Streetwise Professor

October 25, 2009

Prestigious, But Singularly Unpresuasive

Filed under: Derivatives,Economics,Exchanges,Financial crisis,Politics — The Professor @ 10:27 am

New York University Stern School of Business professors Aarya, Engle, Figlewski, Lynch, and Subrhmanyam (AEFLS) contributed a chapter advocating centralized clearing of credit derivatives in the recent volume Restoring Financial Stability: How to Repair a Failed System edited by Acharya and Richardson. Although the authors are a distinguished lot, they utterly fail to make the case. Here are a few of the problems with their analysis:

  • Their argument rests on the propositions that (a) due to the interconnectedness of the financial system and the implicit public subsidy of systemically important financial institutions, the default by a CDS transactor can impose costs on others, and (b) a clearinghouse can internalize this externality. As a matter of logic, second best considerations imply that even if these propositions are true, they do not support the claim that creation of a clearinghouse (central counterparty) for some derivatives products is efficiency enhancing. As I discuss in more detail momentarily, AEFLS’s propositions imply that such externalities are present in every transaction that systemically important institutions undertake. In such circumstances, it is well known that internalizing one of many externalities can reduce efficiency, rather than improve it.
  • Indeed, second best concerns may be most acute when a CCP addresses only one narrow segment of the market—e.g., CDS—as AEFLS do.
  • Even under the AEFLS proposal of mandatory clearing of some of the risks that big financial institutions undertake (i.e., the clearing of only some derivatives), the CCP will only see a sliver of the risk exposure of its member firms. Given that overall portfolio risk determines the systemic risk posed by a financial institution, a CCP will have neither the information nor the incentive to price risk (through its collateral or position limit choices) efficiently (which would require understanding and pricing all of the externalities). Indeed, raising the price of default/counterparty risk on one sliver of transactions can induce banks to substitute to other transactions that impose equal or greater external costs. [The analogy to multitasking agents is apt here. Giving high-powered incentives to such an agent on one dimension can have extremely perverse effects on the allocation of effort. ]
  • That is, the AEFLS argument proves too much. Or put differently, under the AEFLS theory, every loan, every investment, every dollar of interbank borrowing, every capital structure choice by a big financial institution is beset by externalities. Their logic, in a sense, compels them to conclude that market-based allocation of capital through financial intermediaries is inefficient, and that a centralized mechanism is required to ensure that financial institutions do not take on inefficiently large risks. Everything through a clearinghouse!
  • Now, there is some merit to the view that in a world of Too Big To Fail or other implicit or explicit subsidies to bank risk taking that there are (potentially large) externalities. But the foregoing means that a CCP, and especially a CCP focused on a relatively narrow range of products, is a highly defective means of addressing this externality problem. It is defective because it is incomplete, and in a world with multiple externalities, or multiple decision margins with external effects, partial and incomplete internalization does not necessarily enhance welfare.
  • Instead, if externalities in financial intermediation are an important concern, it is preferable to devise policies that either (a) eliminate the common source of the externalities (i.e., eliminate the TBTF subsidies that encourage the excessive risk taking), or (b) price all risks consistently across all decision margins and all decision makers. With respect to (b), this would entail theoretically setting capital charges or risk haircuts based on the entire portfolio of systemically important institutions. I say “theoretically” because measurement issues make this impractical. What’s more, institutions themselves will almost certainly have better information about portfolio risks, and crucially, the marginal contribution of a given transaction or class of transactions, to portfolio risks than will the regulators that set the capital charges/haircuts. This gives rise to adverse selection problems in which institutions load up on risks that they know are underpriced and avoid ones that they know (based on their private information) are overpriced. Moreover, as demonstrated with the Basel requirements that made it very attractive to invest in AAA CDOs, incorrect, centrally-set, universally applicable capital charges can induce institutions to engage in highly correlated investment strategies that exacerbate systemic risk.
  • [Parenthetically, some (including the EC and the US Treasury) propose applying different capital charges to OTC and cleared deals. This is mystifying. If a deal with given characteristics is executed by financial institution A, the effect of that trade on default losses (across all deals that A does) is the same, regardless of whether the deal is cleared or not. Clearing merely affects the allocation of those losses. Indeed, the clearing can exacerbate the systemic consequences of A’s default. For in a clearing structure, the members of the clearinghouse bear the effects of A’s default, whereas in an OTC structure, counterparties do. Since CCP members are likely to be systemically important institutions, clearing can actually increase the financial cost of a default by A to these other institutions. That is, if A deals with hedge funds OTC and defaults, the hedge funds lose. If A deals with hedge funds, and those deals are cleared, when A defaults the hedge funds are protected, but the CCP members incur a cost.]
  • AEFLS also fail to come to grips at all with the information asymmetries and heterogeneity-induced collective action problems that CCPs must face when setting collateral levels.  I’ve beaten that horse’s corpse to a pulp, so I’ll just suggest that those interested in the argument enter “clearing” into the search bar and while away too many hours reading what pops up.
  • To put it as succinctly as I can, AEFLS ignore altogether any analysis of the challenging problem of understanding how CCPs will act in practice, given the information and incentive issues that they must confront.  Instead, as is sadly too often the case in CCP advocacy, in AEFLS clearinghouses are a deus ex machina that descends on the financial stage to resolve all of the dilemmas confronting the financial markets.
  • AEFLS assert that it is desirable to establish margin levels that “ensur[e] minimal, near zero counterparty risk.” They have no basis whatsoever for concluding that zero CP risk is efficient. Indeed, it is almost certainly not the case, just as it is not the case that it is efficient to ensure that flying is perfectly safe. Reducing risk is costly. The necessary task is to balance the costs and benefits of risk reduction. For instance, in a clearing setting, setting margins equal to the maximum possible loss would eliminate counterparty risk. But collateral is costly. As a result of draconian margins, transactors would forgo some trades—including some trades that could reduce systemic risk (e.g., hedging trades that transfer risk efficiently). It is imperative to avoid such simplistic “zero risk” thinking of the type that AEFLS indulge in. The real world is about trade-offs, and those trade-offs must be acknowledged and considered in evaluating public policy.
  • AEFLS recognize that they have to have an explanation as to why CCPs must be mandated if they are such a great thing. They make a rather limp effort on p. 258, stating that “[l]arge players benefit from the lack of transparency in OTC markets since they see more orders and contracts than other players do.” They seem to be arguing that the lack of transparency generates higher profits for dealer firms, and this induces them to resist clearing. Indeed, they have to argue that this private benefit from opacity is so large that it exceeds the benefits that these firms could obtain through better pricing of counterparty risk, netting, etc. (Not that I agree that these benefits exist—that AEFLS do is the point.)
  • I made a similar argument over a decade ago, positing that resistance to clearing could be due to a “raising rivals’ cost” strategy.
  • I am far from persuaded by these arguments, however, both on theoretical and empirical grounds.
  • Theoretically, well hell, big players in every industry want to maintain market advantages that allow them to reap market-power profits. That’s why GM dominates the car business today, and is enormously profitable. Just kidding.
  • Taking AEFLS seriously, they posit that the existing market structure inflates the costs of those trading with dealers relative to costs that could be achieved by trading in a more transparent structure. This creates an incentive for institutions offering more transparent mechanisms to enter the market, thereby reducing trading costs for the dealers’ current customers, and inducing them to trade on the new transparent mechanism rather than the more costly dealer market. Indeed, note that exchanges with central clearing have tried to compete with dealers in CDS trading—and failed miserably even in relatively simple index products.
  • One could argue (though AEFLS do not) that this could reflect the effect of network externalities which create a coordination problem; an alternative system is more efficient than the incumbent (dealer) system if it attracts enough traders to supply liquidity, but it is costly to coordinate the necessary movement of such a number of traders, leaving the inefficient incumbent in place.
  • However, in other segments of the derivatives market, there are exchange traded instruments that are close substitutes for OTC products, yet the OTC market has maintained or increased market share. Eurodollar futures and interest rate swaps are one example. Ditto many FX products and equity derivatives, including equity options. For these markets, there is a closely related, highly liquid exchange traded product, but many prefer to contract in OTC markets instead. Why would market participants agree to be screwed by dealers if the dealer structure is so inefficient, and if dealers profit at their customers’ expense due to opacity and reduced competition? It is incumbent on those advocating a forced change in market structure to answer that question forthrightly and persuasively. AEFLS don’t. Nor do most of the mandate mandarins in Congress, Treasury, or the EC.
  • Relatedly, note that bilateral dealings prevail in many markets where margins are extremely thin, and dealer profits are commensurately small. For instance, bilateral dealings are quite common in vanilla interest rate and FX and equity derivative trades where spreads are very small. In vanilla IR swaps, about 50 percent of interdealer trades are cleared, and since dealer-customer trades are typically not cleared, less than a majority of vanilla IR swaps are cleared. Clearing of vanilla FX swaps is non-existent, and many equity derivative trades are not cleared. Given the competition in these markets (where a large number of institutions actively bid and offer to trade vanilla swaps), the “we don’t adopt clearing to protect profits” hypothesis is quite dubious.
  • That is, although low margins appear to be a necessary condition to result in a migration of products to clearing, they are not a sufficient condition. This calls into question the theory that the failure to adopt clearing is the result of the exercise of market power by self-interested dealer firms looking to preserve their profits.
  • It must also be noted that neither opacity or systemic risk externalities related to TBTF are necessary to explain failures to adopt clearing.  As I noted in an FTAlphaville article over the summer, the CBOT steadfastly refused to adopt clearing for 35 years even though it operated a highly liquid, transparent, centralized exchange.  The same was true of the LME; it failed to adopt clearing until forced to do so even though it also operated a transparent auction market.  Moreover, neither the CBT members nor the LME members were TBTF institutions benefiting from implicit or explicit public subsidies.
  • The experience in energy markets, in which post-2003 (and post the 2002 merchant energy meltdown) the market by its own devices migrated to clearing also demonstrates that there are not insuperable obstacles to the organic development of clearing, sans mandate.
  • Thus, a serious attempt to understand why clearing has not been adopted more widely by voluntary consent must at least contemplate the possibility that bilateral dealings offer some efficiency benefit that must be traded off against alleged external costs when determining whether compulsory adoption of clearing is justified. This AEFLS—and the rest of the mandate mandarins—signally fail to do.
  • Several other issues in AEFLS deserve comment.
  • First, they assert, rather than prove, that the correlation in financial institution CDS spreads demonstrates the existence of a contagion effect (p. 256—the use of the “interconnectedness” concept in this context is the key). This is a hypothesis (at best), not a fact. Exposure to a common risk can do so as well. Indeed, Jean Helwege (in a recent Regulation Magazine piece) argues persuasively that the contagion effect is vastly overstated, and the common-risk exposure effect is the real explanation for the CDS spread correlation that AEFLS discuss. As I mentioned before, virtually all major financial institutions made correlated investments in AAA MBS CDOs.  When these tanked, so did the banks (and AIG).  No need to invoke contagion.  Indeed, this raises the very real possibility that more centralization and standardization can increase systemic risks.  Since clearing entails just such an increase in centralization and standardization . . .
  • Second, they use the difference between the CDS spread and an equity implied spread produced by Hayne Leland based on a particular model to argue for the existence of “counterparty risks and clearing concerns.” To which I say: when given a choice between the market and a model, I choose the market every time (with all due respect to Hayne Leland). Moreover, this could just be a manifestation of the market price of risk which causes deviations between default rates implied by CDS (which are the default rates in the equivalent, pricing measure) and default rates in the true, “physical” measure.  Given the non-linearity of a CDS exposure, and its strong correlation with overall market conditions, it is very plausible that the market price of risk was large and increasing during the period of the crisis.  (A CDS on a Goldman or other big financial firm was akin to a short put on the market.  Such an instrument has a very high systematic risk exposure, which will lead to a substantial divergence between default rates in the true and pricing measures.)
  • Third, they use the AIG episode repeatedly. They ignore the equilibrium implications of the alternative histories that they contemplate (unlike what I have done in It’s a Wonderful Life: AIG Edition and in a recent comment in Regulation.) And again: the plural of anecdote is not data. Moreover, one episode is insufficient to base a restructuring of an entire system based on one episode, and even if this particular episode is so egregious, it is necessary to consider alternative ways of preventing its recurrence. This AEFLS do not do.

In sum, AEFLS fail completely in making a coherent economic case for mandated clearing, especially for a subset of OTC derivatives like CDS. They ignore crucial economic considerations and blithely overlook fundamental trade-offs. If there is a case for mandated clearing, I’d love to hear it. I haven’t heard it yet—least of all from this distinguished group of NYU-Stern professors.

Print Friendly, PDF & Email

No Comments »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress