We must take note of a discrepancy between what India’s government advocates and what the World Bank reports, as it makes a big difference to our poverty count
The most important agreement with some expert analyses of our paper (Bhalla-Bhasin-Virmani, Pandemic, Poverty and Inequality: Evidence from India, IMF Working Paper, April 2022) is on the need to raise the poverty line, possibly a 68 % rise in real terms (from PPP $1.9 per person per day to $3.2 pppd). The present line, in current rupee terms, is approximately ₹52 and we have recommended that it be raised to around ₹88. For a poor family of five individuals, this would mean a household income of ₹1.6 lakh a year. With this poverty line, approximately a fifth of the population will be poor, the ‘right’ definition of relative poverty for a lower middle income economy. Note that poverty reduction is an intrinsic product of economic growth. In 2004, the poverty line was raised by the Indian government (Tendulkar committee) by 18% and this was accepted as a worthy step by all.
Among the substantive points raised in our paper was a simple point about measurement. Somewhat disappointingly, not a single expert, friendly or otherwise, has noted this discrepancy between what India’s government supports and advocates, and what the World Bank reports in its published Povcal reports on the world and individual country reports (hereafter World Bank).
It is a simple matter of definition, on which there should be no disagreement. The facts are as follows.
There are three different definitions of National Sample Survey (NSS) based per capita consumption. The differences have to do with the recall period of consumption for three broad categories, and there is convergence in the academic and policy community (and the World Bank) on the appropriateness of each recall period. The ‘right’ recall period is also supported by common-sense justification of memory and accuracy. For perishables (vegetables and fruits), accuracy is enhanced with a recall period of 7 days. For periodic consumption (e.g. toothpaste, club fees, visits to the doctor, etc), a 30-day recall is considered appropriate; and for durables (e.g. clothing, cars, carpets, furniture, etc), a 365 recall period is deemed proper.
Prior to the 1983 NSS report on consumption, all items were tabulated under a 30-day “Uniform Recall Period" (URP). Starting in 1983, the NSS added the Mixed Reference Period (MRP) with the addition of a 365-day recall period for “durables". And after some experimentation and validation in the NSS surveys of 1999-00, 2009-10 and 2011-12, the NSS organization officially converged to the MMRP (Modified Mixed Recall Period) method. The difference between MMRP and MRP is the addition of a 7-day recall documentation of answers to questions pertaining to the consumption of fruits and vegetables.
It also bears emphasis to note that the Tendulkar committee rejected the URP in favour of MRP in the 2009-10 survey; and post 2011-12 (e.g. 2017-18 and onwards), the MRP was officially junked in favour of MMRP. Somewhat bafflingly, the World Bank, the unofficial ‘gold standard’ of poverty measurement, continues to present Indian poverty estimates for 2011-12 and beyond on the basis of the ‘junked’ and old (pre-1983) method of measuring consumption, and therefore poverty. It is very likely that for no other country does the World Bank not use the official method of measurement, and for no other country does it use a 45 year old outdated method (the last exclusive URP survey was in 1977-78).
It wouldn’t matter if it did not matter. But it does. MRP estimates of extreme poverty are about 3 percentage points lower than URP, and MMRP estimates are about 10 percentage points lower. In 2011-12, for India, it meant that the World Bank was wrongly classifying a 100 million Indian individuals as extremely poor. The World Bank has the slogan that it dreams of a world free of poverty. The practice of using the URP method for India, a country with more than a fifth of the developing world’s population, prolongs the nightmare of a world not free of absolute poverty.
Given the huge importance of the recall period in generating representative estimates of poverty, it is puzzling to note the advocacy and recommendation by World Bank authors Sutirtha Sinha Roy and Roy van der Weide (Poverty in India has Declined over the Last Decade but not as Much as Previously Thought, World Bank Working Paper, April 2022) to use the CMIE Consumer Pyramid Household Survey which has a 4-month (120-day recall) period for all consumption items!
Not all poverty estimates are created equal. It is unfortunate that in India, a one- question consumption estimate, as in 2017-18 onwards labour force surveys conducted by the NSS (the PLFS surveys) is seen to have equal validity as a 33 question-based estimate (pre-2017-18 NSS labour force surveys). Or a 120-month recall period consumption (as in CMIE) for considerably less consumption items has equal validity as 30-day recall questions for more detailed consumption estimation; or a 30-day recall period is preferred by the World Bank for India, when other official and equally detailed estimates are available (as in the 2009-10 and 2011-12 MMRP surveys). For their estimates of non-survey year poverty (note that most countries have at least a 3-4 year gap between national surveys), the World Bank has to rely on a base-year estimate of consumption and national account growth rates for intervening years. This is exactly what we do, with the critical difference that we use the 2011-12 base year MMRP estimate, not the 10% lower 2011-12 URP estimate. Why should the base-year reflect an unofficial lower estimate of consumption is a question not answered by our critics, or by the World Bank.
Surjit S. Bhalla is executive director, IMF, representing India, Sri Lanka, Bangladesh and Bhutan
Source. Mintepaper, 3/05/22
