How to build a better auction?

Though the bulk of the attention today has understandably turned to the U.S. Treasury announcement that $250 billion will be devoted to a U.S. bank recapitalization plan, there does still remain the notion that some of the government’s rescue funds will be devoted to purchasing troubled assets through some sort of auction mechanism. At least that’s the way it is according to Neel Kashkari, the Treasury’s point man on implementing the provisions of the Emergency Economic Stabilization Act of 2008. From Bloomberg:

“In addition to the stock-buying effort, other components of TARP include a whole loan purchase program, a mortgage-backed securities purchase program and an insurance program for those securities.

“[Kashkari] outlined three possible scenarios: ‘One, an auction purchase of troubled assets; two, a broad equity or direct purchase program; and three, a case of an intervention to prevent the impending failure of a systemically significant institution,’ he said.”

Thus, one of the more interesting logistical questions of the Treasury plan remains: How to design an auction that will generate “efficient” prices for assets that the government might purchase. A few days ago, Greg Mankiw linked to a proposal from University of Maryland professors Laurence Ausubel and Peter Cramton, which had earlier been noted by Felix Salmon. The key elements of the Ausebel and Cramton plan:

“An auction that determines a real price for a given security needs to require multiple holders of the security to compete with one another. This can be achieved if the Treasury purchases only some, not all, of any given security.

“Thus, a better approach [than a simple reverse auction for all eligible securities] would be for the Treasury to instead conduct a separate auction for each security and limit itself to buying perhaps 50% of the aggregate face value. Again, the auction starts at a high price and works its way down. If the security clears at 30 cents on the dollar, this means that the holders value it at 30 cents on the dollar. (If the value were only 15 cents, then most holders would supply 100% of their securities to be purchased at 30 cents, and the price would be pushed lower.) The auction then works as intended. The price is reasonably close to value. The ‘winners’ are the bidders who value the asset the least and value liquidity the most.”

Writing in Slate, Steve Landsburg offers up what I think is the same idea:

“Here's (roughly) how a ‘Bils-Kremer’ auction would work: First, put 10 similar distressed assets (such as a series of collateralized debt obligations) up for auction. At the close of the auction, the Treasury pays the winning bids for nine of these properties. The 10th property (chosen randomly) gets sold to the winning bidder.

“The advantage of a Bils-Kremer auction is that the Treasury buys assets and recapitalizes the firms holding those assets while paying only what some private bidder thought each property was worth. Now repeat with 10 more properties. And so on. Under this plan, nine-tenths of the liquidity comes from the Treasury, but ten-tenths of the price setting comes from the assessments of private investors with the incentive to bid judiciously. In other words, the prices can reasonably be considered fair.”

Anyone have a better option?

By David Altig, senior vice president and research director of the Federal Reserve Bank of Atlanta