It is widely accepted that craftsmen feel most comfortable with tools they are most familiar with. A carpenter will opt for a chisel and saw where possible, whereas a plasterer will favour a trowel.
A favoured tool in the investment world is the economic model, which is in turn drawn from statistics. Such models provide a way of examining how different assets are likely to behave under different circumstances. Users of such tools recognise that they are not perfect but they are seen as preferable to relying on human judgment.
With greater political uncertainty in recent years – including Brexit and Donald Trump’s election in the US – the temptation is for investment professionals to apply their approach to politics. They assume that such tools will help to explain the likely impact of such developments.
Although the appeal of such an approach is understandable, it should be resisted. It simply provides a spurious scientific gloss to the understanding of political events. The output will no doubt include numerous numbers but its analytical value is another matter.
To understand these limitations, the best starting point is to consider the law of large numbers. This central premise of statistics is that the more often an experiment is repeated, the more the frequency of a particular event will converge to its average. For example, in any given spin of a roulette wheel there is a one in 37 chance of correctly predicting any number. But if the wheel is spun millions of times it is possible to forecast with a high degree of accuracy how often each number will occur.
The problem with applying statistical models to politics is precisely that they are based on the law of large numbers. They assume that human society is akin to the natural world. If the human world is examined at a large enough scale – rather than focusing on individuals or small groups – then its movement will be akin to a natural system.
Such an approach does not work because human society does not resemble nature. Humans are conscious beings who, unlike balls in roulette wheels, can themselves influence events.
This matters because politics is precisely about discontinuities in human affairs. As Hannah Arendt, perhaps the greatest political theorist of the twentieth century, argued: “The application of the law of large numbers… to politics or history signifies nothing less than the wilful obliteration of their very subject matter.”
Statistical models are based on assumptions of continuity, whereas the key focus of politics is social change. Applying economics models to understand politics is like trying to use a trowel to saw a piece of wood in half.
Daniel Ben-Ami, Deputy Editor