I shouldn't do this, but I think Krugman is wrong in an important way when he describes the run on the shadow banking sector in his review of Geithner's book here:
Worse, runs on shadow banks proved contagious. As investors pulled their funds from shaky firms, those firms were forced into frantic fire sales of assets; such sales depressed the prices of the collateral used by other banks, producing further investor panic.
Why is this wrong? It understates the way the deregulated world of derivatives led to a network of risk that multiplies vulnerability rather than spreading it out.
A falling market didn't lower the market value of assets. The assets were literally guarantees on the assets on other balance sheets. If it were real estate, my neighbor getting a low price for his house means my house has lower value. But what happened in shadow banking was that my asset wasn't a separate house, my asset was a guarantee that the value of my neighbor's house would go up, never down. The collapse wasn't the perception of a run. It was a mechanical process of contractual consequences.
This is important because no matter how liberal or Keynesian you think Krugman might be, he is still a Neoclassical economist par excellance. And Neoclassical economists believe in markets, including markets for risk. The more ways you can go long or short on any economic possibility, the safer the financial system becomes.
The important thing to know about this is that is pure theory. It's truth status is like the existence of the Higgs boson five years ago: either it's true or the theory is wrong.
But here's the thing, just while scientists were finding the Higgs at CERN and providing support for what's called the standard theory of particle physics, the empirical data of the market was disproving the notion that lots of derivatives cause stability. The standard model of economics has been, in part, proven false. And the most famous economist in the world doesn't seem to have noticed!