Streetwise Professor

July 3, 2009

Manfully Attempted, But Still Falling Short

Filed under: Commodities,Derivatives,Economics,Energy,Exchanges,Financial crisis,Politics — The Professor @ 5:50 pm

The European Commission has just released a Commission Staff Working Paper  on “Ensuring efficient, safe, and sound derivatives markets.”  Not surprisingly, it comes out in favor of extending central clearing, especially to credit default swaps.  It must be said that the EC’s paper is far superior to the product of the US Treasury Department (a low bar, to be sure).  That said, I believe that many of the arguments and conclusions are fundamentally wrong.  The paper does the best job I have seen in making the case for extending clearing, but in so doing has unintentionally pointed out the flaws in that case.  Indeed, the basis for many of the paper’s conclusions actually supports the exact opposite of what the paper recommends.

At a high level, the failings of the paper are: (1) it identifies alleged benefits of central clearing, virtually all of which are private benefits, but completely begs the question “if these (private) benefits are as large as the EC says, why haven’t private parties adopted central counterparties (CCPs) more widely?”; (2) relatedly, it does not provide any explanation of why the market share of products purportedly traded on inefficient market structures has increased in recent years; (3) it fails to take into account the endogeneity of market structure, or explain how these structures arose and why–or why not–they are inefficient; and (4) although it clearly identifies the unique risks associated with CDS, it draws the exact wrong conclusion about how default risks from CDS should be borne–bilaterally or through a CCP.

The paper’s main case for CCP clearing (set out in section 2.4.2 and 2.4.2.2) focuses on netting, and the benefits associated therewith including “more efficient use of collateral” and “operational efficiencies.”  But these are benefits captured by the members of a CCP, and provide no rationale for mandating the adoption of clearing.  There’s no externality related to these issues that needs internalizing.  Relatedly, the paper asserts that a CCP is “considered a zero risk counterparty” and hence frees regulatory capital.  Really?  Clearinghouses can fail.  Cf. Hong Kong, 1987, and the near failures in Chicago at the same time, and the near failure on COMEX on “Silver Tuesday” 1981.  If regulators really treat it this way, a CCP would be a source of regulatory arbitrage, again meaning that market participants would have a strong–too strong, in fact–incentive to adopt it.  

With respect to risk sharing and collateral, the paper mistakenly claims that all collateral can be used to cover the costs of a default.  This is obviously wrong–as the box on SwapClear default management process in section 3.2.5.1 shows.  

The paper fails to mention that netting is not necessarily a social benefit, and can facilitate redistribution from one group of firms to another.  This redistribution can also give excessive incentive to form a clearinghouse.  

The paper also fails to recognize that more rigid collateral mechanisms can reduce the utility of a market as a hedging mechanism, and forcing adoption of these mechanisms can actually increase risks some firms face by precluding otherwise mutually beneficial hedging trades.

The paper touts that a CCP “an institution whose sole focus is risk  reduction.” But the world involves trade-offs, and an organization with a “sole focus” is not able to make such trade-offs. If the CCP prices risk appropriately, and therefore charges the right amount of collateral, then other market participants will make the appropriate decisions. But as I’ve argued extensively here, in my Regulation Magazine piece, and in working papers and presentations, for many transactions the clearinghouse is at an informational disadvantage relative to dealers in appraising risk–in which case it is less likely to price risk appropriately.

Moreover, the paper does not consider the incentives that the CCP faces. Yes, it has one job, but how strong are its incentives to do that job? A CCP is typically subject to low-powered incentives (e.g., non-profit form), which is understandable since it is an agent, and an agent of multiple principals; whereas a dealer bears many of the costs of poor counterparty risk management, as an agent the CCP does not. Moreover, although the CCP is touted as a solution to systemic risk, if systemic risk gives rise to externalities that affect those beyond its membership, this benefit is overstated: the CCP does not internalize all relevant externalities and therefore takes too few measures to address systemic risk.

The clear logical implication of the EC paper is: “If clearing is so great, somebody should have done it by now.”  Like virtually all advocates of the idea, the EC paper does not raise, let alone answer, that objection.  To do so, the EC must identify the market failure, and show how its proposed solution would ameliorate it. It has not done so.  It has not even recognized the necessity of doing so.  

The paper also asserts that clearing can “solve[] disruptive information problems” by improving information on positions.  But this can be achieved without the mutualization of risk that clearing entails.  

The paper sets out in great detail the dramatic growth in OTC trading, and the limited penetration of clearing, even in relatively standardized sectors.  Again, given the asserted benefits of clearing and other changes to OTC market structure, this calls for an explanation that is completely absent from the paper.  There must be something else going on here sufficient to overcome the supposed advantages of a CCP.

The paper does provide a very nice description of the variations in structure across different OTC products.  Note that these structures arise endogenously.  This raises the questions: (1) why do different products (e.g., interest rate swaps and CDS) have different structures? and (2) are these structures inefficient?  The endogenous variation suggests that there are fundamental differences across products and participants, and that different structures have evolved to accommodate these differences.  The paper sets out no serious argument or evidence that these variations are accidental, or inefficient.  As a result, it provides no basis for recommendations on how to change market structure via regulation.  

This failure to grapple with endogeneity–the causes of variations in market structure across products–leads to silly inferences, such as: “CCPs have proven to be resilient even under stressed market conditions . . . and showed their ability to ensure normal market function in the case of failure of a major market player” (2.4.2.1).  Yes . . . but this does not imply that similar results would obtain if all products were cleared.  The products that are cleared, and which worked successfully, have attributes that differ from the attributes of products that are not cleared–a point that the paper makes very forcefully.  Therefore one cannot conclude that requiring clearing of other products would work as well.  It could also be stated that other non-cleared products, e.g., CDS, also worked successfully in the face of the failure of a major dealer.

This issue of the differences across products is essential, and one that the paper raises repeatedly before drawing the wrong conclusion.  For instance, in the Executive Summary, and elsewhere, the paper says that CDS are different because of their “binary and discontinuous pay-out structure, concentrated dealer market structure, difficulty in valuing the rights and obligations contained in the contract, especially for the less liquid single name part of the market, lack of solid risk management measures and disproportionate dimension of the derivatives market with respect to the underlying market.”  I agree, completely.  

But I disagree with the conclusion drawn from these differences: “CDS have particular risks . . . all of which are difficult to mitigate . . . . For all these reasons, CCP clearing is desirable.”  But all of these features pose major informational challenges to clearing, and in my view, make it sensible to share risks bilaterally; the conclusion therefore does not follow from the stated premise.   Similarly, the paper points out the valuation difficulties involved in setting collateral from some products, due to the difficulties of valuation.  Again, agreed.  But the relevant question is who can do this better, bilateral parties or a CCP.  For some products, a CCP can do just fine, but for other products, it is quite likely that dealers can do a better job.  

These very considerations are why many operators of clearinghouses–who at first blush would seem to have an interest in a clearing mandate–have come out almost universally against such a requirement.  Note, this is not allegedly self-interested dealers talking: it is the clearing operators.  They recognize the unique risks of clearing some products, notably CDS, and believe that these risks make some products inappropriate for clearing.  That’s been the gravamen of my argument all along, but the paper completely overlooks this point–even though the clearing operators expressed their reservations very publicly weeks before the paper was released.  

Relatedly, in section 2.4.1.4 the paper identifies as a “Potential weakness of bilateral clearing” to be the fact that it relies on “each party’s internal ‘risk engine’ . . . . These risk engines are essential parts of a bank’s comparative advantage and while the broad contours may be similar each bank has a slightly different approach to risk assessment.  Overall, the crisis has highlighted the difficulty of designing models that adequately measure market risk.”  True, again.  But the relevant–unasked–question is: in this imperfect world, who can design the better model?  (Not perfect, but better.)  Due to the very “comparative advantage” aspect the paper identifies, dealers and other big traders have stronger incentives and better information to build better models than a CCP.  (A point I first made back in the 1990s.)  Moreover, an argument can be made that diversification in models is beneficial as a way of reducing the market’s exposure to model risk, whereas a clearinghouse chooses one model, and exposes the market in a big way to its (inevitable) limitations.  

In sum, in my view the paper presents a very thorough and accurate overview of the various OTC markets and products, and the unique characteristics of these products.  It then either ignores the implications of these differences altogether when making its recommendations, or makes recommendations that do not follow from its factual analysis.  Indeed, its recommendations are often the opposite of what a thorough consideration of the implications of these differences would imply.  

That is, the paper is very effective at pointing out the wide variation in the informational and risk characteristics of the various products traded in the OTC market.  It also effectively characterizes the wide variation in trading structures.  But it fails altogether at connecting these two inherently interconnected things.  Moreover, it focuses on the private benefits of its favored “solution” (clearing), even though this cannot be the appropriate basis for its policy recommendation.  There must be some sort of externality or other market failure that results in the adoption of inefficient institutional arrangements, but the paper fails altogether to identify what this externality or other market failure.

In brief, whereas the Treasury Department earns an F+ (and I am a generous grader), the EC merits a gentleman’s C.  It at least provides a factual basis for a considered analysis of the trade-offs entailed with different market structures.  I disagree with the conclusions it draws from its facts, but it’s a start, and a big improvement over Treasury’s arguendo ad AIG.

Print Friendly, PDF & Email

No Comments »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress