Freaks and Geeks

Hah, Clever title. I just now got it. "Freaks" as in "Freakonomics." Very funny. Anyway, hat tip to MP for sending this my way. I hadn't read it in over a year, but it reads well again. I encourage anyone interested in how Steve Levitt fits in the history of economic thought to read it. I think it explains really well what is going on, more generally, in applied microeconomics, and how some of the "identification" stuff has probably jumped the shark by now, but most of us (assistant professors mostly) are paying the price of still having to meet those expectations. Actually, I'm torn on the whole thing. One of the things that makes microeconomics different from sociology or political science is, I believe, the emphasis placed on separating causality from mere correlation. But the problem is that given that most trials are really just observational data, we can't actually do random experiments, so to separate causal mechanisms from merely correlated random variables, we have to exploit randomness in the real world to do the randomized treatments for us. That's kind of interesting when you sit and reflect on it, but as the article points out, there's really only so many questions you can answer this way. There's only so many mistakes made by an agency that accidentally assigned education to someone non-deserving, for instance, or other kinds of weird quirks that help make these assignments for you. But, God forbid you actually try to present a paper to people without identification. I was eaten alive during one talk, early on, because my job market paper (well before going on the job market, actually) had dubious identification. I had given so much thought to the actual question, and the importance of the question, and the model motivating the problem, that the empirics always seemed like an afterthought. That's not good, I know, and I learned and changed through the experience how important it was to think more carefully about the econometrics, but at the time, it was really frustrating since then, and now, I thought to myself, "There is no identification for this other than what I'm doing." Anyway, that may sound like gobblygoop, but I encourage you to read it if you're curious.