We need a bridge consumption survey to analyse poverty trends

India’s 2022-23 data doesn’t allow comparisons with the past but an old-format survey would help

is associate professor at Jawaharlal Nehru University and visiting fellow at the Centre de Sciences Humaines, New Delhi

The surprise release of a fact sheet on the Consumption Expenditure Survey (CES) of 2022-23 has raised more questions than it answers. The CES release was expected later this year, once the companion survey of 2023-24 got over. This was publicly stated by the National Statistical Office (NSO) and also reiterated in Parliament.

The usual NSO practice has been to release the key indicators’ report along with unit-level data. It had so far never released a fact sheet, and the departure is especially surprising because this CES round was supposed to be released with the CES for 2023-24. The fact sheet offers selective data. Among other key indicators, it has no information on quantities consumed and the distribution by fractile groups for all states. Its release weeks before the announcement of general elections, however, has led to attempts at estimating poverty from its findings. This stands in contrast with the fate of the 2017-18 CES, which was fully comparable with the 2011-12 CES but was junked by the government on flimsy grounds. The differential treatment confirms apprehensions that it was binned on account of its poverty findings, which reportedly showed a rise in the 2017-18 round, as against a sharp decline seen in 2022-23. Some commentators have already declared that India has eliminated poverty, with its level below 5% for 2022-23.

The 2022-23 CES is important since it is the first one after the 2011-12 survey that the government has accepted. While it helps fill a data void of more than a decade, it can’t be compared with the 2011-12 survey, given the changes introduced. And while a comparison isn’t possible, an estimate of poverty can yet be made from the fact sheet. Since CES 2022-23 used the Modified Mixed Recall Period method, the only usable poverty-line option is that of the Rangarajan Committee, which was updated using Consumer Price Index (CPI) data. Using the Tendulkar panel’s poverty lines would be misleading, as these use the Mixed Reference Period method.

For academic interest, using updated Rangarajan poverty lines, rural poverty is estimated at 9.1% and urban poverty at 7.7%, with all-India poverty at 8.6%. As against a decline of 2.2 percentage points per annum between 2004-05 and 2011-12, based on official Tendulkar panel estimates, the decline between 2011-12 and 2022-23 is 1.9 percentage points per year. This implies an average of 22 million people emerged from poverty per year between 2011-12 and 2022-23, as against 20 million per year between 2004-05 and 2011-12.

However, given the survey design’s sweeping changes, both in methodology and implementation, estimating poverty will require a completely new set of poverty lines. So, for the time being, the vexed issue of what happened to poverty after 2011-12 remains unresolved. The issue of comparability and validation of CES 2022-23 estimates is also central to its use for the revision of GDP estimates and the CPI consumption basket. It differs from earlier surveys on three counts. First, unlike the established practice of conducting the survey in one visit, data was canvassed via three separate visits, with a month’s interval between each. This raises questions of the possible impact on cited consumption expenditure as a result of changes in respondents or household attrition in subsequent visits. Consumption aggregates are also affected by seasonality and other factors. Second, the inclusion of new items requires analysis of the impact it had on aggregate expenditure. Third, the basic survey design differs significantly from what the NSO had used before even for other socio-economic surveys. These issues affect not just the aggregate estimates we get, but also distribution across households. The only way of resolving its lack of comparability with past surveys is for the NSO to conduct a bridge survey with the old methodology and questionnaire.

However, even a limited analysis of Indian consumption and deprivation is only possible once the NSO releases unit-level data, with full documentation of attrition rates and so on. The CES 2023-24 survey findings must also be released as soon as available, as it was designed as a companion survey, along with a bridge survey. These surveys will be useful for a revision of national accounts and retail inflation estimation. As both these critical indicators are now based on foundational data that is more than a decade old, this exercise should be prioritized. Simultaneously, a new panel should be set up to suggest new poverty lines that are suitably modified for the new CES series to offer us a more reliable view of poverty.