I think Bainbridge's article on efficient markets vs. behavorialism is good. Now, because of my fairly Hayekian/Coasian sensibility, I can't buy the ECMH in it's strict formulation. Indeed, I agree with most behavioralist findings, although I often disagree with behavioralists about the upshot of those findings. Now, I do find it intriguing that Thaler puts his money in index funds, just like you would if you thought the ECMH was true. And it is what I would do if I wasn't so poor I had to beg for money on my blog. Indeed, it looks like the ECMH does a good job of approximating the real world (or the other way around), despite the falsity of a number of its underlying assumptions. So how does it do this? Well, in the gap between idealized behavioral assumptions and actual approximately efficient markets is entrepreneurship. There are folks and firms out there gathering intelligence, keeping their eyes open, trying to cash in on ephemeral assymetries in information, and thereby moving prices to what they ought to be.
The interesting thing about efficiency-enabling entrepreneurship is that it is NOT a built-in assumption of the theory. There are institutional antecedents — legal, moral, cultural — to an effective climate of intelligent, creative proift-seeking. So, despite the fact that Thaler is right about the quirks and foibles of human decision-making, our institutions, both formal and informal, are good enough to induce behavior that reasonably approximates neo-classical efficiency, making it right for Thaler to invest as if the ECMH were true.
The way I see it, the interesting questions are the questions about the way various institutional structures, formal and informal, facilitate efficiency-approximating behavior. We know way too little about this. And here's a connection to so-called “libertarian paternalism,” discussed below. Will Baude says he liked the Sunstein/Thaler paper. I didn't dislike it, exactly. But, like Klein, I found the idea of libertarian paternalism needlessly confusing (and perhaps even willfully and strategically confusing). The interesting thing about framing effects, cognitive biases, and so forth, is that boundedly rational agents like us are not necessarily indifferent between formally equivalent institutional designs. So, if this is true, its pretty obvious that insofar as we're picking institutional designs (as often we must) we should pick the ones under which we're more likely, given our psychological constitution, to satisfy some normative standard, whether it be efficiency, public health, or whatever. Great. Are S & T saying anything more interesting than that, true though it may be? But according to S&T's idiosyncratic usage James Madison is among history's great paternalists. Yet I don't think that's why we call him a founding “father.”