So far, I think that I am the only commentator who has offered a theory of how credit default swaps create systemic risk. I'm not saying that I'm the only person with this theory, but I have only seen it described in my dialog with Will Wilkinson and my primer on systemic risk. The idea probably bears repeating.
A credit default swap is like insurance against default. If you want to buy a municipal bond or a corporate bond but not take default risk, you try to buy a credit default swap. You pay a fee, and in exchange for that fee the seller of the swap will make you whole if the city or corporation defaults.
The seller of swaps collects nice fees, and most of the time the borrowers don't default. But if borrowers do default, then the seller is like an insurance company in a town that was hit by a hurricane.
One strategy for people who sell swaps is to use dynamic hedging, the same approach that was notoriously used to provide portfolio insurance in 1987 and which helped cause the big stock market crash on October 19th of that year. With dynamic hedging, you create a synthetic put option by selling securities on the way down. If you sell fast enough, the money you make on the short sales is enough to offset the cost of making good on the swap.
Suppose I have sold a credit default swap on Sallie Mae. That means that if Sallie Mae defaults on its bonds, I will have to pay some of the bondholders a big chunk of money. One way I can hedge that risk is to sell short Sallie Mae securities. I can sell short their debt, or I can sell short their equity. The closer they are to default, the more I need to sell in order to execute the hedge. However, the more short-selling takes place, the closer they get to default. It is a vicious cycle. Ordinarily, I do not believe that short-selling affects the price, but when there is massive short-selling that is driven by dynamic hedging, I can see where the short selling would drive down prices.
So you could get a vicious cycle: doubts form about a company's soundness; creditors want more protection, so they try to buy credit default swaps; sellers of swaps engage in short-selling to hedge their own exposure; the company's stock and bond prices lose value; creditors get even more worried; etc.
At this point, if the government tries to curb short-selling, all that does is cripple the credit default swap market. It becomes costly to sell default swaps, so now creditors cannot get them at affordable prices. The only way they can reduce exposure is to sell their munis and their corporates and flee to Treasuries. I'm not saying that the curbs on short selling make things worse, but such regulations certainly don't solve the problem--at best, they shift it.
If my theory is correct, then the credit default swap protection is somewhat of a delusion. The contingency plans of individual sellers of swaps cannot be executed collectively. Just when you need to sell short in order to manage your risk, everybody else is trying to do the same thing, and it doesn't work.
It is too late to undo the delusion. In the aggregate, markets under-estimated the risk of the bonds they were buying. The risk premium needs to adjust upward. That upward adjustment is not a credit squeeze--it's a return to reality. Beyond that, there may a credit squeeze, because of the complex interaction among capital requirements, security valuation, investor psychology, and what have you. I don't envy regulators the task of sorting out necessary adjustments from needless panics. However, when in doubt, I lean toward assuming "necessary adjustment."
There are two types of errors that regulators can make. One type of error is to allow a good security to become undervalued or a solvent financial institution to fail. The other type of error is to over-pay for a bad security or to prop up an insolvent institution. The simple way to avoid errors in security valuation is to stay out of that business--don't execute the Paulson plan. As far as institutions are concerned, I think you have to give the FDIC and the Fed the latitude to make those calls, and hope they get as many of them correct as is humanly possible.