MUMBAI : Economist Rayaprolu Nagaraj has been a vocal critic of India’s new gross domestic product (GDP) series and has questioned the manner in which a new corporate database, MCA-21, was plugged into the national accounts by the Central Statistics Office (CSO) without “adequate scrutiny or debate".

In an interview, the professor of economics at the Indira Gandhi Institute of Development Research—a Mumbai-based research institute funded by the Reserve Bank of India (RBI)—and a former visiting professor at the Woodrow Wilson School of Public and International Affairs at Princeton University, argued that certain important questions regarding the new GDP series have not been “convincingly answered yet". The new GDP series has faced fresh questions after field officers from the National Sample Survey Office (NSSO) found holes in the MCA-21 database during a survey.

Nagaraj argues that data is political in all democracies, and that all governments would like to control it. But the challenge of making a democracy work is all about putting in enough checks and balances so that important institutions, including statistical agencies, can function with autonomy and transparency. That task won’t be easy for any institution, he said. Edited excerpts:

In its statement released on Friday evening, the finance ministry has said the key finding from the NSSO survey is that there is misclassification of companies. So the shares of different sectors may be wrong, but overall GDP is not affected. Do you agree with this statement?

There could be an issue of misclassification. But misclassification happens because articles of association are very ambiguous. If a regulator comes to your doorstep and enquires, you can lie about the nature of the activity being conducted. The balance sheets of such companies supplied to the ministry of corporate affairs (MCA) have very sparse information. They are not helpful to figure out where this activity should be classified. They are all ways to fudge information.

For me, the bottom line from the NSSO report is that 45% of the companies (or 38%, if you take a conservative estimate) could not provide even basic information to NSSO. This is the extent of lack of information about such companies. That is a serious matter— if you use a multiplier on all these to estimate GVA (gross valued added), you are overestimating because many of the firms are missing and other firms, which could not be surveyed, may not be producing anything either.

But the finance ministry response says the “blowing up" is for companies that account for only 12-15% of paid-up capital, and hence, even if there is an overestimation, this would at best be marginal and affect GDP levels rather than growth.

Their response assumes that the data for all companies that are filing in a year are usable and can be simply added up to arrive at the GVA. This is incorrect, as several critics and users of MCA-21 data have pointed out.

Even CSO, in its response to my first Economic and Political Weekly (EPW) article (“Seeds of Doubts on New GDP Numbers", EPW March 2015) on this issue in 2015, had admitted that the GVA estimates obtained using the total revenue figure in MCA-21 companies differed significantly from those obtained from the sum of its individual components. It means the balance sheets in MCA-21 database had internal inconsistencies; hence the reported GVA estimates were, and continue to be, questioned.

Mint reported last week that even an NSC committee on real sector statistics, headed by Sudipto Mundle, which included officials from CSO, had pointed to serious inconsistencies in the MCA-21 database in its report published last year. So, there are questions raised by both official and non-official users of the MCA-21 database. And I think these questions have not been convincingly answered yet.

The finance ministry response says the MCA-21 database is being improved...

Improvement does not merely mean striking off companies from the records, though that is important. For the purpose of estimating GVA or GDP, it is important for national account statisticians to know how much of the information being culled from the MCA-21 database can be used to derive a consistent time series. And there are serious questions about that.

The ministry response shows that there are more than half a million companies that file returns every year, but data from other sources seem to suggest that usable data is available for only about half of those firms.

You see, there is an agreement between MCA and RBI to share the data. MCA shares their full data with RBI. RBI is not supposed to share the data with anyone else, but they publish some tables in their bulletin. And their data never goes beyond 2.5-3 lakh (250,000-300,000) companies and they get all the companies for which MCA gets data as per the agreement and they are both official agencies.

This means whenever RBI tries to analyse the data, it cannot do it for more than 2.5-3 lakh companies. I think that’s telling evidence: that there are only 2.5-3 lakh working companies in the country.

The ministry of statistics and programme implementation in its response earlier had said it was planning the seventh economic census, which would create a national business register. Can we rely on these numbers given what the NSSO report has shown about the other enterprise databases?

The NSSO report had three frames. MCA-21 is the largest of the three. There are problems with the other two, but the debate now is about the MCA database. As far as GDP is concerned, MCA-21 is where the issue lies—that’s where the 38% or 45% is coming from. And the NSSO report has highlighted once again that there are serious problems with that database. That’s the crux of the debate.

To my limited knowledge, efforts to create business registers have not been much of a success, as is also evident from the NSSO report.

What is your view on the GDP back-series controversy (an NSC committee that included CSO officials came up with one set of estimates and then CSO came up with another set of estimates)?

How CSO arrived at these numbers is far from clear. They have used a different method, but not explained it clearly. You can do what you want to do, but if you explain to the world that this is what we have done and we think this is right, then we can debate. If you don’t disclose how these numbers are arrived at and say we are right, then it becomes difficult. And if by chance this helps a political stand, unfortunately these numbers lose credibility.

Your views on NSC?

Will only say one thing: NSC came about as an outcome of Dr. C. Rangarajan’s NSC report. We, as users of data, were very happy that finally there will be an independent commission that will oversee everything, but what has actually happened is very different. Mint’s Long Story (“How India’s statistical system was crippled", 8 May 2019) captures most of it. Beyond that, what it should be, I haven’t thought about it.

How do we insulate statistics from politics?

In all countries, data is politics. Especially in democracies: where it is contested. So democracies will have to build checks and balances. These are hard to build and easy to destroy. There is a lot we need to do since our institutions are very weak as we have seen in some recent cases in other arenas.

The question is how you build checks and balances so that they remain independent of politics. It’s very hard but that is part of the challenge of making democracy work.

What is the way forward for estimating GDP better?

One should be honest in admitting things are not right and then there should be an independent commission of the best international experts to do a statistical audit of GDP. In our contemporary experience, many countries have done that. There is no reason why we shouldn’t. Argentina did it just few years ago, with their inflation numbers.

There is a tendency of the ruling parties in various countries to see the numbers they want to see. It’s a global issue. The national elite should take a farsighted view to set the macroeconomic data system in order.

Close