Streetwise Professor

It is apparent that the decisive factor that drove the Treasury to intervene with AIG is the insurer’s immense position in credit derivatives, notably credit derivatives tied to various mortgage securities. Whereas it seems that in the case of Lehman’s slow motion implosion, market participants had the opportunity to move positions and thereby limit the potential for a cascade of knock-on defaults on derivatives arising from a Lehman default, AIG’s much more rapid plunge permitted no such luxury. Fearing that AIG defaults would trigger defaults or bankruptcies by its counterparties (either because they held offsetting positions with others, or because the loss of the insurance contracts would blow holes in balance sheets large enough to force bankruptcy), the government decided to step in and effectively guarantee AIG’s performance on all its obligations.

The wreck of AIG hard on the heels of the collapse of Lehman hard on the heels of the disappearance of Bear has led to repeated calls for the establishment of a clearinghouse–a central counterparty (“CCP”)–for OTC derivatives, including credit derivatives. In this post I will attempt to discuss some of the issues this possibility raise. Bottom line: a CCP could help a lot, but it is not a panacea. Moreover, the benefits of a CCP depend crucially on the answer to the question: Why hasn’t the market already created a CCP?

What does a CCP do? It insures default/performance risk by mutualization, i.e., sharing of these risks across market participants. Note that in the days (or should I say day) before the AIG collapse, the Treasury essentially tried to create such a mutualization mechanism on the fly when it attempted to coax/coerce banks into lending to AIG. Not surprisingly, this effort failed, and in the event, the government stepped in as the ultimate insurer–meaning that taxpayers are essentially bearing the risk.

One way a CCP would have definitely helped is through the netting of offsetting exposures. In the OTC market, as currently constituted, when A sells a contract to B who sells to C who sells to D, (a) B and C have no market risk exposure (because they have offsetting positions), but (b) there are four sources of performance risk; default by A, B, C, or D is a possibility. For instance, B could suffer a loss on another position that impaired its ability to perform on its contract with either A or C. With clearing, or with some sort of mandatory netting (“ringout”) arrangement, B and C would be taken out of the chain of contracts, and only A and D would pose and face performance risk. B and C could blowup and neither A nor D would care.

In brief, central clearing reduces overall credit and performance risk by reducing the overall magnitude of performance risk exposure. Without netting, performance risk is roughly related to gross positions in the market, whereas with netting, performance risk is reduced to the net exposure.

The CCP also centralizes position information, making it easier for somebody (e.g., the Fed, Treasury) to figure out market exposures. I can only imagine the nightmare that regulators lived trying to figure out who traded how much with whom, and then assembling the jigsaw puzzle to visualize the potential for systemic problems. This is especially true given the dodgy nature of documentation in the credit derivatives market. Although participants have supposedly made progress cleaning up back office problems, these problems haven’t disappeared. This would represent another benefit of a CCP–it would force a rationalization of the confirmation process.

Now let’s figure out how things would have worked out in the AIG case had a CCP been in place. I presume that AIG had a big net position (as otherwise it wouldn’t have been facing multi-billion mark-to-market losses.) So, to the extent its positions weren’t fully collateralized (and I return to the collateral issue below), AIG would have failed to the clearinghouse. These losses would have been covered, to the extent possible, by funds precommited by CCP members in the form of a default fund or its equivalent. In addition, most CCPs have credit lines with banks. If the default fund was inadequate to cover the value of the AIG’s defaulted positions, the CCP would have been able to call on these credit lines, which would have become liabilities of the remaining, solvent members of the CCP. Through this mechanism, or through the CCP’s ability to require members to make additional capital contributions, the losses would have been shared by the other CCP members. A member’s share would be, to a first approximation, an increasing function of the size of the participant and its positions.

Thus, the effects of the CCP would be to diffuse the risk of a single default among all of the members of the CCP. It is likely–though I doubt it can be proven mathematically–that this would reduce the likelihood of the default causing a cascade of knock on defaults. Under the current, non-cleared system, the exposure to an AIG default is related to the direct and indirect gross exposures of each trader in the market to AIG. A firm with a big direct, or even indirect, exposure to AIG, would suffer a big loss from the latter’s default. That big loser would be more likely to be unable to perform on its other obligations than an otherwise identical firm with a smaller gross exposure to AIG. Under a CCP, there would be less dispersion across firms in the exposure to AIG. Thus, under most circumstances it appears less likely that one firm would experience a disproportionate loss from an AIG default.

Here’s a simple example that may capture some of what I am trying to convey. A and B are sellers of CDO insurance. C and D are buyers. C and D have portfolios of CDOs that they have financed with debt and equity. To normalize things, let’s assume that each insurance buyer’s debt is 90 and equity is 10. To make it simple, to emphasize the concentration effect, assume that C buys only from A, and D buys from B. A defaults. This occurs when B has suffered a big loss on his CDO position–let’s say that the CDOs have fallen in value to 50, and instead of paying the 50 it owes, A only pays 20. B no longer has insurance over its entire position. It now a CDO worth 50, and a payment of 20 from A–it thus can only pay 70 of its debt of 90. It defaults on the loan. There is a knockon effect.

Now lets say that the default of 30 is shared among B, C, and D in equal proportions. (NB: In reality, sharing rules are much more complicated.) Let’s say too that B is sufficiently capitalized to be able to cover its insurance obligation and its 1/3 share of the 30 default. Here, C and D receive (on net) 40 of the 50 that they are owed, leaving them with assets of 90 and liabilities of 90. They don’t default. Under the CCP, B and D bear some of the loss that C bears alone with no clearing. This reduces the probability that C will default.

Now, of course exposures are endogenous. In my example, C is a fool for having put all his eggs in one basket. Therefore, you would expect traders to consciously attempt to trade with more counterparties and limit their concentration in the absence of a CCP. Nonetheless, various considerations related to asymmetric information–something I will discuss more below–may in fact induce concentration.

It should also be noted that clearing economizes on collateral. Collateral is related to gross positions in the absence of clearing; with clearing they are related to net positions. This can reduce demands for liquidity during periods of systemic stress. Firms will face smaller collateral calls, and thus have less need for credit or to liquidate assets during periods of market stress.

So if clearing is so great, why hasn’t it happened already? On the answer to that question turns the policy recommendation.

One reason is that there is a coordination or externality problem that means that although it is collectively rational to form a CCP, it is individually irrational, or that the costs of coordinating the creation of a risk sharing arrangement are too high.

For instance, I have argued that some market participants (especially big, highly creditworthy players–like AIG before its collapse’-) may oppose the formation of a clearinghouse because a CCP levels the credit playing field. The biggest, most creditworthy firms have a competitive advantage in a non-cleared OTC market, and a CCP might undermine this advantage. The big guys may therefore not want to participate. And if they don’t play, the clearinghouse will never get off the ground.

Moreover, to the extent that a systemic event caused by a chain of defaults has costs that affect parties other than the intermediaries–a plausible hypothesis–the contracting parties do not internalize all of the costs of their decisions. They therefore do not fully internalize the benefits of the formation of a CCP–and hence may not agree to its creation.

These arguments suggest that the mandatory creation of a CCP would be welfare improving.

There are arguments, however, that suggest that there are substantial costs to the formation of a CCP. In particular, asymmetric information makes it costly to share risks. To the extent that participants in a CCP have private information about their own creditworthiness and the riskiness of the positions they clear, moral hazard and adverse selection problems will occur. These are real costs of risk sharing. The greater these costs, the smaller the benefit of forming a risk sharing arrangement.

As I have argued in a working paper, these costs are greater, the more difficult it is to value the instruments being cleared. Valuation is relatively easy in instruments that are standardized and traded on transparent markets with nearly continuously available, competitively determined prices. Valuation is much more difficult for idiosyncratic instruments that are infrequently traded, and/or which are traded in opaque markets.

The former conditions describe centralized exchange markets for liquid instruments, like T-bond futures. The latter, alas, describes many OTC derivatives, especially credit derivatives, and especially credit derivatives on CDO structures. Hence, it may well be the case that the reluctance to move to a CCP system for more complex instruments is due to the fact that the costs of sharing default risk are higher for these instruments.

Of course, these explanations–coordination/externality vs. high cost–are not mutually exclusive. If I had to make a gut call, I would say that the coordination/externality problems are sufficiently acute that some regulatory effort may be warranted to require the formation of a CCP system. At the same time, any such regulation should take into account the moral hazard and adverse selection problems that are inherent in more complicated instruments, and therefore recognize that risks cannot be shared as extensively for such instruments as is the case for vanilla products traded in transparent markets. This regulation should also recognize that pricing information, transparency, and valuation expertise are required to mitigate the costs of risk sharing for more complex instruments. Therefore, the creation of a clearinghouse might also require measures to improve price transparency and share valuation expertise. This raises other possible sources of resistance. Individual traders profit from private information on prices, and hence are reluctant to share this information. Thus, formation of clearing may require the mandating of disclosure of transaction pricing information. Good luck with that.

In brief, the AIG mess would probably haven’t been as messy if a CCP had been in place. Moreover, it is possible that the absence of a CCP is due to a market failure. However, it should also be recognized that clearing of the kind of stuff that got AIG into trouble ain’t easy–it ain’t like clearing T-notes or even OTC vanilla swaps. It is almost certainly the case that it is uneconomic to share performance risks as widely and extensively in more exotic credit and mortgage products as is the case for vanilla, exchange traded stuff.

I should note that this post was written more hastily than I usually like, but I feel that the issue is sufficiently important and timely to justify a more hurried take. If, upon further consideration, I decide that this analysis is incomplete or incorrect, I’ll make the necessary changes.