The Economics of Central Clearing; a commentary

By: John P. Davidson, Chief Executive Officer, Options Clearing Corporation (OCC) Apr 2021

At the recent WFE’s Clearing and Derivatives Conference 2021, I had the distinct pleasure of appearing with Albert Menkveld of Vrije Universiteit in Amsterdam to provide some commentary on a forthcoming paper that he and Guillaume Vuillemey (HEC Paris) wrote for the Annual Review of Financial Economics titled The Economics of Central Clearing. In general, I thought Albert and Guillaume’s paper was well done and useful, particularly to practitioners like myself who are engaged in running CCPs and may not have time to remain fully current with the academic literature on the subject. We certainly need to have much closer collaboration among academics and practitioners in our narrow but vital industry if we are to enable a more comprehensive understanding of central counterparty (CCP) clearing and its role in modern financial markets. It strikes me that there are a variety of issues where the academic literature remains incomplete, and reviewing the paper highlighted several issues for further discussion and study.

CCPs and counterparty risk management

The notion that CCPs mitigate counterparty risk, as Menkveld and Vuillemey state in their paper, is only partially true. The counterparty risk that CCPs mitigate is only that among those counterparties that are direct participants in the CCP. In a market that is predominantly principal-to-principal, as the US futures markets largely were when I first started working in them in 1980, that is a highly useful function. In an agency dominant environment, which is the current characteristic of virtually every exchange-traded market in the world, one can reasonably state the utility of a CCP in promoting financial integrity is necessary, but not sufficient. The member firms of CCPs must also do their part.

What I find most remarkable about the current demand for transparency into how CCPs manage their risks, which has certainly improved dramatically since the global financial crisis with the advent of the Principles for Financial Market Infrastructures (PFMI) standards, is the number of very sophisticated buy-side investment managers who are demanding more and more transparency from CCPs but are silent on the same topic with respect to the sell-side.

When was the last time you saw a broker/dealer or bank disclose its margin breaches or provide client disclosure on the operation of its initial margin and liquidity management models? In virtually every financial intermediary insolvency regime, certainly that of futures and swaps markets in the US, and US securities markets for amounts in excess of the very modest level of SIPC insurance, clients are mutualising the counterparty risk of each other. Once the failed intermediary’s capital is used up what a client receives is only its pro-rata share. Transparency should apply equally to all types of financial intermediaries and should be available to market participants in a similar timeframe and in a fully compatible format.

Theories of central clearing:

The question of collateral, which the paper asserts is the ‘first economic rationale for CCPs’ is a very useful topic that needs to be explored more thoroughly in both theory and practice. In traditional credit risk management theory, there are two concerns: the probability of default (PD) and the loss given default (LGD). In most views on this topic, collateral does nothing to alter the probability of default (that is a characteristic of the obligor well beyond the simple influence of the provider of credit), but it is essential in mitigating the loss in the event default occurs. Two points thus arise from this paper. The initial margin methodology-determined collateral requirements are not the first line of defence. The first line of defence are the membership requirements for admission to the CCP as a member, and the ongoing financial and operational surveillance engaged in by the CCP in collaboration with self-regulatory and regulatory organizations. Think of these as the choice of obligor or know-your-customer elements of sound credit risk management theory. Margin collateral is the second line of defence.

In my view, the two most compelling arguments for central clearing in the paper are the value of multilateral netting, which is not unique to CCPs,[1] and the fact that CCPs through their rules address what Albert and Guillaume describe as the fact that "…investors are imperfectly able to raise bilateral margins". Indeed, it is this second fact that compelled global regulators, as part of the 2009 G20 meeting in Pittsburgh, to encourage mandated central clearing of standardised derivatives. Prior to the financial crisis, I worked at a global derivatives dealer with a large credit default swap book, all with their myriad of attached credit support annexes. Every few weeks we went to the Federal Reserve Bank of New York to report our progress on getting the list of outstanding confirmations successfully compared. The most interesting thing about that list, which was not short, is that the oldest and largest value differences were with the largest other dealer counterparties. Operations professionals and ultimately traders could encourage the smaller counterparties to agree with their valuations and collateral requirements. The larger players, however, would not accept those determinations unless made by the Federal Reserve, and that took time.

With respect to another assertion in the paper, I take issue with the comparison of CCPs and insurance companies. A CCP bears little to no resemblance to an insurance company. The basic business model of an insurance company is to maximize the collection of premiums and minimize the payment of claims. All forms of collateral at a CCP, regardless of type and whether it is posted against initial margin or default fund requirements, are returned to the obligor clearing member once the risk is removed. Insurance companies do not return premiums a when a house, for example, fails to burn down during the coverage period.

The final theoretical argument for central clearing in the paper relates to the mitigation of fire sales. Within that useful section of the paper, I challenge the assumption that auctions are always the best tool for CCPs to utilise in the event of a default. They are certainly very important, but CCPs have many tools and the choice of tools is less critical than the frequency and realism of default management drills. Market liquidation is not an effective tool if the CCP has not opened accounts at liquidation agents in advance of an actual need; auctions cannot be effective if the potential bidders have never downloaded portfolios from the CCP or drilled on how to best ‘slice and dice’ the portfolio into something they can integrate into their existing books.

Another significant benefit of central clearing that was overlooked in this paper is CCPs’ role in the operation of an anonymous Central Limit Order Book (CLOB) marketplace. The primary means of managing counterparty risk in a bilateral market is by varying the price one is willing to pay for a contractual obligation based on the perception of the credit quality of the counterparty. Such adjustments are impossible in a CLOB environment where the counterparty is not known until after transaction execution. By standardising the creditworthiness of the counterparties through novation at the CCP, the CLOB—the defining characteristic of exchange-traded markets—can operate effectively.

Margin breaches

I have the same view of margin breaches as Alan Greenspan has about bank failures—the optimal number is not zero. The function of CCPs is to manage risk, not to eliminate it. In the operation of a CCP in extremis, there is a huge premium for market liquidity. Position liquidation in a default, be it by auction or market action, is much less expensive to the CCP and its participants where the markets for liquidating instruments are liquid than when that is not the case. CCPs have an obligation, subject of course to careful governance including input from clearing member representatives, to strike a balance between a margin methodology that sets the confidence level for breaches above the regulatorily mandated 99%, and one that sets the margin too high, which has a chilling effect that materially diminishes available market liquidity.

Furthermore, as noted above, CCPs do not manage the risk of every participant in the marketplace; they are only managing the risk of the individual and collective risks that the members present to the CCP. Every clearing member has a different risk appetite, and every market participant has a different level of credit worthiness and access to financial liquidity. The old notion of margining clients to exchange minimums, or margining clients at the same level as the CCP margins its members is really nothing more than a failure to engage in proactive credit risk management and to have the tough conversations with clients. The attitude from some commentators seems to be “let the CCP do the heavy lifting” and focus on the level of margin breaches rather than the validity of one’s own risk modelling.

The academic publications cited in the Albert and Guillaume’s section on the design of margins seem to focus on the role of netting, cross-asset class correlations, and adequacy as measured by margin breaches. While there is a reference to a 2020 study which found that “CCPs set margins more conservatively than standard value-at-risk measures would imply,” there is very little probing of the actual margin methodologies and financial models that give rise to initial margin requirements in this collection of academic literature.

OCC has long utilised an expected shortfall approach to modelling the probable change in portfolio value over our defined two-day margin period of risk. This approach is relatively rare among CCPs, which tend to utilise modifications to VAR, or the SPAN methodology, which was introduced to the industry way back in the days when I ran the CME Clearing House, more than 30 years ago. If we can accept the fact that the role of CCP initial margins is to appropriately mitigate the tail risk that is covered by CCP capital and the default fund, then the use of ‘Expected Shortfall’ is clearly superior. Expected shortfall is calculated by averaging all of the returns in the distribution that are worse than the VAR of the portfolio at a given level of confidence. OCC uses that value in setting the margin requirements. OCC’s margin methodology is fully disclosed to our clearing members in a document available to them on our website, as well as through the public comment process that our primary regulator, the Securities and Exchange Commission (SEC) utilises whenever we propose a modification to the constituent parts of that methodology. Margin methodologies and the financial models that generate margin requirements provide a rich area for further study by academics.

I find a very interesting part of the paper’s discussion of the design of margins is this question of the “positive correlation of positions across clearing member positions,” something that Albert characterises as “crowded positions.” I believe this is an important risk, but I disagree with the conclusion that CCPs ‘overlook’ this risk. Since a CCP by definition faces a ‘matched book,’ this risk manifests differently at a CCP than it does to an agency or principal market participant which, almost by definition, does not have a matched book. Second, while not exactly the same thing, OCC’s System for Theoretical Analysis and Numerical Simulations (STANS) large scale Monte Carlo-based margin methodology includes significant add-on components related to concentrated positions relative to instrument open interest as well as correlation/decorrelation analysis among different instruments. Finally, the use of stress testing output, both for the determination of our default fund, as well as the over 100 stress scenarios we utilize for routine portfolio and product surveillance, includes some of this analysis. The risk of the positive correlation of positions across clearing member positions is, however, an interesting subject on which we will do some further work.

Skin in the Game and default waterfalls

The discussion of the role of ‘skin-in-the-game’ in this industry has reached fever pitch. The Menkveld and Vuillemey paper cites one recent study that draws the conclusion that there is a correlation between the amount of Skin in the Game and the number of CCP margin breaches. This strikes me as a nearly classic case of what Socrates would term the post hoc ergo propter hoc fallacy. The assertion that the stock price or even the capitalisation of the ever-larger financial market holding companies of which many CCPs are a part strongly influences the risk management behaviours of the managers of such CCPs is not supported. What does impact CCP management behaviour is when their own ‘skin’ is actually in the game. For example, at OCC my and the deferred compensation of the entire Management Committee is at risk pari passu with non-defaulting clearing members in a default, and first after OCC’s capital in the event of an operational risk loss. That is a true alignment of interests. Now appropriately, this is not material relative to a significant default loss. Therefore, we have a proposal before our regulators for a persistent minimum amount of ‘skin,’ consistent with current European regulatory requirements.

The primary purpose of non-management compensation-related Skin in the Game is to avoid what I would term nuisance draws on the default fund. If the financial surveillance, initial margin methodologies and risk management practices of a CCP are not sufficient to manage the loss-given default of small and medium size clearing members, it is entirely reasonable that such losses be absorbed by the CCP’s capital. We can safely posit, as history has shown, that both idiosyncratic and systemic crises among the largest global or even multi-national financial intermediaries do not derive from their exposure at clearing houses. One can even go so far to say that such risks are not easily discerned from public disclosures or the backward looking regulatorily mandated financial reporting of the legal entities that are typically the financial holding companies point of interface with CCPs. Rather, it is only the bank supervisors with their on-site oversight model and the Comprehensive Capital Analysis and Review (CCAR) and European Systemic Risk Board (ESRB) --type stress tests that have true insight into the risks of these large financial service holding companies and their affiliates. Central and pan-national banks, however, are not in the habit of sharing detailed CCAR results with CCPs.

Recently there have been proposals from certain market participants that a CCP’s Skin in the Game ought to be larger than the largest clearing member contributions to the risk-mutualizing default fund. This proposal, however, would completely alter the incentive structures that characterize modern CCPs. One of the more important characteristics of risk-mutualisation is that it encourages competition among providers of clearing services. By remaining competitive and in the market for clearing services, even for one’s own account, the probability of monopoly or oligopolistic provision of clearing services among clearing members and the associated concentration risk may be reduced or avoided. Most observers will acknowledge that there are clear scale and skill advantages in the provision of clearing services. By establishing a buffer of ‘skin’ as large as 20% of the clearing fund that important incentive to remaining competitive is significantly diminished, which in turn makes the CCP more, not less, risky.

There clearly is an important missing component in the methodology for default funds that needs further study and analysis. While protection against losses stemming from the default of the two clearing member groups creating the largest aggregate financial exposure or ‘Cover 2’ has become the means of determining (through stress testing of extreme but probable default scenarios), the size of CCP default funds in many jurisdictions. However, the methodology does not address the primary issue CCP management would confront in the event of the simultaneous default of its two largest clearing members—the operational and liquidity capacity of the remaining clearing members to absorb those positions.

Regulation of CCPs:

It is understandable that after mandating central clearing for an already highly concentrated set of ‘too-big-to-fail’ dealers in standardised over-the-counter derivatives that the providers of those central counterparty services became themselves systemically important and highly if largely informally interconnected. In addition, one may question whether the essential characteristic for the utility of CCPs in risk managing OTC derivatives was purely the level of ‘standardization’ that characterizes the instruments subject to mandatory clearing, or whether liquidity and market structure should also be on the list of inputs. Regardless, of course CCPs need to be closely regulated. Those regulations should, and generally do, relate to both a CCP’s financial integrity as well as to its operational resilience. A failure on either dimension has likely systemic consequences.

With respect to central bank access for CCPs, which the paper briefly touches upon in the context of moral hazard, in my opinion it is critical that CCPs have access to deposit and custody accounts at central banks. Other than with their own capital, CCPs in my opinion should not be attempting to earn returns on any types of collateral posted by clearing members. CCP collateral, including cash, should be held at institutions with a maximum of both counterparty credit worthiness and operational capacity. Central banks fit that bill for many although not all types of collateral. Access to central bank liquidity facilities, even on a fully (or overly) collateralised basis is a more challenging proposition. I do not believe fully collateralised central bank extensions of credit are a sign of government bail-out, but I also recognise that my perception does not guide political reality. I know that during the stock market crash of 1987 central bank liquidity was extended to CCPs, but only through the intermediation of at the time a very small number of settlement banks. Whether direct access is appropriate is certainly an issue for further examination, both from the perspective of moral hazard and other relevant public policy.

Final comment:

There is one cited article that I had read well before this paper and one which I believe should be mandatory reading by the management of all CCPs. That is the 2020 article in volume 24 of the Review of Finance by Bignon and Vuillemey entitled: The Failure of a Clearing House: Empirical Evidence. The existence of severe agency problems related to clearing member distress and the possibility of inappropriately constrained risk shifting is as real in CCPs today as it was in the sugar markets in Paris in 1974. In many cases it is constrained by appropriate governance and ultimately regulatory oversight. Nonetheless, it remains a challenge.


The forthcoming paper The Economics of Central Clearing highlights the significant progress the academic world has made since the financial crisis in bringing intellectual rigor to our small but critical industry. For those efforts to reach their maximum social utility there needs to be much better collaboration among CCP practitioners and academic finance and economic faculty, both in helping guide subject matter focus, and in providing suitably anonymised data upon which detailed analysis can be performed.

DISCLAIMER: Any expressions of opinion in these remarks are my own and not those of the Options Clearing Corporation

[1] CLS Bank is not a CCP but does engage in multilateral netting for multi-currency settlements prior to its ‘payment-for-payment’ settlement service

The views, thoughts and opinions contained in this Focus article belong solely to the author and do not necessarily reflect the WFE’s policy position on the issue, or the WFE’s views or opinions.