How GDP data misread the economy, complicated policy

At various periods the data signalled strength when the economy was weak, and in others, it suggested that policy should be eased when growth was strong

Recently, the government released a new series of GDP numbers. They did so for two main reasons. First, the weights of the various goods and services needed updating, a step that was long overdue as India’s economy has gone through tremendous changes since the weights were last established in 2011-12. Second, shortcomings in the methodology for estimating GDP needed to be addressed.

The process of changing the methodology has been commendably consultative, for which the Ministry of Statistics and Policy Implementation (MoSPI) deserves credit. But assessing whether the shortcomings have been substantively addressed will take time. And it will ultimately depend on the quality and plausibility of the new numbers.

Prima facie, while there have clearly been improvements, some oddities remain. For example, goods inflation (as measured by the manufacturing GVA deflator) is strangely negative for 10 quarters, which has led to an unusual jump in the sector’s real growth rate. Also, import deflators are radically different from those in the previous series, which has produced sharp changes in real import growth.

But before we can fully assess the new methodology, we need to understand: What were the problems with the old methodology? Were they serious? In a new paper, we try to answer these questions.

Start with the manifestation of the problem. Until a decade ago, there was a close relationship between the GDP numbers and the standard macro-indicators that economists monitor to gauge the health of the economy: Exports, credit, taxes, electricity consumption, sales and the index of industrial production. These relationships broke down, however, after the methodology was radically revised in January 2015.

A key problem was the use of inappropriate data sources, notably proxying the performance of the informal sector by using data from the formal sector. This may have been a reasonable approach in normal times. But after 2015 the informal sector was hit disproportionately by three severe shocks: Demonetisation, the introduction of the GST and Covid. As a result, the methodology overstated informal sector performance — a serious problem, since informal enterprises accounted for over 45 per cent of the economy in 2011-12.

The other major problem was the use of inappropriate deflators. Much attention has been paid to the lack of “double deflation”. But the bigger issue was the use of inappropriate price indices, most notably the use of the wholesale price index (WPI), to deflate services production even though the WPI doesn’t really capture service prices. This was a particularly acute problem over the past decade when the WPI plunged, driven largely by the collapse in the price of oil. As a result, inflation was understated and real growth was overstated.

These problems mattered enormously, causing significant misestimation when the 2015 methodology was applied to GDP growth numbers, initially from 2011-12 onwards and then retrospectively to figures going back to 2004-05.

Growth was overstated by about 1½-2 percentage points on average between 2011-12 and 2023-24, with inappropriate data and deflators contributing roughly equally. As a result, we estimate the economy grew at 4-4 ½ per cent on average instead of 6 per cent over these 12 years. Estimates of India’s potential growth need to be re-calibrated accordingly.

For the earlier period, the 2015 methodology had the opposite effect: It underestimated growth by about 1-1½ percentage points on average between 2004-05 and 2011-12.

The net result is a mischaracterisation of the 20-year trajectory of growth. The boom of the aughties and the subsequent slowdown have been erased from history, replaced by a picture of steady, rapid growth.

The figures show the problem. The macro-indicators surged from the 1990s to the early aughties but then decelerated sharply, falling even below pre-2005 levels (see arrows). Yet official GDP growth figures (the shaded area) portray a picture of stable growth over the entire period.

In this picture, there was no India Shining between 2003 and 2010, when the heady, dizzy dynamism was palpable everywhere, including in all the major macroeconomic aggregates. Nor was there any slowdown after the global financial crisis, any impact of the twin balance sheet problem, or of the four major post-2015 shocks: Demonetisation, GST, the ILFS-triggered credit meltdown and Covid. The official data suggest that the last two decades were an uninterrupted idyll of 6-7 per cent growth.

This misreading complicated macroeconomic policy: At various periods the data signalled strength when the economy was weak, and in others, it suggested that policy should be eased when growth was strong. It also attenuated the urgency for reforms, especially in the period between 2014-15 and 2019-20. After all, why change the policy framework when it was already producing world-beating growth?

In recent years, there have been frequent invocations of puzzles. If growth is strong, why is private investment so weak? Why is net FDI declining? Why are capacity utilisation, wage growth, employment growth so tepid? More recently, why was there pressure on the rupee if real growth was so much greater than anywhere in the world? Each puzzle elicits its own complex and different explanation. But in the spirit of Occam’s Razor, there is one simple, albeit partial, explanation for all of them: Growth has been less strong than it appears!

India has taken considerable pride in being amongst the fastest growing economies in the world. According to IMF data, India has been the fourth-fastest growing major economy between 2011 and 2023. But even on our revised estimates, India will remain amongst the top seven or eight fastest growing economies. National pride may yet be warranted but it does not require a statistical crutch.

Our analysis is both complimentary and complementary to the latest effort by MoSPI. Complimentary because MoSPI has now attempted to address the challenges identified by our analysis and other scholars. Complementary because while the new methodology will apply to future estimates, our research is about the past, focusing on correcting the historical record. Our research will also provide a benchmark for assessing the new methodology when MoSPI undertakes the backcasting exercise to come up with estimates for the last one or two decades

Hopefully, out of the fog of the past there will be light ahead.

The writers are affiliated respectively to the Madras Institute for Development Studies, JH Consulting and the Peterson Institute for International Economics (PIIE). Data and replication code for the paper are available at the PIIE web-site