So I think I agree with pretty much every point Paul Krugman makes in yesterdays' Times magazine about where economics went off the rails. Including his big prescriptive point:

Economics, as a field, got in trouble because economists were seduced by the vision of a perfect, frictionless market system. If the profession is to redeem itself, it will have to reconcile itself to a less alluring vision — that of a market economy that has many virtues but that is also shot through with flaws and frictions. The good news is that we don’t have to start from scratch. Even during the heyday of perfect-market economics, there was a lot of work done on the ways in which the real economy deviated from the theoretical ideal. What’s probably going to happen now — in fact, it’s already happening — is that flaws-and-frictions economics will move from the periphery of economic analysis to its center.

There’s already a fairly well developed example of the kind of economics I have in mind: the school of thought known as behavioral finance. Practitioners of this approach emphasize two things. First, many real-world investors bear little resemblance to the cool calculators of efficient-market theory: they’re all too subject to herd behavior, to bouts of irrational exuberance and unwarranted panic. Second, even those who try to base their decisions on cool calculation often find that they can’t, that problems of trust, credibility and limited collateral force them to run with the herd.

On the first point: even during the heyday of the efficient-market hypothesis, it seemed obvious that many real-world investors aren’t as rational as the prevailing models assumed. ... Behavioral finance, drawing on the broader movement known as behavioral economics, tries to answer that question by relating the apparent irrationality of investors to known biases in human cognition, like the tendency to care more about small losses than small gains or the tendency to extrapolate too readily from small samples (e.g., assuming that because home prices rose in the past few years, they’ll keep on rising).

I'd just add one word of caution: 

There are a number of behavioralists who've done and are doing important, fundamental work--people like Robert Shiller of Yale, whom Krugman mentions, and Richard Thaler of Chicago. (These are really two of the fathers of behavioral finance.) But, if there's been one problem with the sub-field over the years--and I'm more familiar with behavioral work in applied micro than behavioral finance--it's that at times it's fallen into a trap Krugman rightly criticizes the Chicago-school maniacs for falling into:  fetishizing the cleverness of an insight even if its real-world utility is next to nil (albeit for completely different ideological and intellectual reasons).

That is, the first generation of behavioralists--people like Shiller, Thaler, and a guy named Larry Summers--were making deep points about how and when economists' rationality assumptions break down. But, by the time the third, fourth, and fifth generation of behavioralists came along, the discipline was ready to go one of two ways: 1.) Start integrating some of those behavioral insights back into our existing models of the world, so that we can really increase our body of knowledge. 2.) Point out progressively narrower and more trivial ways the traditional model breaks down, which makes for a cool paper and might get you written up in the popular press, but doesn't do much to advance our understanding of the world. As I say, I know less about behavioral finance, but my sense of behavioral work outside finance is that it often veered down the second path, when the first is what we really needed.

I actually discussed this problem (among others) in a piece I wrote back in 2007 about Freakonomics  and its broader effect on the discipline. (Contrary to the popular assumption, Freakonomics isn't the same thing as behavioral economics--in some ways it's the opposite--but it does tend to prize cuteness above knowledge advancement in ways that resemble the less edifying behavioral papers). For the piece, I spoke with a brilliant UC Berkeley Harvard economist named Raj Chetty, who's actually one of the few (though by no means the only) young scholars trying to merge behavioral insights into more standard models to forge a kind of synthesis. Here's what Chetty told me at the time:

There was definitely--in the past ten years--it was a very popular thing to do [i.e., the kinds of behavioral papers mentioned above]. At this point, people are getting a little bit, they’re losing patience with that. There are a hundred different ways the model fails. You need an alternative to be constructive. It's reflected in my work. I have a recent paper showing that salience matters in taxation. The traditional assumption is that if you increase the price of a good by a dollar or [the tax by a dollar], it has the same effect. Everyone is fully informed, etc. In practice, I think that’s not true. I basically tested that experiment at a grocery store. [I posted] the tax-inclusive prices--the price with the sales tax--and checked whether it affected demand. The standard full rationality model failed [i.e., demand fell when the tax-inclusive price was posted, even though the overall price hadn't changed]. What I thought was a critical thing to do was to not stop there. If the standard model fails, you need to come up with an alternative model--bounded rationality, costs of cognition. Then you work out the optimal tax policy. People sense that as valuable, constructive. ...

In 70s and 80s, we weren’t thinking about identification [i.e., isolating what causes what], deviations from rationality. It was all a very traditional framework. In the 80s and 90s, we emphasized identification. With that, we see some failures of the standard model, and there are various ways to apply techniques outside [of the standard model]. The next thing is going to be bringing the two back together. Going back to the traditional models, the original questions in economics. A lot of which [relate to] public policy--integrating these new insights. Behavioral failures, better identification. My research is on changing--trying to find better models of underlying behavior.

As I say, I think the reason Chetty is such a hot commodity is that the number of people actually pursuing--to say nothing of advancing--this research agenda is relatively small. (No surprise that he's back in Cambridge this fall.) To get to the place Krugman says we need to go, that number is going to have to increase by a lot. And, unfortunately, I'm not convinced the field is super-prepared to get there at the moment.

P.S. I hadn't come across this nugget from the Krugman piece before: "Larry Summers once began a paper on finance by declaring: 'THERE ARE IDIOTS. Look around.'" Priceless. 

P.P.S. The one thing that makes me optimistic--and it's a subtext of the Krugman piece--is that I think knowledge in economics is pretty countercyclical. That is, during good times, lots of economists (though by no means most or all) don't feel a particular urgency about solving important, real-world problems. They feel free to stroll off into sumo-wrestling corruption, or whatever.  But in the aftermath of the biggest economic crisis since the 1930s, I suspect a lot more super-smart people are going to focus on fundamental problems like asset-price bubbles and long-term unemployment, to say nothing of how we re-establish self-sustaining growth after a deep recession. Economists are just like anyone else in that respect--they respond to the world around them.