Monday, 28 March 2011

Efficient markets and political misdiagnosis

You have a good idea for a post and then Chris Dillow goes and writes something better. Well, it's more in the same vein of what I was going to write. Rather belatedly I recently got around to reading Expert Political Judgment, and very good it is too. It's also made me think about a few different but linked issues that I'm interested in.

First up, it made me think about the efficient markets hypothesis. Remember the theory is that the reason that no-one outperforms the market is because markets factor in all publicly available information. No-one has an advantage and therefore no sustained outperformance. Yeah, well, maybe. Looking at what Tetlock found, I have to say I lean towards the simple explanation that the future is unknowable and that as such no-one can consistently benefit from how history actually turns out. I mean if political forecasting is so poor, why should its financial equivalent a) be any better and b) not be affected by exactly the same problem? Just a thought. It also made me think about what might be the equivalent of the simple algorithms that Tetlock shows can easily beat expert predictions in politics. I mean in financial markets is it an argument for 'technical analysis'? It makes me feel a bit ill even typing that, but I thought I'd chuck it out there.

Secondly, it's obviously a great book to remind yourself what a load of guff gets talked about politics, especially when there's a convincing narrative. Right now there are a lot of people coming out with a lot of plausible sounding predictions of the next few years based on what look like pretty small scale events. We get all sorts of theories trying to fit the next few years into a repeat of a favoured period of recent history. For Labour commentators, for example, no analysis shall be complete before one has provided a sage reference to 1983. Voters, you see, will react to any suspicion of a leftwards trajectory in exactly the same way as they did last time (an obvious point - at the next election 30-year-olds wouldn't have even been born at the time of that election). I just wish Tetlock was tracking the comment pages of our national newspapers.

I have my own expert prediction for British politics based on an extensive review of polling data and sweeping historical developments. The next election result will look quite like current polling, unless something dramatic happens and it doesn't turn out like that.

No comments: