Bad Economy: Has India’s Poverty Really Decreased?


By Santosh Mehrotra & Jajati Parida

Bhalla, Bhasin and Virmani in a working paper (IMF), claimed that India’s poverty, based on a poverty line of $1.9 per person per day (PPP), was 0.9% of the population in 2020. Thanks to the government transfer of free rations of 5 kg per person per month, it fell to 0.8% (compared to 0.9% in 2019). Roy and de Velt, for the World Bank (WB), claim that poverty fell from 22.5% to 10.2% between 2011 and 2019. Remarkably, the two articles contradict official WB publications, which claim that ‘due to Covid, poverty will increase from 88 to 115 million in 2020, and the biggest contributor to that would be India.

Naturally, the Indian media picked up on these findings and publicized them widely, much to the delight of the government. Therefore, the articles deserve a thorough review, both for their methodology and their conclusions.

First, Bhalla et al use National Accounts Statistics (NAS) to estimate private final consumption expenditure and then use them to estimate poverty. Virtually no one in the world uses NAS estimates of consumption. It overestimates consumption. This never stopped Bhalla from using it.

Second, they do not use Household Consumption Expenditure Surveys (CES), which are widely used to estimate poverty (including by the World Bank, relying on Living Standard Measurement Surveys). Historically, in India, the estimation of poverty has used the CES. The last, in 2017-18, followed demonetization and a rushed GST, which hurt the unorganized sector which employs 85% of non-farm workers. This CES showed a contraction in rural consumption of 8% and an increase of only 2% in urban areas, compared to 2011-12.

Third, they ignore data on inequality, rising unemployment, falling employment rates, falling wage rates and rising food inflation. Inequality had decreased between 2012 and 2019, mainly because all consumption was compressed and trending downward. However, recent inequality reports from PRICE, PEW and Oxfam showed that income inequality has increased during the pandemic, with incomes for 84% of households falling. While the unemployment rate rose from 2.2% to 5.8% between 2012 and 2019, the youth unemployment rate rose from 6% to 17%. The share of the working-age population in employment rose from 38.6% to 35.3% and, for young people, from 42% to 31.5%. The decline in wage rates since 2012 is consistent with the decline in GDP growth rates, particularly since 2016. Additionally, a food inflation rate of 31% between July 2020 and July 2021 (RBI Bulletin August 2021), would negate the impact of the 5 kg of free food. grain; add to that fuel inflation, thanks to voluntary taxes.
Roy and de Velt (2022) find a reduction in poverty between 2011 and 2019, based on CMIE’s Consumer Pyramids Household Survey (CPHS). Their results could be misleading for several reasons. Although the CPHS collects detailed expenditure information for about 115 household consumption items, on a longitudinal basis (covering 174,000 households in 28 states), it suffers from criticism.

The CPHS adopts a measure of consumption that is not comparable to that of the NSS, due to differences in survey instruments. Additionally, Dreze and others have questioned the representativeness of the survey compared to NSS surveys, due to differences in sample design and geographic coverage. These differences will have important impacts on the estimates of poverty in India.

On the design of the sample, the absence of a second degree stratification raises a question about the representation of households from both ends of the income distribution. On the other hand, in the NSS, the representation of urban households from the 1st to the 6th decile of the distribution is integrated into the sampling plan. Homeless people or families living on construction sites are excluded from the CPHS survey. This also contributes to the under-coverage of the poorest households in the CPHS. Then, the instrument differences. The NSSO uses a more detailed consumer module comprising over 345 items, compared to the unique 114 items of the CPHS. In addition, NSS expenditures based on a uniform recall period capture household consumption over the last 30 days, while CPHS collects consumption based on the last four calendar months. Differences in recall periods between surveys can have important impacts on poverty estimates. The shares of households with access to electricity, water, toilets and owning a television and a refrigerator are significantly higher in the CPHS-2015 and -2019 compared to the NFHS of the same years.
Undereducated people are severely underrepresented in the CPHS, with only 2% of the 2018 adult population having no formal education. In comparison, the Periodic Survey of the Active Population (PLFS) of the ONSS of the same year sets the proportion of adults without formal education at 17%. In 2019, adults with no formal education are practically eliminated from the CPHS sample, while the PLFS estimates it at 17%. Basole et al (2021) find that average real earnings in the 2018 CPHS are about 30% higher compared to the PLFS of the same year.

Moreover, as Roy notes, the Gini inequality coefficient using the CPHS weights would rank urban India on par with Sweden, the 25th fairest country. However, NSS-2011 would rank urban India near 60th most unequal.

Thus, comparing expenditure and non-expenditure statistics derived from the CPHS with those obtained from nationally representative benchmark surveys shows that: (1) the CPHS under-represents the poorest households as well as the most rich of the population; and (2) undercoverage of the poor and rich is more pronounced in urban areas. Roy corrects for these differences, but strong differences remain between consumption in the estimates based on the CPHS and the ONSS. Roy then uses regressions to arrive at poverty estimates for India. However, the estimation of poverty based on Roy’s transmission methodology is subject to strong assumptions. Therefore, estimates based on this are questionable. Moreover, the estimate based on the regression of Log consumption per capita also suffers from problems of endogeneity (due to the use of “consumption categories/capita” and “level of education” as explanatory variables) and excluded regressors (family wage income, number of earning family members, land holdings, etc.). These two problems normally produce biased and inconsistent regressor estimates. Therefore, the prediction based on this model may not be reliable.

Meanwhile, based on Tendulkar’s extended poverty line (from 2011-12 to 2019-20) taking into account CPI inflation and after necessary adjustments between CES and EUS gaps from the survey on employment from the ONSS (2011-12), we found that the incidence of poverty between 2011-12 and 2019-20 only decreased by about 1 percentage point (from 21.9% at 20.8%). While the incidence of rural poverty fell from 25.7% to 25.2%, in urban areas of India it fell from 13.7% to 12.4%. But the number of poor people actually increased between 2011-12 and 2019-20 due to the increase in the total population. The increase in the number of poor people in rural areas (from 217 to 235 million) is far greater than the drop of 3 million (from 53 to 50 million) in urban India.

Author Santosh Mehrotra is a visiting professor at the University of Bath, UK, and Jajati Parida teaches at Central Punjab University.


About Author

Comments are closed.