Tue09262017

Last updateSat, 22 Jul 2017 6am

You are here: Home | Poverty | Poverty | Analysis | Prosperity and poverty: A rapid reader on inequality

Prosperity and poverty: A rapid reader on inequality

Increasing prosperity for small sections of the population has been misinterpreted as signifying diminishing levels of poverty for large sections of the population, say S Subramanian and D Jayaraj

 Prosperity and poverty

Photo by Steve McCurry, courtesy Caravan

…were an estimation to be made of the charge of Aristocracy to a Nation, it will be found nearly equal to that of supporting the poor… In stating these matters, I speak an open and disengaged language dictated by no passion but that of humanity.
– Thomas Paine: Rights of Man

Introduction

This note is intended to be no more than a microscopic primer on the subject of inequality in the distribution of household assets in India. This essay is based on earlier work done by the present authors, and published in 2008 and 2013. As such, there is little that is original in this note. Even so, it might serve the purpose of a quick reminder of the fact that we live in a society marked by deep economic inequality, a fact that is only underlined and magnified by the presence of massive levels of poverty at one end of the spectrum, and massive levels of unaccounted income and wealth, at the other. If this is a matter of labouring the obvious, then we would rather labour than not.

Growth in household wealth

Data on the distribution of household wealth are available, over 10-year intervals, at least from 1961-62, in the surveys on debt and investment conducted by the Reserve Bank of India (RBI) and the National Sample Survey Office (NSSO). Table 1 furnishes information, garnered from the sources just mentioned, on the trend in the average value of assets per household. The figures in Table 1 indicate a systematic rise in the real value

 Table 1  The Growth in Household Asset Holdings: Rural and Urban India: 1961-62—2002-03

Year

Value of Assets per Household
(In Current Rupees):
Rural

Value of Assets per Household
(In Current Rupees):
Urban

Value of assets per Household (In 1981-82 Rupees):
Rural

Value of assets per Household (In 1981-82 Rupees):
Urban

1961-62

5,267

-

27,290

-

1971-72

11,343

-

30,740

-

1981-82

36,089

40,566

36,089

40,566

1991-92

107,007

144,330

51,570

69,557

2002-03

265,606

417,158

66,640

104,664

Note: All real values are in 1981-82 prices employing the Wholesale Price Index as deflator.
Source: From Table 4 of Subramanian & Jayaraj’s 2013 paper in Journal of Globalization and Development and Table 6.3 of these authors’ 2008 article in a book edited by Jim Davies. Computations are based on data in the relevant decennial RBI/NSSO surveys on Debt and Investment.

of assets per household. (By the ‘real value’ is simply meant the value adjusted for price changes.) Notice that the growth in the value of assets per household is considerably sharper in the urban areas than in the rural areas. From 1981-82 to 2002-03, the real value of assets per household has risen by a factor of less than two in rural India, and by a factor of more  than two-and-a-half in urban India. The reason is not hard to see. Urban India, after all, is the home of industrial and financial capital, and the growth in the wealth of identifiable corporate houses, as reflected in the Business Standard’s list of the wealthiest households/entities, has been phenomenal.

Subramanian and Jayaraj’s 2008 study reveals the following. If we combine the Business Standard data on the ultra-wealthy with the official survey data for 2002-03, then the suggestion is that, at the all-India (rural-cum-urban) level, the wealthiest 178 households accounted for 0.00009% of all households, and for 2.05% of the country’s estimated wealth What this means is that the ratio of the Business Standard’s ultra-wealthy entities’ share in the value of assets to their share in population was a staggering 23,239! Such acute concentration of wealth is a reflection of prosperity in a small island amidst a vast ocean of generalized deprivation. Unfortunately, the presence of such prosperity  -- especially in the urban areas of the country – has often been taken to be a symptom of the relative absence of poverty. The truth is that high levels of prosperity for a minuscule section of the urban population comfortably co-exist with high levels of poverty for vast numbers of people at the lower end of the wealth distribution profile. This is nowhere more apparent than in the fact that buildings and real estate are the most important component of household assets in urban India: the visual contrast afforded by the shanties and huts and slums of urban India, on the one hand, and its residential palaces and shopping malls and skyscrapers, on the other, is a bleak and direct pointer to the levels of inequality we so easily accommodate in our lives.

It appears reasonable to believe that the largest absolute increases in wealth would have accrued to the wealthier households, though this may well be compatible with roughly equal proportionate increases across all wealth size-classes. To make our meaning clearer, consider a two-household world in which the wealthier household has an asset base of Rs 100, and the poorer household an asset base of Rs 10. If now the  wealthier household’s asset base increases to Rs 150, and that of the poorer household to Rs 15, then both households will have experienced an identical 50% increase in their respective asset levels, although the richer household would have gained an additional Rs 50 in comparison with the very small gain of Rs 5 for the poorer household. Growth will be judged to have been equalizing if we focus only on rates of growth, and dis-equalising if we focus only on absolute amounts of growth. Our judgment on trends in inequality would therefore be mediated by the fact of the nature of class-specific increases in household wealth, and by our interpretation of whether inequality is to be conceptualized in essentially relative or essentially absolute terms. We turn now to this issue.

Alternative measures of inequality: relative, absolute, and intermediate

A relative measure of inequality (which is the sort of measure that is overwhelmingly favoured in the bulk of the measurement literature) is one whose value remains unaltered when we have an equal proportionate increase in all wealth levels. An absolute measure of inequality is one whose value remains unaltered when we have an equal absolute increase in all wealth levels. An illustration based on an elementary two-person world may again help. Here is an example, first, of a relative inequality measure (call it R), and then of an absolute inequality measure (call it A), for such a two-person world. Suppose R is given by the difference in the two individuals’ wealth levels expressed as a proportion of the total wealth, and A by – simply – the difference in the two persons’ wealth levels. Consider the two-person distribution x = (10,20), which tells us that in the distribution x, the poorer person has a wealth level of 10 and the richer person a wealth level of 20. Consider also the distributions y = (20,40) and z = (20,30). It is easy to see that y has been derived from x through an equal proportionate increase in both persons’ wealth levels (their wealth holdings have both doubled), while z has been derived from x through an equal absolute increase in both persons’ wealth levels (their wealth holdings have both increased by Rs 10).

Notice now that the value of the inequality measure R for the distribution x is (20-10)/30 = 0.33, while for the distribution y, R is (40-20)/60 = 0.33. The relative character of the measure R is reflected in the fact that its value (at 0.33) has remained unaltered by en equal proportionate increase of all incomes in the change from x to y. Notice, at the same time, that the value of the absolute measure A is 10 (= 20-10) for distribution x, which rises to 20 (= 40-20) for distribution y. Further, the value of A for the distribution z, at 10 (= 30-20) is the same as for the distribution x (20-10 = 10): the absolute character of the measure A is reflected in the fact that its value has remained unaltered by an equal absolute increase of all incomes in the change from  x to z. As far as the relative measure R is concerned, however, its value declines from 0.33 for the distribution x to 0.2 [= (30-20)/50] for the distribution z. The example we have employed clearly demonstrates that whether we judge inequality to have increased, declined, or remained constant in the presence of growth would depend very much on whether we favour a relative or an absolute measure of inequality.

The French economist Serge-Christophe Kolm, in a couple of papers written in the Journal of Economic Theory in 1976, referred to relative measures as ‘rightist’ ones, and to absolute measures as ‘leftist’ ones. This is because relative measures do not register any increase in value with an equal proportionate increase in all wealth levels, even though this is consistent with a larger absolute increase the richer the wealth-holder is; and absolute measures do not register any decline in value with an equal absolute increase in all wealth levels, even though this is consistent with a smaller rate of growth the richer the wealth-holder is. This ‘rightist/leftist’ characterization, of course, holds only for a situation attended by wealth-growth, not wealth-regression. The qualification will not be repeated in what follows: it will be understood that what is of interest to us is a situation of growth, not decline, in average wealth.

One may regard the values underlying relative and absolute measures to be somewhat extreme, postulated, as they are, on either end of the ideological spectrum. A more moderate approach to inequality conceptualization is an intermediate one, in terms of which inequality is regarded as increasing with an equal proportionate increase in all wealth levels, and declining with an equal absolute increase in all wealth levels.

A well-known relative index of inequality is the (relative) Gini coefficient. An example of an absolute index of inequality is the absolute Gini coefficient (advanced by Moyes in a 1987 paper, and given simply by the product of the mean level of wealth and the relative Gini coefficient). It is not important for us to know precisely how the relative and absolute Gini measures are actually derived or computed, or what the formulae for them are. It suffices to know that one is a relative – and therefore ‘rightist’ – measure of inequality, while the other is an absolute – and therefore ‘leftist’ – measure of inequality. It turns out that one can derive an intermediate Gini coefficient by simply multiplying the relative and absolute Ginis: the resultant index will display the property of registering a rise in inequality with an equal proportionate increase in all wealth levels, and a decline in inequality with an equal absolute increase in all wealth levels. (This, of course, is the defining property of an intermediate inequality measure.) An intermediate inequality measure such the intermediate Gini steers clear of the ‘extreme’ values of right and left on which relative and absolute inequality measures are, respectively, predicated. There is, therefore, a strong case in favour of pursuing this moderate ‘middle path’ in our empirical assessment of inequality in the distributions of consumption expenditure and of wealth.

Trends and magnitudes of alternative versions of the Gini measure of inequality

Table 2 presents information on the magnitudes, over time, of alternative Gini measures of inequality, separately for the rural and urban areas of the country. It is easily discernible that while the levels of inequality in the inter-household distribution of assets are high, the relative measures display no over-time increase – and, indeed, a definite decline, in the urban areas, over the three decades from 1981-82 to 2002-03 (the values of the relative Gini coefficient in 1981-82, 1991-92 and 2002-03 are, respectively, 0.70, 0.68 and 0.66). This picture is very sharply inverted by the trends in the absolute measures. This suggests that our notion of whether or not growth has been inclusive is crucially dependent on whether we are looking for equal rates of growth across wealth classes, or equal absolute increments of wealth across wealth classes. It could be argued with some justice that the former criterion is altogether too conservative, while the latter may be judged to lean ‘excessively’ in the direction of liberalism. Consequently, we may not wish to base our inferences entirely on the behaviour of either relative or absolute measures of inequality.

Table 2 Inequality in the Distribution of Household Assets: Three Versions of the Gini Coefficient: Rural and Urban India, 1961–62—2002–03


Year

The Relative Gini Coefficient

The Absolute Gini Coefficient

The Intermediate Gini Coefficient

Rural

Urban

Rural

Urban

Rural

Urban

1961-62

0.6440

-

17,574.76

-

11,318.15

-

1971-72

0.6564

-

20,177.74

-

13,244.67

-

1981-82

0.6354

0.7037

22,930.95

28546.29

14,570.33

20,088.03

1991-92

0.6207

0.6805

32,009.50

47333.54

19,868.30

32,210.47

2002-03

0.6289

0.6643

41,909.90

69528.30

26,357.13

46.187.65

Note: The absolute and intermediate Gini coefficients are presented in 1981-82 prices. The relative Gini coefficient has been computed by employing the ‘trapezoidal approximation method’ on grouped distributional data.
Source: From Table 5 of Subramanian & Jayaraj’s 2013 essay in Journal of Globalization and Development, Table 1 of Subramanian & Jayaraj’s 2013 paper in Challenge, and Table 6.3 of Subramanian & Jayaraj’s 2008 essay in the Jim Davies-edited book Personal Assets from a Global Perspective. Computations are based on data in the relevant decennial RBI/NSSO surveys on Debt and Investment.

A more moderate view, surely, is afforded by an intermediate conceptualization of inequality. Table 2 clearly indicates that even such a ‘centrist’ view of inequality yields a picture of pronounced over-time increase in inter-household inequality in the distribution of asset-holdings: the intermediate Gini coefficient displays a steady increase in value over time, in both the rural and the urban areas of the country. Indeed, the urban trend of the intermediate Gini precisely inverts the trend of the relative Gini. Briefly, these results are in striking contrast with conventional wisdom on the subject, which is informed almost entirely by a relative, ‘rightist’ view of inequality.

Some conclusions
The somewhat obvious conclusions which this essay yields are summarized in the following points:

  1. Purely relative measures of inequality do not suggest any increase in inequality in the distribution of household assets in India.
  2. Absolute measures of inequality sharply reverse the trend displayed by relative measures.
  3. Arguably, intermediate measures of inequality reflect more ‘reasonable’, and certainly more ‘moderate’, values than either ‘rightist’ relative measures or ‘leftist’ absolute measures. As it happens, the centrist ‘intermediate Gini’ also reflects a clearly increasing trend of inequality in the inter-household distribution of assets.
  4. Both magnitudes and trends of inequality, as reflected by distributional data in official sample surveys, are likely to be understated for various reasons, including underestimation of wealth in the upper tail of the distribution, and a burgeoning unaccounted economy characterized by under-reporting – again at the upper end of a distribution -- of the extent and value of ownership of land and gold, illegal foreign exchange transactions, and false invoicing of imports and exports.
  5. In view of these observations, it is fair to suggest that India is a country with a very large and increasing level of inequality in the distribution of household wealth.
  6. It is important to note that the inequality we speak of relates to an economy with an enormous amount of both income and non-income deprivation in it: inequality in such circumstances would be seriously viewed even by a relatively ‘conservative’ distributional ethic such as ‘sufficientarianism’ (see Frankfurt’s  1987 essay in the journal Ethics) which, loosely, finds inequality objectionable in in the sense, and to the extent, that it co-exists with poverty.
  7. The growth in average levels of wealth and income in this country has been greeted by its elite classes with a great deal of enthusiasm and fanfare (though even the income-growth story is now beginning to unravel). It is a particularly harsh symptom of both ignorance and insensitivity that the presence of large and increasing amounts of prosperity restricted to small sections of the population has been misinterpreted as signifying small and diminishing levels of poverty for large sections of the population. The presence of prosperity is not the same thing as the absence of poverty. Both can co-exist, and do so, quite blatantly, in our country -- and certainly not less so in urban India than in rural India.
  8. One would imagine that a situation such as the one which obtains in our society should attract some serious measure of land-reform; enhanced direct taxation (especially of corporate and agricultural income, and on wealth and bequests); and a minimally principled effort to contain a burgeoning unaccounted economy. What we have, instead, is a failed programme of land-reform (now a matter of the distant past); an under-taxed economy in which budgetary deficits are sought to be addressed by reducing already low levels of spending on the social sector and capital creation, not to mention promises of a complete abolition of income-tax for all but the wealthiest; and a remarkably permissive environment of political, bureaucratic, and corporate fraud and non-compliance.
  9. Apart from the intrinsic moral and political unsustainability of large levels of inequality in wealth holdings and incomes, there are instrumental reasons to set one’s face against such trends of increasing concentration. These have to do with  the deleterious effects of inequality on, among other things, efficiency, social cohesion, aggregate demand, and public health outcomes. These are matters which will not even come up for public discussion and debate if we continue to employ protocols of measurement which are heavily biased in favour of ‘conservative’ relative measures of inequality.
  10.   Finally, it cannot have escaped the reader’s attention that this essay does not enlighten those who already know its drift, which is almost everybody; but it might have the effect of an annoying reminder on those who refuse to acknowledge it, which is no insubstantial constituency.

S Subramanian is an ICSSR National Fellow affiliated to the Madras Institute of Development Studies, and D Jayaraj is a Professor of the Institute  

Recommended further reading
Frankfurt, Harry, (1987), “Equality as a Moral Ideal”, Ethics, 28 (1), 21-43.
Kolm, Serge-Christophe. (1976a), “Unequal Inequalities I”, Journal of Economic Theory, 12(3), 416-454.

Kolm, Serge-Christophe (1976b), “Unequal Inequalities II”, Journal of Economic Theory, 13(1), 82-111.

Krtscha, Manfred (1994), “A New Compromise Measure of Inequality”, In Wolfgang Eichhorn (ed.), Models and Measurement of Welfare and Inequality (pp. 111-120), Heidelberg: Springer-Verlag.

Moyes, Patrick (1987), “A New Concept of Lorenz Domination”, Economics Letters, 23 (2), 203-207.

Subramanian, Sreenivasan & Dhairiyarayar Jayaraj (2008), “The Distribution of Household Wealth in India”, In James B. Davies (ed.), Personal Wealth from a Global Perspective (pp. 112-133), Oxford: Oxford University Press.

Subramanian, Sreenivasan & Dhairiyarayar Jayaraj (2013a), “Economic Inequality in India: Value-Neutral Measurement?”, Challenge: The Magazine of Economic Affairs, 56(4), 26-37, July-August.

Subramanian, Sreenivasan &  Dhairiyarayar Jayaraj (2013b),  “The Evolution of Consumption and Wealth Inequality in India: A Quantitative Assessment”, To appear in Journal of Globalization and Development. (Published Online: 11/29/2013)

www.infochangeindia.org, May 2014