|Financial Services Outlook logo 130|
In the past few months, as the administration has rolled out successive portions of its regulatory reform proposal, it has headlined each new section with the statement, "The Administration's Regulatory Reform Agenda Moves Forward." As its plans have been drawn and quartered even by Democrats on the Hill, these statements have begun to look more aspirational than descriptive. The final piece of the plan, the regulatory regime for derivatives, was issued in early August with a new headline: "The Administration's Regulatory Reform Agenda Reaches New Milestone." Again, the statement has a Monty Python-like ring, as though there were some old milestone we all missed. It turns out that the administration's reasons for imposing new regulations on the CDS market are as faulty as the policy underpinnings of its other regulatory proposals.
In a previous Financial Services Outlook, I described how credit default swaps actually work and why they are important risk management tools for financial institutions and others. That Outlook also addressed some of the myths and misinformation that have grown up around these instruments. In this Outlook, I will discuss the key elements that underlie the administration's proposal, and especially its assumption that the regulation of derivative markets is necessary to prevent a systemic breakdown in the future.
The Administration's Proposal
The administration's draft legislation would require all "standardized swaps" to be cleared by a "registered derivatives clearing organization"--a clearinghouse--and, where possible, traded on a "designated contract market." The definition of a "standardized swap" would be left to a regulation jointly adopted by the Securities and Exchange Commission (SEC) and the Commodity Futures Trading Commission (CFTC). Swap transactions that were not "standardized" would continue to trade in the over-the-counter (OTC) market. Clearinghouses and contract markets would be registered with the CFTC and subject to various rules on financial resources, risk management, reporting, recordkeeping, and disclosure. The requirement for rules issued jointly by the SEC and the CFTC highlights the fact that the legislation does not attempt to adjudicate or fully resolve the long-standing dispute between the SEC and the CFTC concerning which agency has jurisdiction over derivatives of various kinds. As in its other proposals for regulation of the financial system, the administration has apparently decided to avoid the difficult and controversial questions associated with a structural overhaul. In this, it has failed to capitalize on the work of the Bush Treasury, which produced a credible and responsible blueprint for a more streamlined and modernized financial regulatory structure.
In addition, the administration's draft provides for a division of authority on swap regulation between the SEC and the CFTC that leaves open many opportunities for dispute and uncertainty in the future as new instruments are developed. The basic division of responsibility seems to be that the CFTC will have jurisdiction over derivatives based on interest rates, commodities, and foreign currencies, while the SEC will have jurisdiction over derivatives based on securities, loans, and reference entities. The draft requires all "swap dealers and major swap participants" to register with the CFTC and would subject them to minimum capital and margin requirements. However, dealers in security-based derivatives and major participants in the market for security-based derivatives are also required to register with the SEC. In the draft, a swap dealer is "in the business of buying and selling swaps for such person's own account," and a major swap participant is "any person who is not a swap dealer and who maintains a substantial net position in outstanding swaps, other than to create and maintain an effective hedge under generally accepted accounting principles." Capital and margin requirements for banks that are swap dealers or major swap participants will be set by their respective bank supervisors, while those requirements for dealers or participants that are not banks will be established by the SEC and CFTC acting jointly. In both cases, the legislation describes the purpose of the capital and margin requirements as "to help ensure the safety and soundness of the swap dealer or major swap participant."
The reference to the safety and soundness of the swap dealer and swap participant reflects the administration's view that a default by one of these market participants could cause a systemic breakdown and thus must be regulated. In this Outlook, I will argue that, for the following three reasons, there is no sound policy basis for regulating either the swap market itself or the dealers and major participants in that market: (i) there has been no market failure that justifies government entry into this field; (ii) even in the AIG case, where a major swap participant was rescued by the Fed, there is no evidence that AIG's failure would have resulted in a systemic event or even a serious disruption of the swap markets; and (iii) there is no reason to believe--even in theory--that the failure of a large CDS dealer or CDS market participant can cause a systemic breakdown unless that failure occurs after or contemporaneously with the failure of large numbers of reference entities.
Why Regulate Swap Dealers or Major Market Participants?
The most basic question anyone should ask about a new proposed regulation is this: "what problem is this proposal intended to solve?" Regulation is costly, suppresses innovation, and reduces competition, so before we impose these costs on yet another industry--and those who consume its products or services--we should at least ask why regulation is necessary.
The administration's short answer to this question would undoubtedly be "AIG"--the idea, firmly embedded in the media mind, that AIG had to be rescued because, as a major CDS market participant, its failure would have created a systemic breakdown. No one in a responsible position at the Federal Reserve ever said this publicly, and the story has now become a classic case of something having been repeated so often, in so many places, that it somehow becomes true, even though no one can trace it back to an actual fact. Perhaps the administration actually believes that regulation of CDS is necessary to prevent a systemic breakdown, but it is far more likely that the administration is proposing to regulate the swap markets because it can, following the precept of the president's chief of staff, Rahm Emanuel, who famously remarked at the outset of the administration that one "never wants a serious crisis to go to waste."
Even if Fed officials had confirmed that AIG's swap portfolio was the reason for the firm's rescue, there would still be a question as to whether that judgment was correct. Accordingly, this Outlook will start with the more basic question: could AIG's CDS portfolio--no matter its size--have caused a systemic breakdown (or even a major disruption) of the financial markets if AIG had been allowed to fail? As I will explain below, I think the answer to this question is no, and thus I doubt the need for any legislation that regulates this or any other part of the derivatives industry.
Can Credit Default Swaps Create Systemic Risk?
Despite their fancy name, credit default swaps are simply reimbursement or guarantee agreements, much closer to a bank's standby letter of credit (SLOC) than to an insurance contract. A simple example of how CDS work involves three parties. (In this hypothetical, I am not including the intermediary dealers, to be discussed later.) Let us assume that A has made a $10 million loan to B. A is now exposed to the risk that B will default on the payment of interest or principal on the loan. For a variety of reasons, A may want to hedge its exposure to B. For example, A may wish to diversify its portfolio or reduce its exposure to a credit that it believes may be weakening. A can accomplish this objective without selling the loan by entering a CDS with C. In this transaction, C promises to reimburse A for a specified sum (known as the notional amount) if B defaults on its loan obligation, and A agrees to pay a fee (known as a premium) to C over a predetermined period of years. A continues to hold the loan to B on its books, and to receive the payments of principal and interest B is paying. If C has a better credit rating than B, A has both diversified its portfolio and upgraded its quality.
Through SLOCs, banks have engaged in virtually the same transaction for years. SLOCs are routinely used to guarantee the obligations of companies and government entities that are borrowing money from third parties. In an SLOC transaction, a bank takes the credit risk of the borrower, and the lender makes the loan on the strength of the bank's credit rather than the borrower's. Thus, the SLOC serves much the same function as a CDS; it promises a lender that the bank will reimburse the lender if the borrower defaults on the loan. If the default occurs, the bank pays the lender and takes over the lender's position. The only difference between an SLOC and a CDS is that the bank's fee in the SLOC transaction is generally paid by the borrower (B in the hypothetical) instead of by the lender.
In other words, there is nothing mysterious or truly new about the standard CDS transaction. The fact that it is called a "swap" should not cause anyone to think this kind of transaction is impenetrable, strange, complex, or unusually risky. It does not even create any new "interconnections" that had never existed before in the financial system. SLOC transactions create interconnections, as do interbank loans, syndications, deposits, repos, reverse repos, warehousing agreements, joint ventures, under-writing agreements, and dozens of other contractual arrangements that create the web of interrelationships that allow financial institutions to do what they are supposed to do--move funds from places where they are not used productively to places where they are. Calling these arrangements interconnections does not change their character, and it is certainly wrong to claim that CDS suddenly created interconnections that never existed before.
Returning to our hypothetical example, let us suppose that C--which has written the protection for A--goes bankrupt. C's bankruptcy means that A has lost its protection against B's default. But what else has it lost? It is now once again exposed to the risk of B, but in our hypothetical example B has not defaulted: it is still paying principal and interest to A. In this sense, A has not suffered any serious loss. It certainly has not lost the $10 million of the original loan to B, since B is still current on the loan. If A still wants protection against B's default, however, A must go back into the market and contract with a new CDS counterparty. To put this in even simpler terms, if you are a homeowner and your insurance company fails, you have not lost anything unless you have already had a fire and have not collected from the insurer. If these things have not happened, you simply call your local insurance broker and get a new homeowner's policy from another insurer. That is basically what A would do if it still thinks it needs protection.
It could be, of course, that A has lost the value of a favorable contract with C. That would be true if B's credit had declined between the time when A bought protection from C and C's default. In that case, A would be (in swap jargon) "in the money"--it is paying much less for the protection than it would have to pay to buy the same protection today. Because the protection A has purchased is worth more than what A is paying, A can even "sell" its contract to another party not associated with B for a profit. In addition, if A still wants to hedge its exposure to B, the fact that B has weakened since the original CDS was entered with C means that A will have to pay more for the same protection against B's default than it had been paying C. This also represents a real cash loss to A, which may or may not be able to recover its loss from C's bankrupt estate. Obviously, however, a loss of this size is highly unlikely to cause the failure of A. But because of the way the CDS market works, A and the other counterparties of C are likely to be protected against this loss anyway. The typical CDS contract requires the party that is "in the money" to receive collateral to cover its risk of nonpayment by the other party. Thus, if B's credit has been weakening since the CDS was entered between A and C, A is in the money and C would be required to post collateral with A. In most CDS contracts, A is also entitled to receive more collateral from C if C itself is perceived to be weakening. By the time C has gone bankrupt, accordingly, A should have enough collateral to defray all or a substantial portion of the cost of a new CDS to cover its exposure to B. But even if the collateral is insufficient for this purpose, the loss A and others similarly situated have suffered is too small to cause a systemic breakdown.
Setting the Record Straight on AIG
At this point, it is useful to step back and recognize that the key default in the CDS market is not the failure of C--the party that wrote the CDS protecting A--but the failure of B. As long as B is current in paying its debt to A, the failure of C--the party that is in the position of AIG, having written protection on B--would not cause any major loss to A. In that case, A's only loss is the cost of buying a replacement CDS. To be sure, this could be costly, but it is clearly not of a size that might reasonably cause a systemic breakdown. On the other hand, if B had defaulted before C, A's loss of C's protection would have meant a real and substantial loss for A.
Now let us consider a thought experiment involving AIG and the hundreds of billions of dollars (in notional amount) of CDS protection that had been written by one of AIG's subsidiaries (with AIG's guarantee). What would have happened if AIG had been allowed to fail? As the A-B-C discussion above suggests, there would not have been a major loss to any of AIG's counterparties unless large numbers of the reference entities on which the CDS had been written had defaulted before or contemporaneously with AIG's default. As far as we now know, however, there were no widespread defaults anywhere in the world financial system before AIG defaulted, so AIG's default would not have resulted in its counterparties suffering any systemically significant losses. If these counterparties still wanted protection against the same reference entities, they could have gone out into the market and purchased replacement CDS providing the same coverage they had lost. In some cases, this might be costly, but even assuming that AIG's counterparties had not received sufficient collateral from AIG to reimburse them fully for their loss, the size of the loss would not have been large enough in each case to cause their default and thus a systemic breakdown of some kind.
This analysis is valid for all cases in which AIG's protection covered named reference entities, but much of AIG's protection is said to have been written on portfolios of collateralized debt obligations (CDOs). Because of the collapse of the market for CDOs based on subprime or other mortgages, it would have been virtually impossible for counterparties protected by AIG to obtain replacement coverage for those instruments. In that case, the risk that had been concentrated in AIG would in effect have returned to AIG's counterparties. This would have represented an accounting but not a cash loss, as these counterparties were required to put these assets back on their balance sheets. Nevertheless, the fact that AIG's failure would not have caused massive losses was demonstrated in mid-March this year when it was revealed that Goldman Sachs was the largest AIG counterparty, with total AIG protection of $12.9 billion. The others named were Merrill Lynch ($6.8 billion), Bank of America ($5.2 billion), Citigroup ($2.3 billion), and Wachovia ($1.5 billion). When media representatives asked Goldman how large its losses would have been if AIG had been allowed to fail, they were told that the losses would have been "negligible." Consistent with the analysis above, Goldman had received collateral from AIG and had also bought CDS protection against AIG's potential default. If the firms named are indeed the largest AIG counterparties, the losses on whatever CDO protection AIG had written would not have been substantial in relation to their size--even if they had not obtained collateral from AIG before it was rescued. So it is highly unlikely that any systemic breakdown would have occurred if AIG had been allowed to fail. If this is true of AIG's default, it would also have been true of the default of any other CDS counterparty, no matter how large. The question, again, is not the number or the notional amount of the CDS that any such AIG-like market participant might have written, but whether the defaults of reference entities are so extensive or substantial that its counterparties suffer real cash losses.
The "Naked Swaps" Issue
In a sense, AIG was a special case. It was a major player in the CDS market, but not a dealer. It took on a lot of risk but did not appear to do much hedging. As such, it is a great example of the general proposition advanced in this Outlook: a large, unhedged market participant can fail without causing any significant losses to its counterparties. In reality, however, most of the major players in the market are CDS dealers, and they are mostly hedged. The CDS market is a dealer market. The great majority of transactions go through dealers, firms that are in the business of buying and selling CDS for their own account. So in our A-B-C hypothetical example, A would have bought its CDS protection from D, a dealer, and the dealer would have hedged its own liability by purchasing protection from C. This refinement does not change the substance of the transaction. Instead of A being exposed directly to C, it is exposed to D, and D is exposed to C. When C went bankrupt, D (rather than A) would have had to find a new counterparty to hedge its obligation to A.
The fact that the most active participants in the CDS market are dealers is important for understanding two issues that come up frequently among critics of CDS: "betting" and "naked swaps." In the CDS market today, it is possible for two parties that have no connection whatsoever to a reference entity or its debt to enter into a CDS transaction. For example, a hedge fund, E, could sell protection against B's default, even though it has no interest in B's debt. It would do this because it believes B's credit will strengthen, and it hopes to profit from an increase in the value of the protection it has written if this occurs. The value of E's CDS will increase if B's credit strengthens because E is now receiving a premium that is larger than what protection on B now costs. Alternatively, if B actually defaults, E would have to pay the full notional amount of the CDS it has written. This kind of transaction is known as a "naked swap," and to some critics of the CDS market, it is simply betting on B's financial condition. Moreover, the argument runs, this creates risk in the market where none existed before and is thus unhealthy.
The answer on the creation of risk is easy. Yes, E is taking on the risk of B's default, but there is no net additional risk in the market as a whole. If B defaults and E has to pay its counterparty, E is a loser but its counterparty is a winner. Their respective risks net to zero. The notion that E is simply betting on B's default, however, requires more analysis. It is important to recognize that E has a counterparty that is betting exactly the opposite--that B will stay the same or strengthen. Could the CDS market perform its function without these "bets"--that is, without buyers or sellers who are willing to buy and sell protection with no interest in the underlying asset? The answer is that the market would not function efficiently without these bets, just as the equity markets would not function effectively without a constant supply of buyers and sellers speculating--or betting--that a stock will move up or down. Economists would say that all this speculation, in the equity markets as well as the CDS market, contributes to price discovery--determining the price that best reflects the market's judgment about an equity or debt security at any moment in time. Efficient price discovery makes it easier for transactions to occur because it narrows the difference between what the potential buyer of CDS protection is willing to pay and the price that the potential seller of protection wants to receive for taking default risk on a particular reference entity.
It is easier to understand this process in the equity market than in the CDS market, but the process is essentially the same. Say hedge fund E believes that B--the same reference entity in our earlier hypothetical--is actually a greater credit risk than was implied by the fee A is paying to C. E believes that it can profit if B's credit weakens, just as a short-seller in the equity market can profit from a decline in a stock price. So E contacts a dealer and says it wants to buy protection on B. Pursuing E's request, the dealer then attempts to determine what it will cost to buy protection on B, because the dealer will want to be sure that it can hedge its exposure to E if it agrees to E's request. The fact that someone is buying protection against B's default causes the price of protection (known as the credit spread) to rise in accordance with supply and demand. The opposite would be true, and the spread for writing protection on B would decline, if E had gone into the market to offer protection on B.
The price discovery in the CDS market is good enough to act as a guide to the credit condition of thousands of reference entities and conditions in the credit markets as a whole. In the fall and winter of 2008, when there was great anxiety about the stability of many financial institutions, there were over seventy references to CDS spreads in the press, all for the purpose of demonstrating that the risks in the credit markets were either rising or falling at a particu-lar time. One of these statements was by Neel Kashkari, a Treasury official, in congressional testimony: "One indicator that points to reduced risk of default among financial institutions is the average credit default swap spread for the eight largest U.S. banks, which has declined more than 200 basis points since before Congress passed the [Emergency Economic Stabilization Act]." The fact that spreads were declining was taken as an indication that the market was settling down. This was true of spreads as a whole and spreads on individual reference entities. In other words, CDS spreads were thought to reflect real market judgments on credit quality and effective price discovery. CDS spreads would be far less reflective of real credit conditions, and hedging transactions would be more costly in every case, if market participants were not allowed to speculate--through naked swaps--on whether reference entities were weakening or strengthening.
Thoughts on a Clearinghouse
The administration's proposed legislation provides for the creation of one or more clearinghouses--also known (and better described) as central counterparties (CCPs)--which would be regulated by the CFTC. As the name implies, a CCP takes on the role of counterparty for all CDS transactions brought to the clearinghouse by a clearinghouse member. As an example, in the hypothetical transaction we have been using, A seeks protection against B's default by purchasing protection from the dealer, D. In turn, D hedges this obligation by purchasing protection from C, a major market participant. Where a CCP is present, D, a clearinghouse member, in effect transfers to the CCP the obligation to pay A if B defaults, as well as the hedging CDS that D purchased from C. The CCP, in effect, stands in the shoes of D and has what might be called a matched book, just as D might have had if the clearinghouse did not exist.
This arrangement reduces the default risk of A, which now does not have to worry about the financial condition of D. If B ultimately defaults, the clearinghouse pays A and the CCP's loss is first taxed to D. If D and the collateral it has posted does not or cannot fully reimburse the CCP, the balance of the loss is shared by all the clearinghouse members. This makes hedging through a CCP a significantly less risky prospect for A, which now has the resources of the CCP to cover its exposure to B, instead of just the resources of D. In this sense a clearinghouse reduces the default risk of at least one of the counterparties. Another advantage of a clearinghouse could be a more efficient use of collateral. With a CCP, the members post collateral to cover all the risks that they have transferred to the clearinghouse, while in the OTC context collateral may have to be posted for each separate transaction. On the other hand, bilateral netting in the OTC context can allow collateral posted in a CDS transaction to be used in an interest rate swap or commodity futures transaction, so OTC transactions can also use collateral efficiently in some circumstances.
Since the rescue of AIG, clearinghouses have attracted a lot of interest, in part because they are said to offer a degree of transparency by exposing the CDS obligations of all the CCP members. The administration's proposal--correctly--does not require nonstandardized CDS to be cleared through a CCP, and this substantially reduces the usefulness of the CCP as a source of information about the overall risks of the members. Nonstandard swaps will still be handled OTC and will remain essentially invisible. Although a clearinghouse can reduce the default risk for hedgers, that is only true where, as discussed above, the reference entity has defaulted. In most cases, the costs of a CDS counterparty's default would be the cost of a replacement CDS and thus relatively small. The loss could be large for a hedger in the case of a default by the reference entity, but it is questionable whether the costs imposed by a clearinghouse, including the concentration of risks described below, are outweighed by the limited additional default risk mitigation that is provided to the protected party in these cases.
The administration has made clear that one reason, perhaps the major reason, for requiring a clearinghouse for standardized CDS is to reduce the chances that the default of a dealer or major market participant might cause a systemic breakdown. If it is true, as argued in this Outlook, that the failure of a large dealer or participant in the CDS market will create systemic risk only under very limited circumstances, then the relatively weak cost-benefit arguments in favor of a clearinghouse become decidedly weaker. Indeed, it is possible that by concentrating much of the market's CDS risk in one place, the clearinghouse itself could become another entity that the government will feel compelled to bail out in the event of financial trouble.
Risks could concentrate in a CCP because members' incentives will be different when they are taking on business that they will ultimately transmit to the CCP than when taking on business that they will have to hold themselves. Dealer members are likely to be far more prudent in the latter case than in the former. When D has the opportunity to pass along to the clearinghouse the risks of its CDS with A and its hedging transaction with C, D's incentive to evaluate the risk of B and C fully is likely to be substantially reduced. In the future it may be required to put up more collateral if B and C weaken as credits, but that is in the future, and may not actually happen. For this reason, the CCP will be taking on more risk than D would have agreed to assume for the premium it is being paid, and these risks now will be shared with the other members of the clearinghouse. To protect itself, the CCP must have a sophisticated enough model or pricing mechanism to evaluate the risks it is assuming from D and other members. If it cannot do this--and there is reason to believe its incentives for this purpose are far weaker than those who buy and sell CDS in the OTC market--the clearinghouse structure creates a kind of financial tragedy of the commons, in which it would be in each member's best interest to transfer more risk to the CCP than the other members are transferring.
Under these circumstances--and given the CDS market's history of stable operations over the current decade and even during the current crisis--it would not be good policy to force any CDS transactions through a clearinghouse. Because of the adverse incentives it creates for members, a clearinghouse could produce more aggregate risk than the bilateral OTC market that it would replace. Nor can it be shown that a clearinghouse will have any effect in preventing a systemic breakdown. Together, these elements suggest that the unintended consequences of requiring the establishment of one or more CDS clearinghouses seem to outweigh whatever limited benefits might be achieved by a CCP system.
Peter J. Wallison ([email protected]) is the Arthur F. Burns Fellow in Financial Policy Studies at AEI.
1. See the administration's proposed legislation and fact sheets at U.S. Department of the Treasury, Financial Regulatory Reform: A New Foundation (updated August 11, 2009), available at www.treas.gov/
initiatives/regulatoryreform (accessed August 25, 2009).
2. Peter J. Wallison, "Everything You Wanted to Know about Credit Default Swaps--but Were Never Told," Financial Services Outlook (December 2008), available at www.aei.org/outlook/29158.
3. Over-the-Counter Derivatives Markets Act of 2009, §713, as proposed by the U.S. Department of the Treasury, August 11, 2009, available at www.financialstability.gov/docs/regulatoryreform/titleVII.pdf (accessed August 25, 2009).
4. Department of the Treasury, Blueprint for a Modernized Financial Regulatory Structure (March 2008), available at www.treasury.gov/press/releases/reports/Blueprint.pdf (accessed August 25, 2009).
5. Over-the-Counter Derivatives Markets Act of 2009, §711(a) (9).
6. Ibid., §717(e) (1).
7. Peter Edmonston, "Goldman Insists It Would Have Lost Little If A.I.G. Had Failed," New York Times, March 21, 2009.
8. Neel Kashkari (testimony, House Financial Services Committee, December 10, 2008), available at www.treasury.gov/press/ releases/hp1322.htm (accessed August 25, 2009).