How do we want to think about financial crises? We have some idea that there is something unusual going on - something we might see every 15, 20, or 30 years, say, in a given economy. But what is the process that drives a financial crisis? How does the phenomenon get started, and what propagates it?
I'm teaching a second-year PhD course, which is basically financial crisis economics. I gathered together a set of papers, some of which I had seen, and some of which were on conference programs, the key filter being that these papers had to have been somehow inspired by the financial crisis. The list includes two of my own papers, one of which is not quite ready for public consumption (maybe in a month or so).
Here's an idea that struck me in class last Thursday. There are basically two ways to think about financial crises, or the process by which financial factors affect aggregate economic activity. The first is indeterminacy. This is the basic idea behind Diamond and Dybvig's (JPE, 1983) banking model. A bank run, or a panic, is a bad equilibrium. There is a good equilibrium in which the panic does not happen, but there may exist an equilibrium where everyone wants to run to the bank to withdraw his or her deposit, under the self-fulfilling belief that everyone else runs to the bank. The paper by Ennis/Keister on my reading list (and their other recent work) gives you a nice summary of where the Diamond-Dybvig literature went.
The second process potentially driving a financial crisis is amplification - the idea that financial factors can amplify a small shock to the economy and make it a big one. That's the idea in the "financial accelerator" literature, which evolves from early work of mine (JPE 1987), Bernanke-Gertler (1989), and Bernanke-Gertler-Gilchrist. I think the upshot of that literature is that there is not much propagation. However, there's an idea in my JPE '87 paper that Larry Christiano has used empirically, which seems to get some mileage. In a costly-state-verification world (which gives rise to non-contingent debt under some circumstances), more risk will be bad even if economic agents are risk-neutral. Christiano measures the importantance of "risk shocks" for business cycles and finds that the shocks are important in general (and not just in the past 4 years).
What struck me is that my idea of what the real estate bubble was and Jim Bullard's view are quite different. My view is that the bubble was about amplification. A piece of the price of houses is always due to a type of "monetary bubble." Equity in a house is collateral which can be used by the homeowner to borrow; the mortgage on the house can be packaged as a mortgage-backed security, and that security can be used in financial exchange, and as collateral, perhaps multiple times. Thus, through an amplification effect, the housing collateral potentially supports a very large quantity of credit, and that feeds back into housing prices. The financial crisis was about incentive problems that caused the monetary bubble to be larger than was socially optimal, and once financial market participants caught on, that piece of the bubble burst.
Bullard's view is essentially indeterminacy. The real estate bubble was a self-fulfilling good equilibrium, and now we're in a bad one.
The two views get us to the same place. I.e. potential GDP is much smaller than the Old Keynesians are telling us. What you see may be what you get. However, the policy conclusions implied by each view could be quite different.
Addendum: See this paper by Dorofeenko/Lee/Salyer on how risk shocks are propagated through the housing sector.