Poor economics: Need for more clarity on data for poverty estimation


Some circumspection is in order over the recent claims by the International Monetary Fund (IMF) and World Bank researchers that India saw a sharp lowering of poverty numbers in recent decades. The World Bank working paper, released on Sunday, said the share of poor in India’s population—as measured by a daily earning of less than $1.9 in purchasing power parity (PPP) terms, with a dollar being equal to Rs 20.65 PPP—fell from 22.5% in 2011 to 10.2% in 2019, with the reduction more pronounced in rural areas. This report was preceded by an IMF working paper which said that India has nearly eliminated extreme poverty. Even the pandemic couldn’t move it up from 0.8% reached by 2019 because of the Centre’s extensive support programme for the poor, including in-kind support such as free and highly subsidised foodgrains, the report claimed.

Compare the findings of the authors of the two papers with that of the researchers at the Azim Premji University, who estimated last year that the first wave of Covid-19 alone had pushed close to 230 million Indians into poverty. To be sure, the total lockdown of 2020 pushed many into joblessness, and some of that was remedied after the brutal lockdowns ended, but the recovery of jobs and income levels hasn’t been too encouraging. Some other analyses have claimed that the gap between the poor and the rich has widened significantly, exacerbated of late by the pandemic shocks.

It should be noted that the Consumer Expenditure Survey (CES), done by the National Sample Survey Organisation (NSSO), is used to estimate poverty in India, and the data is last available for 2011-12. Data collected for 2017-18—after the demonetisation exercise and the implementation of GST, both of which hit the informal sector in the country—weren’t made public. Against this backdrop, several economists have called for caution regarding the data the World Bank and IMF papers have used. The Bank paper uses data from the CMIE’s Consumer Pyramid Household Survey after adjustments as deemed appropriate by the authors, even as there is consensus among a section of economists that the data doesn’t accurately represent the most poor.

On the other hand, the IMF paper uses the private final consumption expenditure numbers from the National Account Statistics (NAS) because the authors believe—as do many other economists—that the CES overestimates poverty because it doesn’t take into account the monetisation of the in-kind support from the government to the poor, such as highly subsidised grains. However, it is hard to say how safe from criticism the use of PFCE data for the exercise is, given that many noted economists, including Nobel laureate Angus Deaton who has extensively studied poverty, don’t see it as a faultless set. Indeed, the issue of divergence between the NSSO and the NAS data has been taken up officially several times, and the conclusion doesn’t seem to have privileged one as more accurate/better representative than the other because of methodological and design differences, among others.

Besides, whether anti-hunger measures translate into any real reduction of poverty needs to be studied thoroughly before making any leap of assessment. Indeed, factoring in nutritional security of households would mean poverty assessment becomes a different ballgame altogether.

Many economists have also used other datasets to come up with varying estimates—Santosh Mehrotra and Jajati Parida used consumption aggregates from the Periodic Labour Survey to report an increase in the poverty numbers between 2011-12 and 2019-20. This did not remain unchallenged, with critics talking about the “unsoundness” of using PLFS consumption data. This suggests that clarity is needed on data for poverty estimation. It is thus welcome that the conversation in policy circles seems to be moving towards a multidimensional understanding of poverty, which will likely take into account factors such as access to healthcare, nutrition, and education.