26 Şubat 2011 Cumartesi

Why can’t economists predict disruptive events?

Gavyn Davies*     The Financial Times

When the Queen asked asked an academic at the LSE why the economics profession had failed to predict the credit crunch, she raised a topic which continues to resonate. In fact, the IMF’s watchdog criticised the organisation on exactly those grounds yesterday. Although many answers have been given to Her Majesty’s question, I suspect that none of them has really settled the issue. Her question is disarmingly simple, but the answer is not.

The latest academic attempt to tackle the question is this piece by Raghuram Rajan. He is well qualified to write on the matter, having delivered a very perceptive warning about a possible crisis to the entire senior cast of global central banking at Jackson Hole in 2005. They politely ignored him. Prof Rajan now argues that economists had all of the models required to understand the credit crisis, but that the subject suffers from being segregated into increasingly narrow fields. It therefore lacks people with the broad overall view necessary to connect all of the diverse strands. This is indeed a problem, but it may not be the whole answer to the Queen’s question.


Although it is often entertaining to poke fun at economists, they are not the only professional group which finds it difficult to make accurate predictions about disruptive future events. In fact, this seems to be a problem which is alarmingly common among many of the social sciences, if not of the natural sciences. The definitive work on this question was published in 2005 by Philip Tetlock, a philosophy professor at Berkeley. His book (“Expert Political Judgment”, which is well summarised here in The New Yorker) won many prizes when it first appeared, but it still seems to be very little known among economic policy makers. Which is a great pity.

Tetlock systematically collected a vast number of individual forecasts about political and economic events, made by recognised experts over a period of more than 20 years. He showed that these forecasts were not very much better than making predictions by chance, and also that experts performed only slightly better than the average person who was casually informed about the subject in hand. Depressingly, he found that there were remarkably small gains to greater expertise when it came to forecasting accuracy.

This result, which applied to politics and international relations as well as economics, emphasises how hard it is to forecast disruptive or discontinuous events – in fact, events which are not indicated by relatively simple extrapolations of recent past trends. The credit crisis was certainly not one of those “forecastable” events. If we ask why economists failed to predict the credit crisis, we should also ask why political scientists failed to predict the recent Egyptian rebellion, or indeed a terrorist event like 9/11. Or why seismologists cannot predict earthquakes.

Tetlock goes further. He argues that “experts” can be broken into two groups, which he calls hedgehogs and foxes, following the nomenclature first suggested in Isaiah Berlin’s essay on Tolstoy in 1953. (“The fox knows many things, but the hedgehog knows one big thing”.) According to Tetlock, some experts are “hedgehogs”, who hold very strong, often ideological, opinions which are changed very infrequently, if at all. An example would the following viewpoint:
 A rise in the money supply always causes inflation. Therefore the Fed’s policy on quantitative easing will inevitably lead to eventual disaster. It is just a matter of time.

The other group of experts are “foxes”, who express more nuanced opinions which are more often changed as the nature of the evidence changes. For example:

Although the Fed’s policy is increasing the supply of bank reserves, this is not feeding into broader monetary aggregates like M2, and in any event the existence of spare capacity in the economy means that easy monetary policy is more likely to lead to a rise in output growth than to an increase in inflation, at least for now.

It is easy to see that the hedgehogs are by far the more likely group to predict a very big disruptive event like the credit crisis. By temperament, they are willing to be continuously wrong for very long periods in order to be right periodically on very big issues. And that did indeed happen to the relatively few economists who predicted the onset of the credit crisis, often many years before it happened.

But is this the optimal way of forecasting? Tetlock shows that the foxes on balance out-perform the hedgehogs, so the answer is probably “no”. Therefore, it seems to be optimal to make incremental forecasts, which are frequently changed in response to incoming evidence, even though these forecasts may be a whole lot more boring than the more dramatic pronouncements of the hedgehogs. However, the difference between the track record of the foxes and the hedgehogs is not very significant. Neither group forecasts disruptive events particularly well, not because the experts are incompetent, but because the task is inherently so difficult.

As an example of how professional (and presumably rational) forecasters behave in practice, consider the example of today’s Bank of England decision on interest rates. Out of 62 forecasters sampled by Bloomberg, the entire 62 said that there would be no change in rates. But in fact many of them probably thought that there was a reasonably high chance of a rate increase, although it was not the most probable single outcome. If all of them thought that the chance of a rate increase was 30 per cent, they would all have said that no change was the single most likely outcome, even though the subjective probability of a rise was considered to be 30 per cent by the group as a whole.

The same may have been true of the credit crisis. Although very few economists, if any, predicted that this would take place in 2008, many of them probably assigned a positive probability to a crisis occurring at some point in the not too distant future. If that is the rational way to forecast, we should not be too surprised by the outcome (though it does not excuse regulators like the IMF, who should be acting to prevent a disruptive outcome, even if that seems fairly unlikely).

So one answer to the Queen’s question about why economists failed to predict the crisis could be: “Because economists are rational, ma’am”. But I am not sure she would be terribly impressed.

*Gavyn Davies is a macroeconomist who is now chairman of Fulcrum Asset Management and co-founder of Prisma Capital Partners. He was the head of the global economics department at Goldman Sachs from 1987-2001, and was chairman of the BBC from 2001-2004. 

He has also served as an economic policy adviser in No 10 Downing Street, an external adviser to the British Treasury, and as a visiting professor at the London School of Economics.

Source:
http://blogs.ft.com/gavyndavies/2011/02/10/why-cant-economists-predict-disruptive-events/ 

Hiç yorum yok:

Yorum Gönder