World Bank Report on India's Incredible Poverty Reduction Isn't Credible

The Wire

30-04-2025

The World Bank released a poverty and equity brief for India on April 22 which began with the following text:
Over the past decade, India has significantly reduced poverty. Extreme poverty (living on less than $2.15 per day) fell from 16.2 percent in 2011-12 to 2.3 percent in 2022-23, lifting 171 million people above this line.
This would of course be fantastic news, if true. However, for anyone even vaguely familiar with the poverty numbers, the 2.3% headcount ratio for extreme poverty seems a little too good to be true. Furthering the confusion, the World Bank had significantly different estimates in its October 2024 brief:
Following strong post-COVID-19 growth, averaging 8 percent, extreme poverty in India declined by 2 percentage points. In FY 2021/22, 44 percent of the population remained below the lower-middle-income poverty line ($3.65/day 2017 PPP), and 12.9 percent were below the extreme poverty line ($2.15/day 2017 PPP).
Under the assumption that both of these reports are accurate, the 10.6% poverty reduction, from 12.9% in 2021-22 to 2.3% in 2022-23, would surely count as a world record.
In the absence of a vastly underreported magical showering of wealth upon 140 million-plus people in the lowest income brackets in India, it is natural to ask what is behind the sudden change in the numbers reported by the World Bank during the last six months. The mundane answer is a switchover in the methodology for reporting poverty estimates for India. To understand the magnitude of the difference, we have to delve into some details and controversies surrounding poverty estimation in India.
Despite breathless headlines about World Bank data endorsing Prime Minister Narendra Modi's claim, the first thing to understand is that the underlying data in the 2025 report come from the Indian government's Household Consumption Expenditure Survey (HCES) for 2022-2023. The World Bank only calculates poverty statistics from survey data, so ascribing results to World Bank data is misleading.
Beginning in the 1950s, the Indian government used to conduct a Consumer Expenditure Survey (CES) every five years, which formed the basis for poverty related calculations. In 2019, the BJP government decided not to release 2017-2018 CES data, claiming that there were 'data quality' issues, while deflecting insinuations that adverse findings were the reason for its decision. Due to this gap in the data, economists came up with creative ways of estimating poverty ratios.
Two papers published in April 2022, the first by Bhalla, Bhasin, and Virmani (BBV), and the second by Roy and van der Wiede (RW), used very different techniques, and came up with dramatically different poverty headcount ratio estimates for 2019-2020 of just under 2% and 10% respectively. Briefly, BBV used national accounts statistics (basically, those used for GDP calculations) to scale up the consumption expenditure distribution from 2011-2012. On the other hand, RW used an alternative survey, the Consumer Pyramids Household Survey (CPHS), conducted by Center for Monitoring Indian Economy (CMIE), a private entity. The two approaches, and the drastically different results they generated, set off a lively debate that continues to generate criticism and alternative proposals.
In its October 2022 brief, the World Bank decided to adopt the RW technique. This was unusual because of its reliance on a survey conducted by a private entity, even though it had the virtue of producing a more plausible poverty headcount ratio. In April 2025, the World Bank has now switched back to using the Indian government's survey, the 2022-2023 HCES.
Before thinking about the credibility of the results, it is useful to get a sense of the rupee incomes associated with various poverty lines used by the World Bank. For the conversion, we need to use the purchasing power parity (PPP) conversion factor, which differs significantly from current exchange rates. The interpretation of the PPP conversion factor of Rs 20.15/$ is that, given different cost structures in India and the US, people can obtain the same level of goods and services for Rs 20.15 in India that they can in the US for $1. World Bank Poverty Thresholds
(PPP conversion factor ₹20.15/$) Per person per day (2017 PPP) Per person per month Family of four per month Extreme Poverty Line $2.15 Rs 1,300 Rs 5,199 Lower-middle-income Country Poverty Line $3.65 Rs 2,206 Rs 8,826 Upper-middle-income Country Poverty line $6.85 Rs 4,141 Rs 16,563
Source: World Bank; table a ssumes 30 days per month.
So, the 2.3% headcount ratio in the April 2025 brief refers to families living on less than Rs 5,200 per month. This is an absurdly low level of consumption. The very same World Bank report mentions that if we look at the $3.65 per person per day threshold, the poverty rate is 28.1%, which unfortunately is not as headline friendly.
Economist S. Subramanian has provided a thorough analysis of the 2022-2023 HCES data. He computes a headcount ratio of 3% for 2022-2023 using the Tendulkar Committee poverty line, updated to 2022-2023 prices, a number that is comparable to the number in the 2025 World Bank brief. This rules out the possibility that the World Bank calculation methodology is the reason for the low poverty ratio.
Very low poverty lines – which are clearly useful devices to arrive at low poverty headcount ratios – are not confined to the World Bank alone. India's poverty line based upon the 2009 Tendulkar committee recommendation has been long criticised for being too low. The 2014 Rangarajan committee recommended higher thresholds, but those were not adopted. These thresholds, updated to 2022-2023 prices, are shown below. India
(2022-2023 values) Per person per month Family of four per month Rural Urban Rural Urban All India Tendulkar Committee Poverty Line Rs 1,500 Rs 1,866 Rs 6,000 Rs 7,464 Rs 6,527 Rangarajan Committee Poverty Line Rs 1,794 Rs 2,720 ₹7,176 ₹10,880 ₹8,509
All India values calculated by author as weighted averages of rural and urban numbers assuming 64% of the population is rural. Source: S. Subramanian
There are plenty of reasons to find the 2.3% headcount ratio to be implausible. Briefly, the steep drop of 14 percentage points in 11 years is difficult to digest given several macroeconomic events and indicators in this period. There were large shocks to the economy: the infliction of demonetisation and the affliction of the Covid-19 pandemic. As a result of the pandemic and lockdown induced contraction, there was a massive loss of employment and significant reverse (urban-to-rural) migration.
Agriculture's share of output has remained more or less constant for the last two decades, and manufacturing's share has declined, so service sector jobs and wages have to be responsible for improvements in standards of living. This, of course, may have occurred for people in well-paying professions like software development, finance, health care etc. but it is hard to see it having an impact on the incomes of millions of people close to the poverty threshold, especially in rural India. Post-pandemic consumption expenditure patterns were reported to be 'K-shaped' (high-end goods selling well, consumer goods sales flattening or declining), indicating stressed budgets among lower income groups. Urban unemployment, according to the government's quarterly survey data, averaged around 7% in 2022-2023. Given all this, what was the source of the incomes necessary for large scale poverty reduction?
These are all issues that are easy to obfuscate behind opaque data, bad faith argumentation, shifting goalposts, technical skullduggery and relentless propaganda aided by pliant media and a chronically attention-deficit plagued public.
So, here's the simple takeaway: if you choose a very low poverty line, you get a very low poverty rate. Not too surprising that about 2% to 3% of the population lives in households with income levels of less than Rs 5,200 per month.