TPM Café, March 11, 2010
See article on original website
In his survey of financial innovations over the last four decades Brookings economist Robert Litan bravely comes to the defense of credit default swaps (CDS), the instrument that gained so much notoriety with the collapse of AIG. Litan notes the obvious - that CDS must be better regulated and traded on clearinghouses or exchanges - but argues that properly regulated CDS contracts make important contributions to the economy.
For the moment, let's assume proper regulation (I'm an economist, I can do that) and ask whether CDS contracts really provide a net benefit to society. Litan's case is fairly straightforward: CDS contracts allow banks or other financial institutions to hedge the risk that they incur on their loans. This can free up reserves and allow them to make more loans. It can also provide the same sort of protection for businesses that worry a supplier or other customer may default on their obligations.
Litan also notes that CDS contracts may provide useful information. Since CDS markets move more quickly than bond markets they can more quickly transmit information about potential problems in the finances of companies or countries, as market actors rush to buy CDS contracts against companies or countries that are in trouble.
At the most basic level, Litan's claim is obviously true. CDS contracts do allow for a transfer of risk. But is this necessarily good?
The bank that makes a loan presumably knows its borrower better than the issuer of a CDS contract. Similarly, most companies are likely to know their suppliers or their customers better than the issuer of a CDS contract. On the face of it, there is something problematic about an instrument that is designed to transfer risk from the party better able to assess risk to the party less able to assess risk.
The dealings between Goldman Sachs and AIG perhaps present the most extreme case of this sort of relationship. Goldman Sachs was putting together collaterized debt obligations (CDO) based primarily on mortgage-backed securities, which had large concentrations of subprime loans. Goldman then went to AIG and bought CDS contracts written against the CDOs that it had just created. In effect, Goldman was betting that the CDOs it had created would go bad.
As we know, Goldman won this bet with AIG. AIG lost billions of dollars on the CDS contracts it sold to Goldman, as Goldman correctly guessed that the CDOs that it was issuing would go bad. These losses played a big role in bankrupting AIG. It was only as a result of a government bailout that AIG was able to make good on its CDS contracts with Goldman.
In this case, AIG was willing to issue CDS contracts to Goldman because it did not realize how bad the underlying asset was. Presumably, if AIG had better knowledge of the quality of Goldman's CDOs, it never would have issued the CDS contracts. In other words, if the market were working right, Goldman never would have been able to buy the CDS contracts in question. (There are some important agency issues here. It is likely that the AIG employees who actually sold the CDS contracts to Goldman received large bonuses that they pocketed even though the CDS bets went bad.)
Suppose we have a more typical case where a pension fund buys CDS contracts against bonds it holds or a supplier buys CDS contracts against a company to whom it has extended credit. Of course in the vast majority of cases the CDS contracts expire worthless since companies generally will not default on their debts. It could be said that the CDS provided valuable insurance in these cases, even if the adverse event never materialized.
But can we assume that the decision to buy the CDS was well grounded in actual risks and this was not just a pointless transfer of money to the financial industry? There are well-known fashions in investment. For example, putting money into a hugely over-valued stock market in 1999 did not get investment managers fired when it crashed the following year. The reason was that this investment was fashionable - everyone made the same stupid mistake so no one could be held responsible. The same was true for holding housing-related assets in the bubble years.
Suppose that it becomes fashionable to hold CDS contracts against risks that funds and companies would have gladly accepted in prior years. It may not be wealth maximizing for funds or profit maximizing for companies, but for the individuals making the decision, buying CDS contracts could be a safe alternative compared with having to explain the unlikely event where they lost money due to a default.
In this event, CDS contracts are leading companies to over-insure; just like the person who buys service contracts on all the appliances he buys and then never makes use of them. If this is the case, then it means that more income would accrue to the financial sector, and less to everyone else - sort of like what we have seen in recent decades as the financial sector now takes up more than 30 percent of all corporate profits.
In short, CDS contracts do allow for the transfer of risk, but the transfer goes from the party better able to assess risk to the party less able. Transfers in this direction are not generally consistent with economic efficiency. It is far from clear that whatever benefit the buyer of the CDS contract might get from this deal outweighs the cost.
The second point in favor of CDS contracts raised by Litan is that they provide more information to markets about the health of companies or countries. The CDS markets tend to be more liquid and also more volatile than the underlying bond markets since by definition they are heavily leveraged. (The nominal value of CDS contracts is typically more than an order of magnitude larger than their market value.) This means that that CDS markets are likely to respond more quickly to information. However, it is certainly possible that they are often over-responding, effectively reflecting swings in investor sentiment that may not correspond to economic fundamentals. If this is the case, then the CDS markets are not providing information, but rather amplifying fluctuations.
An example of a case where CDS markets were almost certainly not providing real information is with CDS contracts issued on U.S. Treasury bonds. The increase in the price of these CDS contracts last year was seen as providing evidence that investors were losing confidence in the ability of the U.S. government to pay its debts. In fact, many political figures touted the rise in the price of these CDS contracts to advance their agenda for reducing the budget deficit.
This was incredibly sloppy analysis. Any CDS contract effectively involves two bets. First, it is a bet that the company or country against whom the contract is written will default. The second bet is that the party who issued the contract will be able to pay off the contract if the default takes place. If the issuer goes down along with the company or country, then the holder of the CDS contract has lost their money.
If the U.S. government defaults on its debt, almost by definition there will be a massive worldwide financial crisis. In this sort of crisis, there will be few, if any, financial institutions that would be in a position to honor their CDS contracts on U.S. Treasury bonds. In other words, if someone really thought that the U.S. government would default on their debt, they would be very foolish to throw their money away buying CDS contracts instead of tangible assets like land or gold. It is not clear what information the price of CDS contracts on U.S. Treasury bonds is conveying, but it almost certainly has nothing to do with the risk of a default by the U.S. government.
This brings up to the question of proper regulation of CDS contracts. Litan concedes the need to have proper regulation, which includes having most CDS contracts be traded through clearinghouses. This would lead to greater transparency in pricing and also ensure that issuers had adequate capital to support their contracts.
This is easy to say, but in practice can we count on regulators effectively policing CDS contracts in future years? The AIG story was not just an oversight; there was an explicit effort by Congress and the Clinton Administration to protect CDS contracts from oversight in order to foster the growth of the market. Is there any reason to believe that future administrations and congresses will be less susceptible to pressure from the financial industry? (There is also the issue that none of the current bills being considered in Congress would require non-standard CDS contracts to be traded on clearinghouses or exchanges.)
It is also worth noting that at the core of AIG's mistake was the failure to recognize a housing bubble. This housing bubble meant that an event never seen before - a nationwide decline in nominal housing prices - was about to occur. This is worth noting because even if CDS contracts had been regulated, it is entirely possible that the regulator would have made the same mistake in failing to recognize the housing bubble as did so many other actors in financial markets. In this case, they may have judged AIG as possessing adequate capital to cover its potential losses.
A second issue that Litan raises in passing is the misaligned incentives that CDS contracts can create in a bankruptcy situation. In a typical bankruptcy, creditors will negotiate terms that involve substantially less than full payment on their loans to the bankrupt company. However, CDS contracts give a creditor little incentive to negotiate for less than full payment, since they can get paid by the issuer of the CDS in the event of a default.
While Litan suggests a fix to this problem it may not be easily implemented in practice precisely because the market for CDS is so liquid. For example, an investor holding a substantial amount of the debt of a company facing bankruptcy could buy a large amount of CDS contracts on that company. The investor can then take a very tough stand in negotiations with the company. This would increase the likelihood of bankruptcy and also the value of the CDS contracts. The investor can then sell off the CDS contracts at a profit and later agree to terms with the company.
This is a case that would be comparable to buying insurance on your neighbor's house and then setting it on fire. Regulators should not allow this to happen, but does anyone have much confidence that they could prevent it?
In short, while in some circumstances CDS contracts can provide some benefit, it is likely to be relatively limited. In general it is not very costly to sell bonds or even loans that involve more risk than an investor or company is prepared to incur. It is not clear that the additional risks posed by CDS contracts, especially when they are improperly regulated, are worth the limited benefit they provide.
If it is too late to put the genie back in the bottle by banning CDS contracts, there would seem to be a very good case for a modest financial transactions tax (e.g. 0.02 percent of value per transaction) on trades of CDS contracts. As noted above, CDS contracts will often involve bets on the future of companies or countries. Efforts to ban such bets are likely to prove futile, but there is no reason that we can't tax this gambling just as we do when people place bets at Las Vegas or Atlantic City. In cases where CDS contracts do provide real economic benefit, such a modest tax will not deter actors from buying them; however, it may reduce some of the gambling taking place in this market. Such a tax can help to ensure that the financial industry is not the only winner from this innovation.