Poverty has declined from 37.2% in 2004-05 to 32% in 2009-10, according to preliminary Planning Commission estimates based on consumption surveys of the National Sample Survey Organization (NSSO). Some welcome the news; others complain that fast growth has not reduced poverty fast enough. Still others, notably economist Surjit Bhalla, pour scorn on the data, and imply these have been fudged to exaggerate poverty.
I agree that the poverty data are wrong, but not because they have been fudged. Rather, poor people are increasingly lying to surveyors about their true living standards, and this is inflating apparent poverty.
Devesh Kapur, Lant Pritchett and other researchers cite a telling example of this in a research survey last year on the conditions of dalits in Uttar Pradesh. The researchers hired local facilitators to assist the survey. Answering questions, one dalit claimed to be badly off. But as the questions proceeded, the facilitator realized that the person proposed for marrying his cousin was this dalit\’s son. On discovering this, the dalit declared he was actually well off, and had lied about being poor to ensure he did not lose access to subsidies! He jumped from understatement to overstatement in a trice.
Poor people respond to incentives like anybody else. When surveys started some decades ago, the poor probably told the truth, since they did not think this would hurt them. But over the decades, the government has started targeting subsidies more sharply at those below the poverty line (BPL). Till the late 1990s, there was a single subsidized price for all food grains at ration shops, but after that, an especially low price was decreed for BPL folk. BPL targeting is used for other entitlements, such as job cards under NREGA.
In such circumstances, it is absurd to expect that people who rise above the poverty line will admit this to surveyors, and risk losing their benefits. A significant proportion will understate their consumption. So, surveys will capture less and less of what people really consume.
Is this happening? Yes, indeed. Back in 1972-73, consumption measured by the NSSO surveys amounted to 87% of consumption estimated by national accounts (which measure GDP). But this proportion of consumption captured by the NSSO has steadily declined over the years, fell to just 48.8% in 2005-06, and further to an all-time low of 43% in the latest 2009-10 survey.
Leftists claim that this is entirely because the rich are hiding their consumption to evade the taxman, and that the consumption of the poor is accurately captured. Phooey! Income tax rates have crashed from 97.75% to 30%, wealth tax has been abolished and the use of white money has improved substantially in real estate transactions since 1972-73. In every way, the rich have less incentive to hide their consumption today, while the poor have more incentive.
GDP data show Indian consumption rising fast. But NSSO data show consumption rising slowly. Bhalla calculates that in the two years between 2007-08 and 2009-10, NSSO data shows consumption plummeting 7%, as calamitous a fall as during the Great Depression!
This is plainly fiction. We all know that GDP and consumption kept rising, with new cell phone connections increasing by 15 million per month. Explanation: the NSSO captured 47% of consumption in 2007-8 but only 43% of consumption in 2009-10.
Let\’s return to the official poverty estimates, warts and all. A new poverty line has been adopted based on the Tendulkar Committee recommendations. According to this new formula, poverty declined from 45.3% in 1993-94 to 37.2% in 2004-05, a decline of 0.73 percentage points per year. In the next five years to 2009-10, poverty declined by 1.04 percentage points per year. This is a significant improvement, though it looks too low in an era of fast growth. But remember, there was a major drought in 2009-10, which would have depressed consumption below trend.
Besides, a better way of judging poverty reduction is to ask what proportion of the poor rose above the poverty line between the start and end of the period. Measured in this way, poverty reduction between 1993-94 and 2004-05 was 1.65% per year, and improved to 2.8% per year in the next five years to 2009-10. This is a big gain even without adjusting for undercounting of consumption. That\’s welcome news.
In any event, we need to adjust the poverty estimates for the falling proportion of consumption captured by NSS surveys. There is no space in this column to go into the statistical issues involved. But some statistical changes are necessary to bring the NSSO data closer in line with reality.