Thursday, September 25, 2008

At what price bailout?

Treasury Secretary Paulson's bank bailout proposal has now been subject to public scrutiny for a week. Most analysts agree that the bailout is necessary, but many have questions about the specific proposal.

Alan Meltzer, an economist at Carnegie-Mellon, argued against any bailout at all when he appeared on The Newshour this week. If a bailout is necessary, then he suggested mimicking what Chile did in 1982 in response to a financial crisis. At that time the Chilean government offered up loans with interest to distressed companies, and the companies were required to pay back the loan before they could pay out any bonuses or dividends. Most repaid the loans quickly. If applied to the current crisis, then it would help banks deal with short-term liquidity and punish shareholders and bonus-deprived managers until the loans were repaid. If the company were to fail, the government would have first claim on any assets in bankruptcy. Taxpayers would not put up free money, and it would only help banks with short-term liquidity problems, not banks that are doomed to fail anyway.

Paul Krugman provides an excellent discussion of the shortcomings and implicit purpose of Paulson's plan: it only helps markets if the Treasury buys distressed assets at prices higher than current market prices. It's a direct subsidy to holders of securities that are likely to have high default rates. Taxpayer money saves the banks that hold the debt but taxpayers get no equity in the banks in return. It's free money for the banks.

Many financiers and analysts have estimated that private investors are only willing to pay 20% to 30% of face value for the debt right now, and that the debt would probably be worth anywhere from 50% to 65% of face value if held to maturity. That could leave a spread of 30% or so between current market prices and the estimates of fair value. These analysts argue the current credit shortage and a short-term excessive amount of risk aversion are causing the securities to sell for far less than their actual worth. If the default rates on the securities' underlying loans are as low as projected, or if the credit market loosens up in the next couple of years, then these securities would trade at a higher price relative to face value. The gap between the estimated fair value and the current market value would create an opportunity for Treasury and the banks owning the securities to deal and benefit both parties. But how much would Treasury pay, and what assurance is there that Treasury will pay a low enough price to make taxpayers gain? Paul Samuelson has a good discussion of this in the Washington Post.

Paulson proposed a reverse auction, which, as I understand its application here, might work something like the following. Treasury would announce that it intends to buy $10B of mortgage-backed securities from banks on a given date. (It would be best to start small to get the market going and provide price information for future sales. Treasury could buy larger amounts on a weekly basis over a several month period.) Holders of the securities could offer up, say $1B for a price of 40% of face value, or $1.7B at 43% of face value, etc. Treasury would spend the $10B on the securities offered at the lowest percentage of face value; i.e., at the deepest discount to face value.

Even with this format there remains a problem. A reverse auction can work if all the securities being offered are of similar quality, but not all debt is equal. Imagine soliciting offers from car dealers trying to sell you a new car. You are given five envelopes labeled dealer #1, dealer #2, etc., each with a price. But you don't know whether the price applies to a Chevy, a Lexus, or a Yugo. The lowest price probably applies to the lowest quality product, and might not be much of a bargain.

It's hard to know the value of these securities. Each security could represent thousands of mortgages from different housing markets and different brokers. Some bundles could be full of NINJAs (I love this acronym) - No Income, No Job or Assets; i.e., mortgages issued without documentation of income or assets. Others could have been sold by brokers that screened applicants more diligently. Some bundles could include lots of loans from depressed housing markets like South Florida, Las Vegas, and California with high default rates while others come from North Carolina where housing prices have not seen large declines and where default rates are much lower. It seems that the documentation on the securities can be as poor as the documentation on income, so that it would require great effort to assess the quality of the mortgages backing up any given security. The securities are further complicated because they don't just bundle individual loans directly in their entirety but slice them up and combine them into complex instruments. An article in yesterday's NY Times analyzes a set of bonds issued by Bear Sterns. Over 2800 mortgages with a combined value of $1.3B were assembled into 37 different types of securities with varying degrees of risk. The ratings from the debt-rating companies have not been very helpful for these products.

Paulson's proposal gave all the decision-making authority on purchases (and subsequent resale of the securities in future years) to the Secretary of the Treasury, with no review by Congress or the courts. Constitutional government is based on trust in laws, not trust in government officials. Paulson's proposal goes completely against this idea, and the Bush administration has demonstrated amply why lack of oversight is a bad idea. In recent days several different oversight mechanisms have been proposed. It sounds like legislation might require that the government take some kind of equity or warrant, but even that requires trust in the government administrator to negotiate a favorable price for taxpayers to take on a given amount of risk and equity.

I would propose instead to structure the sale procedure to rely on a mechanism that promotes honesty by the debt sellers. Warren Buffet and others have said it's possible for the Treasury to earn a return of 7% or more per year on these securities, but that depends on what the purchase price is. How about using the reverse auction mechanism, with a guarantee from the seller that the Treasury will earn a minimum return? Each deal should guarantee that Treasury makes at least, say a 6% annual return on the purchased security. If Treasury ultimately resells the security for a price less than that or holds it to maturity and the default rate is so high it fails to receive that 6% return, then the original seller of the equity must make up the shortfall plus a 25% penalty, paid with company stock. Any company that hasn't gone bankrupt in the interim will do whatever it can to pay back the difference to avoid the penalty. If it can't, then Treasury gets equity in the company that exceeds the value of the shortfall. With a rule like this, the sellers of the debt will put a check on the government administrator (whether the Secretary of the Treasury or the head of some independent agency) from overpaying for the securities. I'd feel a lot better giving a government agency greater independence selling under rules like that.