Repairing the house that Mahalanobis built
In a July 2011 speech, former Reserve Bank of India governor D. Subbarao had famously described the index of industrial production (IIP) as “analytically bewildering”. More blunt observers have dismissed it as practically useless. The new index released by government bean counters on Friday gives us a good idea why Subbarao had complained about how poor data quality means that Indian policymakers are often flying blind in a turbulent world.
A comparison with the monthly values of the old IIP and the new IIP is revealing. The data for December 2012 is particularly stark. The old IIP had contracted by 0.6%. The new IIP shows that industrial output grew at an impressive 7.2% that month. These are descriptions of parallel universes—one in which industry is in recession and the other in which it is booming.
The situation is magically reversed when we look at October 2012 data. The old IIP expanded by 8.4% while the new one suggests a more modest increase of 3.6%. So what really happened that month? More generally, the industrial slowdown in those years seems less severe than the old IIP had suggested. Those in charge of economic policy at that time could now turn around to claim that they were fundamentally misled by the data being put on their desks by government statisticians.
The new methodology used to calculate industrial output promises to correct some of these problems. Most importantly, the new IIP will hopefully better capture the deep structural change in the Indian economy since the previous index was devised with fiscal year 2005 as a base, for example through the inclusion of renewable energy in the electricity index. The infamous volatility in the index of capital goods production—rubber insulated cables being one of the villains—could also come down since data on work in progress will now be taken into account to smoothen what is otherwise dominated by bunched orders.
The government has also released a new series on the wholesale price index (WPI). The old and new series of the inflation index move far more in sync with each other than what we see for the old and new series for industrial output.
The most important change as far as the new inflation measure goes is that prices of items in the new WPI will not include indirect taxes. India has thus moved a step closer to a producers price index that is used globally to capture economic realities. The change in the WPI will also mean that the price index used to deflate nominal gross domestic product (GDP) every quarter will be net of taxes.
The launch of the new IIP and WPI comes after the move to a new method of measuring GDP. India now also has a national consumer price index that is the formal target for monetary policy. An employment measure to judge the state of the labour market is in the works.
Yet, despite all this, the credibility of Indian economic data has been damaged in recent decades. India once had one of the best data collection systems in the world, thanks to people such as P.C. Mahalanobis and Pitambar Pant.
In their preface of a book on poverty measurement that has been recently republished with a few new essays, economists Abhijit Banerjee, Pranab Bardhan, Rohini Somanathan and T.N. Srinivasan quite rightly point out: “At a time when those in power talk blithely about alternative facts and social media makes its own reality, it is extraordinarily important to underscore the value of objective and careful measurement. When, as in India these days, the representatives of government are forced to voice their scepticism about numbers their own statistical apparatus is producing on things as basic as GDP, policymaking turns into a journey without maps…”
India is not an easy country to capture with credible data. It is dominated by informal economic activities that do not report to government agencies on a regular basis. The widespread use of proxies are inevitable. The government statisticians definitely do not have an easy job. However, there is a deeper problem with the quality of the statistical apparatus—from the design of surveys to the quality of people collecting data on the ground—that policymakers have complained about for at least two decades now.
The government not only needs to dramatically improve the quality of data it puts out but also has to embrace new sources of big data that can give its administrators a better sense of the state of the economy.
In other words, the house that Mahalanobis built needs urgent attention.
Is government economic data reliable? Tell us at email@example.com