India’s draft policy on official statistics misses the mark

The current draft policy appears to be a diluted version of a previous draft published in 2018. That was two years after India adopted the UN Fundamental Principles of Official Statistics which aim to protect the integrity and impartiality of statistical offices while guarding the privacy of citizens. The 2018 draft had a clear pathway to insulate statistical activities from political pressures. It envisaged greater financial and operational autonomy for the National Statistical Commission (NSC).

The NSC has awaited statutory backing for nearly two decades now. The lack of an effective statistical regulator has meant that Indian statistics are not subject to regular quality assessments or audits. Data users have to rely on intuition and guesswork while dealing with data-sets of varying quality in an uneven data landscape. Controversies over statistical output remain unresolved for years on end, and are sometimes unduly politicized. As regular readers of Truth, Lies and Statistics would know, the lack of an empowered statistical regulator has stunted the development of the Indian statistical system.

The 2018 draft recognized this problem in emphasizing the need to empower the NSC. Even the ministry of statistics and programme implementation (Mospi) acknowledged this issue. “Though the NSC, has since its inception, made a number of recommendations to improve the statistical system through its Annual Reports, yet, major structural changes remained mostly unexamined," said the ministry’s vision document for 2019-24 published in 2019. “The mandate of the NSC is far reaching and for it to perform its functions effectively, it needs to be suitably reinforced with appropriate professional autonomy in making implementable recommendations." However, the current draft policy does not have anything to say on empowering the NSC. A code of statistical practice proposed in the earlier draft is also missing. Among other things, that code laid down transparency norms to be followed while releasing data-sets. The 2018 draft policy raised hopes that Mospi was serious about addressing India’s statistical challenges. The current draft belies those hopes.

The current draft rightly emphasizes the need to harness administrative data-sets. Many databases, such as those related to vehicle registration and taxation, are generated as a by-product of routine administrative functions. So they are relatively inexpensive. Yet, the uneven quality and coverage of these administrative records, the wide variation in definitions used by different departments and frequent changes in administrative rules relating to these data-sets often make it difficult to use them for statistical purposes. These challenges aren’t insurmountable, but they must be taken into account while incorporating these data-sets in regular statistical products. The draft policy acknowledges some of these challenges.

However, the draft policy goes on to say that the use of administrative data would minimize the use of surveys. This seems to stem from a serious fallacy. Even countries with far better administrative data-sets such as the US and UK haven’t given up on household or business surveys. The Office of National Statistics in the UK runs 82 business surveys while the US Census Bureau runs more than 100 surveys of households and businesses every year.

Although greater use of administrative data can reduce reliance on some kind of surveys, they do not eliminate their need in many other areas. Most large economies depend on detailed questionnaires sent to businesses to track changes in business activity and economic parameters. Adding these questions in regular administrative forms can raise compliance costs, especially for small firms. Hence, only minimal information is canvassed through administrative records. To obtain detailed data, a randomly chosen set of firms is surveyed. This need is greater in countries that have incomplete administrative records and large informal sectors.

Also, note that surveys and administrative records are not always substitutes. Often they are complementary. In fact, improvements in administrative records can help improve survey quality. For instance, if factory inspectors are able to improve the coverage of factories in each state, it would help to improve the coverage and quality of the Annual Survey of Industries, whose sampling frame is based on lists supplied by factory inspectors. Surveys, in turn, can be used to track under-coverage of administrative data-sets. The Registrar General of India relies on surveys to check the coverage of birth and death registrations.

Overall, Mospi’s draft policy does not seem to provide a clear and coherent framework for rebuilding India’s statistical edifice. We need to set up a high-level statistical reforms commission headed by a non-partisan statistician or technocrat. This body should present a detailed statistical strategy document after consulting all key stakeholders, including India’s community of non-official data users. Our citizens, policymakers and investors deserve no less.

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