Circuit Breaker. What’s wrong with the IIP?

Of late, it seems to be the fashion in market circles to dismiss any positive economic data about India as jhumla , while embracing negative data points, with gloomy relish, as the gospel truth.

This is the only explanation for the market’s overblown reaction to the Quick Estimates of the monthly Index of Industrial Production (IIP) released this week. Yes, the monthly IIP reading for July 2016 sprang a nasty surprise by shrinking by 2.4 per cent year-on-year when the consensus was for it to rise 1.4 per cent. The market also seems to have been spooked by the 3.4 per cent contraction in manufacturing and the 29.6 per cent fall in capital goods output.

But it is surprising that the monthly IIP should be taken so seriously. Of all the macroeconomic indicators one can use to gauge the state of the Indian economy, the IIP is the least robust and reliable.

To understand why, let’s dig a little deeper into this month’s IIP release. Apart from stating that the IIP showed a 2.4 per cent decline, it lists out the large items responsible for this swing.

So, we learn that in July 2016 compared to the previous year, Indian industry sharply (and unaccountably) cut back on its output of rubber insulated cables (down 91 per cent), marble tiles/slabs (down 62 per cent), HR sheets (down 59 per cent), and sugar machinery (down 53 per cent). But it also churned out room air-conditioners (higher by 103 per cent), wood furniture (up 74 per cent), ready-to-eat (up 65 per cent), and colour TV sets (up 29 per cent) at an unusually brisk pace.

Now, if you are a statistician, the magnitude of those monthly changes should lead you to smell something fishy. But things gets fishier once you find out that ‘rubber insulated cables’ alone managed to depress IIP growth by a good 4 percentage points. In short, if one blanks out those pesky cables, the IIP for July 2016 would have grown by nearly 2 per cent, instead of shrinking by a similar number. And lo, we would have a picture of expanding Indian industry, instead of a recessionary one.

Skimming through IIP releases for the last many months would also tell you that ‘rubber insulated cables’ have consistently been the villain of the piece not only in depressing the overall index, but also in pushing the whole capital goods segment off a precipice. But leaving out problematic items doesn’t really address the basic shortcomings of the IIP, which are threefold.

One of the key problems with the IIP series is that it continues to use 2004-05 as its base year even as other macro indicators such as the new GDP series and Consumer Price Index have already moved on to 2011-12.

The antiquated base year results in the IIP index, even today, faithfully capturing the (current) growth in a basket of goods that drove Indian industry a good eleven years ago (2004-05). The series uses what is called a Laspeyres index, where the constituents do not change automatically over time.

To illustrate, while the present-day IIP may capture every small blip in the output of colour picture tube television sets, stereo systems, compact discs, radios, feature phones or desktop PCs, it wouldn’t pay any attention to the burgeoning output of LCD or LED TVs, MP3 players, smartphones or notebook computers.

In fact, this is why the Saumitra Chaudhuri Committee, which was constituted to recast the IIP in 2014, had suggested a new basket of 775 items in place of the present 682 for a contemporary picture of industrial production. It suggested chucking as many as 235 items from the 2004-05 basket, replacing them with 335 new items. But that recommended overhaul of the IIP still remains on paper. The obsolete basket could be one big reason why the IIP consistently presents a bleaker picture of Indian industry than the Annual Survey of Industries, or the manufacturing leg of the new GDP.

If an obsolete basket is one problem dogging the IIP, gaping holes in the primary data are another. The Central Statistics Office, independent researchers and numerous committees have been complaining for years that the wild swings in individual components from month to month are caused mainly by the reporting industries taking a very lackadaisical view of updating data.

The problem really is that no single statistical agency in India is responsible for the veracity of primary data that goes into the IIP. On paper, the IIP is supposed to capture monthly output changes for an ambitious 682 products, reported in by 16 different ‘source agencies’. These source agencies receive their monthly feed through “mail, fax or questionnaires” from designated factories. But many of these factories often skip reporting in their data. In fact, a 60 per cent response rate is considered pretty good.

To make up for the missing data, agencies are allowed to plug in the previous month’s number, repeat the one from the same month last year or average three months’ numbers. You can well imagine how this unscientific ploy can lead to an erratic IIP. A sector that is the beehive of activity may not report in its numbers for many months, and then suddenly show a big surge. Or one that is dying a slow and painful death may keep plugging in old data until it suddenly melts down. The Quick Estimate of the IIP which is based on first-cut data is the most susceptible to this error. The second or third revised estimates would usually capture a bigger dataset.

A third issue with the IIP lies in the high volatility of its key components which, as RBI governor D Subbarao once put it, makes it “analytically bewildering”. One gets a sense of what he is talking about if one plots the growth rates of the capital goods component of the IIP over time. It closely resembles the lurching of a drunken sailor. In the last six months alone, the IIP has shown between a 9 and 29 per cent contraction in capital goods output. Lacklustre private investments and lumpy order flows may account for negative trends. But wild monthly swings can only be explained by flawed data capture.

One explanation offered is that most large pieces of capital equipment (say, boilers or turbines for a power plant) have a production cycle that lasts far beyond a month. As many of them may not hammer out a boiler or turbine every month, their output may go from 0 for several months to a 100 in a single month. A better way to measure their output would be to capture the value of work in progress. Given that the IIP’s flaws are well-known, it is time the Centre and the Statistics Office took up the task of recasting this high-frequency indicator. Right now, the only useful purpose it serves is as a handy tool for GDP sceptics to remind us of the parlous state of the Indian economy.