On the 18th of October the first Berwick’s swans arrived in Britain on their way south from their usual home in Siberia to their winter feeding grounds. This was interesting as it was two weeks earlier than expected, which ornithologists attributed to the likelihood of unusually cold weather arriving soon.
At the time this was dismissed as an ‘old wives tale’, but here we are a few weeks later gripped by the coldest weather the country has seen at this time of year for 17 years. Birds one, Met Office’s multimillion pound computers nil!
Nine days later on the 27th October Charles Bean, the Deputy Governor of the Bank of England, gave a speech at the Royal Statistical Society in which he claimed that the recession could not have been predicted and that the only recessions which are “broadly predictable” are those that are “deliberately policy-induced in order to squeeze inflation down.” He went on to say that “almost all forecasters are in the same boat.”
This is a rather startling admission from the person whose job it is to make decisions based on future predictions, but it possibly provides an insight into where it is all going wrong. For example, when he said; “The moral from this is that one should not expect to be able to predict the timing and scale of these sorts of events with any precision”, did it not occur to him that this statement seems at odds with the various commentators, business people and journalists who had been predicting the recession for many months before it occurred. Perhaps what he actually meant was that it is impossible to predict these things using statistical modelling and computers. If that is indeed what he meant to say, then I would wholeheartedly agree!
The problem is that statistical models rely exclusively on past data. Whereas this is fine for predicting the tensile strength of metal or the fuel load necessary to take a rocket to the moon and back, it is a fundamentally flawed when predicting anything that involves people.
The problem with people is that we do not behave in predictable ways. Using historic data to predict human behaviour is therefore a bit like trying to drive a car forwards whilst looking out of the rear window!
But since predictions involving people are both necessary and valuable, how should we go about making them?
The answer is that you simply need to understand the psyche of the people.
By this I do not mean that you should rely on market research and opinion polls, as that is simply another form of statistical modelling, I mean that you need to be genuinely close enough to the public to understand their mood and sentiment. This, for example, is something Margaret Thatcher used to do extremely well, which is probably why she remained in power for so long, it is also why Delia Smith is a more popular cookery writer than Heston Blumenthal, and why Marks and Spencer is again doing well.
To make accurate predictions therefore you do not need the statistical resources of the Bank of England, you simply need to keep your eyes and ears open to the clues that are all around you.
Perhaps if the Governor of the Bank of England saw his role as being something other than passing comment on the intellect or otherwise of David Cameron and George Osborne and if the Deputy Governor went to speak to the WI rather than the Royal Statistical Society, they might find that predicting booms and busts would become a whole lot easier.