Opinion The poverty debate: Was there really a decline in poverty during the pandemic years?

The attempt by PM to use PLFS data is not new, but they do deserve compliments for reiterating that the PLFS data is fully comparable to the earlier employment-unemployment surveys (EUS) of the NSO and that it “suffers from no known flaws of design and representativeness”. They are also right to point out the superiority and credibility of publicly available data with “benefit of scrutiny and oversight by scholars and statisticians” as against privately collected data.

PM also rightly emphasise the necessity of using comparable estimates of consumption expenditure to arrive at any correct estimate of poverty. Poverty estimates in India have always been based on consumption estimates from the NSO, particularly based on the consumption expenditure surveys (CES), with the last official poverty estimates being for 2011-12, even though a comparable consumption survey was conducted in 2017-18. The latter was junked by the government without any valid reason, denying the benefit of scrutiny and oversight of scholars and statisticians — a principle the NSO has followed since the surveys began in the 1950s.

However, leaked reports are available and these show a rise in poverty between 2011-12 and 2017-18, driven by a sharp 5 percentage point increase in rural poverty and a small decline in urban poverty. This incidentally remains the only valid and comparable poverty estimate for any poverty analysis after 2011-12.

Although the PLFS also collects some consumption aggregates, as PM also reiterate, these estimates of poverty are not comparable with those from the CES. The reason is well-known and this has to do with the different levels of detail in which consumption expenditure data is collected by these two surveys. The PLFS estimates are based on a single question and are therefore neither comparable with CES estimates that are based on much more detailed consumption expenditure information nor the EUS estimates that used an abridged consumption schedule. In fact, this issue of sensitivity of consumption estimates to survey design, the level of aggregation and details has been extensively written about and was at the heart of the “Great Indian Poverty Debate” of the early 2000s.

This issue of sensitivity to the details of questions asked to collect consumption expenditure is not just relevant across different surveys but also across different rounds of the PLFS. While the annual PLFS had a single question to elicit information on consumption expenditure between 2017-18 and 2019-20, in 2020-21 the number of questions was increased to six, along with detailed instructions being given in the questionnaire itself. This change in questionnaire design makes the PLFS estimates of consumption expenditure before 2020-21 non-comparable with those after. In fact, the design not only affects the average estimates but also changes the distribution of the consumption expenditure across income/consumption deciles with the detailed consumption schedule showing lower inequality compared to the single shot question.

Given these data discrepancies, what is it that can be said about poverty trends in recent years, based on comparable estimates of consumption expenditures from the NSO?

The first set of conclusions can be drawn for the period between 2011-12 and 2017-18. Using the CES based full schedule and the leaked report for 2017-18, a rise in poverty can be seen. For a similar time period, the single question asked in the earlier rounds of PLFS can be compared with the 2014-15 (72nd round) NSO survey on services and durable goods expenditure which had exactly the same question in the same block with the same instructions making them comparable to estimates from the PLFS from 2017-18 to 2019-20. These suggest that the poverty headcount ratio was 27 per cent in 2014-15 and rose to 36 per cent in 2017-18, declining to 32 per cent in 2018-19 and remaining at that level in 2019-20. Therefore, despite some reversal in the trend, these suggest a significant rise in poverty in 2019-20 compared to 2014-15.

These trends are consistent with other indicators on wages and could be a reflection of a shock-induced rise in poverty. The twin shocks of demonetisation and roll-out of GST affected the population at the bottom of the distribution. The year 2018-19, which was agriculturally one of the best years, witnessed some decline in poverty, but this decline did not continue in 2019-20 as the slowdown of the economy had set in. The cumulative impact of the twin shocks and the slowdown is also reflected in higher poverty in all years after 2017-18 compared to 2014-15.

Unfortunately, for the period during the pandemic (2019-20 to 2020-21) that the PM paper tries to address, it is difficult to say what happened based on available consumption data because of the questionnaire changes mentioned above. The extended set of consumption estimates from the PLFS which are once again comparable show that the poverty ratio in 2020-21 was 26 per cent and it remained at the same level in 2021-22. This also implies that the so-called economic recovery had no impact on poverty in 2021-22.

Independent data from multiple sources on wages, incomes and consumption of durables confirm the distress in the economy, especially during the pandemic years. But even the PLFS data, which PM use to argue in favour of poverty reduction, has enough indicators to show otherwise. As the PLFS data show, real incomes and consumption have declined after 2018-19, real regular wages have declined since 2011-12 and there is an increase in distress employment.

The issue of what happened to poverty after 2011-12 is not just an academic issue, but is also crucial for policy. Frequent interference in the statistical system through changes in survey and questionnaire design, suppression of data and delaying the release of crucial data are only making it difficult to have a correct assessment of the reality as it exists. While it does affect the credibility of the statistical system, the absence of official estimates on poverty is also a reflection of the lack of political priority of the government on such a crucial indicator.

Currently, a survey on consumption expenditure is being canvassed by the NSO which again follows a completely new methodology and schedule. While it may provide another set of estimates of consumption expenditure, it is unlikely to help resolve the poverty debate.

The writer teaches at JNU