The failures of contemporary macro theory remind me of the time we were interviewing a highly touted graduate student on the academic job market (I believe he was from the University of Minnesota, but I am not totally sure). We asked him how he would teach macro to public policy students at the Kennedy School. He thought for a while, and said: "I guess I would do it all using the overlapping-generations model, and since this is an introductory course, I wouldn't bring money in at all." Enough said.
The good news is that the problem, in my view, is confined largely to macroeconomics. The rest of economics is in good shape, even if, as I argue here, economists haven't done a good a job of putting to use what they know (or should know, from their own models).
The bad news is the world could really use some practical, relevant macroeconomic theory at the moment. Brad DeLong and Paul Krugman are doing a superb job of reminding us of the continued relevance of Keynesian thinking. But they are hampered by the absence of micro-founded models that plausibly deliver the Keynesian remedies they advocate. The economics profession doesn't take an argument seriously until the argument can be laid out with a well-specified model that respects accepted standards of modeling--an attitude that Krugman himself has vociferously defended in another context. Maybe the problem is with those standards of modeling, but other fields within economics have been much more willing to adapt themselves to behavioral thinking, or the fact of missing markets, or the role of market failures.
So what is wrong with macroeconomists?
P.S. Peter Dougherty reminds me that I should have mentioned the excellent new book by Akerlof and Shiller, which introduces behavioral economics into macro. I should have also mentioned Frydman, Goldberg and Phelps' work on "Imperfect Knowledge Economics". Perhaps things aren't as dire for macro as I made it out to be.