Pick your model

I enjoyed this piece by Dani Rodrik at Project Syndicate:

Indeed, though you may be excused for skepticism if you have not immersed yourself in years of advanced study in economics, coursework in a typical economics doctoral program produces a bewildering variety of policy prescriptions depending on the specific context. Some of the frameworks economists use to analyze the world favor free markets, while others don’t. In fact, much economic research is devoted to understanding how government intervention can improve economic performance. And non-economic motives and socially cooperative behavior are increasingly part of what economists study.

As the late great international economist Carlos Diaz-Alejandro once put it, “by now any bright graduate student, by choosing his assumptions….carefully, can produce a consistent model yielding just about any policy recommendation he favored at the start.” And that was in the 1970’s! An apprentice economist no longer needs to be particularly bright to produce unorthodox policy conclusions.

Nevertheless, economists get stuck with the charge of being narrowly ideological, because they are their own worst enemy when it comes to applying their theories to the real world. Instead of communicating the full panoply of perspectives that their discipline offers, they display excessive confidence in particular remedies – often those that best accord with their own personal ideologies.

Is it that bad? Well, statistician Kaiser Fung of the blog Numbers Rule Your World) says that it's actually much worse and that Rodrik doesn't go far enough as he compares Rodrik's point with a critique of economic modeling in Emanuel Derman's new book Models Behaving Badly (which I haven't read yet):

My own view, informed by years of building statistical models for businesses, is more sympathetic with Derman than Rodrik. There is no way that economic (by extension, social science) models can ever be similar to physics models. Derman draws the comparison in order to disparage economics models. I prefer to avoid the comparison entirely.

The insurmountable challenge of social science models, which constrains their effectiveness, is that the real drivers of human behavior are not measurable. What causes people to purchase goods, or vote for a particular candidate, or become obese, or trade stocks is some combination of desire, impulse, guilt, greed, gullibility, inattention, curiosity, etc. We can't measure any of those quantities accurately.