The Rodney Dangerfields of financial instruments, credit default swaps, are now getting disrespected for impeding workouts of companies teetering on the edge of bankruptcy. This argument was first flogged by UT law professor Henry Hu last year. Now, lawyers in General Growth Properties and AbitibiBowater bankruptcies are blaming debtholders who hedged by buying CDS protection on these companies with having little interest in achieving a workout. Several bloggers, including the estimable Megan McCardle and the very good (but not quite so estimable, IMHO) Felix Salmon have expressed similar concerns. The somewhat supercilious, but usually provocative, Charles Davi takes the opposite view.
I come down on Davi’s side, but for very different reasons. This is essentially an argument about the effect of the distribution of property rights/ownership. Those who claim that there are perverse effects of CDS argue that the distribution of ownership has effects on the allocation of resources. There will be a different allocation if some debt is owned by those who also have CDS positions, than if all debt is owned by the unhedged. In particular, outcomes will be less efficient, in this view, if some debts are owned by firms that are hedged.
Well, the proper way to begin any analysis of the effects of the original distribution of ownership is with Coase, and the Coase Theorem. This states, of course, that in the absence of transactions costs, the initial distribution of ownership has NO effect on the ultimate allocation of resources. This result obtains because without transactions costs, the original owners can negotiate mutually beneficial trades that put the resources in their highest value uses. Indeed, they will trade until the final allocation of ownership maximizes value, regardless of the initial allocation.
This means that any “problem” with the distribution of ownership is a transactions cost problem. This leads one to (a) attempt to identify the sources of transactions costs, and (b) determine whether there is any change in laws, regulations, etc., that could reduce transactions costs.
In the CDS context, this means that in the frictionless Coasean world, the “empty creditor problem” that arises when creditors are hedged is in fact an “empty problem”–that is, not a problem at all. It is therefore necessary to evaluate, and do so more carefully than those I have seen weigh in on this subject, whether transactions costs seriously impede value maximizing trades, and if so, what are the most efficacious means of reducing these costs.
Here’s my argument in more detail:
A completely hedged debtholder (i.e., a party that has purchased protection in an amount equal to his holdings) would appear to have no incentive to behave constructively during the reorganization. But this means that a potential source of valueâ€”the control and legal rights attached to the ownership of the debt securityâ€”is not being maximized. There is therefore a set of mutually beneficial transactions that would leave the insured debtholder no worse off than if he acts in the passive manner that the critics of CDS hedgers suggest, but would result in the efficient exercise of the control and legal rights. Specifically, another party could purchase the debt from the insured current holder at a price that is satisfactory to the latter, but which allows the purchaser to capture some of value associated with the control rights. That is, there is a set of transactions that would allow the hedged debtholder to capture some of the benefits of the control rights.
The anti-CDS analysis presumes that there is money left on the table. The total value of the debt and insurance contracts are higher if the control rights are efficiently exercised. This creates an economic incentive to structure transactions to ensure that they are so exercised.
The credit protection that hedgers have is not (from a social perspective) an economic asset. It has zero value because the gain to the buyer is offset penny for penny by a loss to the seller. This suggests another value enhancing transaction. The protection seller and the protection buyer could agree to cancel the contract at a mutually acceptable price that effectively splits the value associated with the control rights between them. This would be mutually beneficial even if the debtholder is overhedged, i.e., has bought more protection than he holds in the debt.
Another alternative is for the protection seller to buy back the protection and the debt. The “empty creditor” problem goes away because there is no hedge once this transaction is completed. The risk profile of the buyers and the sellers is unchanged. The formerly hedged creditor had no risk exposure after the trade because he was hedged; he has no exposure afterwards because both positions are closed. (The former statement is something of a simplification. The “hedged” firm still faces basis risk–a not immaterial consideration, especially of late. Thus, risk exposure actually goes down for this firm.) The former protection seller’s risk exposure doesn’t change either. He was long the debt (via the CDS position) before the trade; he is long the debt itself afterwards.
In other words, CDS critics essentially overlook the Coase theorem, and assume that the allocation of rights is somehow locked in, and cannot be changed by mutually profitable value-enhancing trades.
Of course, transactions costs are not zero in the real world. It may be costly to find the insured holder of the debt (although insurance sellers could always find their counterparties and commence a negotiation.) More substantively, different parties may have differentâ€”and privateâ€”information about the value of the control rights. Similarly, absent some disclosure requirements, the positions in the debt and protection of each market participant are private information. This private information can create transactions costs that impede the transfer of the valuable control rights to those who can exercise them most efficiently.
The CDS critics focus on potential problems in negotiations in bankruptcy, or in pre-bankruptcy negotiations between debtors and creditors, and among creditors. This is a myopic view. That lawyers (pace the FT piece) advance it is not a surprise, in some sense, because that’s their world. It is a myopic view because there are many other potential transactions; between protection sellers, buyers, creditors, and third parties that can reallocate the relevant control rights and mitigate the allegedly perverse incentive effects of hedged positions. If hedges destroy value, by impairing incentives, greedy market participants have an incentive to unwind the hedges to capture that value that would otherwise be destroyed. These transactions can involve the CDS, the underlying debts, or both. The world of possible value enhancing transactions does not begin or end at the law firm door, or the steps of the bankruptcy court.
Thus, at root, as is the case with any asserted inefficiency, the anti-CDS argument is implicitly one about transactions costs. The anti-CDS argument holds only if transactions costs materially impede the consummation of value enhancing bargains, but those advancing this argument undertake no analysis of transactions costs, or of ways to reduce them.
To take the skeptics’ view at face value (no pun intended), is to assume that transactions costs are very high indeed. For the skeptics argue that large amounts of value are destroyed by the empty creditor problem, it must be the case that the transactions costs of circumventing the problem are even larger. Given the presence of large, sophisticated market participants, this seems something of a stretch. And indeed, it’s not just the original holders of the debt and CDS that are potential players here. Hedge funds or vulture funds or others could realize the value opportunity, and structure trades that maximize the value of the control rights.
The fact that a relatively simple trade–the reversal of an earlier CDS trade, combined with the simultaneous purchase of debt–could eliminate the empty creditor problem suggests that this is much ado about nothing. The transactions costs of reversing an earlier transaction, in what is effectively an “EFP” (“exchange for physicals,” a commonplace derivatives trade), are not all that high. Certainly not that much higher–if at all–than the original trade.
Even if my conjecture that the empty creditor problem isn’t that big a deal, or somebody (or many somebodies) would have engaged in trades to make it go away is incorrect, the debate over how to deal with the issue has been pretty much off point. It has a sort of “woe is me what are we going to do with these awful CDS” flavor to it. The more high strung among the critics suggest that trading in CDS should be severely constrained.
A more constructive approach is to identify the sources of transactions costs that are interfering with the efficient reallocation of control rights, and to devise policies to reduce these costs. (And again, I would emphasize that greedy financial innovators have strong incentives to figure out ways of doing this all by themselves. If they aren’t doing it . . . maybe the problem isn’t that serious. Especially since a vanilla EFP will do.)
This suggests some policy responses that would not entail substantial impediments to (and perhaps the banning) the trading in credit derivatives and the underlying credits. One policy response is to reduce transactions costs that would arise when a company approaches financial distress, or perhaps declares bankruptcy. One way to reduce transactions costs would be to reduce the importance of information asymmetries about positions. Like a poker player, someone who knows his positions and knows that nobody else knows them, can bluff and negotiate in a way that impedes efficient transactions. This inefficiency can be eliminated, and efficient trading encouraged, by requiring revelation of all positions in the underlying debt, credit protection, and equity of the troubled company at the time of a credit event. This would facilitate value enhancing transactions between credit protection buyers and sellers by making it easy to identify the economic interests of all parties, and eliminating the ability to bluff about positions and interests in an attempt to extract value from other participants.
Another policy response is to undermine the bargaining power of hedged debtholders, or to remove those with weak or poor incentives from the reorganization process. This could be done in bankruptcy by reducing the voting and representation rights of debtholders by the amount of the protection they have purchased. For instance, the holder of $50 million in debt who has purchased credit protection on $25 million, would be treated equivalently to the holder of $25 million of the debt (and the total holdings used to apportion voting and representation rights would be correspondingly reduced.)
The mechanics of settlement of credit derivatives could also be designed to mitigate the perverse incentives that concern the CDS skeptics. For instance, if payoffs to a credit derivative are determined very soon after the formal declaration of a default or bankruptcy, but before the completion of a reorganization, the concern evaporates. An ISDA auction soon after a default or bankruptcy would establish prices at which all outstanding credit derivatives on the defaulting firm are settled. Once this price is determined, the value of outstanding credit derivatives is fixed, and consequently indifference or strategic behavior by the buyers of protection is no longer an issue. The derivatives effectively disappear, and all that is left is the underlying debt securities, along with their associated control and legal rights, uncompromised by conflicting incentives created by derivatives instruments. The debtholders then have an incentive to exercise control and legal rights efficiently.*
In summary, Hu and others have identified a potential concern about the adverse incentive effects of credit derivatives, but their analysis is incomplete. It overlooks an important consideration that should inform any analysis of an asserted inefficiency; namely, the fact that regardless of the original distribution of economic rights, self-interested market participants have an incentive to engage in transactions to allocate these rights to maximize their value. The skeptics’ analysis presumes that individually and collectively, market participants are not maximizing the size of the economic pieâ€”the value of a firm undergoing a reorganization. This can only happen if transactions costs materially impede the efficient reallocation of these rights, but Hu and the other critics do not analyzeâ€”or even mentionâ€”transactions costs, or the potential for consenting adults to make trades that enhance value. Moreover, since transactions costs may exist in the real world, policy responses designed to reduce these transactions costs deserve serious attention before more draconian restrictions on credit derivatives are contemplated.
Finally, it should be noted that even if credit derivatives do result in some efficiency loss because of their effects on the incentives of debtholders during reorganizations, or the transactions costs that perpetuate these incentives, these costs must be weighed against the benefits of credit derivatives. Credit derivatives also facilitate the efficient allocation of risk among market participants. Moreover, credit derivatives play an increasingly important role in price discovery and enhancing the transparency of the credit markets. These benefits are appreciable, and the phenomenal growth of these instruments is a testament to the magnitude of these benefits. Any policy response to the problem that the skeptics identify should be crafted so as to preserve these benefits. Thus, a light-touch approach that focuses on reducing transactions costs or modifying the bankruptcy process would mitigate the importance of the CDS skeptics’ concern, without jeopardizing the risk management and price transparency benefits of credit derivatives.
* This mechanism would not facilitate pre-packaged bankruptcies, or other negotiations between financially troubled corporations and their creditors. These negotiations would be facilitated by reducing transactions costs as discussed above.