In the old days we used to be good at things like poverty measurement. We may not have been particularly good at actually helping the poor but we were ahead of the world in figuring out how to find them. If my friend and Princeton professor Angus Deaton, who knows more about these kinds of things than almost anyone alive, is to be believed, the first scientifically designed large-scale surveys anywhere in the world were carried out by PC Mahalanobis and his boys at the Indian Statistical Institute.
The survey data they were generating as a part of what soon became the National Sample Survey was then turned into a poverty measure by estimating the fraction of the population that lives under the poverty line. Here again we were a standard-setter. Our first rigorously founded poverty line goes back to 1962, when a Working Group constituted by the Planning Commission came up with a poverty line of Rs20 per month at 1960-61 prices for people living in rural areas, based on discussions about how much it costs to get enough calories to live on and a few other things. The US got its first comparable poverty line only in 1965.
It seems like we have almost entirely lost touch with this piece of our national history. Mahalanobis, if remembered at all, is just the purveyor of the eponymous failed model of planning, not the great applied statistician he surely was. The many enormously sophisticated discussions about the prices that the poor pay seem to be entirely forgotten: Many prominent participants in the recent debate about the poverty line seemed to be unaware of the fact that to convert a rupee poverty line into a dollar poverty line, we need to adjust to the fact that the things the poor buy are cheaper in India than in the US.
The claim, much trumpeted in the press, that the new urban poverty line of Rs32 per person per day was at 65 cents a day, and, therefore, well below the World Bank's dollar a day standard, is just plain false: The World Bank sets its poverty line after adjusting the differences in prices across countries. The rupee-dollar exchange rate in terms of purchasing power works out to be Rs19 per dollar rather than Rs49. Hence the new line is at $1.68 and not at 65 cents.
None of this says that Rs32 is the right standard in some ethical sense. But the fact that it was debated as such, with its proponents insisting that it was enough to cover all necessities and, therefore, the 'correct' standard, and its opponents pooh-poohing it (partly based on the 65 cents conversion), points to a more basic confusion. The poverty line in India does two things.
First, it measures, so to speak, our shame - the number of people who are below what we, as a nation, consider to be an acceptable standard of consumption. This was the original reason for the poverty line. But with the move to a targeted public distribution system, a second dimension of the poverty line became relevant. Since the goal was to target the poor, we had to have a definition of who was poor. This definition was not directly applied to the identification of the poor (and, therefore, the allocation of below poverty line (BPL) cards), which was left to the local jurisdictions. But the allocation of money to the states for funding programmes targeted to the BPL population was based on the number of poor people they had, and to compute that number we needed a poverty line.
This was the process that the Planning Commission was trying to influence by its choice of the poverty line that offended well-meaning people all over the country. Yet, think of what would happen if the line were raised to say Rs50. Raising the poverty line does not make the government richer or generate more resources for fighting poverty - for that we need to raise tax collection, as I have argued before in these pages and elsewhere.
Therefore, when poverty line is raised the same amount of money gets reallocated. Now suppose there are two states, call them B and G, with the same total population: 50% of B's population lives below Rs32 against 25% of G's population, but in the segment between Rs32-50, the proportions are reversed. Hence, if the poverty line were Rs32, B would get twice as much money as G, but if it's raised to Rs50 they get the same amount. So state B ends up with less money for its desperately poor population with the higher poverty line, while the somewhat less favoured people in G benefit. Is that really what we want?
Maybe the way out of this confusion is to call a spade a spade. We need two different poverty lines: an ethical poverty line to describe the standard we should aspire to, which could be Rs50 or more, and an administrative poverty line, which tells us how to best target our limited resources. As we get richer, perhaps the latter will be raised till it is effectively the same as the former. But right now we don't want to hurt the poorest in the name of being more aggressive about poverty.
Abhijit Banerjee is Ford Foundation International Professor of Economics and Director, Abdul Latif Jameel Poverty Action Lab, MIT. The views expressed by the author are personal.