Statistics of poverty suffer from the country’s poverty of statistics

Two different sets of poverty estimates for India were released recently. One of the papers was authored by Surjit Bhalla, Karan Bhasin and Arvind Virmani and the second by Sutirtha Roy and Roy Weide, both affiliated to the World Bank. Both presented estimates for roughly the same period, after 2011-12, but ended up at starkly different estimates of poverty and the magnitude of its decline. This is in contrast with two earlier estimates, both of which showed an increase in poverty after 2011-12 using data on consumption expenditure from the National Statistical Office (NSO), although from different surveys. S. Subramanian presented his estimates in 2019 using leaked data of consumption expenditure from the 2017-18 survey which was junked by the government. On the other hand, Santosh Mehrotra and Jajati Parida used consumption aggregates from the NSO’s employment surveys to arrive at broadly similar estimates as Subramanian of a rise in poverty after 2011-12.

Putting it all together, one gets estimates of poverty ranging from 2.9% by Bhalla, Bhasin and Virmani to 13.6% by Roy and Weide, 25.9% by Mehrotra and Parida and 35.1% by Subramanian. Except for Mehrotra and Parida’s, whose estimates are for 2019-20, the rest are for 2017-18. Clearly, the variation is too vast. The reality doesn’t change irrespective of the way poverty is measured, but that variation is an indicator of the state of affairs in poverty measurement. The differences in poverty estimates are due to the measure of income/consumption used as much as their choice of poverty lines. While Subramanian used leaked NSO data on consumption, Mehrotra and Parida used a short consumption question in NSO’s job surveys. Roy and Weide used the Consumer Pyramid Household Survey of the Centre for Monitoring Indian Economy, with some adjustment. Bhalla, Bhasin and Virmani have used the private final consumption expenditure (PFCE) estimate of National Accounts to create a series. The only one that’s different here is the estimate by Bhalla, Bhasin and Virmani which does not use a household survey; this is the old method that Bhalla has been advocating without success, given the non-comparability of National Accounts data with survey data. Differences also arise from the poverty line: Like Roy and Weide, Bhalla, Bhasin and Virmani used the $1.9 and $3.2 poverty line of the World Bank, while Subramanian and Mehrotra and Parida used the official poverty lines by the Rangarajan and Tendulkar committees, respectively.

This is not the place to examine the merits and demerits of those estimates. Part of the reason there are conflicting estimates of poverty for the same period is the loss of reliable data and a yardstick to measure poverty and inequality after 2011-12. Until then, the responsibility of providing official poverty estimates based on comparable and acceptable criteria was the government’s, in particular the erstwhile Planning Commission. Panels would regularly define and update our poverty line for use with NSO data, all of which was freely available, allowing for a healthy debate on poverty. Indeed, India can rightfully claim to be a pioneer in poverty studies as well as household consumption surveys, which were acclaimed and adapted by such agencies as the World Bank.

We are now in a situation where the last consumption survey of 2017-18 was abruptly junked without any official study or abnormality findings being made public. Even for committees that defined poverty line, the last panel led by C. Rangarajan has not seen any action by the government though its report was submitted eight years ago.

The absence of an official poverty line and consumption data has forced researchers to use alternative ways of estimation. But this has created more confusion than clarity on how hard-up Indians have fared. The question is important in normal times but more so after the economy went through policy shocks such as demonetization and the GST roll-out and also natural calamities such as the droughts of 2014-15 and the covid pandemic. Poverty estimates are not just an academic exercise but are crucial parameters to judge policy outcomes and the overall functioning of the government. Note that poverty statistics continue to be used for allocations under various government programmes and notably for the distribution of resources, financial and otherwise, across states. Poverty and income distribution are issues for public discussion as much as they are instruments of governance and public policy for an economy which still has a substantial population that’s financially vulnerable even if not officially poor.

Himanshu is associate professor at Jawaharlal Nehru University and visiting fellow at the Centre de Sciences Humaines, New Delhi