Central counterparty anti-procyclicality tools: a closer assessment

Atsushi Maruyama and Fernando Cerezetti

Abstract

In the new market and regulatory environment following the 2007–8 crisis, central counterparties (CCPs) need to find the right balance between ensuring the safety of their business and avoiding the creation of additional distress in the market in the form of procyclical margin calls. However, balancing these conflicting goals poses a number of challenges, and no one-size-fits-all solution appears to exist. This paper investigates whether the substantial focus placed on the procyclicality of initial margin reflects both the original concerns at the time of the crisis and the intrinsic modus operandi of CCPs. We argue that the imposition of a procyclicality framework shaped by lessons learned from securities financing and bilateral over-the-counter markets to CCPs may have created unexpected challenges. We support the idea that the best mitigation mechanisms for procyclicality may be achieved through the establishment of an outcome-based approach and the enhanced disclosure of margining practices, especially regarding intraday mark-to-market procedures and variation margin calls.

Download PDF Appendix 1

Introduction

1 Introduction

In April 2008, a report published by the Financial Stability Forum traced the financial system procyclicality observed during the 2007–8 crisis back to two fundamental sources: the limitations in risk tools and their associated risk measurements, and the distortions in incentives for market participants (see Financial Stability Forum 2008).¹¹ 1 The term “procyclicality” was coined to express the effect that changes in the financial system and in related market practices can have with regard to amplifying business fluctuations and exacerbating financial instability during the current cycle (see FSF–CGFS 2009). However, to the best of our knowledge, there is no precise definition of the term that is able either to capture this relationship across the different dimensions from where procyclicality could emerge or to quantitatively determine when this relationship becomes detrimental. In particular, rating-based triggers in over-the-counter (OTC) derivative contracts required additional collateral as credit ratings of counterparties deteriorated. While these triggers were aimed at protecting counterparties against idiosyncratic shocks, they exacerbated procyclicality due to financial distress. Similarly, procyclical behavior emerged from collateral revaluations. Rising collateral prices increased credit availability and leverage during the economic expansion, but when asset prices started falling, investors’ capital to support trading was eroded. In addition, margin practices reduced the amount of collateral held in the period preceding the crisis, when volatility was low, but increased postings were required as market conditions worsened.

As a result of the lessons learned from the crisis, risk practices of financial entities were put under scrutiny. In this new market and regulatory environment, central counterparties (CCPs) face a new conundrum. These entities must now appropriately design their risk frameworks such that the right balance between ensuring the safety of their business, on the one side, and avoiding the creation of additional distress in the market, on the other, is achieved (see Murphy et al 2014, 2016).²² 2 In a similar context, but focusing on the banking sector, Repullo and Suarez (2013, p. 1) state that “the conflicting goals at stake explain why some observers (eg, regulators with an essential microprudential perspective) think that procyclicality is a necessary evil, while others with a macroprudential perspective think that it should be explicitly corrected”. However, a closer assessment of recent developments in the financial markets reveals that the present conundrum may not be as simple as it appears at first glance. First, recent technological developments have enabled CCPs to identify and measure both price and position changes in almost real time. Second, these new technologies were accompanied by similar enhancements in risk models and their capacity to quickly respond to modifications in the risk profile of portfolios. Further, risk management practices have also evolved to cope with this new abundance of information. Together, these developments have enabled CCPs to identify, measure and respond to emerging risks promptly.

There do not appear to be any arguments to deny that such risk-oriented enhancements are key to coping with the fast-changing dynamics of financial markets. Nonetheless, the question is how these new technologies, risk-modeling techniques and risk governance practices can function together to deliver the expected outcome of reducing procyclicality without disincentivizing clearing. As expected, balancing these determinants poses a number of challenges, and no one-size-fits-all solution appears to exist. In fact, there seems to be no consensus as to whether slow reactive risk frameworks would deliver better micro- and macroprudential outcomes than those that respond more closely to market behavior. As presented in Committee on the Global Financial System (2010, p. 12), “the evidence gathered during bilateral interviews suggests that stable through-the-cycle haircuts on SFTs [securities financing transactions] are no panacea, and that significant practical difficulties in implementing such haircuts exist”. Similarly put, uncertainty still remains as to whether it is more prudent to design risk frameworks that respond promptly to market events, where risk managers have the discretion to override models, or to have more stable risk metrics that, when implemented, will create step changes in requirements.

Viewed from a broader perspective, the present conundrum may seem no different from traditional economic problems, where single actions that are rational at the single-agent level can create suboptimal macroeconomic outcomes. However, a closer assessment reveals that nuances exist, and we wish to support and enrich this debate among regulators, practitioners and academics. Specifically, this paper assesses how the new regulatory framework that emerged postcrisis has approached the procyclical behavior potentially fostered by CCP actions, investigating whether the substantial focus placed on initial margin reflects both the original concerns presented at the time of the recession and the intrinsic modus operandi of CCPs. Further, our paper assesses whether some of the anti-procyclicality (APC) tools currently in use by CCPs are delivering as expected, especially in terms of their capacity to address the destabilizing margin calls to clearing members and other CCP participants.

Without questioning the relevance of procyclical behavior for CCPs and financial stability in general, we argue that the imposition of a procyclicality framework shaped by the lessons learned from securities financing and bilateral OTC markets to CCPs may have created challenges in some areas.³³ 3 While regulatory studies tended to focus on particular segments of the financial markets, their policy recommendations had a widespread effect across different areas. First, there seems to exist no consensus in the literature regarding the efficacy of existing APC tools. Moreover, the focus on initial margin requirements as key tools for controlling excessive leverage during stable periods and/or liquidity pressures in downturns may not fully capture the business and operational models of CCPs. Finally, any APC tool based on absolute levels of volatility (eg, absolute margin buffers) may fail to efficiently tackle the procyclical effects potentially created by substantial initial margin changes. Given that the size of the initial margin variations is also a function of the current level of volatility, these absolute-type tools may be creating the wrong incentives for CCP participants.

Against the above background, we support the idea that the best mitigation mechanisms for procyclicality may be achieved through the establishment of an outcome-based approach and enhanced disclosure of CCP margining practices. This view seems to find support not only among practitioners but also among regulators themselves. Committee on the Global Financial System (2009) explicitly stated that an alternative way to promote more stable initial margin requirements could be through greater transparency, while European Association of Clearing Houses (2018) and CCP12 (2018) discussed the idea of transparency and less-prescriptive measures as important mechanisms to deliver efficient and resilient CCPs. Under this principle-based approach, CCP risk governance would play a fundamental role in addressing procyclicality. For instance, accurate and more frequent intraday portfolio revaluations and variation margin assessments, combined with the communication of risks and potential margin calls to market participants as they evolve during the day, could be an effective mitigant of procyclical CCP margining. It not only introduces greater market transparency while preventing big-stepped margin calls, but also allows market participants to start deleveraging riskier positions while liquidity still exists and losses have yet to accumulate.

In addition to this introduction, our paper is organized as follows. Section 2 presents the idea of procyclicality and how CCP regulators responded to it after the 2007–8 crisis. Section 3 discusses the regulatory responses in closer detail, trying to bring different elements to the assessment of procyclicality in CCPs. Section 4 assesses the practical and methodological challenges with the existing APC tools. Section 5 summarizes the results of a case study. Section 6 concludes the paper.

2 Principles of procyclicality and regulatory responses

2.1 The crisis and its lessons

There appear to be no counterarguments to the statement that liquidity is a key factor in a well-functioning financial system. However, the panic of 2007–8 was unique in that the bank run did not occur in the traditional banking markets with the withdrawal of deposits and savings, but rather in the securitized-banking system (see Gordon and Metrick 2012). In particular, the run was on the sale and repurchase (repo) market, which largely provided short-term financing for a wide range of securitization activities and financial institutions.⁴⁴ 4 Although maturity transformation (ie, short-term borrowing and long-term lending) has long been present in the banking sector, at the time of the crisis the importance of this liquidity risk was not properly reflected in the existing risk management frameworks of banks (see Basel Committee on Banking Supervision 2008). Adding to the fact that in many cases repo rates were obscure, market participants increased their repo haircuts to unprecedented levels, and, in a few cases, there was a cessation of repo lending in many forms of collateral. The above findings are supported by the idea that, in modern banking, liquidity risk stems largely from exposure to undrawn loan commitments, the withdrawal of funds from wholesale deposits and the loss of other sources of short-term financing, rather than from the loss of demand deposits as in classic models of banking (see Cornett et al 2011).

Similar conclusions are found in Financial Stability Forum (2008), FSF–CGFS (2009) and Committee on the Global Financial System (2010), which point to the fact that the contraction of liquidity in interbank markets led to severe funding strains for many banks and disruptions to money markets. However, in addition to discussing the mechanics of the crisis, regulators were also interested in understanding the dynamics that led borrowers and investors to substantially increase leverage and risk in the run-up to the crisis as well as to unwind those quickly as conditions deteriorated. Among the different factors contributing to this cyclical pattern, special attention has been placed in the evolution of market-sensitive valuation and risk techniques as well as the type of incentives they create among market participants.

As a result of market and regulatory developments prior to the crisis, fair value measurements, mark-to-market valuation and volatility-based risk metrics have become more widely used in the risk management frameworks of financial institutions. Although it was not clear during the boom period, postcrisis evidence suggests that an important part of the procyclical behavior regarding leverage and risk was driven by these developments. From 2003 to 2007 haircuts and margins were progressively reduced as a consequence of rising prices and a low-volatility environment. However, once the cyclical peak was reached in mid-2008, falling asset prices led to higher asset volatility and higher measures of risk, triggering a reduction in risk appetite and leverage. When market conditions became more unstable and risk-related decisions were updated by firms, haircuts and margins rose substantially.

The regulatory response to the above findings came in the form of macroprudential policy guidance, either directly related to the implementation of quantitative constraints to leverage (eg, the Basel II and Basel III frameworks) or indirectly aimed at affecting incentives in risk measurement, pricing and reporting related to liquidity (see FSF–CGFS 2009).⁵⁵ 5 For details on the Basel II and Basel III frameworks, please access http://www.bis.org/publ/bcbsca.htm and http://www.bis.org/bcbs/basel3.htm?m=3%7C14%7C572, respectively. For instance, it was suggested that initial margin and haircuts for securities financing transactions and OTC derivatives could operate with minimal requirements for periods of low volatility. Similarly, another policy option was related to the promotion of measures of market risk that could take into account the through-the-cycle behavior of asset prices and tail risk. The idea was that when these metrics are more stable across the cycle, and calibrated to include periods of stressed market conditions, some desirable features for addressing financial system procyclicality emerge (see Committee on the Global Financial System 2010). Moreover, it has been argued that more conservative haircuts would also indirectly constrain leverage by increasing the cost of capital employed by banks and other financial institutions.

Although the primary focus of the regulatory response was directed at the identified sources of the crisis (ie, securities financing and bilateral OTC derivatives), the increasing systemic role of CCPs led regulators to transpose some of the more general principles of the policy proposals into new guidance and legal text applicable to CCPs. However, these responses tended to vary significantly across jurisdictions and regulatory bodies, as discussed in the following. Some of these responses were more principle-based, while in other cases they narrowly specified how CCPs should tackle procyclicality.

2.2 International standards

The Principles for Financial Market Infrastructures (PFMI), published in early 2012 by the Committee on Payment and Settlement Systems and the Technical Committee of the International Organization of Securities Commissions (CPSS–IOSCO), further enhanced the existing set of international standards at the time, aimed at increasing the resilience of these entities and of the markets they served.⁶⁶ 6 Previous standards applicable to financial market infrastructures can be found in CPSS–IOSCO (2001) and CPSS–IOSCO (2004). In providing this additional guidance, specific attention was paid to procyclicality in collateral and initial margin arrangements for CCPs. In particular, Principle 5 of the PFMI establishes that an FMI should define stable and conservative collateral haircuts, calibrated to include periods of stressed market conditions in order to reduce the need for procyclical adjustments. In the same vein, but with less prescriptive language, Principle 6 states that a CCP could consider increasing the size of its prefunded default arrangements to limit the need and likelihood of a large or unexpected margin call in times of market stress. However, CPSS–IOSCO explicitly recognizes that it may be impractical and even imprudent for a CCP to establish margin requirements that are independent of significant or cyclical changes in price volatility (see CPSS–IOSCO 2012, p. 54).⁷⁷ 7 For further guidance on the PFMI, see CPMI–IOSCO (2017), which provides additional text on the elements of procyclicality.

2.3 Dodd–Frank and US regulation

The Dodd–Frank Act, Titles VII and VIII, bolstered the existing CCP regulatory structure and provided the US Securities and Exchange Commission (SEC), the Commodity Futures Trading Commission (CFTC) and the Federal Reserve with the regulatory tools to more appropriately oversee the OTC derivatives market.⁸⁸ 8 The Dodd–Frank represents the Dodd–Frank Wall Street Reform and Consumer Protection Act, Public Law 111–203, 124 Stat. 1376 (July 21, 2010). In particular, the Act provided these commissions (SEC and CFTC) with the authority to issue regulation regarding the risk management of CCPs, explicitly stating the need to consider relevant international standards and existing prudential requirements. As such, the promulgation of the act into federal law in 2010 led to subsequent changes in CFTC and SEC regulations.⁹⁹ 9 CFTC and SEC regulations can be found in Title 17, Chapters I and II, respectively, of the Code of Federal Regulations. For instance, in 2011 the SEC proposed Rule 17Ad-22 requiring certain minimum standards for registered clearing houses. This was subsequently amended in 2016 (by adding Rule 17Ab2-2) to establish further requirements for the operation and governance of CCPs. Importantly, as it currently stands, no prescriptive legal text exists in CFTC and SEC regulations regarding the need for CCPs to address the procyclicality of their margin systems.¹⁰¹⁰ 10 Note, however, that in the US and EU equivalence agreement for CCP requirements the inclusion of measures to mitigate the risk of procyclicality is listed as one of the conditions of mutual recognition: see https://bit.ly/2IqUK5Z for details.

2.4 European Market Infrastructure Regulation and the EU regulation

The European Market Infrastructure Regulation was introduced by the EU and aimed at reducing systemic risk and increasing transparency in the OTC derivatives market.¹¹¹¹ 11 Regulation (EU) No. 648/2012 of the European Parliament and Council, July 4, 2012, on OTC derivatives, CCPs and trade repositories, commonly known as EMIR. See European Parliament/Council of the European Union (2012). Among other aspects, EMIR and its associated technical standards define the expected prudential requirements to be implemented by CCPs with regard to, among other things, exposure, margin, the default fund and collateral. Unlike the international standard setting bodies and the US authorities, the EU approach to procyclicality assumed a much more prescriptive form. In particular, Article 28 of the EMIR Regulatory Technical Standards (RTS) states that, when addressing procyclicality at the initial margin level, a CCP should employ at least one of the following options:

1. (i)

   a 25% buffer of the calculated margins;

2. (ii)

   a 25% weight to stressed observations in the lookback period; and/or

3. (iii)

   a floor based on volatility estimated over a ten-year lookback period.

Additional guidance published in European Securities and Markets Authority (2018) provides further details on how CCPs should interpret each one of these options. Similarly, but to some extent via a less prescriptive approach, RTS Article 41 states that a CCP should calculate its collateral haircut in a conservative manner, limiting procyclical effects as much as possible.

3 Regulatory response: a closer assessment

The previous discussion highlighted that CCPs were not directly related to the leverage building and liquidity pressures observed during the 2007–8 crisis. Yet, given the systemic importance of CCPs, authorities preemptively issued guidance and regulation associated with the potential procyclical role that initial margin and collateral arrangements could perform. The fact that no harmonized approach exists among regulatory authorities seems to suggest that procyclicality at CCPs is still a topic under development. Interestingly, the most recent evidence of procyclical liquidity pressures generated by CCPs emerged far from the core areas (ie, initial margin and collateral) where regulatory text exists. During the EU withdrawal referendum in 2016, CCPs were questioned for issuing substantial intraday margin calls in response to the increased market volatility (see Madigan 2016). However, the procyclical liquidity demand in this case emerged mostly from the near real-time marking-to-market of the cleared portfolios (and, in a few cases, realized as variation margin) and the fact that CCP policies and procedures prescribed that margin calls had to be met within one or two hours of being issued. If procyclicality is an evolving and complex matter, when considered at CCPs it warrants additional caution, given the specificities of the business and operational models that CCPs operate. As such, policy responses related to leverage and liquidity need to be assessed against these particularities.¹²¹² 12 There is also an important argument that relates to how CCP margin requirements are cascaded down to clients and indirect clients. CCPs are only one part of the chain, and considerations about procyclicality should also take into account how CCP participants calculate and issue margin to their clients.

3.1 Leverage and the CCP modus operandi

In recent years, a rich and vast literature has emerged analyzing the features that make CCPs different from other financial market institutions, such as banks and insurance companies. Cox and Steigerwald (2017) emphasize the fundamental difference between banks’ and CCPs’ risk profiles. In the words of the authors, “banks are risk-takers” while “CCPs are risk managers”. Moreover, Cox and Steigerwald (2017) and Manning and Hughes (2016) caution strongly against viewing CCP resilience and risk management through a bank lens. As such, if proposed policy actions to address procyclicality focus on reducing leverage during boom periods (see FSF–CGFS 2009), the fact that CCPs do not ordinarily have market risk exposures suggests that altering the structure of incentives at the CCP level may deliver different economical outputs than if they were applied at those entities that deliberately control the amount of risk they take (see Committee on the Global Financial System 2010; Raykov 2018).¹³¹³ 13 In discussing stable through-the-cycle haircuts, Committee on the Global Financial System (2010, p. 12) states that: “In particular, credit terms have several dimensions, which creates a risk that placing restrictions on one or more parameters merely moves the operative dimension elsewhere.”${}^{,}$¹⁴¹⁴ 14 Raykov (2018) argues that due to moral hazard, less-procyclical CCP margins may, in some cases, lead to deleveraging during the crisis (and not before, as intended), intensifying the pressure in the markets. Ultimately, risk exposure at CCPs is determined by clearing members’ – and their clients’ – decisions (see Cruz Lopez and Manning 2017; Cerezetti et al 2019).

The postcrisis policy responses seem to have prudently captured this feature. When the first policy proposals, aimed at controlling leverage, were issued, these were directed at the OTC bilateral derivatives market and securitized financing. It is in this sense that the alterations to the international regulatory framework for banks under Basel III (see Basel Committee on Banking Supervision 2011) and the introduction of the margin requirement for noncleared derivatives (see BCBS–IOSCO 2013) can be interpreted. For instance, as the latter report states: “The recent financial crisis demonstrated that improved transparency in the OTC derivatives markets and further regulation of OTC derivatives and market participants would be necessary to limit excessive and opaque risk-taking through OTC derivatives and to mitigate the systemic risk posed by OTC derivatives transactions, markets, and practices.” (BCBS–IOSCO (2013, p. 1)) Moreover, as stated in Committee on the Global Financial System (2010): “To reduce the financial system procyclicality arising from margining practices in secured lending and derivatives transactions, regulators and authorities should (i) promote the use of properly risk-proofed central counterparties.”

Notwithstanding the above, the fact that there are other elements surrounding, and in addition to, CCP initial margin and collateral that could potentially be functioning as means to prevent undue leverage deserves careful consideration. CCP initial margins, for instance, are usually designed to consider adverse market conditions when determining the lookback period implemented in the calculations (the same is true for confidence intervals and correlation offsets, for instance).¹⁵¹⁵ 15 It is a common practice for CCPs to require additional margin for taking large positions, especially on illiquid products. These margin add-ons affect the market participants’ funding liquidity and could therefore discourage them from building large positions even at times of low volatility. Similarly, CCPs operate under the pillars of membership and waterfall structures, implying that analogous types of control mechanisms also arise from, for instance, the access policy to the CCP, the default fund contributions and the margin add-ons, among others. Pretrading risk systems and position limits are also employed by some CCPs in order to control exposures. In CPSS–IOSCO (2012, p. 33) language: “There are several ways in which an FMI may provide incentives.” Therefore, APC tools applied to initial margin and collateral should not be seen in isolation from other elements affecting the structure of incentives of CCP participants (if CCPs were to be considered entities capable of regulating such a level of exposures).

3.2 Liquidity and the global aspects of procyclicality

If the APC measures proposed by authorities for CCPs did not aim at limiting leverage, as the above suggests, then consistency with the original sources of procyclical concerns would imply that they should be targeted solely at avoiding disruptive margin calls to market participants in periods of stress. However, challenges exist in defining when the potentially disruptive effects on financial stability (ie, the macroprudential perspective) arising from these changes would outweigh the benefits of supporting the resilience of CCPs (ie, the microprudential perspective). For instance, Brunnermeier and Pedersen (2009) suggest that a detrimental margin spiral can arise when margins are increasing in a time of market illiquidity. Conversely, Lewandowska and Glaser (2017) point to the potential inefficiency of the regulatory APC buffer, as little evidence is found to support the idea that the CCP margin setting is procyclical. Therefore, it is important to closely examine how the margin changes relate more broadly to the different business and operational models that CCPs run and the mechanics of the risk management frameworks they operate.

Although in recent years convergence has been observed in the modus operandi of CCPs (see Norman 2011), the markets they serve and the business models they operate are still to some extent bespoke. In particular, CCP margin dynamics are primarily driven by the price behavior and settlement procedures of the underlying products and instruments they clear. The distributional properties of equity indexes, for instance, are different from those of electricity: the latter are influenced by the nonstorable nature of the underlying asset, seasonal features and relevant time dependency. Also, if markets are physically settled and traded as spot contracts, the shorter settlement cycles make initial margin models more volatile as underlying assets are delivered and new positions enter the portfolio. Moreover, nonlinear financial instruments exhibit a different sensitivity pattern to market moves when compared with linear instruments. All of the above suggests that, while model choice and parameter calibration may be used to influence the outcomes of an initial margin, the idea that a single APC approach may fit all CCPs equally well poses some further challenges.¹⁶¹⁶ 16 Similar conclusions are presented in Glasserman and Wu (2017) when assessing EMIR APC tools.

When margin requirements and collateral haircuts are considered, it is common to think that theoretical models are the key drivers of procyclicality. However, most CCPs have a number of channels through which procyclicality may materialize. In particular, while the changes to initial margin requirements and collateral haircuts can cause funding pressures on their own, it is the variation margin that typically reacts first and, most importantly, accounts for the largest portion of the calls issued by CCPs. The dominant role of the variation margin in stressed markets comes from the fact that one-day price variation weighs only marginally on the initial margin calculation (see Appendix 1 online for an illustrative example).¹⁷¹⁷ 17 See Madigan (2016) for CCP margin calls during the British EU referendum in 2016. The majority of estimated margin calls ranging from USD25 billion to USD40 billion on the day related to the variation margin, as a number of assets saw record-breaking moves. Also, as alluded to before, either due to prudential risk management or as required by regulation, most CCPs monitor their risk exposure throughout the day, marking-to-market the positions on a frequent basis. In some cases, these may lead to an intraday variation margin; in others, simply to additional margin calls.¹⁸¹⁸ 18 The term “variation margin” is being used here to denote calls made by the CCP to be paid in cash only and to be exchanged between the different counterparties of a trade. Margin calls (either initial margin or additional margin) represent those requests that can be met with other forms of collateral (eg, bonds, equities, etc) and are lodged with the CCP. One way or another, in these instances margin or collateral models play a small role, and the key factor is how CCPs’ policies and procedures give them enough flexibility to respond to rapidly changing prices and market conditions.

As a practical example, CCPs that operate with very small intraday exposure thresholds might be forced to issue many margin calls as prices oscillate during the day, even if the final end-of-day volatility may not be as severe as that observed during the trading session. Similar concerns may arise regarding the policy that the CCP uses to release excess collateral. If CCPs demand collateral in $T+0$ but only release it in $T+1$, cashflow mismatches may emerge, putting further liquidity pressures on market participants. The fact that risk governance largely influences procyclicality has already been emphasized in the literature. Early on, Committee on the Global Financial System (2010) highlighted that market participants believed raising the initial margin requirements or raising haircuts in normal times to contain financial leverage might not dampen the large and disruptive variation margin calls that can arise in adverse market conditions.

Another important element in the assessment of procyclicality of CCPs regards the fact that the main motivation for central clearing is the risk created by the potential failure of members (see Braithwaite and Murphy 2016). However, unlike insurance companies, for instance, CCPs run a matched book, with their sustainability being based on the capacity to preserve such a balance. Therefore, one of the most important objectives of CCP risk management is to close out the portfolio of a defaulting clearing member in an orderly and timely manner. CCP risk models need to not only be efficient in forecasting the cost of the liquidation, but also consider that this closeout process will take place in a short period of time and in an environment where the risk profile of participants is changing dynamically (see Vicente et al 2015, 2017).

Currently most CCPs operate under closeout periods of up to seven days, depending on, among other aspects, the type of contracts cleared, whether or not client clearing is offered and whether margin is measured on a gross or a net basis. If CCPs are not able to liquidate the defaulting portfolios within these predefined time horizons, they risk not having enough resources to continue to operate. As such, the fact that CCPs need to quickly resolve a default event implies that they, and their risk systems, must respond promptly to any alteration in market conditions. To constrain the ability to closely monitor and act on new market information is to ignore that the sole purpose of CCPs is to ensure the performance of contracts no matter the existing circumstances.

A related argument, but from the perspective of clearing members and their clients, regards the amount of funds pledged as collateral. The purpose of margin buffers, collected by CCPs through APC tools, is to prevent parties defaulting on financial obligations when markets become adversely volatile. However, there is no clarity as to whether the collateral lodged at the CCP could itself be constraining the liquidity ability of CCP participants at the moment they need it most. Moreover, once the party has defaulted, the margin buffer ceases to serve its original purpose and bursts into other prefunded requirements to support CCP default management. Unlike other CCP margin requirements, which are designed to support the CCP in the exercise of its main risk function (ie, the reestablishment of a matched book), the APC buffer has its objectives elsewhere. Under no circumstances are both the APC margin buffer and other CCP margin requirements needed at the same time, since the party cannot be in predefault and default simultaneously (this is sometimes referred as the APC paradox). Therefore, it can be argued, in the spirit of the efficient use of financial resources and the prevention of significant overmargining, the APC margin buffer can be used to offset the other costs of closing out the defaulter’s portfolio.

Faced with the conflicting objectives of delivering APC margining and of ensuring the financial safety of the CCP, the best risk mitigation may be achieved through an enhanced disclosure of CCP margining practices and the historical variability of margins for a wide range of products and instruments cleared by the CCP. A clearly articulated risk statement and CCP tolerance for margin changes should also enhance the predictability of calls for funds, especially regarding the intraday marking-to-market of contracts. Market participants can incorporate comprehensive CCP disclosure into their liquidity planning to ensure a sufficient provision of liquid funds to withstand stress market conditions.

4 Tailoring initial margin procyclicality

The above discussion highlighted that procyclicality at CCPs is a complex matter, and a one-size-fits-all approach may produce unintended consequences. This section builds on the above background, and exemplifies some of these difficulties using the three APC model options available to CCPs under EMIR RTS. Although different challenges exist for the distinct options provided by regulation (see European Securities and Markets Authority (2015), CCP12 (2018) and European Association of Clearing Houses (2018) for some examples), for illustrative purposes, we use options (b) and (c) of RTS Article 28 as references to discuss some of the unintended consequences created by APC tools based on an absolute level of volatilities. Against this background, an alternative approach to more appropriately address regulatory and risk management concerns regarding procyclical margin behavior is proposed.

4.1 Illustration of the issues

While the term “procyclicality” lacks a precise regulatory definition, as alluded to earlier, our discussions for the rest of this paper will use the following (in accordance with EMIR RTS Article 28): “disruptive and big-step changes in margin requirements”. Under such characterization, it becomes clear that APC tools are designed to control the speed of margin changes rather than the absolute level of margin. The distinction is important since the APC margin buffer and the core initial margin serve different regulatory and risk management purposes, therefore affecting how CCPs define their performance criteria. Naturally, each margin component should be evaluated separately, in accordance with its own relevant regulatory and risk management objectives.

It may be more appropriate, for instance, to exclude the APC margin buffer from statistical initial margin model backtests, not only because this could mask potential model deficiencies but also because the buffer is intended to deplete in periods of stress. Consequently, the initial margin model shall be able to demonstrate its performance to the required EMIR RTS Article 24 standard without the APC buffer margin. In a similar vein, the performance of the APC buffer may be reviewed by measures of the speed and magnitude of initial margin changes over the liquidity horizon. Although it is easy to conflate this issue with the level of initial margin, which is reviewed by a separate backtesting program, the APC tools need to be evaluated against their own legal requirements and risk management standards.

Interestingly, option (a) of Article 28 is the only model in which the APC measure is defined multiplicatively to the level of initial margin (ie, 125%) and the speed of initial margin change remains constant in percentage terms throughout the nonstressed market.¹⁹¹⁹ 19 While option (a) is not discussed to the same extent as the other two in this paper, it is not free from its own challenges. Some topics are discussed in the European Securities and Markets Authority (2018) guidelines. The other two options of Article 28 are based on absolute levels: option (b) considers different weighting mechanisms for the lookback period, which is equivalent to blending two distinct volatility components;²⁰²⁰ 20 Note that the term “volatility” is used in a broad sense, as CCPs also apply the option (b) model directly to initial margin rather than variance. and option (c) simply takes the higher margin based on two volatilities calculated from different lookback periods. As a consequence, the effective weight of the APC measure of the two options is dynamic through time, varying in stressed and nonstressed markets. Figure 1 illustrates how the effective weight of the stress component ${\omega }_{\mathrm{stress}}$ changes with respect to the current volatility level using a simple option (b) model framework in which 75% of the current volatility is blended with 25% of the stress volatility.²¹²¹ 21 The same argument holds with respect to option (c), in which the ten-year volatility floor is compared with the current level of volatility. In particular,

$${\widehat{\sigma }}_{(t)}^{\mathrm{art}.28}=0.75{\sigma }_{(t)}+0.25{\sigma }^{\mathrm{stress}},$$

where ${\widehat{\sigma }}_{(t)}^{\mathrm{art}.28}$ is the volatility forecast at time $t$, as per option (b) of EMIR RTS Article 28; ${\sigma }_{(t)}$ is the current volatility at $t$; and ${\sigma }^{\mathrm{stress}}$ is the stress volatility.

Figure 1: Effective weight of the stress component in the option (b) model.

In this illustrative example, we assume an arbitrary 10% stress volatility, noting that the effective weight of the stressed component in Figure 1 is defined as follows:

$${\omega }_{\mathrm{stress}}=\frac{{\widehat{\sigma }}_{(t)}^{\mathrm{art}.28}}{{\sigma }_{(t)}}-1.$$

The stress component attains a 25% weight, equivalent to a 25% margin buffer in the option (a) model, when the current volatility is at 5%. In other words, the APC margin buffer achieves a 25% impact “if the stressed component is set at a level that is double [that] of current volatility” (see European Securities and Markets Authority 2015). The following equation explains this relationship numerically:

$${\widehat{\sigma }}_{(t)}^{\mathrm{art}.28}=0.75{\sigma }_{(t)}+0.25{\sigma }^{\mathrm{stress}}=0.75\times 5\%+0.25\times 10\%=6.25\%.$$

Hence,

$${\omega }_{\mathrm{stress}}=\frac{6.25\%}{5\%}-1=25\%.$$

It is immediately apparent in Figure 1 that the effective weight or the speed of margin change increases exponentially as the current volatility tends toward zero. The stress component weighs just as much as the current volatility (${\omega }_{\mathrm{stress}}=100\%$) reaches 2% or a quarter of the stress volatility level:

$${\widehat{\sigma }}_{(t)}^{\mathrm{art}.28}=0.75{\sigma }_{(t)}+0.25{\sigma }^{\mathrm{stress}}=0.75\times 2\%+0.25\times 10\%=4\%.$$

Hence,

$${\omega }_{\mathrm{stress}}=\frac{4\%}{2\%}-1=100\%.$$

We argue that this is an undesirable property of option (b) (and, equivalently, option (c)) because, while volatilities can modify significantly over time, it is unlikely that they will have such a step change in the short liquidation horizon typically considered by CCPs. The implied returns that would enable this shift in volatility may be not only dependent on the volatility level at the time but also, most importantly, of very low statistical probability, as the following empirical evidence suggests (see Figure 6). Moreover, it is not yet clear to what extent the unwarranted level of high margin could constrain funding liquidity, discourage market participants from clearing or even prevent them from accessing CCPs due to the high cost of entry. If clearing is to be encouraged, as the postcrisis regime entails, it needs to be done over the whole economic cycle, and not only in times of distress. Finally, the above mechanics could present a particular problem for assets or instruments that exhibit a significant long-term peak-to-trough volatility ratio, as the regulatory expectation is that the full available history will be used (see European Securities and Markets Authority 2015).²²²² 22 The peak-to-trough ratio is defined as the ratio of the maximum to the minimum observed values of a variable over a fixed observation period.

5 A case study analysis

To further illustrate the above arguments, we now turn to historical price returns for Brent crude oil covering the period between January 2008 and January 2017, as displayed in Figure 2. A visual inspection reveals that the oil market repeatedly saw distinctive periods of stressed and nonstressed markets across the time period selected. For instance, the market saw significant volatility during the panic of 2007–8 and the extended period of turmoil that ensued. It was then followed by benign market conditions from 2012 to 2014, until the volatility returned to the market as the oil price descended from USD115 per barrel to below USD30 in late 2014, when the global supply glut dominated the market sentiment. Volatility clustering is one of the stylized features of financial markets that has led to the extensive use of conditional volatility models, including autoregressive conditional heteroscedasticity (see Engle 1982) and generalized autoregressive conditional heteroscedasticity (see Bollerslev 1986). The oil market is no exception to this phenomenon of repeated multiple volatility cycles with significant peak-to-trough ratios.

Figure 2: Historical Brent crude oil price returns.

It is understood that regulators are aware of the consequences of an outsized stress impact in a nonstressed market.²³²³ 23 See European Securities and Markets Authority (2015), which states that: “A larger weight to stress observations would reduce the procyclical effects but at the same time increase over-collateralisation during the non-stress periods.” However, the question remains as to what level of price returns is actually needed to generate the buffers prescribed by regulation. Figure 3 traces the effective weight of stress volatility (${\omega }_{\mathrm{stress}}$), discussed earlier, during the ten years covered for both options (b) and (c), with the EWMA smoothing scheme applied to the current volatility.²⁴²⁴ 24 A decay factor of 0.97 is used for the analysis. Note that the stress volatility applied to option (b), based on actual historical returns, is 5.62%, while the ten-year volatility floor for option (c) ranged between 2% and 2.38%. We observe that the weight varied significantly from 0% to 182% for option (b) and from 0% to 203% for option (c).

Figure 3: Effective weight of stress component: option (b) and option (c).

In an attempt to answer the previous question, the hurdle rates, ie, the price returns necessary to generate a preestablished weight of stress volatility, are calculated (see Appendix 2 online for details). These required price returns are shown in Table 1 under both a simple moving average (MA) and the EWMA volatility process, for which multiple liquidation horizons (one, five and thirty days) are considered. In these analyses, the maximum weight observed in the lookback period for option (b) is used as a reference.

Table 1: Hurdle rates for the implied stress returns.

Method	One day	Five days	Thirty days
MA	41.7${\sigma }_{(t-\text{1})}$	18.7${\sigma }_{(t-\text{5})}$	5.7${\sigma }_{(t-\text{30})}$
EWMA ($\lambda =\text{0.97}$)	15.3${\sigma }_{(t-\text{1})}$	07.1${\sigma }_{(t-\text{5})}$	3.6${\sigma }_{(t-\text{30})}$

The results deserve careful consideration. A price return jump equivalent to 15.3${\sigma }_{(t-1)}$ in magnitude is required for the EWMA volatility at $t-1$ to generate a stress weight of 182% in one day and 7.1${\sigma }_{(t-5)}$ over a five-day liquidation horizon. The hurdle rates for the MA volatility are even more extreme at 41.7${\sigma }_{(t-1)}$ and 18.7${\sigma }_{(t-5)}$, respectively. As a means of comparison, the equivalent one-day return observed during the Gulf War (ie, the most severe shock to oil prices in recent history) was 6.02${\sigma }_{(t-1)}$ of the corresponding EWMA volatility.²⁵²⁵ 25 The return is being presented in absolute terms to facilitate comparison, and it corresponds to the value observed on January 17, 1991. Moreover, in the lookback period under consideration, the largest one- and five-day EWMA volatility changes were 46.6% and 63.14% respectively, as displayed in Figure 4.²⁶²⁶ 26 The analysis is extended for the period 1989–2017 in Appendix 3 online, with similar conclusions being obtained. Consequently, to assume a jump to 182% within such short space of time may lack the support of historical evidence.

We believe that a margin impact of such magnitude, typically observed in nonstressed markets, is an undesirable consequence of APC mechanisms that predicate the absolute volatility level. A more appropriate and proportionate approach to APC is to work directly on the potential change of volatility over the liquidity horizon. The following briefly examines such measures for historical oil returns over the years January 2008 to January 2017, now scaled by an $n$-day lagged EWMA volatility (${\sigma }_{t-n}^{\mathrm{EWMA}}$), ${\tilde{r}}_t=r_t{({\sigma }_{t-n}^{\mathrm{EWMA}})}^{-1}$, as displayed in Figure 5.²⁷²⁷ 27 This analysis is extended for the period 1989–2017 in Appendix 4 online, with similar conclusions being obtained.

The EWMA-scaled return series are largely free from the volatility clusters observed in the original price returns (as per Figure 2) and are in proportion to the $n$-day lagged EWMA volatility through time.²⁸²⁸ 28 Price returns were tested for both stationarity (augmented Dicky–Fuller test) and independence (Ljung–Box test). Neither the original nor the EWMA-scaled returns were considered nonstationary. The test for the original returns rejected the null hypothesis of independence at a 5% significance level, indicating the existence of autoregressive terms or volatility clusters. However, when testing for the scaled returns (one-, five- and thirty-day lags), we did not reject the null hypothesis of independent returns. This observation suggests that the stress volatility, to some degree, is conditional on the prevailing level of volatility at $t-n$, leading to our argument that the APC measures can be better defined by a relative change of volatility than by absolute levels.

Figure 4: Historical maximum volatility changes (EWMA and MA). (a) One-day EWMA DVol. (b) One-day MA DVol. (c) Five-day EWMA DVol. (d) Five-day MA DVol.

Figure 5: Historical price returns: scaled by EWMA volatility (lagged by one, five and thirty days). (a) One-day lag. (b) Five-day lag. (c) Thirty-day lag.

By way of illustration, the option (b) model can be adapted to consider a stress component that is directly based on some percentile value (eg, $p=0.99$) of historical realizations of an $n$-day volatility change ($\mathrm{\Delta }\sigma$).²⁹²⁹ 29 In this illustrative example, we simply applied an option (b) formula in which two volatility components add up to 1, ie, $(1-\alpha )+\alpha =1$. However, this constraint can be removed given that the stress component is now in a different order of magnitude to the other volatility component. In particular,

$${\widehat{\sigma }}_{(t)}^{\mathrm{art}.28}=(1-\alpha ){\sigma }_{(t-n)}^{\mathrm{EWMA}}+\alpha \underset{(\mathrm{\Delta }\sigma )}{\underbrace{{\mathrm{\Phi }}_{v_n}^{-1}(p)}},$$

where ${\widehat{\sigma }}_{(t)}^{\mathrm{art}.28}$ is the volatility forecast at time $t$, ${\sigma }_{(t)}^{\mathrm{EWMA}}$ is the EWMA volatility at time $t$, $\alpha$ is the percentage assigned to the stress component, ${\mathrm{\Phi }}_v^{-1}(p)$ is an inverse of the cumulative distribution for historical volatility changes assessed for a percentile $p$, and $v_n$ is the historical realization of the volatility change over an $n$-day liquidity horizon:

$$v_n=\left(\begin{array}{c}\hfill \mathrm{\Delta }{\sigma }_1\hfill \\ \hfill \mathrm{\Delta }{\sigma }_2\hfill \\ \hfill \mathrm{⋮}\hfill \\ \hfill \mathrm{\Delta }{\sigma }_t\hfill \end{array}\right).$$

Finally, $\mathrm{\Delta }\sigma$ is the volatility change over an $n$-day liquidity horizon in which the margin procyclicality is assessed.

This simple additive model can provide an amount of APC buffer margin that is commensurate with the prevailing volatility level, without the unintended model consequence of the blending approach of option (b).³⁰³⁰ 30 This concept can be easily applied to the volatility floor of the option (c) model or indeed when CCPs calculate historical volatility including periods of stress, as per EMIR RTS Article 25(1). As Figure 6 exemplifies for the case of a five-day absolute volatility change, there is no unique alteration that summarizes the behavior of the volatility regime across the assessed lookback period.³¹³¹ 31 The absolute change in volatility was selected for illustrative purposes only. Similar conclusions hold when relative changes are considered. For the latter, this would imply a maximum percentage change of 63.14% for the lookback period considered. Most importantly, it should be emphasized that the volatility change regime is dependent on the level of the volatility per se, as presented in Appendix 5 online. Higher levels of volatility lead to greater expected changes in volatility. This evidence further supports the argument that stressed volatilities may not appropriately represent the volatility change regime in more stable market conditions. Such a relationship can be more suitably identified by, for instance, the use of quantile regression techniques.

Figure 6: Histogram of a historical five-day EWMA volatility change.

6 Conclusion

The analysis presented in this paper reviewed the regulatory response to the undesirable procyclical behavior observed during the 2007–8 financial crisis. Building on both the conflicting evidence presented in the literature and the analysis of the particularities of the CCP business and operational models, we support the idea of an APC framework that is outcome-based and leverages on efficient risk governance (eg, high levels of transparency) to deliver the expected outcome. In particular, we advocate for appropriate disclosure of CCP margin practices, especially those related to intraday mark-to-market and variation margin.

In the case where more prescriptive regulation exists, as with EMIR, this paper discusses the fact that APC tools available to CCPs based on absolute levels of volatility may create challenges for managing risk. While the APC buffer margin erodes naturally as the current volatility rises, the effective weight of a stress component in a nonstressed market increases exponentially as the current volatility level tends toward zero. This would imply that CCPs are required to lock away significant amounts of market participants’ liquidity, which may be unwarranted in light of the empirical evidence around volatility behavior. To amend this and more closely align the CCP risk methodology with our intended regulatory purposes, we proposed an alternative modeling approach that focuses directly on the change of volatility level over the relevant liquidity horizon.

Declaration of interest

The views expressed in this paper are those of the authors and should not be interpreted as reflecting the views of ICE Clear Europe or any other person associated with the Intercontinental Exchange.

Acknowledgements

This paper has benefitted considerably from extensive discussions with colleagues at the Intercontinental Exchange (ICE), ICE Clear Europe, the European Association of Clearing Houses (EACH) and the regulatory community. In particular, the authors would like to thank the anonymous referees, Arnaud Faure, Boudewijn Duinstra, Chuck Vice, David Li, David Murphy, Dionysios Soupios, Finbarr Hutcheson, Jack Peterson, Joanne Rowe and Scott Hill for comments provided. Any remaining inconsistencies are solely the responsibility of the authors.

References

• Basel Committee on Banking Supervision (2008). Principles for sound liquidity risk management and supervision. Guidelines, September 25, Bank for International Settlements. URL: http://www.bis.org/publ/bcbs144.pdf.
• Basel Committee on Banking Supervision (2011). Basel III: a global regulatory framework for more resilient banks and banking systems. Standard, June 1, Bank for International Settlements. URL: http://www.bis.org/publ/bcbs189.pdf.
• Basel Committee on Banking Supervision–Technical Committee of the International Organization of Securities Commissions (2013). Margin requirements for non-centrally cleared derivatives. Report, September, Bank for International Settlements. URL: http://www.bis.org/publ/bcbs261.pdf.
• Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics 31, 307–327 (https://doi.org/10.1016/0304-4076(86)90063-1).
• Braithwaite, J., and Murphy, D. (2016). Got to be certain: the legal framework for CCP default management processes. Financial Stability Paper 3, Bank of England.
• Brunnermeier, M., and Pedersen, L. (2009). Market liquidity and funding liquidity. Review of Financial Studies 22, 2201–2238 (https://doi.org/10.1093/rfs/hhn098).
• CCP12 (2018). RE: Draft guidelines on anti-procyclicality margin measures for central counterparties. Consultation Paper, ESMA. URL: https://bit.ly/2WTdALP.
• Cerezetti, F., Cruz Lopez, J., Manning, M., and Murphy, D. (2019). Who pays? Who gains? Central counterparty resource provision in the post-Pittsburgh world. The Journal of Financial Market Infrastructures 7(3), 21–44 (https://doi.org/10.21314/JFMI.2018.108).
• Committee on Payments and Market Infrastructures–Technical Committee of the International Organization of Securities Commissions (2017). Resilience of central counterparties (CCPs): further guidance on the PFMI. Report, July, Bank for International Settlements/OICU–IOSCO. URL: http://www.bis.org/cpmi/publ/d163.pdf.
• Committee on Payment and Settlement Systems–Technical Committee of the International Organization of Securities Commissions (2001). Recommendations for securities settlement systems. Report, November, Bank for International Settlements/OICU–IOSCO. URL: http://www.bis.org/cpmi/publ/d46.pdf.
• Committee on Payment and Settlement Systems–Technical Committee of the International Organization of Securities Commissions (2004). Recommendations for central counterparties. Report, November, Bank for International Settlements/OICU–IOSCO. URL: http://www.bis.org/cpmi/publ/d64.pdf.
• Committee on Payment and Settlement Systems–Technical Committee of the International Organization of Securities Commissions (2012). Principles for Financial Market Infrastructures. Report, Bank for International Settlements. URL: http://bit.ly/1mcqA8x.
• Committee on the Global Financial System (2010). The role of margin requirements and haircuts in procyclicality. CGFS Paper 36, Bank for International Settlements. URL: http://www.bis.org/publ/cgfs36.pdf.
• Cornett, M. M., McNutt, J. J., Strahan, P. E., and Tehranian, H. (2011). Liquidity risk management and credit supply in the financial crisis. Journal of Financial Economics 101, 297–312 (https://doi.org/10.1016/j.jfineco.2011.03.001).
• Cox, R. T., and Steigerwald, R. S. (2017). A CCP is a CCP is a CCP. Working Paper PDP 2017-01, Federal Reserve Bank of Chicago.
• Cruz Lopez, J., and Manning, M. (2017). Who pays? CCP resource provision in the Post-Pittsburgh world. Staff Discussion Paper 2017-17, Bank of Canada.
• Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflations. Econometrica 50, 987–1007 (https://doi.org/10.2307/1912773).
• European Association of Clearing Houses (2018). EACH response to the ESMA consultation paper “Draft guidelines on anti-procyclicality margins measures for central counterparties”. Working Paper, EACH. URL: https://bit.ly/2WTdALP.
• European Parliament/Council of the European Union (2012). Regulation (EU) no. 648/2012 of the European Parliament and of the Council of 4 July 2012 on OTC derivatives, central counterparties and trade repositories. Official Journal of the European Union, July 27. URL: https://bit.ly/2kNmIfD.
• European Securities and Markets Authority (2015). EMIR review report no. 2: review on the efficiency of margining requirements to limit procyclicality. ESMA/2015/1252, August 13, ESMA. URL: https://bit.ly/2rjX06o.
• European Securities and Markets Authority (2018). Final report: guidelines on EMIR anti-procyclicality margin measures for central counterparties. ESMA70-151-1293, May 28, ESMA. URL: https://bit.ly/2IXtuKp.
• Financial Stability Forum (2008). Report of the Financial Stability Forum on enhancing market and institutional resilience. Report, April 7, FSF. URL: https://bit.ly/2WPGTdh.
• Financial Stability Forum–Committee on the Global Financial System (2009). The role of valuation and leverage in procyclicality. CGFS Paper 34, Bank for International Settlements. URL: http://www.bis.org/publ/cgfs34.pdf.
• Glasserman, P., and Wu, Q. (2017). Persistence and procyclicality in margin requirements. Working Paper 17-01, Office of Financial Research (https://doi.org/10.2139/ssrn.2938515).
• Gordon, G., and Metrick, A. (2012). Securitized banking and the run on repo. Journal of Financial Economics 104, 425–451 (https://doi.org/10.1016/j.jfineco.2011.03.016).
• Lewandowska, O., and Glaser, F. (2017). The recent crises and central counterparty risk practices in the light of procyclicality: empirical evidence. The Journal of Financial Market Infrastructures 5(3), 1–24 (https://doi.org/10.21314/JFMI.2017.072).
• Madigan, P. (2016). Huge Brexit margin calls stoke intra-day funding fears. Risk.net, October 31. URL: https://bit.ly/2XUph1i.
• Manning, M., and Hughes, D. (2016). Central counterparties and banks: vive la difference. The Journal of Financial Market Infrastructures 4(3), 1–24 (https://doi.org/10.21314/JFMI.2016.058).
• Murphy, D., Vasios, V., and Vause, N. (2014). An investigation into the procyclicality of risk-based initial margin models. Financial Stability Paper 29, Bank of England (https://doi.org/10.2139/ssrn.2437916).
• Murphy, D., Vasios, V., and Vause, N. (2016). A comparative analysis of tools to limit the procyclicality of initial margin requirements. Staff Working Paper 597, Bank of England (https://doi.org/10.2139/ssrn.2772569).
• Norman, P. (2011). The Risk Controllers: Central Counterparty Clearing in Globalized Financial Markets. Wiley (https://doi.org/10.1002/9781119205913).
• Raykov, R. (2018). Reducing margin procyclicality at central counterparties. The Journal of Financial Market Infrastructures 7(2), 43–59 (https://doi.org/10.21314/JFMI.2018.106).
• Repullo, R., and Suarez, J. (2013). The procyclical effects of bank capital regulation. Review of Financial Studies 26, 452–490 (https://doi.org/10.1093/rfs/hhs118).
• Vicente, L. A. B. G., Cerezetti, F. V., De Faria, S. R., Iwashita, T., and Pereira, O. R. (2015). Managing risk in multi-asset class, multimarket central counterparties: the CORE approach. Journal of Banking and Finance 51, 119–130 (https://doi.org/10.1016/j.jbankfin.2014.08.016).
• Vicente, L. A. B. G., Cerezetti, F. V., and De Genaro, A. (2017). Estimating “hedge and auction” liquidation costs in central counterparties: a closeout risk approach. The Journal of Financial Market Infrastructures 6(1), 1–27 (https://doi.org/10.21314/JFMI.2017.084).

Download PDF Appendix 1

• Keywords
• CCP
• Initial margin
• Procyclicality
• Regulation AT
• Risk management
• Original research