Demand components were underestimating GDP
even before 2011 and, therefore, cannot be used to link underestimation to measurement changes. Demand-side estimates, based more on benchmarks and thumb rules, are not robust compared to production-based estimates. Even taking demand side, the NAS estimates tell a consistent story. Investment declined from about 34 per cent to 29-30 per cent of GDP, as clearly shown in the official estimates. Underutilised capacity as well as improvements in productivity allowed consumption-led growth that arrested any significant growth decline till 2017-18. Recent slowdown in consumption credit due to problems in the non-banking financial companies (NBFCs), as well as election-related postponement in private investment and few structural issues, explain the decline in growth to 5.8 per cent in the last quarter of 2018-19.
Subramanian argues the Reserve Bank of India (RBI) consumer confidence survey shows a fall in confidence, so growth could not have been consumption-led. Moreover, credit slowed. But credit fell to manufacturing not to services. There was a rise in retail credit as private banks and NBFCs competed to lend to consumers, financing a consumption boom. Consumption credit growth was high until the IL&FS crisis. Household liabilities rose consistently. Their gross domestic savings fell from about 23 per cent to 19 per cent of GDP.
The aggregate credit data he uses cannot capture these compositional changes.
Increasing use of technology has raised productivity. It is possible that outdated methodologies were underestimating growth rates before 2011—changes may be capturing value added, quality and productivity improvements. Intellectuals sometimes get wedded to concepts and ideas of underdevelopment they are unwilling to shed.
Subramanian argues there could not have been growth in productivity because corporate sector profits declined. But this result is restricted to a corporate database where he does not specify the number of firms. National Sample Survey (NSS) in 2017 showed the unorganised sector’s compound annual productivity growth at 7.2 per cent over 2011-2016 exceeded that of the organised sector (3.2 per cent). Demonetisation impacts fall outside the Subramanian estimation period.
Corporate distress in India was narrow, restricted to a few firms largely in infrastructure. That private corporate savings and investments increased from 9.5 per cent to 12.1 per cent and 11.2 per cent to 12.3 per cent of GDP, respectively, over 2011-16 shows diversity in corporate outcomes.
illustration: Binay Sinha
The two possible reasons Subramanian identifies for over-estimation are: (1) use of MCA21 data for industry and (2) absence of double deflation in a period when there was considerable dispersion between different price indices. But issues highlighted are of doubtful validity and also transient. No one has so far showed problems in the MCA data (only the Central Statistics Office has worked on it) except for saying its growth rates are higher than from Annual Survey of Industries and Index of Industrial Production (IIP). Until this is done, underestimation may be as likely as overestimation—there are grounds for both. The database has matured and become more robust based on 3,00,000 active firms. Dummy firms are cleaned out. As divergence between indices dropped sharply in 2017, measured growth should have fallen if deflation was a major cause of overestimation. But growth rose. Double deflation is itself not well-established.
The shift to the enterprise from the establishment approach, which now includes a number of headquarter services, partly explains higher CSO manufacturing growth compared to that from the IIP. Moreover, CSO captures values compared to volume in IIP.
Finally, the NAS data is massive. The CSO uses around 3,000 data sources and more than 300 surveys. No regressions or correlations can substitute it. There are several continuing problems like old benchmarks, informal sector data etc. But, this cannot be used to criticise revision, which is an improvement in these respects. For example, the NSS 2011-12 informal enterprise survey is better than the 2006-07 one used earlier.
The more heuristic arguments Subramanian makes now are, therefore, not more convincing than his empirical estimates.
The present slowdown shows up in the data to its true extent. It is largely due to cyclical factors, although there are some structural issues. Countercyclical polices can be quickly implemented in several sectors to enhance demand, growth and jobs without waiting for structural reforms. Monetary policy can reduce the interest rates further as real rates are still high; abundant liquidity aids transmission. The government can frontload planned expenditure and rebuild confidence.
Dev is director, IGIDR and former acting chairman and member of the National Statistical Commission; Goyal is professor, IGIDR and a member of the PM’s Economic Advisory Council.