By John Kay Financial Times
Published: November 26 2008 02:00 | Last updated: November 26 2008 02:00
In the 1980s, it seemed that computers held the key to economic forecasting. With large models and sufficient processing power, predictions would become more and more accurate.
This dream did not last long. We now understand that economies are complex, dynamic, non-linear systems in which small differences to initial conditions can make large differences to final outcomes - the proverbial flapping of a butterfly's wings that causes a hurricane.
So economic crystal ball-gazing remains unscientific. The trend is the forecaster's friend. Extrapolation assumes that the future will be like the past, only more so. We project current preoccupations - the rise of China and India, global terror, climate change - with exaggerated speed and to an exaggerated degree.
We forget that our preoccupations change. The people who worry about these issues today would 20 years ago have worried about the coming economic hegemony of Japan and the cold war. These issues were resolved in ways that few predicted.
It is a safe prediction - and the only one I shall make - that the topics that grab our attention 20 years from now will differ from those that consume us today and, if anyone has guessed what they are, it is only by accident. The future is unknowable. As Karl Popper observed, to predict the creation of the wheel is to invent it. To anticipate a new political force or economic theory, or even a new product, is to take the main step in bringing it into being.
If extrapolation is the forecaster's friend, mean reversion is the forecaster's crutch. Much of the time, you can predict that next year's figure will be somewhere between this year's level and the long-run average. But mean reversion never anticipates anything out of the ordinary. Every few years, out-of-the-ordinary things happen. They just have.
Still, you might think there would be large rewards for those who succeed in anticipating these events. You would be wrong. People who worried before 2000 that the "new economy" was a bubble, or warned of the terrorist threat before September 11 2001, or saw that credit expansion was out of control in 2006, were not popular. They were killjoys.
Nor were they popular after these events. If these people had been right, then others had been blind or negligent, and the latter preferred to represent themselves as victims of unforeseeable events. As John Maynard Keynes observed, it is usually better to be conventionally wrong than unconventionally right