Gary Gorton has a piece floating around - I would link to if I weren't so inept - arguing that the financial crisis amounts to a classic bank run in the unregulated shadow banking sector. I found the description of the liquidity transformation that goes on in this sector fascinating, but at first I couldn't quite wrap my head around it. The more I thought about it the more it seemed to me a good example of the core idea of the classic Diamond/Dybvig article on bank runs. Let me explain.
The chief action in the shadow banking sector, Gorton says, looks something like this. Corporate treasurers flush with liquid funds buy securities from investment banks under repurchase agreements. My initial question was: where do the funds to repurchase the securities come from? I think the answer must be: from liquidating the securities. Which raises the further question: why don't the corporate treasurers hold the securities directly and liquidate them themselves? This is where Diamond/Dybvig comes in, I think. (Someone who knows what's really going on out there, I'm happy to be corrected. I'm begging to be, but indulge me a little longer!)
The idea in the article is that there is one asset available which is illiquid. If held for one period it yields a gross return of 1 upon liquidation; if held until maturity (for 2 periods) it yields a gross return of 2. Savers prefer a more liquid asset - one with a higher return if liquidated even at the cost of a smaller return if held for 2 periods. Savers may be of two types. One type wants to consume only in one period; the other type only values consumption in 2 periods. They learn their type only after one period. Each has the same probability of being a type 1 and my chance of being type 1 is independent of yours. With a large enough number of us, there is no uncertainty about the proportion who will be type 1's. So say all of us would maximize expected utility with an asset which gave us 1.28 if we turn out to be type 1's and 1.813 if we turn out to be type 2's. And suppose the probability of being type 1 is 1/4. There are 100 of us endowed with one unit, which is the cost of the asset. A "bank" pools the deposits and buys 100 units of the asset. After 1 period, they liquidate 32 units of the asset for one unit each, allowing them to pay each of the 25 type 1's 1.28 each. The remaining 68 units of the asset mature next year, paying 2 each, allowing them to pay 136/75 = 1.813 to each of the 75 type 2's, as promised. So the "bank" doesn't hold liquid reserves at all - like Gorton's shadow banks as I understand them. Mutatis mutandis: think one-period repos. 75 of the corporate treasurers are happy to renew their purchases after one period, so what the bank owes them is exactly offset by what they owe the bank and no funds move in either direction. The other 25 do not want to renew, so the shadow bank liquidates 32 of the assets.
Does this sound remotely plausible?