Dry statistics, especially that of a GDP back series going back a decade, are not supposed to set off furious political debates. But when Niti Aayog Vice-Chairman Rajiv Kumar, with CSO chief Pravin Srivastava in tow, decided to announce the back series calculations of 2004-10 finally on November 28, it immediately led to a war of words between the BJP and the Congress. Worse, it led to neutral professional economists scratching their heads and wondering what exactly was India's growth rate between 2004/05 and 2013/14.
The problem was that there were now three sets of figures for each of these years. One, just calculated by the CSO under the supervision of Niti Aayog, a somewhat unusual occurrence, and given the official stamp of approval. The second was a back series accidentally released by the CSO in August based on a rigorous calculation done by the Committee on Real Sector Statistics formed by the National Statistical Commission.
The National Statistical Commission had India's then chief statistician TCA Anant, as its secretary, while the Committee on Real Sector Report was chaired by Professor Sudipto Mundle. Finally, there was the original GDP data released by the CSO during the UPA years till 2013/14, until the methodology was drastically revised and the base year changed after the Modi government took over.
All three methodologies agree directionally on the GDP growth but the actual rates for each year are very different. While the release of the "official back series calculations" and its timing gave rise to a heated debate on how much one could actually depend on them, the GDP debate has actually started in 2015, when the CSO revised the methodology completely and started using new data sources, apart from changing the base year from 2004/05 to 2011/12
The base year revision is a fairly routine thing and it takes place at five-year intervals generally. Even methodologies and data sources are changed from time to time to bring in more accuracy to GDP calculations as the product and service landscape changes. Each time this is done, there is generally a calculation done for back series, but the exercise rarely results in any controversy. That is because the growth figures of earlier years do not change dramatically. Statisticians take care to ensure that no big distortion takes place because of methodology and data source changes.
The problem was that in 2015, when the CSO unveiled the GDP using the new methodology and using the base year of 2011/12, it also released its calculations for GDP of previous years only till 2012/13 - and the first thing almost everyone noticed how much higher the GDP growth under the new methodology was compared to the old one. For FY 2013/14 for example, the old GDP calculations showed a meagre 5 per cent growth - while the new methodology showed a much higher 6.4 per cent growth. Similarly, for 2012/13, the old methodology showed a growth rate of 4.7 per cent, while the new methodology came up with a more respectable 5.5 per cent.
The CSO said that the new methodology was more in line with global practices, and it also used a new database for estimating corporate growth and investment - the MCA 21 database of the Ministry of Finance. This is a digital database on which all listed and unlisted companies need to file their annual reports. Thus it makes available on one platform, enough data on corporate results to estimate industrial and services growth more accurately. However, the database was only available since 2006/07 and it had not stabilised till 2010/11.
The fact that the CSO released the figures without fully explaining the details of the methodology and because the estimates differed from other statistics gathered by other bodies, a number of economists were skeptical about the numbers. The initial GDP calculations, the new methodology and the variance in numbers confounded even someone like the then RBI governor Raghuram Rajan, who said this during a lecture in an institute: "There are problems with the way we count GDP, which is why we need to be careful sometimes just talking about growth,"
Though eventually the noise died down, the doubts did not go away altogether that the government was prone to pick data and methodology that would show it in better light. For example, despite the enormous disruption created by demonetisation, the GDP statistics released for the quarter immediately after it showed hardly any effect of the radical move initially. The problem was the initial GDP estimates were done using the early filers in the MCA21 database, which were all big companies least affected by demonetisation, and the extrapolation did not show the real effects on smaller companies and the informal sector. Despite that, the GDP controversy was slowing dying out, and economists and commentators were beginning to use it for all calculations.
The then chief statistician, TCA Anant, steadfastly refused to give a date for the release of the back series using the new methodology and finally retired on January 31, 2018. In July, the Committee on Real Sector Statistics submitted its report. The report was voluminous (178 pages of recommendations and explanations) and the back series calculations formed only a part - one chapter out of seven. The rest went into great detail about the problems with the existing data on industry, agriculture and services gathered by the government. It also gave a series of recommendations to fix those.
But the chapter on back series calculation grabbed attention. It showed that the UPA years showed growth that was better than originally estimated. The committee gave detailed explanations about how the calculations were done. The release was greeted with jubilation by the Congress and dismay by the BJP. The Congress crowed about how superior its handling of the economy was. But within two days, the report vanished from the CSO website and the government said the calculations were not official.
Meanwhile, Rajiv Kumar was increasingly becoming the government's main spokesperson when it came to defending its economic performance. He was also getting involved in trying to find alternative data sources which would help the government's image. One of his initiatives was to give two outside researchers access to Employee Provident Fund and other such data to estimate formal job creation in the economy, something that should have been handled either by the labour ministry or the CSO.
The controversy over the official back series now arose because Rajiv Kumar did much of the talking and despite the press note, there weren't enough details to make sense of why there was a sudden drop in growth estimates for all years before 2010/11. The new calculations had to use alternative data sources for almost all sectors because comparable data sources (including the MCA 21) were not available or comparable before 2010/11. Moreover, they made a number of assumptions - which also had an effect. However many economists found the calculations rather dubious given that the new numbers were at variance with other available statistics in that period - ranging from export growth to private sector investment to private consumption and even agricultural growth. Also, many economists pointed out that between 2004-08, and then again in 2010/11, people felt the economic growth in rising asset prices, jobs created, investments and increments. In fact, it was a puzzle how 2011/12 under the new calculations showed a worse slowdown than 2012/13, effectively reversing the calculations done under the old series.
So far, the government has not really come clear to explain the discrepancies fully. The problem is that it has muddied the water of the GDP so much that almost no one knows what to do with the numbers of the back series. As Vin Sculley, a former sportscaster in the US once memorably said: "Statistics are used much like a drunk uses the lamp post. For support, not for illumination." He said it in a different context, but it seems to fit the current situation quite well.