Wednesday, February 3, 2016

Scott Sumner on behavioral economics

Scott, who I generally love, has had two posts recently allegedly criticizing behavioral economics. But he's actually criticizing a small subset of behavioral economics, and on that subset, to the extent that he's right, he shouldn't be.

In a nutshell, he's saying that the problem with behavioral economics is that its practitioners are too quick to judge actions as irrational. He provides various examples that one might attribute to irrationality that are actually easily rationalized. But they are easily rationalized using concepts from behavioral economics! He uses social preferences to rationalize Christmas giving, anticipatory utility or something closely related to rationalize buying lottery tickets, any number of possible behavioral economic sources of utility to rationalize voting, and loss aversion to rationalize buying warranties on small purchases. Far from building my career on the idea that most people are stupid, I'm building my career on the idea that people are much less stupid than the average classical economist might think.

He even explains why himself: "This kind of thinking led Deirdre McCloskey to turn away from "maximizing utility" models of behavior. I see her point. But I don't see utility as the problem, but rather a lack of imagination as to all the subtle ways that people can derive utility." I completely agree. And a vast majority of behavioral economics is exactly in the business of imagining the subtle ways that people can derive utility and adapting classical models to incorporate them.

This definitely isn't a refutation or overhaul of classical microeconomic models**. It's a set of tweaks to deal with of situations in which those models can't handle the reality of utility. You don't need to tweak the standard model if people are discovered who really love drinking sour milk: that's a weird preference and seems pretty mistaken to me, but classical utility functions don't care what your preferences over regular goods are. De gustibus non est disuputandem. You do, however, need to tweak the theory to account for social preferences, because social preferences interact in ways that classical preferences aren't equipped to represent. Same thing for loss aversion, anticipation, ambiguity, ego, social image, guilt, etc etc etc etc.

I'm going to go out on a limb and hypothesize that the reason Scott seems to be fixating on a particular subset of behavioral economics (the part that catalogues the systematic ways in which people make mistakes*) and getting irate at the implications a particular subset of behavioral economists often prematurely jump to (that they therefore need help making better decisions), and unfortunately accusing the rest of the field of practices that I wholly agree are ill-advised, is that he is libertarian. There's quite a bit of animosity towards behavioral economics from libertarians, of exactly the type Scott is stating, due to their aversion towards meddling in people's choices. They worry that if behavioral economists are busy showing how people need help making decisions, someone will use this as justification for government meddling. I, as a basically-libertarian behavioral economist, share these concerns and it drives me nuts when I see my colleagues jump to these conclusions prematurely (that's a whole other argument).*** But the correct response is to appeal to better behavioral economics, which luckily is most of behavioral economics. It's the solution, not the problem.

~~~

*I am intentionally saying "make mistakes" here, as I really hate how the term "irrational" is thrown around when "mistake" is what is meant. Of course people make mistakes, even systematic mistakes. The fact that they correct (most of) them when they're pointed out means they're still perfectly rational.

**Contrary to popular belief among non-professionals, but certainly not among behavioral economists.

***It also worries me extremely how immediately students jump to "to help policy makers" as a justification for any economic study. Usually to the absence of any other justification. Sure that makes sense for public finance and monetary policy and such, but shouldn't be anywhere near the top of the list for most behavioral economic studies. (It often is, but I maintain it shouldn't be. That's that whole other argument I was mentioning.)

No comments: