copy of poverty and inequality in india

7/27/2019 Copy of Poverty and Inequality in India

1/64

A Study on Poverty and Inequality in India

A REVIEW OF THE STUDIES ON POVERTY AND

INEQUALITY IN INDIA

1


2/64


CONTENTS

Introduction Page

Chapter 1: 1

Concept of Poverty and Inequality

Concept 2: 4

Poverty in India

Chapter 3: 20

Studies on Inequality

Chapter 4: 63

Causes of Poverty and Inequality in India

Chapter 5: 75

Conclusion: The Anti-Poverty Programmes

Notes 82

Bibliography 99

2


3/64


INTRODUCTION

Poverty and inequality are complex phenomena and their major discussions

and consequent policy implications deserve the attention of all concerned. A critical

review of the academic research on these phenomena reveals a variety of

perspectives. Major differences in this respect have characterized the discussions of

the definition of poverty and its measurement on the one hand, and of the elaboration

of the concept of inequality in the context of a developing economy on the other.

Policy implications also differ in different studies, which is a natural outcome of the

differences in approach. One of these approaches starts from the concept of

subsistence level for defined with some standard norm of nutrition. A second

approach has been to use income and expenditure data for constructing a poverty linebelow which poverty will be said to exist. In this approach data relating to per capita

income or expenditure are utilized to establish the concept of a poverty level.

Broadly speaking, these two approaches distinguish the major studies on poverty so

far undertaken in India.

Measurement of economic inequality can be regarded as one aspect of the

wider problem of securing social equity and justice. In a sense it is perhaps the most

important aspect of the broader concept. Economic inequality prevents the realization

of the goals of social equity and justice. Economic justice becomes an important

issue in the context of widespread poverty; the contrast becomes sharper when there

exist small islands of affluence in an ocean of poverty. The studies on inequality in

India generally use such standard measures as the Lorenz ratio though other more

refined measure have also been used. Poverty and economic inequality can be

regarded as the twin aspects of the same problem. In India again the problem of

poverty and inequality is further aggravated by caste stratification which is a

prominent feature of her social and community life over and above the economic

stratification as commonly found in capitalist economies.

The objective of this dissertation is very modest. It attempts to make a brief

survey of the major studies on the measurement of poverty and inequality in India.

At the outset the theoretical position in respect of the concepts of poverty and

inequality has been explained, and subsequently the various Indian studies on the two

topics have been analyzed. Thus the dissertation has been divided into five Chapters.

3


4/64


In the first chapter, we review economic theory relevant to the concepts of poverty

and inequality. In the second chapter we deal with different measures that have been

evolved for the study of poverty in India. In the third Chapter we take a critical look

at the studies made about economic inequality in India. The fourth chapter is devoted

to an attempt to piece together the factors which the different studies on poverty and

inequality have shown to be the underlying causes of these disturbing phenomena. In

the last chapter we have made a brief survey of the programmes adopted in India to

launch a direct attack on poverty in the rural areas.

4


5/64


CHAPTER-1

CONCEPT OF POVERTY AND INEQUALITY

An analysis of the Indian economic situation reveals two features: First, while

the economy has attained a modest trend rate of growth, the problem of poverty

perpetuates. Secondly, in our plan documents the problem of distribution has not

been completely left out of discussions relating to production, and the possibility of

conflict between growth and distribution has not been explicitly recognized. 1 The

essence of this argument in favour of this is as follows: A very slow rate of growth

along with inequality in income distribution has perpetuated poverty in India. A large

proportion of population has to live without even the most essential needs of daily life

because total national income is too small relative to the size of population; and

secondly, the distribution of this income is very uneven. This problem cannot be

overcome if emphasis is not put simultaneously on economic growth and reduction of

inequality. Even the highest attainable rate of growth cannot make a major impact on

the problem of poverty in the foreseeable future if inequality is not reduced.

There has been a clear recognition, from the First Five Year Plan onwards, of

the need for special policies for the benefit of the poor. In fact, the justification of

policies like land reforms, subsidies to the village industries, economic assistance to

the backward communities and so on is to be partly found in the recognition of their

problem, while the problem is explicitly stated in the plan documents, there always

appears to exist a gap between the formulation and execution of policies which could

be regarded as intended primarily for the benefit of the poor. Indeed, the attempt to

identify the poor in operational terms has started only in recent years.

At present the problem of poverty is being dealt with both unprofessional

writing and in government policy documents and programmes with increasing

frequency. Poverty is a complex phenomenon and at least its major dimensions

require considerable attention if the problem is to be properly comprehended. We

should, therefore, start with the basic question: What is poverty?

A review of the academic research in the field and of government programmes

dealing with this subject shows several perspectives which have characterized

5


6/64


approaches to the definition of poverty. Three such perspectives can be clearly

discerned from the literature.

One such perspective to the definition of poverty uses income and expenditure

data in order to establish a bench mark income figure below which poverty may be

said to exist.2 Thus a size distribution of incomes is constructed and per capita income

data are widely used as a means of establishing the bench mark income figure. We

can distinguish two aspects of the so-called size distribution of incomes. The first

focuses attention on one end of the distribution, those with the lowest income or the

poor, choosing a more or less arbitrary bench-mark income figure. Here one can

study the magnitude of poverty either in terms of the absolute number of the poor, or

one can look at the different degrees of poverty within the group designated as the

poor.3 The second aspect comprises the degrees of inequality in the distribution of

income and is concerned with the whole of the distribution. The degree of inequality

is generally measured by the Lorenz Curve or Gini Coefficient. While the first aspect

deals with the absolute poverty level, the second deals with poverty in its relative

sense.4

The first approach has its limitations- incomprehensive coverage, inaccuracy

of measurement, inconsistency of definition and a failure to include all the sources

other than income which support the consumption of people in the lower rung of the

economic ladder. In the Indian context, measures of supporting resources such as

ownership of land and other assets, income in kind and private gifts should be

included and these are very difficult to compile on the part of the statistician. Thus,

income figures are not accurate especially for the two extreme economic classes- the

poor and the rich. While the poor are ill-equipped to provide correct informationrequired for proper income and expenditure studies, the rich are least inclined to

reveal their true economic power. As a result, income and expenditure data tend to

have a downward bias for both the groups: for the poor, lack of literacy and

consciousness and the subsistence production system largely play their parts; for the

rich, the reasons are completely different-lack of organization in the economy and

steep progression in the direct tax laws include the rich to record their income on the

lower side.

6


7/64


The second perspective can be specified as a subsistence approach to the

concept of poverty. People who cannot afford the minimum necessities for bare

subsistence are defined as poor. But how are we to define the required minimum

level? The minimum need is defined in terms of food consumption or more

specifically, in terms of nutritional requirements. This is then converted into an

income level for a particular base year.5

But in a country of Indias size, with wide differences in geography, climate,

habits and customs, nutritional requirements are found to vary across levels and

patterns of living and diet. In such a situation, to develop an index of minimum needs

it is necessary to take account of customary behavior. Therefore, such a subsistence

definition of nutritional requirements; it also depends on subjective consideration like

preferences and prejudices.

A third perspective to the definition of poverty is concerned with the degrees

of differences of welfare among different group of people who can be designated as

poor by the first two approaches mentioned in earlier paragraphs. This approach

seeks to establish a measure of poverty by attaching appropriate weights, which vary

inversely with the difference of income levels of the poor from the chosen poverty

line.6

Thus from the first approach, we have an income threshold definition of

poverty, while the second gives a multinational definition of the same. But these

approaches divide the population into poor and non-poor and cannot take into account

the differences in the degree of intensity of poverty in the different income layers

below the poverty line. This weakness is sought to be removed in the third

approach.

What lessons can be draw from the three approaches to the definition of

poverty? One point which emerges is that these approaches try to define poverty in

the context of economic factors only and, again, these definitions are concerned with

the people in the lower strata of the income scale. But the concept of poverty should

be seen in the context of society as a whole. As society is better seen as a series of

stratified income layers, poverty should be conceived in the light of how the lower

strata fare as compared to the people in the upper layers of distribution. Moreover,

7


8/64


poverty means helplessness of the persons designated as poor as these people are at

the lower end of a two-fold hierarchy of stratification along economic and social

dimensions. In short, the poor are part of a set of stratification system within the

society and they are ranked at the bottom of each hierarchy. The economic factors

are significant in the sense that these constitute the most important dimensions of

poverty; but these alone cannot fully describe the condition of the poor. Therefore,

the definition of poverty should be broad-based, for it is only a proper identification

of the poor that can determination of policies to solve the problem in a proper way.

Measures of Inequality

A variety of indices for the measurement of the degree of inequality are

found in the literature on income distribution. These indices emphasis different

aspects of the inequality phenomenon. Theoretically, these indices are not

completely satisfactory; sometimes, they exhibit contradictory tendencies in the

distribution of income unless the latter adapts itself to a well defined family of

distributions whose properties are completely known. Because of the limitation

inherent in all measures of inequality, one may adopt a more general approach to the

problem by laying down that any measure of inequality should be concerned with the

distribution of a size variable (x) relative to the mean (M) or any other typical value

of x. This principle is recognized in all the known measures of inequality. These

well-known empirical measures of inequality are Lorenz Curve, the Gini Co-efficient,

the Co-efficient of variation, the S.D. of logarithm etc. Before going into details

about these measures, we discuss the methodology regarding the measurement of

inequality of income.

Conceptually, we can draw a distinction between the objective and the

normative approaches to the measurement of inequality of income. As objective

measurement, in itself, is not of interest and presupposes appraisal; the difference

between the two approaches, in fact, emerges at the level at which normative

considerations come up. Therefore, use of the word equality in the context of

distribution of income necessarily involves normative judgment. This is so in both

the situations. We may be interested in inequality in the distribution of income of a

8


9/64


country at different points of time or, we may compare the inequality in the

distribution of income of different countries at the same point of time.

In the literature, we find two broad categories of the measures of inequality.

On set of measures try to quantify the size of inequality in some objective sense by

using some statistical measures of relative variation of income. Such measures

include the variance, the co-efficient of variation, the Lorenz Curve, the Gini Co-

efficient etc. On the other hand, there are indices that try to measure inequality in

terms of some normative notion of social welfare- that is , when the level of income is

given, a higher degree of inequality corresponds to a lower level of social welfare. 7

We now discuss the measures of inequality. It would not be wrong if wediscuss some measures briefly while concentrating on others which have been used

widely in the literature on inequality in income distribution in India.8

Long ago Pigou maintained that any transfer of income from the poor to the

rich, ceteris paribus, should increase inequality and diminish welfare.9 The common

statistical measure of dispersion-variance-does satisfy the Pigou condition. But the

dependence of variance on the mean income level makes its position weaker; since

one distribution with lower mean income level but greater relative variation around

the mean registers smaller variance than another distribution with higher mean but

smaller relative variation around the mean. To remove this weakness we take the

square root of the variance and divide it by the mean to make it mean-independent.

This is the co-efficient of variation.

The Co-efficient of variation is sensitive to income transfer for all income

levels. But sometimes it is desirable that we should attach lower importance toincome transfers at the higher levels of income and greater importance to income

transfer at the lower end of the income scale. This objective is fulfilled in another

measure- the standard deviation of logarithm. Since standard deviation is derived

from the logarithm of the actual income levels, the staggering effect of logarithm

highlights the income transfers at the lower end of income distribution and minimizes

the effect of the transfer at the higher levels. This feature makes this measure

attractive, but this creates difficulty for other reasons. The severe contraction the

income levels suffer-as they get higher and higher-denies this welfare measure its

9


10/64


concavity to be a concave function of individual incomes, this measure can create

difficulties.10

When we get interested about the share of different deciles of population from

the lower end in the national income, we take recourse to the Lorenz Curve. This

curve depicts the percentages of the population arranged in ascending order according

to their positions of the income scale on the horizontal axis and the percentages of

total income enjoyed by each of these groups are shown on the vertical axis. If

everyone has the same income, the Lorenz Curve will by the diagonal, which is called

equality line. The deviation of the Lorenz Curve from the line of absolute equality

is an index of inequality. One distribution is unequivocally more unequal than

another only if the Lorenz Curve for the former lies below the Lorenz Curve for the

latter throughout its range.

The advantage of the Lorenz Curve lies in its use of the arithmetic scale. It is,

again, independent of any assumption about the form of the distribution to which the

observed data must conform. However, when the form of the distribution is known

with certainty, the corresponding Lorenz Curve is uniquely determined by the values

of the distributional parameters. For different values of these parameters Lorenz

Curves are obtained which are completely non-overlapping.11

The use of Lorenz Curve as a means of measuring income inequality within

the utilitarian framework has been justified in the literature. 12 It has been shown that

if social welfare is additively separable and if it is the sum of individuals-then the

partial ordering of income distribution according to Lorenz Criterion is identical with

the ordering implied by social welfare. This result holds irrespective of the form of

the individual utility functions as long as they are concave. 13 Intuitively, one might

expect that the equivalence of partial orderings would hole for any symmetric social

welfare function which is quasi-concave and increasing in individual utilities. 14

When the Lorenz Curves of two distributions intersect, there is ambiguity in

comparison; that is, for comparative purposes, it is difficult to say which Lorenz

Curve represents greater inequality. Gini introduced an analytical measure called the

concentration ration or the Gini Coefficient, which is the ratio of the area between

the equality line and the Lorenz Curve to the triangular region below the diagonal. It

10


11/64


is defined as exactly one half of the relative mean difference-this is the arithmetic

average of the absolute values of differences between all parts of incomes.

==

=q

j

ji

q

i

yymq

G11

2||

2

1

Where m is the mean income of the poor, and q is the number of persons.

After a bit of manipulation the formula reduces to 15

( )=

++=q

i

iqyimqq

G1

21

211

for y1 y2 y3 . yn

Since the Gini Co-efficient takes note of differences between every pair of incomes, it

is a direct measure of income inequality; it is also sensitive to transfer from the rich to

the poor at every level.

Gini Co-efficient is mean-independent, that is, it is invariant with respect to

equi-proportionate change of incomes of all individuals. If the mean income of the

distribution changes, while there is no change in the inequality measure, the effect on

the measure of absolute poverty is certain. It moves inversely with the changes in the

mean income. But measure of inequality being mean-independent, changes in mean

income introduce complex difficulties in measuring of poverty in a relative sense. In

a society with higher mean income inequality causes greater injustice compared to a

society with lower mean income. This fact introduces ambiguity in the measurement

of poverty in a relative sense.16

Since the Gini Co-efficient is an index of inequality, the negative of it can be

treated as a welfare function. But this function, being a linear function of income

levels, is not a strictly concave group welfare function. As strict concavity is a

desirable criterion for a welfare function. Gini co-efficient is thought to be weak on

this point.17 But this measure takes into account any transfer of income from the

poor to the rich in appropriate direction, and it is, though not strictly concave, is

concave alright.18

Atkinson proposed a new measure of inequality19 by an index:

11


12/64


m

YI

*1* =

Where Y* is the level of income per head and m is the mean of the income

distribution. If the level of income is equally distributed, Y* would give the same

level of social welfare as the present distribution.

Moreover, in a separate measure,20 Atkinson introduced distributional

objective through an explicit parameter, say E, which represents the weight attached

by the society to inequality in the distribution. When the society is indifferent

regarding the income distribution, the value of E becomes zero, and when the

society is concerned only with the position of the lowest income group, the parameter

becomes infinity. The index is:

En

i

E

fiY

YiI

=

=

1

1

1

1

1

Where Y1 is the income of those in the 1-th income range, f1 is the proportion of the

population with incomes in the 1-th range and Ythe mean income. This index

indicates the proportion of the present total income that would be required to achieve

this same level of social welfare as at present if income were equally distributed. The

measure is an index of the potential gains from redistribution.

The inadequacy of the index of income distribution as an index of the

distribution of welfare has been pointed out by Bentozel.21 In this view, it is the

distribution of consumption which has a direct bearing on the distribution of welfare.

From the welfare point of view the welfare-creating properties of the distribution of

income or consumption are important; these properties should be represented in the

analysis of income inequality. Bentzels new measure seems to fulfill this condition.

The construction of this index is as follows: Assuming that a group of individuals

have the same welfare function u(c), where c denotes the level of consumption and

f(c) is the frequency function of the level of individual consumption, the aggregate

welfare (w) can be written as

dccfcunW )()(0

=

12


13/64


Where n is the number of individuals. Suppose W* is the maximum

aggregate welfare that could be obtained by a redistribution of consumption between

the individuals. The difference between W* and W can be interpreted as the total

welfare loss caused by the inequality in the individual consumption measured by

*1**

W

WI =

The ratio W/W* has been called welfare efficiency of distribution by Bentzel. But,

in this analysis the precise relationship between distribution of consumption and

distribution of welfare is not brought out very clearly. Again, since the shape of the

welfare function u(c) is unknown, the definition of welfare efficiency of

distribution is not useful for empirical analysis.22

Sens Measure

Sen has constructed a measure of poverty in an axiomatic framework.23 This

measure takes into account the income distribution among the people designated as

poor, that is, people below the poverty line. If yj is the income level of individual

j, the individuals may be numbered in a non-decreasing order of income, satisfying

y1 y2 . yn

If z is the poverty line and q is the number of people with income yi z, then for

any person i q, there are exactly (q+1-i) people among the poor with at least as high a

welfare level as person i. Let n be the total population. The poverty-measure P

can be written as

)1)(()1(

2iqyiz

nzqP +

+

=

This measure of poverty P is closely related to Gini Co-efficient, and Sen has

established a correspondence between Gini Co-efficient and his poverty measure.

Defining the head-count ratio H as the ratio of the number of people below the

poverty line, Sen(1976) has proved the relation

P = H [I*-(1-I*)G]

Where G is the Gini Co-efficient and I* reveals the percentage of the mean shortfall

of incomes of the poor from the poverty line. 24 From this relation an important

13


14/64


policy implication follows: The aggregate welfare of the population can be increased

by minimizing the inequality in the distribution of income.

In any actual situation, the use of the measure P involves the specification of

the welfare function. In such a choice, value judgment cannot be avoided. Moreover,

this measure is concerned with only a part of the income distribution.

Our analysis of the different measure of inequality leads finally to the

following points: First, the literature is full of controversy regarding the practical

relevance of the implicit welfare function, the shape of the welfare function and the

question of ordering in the social welfare. Second, while descriptive measures like

coefficient of variation, standard deviation of logarithm etc. lack motivation, purelynormative measures seem to leave out of consideration several important

characteristics of inequality.

14


15/64


Chapter-2

Poverty in India

The problem of poverty had attracted the attention of economists even before

independence25 but as we are interested in the study of the post-independence period,

we will not discuss this here. After independence a vast literature on poverty has

emerged, the most important of this being Dandekar and Rath(1971), Ojha(1970),

Minhas(1970), Bardhan(1970,1973) and Vaidyanathan(1974). Most of these studies

take 2004-05 as a bench mark year for the purpose of comparison, and the data used

by these studies cover periods up to 2008-09. These different studies come to

different conclusions both as regards the definition of the minimum standard of living

and regarding the number of people below the different studies are four in number: (i)

differences in estimates of income and consumption; (ii) different concepts of

adequate nutritional levels; (iii) differences regarding current price deflator to be

used; and (iv) arbitrariness in the choice of some key conversion factors to overcome

the lack of availability of appropriate disaggregated empirical data. Let us now

explain these in turn.

First, for the calculation of the size distribution of income and/or consumption

most of these studies depend on the data yielded by the National Sample Surveys.

But there is discrepancy between NSS data and national income statistics prepared by

the C.S.O. It is alleged that NSS data underestimates consumption expenditure for

the upper income classes,26

which implies that the measure of poverty will not beaffected but the measure of inequality will be. Secondly, the concepts of a minimum

adequate level of nutrition and its purchasing power equivalent form the basis of all

the studies that have tried to estimate the numbers of people living below the national

poverty line. But different authorities give different estimates of nutritional

requirements. Among these are: PPD, Planning Commission(1974), Sukhatma

(1971) and Gopalan, Sastri and Balasubramanian(1971). Thirdly, as most of the

income and consumption expenditure data are available initially at current prices, one

has to deflate them to obtain real values for the purpose of inter-temporal

15


16/64


comparisons. But the choice of a correct price deflator is difficult as different income

groups possess different preference patterns, some buy commodities at ex-farm prices

and some obtain commodities through exchange in kind. Again, prices for the

identical commodities vary region wise. Thus conclusions vary because the

researchers use different price deflators. While Minhas(1970) and Dandekar and

Rath(1971) use the national income deflator, Bardhan(1970) uses the official

Agricultural Labour Consumer Price Index for deflating the values of consumption of

the rural poor and the Official Working Class Consumer Price Index for deflating the

consumption of the urban poor.27 In this connection, it will not be out of place to

mention the nature of price movement of different commodities in different parts of

India. Mahalanobis(1962) had drawn our attention to the unequal movement of the

prices of cereals for different decile groups of the population in rural India.28 This

was later corroborated by extensive tabulation of NSS household budget data

undertaken for the Government of India Committee on Distribution of Income and

Levels of Living. Since cereals occupy a very important place in the consumer

budget, specially for rural households, the above finding stresses the need of

constructing inter-temporal consumer price indices separately for different fractile

groups of population. Fourthly, in different studies, conversion factors have been

used to build up a complete picture from fragmented data. The assumption in

Dandekar and Rath(1971) is that the top 30 per cent of the population would have

fared no worse in terms of consumption between 2004-05 and 2010-11 than the lower

income groups. On the other hand, Bardhan(1970) uses a conversion factor

appropriate to the bottom 50 per cent of the population for the derivation of an

estimate of total expenditure from expenditure on food items. From these it follows

that there cannot be a correct estimate that can be derived from the assumption that

are made in the studies noted above.

With these preliminary observations we will now turn to an analysis of the

Studies on Indian Poverty.

In July, 1962 the Government of India set up a Study Group29 which

recommended that a per capita annual consumption of Rs.240 at 2004-2005 prices

should be considered as the nationally desirable minimum level of consumer

expenditure. This excluded expenditure on health and education, which was

16


17/64


18/64


allowances for India in terms of availabilities and consumption habits; he has taken

into account different requirements of persons in different age and sex groups and

thus worked out two food baskets corresponding to a minimum concept and a

medium concept. The cost of the minimum food basket (per day) has been worked

out as Rs.0.5238 at 2004-05 prices or Rs.15.71 per month at 2004-05 prices. As

compared with Sukhatmes estimate, the cost of a minimum diet prescribed by the

FAO33 is Rs.18.26 at 2004-05 prices.

Sometimes alternative standards have been defined in terms of minimum

Calorie requirements, as in Ojha (1970) and Dandekar and Rath (1971). They have

taken 2250 Calories is met from food grain and considering the food grains intake per

person in different expenditure brackets as given by NSS data they have estimated the

proportion of population below the poverty line.34

On the basis of the recommendations of Dr. Patwardhan as mentioned in the

Report of the Central Government Employees Pay Commission (2001-59), Bardhan

(1973) has worked out the cost of the minimum diet recommended for an built in

moderate activity.35 The diet consists of 15 0z. of cereals, 3 0z. of pulses, milk (4 0z)

sugar and gur (1.5 0z), edible oils (1.25 0z), groundnut (1 0z) and vegetables (6 0z).

These give 2100 calories and 55 grams of protein per day. Because of non-

availability of prices, Bardhan leaves out the last two items i.e., groundnut and

vegetables. For the determination of the cost of this minimum diet Bardhan adjusts

the NSS rural retail prices to take account of the lower prices for non-monetized

consumption. First, the cost of this minimum diet for adults is determined. Then he

adjusts it by an adult-equivalent ratio to obtain the cost of the minimum diet for an

average person. Then he blows it up by using food to non-food expenditure ratio of

the relevant expenditure group of the rural population from NSS data, to get the

estimated minimum level of living for rural India to be Rs.28 per capita per month in

2008-09 in current prices.

Bardhan (1973) has taken the food basket from the report of the Second Pay

Commission and has used the rural retail prices from NSS estimates as is done by the

Commission; but he has arrived at a much lower value. While in the former the cost

of the minimum diet is Rs. 14.51 per month per adult unit, that is, Rs.11.61 per month

per person basis, the corresponding figures in Bardhan are Rs.11.61 and Rs.9.61

18


19/64


respectively. The lower figures of Bardhan can be traced to the following two

elements in his calculation: First, he has neglected vegetables and ground-nuts from

the food baskets; and secondly, he has adjusted the NSS based average retail prices to

take into account the consumption of home grown articles on the part of rural

consumers. Regarding the second consideration Rudra (1974) has taken exception on

two points.36 First, according to Rudra, the procedure adopted by Bardhan (1973) is

arbitrary, as he overlooks the fact that it is not only the home grown part that is

evaluated in the NSS data in prices different from the average retail prices; the cash

purchased part in NSS expenditure data is evaluated in actual prices paid by the

consumers, and not by the average retail prices independently estimated and

separately presented by the NSS. While the price average in the NSS expenditure

data regarding cash purchases is a weighted average, the rural retail prices are simple

averages. Secondly, Rudra points out that the blow-up factor (46%) is not correct.

We have discussed the nature and inadequacy of the concept of an absolute

poverty line in India. Let us now turn to the studies on Indian poverty.

Minhas (1970)37 has used the computations of Tewari (1968)38 to derive the

estimates of per capita private consumption expenditure, which are at 2004-05 prices.

He has, then, applied the NSS ratio of rural to urban consumption to obtain the

estimates of rural average per Capita Consumption. These estimates of per capita

overall consumption in rural area in 2004-05 prices, together with the percentage

shares of different fractile groups of population in total consumption, derived from

different NSS rounds, have been used to derive the average per Capita Consumption

in 2004-05 prices of each fractile group of the population over several years.

It will not be out of place if we reproduce here a table from Minhas:

19


20/64


Table: Average Per Capita Consumption by Fractile Groups at

2004-05 prices in 2000-01 and 2010-11: Rural India

Fractile Group2000-01

(Rs.)

2010-11

(Rs.)

Poorest 5% 63 81

5-10 88 110

10-20 108 137

20-30 133 166

30-40 155 194

40-50 180 222

50-60 207 254

60-70 240 29270-80 283 240

80-90 351 414

90-95 443 512

Richest 5% 731 723

Source: Minhas (1970) Table 2. Abridged

We see that the richest 5% of Indians could enjoy a per Capita

Consumption level of about Rs.2 in 2004-05 prices per day in the years 2000-01 and

2010-11. Even this group cannot be termed as rich by the world standard, though

they are at the topmost income level in the villages of India. The poorest 30% or the

people, and the other extreme, live a life beyond the description and analysis of the

economists and nutrition experts.

If the level of per Capita annual consumption expenditure of Rs.240 at 2004-

05 prices is accepted as a bare minimum, the percentage of people in rural India

below this poverty line was 65 in 2000-01 and 50.6 in 2010-11. But in Minhas

estimate the absolute number of people below the poverty line in rural India remained

almost same 215 million in 2000-01 and 210 million in 2010-11. If, however, a

level of Rs.200 at 2004-05 prices is taken as the minimum, the proportion of people

below the poverty line would be 52.4 per cent in 2000-01 and 37.1 per cent in 2005.

The steady decline in the proportion of people below the poverty line is, according to

Minhas, largely explained by the growth of an average per Capita Consumption at a

rate of 1.25 per cent per year over the period 2000-01 and 2010-11.

20


21/64


As regards the identification of the rural poor Minhas be specified several

classes to which most of them belong and these classes are:

(a) Agriculture labour households without land, which formed 58 to 61 per

cent of all agricultural labour household between 2000-01 and 1963-64.

(b) Other rural labour household without land.

(c) Agricultural labour household with land, which formed between 42 to 39

per cent of agricultural labour households between 2000-01 and 1963-64.

(d) Marginal farmers operating holdings below 5 acres in size.

According to the estimate of Minhas, the total number of people in agriculture

labour households in 2004-05 was 66.5 million of which about 26.5 millionsbelonged to households who operated some land as well. The corresponding

estimates for non-agricultural rural labour were 15.8 and 6.8 millions respectively.

Therefore, the total rural labour population in 2004-05 was estimated to be 82.3

millions and of these 33.3 millions belonged to households who operated some land

also. Minhas estimated that about 79 per cent of the population of these households

had a per capita annual level of consumption of less than Rs.200 at 2004-05 prices in

the year 2000-01. Minhas further stated that the percentage of poor among the

agricultural labour population declined to 75 in 2004-05, which corresponded to

about 50 million in absolute numbers.39

While Minhas is concerned with rural poverty only, Ojha (1970) deals with

both rural and urban poverty for the year 2004-05 and only rural poverty for the year

2010-11. In his study Ojha accepts 2250 Calories as the barest minimum intake per

capita per day for a representative Indian. He further assumes that people in urban

areas derive 66 per cent of 2250 calories from food grains; for the people in rural

areas that percentage is 80. he then compares the actual food consumption in grams

per head of rural and urban population using the NSS data [16 th round].41 This means

that to attain the nutritional minimum per capita daily consumption should be 518

grams in rural areas and 532 grams in urban areas. After some modifications of the

estimates of food grains intake he finds that the average level of food grains intake

per person was below the standard for expenditure levels up to Rs.15-12 per capita

per month at 2004-05 prices. Regarding the urban population the corresponding

expenditure level for which deficiency in consumption existed was Rs.11-13 per

21


22/64


capita per month at 2004-05 prices. On his basis Ojha finds that in 2004-05 about

51.8 per cent of rural population lived below the poverty line, and for the urban

population the proportion of those considered as poor was only 7.6 per cent.

Ojhas conclusion regarding poverty in the urban area is surprising because of

the following reasons. First, the urban households tend to spend a smaller percentage

of their incomes on food grains, which Ojha has accepted in his calculation. But this

again means that urban households derive a comparatively larger fraction of the

required 2250 calories from food other than food grains gives greater calorie value

compared to the same spent on items other than food grains, the minimum

expenditure levels for the required 2250 calories should be higher in urban areas.42

Secondly, since unit prices of food grains in urban areas are higher than in rural areas,

expenditure in urban areas for food grains should be higher. 43 Though Ojha takes the

nutritional norm for urban are as 432 grams, which is 86 grams lower than the rural

requirement of 518 grams (and this means that smaller volume of expenditure is

involved), the urban people must procure additional calories from food items other

than food grains, and again they are to purchase these at higher prices. These points

seem to have been missed by Ojha and that is why his estimated poverty line for the

urban area is as low as Rs.11-13 per Capita per month, and much lower than the

figure estimated for the rural areas. From these it seems that the proportion of

population in urban areas below the poverty line is much higher than the estimate of

Ojha indicates.

Regarding the year 2010-11, Ojha calculates that about 70 per cent of rural

households were living below the poverty line. Moreover, the degree of nutritional

deficiency was higher for each expenditure level. Ojha does not consider the question

of urban poverty for the year 2010-11.

The study of Dandekar and Rath (1971) has some important features which

are as follows.44 First, this study is extensive in the sense that it covers both rural and

urban poverty condition for the relevant period. Secondly, in this study the authors

have made a number of arbitrary adjustments to the NSS data. Thirdly, these authors

also assume that a daily intake a 2250 Calories per adult person is the required

minimum for subsistence, and this figure is derived from the estimate of Sukhatme

(1971). Fourthly, these authors have not only analyzed the nature of Indian poverty

22


23/64


in greater detail, but they have also examined the feasibility of the Planning

Commissions perspective regarding the reduction of poverty in India during the

planning period.45 In fact, a large part of their study is devoted to prove that the

Planning Commissions perspective appears to be completely out of line with the

performance of the economy in the past decade and is therefore unlikely to be

realized.

Moreover, Dandekar and Rath are critical about the NSS consumption

expenditure data, which, according to them, do not reflect fully the pattern of

inequality existing in the economy because of its downward bias regarding the

consumer expenditure of the richer sections. However, we shall examine this point

later on.

The method followed by Dandekar and Rath is similar to that of Ojha (1970)

but with difference that they have assumed that about 200 calories would be obtained

per Capita per day from food other than food grains. Now according to Dandekar and

Rath (henceforth D R), in rural area, the per Capita daily consumption of food

grains and substitutes reaches 616 grams for households with per Capita monthly

expenditure of Rs.170.8. If one gram of food grains, including substitute, gives 3.3

Calories, 2033 Calories per capita per day can be obtained with the consumption of

616 grams of food grains. Another 200 Calories are obtained by these households

from the consumption of items like edible oil, ghee and butter, sugar and gur and a

little of milk, meat and fish. Thus the total intake of food at this level seems to give

about 2250 Calories per capita per day. It implies that, and annual per Capita

consumption expenditure of Rs.170 (at 2004-05 prices) is essential to give a diet

adequate in respect of calories in 2004-05. Now D R find that about 33.12 per cent

of rural population had their per Capita annual consumption expenditure below the

stipulated minimum i.e., Rs.170 at 2004-05 prices.

Regarding urban consumption, D R find from NSS data that urban

households secure the minimum Calories requirement of 2250 at levels where their

consumption of food grains and substitutes reaches 490 grams per person per day.

Again, while a rural household at Rs.170 per capita per annum spends 78.56 per cent

on food, an urban household at Rs.271 per capita per annum spend only 70.26 per

cent on food; again of this 70.26 per cent only 36.45 per cent goes on food grains and

23


24/64


substitutes and the remaining 33.81 per cent is spent on other items of food. Finally,

the prices of food grains which an urban household at an expenditure level of Rs.271

per capita per annum pays are almost 25 per cent higher than the prices paid by a

rural household at expenditure level of Rs.170 per capita per annum. Hence an

annual per capita urban expenditure of Rs.271 is to be regarded as equivalent to an

annual per capita rural expenditure of Rs.170. Now D R find that about 48.64 per

cent on urban population in 2004-05 could not afford the diet consumed by the group

with an annual per capita expenditure of Rs.271. Thus 48.64 per cent of urban

population lived below the poverty line.

According to D R the NSS estimate of average per capita consumption in

2010-11 is an underestimate. What is unacceptable to them regarding the NSS

estimate is that NSS estimate of rural per capita consumption estimate of Rs.239.8 (at

2004-05 prices) in the year 2010-11 is about 7 per cent lower than the corresponding

NSS estimate in 2004-05; again, it is about 11 per cent lower than the consumption

estimate for 2010-11 derived from the national income statistics. Using the national

income deflator D R find that the average per capita consumption of the lowest 5

per cent rural fractile group in 2010-11 is 94 per cent of that in 2004-05, whereas for

the topmost 5 per cent fractile group the estimate of average per capita consumption

is only 73 per cent of that in 2004-05. Now such a large differential decline in per

capita consumption of the richest 5 per cent is not acceptable to D R, as this goes

against their a prior judgment that during the sixties there has been an obvious

increase in inequality in the structure of the Indian rural economy.

It is clear that the position taken by D R rests on whether the NSS estimate

of average per capita consumption is an under estimates or not. On this point

Bardhan (1974)46

is of the view that the deflating of private consumer expendituredata by the national income deflator is improper. Bardhan cites the study of

Radhakrishnan, Srinivasan and Vaidyanathan (1974) who have worked out a general

consumer price index on the basis of NSS weights for ten commodity groups and

wholesale prices data issued by the office of the Economic Adviser to the

Government of India.47 The Radhakrishnan, Srinivasan and Vaidyanathan (R-S-V)

index is 178.1 in 2008-09 taking 2004-05 figure as 100, but for the same period

national income deflator increases from 100 to 169.8. Now if the R-S-V index is

used, not only is the NSS estimate for per Capita real consumption for 2010-11 below

24


25/64


that for 2004-05, but the official estimate for 2010-11 will also be lower than that for

2004-05. Thus Bardhan (1974) asserts that NSS estimate should not be discarded

simply because it establishes a decline in the per Capita real Consumption during the

sixties.

But the observation of D-R regarding the increasing under-estimation of the

consumption of the rich in NSS estimate is endorsed by Bardhan (1974) and other

economists,48 though no satisfactory explanation of this is yet available. The

explanation given by D-R is as follows:

The NSS secures its estimates of consumer expenditures by interviewing a

random sample of rural and urban households and inquiring from them about their

consumer expenditure during the previous month. The limitations of this procedure

are well known. For instance, a number of items of consumer expenditure, such as

clothing and other consumer durables, which a household would purchase less

frequently than once every month, are likely to be missed by this procedure and hence

the expenditure on them is likely to be under-estimated. These are the items which

are more important in the consumer expenditure of the rich than the poor and they

become more important as the rich become richer. It is also known that the upper

middle and the richer households, both in rural and urban areas, have become

increasingly inaccessible to the NSS investigators who are after all class III

government servants.

But the relative infrequency of the purchase of some consumer durables and

articles of high consumption need not lead to underestimation as the NSS is

canvassed in a number of sub-rounds spread throughout the year and so the

seasonality, if any, in the purchase of consumer durables should not lead to any bias.

Moreover, in any given sub-round the random sampling procedure of the survey will

ensure that purchasers of cloth are not systematically excluded.

Now R-S-V have shown that the estimates of consumption at current prices

derived from National Income and the NSS data a are in close agreement for the

period 1954-55 to 1963-64, but thereafter the two series differ considerably. These

authors conclude that the relatively close agreement between the NSS and official

series for some years does not necessarily indicate the absence of systematic bias in

the two series and similarly the divergence between the two series for the later years

25


26/64


of the sixties does not necessarily indicate the presence of such bias. But R-S-V have

not given any explanation for the divergence of the two series from 1963-64 onwards.

Turning now to other studies we find that the study of Vaidyanathan (1974a,

1974b) takes an income level of Rs.132 per year to denote poverty and finds that the

proportion of rural population living below the poverty line is 15.7 per cent in 2004-

05. Now the poverty line drawn by Vaidyanathan is much lower compared to the

estimates of others and there is little surprise that the number of the poor will be

considerably lower.49 However, Vaidyanathan (1974a) has presented the following

estimate of the proportion of rural population with per Capita Consumption

expenditure below Rs.20 per month at 2004-05 prices. According to NSS estimate,

the percentage of people living below the poverty line was 59.5 in 2004-05; it

increased to 67.8 in the year 2010-11. Again, according to official estimate, the

respective percentages are 58.8 and 57.8, which shows that the level of poverty

remained more or less same during the sixties, though the number of people below

the poverty line increased.

The study of Bardhan (1970, 1970a, 1973) uses NSS data for distribution of

consumer expenditure, but uses a different minimum level of income of Rs.15.00 per

month at 2004-05 prices and Rs.28.4 at 2008-09 prices for the two years respectively.

Bardhan then argues that the national income deflator (which rose from 100 in 2004-

05 to 170 in 2010-11) as used by Minhas (1970) and Dandekar and Rath (1971) is not

an appropriate price index to use, as it does not accurately reflect the set of prices

facing the poor consumer.50 This is due to the following reasons. First, national

income includes both investment and consumption goods and as such the national

income deflator cannot be a suitable index for studying changes in consumption.

Secondly, the national income deflator covers the prices the prices of both

agricultural and industrial commodities. The weight of the latter items in the budget

of the poor is much lower than the national average and hence the national income

deflator is very likely to have understated the rise in the prices paid by the rural poor.

This is more so when the prices of agricultural commodities have risen at a much

faster rate than the prices of manufactured items.

For the above reasons, Bardhan uses an alternative deflator which is the

agricultural labour consumer price index. This index is based on the Labour Bureau

26


27/64


series of consumer price index number for agricultural labourers constructed on the

basis of NSS rural retail prices and weighting diagrams obtained from the Second

Agricultural Labour Enquiry. Bardhan finds that the number of the rural poor

increased from about 135 million in 2004-05 to 230 million in 2008-09, that is, the

percentage of the rural population living below the prescribed minimum increased

from somewhat less than 38 per cent in 2004-05 to about 54 per cent in 2008-09.

This result is in sharp contrast to the conclusion reached by Minhas (1970).

It is contended that the high figure of population below the poverty line in

2008-09 may be traced to the consumer price index used by Bardhan. 51 But Bardhan

has defended these indices by computing several alternative indices and showing their

agreement.

The study of Bhatty (1974)52 is concerned with the extent of rural poverty in

2008-09. The object of his study is to present a measure of absolute poverty in terms

of per capita income, which takes into account the inequality of income distribution

among the poor. Moreover, Bhatty is interested about the incidence of poverty in

different regions of rural India and among different classes of the rural population.

To avoid arbitrariness, Bhatty has set five poverty levels, which, in terms of per

capita annual income in 2008-09 prices are: Rs.180, Rs.240, Rs.300, Rs.360 and

Rs.420. Since the study is concerned with a single year, 2008-09, no comparison is

undertaken regarding the change in poverty level of the country. Taking Rs.360 as

the minimum per capita annual income, Bhattys study reveals that about 67.15 per

cent of all rural households lived below the poverty line. If only agricultural labourer

households are considered, then about 82.84 per cent of them live below the poverty

line, whereas for the non-agricultural households this percentage is 69.7. Thus the

incidence of poverty is most severe among agricultural labourers.

Bhatty has also tried to calculate for India the magnitude of Sens poverty

measure P, the result of which is summarized in the following table:

Sens Poverty measure P for India, 2008-09

Poverty line: rupees per Capita per annum180 240 300 360 420

27


28/64


(i) All occupation classes 0.101 0.188 0.279 0.363 0.435

(ii) Cultivators 0.107 0.187 0.269 0.345 0.415

(iii) Agricultural labourers 0.098 0.220 0.336 0.440 0.521

(iv) Non-agricultural workers 0.171 0.155 0.251 0.344 0.425

Source: Bhatty (1974) Tables 7-10, pp. 316-17

From the table we find that of the three occupational classes, agricultural labourersare the most deprived. Bhatty has also analysed in detail the level of poverty in

different states. He has identified the five poorest states in the country - Gujrat,

Tamil Nadu, Madhya Pradesh, Rajasthan and Orissa and explained the region wise

variations of poverty level. He shows that while the incidence of poverty among the

agricultural labourers is the highest in Gujrat and the lowest in Punjab, the non-

agricultural workers are worse off in Madhya Pradesh. According to Bhatty, the

relative incidence of absolute poverty in the rural population of different states

depends on many factors, such as, land-man ratio, topography and quality of land,

rainfall, irrigation, cropping pattern, rural institutional structure etc. Thus this study

points to the fact that any study on poverty should be based on the diversity of

different regions and it is pointless to calculate any uniform measure of poverty for

India as a whole.

The importance of regional study for ascertaining the poverty level has been

emphasized also by Panikar, whose study is based on condition in Kerala. 53 The

author is concerned with the question of choice of a nutritional measure used by

Dandekar and Rath and the Nutrition advisory Committee. Panikar establishes the

fact that D-R have overestimated the number of the poor in Kerala, as the minimum

diet accepted by D-R is in fact the India average.

Indira Rajaraman (1975) discusses the incidence of poverty in the Punjab

between 2004-05 and 1970-71.54 In the year 2004-05, the lowest two deciles of

population accounted for 10.4 per cent of total consumption, while in the year 1970-

28


29/64


71, they accounted for only 8.9 per cent of total consumption. Again, during the

same period the share of the topmost decile increased from 23.2 per cent to 24.7 per

cent. Taking a consumption level of Rs.16.36 at 2004-05 price, Rajaraman finds that

about 18.4 per cent of total population of Punjab lived below that consumption level.

The poverty line for the year 1970-71 has been estimated at a consumption level of

Rs.33086 at current prices, and the author has shown that the percentage of the poor

has increased from 18.4 to 23.3 in ten years. It is seen in this study that poverty is

most intense among the agricultural workers, as about 22.6 per cent of households

lived below the poverty line in 2004-05 and in the year 1970-71, this percentage went

up to about 40.5. In this respect, Rajaramans study is consistent with those of Bhatty

(1974) and Bardhan (1970).55 Like Bhatty, Rajaraman also finds that incidence of

poverty among the cultivators is comparatively low. While the increased incidence of

poverty among the agricultural labour households is significant statistically, this is

not the case with the cultivating households.

In a different type of study based on the socio-economic survey data derived

from the sample households distributed all over Bangalore City, Hanumappa (1978)

has shown that 24.35 per cent of all households can be considered as poor in the

context of their monthly income.56 But the study is mainly concerned with the effect

of family size and education on the pattern of income distribution and so we will not

pursue it further.

We now take up for consideration an important study based on Kerala which

will explain the difference between the conclusions of D-R and Panikar.57 In the

study of D-R, Kerala is seen to have the largest percentage of population below the

poverty line, 90.75 per cent in rural areas and 88.89 per cent in urban areas. The

study on poverty by Centre of Development Studies (1975) has prepared a balancesheet compiled for Kerala and then compares it with the NSS data used by D-R. It is

revealed that part of the explanation for the high proportion of the poor, as given by

D-R, is that the NSS has underestimated certain items of food such as banana,

coconut and fish entering into the diets of these people and as a result the extent of

main nutrition in Kerala has been somewhat exaggerated. Whereas the per capita per

day calories intake is 1619 calories for all food items according to the food balance

sheet of this study (1975). Thus, the study reveals that 48 per cent of the population

in Kerala lived below the poverty line in 2004-05.

29


30/64


But the important point which the study reveals is that per Capita

Consumption of food does not depend on per capita income alone, so that while a low

consumption of food may indicate under-nourishment, it need not necessarily mean

poverty. Moreover it is shown that per capita consumption of food in a region

depends on per capita production of food and the pattern of the distribution of land

holdings; further, the degree of inequality is negatively correlated with increases in

food consumption. One conclusion of this study is that the availability of food

cannot be treated as a function of income and price alone, but it may depend also on

physical factors such as output in the region and institutional factors such as

distribution of land holdings.

Further, the above study suggests that we require more through scrutiny of

regional variations in diet patterns before we draw the poverty line by counting

calories with the help of size distribution of consumption expenditure derived from

NSS data.

A lack of uniformity in the structure of poverty in India has been highlighted

by Vaidyanathan (1974), D-R (1971), Bhatty (1974) and Bardhan (1973).

Vaidyanathan finds that six States- Andhra Pradesh, Kerala, Madhya Pradesh, Tamil

Nadu, Orissa and Uttar Pradesh had a higher percentage of people below the

poverty line than the national average in the year 2004-05. Again, the level of

poverty is very much lower in Assam and Punjab. But in D-R study the intensity of

rural poverty is greater than the All India average in eight States and these are:

Kerala, Andhra Pradesh, Maharashtra, Tamil Nadu, Assam, West Bengal, Orissa, and

Bihar. Again, barring Rajasthan, U.P., Bihar, Jammu and Kashmir and Assam, the

other eleven States show a greater percentage of the poor living in urban areas

compared to the All India average in 2004-05.

While in Bhattys study a greater concentration of poverty is seen in four

States (Gujrat, Tamil Nadu, Rajasthan and Madhya Pradesh), Bardhans study reveals

that higher levels of poverty have been registered in West Bengal, Bihar, Kerala and

Orissa. Such inter-state variations in the percentage of people below the poverty line

underscores the importance of the study of poverty on a regional basis. One cal also

infer a high degree of correlation between the incidence of poverty and the pattern of

30


31/64


the production of cereals, as, state wise, poverty seems to be higher in the non-wheat

zones stretching from eastern India to the Malabar Coast. While this is just a

reflection based on the available data, such a hypothesis (i.e., incidence of poverty

being highly correlated with the pattern of cereal production) requires further detailed

investigation for its acceptance.

All the above-mentioned studies suffer from two general weakness, if any, in

NSS data will render the studies inaccurate. NSS data are based on stratified random

sampling and so the sample households must represent all the rural and urban

households in the country. But, as definitions have changed to some extent in

different NSS rounds, doubts can be expressed regarding the comparability of NSS

data in different years. Moreover, samples from the same universe give comparable

results, but when sampling is done at two different points of time, the Universe is

bound to change. This is likely to be reinforced by planned economic development.

Whether data from different samples drawn at two separate points of time are

comparable or not is a matter of considerable doubt.

Secondly, though rural consumption in NSS data includes non-monetised

consumption, the latter may be of two types commodities produced at home and

commodities simply derived from nature. While the first category can be valued by

using ex-farm prices, the second category cannot be valued at all. It is our contention

that the second category of commodities represents some sort of free good derived

from nature which is unique in a rural agricultural country with a largely

disorganized and demonetized economy. To an NSS investigator such commodities

are either not mentioned at all or, even when mentioned, are likely to be neglected

because of their largely no-economic character. This is more so because of the

existing studies that the per capita per day barest minimum consumption should be

2250 calories and a sizeable section of Indian population cannot afford it. But the

number of such poor people is increasing, though percentage-wise the size of the poor

may be falling. Should we say that a man may survive without the barest minimum

consumption day after day? Accordingly, either the barest minimum has to be further

lowered, or we should concede the fact that people derived their necessary calories

from the consumption of commodities, only a fraction of which can be brought within

the economic categories.

31


32/64


While the first weakness is common to any method of induction in statistics

i.e., any representative sample may lose its representative character with change in

the universe, second weakness points to the more fundamental fact that the economic

categories used for the measurement of rural consumption should be more broad-

based so that the way of living of the rural poor can be analyzed in greater detail.

Chapter-3

Studies on Inequality

32


33/64


Wide disparity in the living standards of the Indian population in both rural

and urban areas has attracted the attention of a number of economists who have tries

to measure the change in the degree of inequality in the income distribution and that

in consumption expenditure. Moreover, attempts have been male to compare the

degree of inequality existing in India with the same in other countries. A survey of

the existing literature in this respect should be preceded by a discussion of the

following points, as clarification of these points will facilitate the evaluation of the

existing literature.

First, it is difficult to compare the per capita income of a developing country

like India with that of a developed country because of the limited scope of national

income accounts in the former.58 Moreover, objection has been raised in the use of

exchange rates to convert all figures to a common standard and it is recognized that

exchange rates may be poor guide to purchasing power. As the study of David

(1972) suggests the true gap regarding the real per capita income between the

United the United States and other countries is only four-ninths of that indicated by

exchange rate conversion.59

Secondly, while it is difficult to generalize about inter-country differences

because of heterogeneities of different sorts historical, physical and regional in

addition to the purely economic, we can still arrive at some conclusion regarding the

effect of growth on size distribution of income by shifting to inter-temporal

comparisons. This is as follows: Though inequality is generally low in the pre-

industrialization stage, it tends to rise with the growth of towns and cities which

emerge and flourish with capitalistic enterprise and growing commerce.

Concentration of capital occurs with the growth of firms and urbanization increases

regional disparity. But beyond this early phase it is difficult to generalize about the

pattern of the change in the inequality-index as economic development proceeds.

Kuznets maintains that the degree of inequality is lower in the developed economy

compared to the less developed countries.60 Kuznets has been supported by others

and these authors, after a careful scrutiny of the size-distribution of income of some

countries, maintain that the distribution of income tends to be more equal, the longer

33


34/64


and the more thoroughly the country has been exposed to the processes of social and

economic transformation after the advent of industrialization.61

We may now turn our attention to the studies on inequality in India. The RBI

Study (1962) gas two parts: in the first, the method of estimation and the average

state of income distribution during the period 1990-91 to 2000-01 have been

described; and in the second, changes in the income distribution from Period I (1990-

91 to 1954-55) to Period II (1955-56 to 2000-01) has been analysed. This study was

undertaken under the guidance of Ojha and Bhatt (1964). Taking the household as

the income-receiving unit, it attempts to present the pattern of income distribution in

the households sector only, which comprises household, not-corporate business

(including agriculture), and private collectives like temples, educational institutions

and charitable foundations. The household sector is divided into three income

groups: (i) low income group with annual income below or equal to Rs.3000, (ii)

households with an annual income between Rs.3001 and Rs.25,000 and (iii) top

income group with annual income above Rs.25,000.

The essence of the method of estimation used in the study is the integration of

the income tax data with the consumer expenditure data from the National Sample

Survey (NSS). The integration is indirect as the study does not use either the actual

expenditure given in the NSS data or the actual income for those with annual

expenditure equal to or below Rs.3000. Again, the proportion of population and the

size of the households in various expenditure brackets given in the NSS data on

consumer expenditure are used for the following: First, to estimate independently

from the population data (i) the distribution of rural and urban households between

low income groups and high income groups, and (ii) the total number of households

in the rural and urban areas separately. Secondly, to estimate independently from the

national income data the distribution of personal consumption expenditure between

the rural and urban sectors and between low income and high income groups within

each sector. The savings made and taxes paid are then added to derive personal

disposable income and personal income respectively of the various income groups.

Moreover, personal incomes accruing to households in the top income group

are obtained directly from the income tax data on the assumption that each salary

earner assessed to tax represents one household. Income tax data are also used in the

34


35/64


estimation of personal income accruing to households in the high income non-salary

earners group both in urban and rural areas in so far they are obtained as residual

magnitudes. The distribution of incomes and households in this group is also done

from income tax data. The independently estimated households and incomes are then

put together to derive the pattern of income distribution.

The study reveals that for the period 1990-91 to 2000-01 the top decile

accounts for 28 per cent of personal income, while the bottom decile obtains only 3

per cent. The Lorenz ratio for disposable income is only slightly lower (0.335) than

that for personal income (0.340). Income distribution is more uneven in the urban

sector than in the rural sector; the urban sector concentration ratio for personal

income is 0.40, while this ratio for the rural sector is only 0.31.

Ojha and Bhatt (1964) then conclude as follows:

Contrary to general impression, the degree of inequality in income

distribution in India does not seem to be higher than in some of the advanced

economies.

This conclusion of the authors goes against the thesis of Kuznets and the empirical

findings of Morgan (1953) and others. This view has been challenged by Ranadive

(1965), Swamy (1965) and Mueller and Sarma (1965). Though the debate is primarily

concerned with the above conclusion of Ojha and Bhatt is primarily concerned with

the above conclusion of Ojha and Bhatt rather than with the index of inequality as

revealed in their study, this has cast some doubt on the basis on which the study of

Ojha and Bhatt depends.

According to Ranadive (1965), the RBI study is marred by a lack of

appreciation of the need for an appropriate concept of personal income, an incorrect

use of NSS data for deriving size distribution of household income and by what

seems to be a methodological error which has resulted in over-estimation of

households in the high income groups. A close scrutiny of the RBI study shows that

the number of households in the high income group is over-estimated, which is

revealed by the fact that the number of households in non-salary earners group is

about six times higher than the number of the corresponding group in the tax data.

Again, according to Ranadive, the concept of income in the study excludes some

elements and so the personal disposable income in high income groups is likely to be

35


36/64


underestimating. Thus the under-estimation of income and/or over-estimation of

households in the higher income groups might account for the sharp difference of this

study with the thesis of Kuznets (1963) supported by some empirical studies.

Mueller and Sarma (1965) have pointed out that the assumption in the study

of Ojha and Bhatt leads to a downward bias in income inequality and they have

ignored an important body of data, which is a survey conducted by NCAER in 2004

with a stratified probability sample of 4400 families in 30 cities and towns all over

India.62 Mueller and Sarma have shown that NCAER income distribution is much

more unequal than the OJha and Bhatt distribution. Moreover, the saving estimates in

RBI study do not correspond with the NCAER estimate. Criticising the thesis of

Ojha and Bhatt, Swamy (1965) also contends that the pattern of income distribution

in India supports the thesis of Kuznets.

Lydall (2004) assumes that Pareto Law of distribution63 holds in India. He

then makes use of income-tax statistics of individuals and Hindu undivided families,

and converts the NSS 10th round data from household to a per-person basis, the

income of each household being divided by the number of persons in that household.

He further assumes that average number of persons covered by each tax assessment is

about three. The result of his study is as follows. The top ten per cent of Indians

account for 34 per cent of pre-tax income and 33 per cent of post-tax income in 1955-

56. The corresponding figures for United Kingdom in 1954 are 30 per cent and 25

per cent respectively. But Lydall is cautious regarding any comparison of income

distribution because coverage of income-tax is much smaller in India compared to

U.K. and the estimate of direct income distribution in India is absent.

Since Lydalls study is based on income tax data, we do not get any reliable

picture of the pattern of income distribution as a whole in the fifties.

Mukherjee and Chatterjee (1967) have utilised NSS data and the national

income estimates to indicate broadly the behavior pattern of income distribution. The

premises of the study are as follows:

(i) First, statements are made about the nation as a whole and hence reliance

has been placed on national income and allied information which relate to

the country as a whole.

(ii) Secondly, the reference period is 1950-51 to 1965-66.

36


37/64


Moreover, at attempt has been made to construct size distribution of

income at constant prices.

The study reveals that disparity of private consumption expenditure at current

prices showed some reduction at the All India level during the period 1990-91 to

2005-62. But disparity of private consumption expenditure in real terms showed a

large increase from the First to the Second Plan period and then maintained a high

level. Again, the evidence is not conclusive on the movement of inequality in the

distribution of personal income by size reckoned at current prices, but the

overwhelming suggestion is that there has been some increases in inequality after

1990-91 and also towards the end of the period. Moreover, there has been a marked

increase in disparity in distribution of personal income by size reckoned in real terms

throughout the period.

Mukherjee and Chatterjee implicitly assume that prices of cereals and non-

cereals change in the same direction and also by the same proportion. While the first

part of the assumption is generally true, the second part is not borne out by facts. 64

Again, NSS data give changes in the prices in the prices of individual cereals as well

as shift in the composition of cereals and this must be adjusted to get a proper index

for deflating the consumption expenditure.65

Swamy (1967) has adjusted the NSS data for reference period biases and

differences in valuation. He establishes that inter-sectoral inequality is connected

with the shift in the size distribution and the overall size distribution of income has

little meaning unless it is examined along with the components of this distribution.

According to Swamy about 85 per cent of the increase in inequality in the size

distribution of consumer expenditure over the decade 1951-60 was due to structuralshift in the economy and only about 15 per cent was due to intra-sectoral inequalities

changes in inequality.66 Thus Swamy emphasizes the importance of the study of

structural parameters, as a study of intra-sectoral inequalities alone would not truly

indicate the changes in the size distribution of income for the country as a whole.

The implication of Swamys study is that the process of industrialization

causes shifts in relative weights of different sectors; and without radical changes in

the institutions, inequality in the income distribution must inevitably rise over time.

37


38/64


Neglect of this aspect, according to Swamy, is the principal cause why some income

distribution studies show a decline in the disparity in the income distribution. Further,

Swamy has estimated that inequality remained more or less stable in rural areas, but

increased in urban areas, which is reinforced by the fact that the proportion of

population increased in urban areas over the period. This increased the disparity

between rural and urban areas.

Modifications tot the Ojha and Bhatt method have been suggested by

Ranadive (1973) to allow for not dissaving by poorer income groups and for possible

tax evasion in the top income groups. Ranadive adopts two extreme alternatives: (i)

where households with annual income less than Rs.2000 in the urban sector and with

income less than Rs.720 in the rural sector are assumed to have zero net savings and

all evaded tax payments are assumed to have zero net savings and all evaded tax

payments are assumed to be fully reflected in consumption and/or savings, and (ii)

where households with annual income less than Rs.3000 in the urban sector and with

income less than Rs.1200 in the rural sector are assumed to have negative savings,

which constitute 25 per cent of the total urban savings in the case of the former and

14 per cent of the total saving in the case of the former and 14 per cent of the total

savings in the case of the latter. As for evaded tax payments, they are not reflected in

consumption and/or savings, so that the estimated amount of tax evasion is added to

the disposable income of the tax-paying classes. Now case (ii) should show higher

inequality than case (i) due to the assumptions relating to savings and tax evasion.

Ranadives estimate shows that in the year 2005-62, the bottom 20 per cent of

population accounted for 7.6 to 7.8 per cent of total income, and top 20 per cent

accounted for 45.5 to 46.7 per cent. The Lorenz ratio was between 0.351 and 0.367.

The assumption of Ranadive that in the savings group total saving is distributed in

proportion to consumption expenditure has been questioned by Bardhan (1974).

Ahmed and Bhattacharya (1972) have tried to integrate the size distribution of

consumer expenditure obtained from NSS data with the size distribution of income

before tax, obtained from income tax data, to estimate the size distribution of per

capita personal income in India in three different periods, 2000-01, 2004-05 and

1963-64. They have followed the technique developed by Lydall (2005) and their

study is a more systematic and rigorous extension of the earlier attempt of Ahmed

38


39/64


(1971). This approach is based on two assumptions: (i) income (before tax) equals

consumer expenditure in the lower ranges of per capita consumer expenditure, and

(ii) the distribution of per capita personal income before tax is symptotically Paritian

for high values of per capita income and has the same slope as the distribution of

assesses by size of incomes before tax.

The results of the study of Ahmed and Bhattacharya are as follows: For the

first fit, where Pareto Curve is fitted to the size distribution of income before tax

taking all income classes above Rs.20,000, the Lorenz Ratio is 0.418 for 2000-01,

0.379 for 2004-05 and 0.372 for 1963-64. Again, for the second fit, that is where

Pareto Curve is fitted taking the last interval of income before tax as Rs.100,000 and

above, the Lorenz ratio is 0.408 for 2000-01, 0.382 for 2004-05 and 0.361 for 1963-

64. However, this result has been qualified by the authors with two points: First,

considering the fact that price increases have been more sharp for the lower income

groups than for the higher income groups, this decreases in the inequality in nominal

income distribution may be more illusory than real. Secondly, this decline, the

authors suspect, may be traced to the inherent weaknesses of the two sets of data.

Bardhan (1974) points out that the first assumption of Ahmed and

Bhattacharya rules out dis-savings in the lower income brackets and so it leads to

some understatement of inequality. Moreover, considering the fact that the number

of income tax assesses is not even 1 per cent of Indian population and rural rich are

mostly beyond the net of income tax authority, the technique of fitting Pareto

distribution in the Indian contest may very well distort the picture.

On the basis of NSS data Vaidyanathan (1974) analyses inter-State variations

i

copy of poverty and inequality in india

Documents