-with Adelman and Taft Morris data it
becomes obvious that they contain.a number of mistakes and
inconsistencies. The most obvious type of mistake was
disaggregation of data between quintiles by straightforward pro
rata allocation. This was remedied by recalculation of original
or intermediate data from the "Anatomy" paper or from original
source's' wi'th the help of logarithmic probability graph paper.
This method (based on the' u.'sual assumption that lower and middle
incomes are log normally distributed) is not entirely precise but
is greatly superior to pro rata disaggregation and should be
sufficient for the degree of precision shown here. When
disaggregation at the top end was necessary, the double
logarithmic graph paper method (Pareto) was used.
Investigation showed that original data for three countries
are so bad that they are unsuitable for any further calculations,
and information about Kenya (l9'6l/2) , Libya (1962) and Rhodesia

*. 40 -

(1966) has "been "therefore omitted.

Data for one Asian country are

also left out. Data for Argentina, Mexico and Trinidad and Tobago,
although not had, have "been replaced by superior data from other
sources.

The original Adelman - Taft Morris data which were

recalculated by me cover Bolivia, Chad, Costa Rica, SI Salvador,

Greece, Jamaica, Madagascar,. Niger and Tunisia.

It should be

stressed, however, that even'after elimination of some countries

and.after recalculations, data for some countries (in particular
Burma,

Chad, Morocco, Higer, Nigeria, Sudan and Tunisia) a ? e f

rather doubtful value.

To' the Adelman - Taft Morris data. I have added information
about a number of other developed and developing countries.
Although in a few places a disaggregation was necessary, they . . .
are mostly taken airectljr (or by a straightforward aggregation) .
from the following sources:
. ,
Austral_ia from IT. Podder, Distribution of Household Income :'
in Australia, The Economic Record, June 1972, pp.
181-200.
USA from U S Bureau of the Census, Current Population
Reports, Consumer Income 1969, Washington 1970, p.56. .
United Kingdom, Western Germany, 2Te_tlierland_s, Denmark,
Norway, _Sw_ed_eii, Prance and'Pinland, from UN, Incomes
in Postwar Europe! A Study of policies, growth and.
distribution, Geneva, 1967, Chapter 6, page 15
Mal^ava_(l.9_57/8)_ and Siouth Korea from H. Oshima, Income
Inequality and Economic Growth: The Post War
Experience of Asian Countries, in Malayan Economic
R-.eview, October 1970, p. 13.
Piji from Michael Ward, The Role of Investment in the
Development of P'i.ji, Cambridge 1971 > p* 109.

- 41 -

jCrini.dad_and_I-_obag_o from A. Amram, Distribution of Income

in Trinidad and Tobago and Comparison with Distribution
of Income in Jamaica, in Social and Economic Studies
(Kingston), June 1966, p. 105.
I_talv and 3arbad_qs from S.. Kuznets, Quantitative Aspects
of the Economic Growth of Nations, VIII. Distribution
of Income by Size, in Economic Development and
Cultural. Change, vol. XI, Number 2, Part II, January
1963, p. 13.
J^^^Jlic.2> &reQ''ciLn:

^ leZ?-P ^ r0E1 ^* Weisskoff,

ar:

Income Distribution and Economic Growth in Puerto

Eico, Argentina and Mexico, in the Review of Income
i^llJLe&ki^> December 1970, p. 312.