What really happened?
Welcome to the world of frequent revisions of economic data in India! The CSO releases as many as six estimates of the country’s annual economic output growth. Believe it or not, these estimates for the same year’s economic output are released over a period of three years!
Thus, the GDP growth figure for 2016-17 was first released in January 2017 as the First Advance Estimate. The Second Advance Estimate was released a month later by the end of February 2017. The third estimate is called Provisional Estimate and was released in May 2017. And then there is a long gap of about eight months and the fourth estimate, the First Revised Estimate, was released by the end of January 2018. Two more estimates for 2016-17 GDP were released — after a gap of one year each. The Second Revised Estimate for 2016-17 was released on January 31, 2019, and the Third Revised Estimate for 2016-17 was released exactly a year later on January 31, 2020.
Note that there were no changes in the first four estimates for GDP growth in 2016-17, all of which placed it at 7.1 per cent. The sharp variation began with the Second Revised Estimate released in January 2019 (8.2 per cent), followed by a further change in the Third Revised Estimate released in January 2020 (8.3 per cent).
The trend of sharp variations in the GDP growth was maintained in 2017-18 as well as in 2018-19, although the trajectories were a little different. For 2017-18, the First Advance Estimate put GDP growth at 6.5 per cent, which was revised upwards just a month later to 6.6 per cent as the Second Advance Estimate. The Provisional Estimate raised it again to 6.7 per cent, when it was released in May 2018. And the First Revised Estimate, released in January 2019, revised it further to 7.2 per cent. The Second Revised Estimate, released in January 2020, scaled it down a bit to 7 per cent. The Third Revised Estimate will be out next January. And a new number may well be released then.
For 2018-19, only the first four estimates have been released so far. In a space of one year, the GDP growth figure has already seen four different versions — starting with 7.2 per cent in January 2019, it has gone down to 7 per cent in February 2019, 6.8 per cent in May 2019 and further down to 6.1 per cent in January 2020. Unlike in the previous two years, the different estimates for 2018-19 have successively brought down the growth rates. For the 2019-20 growth number, the first two estimates are out, both placing it at 5 per cent and without any variation.
It is not just the GDP growth numbers that have seen such variations in different estimates, but also other parameters like the gross value added at basic prices, private final consumption expenditure, government final consumption expenditure and gross fixed capital formation.
Economic output growth numbers are an important tool for policy makers to assess the economy and recommend necessary policy measures. If these numbers are changed significantly in six different estimates, the efficacy of such policy intervention will naturally suffer. The problem arises not just from the revisions, but also from the inordinately long period of three years over which such changes take place. And every time one of these six estimates undergoes a change, either way, the estimates for the following years are also necessarily revised.
Of course, the CSO explains every time there is a variation in a fresh estimate. But that does not help tackle the challenges for policy makers. Questions on data reliability or integrity are also not addressed. Additionally, such changes make the data release system vulnerable to political pressures to modify or tinker with numbers to suit a specific political narrative. While the CSO must look into the methodologies of arriving at its various estimates and examine necessary remedial steps to make the system more robust, it must also evaluate the need for as many as six different estimates of the same output growth numbers and, that too, over a period of three years. Surely, reducing the number of estimates and the span of time over which they are released would go a long way in addressing the growing concerns over data reliability and its vulnerability to political pressure.