Get all your news in one place.
100’s of premium titles.
One app.
Start reading
Pramit Bhattacharya

How to correctly assess the debate on extreme poverty in India

Photo: Bloomberg

In the absence of any nationally representative data on household consumption expenditure (required to calculate household poverty), both studies have used a number of heroic assumptions and imputations to estimate India’s poverty rates. Such heroism would not have been necessary had the Indian government been able to conduct and release consumption surveys on time.

The last National Sample Survey (NSS) consumption round, conducted in 2017-18, was suppressed by the government. There hasn’t been any fresh survey since then. The Centre appeared to reject those findings because it presented an unflattering picture of rural consumption. Had the results been accepted, it would have shown a bump in poverty.

The statistics ministry said it was holding back the data because of differences between its findings and consumption data from the national accounts statistics (NAS). It cited an adverse report from an expert committee to justify its decision, and sought to find flaws in the survey post-facto (see ‘After junking report on consumption, NSO now looks for reasons’, Mint, 22 November 2019). But the expert committee in question did not recommend suppressing the report. It noted the difference between NSS consumption estimates and those of the NAS, but as a member of that panel pointed out to this writer, such gaps are not new and do not justify suppression of the report or raw NSS data.

Discrepancies between consumption data reported by the NSS and NAS (the latter is used in calculating gross domestic product or GDP) has been a matter of debate since at least the 1980s. In a major 1988 study, the renowned economist B.S. Minhas examined this issue and argued that there is no reason to privilege NAS estimates over those of NSS surveys. Minhas’ findings led the Planning Commission to abandon its practice of scaling up NSS estimates to match NAS figures. Several years later, in 2005, Nobel-winning economist Angus Deaton noted that Minhas’ observations held true even then. Deaton argued that the NAS estimates were based on several imputations, used rates and ratios going back decades, and should not be assumed to offer a more accurate picture of consumption trends compared to surveys.

The most recent official panel to examine this issue was led by A.K. Adhikari, a former professor of Indian Statistical Institute (ISI) Kolkata; it included senior statisticians from both the national accounts and survey wings of the statistics ministry. In its report submitted in 2015, the committee recommended that the national accounts division (NAD) should update the old rates and ratios (some of which date back to the 1960s) in a phased manner after conducting country-wide studies. It also recommended that the NSS team should consider shortening the survey questionnaire to lower respondent burden and generate more accurate responses, but after doing experimental surveys with alternate questionnaire formats.

Hence there is already considerable agreement within the statistical establishment on resolving the NSS-NAS divide and produce more accurate data on consumption, poverty, and GDP. While it is likely that NAS overstates consumption expenditure, it is also likely that NSS underestimates it. Over the years, NSS consumption schedules have grown longer as new items have been added to reflect changed patterns. However, queries on new items are typically placed at the end of the question set, so respondents may be too fatigued to reply properly, even though these items make up a large part of India’s recorded GDP.

The key reason why old items have not been junked in the NSS schedule relates to the issue of comparability. Any change in a survey instrument—in either the choice of questions that are asked or in their sequence —can impact comparability with past rounds. Given that poverty estimates derived from consumption surveys are used to assess India’s economic progress over time, NSS staff have been extremely cautious in introducing such changes. But there’s a trade-off between comparability and accuracy.

It is important to recognize this trade-off, and find a way to shorten the survey questionnaires that improves accuracy without compromising comparability too far. It is unrealistic to expect full comparability. But it is not unreasonable to expect our statistical system to generate an adjusted back-series of consumption data if the new survey instruments can’t be made broadly comparable with the past. Such a back-series should be published while rolling out the changes rather than years later. This would prevent a repeat of the GDP back-series fiasco.

As recommended by the Adhikari panel, these changes should take place only after pilot surveys have been conducted to assess the impact of changing survey schedules. The results of these experimental surveys should be released publicly, so that data users, including policymakers, academics and journalists, are well aware of the implications of the changes. This will ensure that the changes are not unduly politicized.

The process by which these changes are effected will ultimately determine whether they are seen as well-considered and well-intentioned reforms.

Pramit Bhattacharya is a Chennai-based journalist. His twitter handle is pramit_b

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
One subscription that gives you access to news from hundreds of sites
Already a member? Sign in here
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.