Opinion | Inside the Rashomon world of Indian macroeconomic data

Twitter@Christophmolnar
Twitter@Christophmolnar

Estimating economic activity is never going to be an easy task in an economy that has poor data standards

Indian statistics often resemble a hall of mirrors that the bewildered have to walk through. The images change with every step. The revisions in economic growth estimates by the government are only the most recent example.

For those who came in late, official statistics now show that the Indian economy grew at its fastest rate this decade in the year of demonetization. That flies in the face of all previous evidence, including the most recent labour market data that shows a sharp increase in unemployment.

Economic growth in fiscal year 2017 is now reported to be 110 basis points higher than the previous estimate. The estimate of economic expansion in fiscal year 2018 is a further 50 basis points higher than the earlier one. Even nominal growth for fiscal year 2018 has been bumped up by 130 basis points despite a sharp fall in inflation.

The consensus view that the Indian economy began to lose momentum in the first quarter of fiscal 2017—and that this slowdown was exacerbated after the demonetization in November 2016—now stands demolished. Such drastic data revisions have profound implications—for policymakers, investors and companies. A lot of PowerPoint presentations for the conference circuit will now have to be redone.

In July 2011, Reserve Bank of India governor D. Subbarao gave a speech to explain how the poor quality of data handicaps policymakers. It was ironical that he was speaking on the birth anniversary of Prasanta Chandra Mahalanobis, who had built a statistical system that was once the envy of the world. In a May 2017 essay in this newspaper, Pramit Bhattacharya had written an excellent overview of the house that Mahalanobis built, and the cracks that have developed in recent decades (bit.ly/2MFyo0Y).

Subbarao gave a range of examples to make his point. He first dealt with the changes imposed by new data series that are generated by changing the base year, before moving on to the frequent revisions within an existing data series. The older series of the index of industrial production (IIP) had shown that industrial output between December 2008 and June 2009 was positive despite the impact of the global financial crisis of late 2008. The new IIP series, with 2004-05 as the base year, showed that industrial production had actually declined in these seven months! Indian industry was in recession, but the original data did not show it.

Then there are data revisions without any change in the base year. Subbarao spoke about how the estimates of economic growth for the year 2009-10 fluctuated wildly. The advance estimates released in February 2010 put growth in gross domestic product (GDP) at market prices at 6.8%. The revised estimates put out in May 2010 said that Indian economy would grow at 7.7% that year. The quick estimates in February 2011 were for GDP growth at 9.1%.

Meanwhile, the inflation numbers were also dramatically changed. Wholesale price inflation was at 8.2% in January 2011 and at 8.3% in February 2011, according to the first estimates. These were bumped up by 120 basis points after a few months. Pity the central banker who has to set interest rates based on such data.

All these were admittedly months when there was immense economic volatility—first when output was hit by the global financial crisis followed by a V-shaped recovery because of the coordinated global stimulus as a response to the crisis.

However, the fact remains that policymakers had to take calls based on extremely unreliable numbers of output and prices. Remember: these are not debates about inflation forecasts or estimates of unobservable variables such as the output gap. This is a case of the past being as uncertain as the future.

The recent controversies about official statistics are bound to dent confidence in the numbers released by the government. Estimating economic activity is never going to be an easy task in an economy that is undergoing structural change, is dominated by the informal sector and has poor data standards. Yet, the recent events are definitely cause for concern. Perhaps some diligent researcher should chart out the revisions in Indian macro data over the past few years, and check whether the extent of the revisions has increased.

A good rule of the thumb for anybody tracking the Indian economy is to look at a range of data sources rather than any one number put out by an official agency. My own preference is for high-frequency data (car sales, bank credit, railway freight, tax collections, capital goods imports, etc.) as well as data on prices (rural wages, yield curve, real estate prices, etc.). These can never substitute the national income accounts, but they do offer some extra information that is useful when the macro data is volatile.

The Japanese film maker Akira Kurosawa dealt with the fragility of truth in his 1950 classic, Rashomon. Four witnesses to a crime provide subtle yet different interpretations of what they saw. Kurosawa dealt with the uncertainty of facts by using conflicting narratives of four separate people. The recent events in the Indian statistical system are about different narratives of the past coming from the same data collectors rather than four independent witnesses, but the result is not cinematic brilliance but policy confusion. 

Niranjan Rajadhyaksha is research director and senior fellow at IDFC Institute. Read Niranjan’s previous  Mint columns

Close