R Nagaraj, professor at the Indira Gandhi Institute of Development Research (IGIDR), had disputed the Central Statistics Office (CSO) methodology on non-financial private corporate sector, which is a segment in the new GDP data. His criticism was mainly on the ground that CSO altered the methodology in 2015 from what was agreed on by a sub-committee, in which Nagaraj was a non-official member, in 2014. CSO gave a rejoinder to Nagaraj’s criticism in the Economic and Political Weekly. Nagaraj tells Indivjal Dhasmana why he was not convinced by the reply. Edited excerpts:
CSO has replied to your doubts over differences in the data on the non-financial private corporate sector that was given in 2014 by a CSO sub-committee and the one finally submitted in 2015. Why do you say seeds of doubts remain?
In its rejoinder, CSO sought to explain the reasons for the difference between the 2014 edition of the sub-committee report on “private corporate sector and PPP”, and its 2015 edition. The rejoinder said the differences were in the levels of the estimates but not very much in their growth rates for 2011-12 and 2012-13.
The differences in the levels were on account of two changes CSO made in the methodology: (i) It replaced aggregate revenue estimates by disaggregated revenue estimates, as CSO felt the latter were a better representation of corporate sector performance, and (ii) scaling up (blowing up) the estimates obtained for 520,000 companies, to 900,000 “active companies” which had filed their financial returns with the Ministry of Corporate Affairs (MCA) for at least one year of the past three years. We have argued that both improvisations are questionable and, hence, the seeds of doubts remain.
You countered CSO’s methodology to use disaggregated data of companies’ revenue stream of MCA-21. What is the problem in that methodology?
In a company balance sheet, by definition, aggregate revenue should be the sum of its components. If it is not, the balance sheet is faulty. One cannot, beforehand, claim the components of the revenue show a truer picture than the total revenue reported, unless one has prior knowledge to make such a claim.
CSO has in fact made such a claim when it said: “… it was felt that the individual components under total revenue would reveal the real picture of the economy in a better way. Hence the restriction was removed and output as per database was estimated using the individual components under total revenue in the database” (p 87). On what basis could CSO make such a claim? This is not for a single company but 520,000 companies that submitted their financial returns. If the components did not add up to the total revenue, the database is incomplete and inconsistent. Hence, the methodology seems faulty.
You also opposed the move to blow up estimates of companies given in MCA-21. But, CSO says that blowing up was used even in Reserve Bank of India (RBI) data. What is your objection?
Historically, when the non-financial private corporate sector (PCS) was small, RBI’s purposive sample of about 2,500 high paid-up capital companies was used to obtain the estimates which were then “blown up” (multiplied) for the universe of registered (or working) companies. However, with a sharp rise in company registrations after the economic reforms, but without all these new companies filing statutory financial returns, the true number of working companies has become hard to know (as acknowledged in the National Statistical Commission Report in 2001). Hence, the blowing up procedure got discredited as it overestimated the size of the PCS. In fact, the principal reason for replacing the RBI sample with the MCA database was precisely to avoid the much-discredited procedure.
You said that blowing up was much higher in 2013-14 data because the number of companies that reported their data to MCA-21 was just 300,000 instead of 500,000 in earlier estimates. Do you think that once the final results of MCA-21 comes, and GDP data are revised, we will be closer to reality?
I do not know. As the method is faulty how can one be sure of the veracity of such estimates?
Why do you say blowing up may be statistically correct, but not right to economists. Is GDP calculation a statistical method or a subject of economics?
GDP is an economic concept. Statistical methods are the tools used to estimate GDP. When I said the blowing up might be correct statistically, it means that from a statistical viewpoint, we have estimates for the universe (however defined) of companies. But the underlying economics of such estimates were suspect because most of the companies remained on paper, without producing goods and services. Hence, one may be estimating the contribution of companies that may be producing little economic output.
What is an alternative to blowing up methodology then?
This is a challenge before economic statisticians. My suggestion is as follows: Get the data for all public limited (or listed) companies on a census basis, and obtain the estimates for private limited companies on a stratified sample basis, and add them up. This would perhaps reduce (i) the error on account of the blowing up procedure, and (ii) the yearly fluctuations that are likely to occur in the number of companies filing the financial returns.
In layman's term, could you explain whether CSO overestimated GDP data for 2013-14 or also for 2012-13 and 2014-15?
As I have shown in my reply (and many others have also reported it) the growth rates as per new GDP estimates for 2012-13 and 2013-14 (full year) are distinctly higher than the estimates reported in the older (2004-05) series. Since the higher growth rates are at odds with other economic indicators (such as growth in bank credit) there is widespread scepticism about the new GDP estimates.