Economic forecasters are in the dock. Last week, none other than the chief economist of the Bank of England, Andy Haldane, was confessing the crimes of the profession.
The failure to predict the financial crisis was, Haldane said, economic forecasting’s “Michael Fish” moment. Thirty years ago, the BBC weatherman predicted that the UK would avoid the hurricane which devastated the south of England.
Haldane went on to admit that the Bank had “not anticipated” the strength of the economy after the Brexit vote. This is something of an understatement. The Treasury predicted an immediate recession, with the economy shrinking dramatically in the July to September period at an annual rate of up to 4 per cent. The latest Office for National Statistics estimates show that it grew at an annual rate of 2.5 per cent, almost the complete opposite of the prediction made in June.
He attributed these forecasting failures to the fact that economic models assumed that people behaved rationally. However, they had been “irrational” in continuing to spend after the Brexit vote.
This tells us a lot about the mindset of the economics profession. Economists were overwhelmingly in favour of Remain. If you think that the European Union is a great economic success story, bursting with dynamism and innovation, then it would be irrational to continue to spend after a Leave vote. You should be saving for a rainy day. Consumers have taken a different view, but one which may very well be rational.
A serious problem for economic forecasters, and indeed the profession as a whole, is groupthink. Writing about the crisis of the 1930s, Keynes said: “a sound banker, alas, is not one who foresees danger and avoids it, but one who, when he is ruined, is ruined in a conventional way along with his fellows, so that no-one can really blame him.” There is a real reluctance to go out on a limb and risk being proved wrong.
But there is a deeper reason why forecasts repeatedly fail. The problem is that the data on key variables such as GDP and inflation does not contain very much genuine information.
To make successful forecasts in any scientific setting, the data needs to have regularities which can be identified. No-one, for example, can successfully predict over time the outcome of the next shake of a fair dice. The outcome is completely random.
Economic data is not quite as bad as this from a prediction perspective. But the economy is bombarded continuously by so many different events that it is hard to pick out any underlying structure.
Imagine watching a hospital soap in which a screen shows the regular heart beat of a patient. Imagine now the screen plagued by constant interference. It would be difficult to distinguish the “signal” (the heart beat) from the “noise” (the interference). This is what economic data is like. And this is why forecasts are often wrong.
As Clint Eastwood said in Magnum Force: “a man’s gotta know his limitations”. Time for forecasters to wake up to this fact.