Thursday, April 14, 2011

Today's comment on Yglesias.

He's tromping all over my turf.

He went out of his way to say he isn't worried about financial arson. He taught me something I should have known (I should have read the article and should have thought of it myself). I could unfairly attack him by ignoring his use of the qualifier "could" in a, just to be polite, qualified concession "to be fair" to those with whom he disagrees. I could pretend that he really believes that that which he concedes "could" be true, definitely is true.

So I will.

Fascinating post. Thanks for telling me about the Gorton article.

I do think that Gorton is trying very hard to be as orthodox as possible. Effectively he assumes that the semi strong form efficient markets hypothesis is true. He notes that, even if CDS prices move only in response to valid information, they might cause problems.

In the real world another issue is that asset prices bounce around for no good reason. This means the cost of speculation in liquid assets include those described by Gorton plus many more. The part of the post which doesn't convince me is "In the long-run, the availability of naked CDS and therefore more information about firms could be a good thing that enhances the operation of the economy." This is only true if CDS prices are based on information and not noise, bubbles, panics, fashion and well all the things which really make asset prices bounce around.

CDS played an important role in the economy before the crisis. Firms designing and rating CDOs used the correlation of different CDS prices to estimate the risk of default of tranches of the CDO. This is using information in CDS prices. It didn't work out well. The "information" was noise and it caused a gross underestimate of the probability of joint default of different underlying bonds (of the correlation of the default events). Now the problem was partly that the same firm designed and rated CDOs creating massive conflict of interests. It was largely that the calculations were performed with the absurd assumption that all stochastic variables are jointly normal.

But the fact remains that use of the alleged information in CDS prices, impaired the operation of the economy. Massively.

Gorton's point is that, even if the semi strong form efficient markets hypotheses were valid, new liquid assets can make things worse. This is an impressive result, but there is no reason to take the semi strong form efficient markets hypothesis seriously as it fits the data about as well as the flat earth hypothesis.

No comments: