National Council of Applied Economic Research
Opinion: Sonalde Desai.
Competing newspaper op-eds by PM’s Economic Advisory Council and two former chief statisticians have led to back and forth about the weakness of our statistical systems and political motivations behind impugning statistics. However, they offer a silver lining by highlighting the importance of statistics for evidence-informed policymaking. Several observations are noteworthy:
Indian national statistical systems are in crisis. Either data needed for policymaking is not available or not deemed reliable. Examples of this crisis abound. The sitting chief statistician of India argued that data from the flagship consumption expenditure survey of the National Sample Survey from 2017-18 was unreliable and suppressed the results. This left us floundering without official data on poverty. India’s unbroken record of conducting a decennial census since 1861 was broken during the pandemic. While this delay was understandable at the height of the pandemic, a continued delay is inexplicable.
Indian statistical infrastructure has not kept pace with global advances. Past decades have seen tremendous advances in methods for data collection and collation. Computerised surveys, including in-person, telephone, web, and text-based interviews, require a different way of thinking in questionnaire design. They also allow better quality monitoring using audio recording, interviewer prompts, and keystroke analysis. Advances in sampling methodologies include responsive survey design, where the interviewers are asked to make a greater effort in interviewing individuals with rare characteristics (eg, college graduate tribal women) than those with more common attributes. We have not developed institutional mechanisms for keeping up with global advances or ensuring that unique Indian conditions (eg, linguistic diversity) are incorporated into these advances.
We tend to react to global discourse rather than lead it. Global index construction has become a minor industry, with international “experts” often unable to incorporate regionally specific conditions. This can both inflate and deflate India’s position in the world with little relevance for ground realities. The now discredited Ease of Doing Business index of the World Bank showed significant improvement for India because it used electricity connections as an indicator, not the reliability of electric supply. In contrast, India performed poorly on the World Economic
Forum Gender Gap Index. It includes the gender gap in wealth but does not include the gender gap in poverty. India would fare better on gender gap in poverty than the US because while the overall standard of living is higher in the US, American women suffer from greater poverty than men due to high rates of divorce and single parenting, which is not a factor for India. Advocacy for adding contextual reflections to the applicability of global standards is not always politically motivated, although in some cases, it can be.
While acknowledging these mounting challenges, how can we rebuild India’s once-vaunted Mahalanobis-designed statistical system? This cannot be done without a thoughtful redesign of our statistical infrastructure with a substantial government commitment.
The first principle of a new system should be that statistics are too important to be left to statisticians. In a modern world, we need an interdisciplinary approach that brings together diverse social science and public health domain experts, research methodologists, statisticians, data scientists, and computer scientists. Our investments in research methodology stopped when the research unit for National Sample Survey housed in Indian Statistical Institute was disbanded. We need a new interdisciplinary institute, possibly under the aegis of the NITI Aayog, where innovations in research methodology can be explored, allowing us to keep up with global advances while responding to uniquely Indian conditions. Holistic training programs in research methodologies should be developed in diverse universities.
Second, as Pramit Bhattacharya notes in a paper, the ministry of statistics and programme implementation barely receives 0.2% of central government expenditure, and 3/4 of that is for the MPLAD scheme, unrelated to statistics. We need greater monetary investment in the data collection systems combined with greater demands for data quality and oversight. The dissolution of the NSS governing council has left a hole that must be plugged. The National Statistical Commission should be fully staffed and empowered.
Third, we must be realistic about what we expect from statistical systems. The demand that the National Family Health Survey (NFHS) provide district-level estimates of health indicators led to a seven-fold increase in sample size between NFHS-3 and NFHS-4, which has affected its data quality, as recorded by several academic articles. Moreover, even with this explosion, sample sizes are insufficient to provide district-level estimates of indicators such as infant mortality.
Trust in statistical systems is much easier to lose than to gain. Without a consistent commitment to systemic reforms, this war of words will diminish our confidence in statistics and not help in policymaking that relies on evidence rather than ideology.