Aniruddha Chowdhury/Mint
Aniruddha Chowdhury/Mint

Opinion | The Central Statistics Office is an institution that matters

The new GDP series continues to pose a problem because of the constant revisions being done within its period of coverage

We don’t need economists to tell us this, but institutions matter. In an empirical literature too large to cite, the strength of institutions is shown to be what sustains any country through good times and bad. We have many strong institutions in India, only we don’t think about them. The best institutions don’t shout their contribution from the rooftop, but do their work quietly to keep the country humming.

One such is the Central Statistics Office (CSO, renamed from what was once the Central Statistical Organization). National accounts estimation in colonial times was mostly confined to economic activity within the boundaries of British-administered India, and with a limited focus on goods and large-scale service networks like the railways. After Independence, the territorial coverage had to be expanded to include the former princely territories, and the service sector needed inclusion in all its variety and spatial spread.

Extending the post-Independence estimation back to the start of the 20th century was painstakingly performed by a statistician named S. Sivasubramonian, who established that per capita income had remained essentially stagnant during those 50 years. When we remember that several segments of society saw substantially enhanced income during colonial rule, it became clear that the remainder must have become worse off if the average had remained the same. No further explanation is needed for why the movement for political independence gained such widespread traction.

Fast forward to the present, when the gross domestic product (GDP) issued by the CSO is the universal denominator, used to judge fiscal performance, public debt trajectories, financial sector depth, you name it. We took the GDP number for granted because of the professional care which had gone into its estimation, just as we sit on any chair presuming that the skill of the carpenter who made it will ensure the chair supports our weight.

Some years ago, a new series of GDP was issued with base year 2011-12, with a substantially altered method of estimation. Extending the new series back to earlier years has been mired in controversy.

The back series controversy is not that serious, only because people can continue to use the old real growth rates for years prior to the new base year, and shift to the new growth rates for the years after. As far as the nominal aggregates are concerned, a simple splicing factor can be constructed by every user.

But the new series continues to pose a problem because of the constant revisions being done within its period of coverage. The entire series going back to 2011-12 was revised two years ago, on 31 January 2017, followed by another overhaul of the entire series on 31 May 2017, and yet again on 31 January 2018. On these standard issuance dates, only recent year data are normally revised. This year, on 31 January 2019, there has been yet another overhaul of the figures going all the way back to the base year 2011-12.

This latest revision does not alter the GDP growth rates for the early years too much, but growth in 2016-17 has been radically revised upwards, from 7.1% to 8.2%. There is a less radical upward revision for 2017-18 as well, from 6.7% to 7.2%. There will have to be a corresponding revision in the second advance estimate for 2018-19, due on 28 February 2019.

Gross value added (GVA), which is a measure of production without the overlay of indirect taxes like the goods and services tax (GST), or the corresponding pre-GST levies, has been revised upwards for 2016-17 as well, from 7.1% to 7.9% (and less radically for 2017-18, from 6.5% to 6.9%).

I prefer to look at what I call core GVA, after removing the agriculture and government sectors, not because they don’t matter but because they are subject to exogenous vagaries like the weather and Pay Commission upgrades.

The upward revision in core GVA for 2016-17 is more startling still, from 6.7% to 8%. The further widening was on account of the government sector, where the growth rate has been revised sharply downwards from 12.7% to 8.6%. Such a sharp fall, by 4 percentage points, is most puzzling for the government sector, since final audited figures of accounts for 2016-17 would have been available at the time of the last revision, on 31 May 2018.

With government sector growth revised downwards, core GVA sectors have seen a more sharp upward revision in growth rates than GVA as a whole. Growth in the construction sector in 2016-17 (the year of demonetization) has been revised upwards from 1.3 % to 6.1%. Construction estimates are normally based on upstream output of manufactured inputs like steel and cement, and are not normally revised so substantially. Other sectors that have been revised upwards, though not as much, are financial services and real estate (6% to 8.7%), trade and transport (7.2 % to 7.7%) and electricity and utilities (9.2 % to 10%).

A few points in conclusion. Most analysts I know are entirely apolitical, and are quite willing to accept the new revised growth rates, if they are satisfactorily explained. The notes accompanying the 31 January revision do not do this. We had a statistical commission, which was supposed to approve any revisions before they were made public. This time the revisions were not passed by them, and two highly respected members resigned as a consequence.

Policy on all fronts will be stymied if the GDP numbers are fluid and subject to constant change.

Indira Rajaraman is an economist.

Close