Auctions for Mortgage-Backed Securities

The Emergency Economic Stabilization Act (EESA), which was enacted in October 2008, included a plan to buy mortgage-backed assets from troubled financial institutions. The U.S. government likely would buy those assets through some form of auction. This note discusses the dynamics and the pitfalls associated with various types of auctions. The EESA calls for the Treasury to buy mortgage-backed securities to provide liquidity for troubled financial institutions thus alleviating the credit crisis. The government would later sell the securities, hopefully recapturing most of its initial expenditure. Because the government is buying instead of selling, the auction would be called a reverse auction.

Auctions could be either open or sealed bid. In an open bid auction, bidders continuously submit bids and can see the previous bids of other bidders. This form of auction is more dynamic than a sealed bid auction, where bidders submit their bids simultaneously in a sealed envelope. Dynamic open bid reverse auctions work best when the objects for sale all have the same value to the sellers. They could fail miserably in a case such as this, where the various mortgage-backed securities vary widely in terms of their underlying value.

For example, suppose the government announces it will buy securities with a certain total face value at a certain percentage of their face value. It begins by offering 100% of the face value and determines what volume of securities is offered to it at that price. As very few if any mortgage-backed securities are worth their face value, the government will likely be offered many more securities than it wants to buy. It can then lower the share of the face value it will pay until the volume of securities that it is offered falls to the level that it wants. The problem with this method is that the securities that would be offered for sale would be those with the least value. The government would end up buying only the worst of the mortgage-backed securities.

A partial solution to this problem would be to classify the underlying securities into separate categories and have a separate auction for each category. The government would start the auction at a separate reference price for each category, that price, and then progressively lower the price until the supply equals demand for that category. Determining those reference prices, however, would be very difficult. One possibility would be to run a preliminary auction to determine the reference price.

Because sellers know that the first-stage auction will affect prices in the second-stage auction, however, they may bid strategically to raise price in the first stage. Thus, the second-stage reference prices should be below the first-stage prices. The government could ask bidders to specify the price, given as a percent of face value, at which they are willing to sell their assets and then buy assets with the lowest specified prices until it has acquired the face value of assets that it wants. One question would be whether each successful bidder should receive the price specified in its bid or whether all bidders should receive the same price. While paying a uniform price may seem less desirable, as some successful bidders would have been willing to accept less, it has the advantage of encouraging low bids. Paying uniform prices, however, may allow large bidders to have undue influence on the market clearing price and may increase the chance of collusion. If a uniform price system is used, it may be advisable to open the auctions to a wider range of buyers. If investors are allowed to bid to purchase the securities, their bids would reveal the information they have concerning the value of the securities.

The government is on the verge of spending large sums for mortgage-backed securities. How it designs the auctions through which it acquires those securities will have significant implications for the cost of that program to the taxpayer and perhaps for the success of the stabilization program itself.

Sudip Gupta is a Professor at the University of Maryland and the Indian School of Business, and has done extensive work developing and estimating auction models.