The Devil’s in the Data

gdp back series data

I could end the deficit in five minutes. You just pass a law that says that anytime there is a deficit of more than 3 per cent of the GDP, all sitting members of the Congress are ineligible for re-election, Warren Buffet, the maverick investor, famously said. A ballooning fiscal deficit and sliding growth are every government’s nightmare. Data can make or mar reputations. In India, as the economy grows and diversifies, the problem of capturing the growth story in numbers has left policymakers flummoxed. GDP revisions are the norm, as are various other data revisions, including of industrial growth and trade, which form the basis of GDP calculation. Such revisions seldom create a stir, except when they impact past data and put a former government in dimmer lightas we witnessed last week.

The back series of the GDP for the years 2004-05 to 2011-12 has kicked up quite a storm, with some sections of the business media even calling for its withdrawal. The reason? The data throws up numbers which suggest that in the years when the previous government was in power, growth was lower than widely believed, and more importantly, lower than the growth record of the present dispensation. This statistical assertion has surprised economists and analysts alike, since it contradicts virtually all other data on the real’ economy, including on corporate sales, investment, credit growth and revenue from taxes, among others. The debate opens a can of worms in a country where both the dearth of data and the credibility of available data have become a matter of concern. For instance, the credibility of jobs data has been debated over. A new way to gauge job creation, by using data on new additions to the government’s provident fund scheme, has faced flak for not reflecting the actual situation on the ground. There have been conflicting theories on the impact of demonetisation, with the agriculture ministry first saying that the withdrawal of high-value currencies announced in November 2016 actually hurt farmers, only to reverse its opinion within days and put a question mark on the data used for the analysis. The biggest casualty of frequently changing or unreliable data are new investments, because investors cannot take a secured call on where to put their money. Frequent and drastic changes in data can send conflicting signals to investors, says D.K. Joshi, chief economist at ratings agency Crisil. For a country where private investments have been on the downslide for some time now, more unpredictability on data is a further dampener, which could prompt them to shelve investments or put their money and resources to use elsewhere.


The concerns that have arisen about the backdated GDP data under the new series, with 2011-12 as the base year (compared to 2004-05 as the base year earlier), are numerous, but the one that has caught many by surprise is the comparative data on the United Progressive Alliance (UPA) and National Democratic Alliance (NDA) years, now suspiciously resolved to render the NDA years superior. The new data says that the economy grew at an average 6.7 per cent in four years of the first term of the UPA government (2005-06 to 2008-09) as well as in its second term (2009-10 to 2013-14). This is lower than the earlier estimates of 8.1 per cent and 7 per cent average growth rate (calculated with 2004-05 as the base year), respectively. These growth rates are lower than the average 7.4 per cent growth rate (calculated with 2011-12 as the base year) seen during the first four years of the present NDA government. Experts have questioned some of the glaring anomalies in this calculation. For instance, growth during the economic boom’ of 2007-08 has now been downgraded from 9.8 per cent under the old series to 7.7 per cent under the new series. This is only a shade higher than the 6.4 per cent growth registered for the year 2013-14, the last year of the UPA government, which witnessed considerable turmoil, marred by Coalgate and the 2G scams, flight of capital and policy paralysis.

This back series issue has generated many more questions than it has answered, says Maitreesh Ghatak, professor of economics at the London School of Economics. I am not aware of any other country which uses these back seriessometimes you have to combine old and new series and there are standard methods of doing that. He says that anyone who deals with statistics knows that by choosing a suitable base year or suitable price deflator, one can have considerable leeway in terms of how one can make growth look in a particular sub-period. (A deflator is a value that allows data to be measured over time in terms of a base period, usually through a price index.) This seems a purely political exercise to make growth under the current regime look good, says Ghatak. And, yes, it does damage the credibility of our economic statistics, which is especially unfortunate because despite being a low-income country, India has a very distinguished history of producing high-quality economic statistics and enjoys worldwide respect for that.

If the performance of certain individual sectors was a key indicator of economic growth, then the UPA years clearly saw better growth than the NDA years. Reports suggest, for example, that the average annual growth in domestic car sales between fiscals 2005 and 2012 was 13.8 per cent, while it was just 1.1 per cent between fiscals 2012 and 2018. Two-wheeler sales growth in the first period was 11.6 per cent as against 7.1 per cent in the second. Growth in corporate tax was 21.5 per cent in the first period compared to 10 per cent in the second, while growth in non-oil exports stood at 18.4 per cent compared to just 1 per cent in the second period.


The average annual growth rate between 2005-06 and 2011-12 got reduced by 1.3 percentage points, from 8.2 per cent in the old series to 6.9 per cent in the back series. This is unprecedented, says R. Nagaraj, professor at the Indira Gandhi Institute of Development Research, Mumbai. In any such exercise, it is important to check if the estimated values are broadly consistent with related macro variables, he adds. The reduction in GDP growth rates for the second half of the past decade does not seem to square with trends in saving, investment, foreign capital inflows, exports and so on, and hence the doubts about the veracity of the back series. The Central Statistics Office (CSO) should publish the details of the underlying methods and procedures in a transparent manner to silence the critics, adds Nagaraj.

Contrary to practice, CSO and NITI Aayog jointly held a press conference on November 28 to release the new series data, raising eyebrows. It’s a clear shift that the NITI Aayog got involved in the generation of the new series. One gets the suspicion that it was not done by professional statisticians, says Pronab Sen, a former chief statistician of India. Sen has criticised the series because it relies on value. For instance, if earlier the number of telecom subscribers was accounted for, the new series looks at the number of minutes consumed. I don’t think this is a better way of doing things. There has been no improvement in productivity or quality, adds Sen. Another economist, who did not want to be named, suggests that while the new methodology may be good, the data inputs are of poor quality. It’s like running a Ferrari on adulterated fuel, he says.


Defending the new back series, Rajiv Kumar, the government-appointed vice-chairman of NITI Aayog, says that economists and researchers had been demanding it for the past two years and former chief statistician T.C.A. Anant had said that the new data would be released by 2017. However, because of several coverage and methodological challenges, the process took longer than expected. When the CSO approached me about the work they had done, I suggested that they should get it validated by senior statisticians. I approached the series as a professional economist, he insists. Kumar rattled off a list of five statisticians who were involved in the vetting of data before it was presented (see interview).

The growth estimates for the UPA years run lower than what was presented by Sudipto Mundle, professor and member of the board of governors at the New Delhi-based National Institute of Public Finance and Policy, in August. At that time, the government had dismissed the new back series GDP data as not official, an argument it maintains even now. Mundle had reportedly called the estimates the second-best option in the absence of comparable data for the years up to 1993-94. The new back series uses data from the ministry of corporate affairs and is technically compliant with the United Nations guidelines in the System of National Accounts 2008.

One possible explanation for the big change in numbers is inflation, which was much higher during the UPA regime, in turn inflating nominal GDP growth. The decline in inflation in the past four years, says Joshi, accounts for the difference between real GDP (GDP adjusted for inflation) and nominal GDP growth. Consumer inflation between fiscals 2006 and 2014 averaged 8.5 per cent as against 4 per cent in the past four fiscals. But many economists remain sceptical. Yes, it clearly has to do with the deflator that is used, says Ghatak. But the problem is that the real GDP growth rate in the UPA years under the original CSO figures looks pretty good too. It does so under the back series that the NSC (National Statistical Commission) generated a few months ago. So, it is not just a matter of inflation being higher under UPAeven though that is true. It has to do with the specific price index use under the very latest back series calculations.

Is real GDP’ the wrong way to measure growth? Real GDP is not the yardstick. It should be nominal GDP. And that will show a different picture, says Joshi. Nominal GDP growth averaged 10.5 per cent in the past four years, or 500 bps lower than the trend growth of 15.5 per cent in the previous nine years. This partly explains lower credit growth, corporate revenues and tax collections in the past four years compared with the earlier period.

The controversy around the GDP back series data has once again underscored the need for trustworthy data. Whether it’s jobs, trade or factory output, all talk about data comes with some scepticism. Add to that the new data dashboards on every ministry’s website, which make interestingly sunny claims, and the confusion is quite palpable.