India’s latest GDP numbers show the economy growing at 7.8%, but how much confidence should we place in that figure? In this episode of How India’s Economy Works, Puja Mehra speaks with Rajeswari Sengupta, Economist and Associate Professor at Indira Gandhi Institute of Development Research (IGIDR), about the complexities of measuring India’s GDP, the latest base-year revision and the methodological changes behind the new estimates.

Sengupta explains why India’s large informal economy, limited data, rapid structural change and the absence of comprehensive income data make GDP estimation particularly challenging. She also breaks down the shift from single deflation to double deflation, the use of producer price indices, and why the statistical office’s lack of detailed sources-and-methods documentation has raised concerns.

The conversation also examines the use of GST data and surveys of the informal sector, the credibility of the new GDP series, and the unusual spending boom implied by the latest quarterly numbers despite a sharp rise in import prices. Sengupta argues that the debate should be about transparency and statistical methodology—not politics.

Tune in for insights on India’s GDP measurement, the new base-year revision, double deflation, data transparency, and the puzzle surrounding the 7.8% growth figure.

CHAPTERS

(00:00) Introduction to Indian GDP Estimation

(01:19) Challenges in Indian GDP Measurement

(09:47) Uncaptured Services and Gig Economy

(15:19) Controversies of Past Base Revisions

(21:42) Single Versus Double Deflation Explained

(26:04) Shift from WPI to PPI Deflation

(32:43) Missing Methodology and Sources Document

(35:10) Doubts Around Quarterly GDP Numbers

(40:01) Mismatched Identifiers in GST Data

(46:12) Political Pressures on Statistical Agencies

(52:48) The Terms of Trade Shock Puzzle

—

TRANSCRIPT

NOTE: This transcript is done by a machine. Human eyes have gone through the script but there might still be errors in some of the text, so please refer to the audio in case you need to clarify any part. If you want to get in touch regarding any feedback, you can drop us a message on feedback@thecore.in.

Puja Mehra: Rajeshwari, welcome to the show once again. Thank you so much for coming.

Rajeswari Sengupta: Thank you, Puja. It's always a pleasure to be here.

Puja Mehra: So, GDP estimation remains in controversy as it has been for a long time, intermittently. There could be many valid and not so valid reasons for this, but I feel one basic reason is that it's very few people, even many economists, don't fully grasp how GDP is estimated, what happens when there are revisions, when the base changes, and all the challenges that there are in GDP estimation, especially in a country like India where you do not have too many data sources for large swaths of the economy. So, in these circumstances, it's just natural that GDP estimation sometimes becomes controversial. I would like you to start by first helping us understand this at a very conceptual level, and then we can probably get into the latest GDP debates, the latest number of 7.8% that is now in the news, etc. But let's start with this first. Sure, absolutely.

Rajeswari Sengupta: So, exactly as you pointed out very rightly, Puja, that estimating GDP is an extremely complicated exercise, particularly for a country like India, again, exactly as you said, because there are vast parts of the Indian economy that are in the informal sector that we really can't measure very well. And there are many other problems. For example, the economy is constantly undergoing very rapid structural changes.

From manufacturing, we have become predominantly a services-orientated economy. Then measuring services output becomes extremely complicated. Then we don't have, for example, conceptually in any economy, GDP is measured in three different ways, right?

So, we have a production approach, we have an expenditure approach, and we have an income approach. In production approach, we're looking at the output that is produced by industry, agriculture, and services, the three main sectors. In this, I mean, that is the main method that is used in India for calculating GDP.

And we have some idea about agricultural output, because we know the crops that are being produced. When it comes to manufacturing, it has been a little bit of a hit and miss, because we shift data sources depending upon how much of informal versus formal manufacturing we are able to capture. And a big portion of manufacturing is, of course, in the informal sector that we always have struggled historically to get a good sense of.

And then services becomes even harder. For example, in the older times, we would have this database called Annual Survey of Industries, which is ASI data. ASI data would give us a very good picture about factory-level output that is being produced.

But as you can understand, that entirely applies for the industrial sector or the manufacturing sector. We can't get an idea of services sector output if we are using ASI data. Now, after reforms for the last 30, 35 years, when you see that economy is becoming so services-orientated, how do you capture the GDP produced in the services sector?

So that is when, about 10, 12 years later, when the last GDP revision was done, the Statistical Office decided to move towards a different database, which is called MCA data, Ministry of Corporate Affairs data. The reason being that gives us information on financial returns of companies. And when you talk about financial returns, as opposed to factory output, it doesn't matter whether the company is a manufacturing or a services, because every company will report financial returns in their annual report.

And that becomes the source of our calculation of gross value added. So that is a very big change that had to be done because the economy is structurally becoming more services-orientated, and ASI was not really equipped to handle that. And then from the production approach, which we are doing a relatively good job of, barring, of course, the problems of capturing the informal sector, which still remains a little bit of an issue, then comes the expenditure approach.

There you have to collect data on consumption of households, investment by private sector and government. Then you have to do government expenditure itself, exports, imports, all of that. In that, it's relatively easier to get exports and imports data, because as you can understand, this is also correlated with other countries' exports and imports.

It's a balance of payments framework, so you can cross-validate how much you are doing in terms of exports and imports. But then you come to more murkier territory with consumption expenditure. How do you calculate consumption done by millions of households of India?

Our survey methodology, given that we are a poor developing country with very weak state capacity, our survey methodology is not as sophisticated as it is in the advanced economies. Of course, we are trying with every passing decade to improve it. But there are many loopholes and gaps.

Incentives of survey, survey methodology is designed all of that. So consumption is entirely based on survey. So there are caveats to that.

Then there is investment data coming from the private sector, from the government. That, I believe, is slightly better measured. Then government expenditure also you get from different ministries.

So in the expenditure approach also, by and large, we are doing a decent job with the caveat that consumption expenditure is mostly calculated like a residual. Like you have total GDP, you subtract all the other components. And then the third approach, which is income approach, which is again something that is used in all developed countries, we don't even have that.

In every GDP release that you'll notice, we are only getting data on the production side and on the expenditure side, but we don't get any data on the income side because we do not measure wages and salaries or income in general in the Indian economy. And that is a very, very big gap because you essentially know how much people are spending, but you don't know how much people are earning. Now, spending could also be out of borrowed amount.

For example, now households are borrowing a lot, which is a big reason why credit growth is going up. Many people are confusing that to mean the economy is doing well because the credit growth is increasing. But if households are borrowing to consume, I would be worried about that because that's distressed borrowing.

That means you're not earning enough. But I can't cross-validate that because I don't have good data on income. So we don't have any GDP estimates coming from the income side, which is a big gaping hole.

Again, a classic sign of a developing country with not enough good data and not enough state capacity. So that is an overarching framework within which we should view the way the GDP measurement exercise is done. Within that, of course, one is, as I said, the economy is structurally changing very rapidly and the statistical machinery has to constantly keep pace of updating data and updating methodology to capture those changes and also to keep pace with international standards.

Because remember, when we only talk of Indian GDP, we sort of lose perspective of the fact that there are more than 100 countries of the world whose GDP are constantly being compared to each other. There has to be international standards that every country has to satisfy. India, this whole system of national accounts, SNA, in the 2011-12 revision that we did of the GDP data, there was a very active and conscious decision made by the statistical machinery to become even more consistent with the international standards.

That meant we had to bring about very big changes in our methodology and data sources. So in keeping times with the structural changes of the economy and also to be consistent with international standards, we have had to introduce many changes. So as it is, GDP estimation is complicated for a vast heterogeneous country like India, where you have a very big informal sector, more than 40% of GDP that you really can't measure very well.

On top of that, you have data inadequacies simply because it's weak state capacity, not enough surveys happening, not enough maybe extremely good quality of service is happening. And then on top of that, you have to constantly keep updating to keep pace with the changing structure of the economy and also to be consistent with international standards. So then I would say it's such a complicated exercise.

And of course, GDP gets a lot of attention because we don't have income data and unemployment data. If GDP has measurement problems, unemployment data, I wouldn't even know where to begin. So without good unemployment data, without good wages and income data, the only macro indicator that you have to assess the performance of the economy is GDP.

That is why every time the GDP data gets released, there is so much of debate and discussion happening. And to some extent, my sympathy goes out to the statistical machinery because they understand what they are trying to accomplish. It's actually pretty herculean.

But I think that's where my sympathy will stop and we can talk about that. But it's also kind of unfair that everybody then wants to opine about the GDP data. And I would go a little bit further to say not all of these people understand what is happening in the GDP data.

So it's very easy to catch the headline and say, oh, it's cooked and it's overestimated or whatever it is. But as I just explained, it is a very complex exercise. And most of the time, we don't have the right data and the methodology required.

Or we have to constantly keep changing it, which makes it very hard to get a sense of how exactly the economy is doing.

Puja Mehra: So to take an example, for instance, we are recording this podcast and I'm going to get paid for recording this podcast. At what level of estimation does this enter the GDP estimate or does it get captured at all?

Rajeswari Sengupta: Very good question. So, you know, when I said rapidly changing structure of the economy, think of how our own consumption basket has changed over the last 10-15 years. For example, we are doing this recording virtually using laptop.

10-15 years ago, the widespread usage of laptops in Indian households was unheard of, right? After this, maybe we will exchange some WhatsApp messages on a smartphone. 10-15 years ago, we did not have smartphones.

There was a point of time when we were using typewriters and fax machines to communicate with each other. Not so long back because we are a poor country. Typewriters and fax machines are completely gone out of consumption basket now.

This service itself that we are rendering, ideally this should get captured at the income level because you are getting some income out of it. But because your income is not getting captured in the income method, this does not really get captured. So whatever we are doing now sort of falls out of the GDP calculation.

So there are many changes that's happening and the statistical machinery is always going to be behind in trying to capture all of these things.

Puja Mehra: But that does tell me, Rajeswari, that there must be so much work like this happening which does not get captured in official GDP estimates. So probably our GDP is larger than what we are looking at in the estimates.

Rajeswari Sengupta: Yes, I wouldn't be surprised. You know, I mean, I can make arguments on both ends. There are many things that are still not getting captured.

I'll give you one other very pertinent example of the gig economy, right? Until recently, till COVID happened, we were not ordering out of Swiggy, Zomato, Blinkist of the world. But now we can't live in most urban areas or even rural areas, very urban areas.

We can't live without these services. But until recently, the gig economy was not really getting captured very well in the GDP data. But now they have to update it to figure out how this can get captured.

Or, for example, fintech. Fintech is such a big part of our urban life today. Where was fintech 10 years ago?

Or any of those technology innovations of edtech, healthtech, and you name it, that wasn't there earlier. So, which means earlier, just about seven, eight years ago, all of these things were not getting captured in GDP data. Now, I know MOSFET has made a conscious effort to try and figure out how this can get captured.

On the other hand, there are still things which probably are getting captured that are not relevant. Because, again, the new Bayesia revision will happen X number of years later, when many of these goods and services may be rendered redundant in the world of artificial intelligence and what have you may. So, there is a very fast pace of change that's happening.

So, to think that the GDP number alone is just that one single 86 or 88 lakh crore is encompassing everything in the economy, is a little bit of a stretch. So, we should really, and particularly quarterly data, mind you. I mean, I really get very amused when there is so much of hoo-ha-ha about quarterly data.

Because we all know, those of us who have studied national income accounts over the last 10, 12 years, we know that quarterly data really needs to be taken not with a pinch, but with a fistful of salt. Because at this level, the statistical machinery is working with very limited raw data. They're mostly working with very limited high frequency indicators and a very small sample of companies.

So, to think that that 88 lakh crore is basically what the economy is doing right now, is really not a very good way of thinking about the GDP growth. It is an indicator, definitely. It is suggestive evidence that this is the way the economy is moving.

There are some sectors doing well, some sectors not doing well. But the quarterly data is not nearly comprehensive. It is subject to revisions, because as more and more data come up, in a country like India, you first have, let's say, five data sources, based on which the statistical machinery will release some quarterly estimates.

Then with every passing month, every passing quarter, more and more data keeps coming up, because new surveys are happening, new high frequency indicators are coming up, and you keep updating the data. So, revisions are a very standardised part of GDP release, particularly in a country like India. It happens everywhere in the world.

And by the time we will actually know how Q1 was doing, or how 2025-2026 was doing, it's going to be two, two and a half years from now. So, to pin so much of detail on just that one number, is definitely a bit of a simplification that is happening. But yes, as I said, it is suggestive evidence, and we should debate and discuss it.

Also, because there is a very big change in method and data that has happened. So, that definitely calls for the debate.

Puja Mehra: Yeah, my next question is that, you know, you've explained what the revisions are, and that happens according to a pre-announced calendar, and they say what data will be plugged in for the initial estimate they have to be defined. But my next question is that, you know, what happens when they say they have revised a base series? A lot of people, their guess is that, you know, like you were explaining, that when they say they have updated methodology, what it means is that because the economy is changing, new goods and services are being used more and more, some are falling out of the consumption baskets.

And that is all that there is to it. But from what you're saying, my question about our podcast, for instance, it isn't just that there is a new service that has come about, but also that there isn't any data source to capture that service. It isn't just that, you know, you're having new categories, but how do you capture it?

What is the data with which you will capture it, right? So, isn't that what it is?

Rajeswari Sengupta: Absolutely. And because we are talking about this specific base year revision, I think to understand it better, it helps to talk a little bit about what happened in the last base year revision, just to give a context. So, when the Statistical Office revised the national income accounts and used the base year of 2011-12, and that data was released in 2015, what happened was there was a tremendous amount of controversy and criticisms that were being talked about by several economists and statisticians.

Because, A, you suddenly, because of internet, we all had access to many high-frequency indicators, right? And many private agencies started doing their own calculation and surveys to figure out how parts of the economy were doing. So, something that hadn't happened in the first 50 years of our statistical machinery started happening then, that all of us, we are not associated directly with the statistical machinery, would use all of those data sources to check what is called the smell test.

That is the number released by the Statistical Office matching the smell test. Never before was this thing discussed. For example, in the 2004-05 basis, nobody bothered.

And the smell test means, oh, you have 70 indicators, you have a GDP data, 40 of these don't match, 20 of these match, all kinds of debates and discussions started happening. And then it also became clear, because people started becoming more aware of, knowledgeable about the methods being used internationally, it also became clear that the Statistical Office was not following certain methods that should have been done earlier. Of course, as you know, in the recent debate, double deflation has become a very common phrase.

I mean, I'm just amazed that so many people even understand what double deflation is, because it is an extremely complicated concept. Ten years ago, if you asked anybody, do you know what is double deflation? They would just look at you physically and say, what are you even talking about?

But now we have many, many experts talking about double deflation. So it's kind of amusing. But coming back to my point, you know, I mean, suddenly people started pointing out that the Statistical Office is not doing double deflation.

So that was a controversy. Or it became clear that you're not capturing the informal sector correctly, right? You're using some proxies of the formal corporate sector and using that to extrapolate the informal sector, which became problematic because, as you know, after demonetisation, COVID, etc., the two sectors diverged significantly. So there is an element of overestimation, presumably, happening there. Or, for example, when they used the MCA-21 database, which is what I was alluding to earlier, it became clear that many of these companies were shell companies. They did not really exist.

And they existed on paper in order for tax filing purposes, etc. And there was, in fact, a survey done by the NSO, by the National Survey Organisation, in 2017-18 period, which showed very big number of companies being shell companies, which means if the Statistical Office is using that data to calculate value added, there is a very real chance of overstatement because you're counting the value added of companies that actually don't produce any. And that became a very big controversy, and the Statistical Office had to figure out how to address that.

So what I'm trying to say is that for the first time since the 2011-12 base year GDP series was released, many controversies and criticisms started getting talked about extensively. And this was also the time when the way they were deflating the GDP, and we can talk about it later. We should talk about it because that is so important now.

The way they were deflating the GDP to go from nominal to real started becoming a problem. Unfortunately, what happened was that the Statistical Office did not respond in a satisfactory manner. In fact, they became more defensive.

And unfortunately, surveys started getting discontinued. The base year revision that should have happened five years after 2011-12 did not happen. And that further entrenched the suspicion in people's minds that maybe something is indeed wrong with the data.

Now, base year revision, slightly conceptual part, it is a very standard best practise all over in the world because every few years you have to do a base year revision of the GDP data because your price indices, your inflation series is changing, your prices are changing. So you have to account for that, because ultimately what you want to know is the physical output that is being produced in the economy taking away the effect of crisis. But because there is a constant inflation happening in the economy, you need to be able to keep revising what is the base year with respect to which you will calculate that physical output.

The base year means you're holding prices constant in that year and you take out the effect of the crisis. And then going forward, you're just looking at what is the physical output being produced in the economy. So it is imperative that you do the base year revision every five years.

This is what happens in every country of the world. But in India, over and above this price effect, we have these structural changes. So new data is coming up.

Whatever new data you can access, whatever new surveys you can initiate, whatever new methodology you can introduce, like the double declaration, etc., you have to keep doing that also. And because they didn't do all of these revisions in the standard five-year period, essentially they let it slide. And the controversy and the criticisms became stronger because, remember, the economy is going through so many shocks, demon, COVID, GST, etc., and the suspicion is getting stronger. So then suddenly, after almost 12 years, you're releasing a base year revision, which is far overdue, and you're introducing so many changes, underlying data, underlying methodology, underlying new surveys. Of course, everybody will want to understand, OK, how have you corrected all the controversies and criticisms of the last series? And how has that impacted the nominal and the real growth rates now?

So it's a lot going on.

Puja Mehra: So I'll come to what all has been changed in this latest methodological changes that they have introduced. I'll come to that. Before I come to that, I want to ask you, because we've talked about it, listeners must be wondering, what does deflation mean and what is double deflation?

Absolutely.

Rajeswari Sengupta: So when a statistical office collects data on, let's say, value added, value added is output minus input. And simplest way to think about it is just think of profit for a company. That's easier to visualise for us.

So profit for a company is whatever you're producing minus whatever you're using as an input. Now, earlier in the 2011-12 base year series, until that point, until recently, the statistical office used some method called a single deflation. What is that?

When you're collecting nominal data on rupee terms in the value added, let's say profit of 100 rupees or value added of 100 rupees, how do you go from there, from nominal to real? Because real, as I was mentioning, you strip out the effect of prices and you only want to focus on what is the actual output being produced by the company. And to do that, they were calculating the nominal value added in rupees term.

Then they were dividing it by a single price index. And that's how they were taking out the price effect and they were getting the real value added, which is the actual output being produced. And what they were doing is they were using predominantly the wholesale price index as that deflator, as the single deflator.

The single deflation is what they were doing. Now, the problem is all over the world, in most OECD countries and most G20 countries, that is not the standard practise. The standard practise is to do something called a double deflation.

What is a double deflation? Instead of dividing nominal value added or nominal profit by a single price index, what you do is you deflate twice. You deflate the input value first, then you deflate the output value.

So you get that two different deflated numbers and then you subtract to get the real value added. Why is this important? When input and output prices are moving together, then single deflation is fine.

WPI and CPI are basically the same. You're not really doing much of a problem. But if, let's say, wholesale price index falls or input prices fall massively, but the output price, which is, let's say, CPI, which is retail price, hasn't fallen by that much, what you're getting is, think of the profit of the company, you're getting an artificial boost in the profit because you're just using cheaper inputs.

But your output price hasn't changed much, so your profits will artificially go up. Your nominal value added will artificially go up. If you don't adjust for that, when you go from there to real using just WPI, your WPI will go up.

Think of the ratio as nominal value added divided by WPI. As WPI goes down, sorry, if the input prices are falling, the overall ratio will go up. So your real value added or real profit artificially gets overstated.

That's a problem because that means your GDP, which is basically go from value added to GDP, that is getting overstated. So that is sort of what was happening in the old 2011-12 base year series. Ever since they released the data in 2015, input prices collapsed.

We did not have any oil shock. WPI was almost in deflation territory. WPI inflation was negative.

And CPI was still around 3%, which means you've got this sudden wedge between input and output prices. You've got an artificial bump up of nominal value added. And because they were only doing single deflation using WPI, they were overestimating real GDP growth rate.

That was a big criticism that was being discussed. So in order to correct for that, the international best practise is, as I was saying, do double deflation. You deflate both.

You deflate both input and output to take away this increase that is coming from fall of input prices. And then what the real value added that you get is adjusted for all of that price effect. That's the correct way of doing it all over the world.

And the statistical office claims that that's what they have done now, which is one of the biggest changes that happened.

Puja Mehra: So I read an interview with Dr Pranab Seng where he said that India has been trying to do this double deflation for many, many years, but we couldn't because we do not have so many price indices to be able to deflate the various inputs. And I think the ministry has not as yet released the information required for experts like you to take a look at how they have done this double deflation. But what are your thoughts on this?

Have they done a good job? Do you think they have all the information they need to be able to do it now when they couldn't do it for so many years?

Rajeswari Sengupta: Yeah. So this is where we get into the current controversy, so to speak, that is brewing. So the statistical office claims that they have done the double deflation.

To do double deflation, you need output prices, which is relatively straightforward. We can use consumer price index as the output price deflator, output value deflator, but you need input value deflator. Now, there are hundreds of thousands of inputs being used in the economy as we speak.

In order to be able to correct double deflation, you need prices on all of those inputs, or at least most of those inputs, right? Which means you need a very comprehensive input price index. Like you have a CPI for an output price index, you need a very comprehensive input price index.

I remember in February 2026 when they first released the new base year data, they said that they are using WPI as an input price index, granular level WPI data for different sectors as the input price index. Then in the current 31st whatever August data that they released, they said that they are using producer price index as an input value deflator. And that's where people likely start getting nervous, and I completely agree with Dr. Pronab Sen, WPI, we know what the WPI data looks like. We have had WPI data ever since we have had GDP data, and we know the subsectors, we know the granular level. Now you're claiming that you're discontinuing the WPI data, you've moved to PPI data for input price deflation. Problem is, for the last 20 years, the statistical office in India has been trying to construct an input level producer price index.

They know more than any of us that double deflation is the holy grail that they should move towards, and they have been trying to do it. The reason they were not able to do it is because they did not have an input price level producer price index because producers do not give them data on input prices. And producers have different input prices for different distributors.

So they have to give an average price index and all of that.

Puja Mehra: I recall that when the last controversy had happened, and I had interviewed Professor TC Anand, who was the then chief statistician and was my professor of statistics in college. And he had said that, I think it was the Rangarajan committee that had looked at statistics reforms required, and they had recommended that there should be a law. By law, firms should be required to give payroll data, producer pricing data, all kinds of data that is needed for these purposes, not for taxation or ED or this or that, but for these purposes.

And there was so much resistance, and India is one of the few countries, in fact, that does not require firms to share this data by law. And therefore, we are not able to do all these exercises required for GDP estimation.

Rajeswari Sengupta: Absolutely correct, Puja. So as I said, they have been struggling, the statistical machinery has been struggling for years to try and get a producer price index, which is something that most developed countries of the world have. Because without a good producer price index, you cannot do double diffusion.

Because wholesale price index is actually not a very, it's like a very rough proxy of input prices, but it is a very limited set of commodities like steel, cement, et cetera, oil, that you're using at inputs. There are many, many other inputs that you're using that WPI does not capture. They did the best they could using WPI.

Of course, they were doing, that's the reason they were doing single deflation, because they didn't have data for all the inputs. So now when the statistical office says that they are doing double deflation using producer price index, of course, some people who understand this really well will get nervous, because for 20 years, you did not have a producer price index. There is no law as exactly was recommended by the Raghuram Rajan Committee.

There is no compulsion for the Indian producers to provide data on input prices. How are you then coming up with this input price index? Now, I'm going to add another issue to that.

What the press release of the ministry says is that they have an output producer price index, they have an input producer price index, and they have a services producer price index. Apparently, there are three PPIs, out of which the input PPI is in a trial stage. I don't understand what they mean by input PPI being on a trial stage if they're saying that they're using producer price index to do double deflation and releasing the data officially.

Maybe I have missed something, and I'll talk about that later, why I feel I may have missed something. And the services PPI, they only have captured seven services sector, most of which are banking, financial services, insurance. Then there's railway freight traffic and airport passengers, and that's about it.

Services is much, much bigger than that. If you're just able to capture prices from seven services sectors, why release it at all? Why even talk about it?

And I don't know what is this output PPI. I was under the impression that CPI is being used for deflating the output value. If they're using output PPI to deflate the output, I will be doubly nervous, over and above being nervous about what is the input deflator that they've used.

So bottom line, yes, they're saying they're doing double deflation, but their own claim in the press releases have changed between February 2026 and August 2026 when they've gone from WPI to PPI. In the PPI domain, we do not know how they have gotten the producer price index. They are saying they're using granular level producer price index, which is even more worrying because I don't see that granular level producer price index data anywhere in the public domain.

And if you are doing three different PPI series, the first thing they should have done before releasing GDP data is release the producer price index data. That to me is the biggest missing point here that if you're going to switch to something so complicated as double deflation, and if you're going to be using a completely new price index, where is the data on that price index for people like me or Dr. Sen or other experts to look at the data and understand just there is veracity and validity in this price index that can be used as a deflator. Without all of that, you're basically told us, oh, you have to trust us because we are using some PPI and we are using double deflation with those PPI.

I don't understand how you're doing double deflation. I don't understand what pricing this is using for double deflation.

Puja Mehra: How are you going to trust them when they're saying that they have data for only seven services in a services-orientated economy? Most of our GDP comes out of services. Absolutely.

Rajeswari Sengupta: So two big missing points. One is, of course, as I said, not releasing the producer price index data. And the other, even more important, is not releasing the sources and methods document.

I mean, in the run-up to the GDP release, they did a lot of stakeholder consultations, lots of conferences all over the country to get a buy-in from the data users, a lot of FAQs being released, and CREST releases being released. So clearly, the information is there scattered across multiple documents. Why not put everything together and release a very detailed, comprehensive sources and methods document, which is even more important than ever before because you've waited for 12 years to do a base year revision and you've brought in fundamental changes in the methodology as well as in the data.

And I'll talk about what changes they have brought about other than this. But without that document, I don't understand how you've done double deflation for which sectors, what price indices have you actually used, and where is the data on those price indices. And the reason I say how I don't understand also how double deflation is done is because when I was attending these stakeholder conferences, Puja, they said something like, for manufacturing sector, they are doing double deflation, the way you and I talked about.

And then for some services sector, they did something called a volume extrapolation. Now, in the latest press release that came out, or in the entire conversation that is happening, there is no discussion about volume extrapolation or how they have done it for the services sector. So are you doing double deflation for the services sector, single deflation, volume extrapolation?

I don't know. So how am I supposed to evaluate more than 50% of the GDP calculation, which is services sector, and you have services PPI only for seven services, as you rightly mentioned. So my point is, I can't evaluate whether they have done a good job of it, despite all the noble efforts, because I don't know what they have done, as simple as that.

Exactly why Dr. Pranav Sen is also wondering, I am wondering, several others are also wondering, what is it that you've actually done? Just make it clear in a document and come out clean and transparent with it. And then we can sit and debate whether it's right or wrong.

I don't even think we have the correct information required to evaluate that.

Puja Mehra: So what do you make then of the latest quarterly GDP estimate of 7.8% which is being discussed so much?

Rajeswari Sengupta: Because of all the reasons that I mentioned, I don't believe the number. I don't know where that number is coming from. Maybe it is a correct number.

Maybe they have actually done a fantastic job. They are just not releasing the information and maybe it is actually 7.8%. Maybe. But I don't believe it because I don't know what they have done.

I need to see a detailed sources and methods document to understand how they have done the methodological changes, what data have they used for which subsector, what method has been used for which subsector. Very detailed information is needed. And secondly, I know that other than double deflation and using producer price index, they have done many other changes.

For example, we were talking about informal sector not being measured correctly in the previous data revision. So now they have a survey called ASUS survey, annual survey of unincorporated sector enterprises which is happening on annual frequency, which is great. Kudos to the ministry for that.

And apparently now they are also doing quarterly survey. Now, when I saw that they are doing quarterly survey, I got very excited thinking, oh, they must have used the quarterly data from the survey to do the quarterly GDP estimation. But no, that's not what they seem to have done.

They seem to be still using high frequency indicators and benchmarks like GST data for doing informal sector estimation for the quarterly GDP and not the results of the ASUS survey. Again, I don't know why and I don't know when the quarterly data from the ASUS survey will be used for the quarterly GDP calculation. So again, this is the second thing other than double deflation.

That's my worry. And the third is, remember what we said that MCA21 data has problems. There are shell companies.

The ministry said that they did a sweeping effort, the Ministry of Corporate Affairs, sweeping effort sometime in 2019 or so to weed out the shell companies, which is great kudos to them. Now, we are in 2026. I don't know if more shell companies have crept into the data between then and now.

I don't know how the statistical office is adjusting for those shell companies. How many shell companies are there if at all? Is it zero?

Is it a few thousand? Is it a few lakh? I don't know.

No discussion has happened on the sample frame of the MCA21 data at all because everything has gotten focused on the deflation point. This is a very important point. MCA21 is the main basis.

Of course, in the quarterly data, it is mostly results of listed companies that we focus on. But still at some point, annual data will come out and we need to know what is your data frame. I remember the stakeholder conferences, we were told that GST data is being used to validate the MCA21 data frame.

But we also know that the ministry had access to GST data of only one year or two years. So what happens going forward? If there are new shell companies coming into the sample, you have GST data of only two years, how do you validate the sample frame to give us the confidence that it is a legitimate data set, that we are not blowing up shell companies' GDP data?

We don't know that yet. There has been no discussion about that. All of that is happening.

Why? Because we don't have the sources and methods documents. So three changes.

Deflation, I don't know what's going on. Informal sector, I don't know what's going on in the quarterly data. MCA21 sample, I don't know what's going to happen when the annual data comes out.

So all the changes that they've made, they've not explained well. There are FAQs spread over the website. There are press releases spread over the website.

If you pore over all of that and try to draw a picture, you will get even more confused, which is where I stand today. Because as I said, I thought they were doing WPI for deflation. Now I know they are doing PPI.

I don't know what that PPI is. I thought they were using quarterly ASUS for quarterly GVA, but apparently they are not. So I get more confused because I'm trying to piece information and try to do 2 plus 2, 4 from their documents, when all they could have done is just release one document so that we could just study it.

And it makes me nervous that why are you not releasing it? Why did you shift the publication date from August 2026 to September 2026? Are you not ready with the document?

That could make me even more nervous. How did you do all of these fitting changes without there being an underlying document for it?

Puja Mehra: And in the question of GST, again, I go back to that conversation I had with Professor TC Anand, where he told me that GST data was very difficult to use for GDP estimation, although ideally he would like to use it, he said. But he said because data is not forthcoming. Not only that, because companies have not given permission for it to be used even after it is anonymised.

There is no framework, legal framework for government to be able to do that. But two, also because they did not have it in the formats with which, I think listeners may not know that there is a coding. You may want to explain the whole NIC code.

I think the GDP NIC code does not match the GST code is what I understood from ID quality he had told me.

Rajeswari Sengupta: Yeah, so first of all, there is a problem because GST data is collected by let's say Department A. GDP estimation is done by Department B. And there is nothing connecting the two departments legally to say that you have to pass on the GST data from Department A to Department B, right?

It is a favour if they give it, but they can also turn around and say for a particular quarter, we will not give it. So it cannot be taken as a consistent source of raw data for GDP estimation going forward. That is a big concern.

So even if you somehow got access to some GST data for this quarter, doesn't mean that you'll have access to it for all the quarters going forward. So why use something that is not inbuilt into the GDP data sources? The second thing is exactly as you said, the NIC code essentially think of it as an identifier of a particular enterprise, right?

You have one identifier that Department A is using for collecting GST data. You have another identifier that Department B is using for collecting data on returns, value added, et cetera, from enterprises. The two identifiers don't match.

So how do you connect that taxes being paid by one particular enterprise into the value added by that particular enterprise when you can't match the identifiers? So that is definitely a concern. But now it seems that GST data will be used for informal sector value added calculation on a quarterly basis.

We just need to get more information about how this is happening.

Puja Mehra: So like, for example, if let's say there is a tailor shop down the road and they pay GST, when you pay GST, you have to fill in a code, an HSN code, which identifies what industry you are because that corresponds to a certain tax rate. So that's why you have code. So you pay your tax and that gives you, you know, you get identified under that code.

Now for your value addition to go into GDP estimation in that particular industry, that code has to match and two codes don't correspond to each other. So you don't know how to take the value added by this particular tailor enterprise into the GDP estimation. The two series don't talk to each other basically.

Rajeswari Sengupta: Exactly. Now it is possible that if the statistical office is saying that they have used and they have actually written that in the press release, that they have used GST data as one of the high frequency indicators for calculating the quarterly GDP for this quarter, maybe they have figured out a way of doing this concordance. Now, the reason I'm saying maybe is again I go back to my favourite question so that I don't know.

Unless you explain to me how you have used the GST data, which is again a new data to use, right? Every new data that you have used and every new methodology that you have used, it would have been best practise and decent practise, in fact, to just be very clean and transparent about the how. And because they have not, I see that they've used GST data.

Maybe there is a way that I don't understand, but I need to see that and you have not explained to me how. In the last 10 years of GDP data controversy, in no other country GDP data creates the kind of controversy that probably does in India, which is very fascinating. And the reason, one of the big reasons is because lack of communication and transparency from the statistical office.

Singular factor, right? Yes, our understanding is bound to be limited because we are not a part of the statistical machinery. We are not sitting and doing the GDP estimation, which is a very complicated task.

So help us understand better because we are the data users. If you're expecting us to trust your data and to have faith in the credibility of your data, you need to give us information so that we can understand what is it that you're doing. It's as simple as that.

This doesn't have to get political. It doesn't have to get sentimental. It doesn't have to get anything.

It's just pure statistical exercise. So if you don't help us understand what you are doing, of course, we are going to be suspicious because A, this is all smell test question, right? Do I really feel the economy is booming at 8% around us?

I don't think so. And then B, the more reticent you are to release documents and information, the more I start getting suspicious and start saying strange things like base year revision and whatever, whatever. So yeah, I think the onus is on them to explain better, as simple as that.

Puja Mehra: No, and they tend to be slightly, you know, whenever I, in my reporting career, have interacted with people who work in the NSO, they tend to be very restrained in what they say. And what I have gathered from conversations with them over the years is that, like you said, the smell test, the pressure on them and therefore the incentive for them used to be earlier to somehow produce estimates, you know, that would conform to the smell test in advance, anticipating what the smell test would be applied once they release, you know, the estimates. Now, increasingly, I suspect the incentive is to do much better than what the general consensus expectation is.

That's just how the political narrative is. Unfortunately, something so technical and complex and so dependent on methodology, on data sources, on understanding of the economy has somehow got to be influenced by what the political narrative is going to be, one way or the other. Earlier, you wanted to conform to what the smell test would be applied and now you want to do literally better than that because you always want to look much better than what people thought the economy is.

So, this brings me to the question of, I'm not saying that GDP estimates are fudged. I have far too much professional respect for people who do GDP estimation in the National Statistics Office to ever, ever say or suspect something like that. But, the question of influence and on the margin, when you have a call between whether to do this or to do that, both decisions will be completely professional.

Do you think that pressure sometimes leads to calls getting taken because of the whole environment that we are in?

Rajeswari Sengupta: It's actually a very tricky question. So, first of all, let me say that I also have utmost respect for the very, very complicated task that they are trying to achieve because being consistent with international standards in a country like India where we don't have half of the data sets required is a really, really difficult task to do. And they have, to be very fair, they have taken a lot of effort to address all the criticisms that were being discussed over the last 10 years.

They could have just easily ignored all of that too and we couldn't have done anything about it. But they have taken a lot of effort, they have engaged with stakeholders, they have released as much information within their constraints, I'm guessing, that they can. Now, yes, I think there is going to be a subtle pressure on them at all points of time, but you suddenly cannot show that the economy is growing at 6%.

I'm not saying the economy is growing at 6%. I'm just saying that there is a general narrative that the Indian economy is doing spectacularly well. And that narrative has continued since 2011-12 or since 20-15 years that the Indian economy is growing at 7-8% in general.

So to suddenly deviate from that and show that the Indian economy is not growing at 7-8% will probably make their lives much harder. And that's why it's possible that there is some kind of general incentive to show that the economy is doing well. I'm not saying that they are trying to show it even when the data doesn't.

But I'm saying that the reticence to be so transparent and the reticence to engage more... For example, I'll give you one instance. Like the recent controversy when the ex-Finance Secretary said whatever he said, and that is basically what triggered all of the debate over the last 3-4 days.

I thought that they should have just ignored it. You know, I mean, this is not a very well-thought-out, excellent, solid criticism. Not at all.

Those of us who understand the methodology even a little bit know that that point was not at all meaningful. So then the experts, the real experts who are doing this work, they know even more that how pointless that criticism is. So then why do you add credence to it by trying to refute it and saying that that is not correct?

Because then I feel you're triggering suspicion that all you said is that that is not correct. But now after that, you're defending using high-frequency indicators and all of that, which seems a little bit wishy-washy. I would have just expected them to completely ignore it and let it die down and just stand by the fantastic work that you would have done and just release more information for the experts, the economists, and statisticians to take a look at and validate the claims that, yes, the economy is growing at 7.8%. That would have given me a lot of confidence. But instead of that, this whole debate that has happened kind of makes me question that is really something strange going on here?

Puja Mehra: No, in fact, it's not even a debate, Rajeshree. I think it's just so much noise. Like you're saying, the ministry has unfortunately added to the noise instead of either letting it die, let the inconsequential criticism die and then just respond to questions like you're raising about the methodology coming out of their own lack of transparency.

But also, to your point, they did release, I think it was what, 4% GDP growth in 2019, just before the Lok Sabha elections that year. So they do under pressure do release estimates where growth is not looking very good. So all the critics of GDP estimation should be mindful of that as well.

Rajeswari Sengupta: Or even, for example, now when they release this data, they did show that the earlier base year had overestimated growth rate. So they did bring down the nominal data from 86 lakh crore to 80 lakh crore. They are admitting that at that point of time, the nominal GDP was only 80 lakh crore and that can actually wreak havoc for the government's fiscal calculations.

Because remember, all fiscal calculations are happening on the basis of nominal GDP data. So suddenly, when the statistical office says no, your nominal GDP is so much lower than the fiscal calculations go haywire. So they did do that.

It's because the data and the methodology showed that that is what the nominal GDP was. And so yes, from time to time, whenever the data has shown, they have revealed it. But I guess the problem is mainly because they have been criticised so much that they have gotten a little bit more defensive.

And now, even with all the efforts that they have taken to improve the methods and the data, their response is still not, I feel, what it should have been, which is just hold not responding to any of these criticisms which don't make any sense and just release more information and data for people like us to validate the GDP data. That would have been the best thing. And also, I like what Dr. Pranab Sen also mentioned, that release the old and the new series simultaneously for the next few quarters and let us see in parallel what is happening in the old data and what is happening in the new data. These are the technical ways of responding to criticism that they should have done or they should do, in fact, there is still time. Or, for example, release the back series. You know, releasing a back series is so much more important so that we know how did the economy perform in the new series, how did the economy perform in the old series.

We can see for ourselves how much is the divergence. These are the technical details that need to be done to gain the credibility instead of just getting into this your word versus my word kind of an exchange which really doesn't help. There are some problems with the numbers that do make me a little bit worried, which is why I said I don't believe in the number.

Puja Mehra: But I think, you know, here what you're saying, there was also a role for the government's economists, the economists that sit in the government and the Prime Minister's Economic Advisory Council in the Finance Ministry to help clear the air like you are by raising questions. You're not raising doubts, you're raising questions. Instead, I saw social media posts from them trying to beat it to any criticism.

I mean, there was one post which was factually incorrect where a renowned economist has said that like you're saying, you know, the methodology has not been released, the data has not been released, the ministry should very quickly release it. And this economist, government economist said this is a lie, they have already released the data. And, you know, to attack journalists asking these questions, I just thought that was so unnecessary and I don't know how it helps anybody.

Rajeswari Sengupta: Exactly. This is where, Puja, I told you, I find it fascinating that India is one country where the release of GDP data incites so much of discussion and debate. You know, it's in every other country the number gets released, we look at it, we move on.

It just becomes politicised. Exactly. It becomes politicised because it gets polarised.

Because suddenly the camps get formed. Some people are saying it's manipulated, some people are saying no, the economy is doing great. And in that noise, the actual substance gets completely lost.

So the real questions that need to be asked and the real issues that need to be addressed don't get talked about because everybody is now trying to curry favours with whatever powers that be that they want to to either defend or attack. And I just don't find it useful at all because it is the most important macroeconomic indicator that tells us about the health of the Indian economy. We need to understand it better.

We need to understand the technicalities are better. The ministry and the statistical office need to help us to do that better. This is not political.

This is not sentimental. It's just technical statistical data. But for example, I'll give you one quick point on this.

You know, we say that the economy is growing at 7.8% in real terms, which means it's booming in a quarter when India was subjected to the most severe energy shock. Right? We are a net importer of oil.

And at a time when oil and LPG and LNG were in short supply, irrespective of what the government did to tide over the shortages, it still remains that it was a huge terms of trade shock. What I mean by that was import prices went up significantly. And the NIA, the National Income Accounts data itself shows that the value of imports in India in that particular quarter went up by 31%.

But the actual volume of import, which is in constant prices you strip out the price effect, fell by 1%. That means that the total import prices in one quarter increased by 32%. That is a huge terms of trade shock for a country like India.

Import prices increasing by 32%. Now, in my mind, in my limited knowledge, if import prices are increasing by 32%, somebody has to pay for it. So in a country where in a quarter import prices have gone up by so much and there is such a big terms of trade shock, how is it that your investment spending is increasing by 12%?

Your consumption is increasing by 7%. Your government expenditure is also increasing. The spending is all increasing when you have such a big terms of trade shock.

So who's paying for that? I don't get this. This is, for me, it's pure arithmetic that you have a terms of trade shock which all economists will understand is a real income shock.

When you have such a big real income shock, how do you have a spending boom? I don't get it at all. This is, for me, when I looked at the numbers, that's one thing that stood out to me that this itself is not making sense.

I'm not even going into, you know, GST cuts did some auto sales increase or, you know, services, exports are growing in real terms in 12%. Fine, all of that is great. Exports are growing because exchange rate depreciated.

There was some positive impact that came out of it. GST cut was very, very small in magnitude to compensate for a terms of trade shock of 32% increase in import prices. Our government capex, states are not spending.

It's only central government. That too not as high as they were doing in the immediate post-COVID period. So what explains the spending boom happening in the economy in a quarter when you have such a big energy shock that manifests in the form of an unfavourable terms of trade?

So far, I have not seen anybody in the public discourse who has been able to give me a satisfactory answer to this. Neither the government economists, nor the other. Nobody, basically.

Because all we are talking about is is this base year revision of whatever old series, new series, 2.5%, whatever it is.

Puja Mehra: That's really not the debate. Nobody has asked this question. You're the only one who's pointed this out, Rajeswari.

You were saving it for the end. Maybe I should have asked you the question sooner than this. Because it's a puzzle.

It is so important. But thank you so much for this conversation. It has been so illuminating. Thank you so much

Rajeswari Sengupta: I really hope it helps.