On February 28, as per its release calendar, the National Statistical Office (NSO) put out the third quarter gross domestic product (GDP) estimates, that is, for October-December 2019. It showed that domestic output grew at 4.7% at constant prices (that is, net of inflation), compared to the same period the previous year. As the third quarter GDP was marginally higher than the second quarter (July-September 2019) figure of 4.5% (as reported in the earlier data released), experts in the media were quick to infer that the economy is turning around. This was also in line with expectations of many forecasters. Hence, many concluded that the economic slowdown witnessed during the last six quarters has “bottomed out”; government spokespersons endorsed the view.
However, a closer reading reveals that the latest data release has revised the estimates of the first two quarters of the current year (2019-2020) upwards to 5.6% and 5.1%, from the earlier figures of 5% and 4.5%, respectively. The upward revisions have, perhaps unwittingly, changed the interpretation of the current year’s Q3 estimate: the slowdown has continued, not bottomed out; hence, there is no economic revival in sight as of now.
Thus, we have competing views of what the third quarter performance really means for the economy, giving rise to the suspicion of the integrity of the latest revision.
The question therefore is why did the current year’s Q1 and Q2 GDP estimates get revised upwards? The answer is this was simply because the corresponding figures for the previous year (2018-2019) got revised downwards. Many viewed the revision of last year’s estimates as evidence of lack of credibility of the NSO’s revision process. Such doubts are well taken, given the long-standing debate and unresolved disputes on the veracity of GDP figures put out since 2015, when the statistical office released the new series of National Accounts with 2011-2012 as base year.
Constant revisions
We will explain why the annual GDP estimates undergo revisions, and how quarterly output estimation is related to the annual figures.
GDP is a statistical construct — unlike the temperature on a thermometer — prepared using many bits of quantitative information on an economy’s production, consumption and incomes. Many statistical models and methods are used following standardised analytical procedures in line with the international guideline called the UN System of National Accounts (UNSNA). The GDP revision followed the latest (2008) edition of UNSNA. As there are lags and unanticipated delays in obtaining the primary data, the GDP estimates undergo several revisions everywhere (except in China).
GDP estimates are revised five times in India over nearly three years. The initial two rounds, the advanced estimates, are prepared mainly using high-frequency proxy indicators (which probably contain more noise than information), followed by three rounds based on data obtained from various sectors.
More based on projections
Since 1999, quarterly GDP estimates are being prepared, as per the International Monetary Fund (IMF)’s data dissemination standards. Their quality is subpar as the primary data needed quarterly are mostly lacking. For example, nearly one-half of India’s GDP originates in the unorganised sector (including agriculture), whose output is not easily amenable to direct estimation every quarter, given the informal nature of production and employment. Hence, the estimates are obtained as ratios, proportions and projections of the annual GDP estimates.
The National Accounts Statistics (NAS)-Sources and Methods 2012, the official guide for national accounts estimation, states it as follows: “The production approach is used for compiling the QGDP estimates, in terms of gross value added (GVA) and is broadly based on the benchmark-indicator method. In this method, for each of the industry-groups... a key indicator or a set of key indicators for which data in volume or quantity terms is available on a quarterly basis are used to extrapolate the value of output/value-added estimates of the previous year… In general terms, quarterly estimates of GDP are extrapolations of annual series of GDP. The estimates of GVA by industry are compiled by extrapolating value of output or value-added with relevant indicators.”
The NSO continues to follow these practices.
So what can we make of the disagreements over the quarterly GDP growth estimates for 2019-2020? The revisions were probably in line with the latest changes in the annual estimates (second advance estimates). The press release stated: “Quarterly estimates of the previous years along with the first and second quarterly estimates of 2019-20 released earlier have undergone revision in accordance with the revision policy of National Accounts.” True, there were considerable variations at the sectoral estimates after the revision, which probably contained more noise than information. For now, there is little ground to question the revised estimates based on the publicly available information.
Undeniable decline
However, if we accept the latest data, it is clear, though in an alarming way, that there has been an undeniable decline in the GDP growth rate over seven consecutive quarters, from 7.1% in Q1 of 2018-2019 to 4.7% in Q3 of 2019-2020. Considering that physical indicators of production, such as the official index of infrastructure output, or monthly automotive sales, continue to show an unambiguous deceleration, the economic slowdown has apparently not bottomed-out as the government would like to believe.
More seriously, the quarterly GDP deceleration comes over and above the annual GDP growth slowdown for four years now: from 8.3% in 2016-17 to 5% in 2019-20 (as per the second advance estimate). Further, it bears repetition that many have questioned the entire GDP revision since 2015 to the new base-year for possible over-estimation of output growth. If the validation exercises of former Chief Economic Advisor Arvind Subramanian and others have merit, the actual GDP growth rate during much of the 2010s may have been lower than the official annual estimates by 2-2.5 percentage points.
To conclude, India’s quarterly GDP estimates have limited primary information in them. Their revisions are largely extrapolations and projections of the annual figures. Hence, one should be cautious in reading too much into the specific numbers. They are helpful to discern the broad trends in economic activity, which appear grave at the moment. Economic growth continues to drift downwards, from a peak of 7.1% in the first quarter of 2018-19 to 4.7% in the third quarter of the current year. It probably suggests more pain ahead, as the green shoots of economic revival seem nowhere in sight.
R. Nagaraj is currently with the Indira Gandhi Institute of Development Research, Mumbai. E-mail: nag@igidr.ac.in