Saturday, September 30, 2006

Useless Competitiveness Surveys

The RBC's James Wimberley posts on one of those wonderful "competitiveness surveys" rightwing people like. Apparently the US has fallen from 1st to 6th, which would be important if it meant anything. Wimberley did a more formal and scientific version of this old post of mine, in which I plotted GDP growth rates and rankings for OECD countries and concluded that it was little better than a random walk. I did that one to back up some complaints of mine about methodology. He did the sums, though, and arrived at a negative correlation of -0.2 - that is to say, falling competitiveness on these things is a signal of faster economic growth.

The important message, though, is not just that they get it wrong. It's about the entire project. This is not science, and it ain't economics. If you were doing science, you'd want different inputs - for example, you'd want the actual growth numbers, not assessments of the policy - and you certainly wouldn't proceed without defining "competitiveness". In fact, it's an exercise in what Richard Feynman called cargo cult science. It looks like science, but it's not the real thing, and that's why the planes don't land.

What it is very much like is the Labour Party's version of evidence-based policy. It's not about deciding what to do on the basis of evidence, it's about using statistics to coerce underlings into doing what the elite thinks they should, or in this case, to promote policies it likes.

No comments:

kostenloser Counter