VoxEU Column Financial Markets

What is a reverse auction?

The Paulson plan envisages that the US Treasury will purchase financial assets held by distressed financial institutions for which there is currently no market. In order to set a price for these toxic assets, reverse auctions have been proposed. As this column explains, one has to be very careful in designing these auctions in the presence of asymmetric information.

One of the most controversial aspects of the recent plan focused on supporting American financial institutions is the possibility that a Treasury agency may buy financial instruments,  in particular so-called Mortgage Backed Securities (MBS), for which there is currently no market. The Treasury states that this purchase can be a good deal for taxpayers, since now these MBS can be bought at a very low price, but they can be expected to appreciate in value as time goes on. Obviously this is only likely to happen if MBS are bought by the Treasury at a sufficiently low price. We do not know yet in detail how the purchase mechanism will be designed. It is obvious though, that if there were to be total discretion in the determination of prices and in the choice of financial institutions to buy from, this would open enormous possibilities of abuse and conflicts of interest. It is therefore advisable to adopt a clear and transparent mechanism for purchases.

On the 23rd of September, in a testimony before the US Senate, Ben Bernanke suggested adopting a method called reverse auctions. What are they? What problems can be generated if they are used to buy Mortgage Backed Securities?

What is a reverse auction?

In a reverse auction the buyer announces an amount he intends to buy and a maximum price. At this point sellers offer the quantities that they are willing to sell at that price. If the quantity for purchase exceeds the quantity the buyer intends to purchase (if supply exceeds demand), the price will be lowered and a new offer will be generated. This practice continues until supply equals demand.

In our case, the buyer is the US Treasury and the quantity to be bought is a certain amount of bonds, say MBS with ten-year maturity and face value of 100 million dollars. At this point potential sellers, i.e. banks that have these securities in their portfolio and want to sell them, announce how many they want to sell at that price. If supply exceeds demand, i.e. in our example the bonds on offer exceed the face value of 100 million, their price will be lowered. Presumably at the lower price, banks will reduce the amount of bonds they wish to sell. If the reduction is sufficient to match supply and demand, the auction will be over and the bonds are sold at the last price announced. Otherwise the price will be lowered yet again, until supply and demand are matched.

This is the simplest design. Another one has the buyer announcing the quantity he wishes to purchase and all potential sellers announcing the prices at which they are willing to tender their bonds. Once the process ends, the buyer purchases from those who announced the lowest prices.

These two designs are not equivalent in terms of final prices that can be obtained, but the choice of design is not particularly relevant in relation to the specific issues discussed in this article.

Possible problems in a reverse auction

Generally when the items on sale are homogeneous and have clearly verifiable characteristics, these auctions generate an efficient outcome for the buyer. Unfortunately this is not the case with MBS, which are financial assets whose cash flow is tied to the repayment of the mortgages on which they are based. Two apparently identical MBS, say with the same maturity and interest rate, can have very different values depending on the solvency of the underlying mortgages. The main problem is the asymmetric information on the solvency and therefore on the cash flow likely to be generated by these mortgage loans: a financial institution that has certain securities in its portfolio, invariably knows better than others how reliable those securities are.

Let us imagine for example, that as a result of the auction, a bank sells securities for 10 million dollars at 50% of the face value. Out of all the securities that can be used to satisfy the buyer's requests, which ones will the bank choose? Obviously those that they know have a lesser chance of being repaid, typically those that are worth less than 50%. The result is a loss for the buyer.

This is exactly the reason why these securities are so illiquid. Indeed, if there was no asymmetric information, private parties would start trading these securities, liquidity issues would disappear and it would be easy to establish their market value. Because of the market’s failure due to asymmetrical information, the Treasury decided to intervene. But quite obviously “you can’t have your cake and eat it”.

Buying securities under these conditions inevitably creates the risk of overpaying

An easy way to eliminate the adverse selection problem is to buy all the securities of a specific category, or almost all of them. Unfortunately, in this case a reverse auction is not a good selling mechanism. As a matter of fact, the higher the demand, the higher the buying price will be. If there is demand for all the securities of a specific category, the selling price will be the opening one and will have to be high enough to induce sellers to get rid of all securities, even the most reliable ones. Therefore, the problem reappears again in a different form.

How to alleviate these problems

In the presence of asymmetric information, it is very difficult to avoid inefficiencies. However, there are ways to alleviate the problems. To begin with, several different auctions are needed for different securities, instead of fewer auctions with rough specifications of the securities for sale.

For example, instead of having a single auction for securities with a 5 year maturity, one could have different ones depending on the date at which mortgages were signed, depending on whether the interest on them is variable or fixed, etc. The more detailed the specification of securities, the lesser the adverse selection problems. Even so, a certain degree of heterogeneity is bound to remain. In a very recent paper on reverse auctions, the economic consultancy firm NERA estimates that there are more than 100,000 different types of MBS. It is obviously impossible to have an auction for every different type, especially if time is limited.

In an article on the Economists’ Voice, Larry Ausubel and Peter Cramton, two influential auction theorists, suggested specifying before the auction that securities bought from the Treasury will be resold in a certain amount of time and that selling banks will be fined if prices are lowered.

For example, let us imagine that the Treasury buys an MBS from a bank with a face value of 100 at a price of 30. If the mortgages behind the MBS do not go in default and payments continue, in a year’s time the MBS will be worth more. If payments, instead, are interrupted, the MBS will be worth less and the Treasury will suspect that the bank knew about the solvency risk of that MBS and played a dirty trick. In this case Ausubel’s and Cramton’s suggestion is for the Treasury to resell the security through another auction and ask the bank to pay at least in part for the price reduction.

Obviously such a mechanism would discourage banks from “being too clever”, since they would be paying more at a later date. The main risk is that this mechanism may be very ineffective with more troubled banks. If there is a high chance of bankruptcy over a short horizon, then the fear of being fined in the future will have a small deterrent effect.

Editors’ note: This column first appeared in Italian on our Consortium Partner’s site www.LaVoce.info.

315 Reads