Jayson Lusk recently discussed a forthcoming article in the American Journal of Agricultural Economics. As he explains, the article uses behavioral economics to examine consumer responses to taxes or subsidies on goods. The gist of their conclusion is that the tax (subsidy) itself makes people want to consume more (less) of the good and less (more) of the things politicians prefer.
This is certainly an interesting result, and I enjoyed Lusk's discussion. Specifically, he points out that regulators and politicians are also flesh-and-blood human beings who are subject to the whole range of irrational or counter-intuitive results of behavioral economics. Behavioral economics can certainly help us think about certain situations, but I think the behavioral folks and the anti- (non?) behavioral economists both often miss out on important aspects of human behavior that have impacts on economic analysis.
For instance, I think Lusk is correct to apply behavioral insights to the actions of paternalists. However, even if the problem he identifies didn't exist and economists could design optimal policies with functional analysis, and even if politicians were not subject to these "irrational" results, we still couldn't be very confident that such optimal policies would be implemented. Even if these politicians were perfectly rational, they have an incentive and the ability to enrich themselves at their constituents' expense. All we have to assume to be skeptical of the actual effects of policy is that politicians are potentially as self-interested as the average person. Certainly it's important to pay attention to the effects of cultural rules and institutions on economic behavior which have been studied in depth by Buchanan, Tullock, Ostrom, and many others.
While the behavioral economics crowd has some important insights, I think they, just like many other economists, go too far. This typically takes the form of stating that any deviation from short-run selfishness is "irrational." I think John Cochrane's recent criticism of behavioral economics is a bit harsh, but David Henderson has some interesting thoughts here and here. Apparently Henderson agrees with me that standard neo-classical theory tempered by the insights of the behavioral school and informed of the importance of cultural rules and institutions is likely the best way to approach economics generally. Pete Boettke famously refers to this as "mainline economics." I certainly think he's on to something.
For another interesting example of the need for paying attention to behavioral, cultural, and institutional insights, see this post by Bryan Caplan.