Thursday, March 26, 2009

A Slight Gap in the Analysis

I finally got around to reading the proposal by Lucian Bebchuk that informed the latest incarnation of the Geithner plan. Recall that a central problem is pricing toxic assets for which no current market exists. The old approach (Paulson) was to have the government simply dictate prices in the course of buying up a few hundred billion dollars of them, presumably to overpay and surreptitiously subsidize the sellers. The new plan is to encourage a multiplicity of private funds, stuffed with mostly public dollars, to bid for the assets. The key paragraph reads:

The existence of such a significant number of private buyers armed with substantial capital will produce a well-functioning market for troubled assets. This will be a market in which many potential sellers (banks) face a significant number of potential buyers (the funds). The profit share captured by the funds’ private managers will provide these managers with powerful incentive to avoid overpaying for troubled assets. At the same time, the profit motive of the selling banks, coupled with the presence of competition among the private funds, will make it difficult for funds to underpay for troubled assets. As a result, we can expect the market for troubled assets to function well, with prices set around the fundamental economic value of purchased troubled assets.

Remember that old Gary Larson cartoon in which two scientists are standing before a blackboard crammed with math? One furrows his brows and says he has doubts about Step 3. Standing apart from all the Greek letters and operators above and below it, Step 3 says, “And then a miracle occurs....”

This paragraph is Bebchuk’s Step 3. With so much tweaking of fund managers’ incentives needed to get them to participate in the program, it is not at all a given that they will maximize expected profits by bidding to the expected value of the assets on offer. In fact, it is easy to show that, the more dispersion there is in their subjective probability distributions around the assets’ expected values, the more distortion there is in price discovery. Paul Krugman picks a maximally dispersed example (all the density at the two extremes) to demonstrate the problem his post from three days ago.

It’s funny how “competition” can take on magical properties for some people. It seems that Bebchuk was so pleased to have found a way to inject competition into the “bad bank” strategy that he didn’t inquire into how well it would perform.


Eleanor said...

The cartoon is by Sidney Harris, and it's wonderful. Since Larson and Harris are the two cartoonists that scientists love, it's easy to confuse them.

Ken Houghton said...

"it is not at all a given that they will maximize expected profits by bidding to the expected value of the assets on offer."

I'm perfectly willing to assume that they would be willing to do that, and that the new E() is higher than the current trading price in the market (ca. 30%).

However, the absurdity of the DeLong position (Thoma made clear that he doesn't expect the result to be Price Discovery, and therefore his advocacy of lighting money on fire is more understandable, if even less reality-based) is the idea that playing with 85% house money will lead to a trading price closer to 80% (where Citi has such securities marked) than 50%, even as the underlying assets are depreciated.

Even Tim Harford has realised that the result will not be Price Discovery in any legitimate sense of the word. So the question for DeLong remains: when this doesn't work, under what delusion will this make George Voinovich the 60th vote--and what rationale makes this a fair price, when we could "bribe" someone else for significantly less—or, in fact, prop up the housing market and make the underlying assets more valuable for the same amount.

If we were talking a difference between 30 and 35, or even 40, I might believe that raping the FDIC could be cost effective. 30 to 80 is one of those gaps where lying to yourself helps no one.