Fuzzy math? Puzzling out the GDP numbers

Written by  //  September 1, 2010  //  Economic & Social Policy  //  4 Comments

The IMF was the cause of some embarrassment for Indian finance minister, Pranab Mukherjee, a few weeks ago. In an update to its biannual World Economic Outlook of April 2010, the IMF revised upward its estimates for India’s output growth in 2010 from 8.4 per cent to 9.4 per cent. “Always, we have dispute with the IMF”, beamed a happy Pranabda. And then, “I was inclined to accept the IMF assessment for India’s growth for the year when they projected it 9.4-9.5 percent. But I am being conservative in my assessment…I will be happy with 8.5 percent plus growth”.

The faith of the IMF appears to have been vindicated, at least for now. The latest GDP figures for the June quarter were out yesterday with the claim that between April and June, the Indian economy grew at 8.8 per cent – the fastest in almost 3 years. Industrial output continues to lead the way with manufacturing continuing its sensational run growing at 12.2 per cent, mining at 9 per cent, services at 9.7 per cent, and even farm output picking up to 2.8 per cent after actually shrinking in 2009-10.

So far so good? The worry is that these big numbers show up despite a slowdown in exports, imports, private consumption and government consumption. I’m not joking. Export and import growth is slowing this quarter, private consumption declined from 0.3% y-o-y in Q2 from 2.6% in Q1 and fixed investment crashed to 3.7% from 17.7%. Government consumption growth was actually negative. FDI inflows declined as well in this period. Gross domestic product numbers that look so good and domestic demand numbers that look so bad is a bit like Michael Bevan having an ODI batting average of over 50 and a test batting average of less than 30. You’ve got to ask yourself: where’s the catch?

Whenever I see big, bright numbers released in the name of growth data I have two concerns. The first is that the statistical calculations that underlay the data which is released often conceal important dynamic trends about the economy and where it sits in the business cycle. This is compounded by the fact that these releases form the basis of important policy decisions. For instance, the first thing I do when I reach my desk every morning is look for the daily Economic Update – which is basically an attempt to sum up a whole day’s work in a bunch of numbers. I’m wondering what the RBI will make of these figures. Is this the green light for further tightening of interest rates on September 16? Inflation, after all, is still close to 10 per cent. Or will they see through the chinks in the shiny armour and hold back from possibly choking domestic demand which already looks like it’s on life support. The numbers are crucial. It’s just worrying that so much hangs on data which is so sensitive to individual events – like, maybe the Commonwealth Games, or the revenue leap from the 3G spectrum sale. 

The second alarm bell that starts ringing when I see these big numbers is the sense of complacency that starts to set in. The media will soon start talking shrilly of “overtaking” China (although, how one overtakes an economy 3 times the size of ours and growing 2 percentage points faster is beyond me). Politicians begin to indulge in some backslapping (though the Congress-led UPA has shown admirable restraint so far). And the spotlight shifts off sectors that are screaming for attention. One of my personal bugbears, for instance, is the miserable performance of farm output, which ironically coincided with the emergence of India as a high-growth economy in the last decade. Another is the impact of sustained levels of inflation on the lives of ordinary people. 9 per cent growth doesn’t look so good to someone facing 10 per cent food price inflation.

I’m not going to go into what I see as serious obstacles to growth and reduction in poverty in this post (the agricultural sector would certainly feature prominently). But a first step would be recognizing that numbers don’t tell the full story.    

4 Comments on "Fuzzy math? Puzzling out the GDP numbers"

  1. Arghya September 2, 2010 at 5:36 am ·

    Anisha, you say that important policy decisions are taken on the basis of these numbers. I really don’t know much about this and writing as a cynical lawyer, isn’t there any form of rationalisation before these numbers are accepted as the gospel truth? Questions such as who published it, what were the parameters considered, what were the parameters not considered, why such numbers are published etc. are questions which came to my mind as I was reading these articles. I wonder if policy makers weren’t even more critical of the numbers before finally accepting it or is it just lawyerly to be needlessly looking for the devil in the details?

  2. Anisha September 2, 2010 at 6:44 am ·

    Arghya, thanks for your comment. The data in this case is published by the CSO, and to question the CSO’s methods and motives each time data is released would be like reinventing the wheel. The economy/markets/government must trust someone to do their numbers for them and that someone in India is the CSO. What everyone seems to forget is that the science of the calculation has always been inexact. Or, should I say, they forget how inexact the calculations are when the news is good (but are always quick to complain if the news is bad). In the face of imperfect data, the quality of analysis becomes very important. God knows the media and sometimes the markets lack good judgment; one can only hope the government does a better job.

    Interestingly, the government has revised the demand component of the figures upward as of yesterday. The GDP figures have remained the same, but the demand component was viewed as being too low (see what I mean about complaining only about the bad news?). Why, how, using what methods: all of this is unclear. Pranab M simply says that with random sampling, “there are some errors here and there”.


    Maybe I’m being unfair and there was some genuine miscalculation that has now been set right. The original set of figures certainly seemed like a right mess. That notwithstanding, I think we could all be a little more discriminating when it comes to interpreting economic performance as measured in the GDP numbers.

  3. Arghya September 2, 2010 at 11:36 am ·

    Hi Anisha, thanks for the clarifications- makes much more sense. Two quick follow-ups: One wouldn’t it always make sense for institutions other than the CSO also to undertake similar analyses? Of course the government needs to refer to some figures and that can be the CSO’s but is there a possibility of a more independent assessment as well, which can ensure the CSO keeps doing its job well and there is some degree of rationalisation with which figures are seen by the market? Or is the task too enormous and data entirely in the hands of the government, for any independent institution/ agency to do the job. Second, when you say quality of the analysis becomes important in case of data which is considered dodgy, are there independent standards regulating quality or is the CSO by virtue of being the CSO deemed to be of sufficient quality?

  4. Anisha September 2, 2010 at 2:57 pm ·

    Arghya, in answer to your first question, the scale of the task is such that only a government can mobilise the resources to collect data. An entire Ministry (of Statistics and Programme Implementation) is devoted to the collection and analysis of statistics; CSO sits within this Ministry. They pull together information picked up from field surveys of individual households and producers at the district and state levels to compile thousands of indicators. It really is a stunningly difficult task in a country the size of ours.

    With respect to the second question: there certainly ought to be some oversight over the methods of sampling, the technology platforms, quality control and so on. There is a National Statistical Commission which was set up a few years ago to make recommendations for improving the quality of statistical information, for instance, employment and inflation data, which is notoriously poor. But, as we saw yesterday, there is still a wide margin for error, “here and there”. Till the CSO makes substantial improvements in the quality of data, policy makers, markets and the very vocal media must be circumspect in drawing conclusions from it.

Comments are now closed for this article.