Tuesday, September 20, 2011

Quote of the day: On prediction

It is not that I am somehow against prediction. I think that if forty years of econometrics has revealed anything of value to us it is that you cannot very often make successful predictions of the sort that economists seek. Despite the claims of some econometricians, most results they achieve are pretty useless. Anyone can run millions of regressions with a set of data and report a result that seems to pass all tests—though the fact that millions of regressions are run means that most of the conditions of the tests are violated. But even with such results we find that as soon as new data come along the previously reported results or models typically break-down. What I am saying is that no matter how interested in successful prediction some economists may be, this interest does not make it feasible.

- Tony Lawson, Faculty of Economics at Cambridge University in Erasmus Journal for Philosophy and Economics, Volume 2, Issue 1, Summer 2009.

No comments: