Further grist to the mill of our great Indian debate on poverty
PremiumAn IMF paper that claims the near elimination of extreme poverty in India deserves close scrutiny
Nobel Laureate Angus Deaton along with Valerie Kozel of the World Bank edited a book titled The Great India Poverty Debate (2005). This debate, while important to us, has global relevance too. India has had a rich tradition of systematic and disciplined data collection at the household level, subject to sophisticated statistical techniques. As Deaton and Kozel said, the debate was important for several issues involved. For instance, the wide difference in an estimate of total national consumption spending based on extrapolation of data from household surveys and that obtained from national accounts. A small discrepancy can be ignored, but what if it’s 20-30%? That would be shocking. In India, the National Sample Survey Organization (NSSO) data consistently shows much lower consumption spending than the National Accounting System (NAS) does. The latter’s proponents say it is more reliable and so the NAS should be used in poverty estimates. But wait, what if poverty estimates based on it, invariably lower, are not corroborated by NSSO data on living standards? Surely, NSSO data cannot simply be ignored, or junked.
Apart from that discrepancy, another issue that animates the debate is the questionnaire’s design. Does it have any leading questions? Do the poor get oversampled? Are there consistency checks? Of course these questions are not unique to NSSO questionnaires. One issue in administering it was the recall period. If asked to report on items purchased once or twice a year, like durable goods, there are bound to be errors of undercounting. Hence the approach was modified to include multiple recall periods (365 days, 30 days and 7 days) to get a more accurate estimate of expenditure. This has still not reduced the gap between NAS and NSSO consumption estimates to statistically negligible levels. More explanations are needed. Finally, there are other issues such as non-responses, repairing faulty responses, and so on. And, yes, also the hugely important issue of what the poverty level should be.
So it is not surprising that the debate was fierce and Deaton and Kozel did an excellent job of collecting the main arguments. Their book did not settle the matter. If economic reforms accelerated the pace of poverty reduction, the methodology also overstated the gains. If national accounts gave a rosy picture, sample surveys did not offer adequate corroboration. If income poverty went down sharply, the human development indices did not show corresponding improvements. These findings almost reached the level of Joan Robinson’s famous quip that everything that you say about India is true and so is its opposite.
If the debate was raging fiercely then, it was fuelled by a recent IMF paper by Surjit Bhalla, Arvind Virmani and Karan Bhasin. The authors contend that India’s extreme poverty (as defined by the World Bank as $1.9 purchasing power parity consumption per day per person) is as low as 0.8% and remained low during the pandemic period, thanks to substantial in-kind transfers of food grain as part of the Pradhan Mantri Garib Kalyan Anna Yojana. This ran for two years after covid struck and was recently extended to September 2022. The authors value these transfers at high market prices, thereby inflating the consumption estimates of the poor. These corrections, they claim, are crucial to get a correct estimate of “extreme" poverty; in most of the world, poverty worsened during the pandemic, but not so in India; in-kind transfers also led to the largest decline of consumption inequality in the past 40 years. India was now on the verge of eliminating extreme poverty.
This claim is problematic for several reasons. Firstly, inflating the value of consumption using market prices of free foodgrain distributed to more than 75% of the population is like imputing huge wealth to a billionaire dominant shareholder based on a tiny slice of traded shares on a stock market. (The free food grain was not even traded!) Secondly, while NAS estimates are used by the authors to get aggregate consumption spending, the shares of the poor (and all deciles) are taken from NSSO. And then consumption spending of all deciles is supposed to grow at the same national growth rate. That belies the observation of worsening income and expenditure inequality. The rich capture a disproportionate share of national income gains than the poor. And surely the national consumption spending statistics include much more non-food expenditure than what’s in NSSO data. In fact government “consumption spending" has been a big contributor to growth in recent years.
There are reasons to be sceptical beyond a quibble about the $1.9 poverty line. Can poverty removal be sustained simply by sustained food transfers funded by taxpayers? Is not poverty reduction about reaching a permanently higher plateau? Also, as mentioned in Deaton’s book, should not such fantastic poverty reduction be manifest in other indicators, such as the Multi-Dimensional Poverty Index (MPI) as reported by Niti Aayog, or the world hunger index, on which India ranks No. 101 among 116 countries. The ultimate validation of a sharp poverty reduction would be an absence of long queues at government recruitment camps and declines in hunger and malnutrition. The IMF paper has just added grist to the mill of the great Indian poverty debate.
Ajit Ranade is a Pune-based economist