Point, Counterpoint: Arvind Subramanian Defends His GDP Analysis

In a paper on India’s economic growth in June 2019, former Chief Economic Adviser Arvind Subramanian claimed that India’s annual average GDP growth may have been about 4.5 percent between 2011 and 2016, instead of the 7 percent estimated by the Central Statistics Office.

The paper was critiqued by several prominent economists, including the Prime Minister’s Economic Advisory Council. Subramanian formally addressed the questions raised by other economists at the India Policy Forum by the National Council of Applied Economic Research in Delhi earlier this month.

On Thursday, Subramanian published another paper defending the methods he used to estimate whether India’s GDP estimates are accurate. Some of his counter-arguments are detailed below.

PMEAC: A critique of official GDP estimates must specifically critique coverage or methodology, the author does neither.

Subramanian states that estimating GDP is a detailed exercise. Since the underlying data isn’t available publicly, nobody outside the CSO can estimate GDP, he said. Outsiders can only check to see whether the GDP estimates are plausible, broadly satisfying some macro-consistency checks.

Accordingly, the GDP paper attempted a validation of the numbers through other macroeconomic indicators, explained Subramanian.

He added that while the CSO compiles GDP data from the production side, he attempts a macro-validation of the data from the demand side.

He relies on the basic macro identity: Y= C +I +X -M, where Y= GDP; C= Consumption; I= Investment; X= Exports; M= Imports.

PMEAC: Subramanian is cherry-picking high-frequency indicators

Subramanian said he used the above equation for GDP growth and then looked for reliable proxies for each of the macro indicators. He uses independently produced proxies in place of each variable for investments, exports and imports. While balance of payments data are used for exports, bank credit is used for financing investments.

Both theory and evidence point to a critical role for investment and exports as proximate drivers of growth, Subramanian said. As consumption is harder to proximate, and considering that consumption cannot sustain medium-term growth without increase in investment, consumption is excluded from his analysis, Subramanian said. He conceded that this creates some problem in assessing the correctness of GDP.

Subramanian said that as an added measure, he has used electricity consumption as one of the indicators for the econometric analysis. As it’s a weak proxy for investment and exports, a second regression analysis has also been run, dropping electricity generation as a variable, he said

PMEAC: The correlation of indicators changes over time. Subramanian doesn’t account for this.

Subramanian finds it difficult to understand how key macro-indicators he has used can negatively correlate with growth suddenly in a span of five years. From 2002 to 2011, India measured GDP growth exhibited a strong correlation with investments and exports, he said.

During the period 2011-2016, the Indian economy was hit by a series of shocks. Three of these shocks are likely to have impacted growth directly. Export collapsed, balance sheet of banks came under stress and corporates were increasingly indebted. However, oil prices declined, boosting growth by about 1-1.5 percentage points.

The net effect of these shocks is likely to have dragged GDP growth by 3-4 percentage points, he estimated.

PMEAC: Choice of other sample countries random. Cannot draw inferences from cross-country analysis.

Subramanian used a set of six large emerging economies he believes are obviously comparable to India. His analysis showed that countries which suffered fewer or smaller shocks than India experienced larger declines in GDP growth.

The earlier paper found that the chosen indicators move in line with GDP growth rates in sample countries. These indicators explained between 60 and 75 percent of the variation in GDP growth, across the sample of countries. In the case of India, these variables only explain GDP growth in the pre-2011 period. However, in the post-2011 period, India becomes a marked outlier. “Put more directly, the deviation suggests that the new methodology overestimated growth substantially in the post-2011 period,” he said.

PMEAC: The analysis does not pass the smell test.

The previous paper has widely criticised for its estimate of the extent of GDP overestimation. Subramanian said that the correct interpretation of the result is that growth is likely to have fallen in a range of 3.5 percent to 5.5 percent, with 4.5 percent as the central estimate within that range. “In other words, the confidence level is applied to the range, not the central estimate.” he stated.

Changed economic circumstances and higher GDP level mean that even growth rate of 4.5 percent may not be as low and unreasonable as thought to be, he asserted.

Subramanian rules out rising productivity as an explanation to the deviation in high frequency indicators and GDP growth on account of the collapsing profits of India’s corporate sector. Also, higher tax collections may have to do with substantial changes in tax policy, he said in response to those who pointed to the strong tax collection in the period in which he said growth had weakened.