T N Ninan: Choose your number

WEEKEND RUMINATIONS

How much fruit does the average Indian family eat in a day? If the government told you it was half a kilogramme, would you believe it? It does strain credulity, but that is what the official statistics put out by the National Horticulture Board say: that India has produced 52 million tonnes of fruit this year. For 200 million families, that is about 500 grammes per day. Since everyday observation suggests that that does not seem to be the case, could it be that a lot of fruit gets processed and exported? No, because only 2 per cent of fruit production in India is processed, and exports are very small indeed. Could it be that a lot of fruit simply goes waste? Perhaps, but even if 20 per cent rots and is not eaten, the overall numbers still don't make sense. So what does that say about the official statistics?

In the week when milkman V Kurien has finally had to call it a day, it is also relevant to look at the country's milk statistics. My colleague, Surinder Sud (who is the agriculture editor), recalls the time five or six years ago when the UN's Food and Agriculture Organisation announced that India had surpassed the United States and become the world's largest producer of milk. Surinder asked FAO where they got the statistics. Your agriculture ministry, they said. So Surinder asked the people in Krishi Bhavan, who said they got the statistics from the National Dairy Development Board. When the tireless Surinder asked the NDDB how they got the statistics, the answer was quite straightforward: "We don't collect milk production statistics."

The problem is not just with agricultural statistics""which have become less reliable over time because the village patwari is a declining breed (following the abolition of land revenue in many states). Food production numbers therefore are often based on crop-cutting experiments done by block development officers, and if you were a betting person, you would not depend on the accuracy of the extrapolation that is done from this. The numbers on industrial production are just as shaky. For instance, the industrial production numbers ignore the small-scale sector altogether. And with de-licensing, even the large companies are not required to report their production numbers (since there is no need any more for import licences, for instance). What they do submit by way of returns is done voluntarily, and the head of the Central Statistical Organisation once confessed that half the companies were reporting the production of items for which they had no licence to begin with! In the sphere of services (which account for more than half of GDP), it gets even worse because the methods used for assessing the value added by doctors, lawyers, private traders and many others are simply not worth the paper they are written on.

If you want a genuine fix on what is going on, therefore, you have to look at other numbers that reflect reality somewhat indirectly. Trade figures are a good index, because there are counter-party numbers in other countries for a cross-check. Price trends can point to abundance or shortfalls in a crop, because the trade always knows. And for industry, corporate sales numbers are a good enough indicator of what is going on. In services, the fate of the hotels, airlines, banks and insurance companies can all be gauged fairly accurately. But at the end of the day, there is no substitute for a reliable, comprehensive statistical system.

Since India does not have that, you can say anything you want and get away with it: that the number of poor in India has grown, and also that it has declined; that almost no one goes to bed without having had two square meals, and also that malnutrition and hunger are widespread; that unemployment has been coming down, and also that it has been increasing. This confusion about the reality is an important reason why some of the debates on economic reform and public policy go on forever, without anyone convincing anyone else. At the end of the day, your bias is as good as mine, and the government's statistics can support both.