Surjit S Bhalla: In(equality) we believe

CALLING THE BLUFF

The next few articles will survey the evidence in India on the contentious trinity issues""what happened to inequality, growth, and poverty in India before, and after, the introduction of major economic reforms in the early nineties.

This examination is critical because the present Indian government has raised the stakes""it believes that past reforms did not have a human face, i.e. reforms to date have made "the rich richer, the poor poorer".

Translated, this means that two propositions hold: first, incomes of the poor have proceeded at a slower pace than incomes of the rich; second, that the speed of poverty decline has not kept up with growth.

The first proposition has to do with changes in inequality, the second with changes in poverty (the subject of the next article). We all have our views about the magnitude of inequality we find tolerable; however, the only tolerable level of poverty is zero.

Today, I shall talk about changes in inequality, i.e. what has happened to the pace of growth in incomes of various strata of the population.

Note that this discussion says nothing about the level of absolute poverty. So what views you have on the nature of poverty in India, save them for the next installment of ideology.

The official source of information on inequality (and poverty) is the household surveys conducted by the National Sample Survey Organisation (NSS).

Data provided are very unambiguous: inequality did not increase in India. Somewhat surprisingly, this simple, accounting result has been hotly debated: internationally respected experts believe, on the basis of the very same NSS data, that the truth is very different.

They assert that inequality has significantly worsened. How is it possible that the same data yield such divergent accounting results? Read on. And remember, this discussion is not tainted by one's ideological views on political leadership ""the analysis ends in 1999-00, and so ignores the impact of the BJP/NDA on inequality, etc.

If inequality worsens, then that is equivalent to the statement that economic growth has not been shared evenly, that trickle-down has failed to materialise, that the rich have gotten relatively rich, etc.

So inequality worsening is a "bad", is suggestive of government action to curb the rich, increase their taxes, etc. My own take on discussions on inequality is that it is equivalent to the "economics of envy"; what should really concern us is the growth in incomes of the poor. But I am not here to defend that position, but to present the conflicting interpretations of inequality change in India.

In contrast to the international experts, NSS data suggest, if anything, that inequality in India has shown a slight improvement. One inequality measure is the Gini, an index whose value is 100 if one person has all the income, and zero if everyone has the same income.

India's Gini level of 32.6 in 1993-94 places it amongst the top 10 per cent (rank 11 out of 86 with rank 1 being the most equal).

In 1999, India's Gini improves to 32 and its world rank improves to 6. So relatively, India's position improved. If instead of Gini, alternative inequality measures are used (e.g. the variance in log expenditures or the share of expenditures accruing to the poor), the same result holds""improvement: i.e. the poor increased their expenditures at a faster rate than the rich.

So no matter how or where one looks, the extremely robust conclusion is that inequality did not worsen in India; and may have marginally improved after the advent of reforms in India in 1991. So how come the pervasive view, especially in the development international aid NGO community (Oxfam, World Bank, IMF, UNDP, ADB, etc.), that inequality worsened in India, and did so certainly after economic reforms?

Perhaps because several inequality-poverty experts have argued so. First off the block was World Bank poverty-inequality expert Martin Ravallion; in an influential article he stated that there was "compelling evidence of rising consumption inequality in the NSS data for 1983""1997". What was this compelling evidence?

His large inequality estimates for two years: 1994 and 1997. It turns out that Ravallion's Ginis are close to those derived from NSS unit-level data for all years except 1994 and 1997. For these exceptional end-of-period years, the estimates are very, very different.

For 1994 and 1997, Ravallion computes Ginis of 36.3 and 37.8, respectively, whereas official unit-level data yield Ginis of 34.9 and 34.1. Ravallion's conclusion that inequality in India increased appears to depend entirely on reliance on calculations that are not accurate.

Two other studies (senior economic adviser to the UPA government Jean Dreze is associated in these with both Nobel Laureate Amartya Sen, and distinguished economist Angus Deaton) also conclude that inequality worsened in India in the nineties.

Both studies emphasise that inter-state disparities in income have increased, and that urban incomes have increased at a faster pace than rural incomes. Even if both the statistics are true, neither statistic may have much to do with trends in aggregate inequality. And this theoretical result is empirically supported by the NSS data: no inequality measure based on it shows an increase in overall inequality.

There is little to support the bold conclusion reached by Dreze""Sen: "Note should also be taken of the related issue of rising economic inequalities in the nineties.

...Given the adverse social consequences of economic inequality (ranging from elitist biases in public policy to the perpetuation or reinforcement of other types of inequality), this accentuation of economic inequalities, from an already high base, is not a trivial matter" (Dreze""Sen, p. 329; emphasis added). And the above data on India's rank suggest that the high base that the authors talk about is actually very low.

Deaton""Dreze (DD) base their claim that there was a "pervasive increase in inequality" on estimates derived from some adjustments to the NSS data. According to their calculations, adjusted inequality in 1999-2000 [variance of logs (VL) equal to 0.32] is 10 per cent higher than unadjusted inequality in 1993-94 (VL equal to 0.29).

The method used to adjust the 1999-00 data is not of concern here [see Bhalla (2003) for a detailed analysis]. It should be emphasised, however, that there is nothing wrong with the method; indeed, it is ingenious.

The problem with the DD study is that it does not compare like with like. DD compare unadjusted inequality for 1993-94 with adjusted inequality for 1999-00, whereas they should be comparing adjusted inequality for both the years.

If this is done, using exactly the DD method of adjustments, then one finds that there is virtually no change in inequality between 1993-94 and 1999-00 (VL of 0.31 and 0.32 for the two years).

Thus we see that there is no credible evidence to support the conclusion, offered by Ravallion""Dreze""Sen""Deaton (and accepted by ideological types in the UPA government) that inequality has worsened, let alone that India has witnessed a pervasive increase in inequality.

The reality is just the opposite, at least as revealed by the government's own NSS data: per capita expenditures of the poor have grown at a slightly faster pace than those of the rich.

It is likely that sometime during its course of development, India will witness increasing inequality. It is equally likely that India's past experience has been highly unusual.

But there is no reason to cry wolf before such an increase has occurred. It may be a politically correct position""but that does not mean it is right.

ssbhalla@oxusresearch.com