Models need human factor

Here is my latest comment from Fund Strategy.

Numbers can be an essential aid to understanding financial markets and economies. They can also be highly misleading.

One of the most valuable aids to gauging the size and performance of economies is GDP. It should only be seen as a rough estimate of economic output but with that caveat it is extremely useful.

The critics, including Nobel prize winners, who argue that GDP is a poor measure of well-being miss the point. Since the inventors of GDP conceded from the start that it was not a measure of welfare, the criticism is hardly a revelation.

In fact, contrary to the critics, the opposite point is much more valid. The problem with GDP is that it substantially underestimates the contribution of ­economic growth to improving human welfare.

On the other hand, numerical predictions of the likely market impact of the elections are in a long line of dubious mathematical extrapolations. Doing the maths is easy enough but, as with any such model, the axiom “garbage in, garbage out” should be remembered.

A typical procedure is to compare how the markets performed during different governments dating back to the 1930s. It then suggests that, based on past performance, a Conservative government is likely to be better than a Labour one.

There are many problems with this approach but the most obvious is that it assumes the past is a good guide to the future. Since the division between left and right ceased to have any meaning in the 1980s the approach makes no sense. Until then the Conservatives were a traditional pro-market party of business while Labour represented the trade union movement and stood for extensive state ownership. But under Margaret Thatcher’s Tory regime the idea of an alternative to the market was ditched while the old-style trade union movement was destroyed.

This particular misuse of statistics points to a ­common problem with the way figures are used to make points about social questions. The role of human consciousness is often downplayed or even ignored completely.

It makes no sense to model the social world in the same way as the natural one. Atoms and molecules have no mind of their own. Nor can the sun decide not to rise in the morning. Yet too often models of human societies are essentially crude applications of the type of models found in physics.

Numbers certainly have their place but they should not be used unthinkingly to draw dubious ­conclusions about economies or the social world more generally.