Economic forecasting will never be an exact science


John explains why reliable economic forecasting is, in principle, impossible

In last week’s column, I warned against DIY economics – the false propositions that people who know little about economics think are true. Since then, my mailbox has been filled by people who say they are sticking to DIY because they have been let down by cowboy tradesmen. Not dodgy decorators or crooked carpenters, but economic forecasters.

Most economists do not do forecasting and, like honest craftsmen, they resent those who earn more money for shoddier work. Several decades ago, in the first flush of naive enthusiasm for the potential of computers, many people believed that large models would describe the evolution of the economy ever more perfectly. But these hopes were unfounded. It is not that economists are insufficiently clever, or their tools insufficiently powerful. Predicting whether the dollar will rise or fall in the second quarter of next year, or the level of the Standard & Poor’s index at the end of 2004, is in principle impossible.

The reasons fall into two main groups. Many of the processes we find in commercial and economic life are dynamic and non-linear. What this means in everyday life is that small differences in where you start can make a large difference to where you end up. Almost everyone is now familiar with the metaphor of the butterfly whose flapping wings provoke a tornado weeks later, thousands of miles away. If small events can have large effects on complex systems, even knowing 99 per cent of what you need to know leaves you vulnerable to large errors. And 100 per cent knowledge is impossible.

That is why we can never aspire to accurate forecasting of economic events, just as we can never know whether it will rain on June 4 next year or the date of the next Tokyo earthquake. But economics is even harder than meteorology or seismology. Most economic systems are reflexive – what happens is influenced by how we perceive what will happen. Good mechanisms for forecasting stock prices will not be found because their very discovery would affect stock prices. If there had been reliable predictions that the value of the euro would move from $1.17 to 81 cents and back again, these movements would not have occurred.

But we can identify earthquake zones even if we cannot predict earthquakes and we can look forward to summer even if we cannot forecast the weather on June 4 next. Seismologists tell us where not to build our houses, and meteorologists help us know where to sell sun cream and when to take an umbrella. Useful economic knowledge is of a similar kind.

When the value of the euro fell below a dollar, it was virtually certain to appreciate, because it is rare for the currencies of rich countries to deviate so far from purchasing power parity for long. But no one could have successfully predicted just when the euro would appreciate, or that it would first fall to almost 80 cents, although the pattern of momentum in the short run and mean reversion in the long run is a common feature of speculative markets. And, in answer to the questions some readers may be asking: you need to take an economics course to learn about purchasing power parity and mean reversion, and a physics course to learn about momentum, and yes, I did fill my boots with euros at an average price of 87 cents.

But, despite the inevitable failures of economic forecasting, people continue to want the knowledge it would provide. As even DIY economics will tell you, where there is demand there will be supply. Cowboy tradesmen remain in business because there are always gullible customers, and the same is true of economic forecasters.

But when someone tells you that the dollar will appreciate in the second quarter of next year, or what the level of the S & P will be at the end of 2004, they do not know what they are talking about. If they did, they would not be making these predictions. But if you stop asking economists to forecast the future, there are other interesting things they can tell you. I predict this column will appear again next Wednesday, and hope you will find something in it to enjoy and profit from.

Print Friendly, PDF & Email