T C A Srinivasa-Raghavan: An equal and opposite statistic

There used to be a tutor at the Delhi School of Economics who would inevitably end all theoretical discussion, for which the institution was famed, by saying, "show me the numbers". In a country where people despise data, that was an insurmountable challenge, especially 35 years ago.

There was also a professor, now deceased, who used to say that an abstraction from reality was a theory but an abstraction from an abstraction was nonsense. This acted as balm for those who felt oppressed by all that fancy mathematics, including, unbelievably, the calculus of variations.

Over the years, I have used these two devices to stop people in the full flow of uninformed opinion. It has worked magically.

But since I also completely disapprove of those whose algorithm-based empiricism is unsupported by any kind of foundation in economic theory - engineers-turned-economists are especially guilty here - I have enunciated my own little talk-stopper.

And that is: "In economics for every statistic, there can be, and is, an equal and opposite statistic".

So I am very pleased that the economics Nobel has gone to Angus Deaton this year. Deaton has provided us with a method to measure things more sensibly, if not very satisfactorily or completely.

By doing so, he has proved my point - and that of a million others, I daresay - that a lot of the debate in economics, being post facto and based on context sensitive data, is futile.

If only economists would be less in-your-face and certain about themselves when they debate issues, they would not sound as stupid as they do now. They owe Deaton that much.

All in all a richly-deserved prize. Well done, Nobel Prize committee.

What about the econometricians?

The only other economist, in my view at least, who was as richly deserving of the Nobel as Deaton was the late Richard Stone, who won the prize in 1984. He showed the world how to formulate national accounts.

He wasn't the first, of course. There had been Simon Kuznets before him. But Stone introduced the equivalent of the double-entry booking system in national accounts so that there would be consistency. Income on the left hand side had to be matched by expenditure on the right hand side. This put paid to a lot of nonsense.

The other Nobel winners, who were as fully deserving as Stone and Deaton, in my view again, are the econometricians because they invented methods for analysing data properly. But it says something about the prize committee that of the 60-odd winners so far, only a handful is econometricians. Four, to be precise.

This is quite extraordinary considering that the first economics Nobel went to Jan Tinbergen (shared with Ragnar Frisch). After that, econometrics took a back seat. Lawrence Klein, Robert Engle and Clive Granger are the only three other econometricians to have got the prize.

Typing the errors

When the data was not there, economists had to make do with theories. Now that the data - big data! - is there, they are making do only with it.

This is perhaps not as bad as it sounds because the availability of alternative theories can confuse policy-makers into either/or errors or Type One and Type Two errors.

Type One errors occur when a correct hypothesis is rejected. Type Two errors occur when a wrong hypothesis is not rejected.

But the availability of a large body of numbers can lead to both types of errors because as I have enunciated, for every statistic, there can be, and is, an equal and opposite statistic. You just have to choose the ones that suit your purpose.

Governments and political parties thrive on this. That is why they make so many mistakes.

Real time or time series?

The point, in the end, is this: government intervention in the economy, whether macro or micro, can be effective only if it is based on real-time data. The sort of data that economists use is pretty useless for reasons Deaton has said.

But even real-time data is useless if the government lacks the capacity to intervene effectively. Since this has largely been the case in India - no real-time data and little capacity to intervene - I would say our historical experience with poverty intervention is totally unsuited for devising future policies.

Deaton is therefore right: most of the time we don't know what we are doing. But whether the answer is in more time-series data or more real-time data is the real question.

Manmohan Singh's pioneering Unique Identification initiative and Narendra Modi's ambitious programme for Digital India are crucial in this regard. They will provide real-time data.

The Supreme Court, in making Aadhaar optional, has committed both Type One and Type Two errors.