I agree with Andrew that as illuminating as Paul Krugman's Sunday New York Times Magazine article was regarding the internecine warfare between "saltwater" and "freshwater" economists, it didn't answer the basic question in the title of the article: "How Did Economists Get It So Wrong?"

One thing you learn if you study advanced physics is that what you observe is crucially dependent upon your point of observation.  The same is true in economics.  I can't fault Paul Krugman for viewing the world from academia.  I would note, however, that the financial market crash of 2008 resulted from Washington policy and financial market decisions that went badly awry.  We deregulated financial markets and promoted derivatives in ways that we believed extinguished risk, when, in fact, we just shifted the risk off balance sheet and pretended it didn't exist.  To our dismay, we proved that markets can consistently ignore financial risk and overvalue housing assets for years, and hardly any economists, except Robert Schiller, noticed.

I've always been amazed at the number of economists who believe their models are superior to real world experience.  They really believe that risk can be continuously measured with a probability of loss and cost of loss functions.  In my experience, lots of discontinuities occur.  Actual probability outcomes seem to defy our distributions, and our loss estimates look ridiculously small after the fact.  When I raise such concerns with top academic economists, I'm usually dismissed out of hand for failing to rigorously model what data that exists and for criticizing that data as inadequate.

Any economist working in a frontline policy job is sooner or later confronted with a hard choice: Do I go with what my models tell me or with what market participants and legislators tell me?  Those who temper their reliance on the models do a lot better and last a lot longer in their jobs.  What was going on in the mortgage market never made it into the models.

Don't get me wrong.  Models are very good for rank ordering some micro decisions for policy makers.  They impose needed consistency in considering policy options.  They can even predict some outcomes well, but we are a long way away from the day when a macro model can predict a sharp turn in the economy.  The world is just too complicated; the underlying assumptions are necessarily too strong; and the vagaries of human nature (Is it time to withdraw all of my money yet? Do I follow the herd?) are way beyond what can be incorporated in current models.  Wall Street risk/reward models made some people rich for a few years before they crashed and dragged the rest of us down with them.