Why the IMF doesn’t buy India’s GDP data

The International Monetary Fund’s (IMF) unflattering report on India’s national accounts has once again focused attention on the dubious nature of the country’s macroeconomic data. In its 2025 ‘Article IV Consultation Report on India’, announced on 26 November, the Fund has accorded a ‘C’ rating for the data used in India’s national accounts because ‘the data provided [by India] to the Fund have some shortcomings that somewhat hamper surveillance’.

While the IMF is mandated to accept government data on GDP, it is saying the data is not reliable. In plain terms, a ‘C’ grade implies that India’s official data is not up to the mark — in other words, the IMF would have us take the latest GDP growth figure (8.2 per cent for Q2 of FY26) with a pinch of salt.

In fact, there are indicators that suggest growth cannot be that high. For instance, reports of investment projects being withdrawn or curtailed, and of net FDI turning negative. These are not the signs of an economy that is growing rapidly.

Among the shortcomings in data the IMF has flagged are the use of an outdated base year (2011–12); sizeable discrepancies in GDP data, possibly due to the lack of informal sector data; weak statistical techniques used in the quarterly compilation of national accounts; and the lack of consolidated data on states and local bodies after 2019.

These points have been raised by several analysts since the demonetisation of November 2016. The economy also experienced shocks due to the faulty implementation of GST in 2017, the NBFC crisis in 2018 and the pandemic in 2020. Each of these crises aggravated the issues flagged by the IMF.

Analysts have, in fact, raised a deeper question about the GDP in the new series (base year: 2011–12) that the IMF does not touch upon. The new series was announced in 2015 during the NDA years even though work on it had started during UPA II.

A government committee was asked to rework the series, but its work was rejected when it showed higher growth during the UPA years than in the NDA period. The Niti Aayog was now asked to produce the new series, even though it wasn’t qualified for the job. It did produce a series nevertheless — which showed higher growth during the NDA period, fulfilling the political ask, and was duly accepted.

But former Chief Economic Advisor (CEA) to the government Arvind Subramanian showed through econometric modelling that the GDP was being overstated by 2.5 percentage points or more.

Next, out of 18 lakh companies in the MCA21 database, three lakh were removed as being ‘shell companies’. As this author had pointed out then, this should have impacted the GDP estimate — as shell companies are typically used for under- and over-invoicing to divert income from regular companies — but this did not happen.

Also, 35 per cent of companies couldn’t be found at their given addresses—so these were likely fake companies putting out fake data. All these discrepancies put a question mark on the GDP data. Finally, GDP calculations cannot account for the black economy — a whole different can of worms. Bottomline: the new GDP series (base year: 2011–12) has been manipulated and is seriously flawed.

More data manipulation

The current government has been systematically rejecting or withholding adverse data. For instance, the 2017–18 consumer survey was not released.

Unemployment data was withheld before the 2019 elections, because it showed that joblessness was at a 45-year high. The multi-dimensional deprivation data is being manipulated to show lower poverty.

A comparison is made between 2015–16 and 2019–21. This contains the pandemic year 2020, when people’s income fell and education and health deteriorated. So, how could deprivation (and poverty) have decreased?

Our data is suspect also because samples are drawn on the basis of the outdated 2011 Census. India skipped the Census exercise in 2021; it’s now supposed to be held in 2026, while it could have been done in 2022 or 2023 as many other countries did. It matters because there have been demographic changes since the last Census, so samples drawn from the 2011 Census will not yield correct estimates and the conclusions will not be reliable either. This further strengthens the IMF’s point.

Even the Consumer Price Index (CPI) is based on 2011–12 data. The distribution of income has since changed, which impacts consumption patterns. Newer goods and services have become available, which must be included in the analysis but can’t be if we continue with 2011-12 base data.

The IMF report also flags discrepancies between the production and expenditure approaches to measuring GDP.

Definitionally, there shouldn’t be a difference between these two estimates, but as pointed out by this author earlier, both estimates have errors for lack of independent data for the unorganised sector. This impacts the two estimates differently, so the discrepancy too changes from year to year.

When there are shocks to the economy, the divergence increases, as has happened in India since demonetisation. Not only has a small variation become much larger, it is also swinging wildly from positive to negative and back, thereby indicating unreliability of data.

In quarterly estimates of GDP, the lack of current data necessitates the use of proxies. But the growing organised sector cannot act as a proxy for the declining unorganised sector. That is why there is overestimation of GDP. In fact, higher the growth rate of the organised sector, higher the mis-estimation of the unorganised sector.

Extrapolations are common in the compilation of annual GVA (gross value added) series. But when this is done for a shock year, there is overestimation — the actual decline in economic growth is not captured. No wonder official data for the demonetisation year (2016–17) showed a high growth of 8 per cent when the economy had, in fact, contracted.

When the use of proxies leads to overestimates in the contribution of the unorganised sector, it inflates estimates of the production of consumption goods and services, thereby leading to overestimation of consumption in the economy.

It is argued that since data is not available for the unorganised sector, some assumptions must be made for estimation. That is true, but conditionally correct assumptions become invalid when the economy experiences shocks like demonetisation. So, the assumptions must change, but this has not been done, leading to wide gaps between reality and official data.

The IMF presumably felt compelled to finally acknowledge the unreliability of India’s GDP data. Unfortunately, what the cheerleaders of this government do not see is that you need reliable data for sound policymaking — that is, if your government is in the business of good governance.

Arun Kumar is retired professor of economics, JNU, and author of Indian Economy’s Greatest Crisis: Impact of the Coronavirus and the Road Ahead