Thread by @pramit_b

India's statistical system, once the envy of the world, has been decaying over time. But the crisis has never seemed as profound as it seems today.
This #thread is a summary of the symptoms, causes, implications, and possible remedies of that crisis (1/n)

The latest blow comes in the form of a misguided attempt to dig up dirt on the consumption survey report after deciding to bury the report, which had some unfavorable evidence on consumption spending. (2/n)
livemint.com/news/india/aft…

All the evidence so far suggests that the stats ministry's decision to bury the report was taken in haste after @someshjha7 blew the lid off the report, which had been inordinately delayed. (3/n)

https://twitter.com/someshjha7/status/1195147866559385600?s=20

The key reason for suppressing the report: the reported decline in real consumption, and the dissonance that it suggests between actual survey findings and the imputed consumption figures used in GDP calculations. (4/n)

Some people assume that GDP figures are always superior to survey figures but this is a mistaken notion, as a long list of research papers on this issue has shown. (5/n)

In a landmark 1988 study on the survey-national accounts discrepancy, late B.S. Minhas of I.S.I showed that national accounts estimates were more (not less) error-prone compared to the survey estimates. (6/n)
jstor.org/stable/2505255…

Many years later (in 2005), Nobel-winning economist Angus Deaton echoed Minhas' findings and rued the fact that not many discussing the issue had paid attention to Minhas' work (even in 2019, that lament rings true). (7/n)
princeton.edu/~deaton/downlo…

Minhas incidentally was perhaps the first case of a statistics-related resignation in the country; he resigned from Mrs. G's Planning Commission over inflated growth projections (not GDP estimates) (8/n)
nytimes.com/1973/12/10/arc…

Despite his protest, he was not shunned and his advice continued to be sought by both the statistical and planning establishment. In fact, his 1988 paper led the Planning Commission to abandon the practice of scaling up survey estimates to match national accounts estimates. (9/n)

While the average citizen may be unaware of all this, India's statistical leadership is fully aware of these issues. Hence, their attempts to portray national account estimates as statistically pure and survey estimates as somehow more contaminated is problematic. (10/n)

This is especially troubling because despite all its flaws, survey work is transparent and raw data is available for scrutiny. Neither is true in the case of GDP estimates; and even experts involved in the latest series have grave doubts over it. (11/n)
livemint.com/politics/polic…

To this date, the ministry has not opened up the controversial MCA-21 database to the public nor come clean on how exactly it is refined for use in national accounts even as serious problems relating to it have come to the fore. (12/n)
livemint.com/news/india/new…

A summary of some of the key issues relating to the GDP series is here (livemint.com/news/india/the…) and also in this earlier thread:

https://twitter.com/pramit_b/status/1138732530767618048?s=20

(13/n)

The refusal of India's statistical leadership to engage with data-related questions, and their attempts to bury anything that is not politically suitable is rapidly eroding trust in official statistics, and hurting India's reputation (14/n)
cgdev.org/article/indias…

This is especially unfortunate since India was once renowned for its statistical institutions globally, and set benchmarks for the rest of the world (including advanced economies) to emulate. (15/n)

The roots of the statistical crisis lie deep in India's statistical architecture and is a legacy issue. (16/n)
livemint.com/news/india/how…

But this govt. has let matters slide even further, instead of helping the statistical system reclaim its past glory. And it has set dangerous precedents that endanger our ability to hold future governments to account using facts and figures. (17/n)

Can this slide be reversed? Yes, it can. But it would require a lot from our statistical and political leadership to rebuild trust in our statistical systems once again. And maybe it would require new institutions (such as a statutory NSC/statistical auditing systems) (18/n)

If they are really interested in this agenda, they need to just look at our own history of what an inspired and talented leadership could achieve even in a poor country, even if not everything was perfect even then. (19/n)
livemint.com/Opinion/aMKJ70…

Recovering from the statistical crisis, and keeping stats politics-free won't be easy, but as Nagaraj had pointed to us sometime back, that's part of the hard work involved in making democracy work. (20/n)
livemint.com/news/india/gov…