Take Piketty with a pinch

While entirely appropriate for developed economies with good tax data, the Piketty method of estimating income distribution is entirely inappropriate for developing economies with bad tax data.

Written by Surjit S Bhalla | Published:January 23, 2016 12:17 am
Illustration by C R Sasikumar Illustration by C R Sasikumar

On his visit to India, Thomas Piketty, quoting the Credit Suisse (CS) report and his own considerable work on income inequality, stated that “India has one of the worst inequalities in income and wealth globally”, in comparison with high-inequality countries like South Africa and Brazil. Is this true?

First things first. Wealth, as defined by the CS report, is the “value of financial assets plus non-financial assets (principally housing) owned by individuals less their debts”. Given that human capital is a major wealth asset

for most individuals in the world (other than the top 1 per cent in the Western world or the 62 world billionaires of which, the CS report kindly informs us, 53 were men), this is a rather glaring omission.

It is likely that if education were included as wealth (why not?) the results would not be as drastic, and therefore less “saleable”, as the following “rich are getting richer, poor are getting poorer” polemical CS and Oxfam conclusions: First, while the bottom half of adults collectively owns less than 1 per cent of total wealth, the richest decile holds 87.7 per cent of assets, and the top percentile alone accounts for half of total household wealth. Second, the richest 1 per cent have now accumulated more wealth than the rest of the world put together. Third, since the turn of the century, the poorest half of the world’s population has received just 1 per cent of the total increase in global wealth, while half of that increase has gone to the top 1 per cent.

In contrast, while Piketty approvingly quotes the heavily flawed CS report, his and his associates’ work on income inequality is brilliant and compellingly correct. Piketty uses income tax data to derive inferences and conclusions about income inequality. The problem with household data, as used by CS, is that individuals tend to understate their income. Income tax data, on the other hand, is considerably more accurate about the distribution of income.

Piketty speculates that income distribution in India is likely to be similar to that in Brazil and South Africa; that is, among the worst in the world. According to the latest household data, India would place among the five most unequal economies in the world. The South African Gini of 70.1 is the worst and Brazil’s Gini of 60 is close to the 10th worst.

Is Piketty right in his speculation on inequality in India? There are two methods of estimating household income inequality in the absence of credible household data. The first is by estimating it from available consumption inequality data; the statistical rule of thumb is to add six Gini points to consumption inequality. This would place Indian inequality at around 42; that is, squarely at the world mean (and median) of household inequality. A movement of India from the worst percentile to the 50th percentile, if correct, is a very large speculative error.

There is, of course, the Piketty method of deriving income distribution from tax data. There are two major problems with the use of income tax data (even when available!) for countries like India. The first, widely noted problem is that in India, very few individuals are in the tax net and, of these, only about half pay taxes. In case you are dumbfounded by these low figures, note the following: The universe of Indian taxpayers is not the total population but the worker population, and this is 40 per cent of the total. Second, only non-farmers pay income taxes and farmers are about 10 per cent; so the relevant population universe is 300 million, not a billion. Further, of these 300 million, only about 80 million workers are eligible to file tax returns — the eligibility criteria is based on a minimum income level of Rs 2.5 lakh a year. This means that even with full compliance, there will be tax data for only 7 per cent of the population. But compliance ratios in India, though improving, are still around 50 per cent; that is, Piketty, and I, and others, will only have tax data for 3.5 per cent of workers in India. One would have to be bravely foolish to estimate an income distribution on the basis of such scarce tax data. It can be done but only with recourse to household distribution data, which the tax-based income distribution logic explicitly rejects.

And even with heroic assumptions, the estimate of tax-based income inequality in India is likely to be an overestimate, possibly of large proportions. This is because an examination of Indian tax data will reveal that tax compliance in India is U-shaped — people with incomes between Rs 2.5 and 5 lakh have relatively high compliance ratios, as do the rich (incomes greater than Rs 10-15 lakh). The large middle of the taxpaying public (around the 90-95 percentile of the worker income distribution), comprising of professionals and other tax absconders, has a considerably poorer compliance ratio.

Piketty also made a plea for the release of income tax statistics in India. Irrespective of the robustness of the conclusions that could be drawn using this data, this request is supported by most Indian observers — I myself, in several articles published over the last few years, have argued for the same. In ‘Blinded by tax revenue’ (The Indian Express, January 12, 2013), I stated that “Sensible policy can only be made if the data are credible. Can there be a right to information about aggregate income tax data for each year, as done in most of the world except perhaps China?” The fact that a renowned tax and income distribution expert is requesting the same should increase the chances that at some point, the Indian government will conform to basic international standards of transparency. But I wouldn’t bet on that, for the Indian tax department in the ministry of finance has been known — independent of the political party in power — to be the most obtuse and convoluted tax department in the world.

What the Indian tax department has never answered is why, and how, it or anybody else will lose if tabulated tax statistics, with no information about individual taxpayers, are released. It is also surprising that no RTI has been filed for aggregate tax data to be released. Think about it: The government makes plans on revenues and tax rates, the CAG periodically disseminates the number of taxpayers and aggregate tax collected for three or four income ranges, yet more complete data is not made available. Why?

Even if not very useful for derivation of income distribution, the fact remains that aggregate Indian tax data should be made public. On this request, Piketty is spot on. Regarding speculation about Indian inequality being among the worst in the world, I am afraid Piketty is likely to be grossly wrong.

The writer is contributing editor, ‘The Indian Express’, and senior India analyst, The Observatory Group, a New York-based macro policy advisory group.
Have You Tried These 15 Minutes Recipes Yet?

For all the latest Opinion News, download Indian Express App

    Live Cricket Scores & Results