Poverty of Numbers
_________________________

The actual figure of those living below the poverty line
is likely to be nearly half of the Government’s estimate,
writes SHEFALI MISRA

Can the economic reforms of the Nineties be justified if they have failed to tackle poverty? No, of course not. This argument has long been the ammunition of critics of reform. Bolstering them have been the Government’s figures which suggest that poverty has thrived mostly undisturbed through the decade of reforms. But if a lively new debate on numbers is any indication, the firebrands may soon have to get their fire elsewhere. The number of the absolutely poor are in all likelihood far smaller than the Government’s figures and may have fallen sharply after the reforms began. What’s going on? A numbers game and some economists taking issue with the Government on alleged gross underestimation of consumption and overestimation of poverty.

The problem seems to lie in the way people are asked questions about how much they spend on consumption, which is used in concluding how many live below the poverty line. (Consumption spending shows if people manage to get their required calorie intake.) The National Sample Survey (NSS) ask about 1.2 lakh households every five years what they spent on consumption in the last 30 days. In this apparently innocent question, however, lies the trouble. Consider the following:
The NSS recently released ‘‘startling’’ data for the first six months of 1998, as well-known economist Surjit Bhalla puts it. According to this data, real spending (after accounting for inflation) per person fell by six per cent from 1997. This meant that per capita spending had increased only seven per cent over 11 years (1987-1998). The number of those under the poverty line increased from 38 to 42.6 per cent of the population. Seems to contradict common sense and observation?
Now, change the question to how much people spent in the last seven days, and guess what happens? The number of those under the poverty line is nearly halved from 42.6 to 23.6 per cent! No profound mystery here. People remember their spending of the last seven days far more clearly than of the last month. The seven-day method is far more reliable and India is possibly the only country to use the 30-day method.

The difficulty, as Planning Commission member Montek Singh Ahluwalia pointed out in a paper (EPW, May 6), is that the seven-day NSS sample is available only since 1994-95 and can’t be used to confirm poverty trends for the Nineties.
But even with the seven-day method, the NSS poverty figure for January-June 1998 (23.6 per cent) is higher than for January-June, 1994-95 (22.8 per cent). This is important because, as economists for once agree, the most important thing is the poverty trend: whether or not it is falling over time and how fast.

This is where evidence to the contrary has been forthcoming lately. The so-called MISH (Market Information Survey of Households) of the National Council for Applied Economic Research finds the poverty ratio to have been 38.8 per cent in 1987-88, but says it fell to 26.2 per cent in 1997-98 (the NSS figure is 37.6 per cent, based on the old method). MISH covers three lakh households, but does not record spending, only reported incomes.

NCAER chief economist I. Natarajan, the man behind MISH, is at one with everyone else in saying that the NSS hugely underestimates consumption spending. But he stresses that whatever the method, the trend at least should be correctly captured. He attributes the scale of discrepancy in the NSS and NCAER data to the NSS underestimating everything, from household size and the number of life insurance policies held to spending on footwear.

Pravin Visaria, chairman of the National Sample Survey Organisation and director of the Institute for Economic Growth, readily concedes the methodological problem with the 30-day recall question. But he says the task of moving from that to a seven-day recall is a delicate one and can only be implemented cautiously. He is right because any sudden switch showing lower poverty could promptly be shouted down as a game of political manipulation.

Visaria says that the poverty trend between 1987-88 and 1993-94 is certainly downward, and expects that the 1999-2000 figures now being prepared will show a much lower poverty line percentage, closer to the NSS’s own seven-day recall figure of 26.2 in 1994-95.

Meanwhile, perhaps the most politically provocative point is made by Surjit Bhalla. He finds fault with the fact that the country’s economic growth numbers are derived from the National Accounts Survey while the poverty numbers are derived from the NSS. The one shows economic growth and drastically-reduced poverty, the other shows low consumption growth and very slowly falling poverty. They contradict each other, but are internally consistent. For the reform period of 1987-98, the NSS shows a spending increase of just 10 per cent, and a poverty decrease of just four percentage points, from 38 to 34. The NAS shows a real spending increase of 39 per cent and a drastic poverty reduction from 38 per cent to just 11 per cent!
Using growth and poverty data from two different sources makes them liable to misuse by anti-reformers, who can say that growth is not trickling down to the poor. ‘‘All I am saying is be consistent,’’ he says. And accurate, we might add.

Next - How To Use Napster