The argument that the new methodology over-estimates GDP numbers requires deeper analysis
Former chief economic advisor, Arvind Subramanian, has brought back the importance of reliable data for “correct diagnosis and effective policy response” by the government.
In a recent lecture at the Indira Gandhi Institute of Development Research, Mumbai, Dr Subramanian spoke on the issue of gross domestic product (GDP) estimations based on the old and the new series. He compared some macro indicators (based on a research paper co-authored with Josh Felman, former head of the International Monetary Fund office in India) to understand the underlying growth story since the middle of the 1990s. The lecture broadly concluded that the narrative on growth differs depending on the statistics that is considered. By extending the old series, he estimated the GDP to have collapsed and moved to negative territory (close to -2 per cent) in 2019-20 itself, before the pandemic hit the economy.
This research is a continuation of the authors’ other two studies on the subject of GDP measurement that were published by the Harvard Kennedy School.
While it is laudable that these studies have kept the issue of (mis)measurement of Indian macro data alive, and helped in strengthening future efforts, there are issues with respect to conclusions drawn on growth narrative. It appears that the authors have stretched the estimates beyond their limits. Upfront, it is necessary to put forth our view that the old and new GDP estimation and their methodologies are not strictly comparable. The main reasons for this, as the Business Standard editorial opined very rightly, the data sources, coverage and the methods of GDP estimations between the two are quite different. Indeed, for the same reasons, the National Statistical Office (NSO), for a long time, did not attempt to generate the back series based on the new methodology. As a second-best alternative, the Mundle Committee Report (the National Statistical Commission’s Report of the Committee on Real Sector Statistics) has gone for a method that retains the basic structure of the Indian economy. Following this, the NSO brought out a new official back series that was different from the numbers in the Mundle Report, which Dr Subramanian calls “The Great Revision”. With respect to these two back series, it is imperative to highlight that, as the base year revision is always undertaken on the nominal series, in terms of growth rates of nominal series, they almost coincide with each other (as shown in graph-1). Where both series differ is in the real GDP growth rates, thus, suggesting that the differences in the two back series are almost entirely explained by the differences in the deflators used. Indeed, as Dr Subramanian rightly pointed out, large differences were found in the Private Final Consumption Expenditure (PFCE) series. However, even this difference can be explained through differences in its deflators. One can clearly see this in graph-2, where the two series differ quite sharply between 2005-06 and 2011-12. Now, it is not clear from the official sources why the deflators have changed. One could only understand that as the consumer price index (CPI) and its components are majorly used under the new methodology (compared to the wholesale price index and its components in the old GDP methodology), it appears that the same CPI and its components are used for the estimation of real GDP in the official back series. Here, one may also recollect that in the period before the global financial crisis, CPI-based inflation was much higher than WPI-based inflation and that might give a clue with regard to differences in the two back series.
Dr Subramanian argues that the trends in most leading indicators for the period from 2003 to 2010 are more correlated with the back series numbers of the Mundle Committee than with the official back series. And this seems to be a reasonable argument.
However, where it stretches is extending the old GDP methodology to the post-2011 period and argue (both in the lecture as well as in his working papers published in 2019) that the new methodology over-estimates the GDP number. Further, it is argued that the estimates based on the new methodology do not correlate with the trends in some of the macroeconomic indicators. The Mundle Committee does not cover this period as it was not part of the terms of reference. And there is a strong rebuttal on such conclusions by the Economic Survey of 2019-20 (Chapter-10) and by other researchers as well. To be fair to Dr Subramanian, this needs more research and perhaps more comparable and disaggregated data sets before one could dismiss the arguments. We also urge any estimations that relate to base year changes need to be based on nominal terms while retaining the implicit prices, as base year change does not alter the implicit price series. Indeed, the “Official” and “Posterity” estimates shown in Dr Subramanian’s lecture appear to be almost similar to the trends in WPI and CPI inflation where “Official” estimates show CPI inflation trend while “Posterity’ estimates follow WPI inflation.
There are other issues that were raised in the lecture, such as how to measure potential growth or what is “normal” growth for the country. While this is a forward-looking issue and has implications for future policy recommendations, it could be separated from the debate on GDP revisions, which is already a decade old. But many commentators do suggest that the potential growth has declined in recent years and Covid-19 only accentuated the decline. The good news is that several reform measures have been introduced, many of them as part of the Aatmanirbhar Bharat package and some through the Union Budget, and the potential growth could have only improved from the pre-Covid period and not decelerated. In sum, the first half of the arguments that Dr Subramanian makes appear to be largely reasonable, but the second half needs a separate and much deeper analysis.The writer is vice chancellor, Dr BR Ambedkar School of Economics University, Bengaluru. email@example.com. Views are personal