10
The Impact and Risk of Clearing and Margining

Mandatory central clearing of over-the-counter (OTC) derivatives is not a panacea, and central counterparties (CCPs) create risks. For instance, the difficulties faced by CCPs in the stock market crash of 1987 posed a serious threat to the entire financial system. For the past century and longer, clearing has been limited to listed derivatives traded on exchanges. Bilateral OTC markets have been extremely successful, and their growth has been greater than that of exchange-traded products over the last two decades. Whilst certain OTC derivatives (notably some interest rate swaps) have been centrally cleared since 1999, the majority of OTC products have not moved to central clearing by means of natural forces.

The trouble with clearing OTC derivatives is that they are more illiquid, long-dated, and complex compared to their exchange-traded relatives. Hence, they may prove a challenge for traditional CCP risk management methods, especially with cross-border activity being so important. What is indisputable is that centralised clearing will have significant structural and behavioural effects on the management and allocation of risk in financial markets, causing a profound change in market structure and trading practices.

A first obvious and almost paradoxical problem with mandatory clearing is that CCPs clearing OTC products may become systemically important, creating a potential moral hazard if it is clear that government financial support will be forthcoming in the event of a CCP risk management failure. After all, bailing out a CCP is ultimately no better than bailing out any other financial institution. CCPs do not make counterparty risk vanish, and forcing derivatives through CCPs could create sizeable financial risks via concentrating counterparty risk within a single systemic point in the system. As CCPs clear more complex, less liquid, and longer-term instruments, their potential risks will likely increase.

A second concern is the costs and instabilities that CCPs and bilateral margin requirements will introduce through requiring a significant amount of liquid margin to be posted by members and their clients, with various estimates putting this increase in the region of trillions of US dollars. There is a question over the economic impact of such margin, which may start in terms of financial institutions being less profitable, but will eventually have an impact on economic growth in general. The more subtle problem is that margining can transmit systemic disturbances as requirements change. This can occur, for example, when firms that must meet large margin calls respond by selling assets and reducing positions in ways that exacerbate the price changes that caused the margin calls. Moreover, initial margins generally increase in volatile markets, which could have the effect of catalysing rather than resolving financial distress via a damaging, system-wide liquidity drain.

A third potential problem is related to the loss mutualisation that CCPs use, whereby any losses in excess of a member's own financial resources (mainly initial margin) are generally mutualised across all the surviving members. The impact of such a mechanism is to homogenise the underlying counterparty risk, such that all CCP members are more or less equal. The most creditworthy market participants may see less advantage to their stronger credit quality with CCP clearing. As with any form of insurance, adverse selection – when the insured know more about risks than the insurer – may be a problem and make risk-sharing costly. In a clearing context, to the extent that firms that trade OTC derivatives know more about the risks of particular cleared products than the CCP, these firms will tend to overtrade the products for which the CCP underestimates risk, and undertrade the products for which the CCP overestimates risk. CCPs could encourage excessive risk-taking compared to bilateral trading, since an institution knows that its potential losses are mutualised amongst other members.

CCPs are often touted as reducing systemic risk. Yet the clearing mandate covers OTC derivatives and so CCPs can therefore, at best, reduce systemic risk in OTC derivatives markets. Of course, based on past experience, reducing systemic risk in derivatives (aka weapons of financial destruction) may be rather close to reducing systemic risk in financial markets on the whole.

10.1 RISKS OF CENTRAL CLEARING

Regarding the general risk posed by CCPs (especially those clearing significant volumes of OTC derivatives), two obvious concerns arise:

  • Central nodes of failure. Mandatory clearing will increase the importance of CCPs as key central nodes within financial markets. As such, their failure or even distress could initiate a major disturbance. This also creates a moral hazard and too-big-to-fail dilemma: if a CCP needs to be saved from failure, then this would likely be at the expense of the taxpayer.
  • Shock amplifiers and propagators. Given their size and interconnections, CCPs could amplify systemic shocks and allow a financial disturbance to propagate through the market as a result of their actions (e.g. in relation to margin requirements).

10.1.1 Historical CCP Problems

There have been a number of failures of CCPs, and similar clearing-focused entities, that give some insight into the risks of central clearing

Whilst not a CCP default, the recent default fund losses at the Nasdaq are also an interesting case study and will be discussed in Section 10.1.3.

10.1.2 The 1987 Stock Market Crash

Whilst HKFE represented the only CCP failure as a result of the 1987 ‘Black Monday’ stock market crash, there are other examples of CCP distress in this period. The CCPs of Chicago Mercantile Exchange (CME), Options Clearing Corporation (OCC) and Chicago Board of Trade (CBOT) were very close to failure, and only prompt action from the Federal Reserve prevented a catastrophe (Pirrong 2010b).

Due to the very large price moves on and around Black Monday, the magnitude of margin calls was extremely large. This created a number of inter-related problems:

  • Difficulties in receiving variation margin payments due from those with losing positions, despite, in some cases, multiple intraday margin calls being made.
  • CCPs ‘absorbing’ significant amounts of liquidity by collecting some variation margin payments but not always paying out on the winning positions in a timely manner.
  • Large increases in volumes of trades and unprecedented price volatility, creating operational problems such as errors and delays in confirming trades, made worse by a lack of automated payment systems in some cases. Since some trades were reconciled only at the end of the day, many leveraged positions had losses well in excess of their posted margins.
  • A lack of linked clearing arrangements so that, for example, members hedging options (OCC) with futures contracts (CME) did not have gains and losses offset and were therefore caught in a ‘variation margin trap’. OCC had asked CME to agree to cross-margining prior to the 1987 crash (Norman 2011).

The net result of the very heavy liquidity problems caused by the above issues was that CME came close to having insufficient margin to start trading on 20 October 1987. Failure of CME was only averted due to its bank advancing the CCP $400m just minutes prior to the market opening, so that it could make variation margin payments totalling $2.5bn (IMF 2010).

At OCC, a large clearing member had difficulties in paying margin (and only did so after an emergency loan from its bank) and OCC itself was late in making payments to clearing members and suffered a significant loss due to a clearing member default. Around three-quarters of this loss (after using retained earnings) caused a hit to the default fund, with rights of assessment used to bring the default fund to its previous level (although the default fund hit was only a fraction of the total default fund size).

Without the strong support of the Federal Reserve (both explicit and implicit in terms of liquidity injections and public statements, respectively), a large CCP failure in 1987 was a significant possibility. An interesting review of this crisis is given in Bernanke (1990).

It is important not to overstate the importance of historical CCP failures in assessing the current benefit of central clearing. On the other hand, it is also important not to overuse the previous experience as a justification for the benefits of clearing.

10.1.3 The 2018 Nasdaq Case

A recent event at Nasdaq's commodities clearing division has provided a reminder of the risks of CCPs. Nasdaq Clearing is a CCP for various products including certain commodity contracts. One of the clearing members was Einar Aas, a proprietary trader who had traded positions based on the difference in price between Nordic and German power contracts. Following a large move in these underlying markets on 10 September 2018, the value of Mr Aas's portfolio fell substantially, and he failed to make an intraday margin call. This led to Mr Aas being declared in default, and the underlying portfolio was closed out in two days following a second auction (the first having failed to produce any adequate bid). Only four clearing members were involved in this auction.1 The result of the auction was a loss that exceeded the ‘defaulter-pays’ resources, which led to a loss of €7m for Nasdaq Clearing (‘skin-in-the-game’) and an approximate €107m loss to the segregated commodity default fund. This default fund loss was in the region of two-thirds of the total default fund, which clearing members were then contractually obliged to replenish.

The initial margin held was based on a Standard Portfolio Analysis of Risk (SPAN) methodology (Section 9.2.2) with a margin period of risk (MPoR) of two days and a confidence level of 99.2%, as well as stressed market conditions and including a buffer of 25%. The position did not trigger an initial margin add-on for concentration risk. The market move experienced was in excess of the initial margin of the defaulter, but only by about 40%. Most of the loss was, therefore, a result of the difficulty in auctioning the large position.

This episode is a cautionary tale, showing how clearing members can experience default fund losses as a result of the default of other clearing members. It highlights several potentially problematic areas of CCP risk management, notably:

  • membership criteria;
  • initial margin calculation methodologies and choice of parameters (such as the MPoR);
  • concentration add-ons for initial margin;
  • the auction process; and
  • the skin-in-the-game of CCPs.

For more details, see BIS (2018).

10.1.4 Risks Faced by CCPs

The primary risk for a CCP is the default of a clearing member(s) and the possible associated or knock-on effects that this could cause. Knock-on impacts could include the default or financial distress of other clearing members, failed auctions, business being moved away from the CCP, and reputational problems.

Although CCPs typically have a large number of members, the majority of clearing is concentrated amongst a much smaller number of participants. For example, Armakolla and Bianchi (2017) analyse European Union CCPs and report the proportion of total initial margin posted by the largest five clearing members (‘IM5/IM’). Out of a total of 11 CCPs, this ratio is more than 50% for 8, with the largest being 73%.

CCPs could potentially suffer losses from other non-default events such as fraud, operational problems (e.g. systems failures), legal issues (e.g. if the law in a given jurisdiction does not support the rules of the CCP), or investment (of cash and securities held as margin and other financial resources, or due to a deviation from this policy (e.g. a rogue trader)).

Other risks faced by CCPs include:

  • Model risk. CCPs have significant exposure to model risk through margining approaches. Unlike exchange-traded products, OTC derivatives prices often cannot be observed directly via market sources, so that valuation models are required to mark-to-market (MTM) products for variation margin purposes. The approaches for MTM must be standard and robust across all possible market scenarios. Initial margin requirements introduce even more complexity. Particular modelling problems could arise from misspecification with respect to volatility, tail risk, complex dependencies, and wrong-way risk. A lesson from previous CCP failures (Section 10.1.1) is that initial margin methodologies need to be updated as a market regime shifts significantly. On the other hand, such updates should not be excessive as they can lead to problems such as procyclicality (Section 9.3.5). Another important feature of models is that they generally impose linearity: model-based initial margins will increase in proportion to the size of a position. It is important in this situation to use additional components such as margin multipliers to ensure that large and concentrated positions are penalised and that risk is adequately covered (see the Nasdaq discussion in Section 10.1.3).
  • Liquidity risk. Large quantities of cash flow through a CCP due to variation margin payments and other cash flows, and a large amount of initial margin is held. CCPs must try to optimise the investment of some of the financial resources they hold, without taking excessive credit and liquidity risk (e.g. by using short-term investments such as deposits, repos, and reverse repos). However, in the event of a default, the CCP must continue to fulfil its obligations to surviving members in a timely manner. Although CCPs will clearly invest cautiously over the short term, with liquidity and credit risk very much in mind, there is also the danger that the underlying investments they hold must be readily available and convertible into cash. For example, CPSS-IOSCO (2012) principles require ‘prearranged and highly reliable funding arrangements, even in extreme but plausible market conditions’, so that bonds (including government securities) may only be counted towards a CCP's liquidity resources if they are backed with committed funding arrangements.
  • Operational and legal risk. The centralisation of various functions within a CCP can increase efficiency but also expose market participants to additional risks, which become concentrated at the CCP. Like all market participants, CCPs are exposed to operational risks, such as systems failures and fraud. A breakdown of any aspect of a CCP's infrastructure would be catastrophic since it would affect a relatively sizeable number of large counterparties within the market. Legal challenges and litigation as a result of the default management process (e.g. see the Caisse de Liquidation case discussed in Section 10.1.1) represent a significant problem.
  • Concentration, sovereign risk, and wrong-way risk. CCPs may hold margin in currencies and securities which are correlated to the default risks of their clearing members. For example, there may be direct exposure to the knock-on effects of a sovereign default in terms of the failure of members and the devaluation of sovereign bonds held as margin.

10.1.5 Risks Caused by CCPs

With respect to the historical cases discussed above, the common sources of failure can be identified as:

  • large underlying price moves (e.g. commodities, stocks, FX);
  • defaults stemming from losses in relation to large market moves;
  • initial margins and default funds being insufficient to absorb losses;
  • liquidity strains arising from delays in the flow of variation margin in and out of CCPs, caused by CCPs and/or clearing members (for operational reasons, as well as due to solvency problems);
  • initial margin calculations making inappropriate assumptions (e.g. relative versus absolute returns, discussed in Section 9.3.3) and not being updated according to changes in market conditions (e.g. after a large asset price and associated volatility increase); and
  • operational problems arising from excessive volumes and large price moves.

In summary, failures are a result of a combination of high market volatility, failure of members, and large asset price shocks, coupled with inadequate margins and default funds. There are several overall lessons to be gleaned from these derivatives CCP failures and near-failures:

  • Operational risk must be controlled as much as possible.
  • Variation margins should be recalculated frequently and collected promptly (intradaily in volatile markets).
  • Initial margin and default funds should be resilient to large asset shocks or gaps in market variables and to extreme dependency (e.g. the concept of correlation increasing in a crisis).
  • A CCP should carefully monitor positions, penalise concentration (e.g. by increasing initial margin), and act quickly in the case of excessively large positions.

10.2 ANALYSIS OF A CCP LOSS STRUCTURE

10.2.1 Review of the Loss Waterfall

Figure 10.1 shows again the loss waterfall (Section 8.3.4) and a potential liquidation scenario causing losses that exhaust the default fund over the close-out period or MPoR. In this extreme situation, the CCP needs to apply other methods of loss allocation. The definitions ‘first loss’ and ‘second loss’ are used to denote the defaulter-pays and remaining shared financial resources (survivors pay), respectively.

Mutualisation of losses can be an efficient insurance mechanism for risk sharing. It works well when risks are relatively idiosyncratic and independent (e.g. car insurance). However, it works less well as risks become more correlated and systemic, as illustrated by monoline insurers (see Section 2.4.4). Therefore, if the failure of a clearing member is an idiosyncratic event, then the loss mutualisation via the default fund of the CCP may enhance stability. However, in the case of a more systematic failure, this process may be less helpful.

Since initial margins are designed to provide coverage to a high confidence level, it is likely that any sizeable hit to the mutualised default fund would only arise due to an adverse combination of a number of underlying factors. Whilst such a scenario is by construction unlikely, there must be a well-defined allocation of losses and/or recapitalisation of the CCP as part of the CCP ‘rule book’. Indeed, the mere existence of such a well-specified waterfall may add stability to the CCP, since members will be better able to measure and manage their own risk if the CCP rules are clearly defined.

The general way in which losses above the defaulter's resources are handled is either:

  • to call for additional financial resources from clearing members (rights of assessment); or
  • to reduce the claims of the clearing members (other loss allocation methods).
Schematic illustration of the loss waterfall and potential risk in the event that losses exceed the primary financial resources of the CCP.

Figure 10.1 Illustration of the loss waterfall and potential risk in the event that losses exceed the primary financial resources of the CCP.

Table 10.1 Analysis of different components of the CCP default waterfall.

Paid by Funded or unfunded
Initial margin (defaulter) Defaulter Funded
Default fund (defaulter)
CCP equity (‘skin-in-the-game’) CCP shareholders Unfunded
Rights of assessment Surviving clearing members
Other loss allocation methods Surviving clearing members (and potentially their clients)

Ideally, loss allocation rules will also create the correct incentives both before and during the default management process. The nature of the different components of the loss waterfall is shown in Table 10.1.

For a CCP to recover from a member default, it needs to re-establish a matched book. This is normally achieved by replacing the defaulter's positions – for example, by selling long positions to (or buying short positions from) surviving participants through an auction process (Section 8.3.3). In a severe scenario, an auction may not clear at prices consistent with the CCP remaining solvent. In other words, the best prices demanded by surviving clearing participants to take on the defaulter's positions may exceed the financial resources available to the CCP. In an extreme scenario, it may be that the CCP receives no bid at all for one or more sub-portfolios. Hence, other loss allocation methods may be needed (Section 10.2.4).

There are a number of inter-related reasons why a default may lead to losses that may, in turn, breach the financial contributions of the defaulted member and cause losses in the mutualised default fund (and beyond). These are:

  • Delay since last variation margin was received. The delay before last receiving variation margin prior to the clearing member being declared in default. Due to variation margin practices of CCPs being at a minimum daily, and potentially also intradaily, this period is intended to be short. However, it should be considered how long a CCP might delay before declaring a large clearing member in default.
  • Market volatility in the hedging period. Whilst macro-hedging the risk, the CCP will be exposed to further market volatility and potential bid-offer hedging costs for large positions. For more complex portfolios, such as OTC derivatives, such hedging costs may be significant.
  • Auction costs. Finally, in attempting to auction the positions of the defaulted member, the CCP will be exposed to liquidation costs (although, as noted in Section 8.3.3, the above hedges should improve the auction process by creating less-directional portfolios). This could be related to bid-offer costs, but in more extreme situations there might be severe downwards price movements due to ‘fire sale’-like dynamics. An extreme case of this could be a failed auction, where a CCP does not consider that it has received reasonable bids for one or more portfolios. In liquid markets, such as exchange-traded ones, even large portfolios can be closed out relatively easily. For OTC derivatives, such a resolution could take at least a few days and may even be simply unachievable.

There is a key difference between losses due to market volatility and risk premiums experienced as liquidation costs. Market volatility may lead to losses for a defaulted member, but these must lead to equivalent gains for surviving clearing members. This means that within the set of positions cleared by the CCP, there are possible ways to consider offsetting losses – for example, by not posting margin against the gains of surviving members. On the other hand, when a CCP experiences bid-offer costs and/or risk premiums, these are costs that are not balanced by equivalent gains in the opposite cleared trades. The result of this is that loss allocation is likely to be more problematic. Put another way, highly-volatile products may clearly present a challenge, but illiquid products (or those that become illiquid in the aftermath of a major default) will be even more difficult.

Loss allocation methods are typically needed in the event that the defaulter-pays resources have been exhausted. Most CCPs use ‘rights of assessment’, which represent an unfunded obligation to contribute additionally to a default fund that has been depleted by losses. Typically, such a recapitalisation will be invoked if a significant fraction (e.g. 25%) of the default fund is used. If this right were unlimited, then the CCP would be unlikely to fail (unless all its clearing members did so). However, this would create significant moral hazard and unlimited exposure for clearing members, and may destabilise markets as members would be required to contribute to default funds at the worst possible time. For these reasons, rights of assessment are generally capped, an example being 100% of the last default fund contribution subject to a maximum of three defaults in a six-month period (see LCH 2017; RBA 2017).

10.2.2 Impact of Default Fund Exposure

Surviving clearing members have a ‘second loss’ position via their default fund exposure (Figure 10.1). This means that clearing members may not be incentivised to bid aggressively within an auction if they believe that the ‘defaulter's resources’ (initial margin and default fund) will provide sufficient loss absorbency. Put differently, auction bidders may take on positions at a profit, knowing that the defaulter's resources that are held by the CCP will pay for such a profit. Indeed, the clearing members' optimal strategy is to utilise at least 100% of the defaulter-pays resources, which otherwise will only be returned to the bankruptcy administrator. Indeed, it is actually optimal to utilise the default fund as the winning bidder stands to gain from the entire amount of default fund lost, whereas any allocation of default fund losses via rights of assessment will be pro rata.

There is some evidence of the above optimal behaviour in practice. For example, in the Lehman bankruptcy, there were claims that CME members profited from participating in the auction,2 although such firms seem to be immune from prosecution.3 This may also explain the large default fund losses in the Nasdaq case (Section 10.1.3). Such effects will be a significant detriment to other creditors (Section 10.3.5).

Related to the above is the concept that a CCP waterfall may behave rather like a synthetic securitisation or collateralised debt obligation (CDO), which has been noted by a number of authors, including Murphy (2013) and Pirrong (2013). The comparison is that the ‘first loss’ of the CDO is covered by defaulter-pays resources, and clearing members – through their default fund contributions and other loss allocation exposures, such as rights of assessment – are exposed to the second loss position on the hypothetical CDO. Of course, the precise terms of the CDO are unknown and ever changing as they are based on aspects such as the CCP membership, the portfolio of each member, and the initial margins held. However, what is clear is that the second loss exposure should correspond to a relatively unlikely event, since otherwise it would imply that initial margin coverage was too thin.

The second loss position that a CCP member is implicitly exposed to is therefore rather senior in CDO terms. Such senior tranches are known to be heavily concentrated in terms of their systemic risk exposure (see Gibson 2004; Coval et al. 2009; Brennan et al. 2009; Gregory 2014). It is well known that such structures are concentrated in systemic risk and perform very badly during large, market-wide shocks. Furthermore, a consequence of such structures is that they concentrate wrong-way risk (Gregory 2011).

The implications of the systemic and wrong-way risk concentration via the senior tranche exposure created by a CCP, by analogy to CDOs, would be:

  • the risk concentrated within a CCP will be systemic in nature;
  • correlation between losses from different clearing member defaults is likely to be high, as losses will hit the mutualised default fund precisely when surviving clearing members are under financial stress (wrong-way risk), and these members will be put under stress by the impact of losses of their default funds, rights of assessments, variation margin gains haircutting, and other loss allocation methods which could cause them to fail also;
  • loss allocation methods increase interconnections between CCP participants during periods of stress; and
  • default funds should, therefore, be very large and expensive.

10.2.3 The Prisoner's Dilemma and AIPs

If members believe that losses may exceed the defaulter-pays financial resources (Figure 10.1) and be mutualised back to them via the default fund, then this clearly incentivises better behaviour. However, once losses move into the survivors-pay region, moral hazard becomes a problem, and there is a danger of certain gaming behaviour by CCP members. The ‘prisoner's dilemma’ refers to a situation where it is in the individual interests of each party to take a particular course of action, but it is damaging for a large number of parties to take this route simultaneously. For example, clearing members may not participate actively in an auction process, even though collectively it is in their best interests to do this.

As long as they believe that initial margin may not be sufficient, it is collectively in the interests of the surviving clearing members to participate actively in an auction to ensure that the CCP can close out the portfolio most efficiently, so as to minimise losses to the default fund. On the other hand, an individual member potentially has the incentive not to participate actively in the auction in the hope that the other members will instead devote their time and balance sheets to ease the problem. Of course, when all members adopt this view, then the situation can become highly destabilising. Hence, the CCP should have alternative methods to close out positions and allocate the resulting losses. Ideally, these methods would also go so far as to penalise members behaving outside the bounds of the common good. This would then, in turn, incentivise good behaviour in the auction.

The first part of loss allocation involves the allocation of the finite default fund to cover losses. Some CCPs may operate separate default funds to align the losses of members with the products they clear. This would ensure, for example, that clearing members clearing only futures would be protected from default losses resulting from an OTC derivatives portfolio being cleared with the same CCP.

Furthermore, some CCPs may allocate losses within a default fund in tranches to mitigate the prisoner's dilemma and incentivise members to act in the common good. One way to achieve this is to make default fund loss allocation heterogeneous and linked to the auction bidding process. A basic form of this is to ‘juniorise’ the default funds of non-bidders – or bidders outside a given range – so that any losses are allocated to their default funds prior to those of other bidders. However, this introduces a related problem in that different participants may make more natural bidders in a particular auction due, for example, to their size, sophistication, or market activity in a participant type of product.4 For example, a bank not clearing swaps in a given currency, or inflation swaps, should not be expected to bid in an auction on this particular sub-portfolio of a defaulter.

A more sophisticated approach considers a hierarchy of bids and sub-allocation of default funds in line with sub-portfolios that will likely be auctioned (e.g. currencies of interest rate swaps). An example of the above concept is provided by auction incentive pools (AIPs), used in OTC auctions by LCH.5 With AIPs, each member's default fund contributions (pre- and unfunded) are allocated proportionally to a product and currency-specific bucket, in an amount that reflects the relative market risk that the member contributes in that currency. These currency-specific default funds will be used to absorb losses in auctions for products in each currency, respectively. If a loss exceeds the AIP in a currency, this will be allocated to the remaining default funds in other AIPs and any unallocated default funds. This means that, for example, a regional bank clearing only products in its own currency would be partially protected in the event of the default of a European bank clearing mainly euro-denominated products.

Another feature of AIPs referring to the allocation of losses in tranches is related to members bidding in an auction for a particular currency. Losses in each LCH AIP are distributed in sequence according to the competitiveness of members bidding in the auction:

  • non-bidders (who were expected to bid);
  • other (unsuccessful) bidders; and
  • the winning bidder.
Schematic illustration of tranching of default fund losses via auction incentive pools.

Figure 10.2 Illustration of tranching of default fund losses via auction incentive pools (AIPs).

The CCP would inform members of their relative contribution to an AIP (compared to other members). An illustration of the AIP concept is shown in Figure 10.2. This shows relatively small losses on the default fund for AIP 1 and larger losses for AIP 2. This means that all bidders on AIP 1 would be protected from default fund losses, which would be absorbed by non-bidders. Moreover, the relatively large losses on AIP 2 would not cause losses for members clearing trades only in AIP 1. However, note that the bidders in AIP 1 could experience losses in two ways: first, via larger losses on AIP 1, and second, via losses in excess of the total AIP 2 default fund. In the latter case, this is because, as mentioned above, losses in excess of an AIP can be absorbed by the remaining default fund (either unallocated or allocated to other AIPs).

The aim of the above is obviously to give clearing members the incentive to participate actively in the auction process within the subset of products they clear. Note that a member not clearing a given currency need not bid in that auction, as they will not have any default fund allocated to that AIP.

10.2.4 Other Loss Allocation Methods

In the event that rights of assessment are insufficient to absorb defaulter losses, the CCP could either fail or have other methods for allocating losses. CCP insolvency would clearly represent a highly-contagious event, and it is therefore important for there to be a robust mechanism for wind-down or recovery. In contrast to formal bankruptcy, CCP loss allocation potentially allows a timely and orderly resolution of an extreme loss event that is likely preferable to CCP failure.

The basic idea of other loss allocation methods is that defaulting members with out-of-the-money positions will be unable to pay variation margin to the CCP, and this will, therefore, be offset by some reduction in the claims of the other surviving members. This will clearly lead to a heterogeneous setup as losses will depend on the extent to which surviving members have a similar portfolio to that of the defaulter(s).

Schematic illustration of VMGH. The diagram on the left-hand side depicts the initial situation where financial resources are not balanced, and the right-hand side represents the balancing of resources by using the defaulter’s initial margin and (mutualised) default funds with VMGH.

Figure 10.3 Illustration of VMGH. The diagram on the left-hand side depicts the initial situation where financial resources are not balanced, and the right-hand side represents the balancing of resources by using the defaulter's initial margin and (mutualised) default funds with VMGH.

Variation margin gains haircutting (VMGH) is probably the most common alternative loss allocation concept (see ISDA 2013b). The idea of VMGH is that gains which have accumulated since the start of the default management process can be reduced pro rata so as to absorb the amounts owing to the CCP by the defaulted member. This means that clearing members whose positions have increased in value since the default6 will not receive the full margin to cover their gain, whilst those owing money to the CCP will still be required to pay in full.

An illustration of VMGH is shown in Figure 10.3. This shows a CCP owed variation margins of 5 and 15 by members B and D, the latter being in default. Correspondingly, the CCP owes the same total (20) of variation margin to two other members (A and C) in the amounts 12 and 8. Since the defaulter's initial margin and default fund contributions (10) are exceeded by the net amount owed (15), then the CCP must allocate a loss of 5. In the absence of any rights of assessment (or after such rights have been exhausted), this could be allocated via haircutting the variation margin owed to members A and C on a pro rata basis, as shown.

There may possibly be a cap on the variation margin haircut that can be applied. For example, LCH Ltd (LCH 2017; RBA 2017) caps VMGH at £200m or twice a participant's default fund contribution, and the loss distribution process is limited to 10 days. Surviving members vote if the cap or time horizon is likely to be breached.

Another interesting feature of VMGH is that it mimics the economics of insolvency in bilateral markets since parties with claims on the defaulter lose in a pro rata fashion. It may be argued that clearing members on the other side of the defaulter's trades – and therefore more likely to incur VMGH – may more actively bid in auctions to avoid this. However, it may not be fair to penalise a member just because they were on the correct side of a market move (e.g. their positions may be hedges).

If a CCP can auction a defaulter's portfolio at mid-market prices and suffer no other losses, then uncapped VMGH is guaranteed to work, as the losses from the defaulter must be matched by equivalent gains for other CCP members. This may be seen as particularly relevant if the underlying portfolio is liquid and the default has been associated with a large market move, creating some large ‘winners’ and ‘losers’, as may be the case in exchange-traded markets (as seen from some historical CCP failures, Section 10.1.1).

The obvious reason why VMGH may fail in practice is if the CCP has to close out at a significant premium to mid-market prices. In such a case, unless the risk premium required can be covered by the member's initial margin and default fund contribution, then VMGH will fail. Any non-default losses or caps in VMGH may create the same effect. One more drastic approach in such a situation is ‘tear-up’ with a similar method being ‘forced allocation’.

Tear-up involves cancelling contracts with surviving clearing members that represent the opposite position to that of the defaulter. Such contracts could be terminated with a cash settlement based on the current mid-market price, or the equivalent price when the default occurred or when variation margin was last exchanged. There may also need to be a pro rata reduction if the CCP is unable to pay such prices in full. The aim of the tear-up is to return a CCP to a matched book by terminating the other side of a defaulter's trades (or at least those that cannot be auctioned). All other contracts (possibly the majority of the total contracts cleared) could remain centrally cleared. As with VMGH, this option as a backstop may incentivise active bidding in an auction if members fear they may otherwise have contracts torn up.

The major difference with tear-up compared to VMGH is that the CCP is not exposed to any risk premium in auctioning transactions as it pays (at most) only the current market value of the transaction. To understand this, consider the illustration in Figure 10.4, which is similar to the VMGH example in Figure 10.3, except that it is assumed that in the auction there is a risk premium of 25 charged (represented as a payment to the market – M). This risk premium means that VMGH will fail since, even if the haircut is 100%, the CCP's total financial resources (15 + 5 = 20) are insufficient to pay the risk premium. In this scenario, an alternative would be to tear up all the underlying trades with members A and C and avoid the risk premium. Note that the CCP can pay the full (mid-market) amount owed to A and C without any haircut or other reduction (such as using a valuation when margin was last posted).

A CCP may reasonably attempt to tear up the smallest subset of trades that will return it to a matched book and allow it to continue operating. In the above example, if there were an additional 5 of financial resources (e.g. if the CCP had a larger mutualised default fund), then it would only be necessary to tear up 80% of the positions.7 Furthermore, if the CCP pays out less than the current market value of the contracts being torn up, then the tear-up fraction can be lower.

Tear-up represents a dramatic loss allocation process and has a number of important disadvantages:

  • Unlimited liability. Non-defaulting members theoretically face an unlimited liability since the amounts they are paid for torn-up trades will differ from the market prices at which these members can enter into replacement transactions.
  • Replacement impact. Whilst tear-up avoids the CCP needing to replace trades via the auction, a similar dynamic may occur where clearing members have to hedge their risk from trades subject to tear-up. Such replacement hedges could create market instability, especially since the CCP has failed to execute similar trades via the auction. A clearing member could even be sent into default because of tear-ups.
    Schematic illustration of partial tear-up. (Left) The initial situation where financial resources are not balanced, and (Right) the balancing of resources by using the defaulter’s initial margin and default funds together with tear-up.

    Figure 10.4 Illustration of partial tear-up. The diagram on the left-hand side depicts the initial situation where financial resources are not balanced, and the right-hand side represents the balancing of resources by using the defaulter's initial margin and (mutualised) default funds together with tear-up. M denotes the auction market where a risk premium of 25 is included in the price and therefore represents an extra cost for the CCP. Tear-up prevents the need to pay this risk premium.

  • Portfolio bifurcation. Partial tear-up may alter the balance of surviving members' portfolios and therefore the exposure to the CCP. This, in turn, may trigger additional initial margin or capital requirements if the tear-up results in a less diversified portfolio.

There are other loss allocation methods that are similar to tear-up. One is ‘forced allocation’ or ‘invoicing back’, where clearing members are obliged to accept certain portfolios at prices determined by the CCP. This has a similar effect since instead of tearing up an existing trade, a CCP might impose the reverse trade on a member. However, forced allocation may be more flexible as any member can be allocated positions, whereas tear-up requires a member to have the appropriate trades in their portfolio. Unlike tear-up, it may not be possible for a clearing member to pass on the impact of forced allocation to its clients.

Most CCPs use rights of assessment (capped) together with VMGH and/or tear-up as possible loss allocation methods. Other approaches, such as initial margin haircutting, have been suggested (Elliott 2013) but have not been used in practice, and also regulation may prevent them – for example, European Market Infrastructure Regulation forbids CCPs from using initial margin of non-defaulted members to cover default losses.8 Note that regulation may incentivise clearing members to pass on aspects of loss allocation to their clients.9

Pirrong (2011) notes that a CCP could be solvent but illiquid due to struggling to meet the immediate requirements to pay margin. In such a situation, a CCP would require access to external liquidity. This implies that a liquidity injection from a central bank is the only way to avoid a potentially catastrophic, liquidity-induced CCP failure. Not surprisingly, regulators view this liquidity support as a last resort and may require them to have facilities such as lines of credit in place to avoid this necessity.10

10.3 IMPACT OF MARGIN

10.3.1 Background and Historical Examples

Mandatory clearing and bilateral margining requirements will have one stark impact on the OTC derivatives market: a significant increase in margin requirements. Whilst it is not surprising that post-GFC regulation was focused on the reduction of counterparty risk, a concern is that the related funding and liquidity impacts of such regulation are in danger of being overlooked. Assessing the impact of increased margin requirements is difficult because it will be driven by subtle aspects such as the quality of margin required, segregation, and the functioning of margin transformation trades (e.g. repo markets). Nevertheless, it is important to characterise the impact of the increase in margin as a balance between reducing counterparty risk and increasing funding liquidity risk.

During the GFC, the reliance of financial institutions on short-term debt made them particularly vulnerable to the outbreak of problems in longer-dated markets (such as sub-prime mortgages). Excessive reliance on short-term funding markets has been cited by many as an important contributor to the severity of the GFC. It is therefore important to consider the funding liquidity risk created by clearing and margin requirements. Generally, it is relevant to consider two broad effects that arise from both clearing and bilateral margining requirements:

  • increases in variation margin requirements due to the tighter requirements compared to bilateral markets; and
  • the impact of the additional initial margin requirements.

The operation of margining mechanisms relies heavily on the extension of credit (Bernanke 1990). This could imply that in certain cases margining may be irrelevant or even lead to problems that would not exist in a non-margined world. There are some historical examples that shed light on these aspects:

  • American International Group (AIG). The example of AIG (Section 2.4.4) is a well-known example of the funding liquidity problems that can be induced by margin posting. In September 2008, AIG was essentially insolvent due to the margin requirements arising from credit default swap trades executed by its financial products subsidiary (AIGFP). Due to the systemic importance of AIG, the Federal Reserve Bank created a secured credit facility of up to $85bn to allow AIGFP to post the margin it owed and avoid the collapse of AIG. In this example, one of the key aspects was that AIGFP posted margin as a function of its credit rating. This created a ‘cliff-edge’ effect as a downgrade of AIG led to a very large margin requirement. Linkages of margin requirements to ratings have generally been removed or diluted in recent years in both bilateral and centrally-cleared markets.
  • The BP Deepwater Horizon oil spill. In 2010, British Petroleum (BP) experienced the largest accidental marine oil spill in the history of the petroleum industry. This caused loss of life, severe environmental problems, and financial losses for BP itself. Not surprisingly, in the aftermath of these problems, some of BP's trading partners (including banks trading OTC derivatives) had credit risk concerns which were compounded after BP's credit rating was downgraded by the major rating agencies (Moody's, Standard & Poor's, and Fitch) and its credit spreads widened significantly. There is some anecdotal evidence of banks giving flexibility in terms of margin posting to prevent BP from suffering further problems. BP then borrowed a total of $5bn from banks to ease its liquidity problems as a result of additional margin demands arising from the aforementioned credit rating downgrades.11 One interpretation of this is that banks were reabsorbing the credit risk of BP that had been previously transferred to BP as funding liquidity risk via margin requirements. Furthermore, this was being done at the very time when the margin arrangements were most important.12
  • Ashanti. Ashanti (now part of AngloGold Ashanti Limited) was a Ghanaian gold producer. When gold prices rose in September 1999, Ashanti experienced very large losses of $450m on OTC derivatives contracts (gold forward contracts and options) used to hedge its exposure to a falling gold price. Market participants commented that Ashanti had hedged an unusually high proportion of its total reserves. The negative value of Ashanti's hedge book meant that its OTC derivatives counterparties were due variation margin payments totalling around $280m. Ashanti had a funding liquidity problem: it had the physical gold to satisfy contracts, but not the liquid assets to make the interim variation margin payments. Ashanti then struck an agreement, making it exempt from posting margin for just over three years.13 This is another example of liquidity risk being converted into credit risk as margin agreements are deemed impractical, which in turn takes place at the time that they are most important. It could be argued that this is a good argument for exemptions for end users hedging with derivatives. However, it seems to be generally believed that Ashanti went too far in using derivatives as a risk management tool and hedged excessively (around 10 times its annual gold production at the time).14
  • MF Global. MF Global entered into trading strategies that it was unable to fund. Due to having insufficient liquidity to sustain its investment strategies (despite, allegedly, resorting to segregated customer funds, as discussed in Section 7.5.4), MF Global defaulted in October 2011 (see Heckinger 2012).
Schematic illustration of a potential feedback loop involving margin posting.

Figure 10.5 Illustration of a potential feedback loop involving margin posting.

Margin has the potential to cause feedback effects in volatile markets, as illustrated in Figure 10.5. This could be driven by the risk sensitivity (and in the extreme, procyclicality) of initial margins leading to larger requirements in volatile markets, in turn creating liquidations and large price moves. Such problems may spill over into other markets, as if CCP members have to meet unexpected margin calls in one market, they may sell assets in another and therefore drive prices down. Of course, methodologies for initial margin calculation that produce stable results can reduce these problems.

10.3.2 Variation Margin

It can be argued that variation margin is not expensive since it only represents a settlement (actual or proxy, see Section 7.2.6) of running profit and loss, and as such is a zero-sum game (one party's loss of variation margin is another's gain). This zero-cost variation margin idea has been expressed in various ways – for example:

In the case of variation margin, the BCBS and IOSCO recognise that the regular and timely exchange of variation margin represents the settlement of the running profit/loss of a derivative and has no net liquidity costs given that variation margin represents a transfer of resources from one party to another. (BCBS-IOSCO 2013b, emphasis added)

The variation margin payments, on the contrary, should not have a first-order effect on the demand for collateral, as variation margin is a one-way payment and hence does not affect the net demand for collateral [margin] assets. (CGFS 2013)

These statements suggest that margin costs and benefits are symmetric and/or that the ‘velocity of margin’ is infinite. In reality, this is not true, due to a frictional drag on margin from operational delays between margin posting and settlement, and also since institutions may have to hold excess funds to fulfil potential variation margin calls. Large derivatives players have hundreds of margin calls per day in both directions, representing potentially hundreds of millions of dollars in cash and securities.15 Margin velocity is also subject to resistance, as margin flow (especially high-grade liquid margin) within the financial system has fallen significantly and adversely influenced global liquidity in recent years. Singh and Aitken (2009b) show that counterparty risk during and in the aftermath of the GFC resulted in a decrease of up to $5trn in high-quality margin due to reduced rehypothecation, decreased securities lending activities, and the hoarding of unencumbered margin.

It can be argued that variation margin is not without additional liquidity costs for two reasons. Firstly, variation margin must be paid in relatively liquid securities or cash (CCPs typically only allow the latter), whilst some market participants will only have access to illiquid non-eligible assets.16 This could lead to a large liquidity squeeze in the event of a large market movement catalysing large variation margin requirements. This might be especially difficult if standard methods for transforming assets into cash, such as the repo market, become more strained in such situations. This could be a good example of why, for an end user carrying out hedging trades, these trades should be exempt from clearing and margin requirements. This is highlighted by the following quote:

The economic effect of the requirement to provide cash collateral [margin] is to convert the primary risk for companies from that associated with counterparty exposure into liquidity risk. Non-financial companies are highly experienced in managing their counterparty risk with financial institutions; managing liquidity risk in collateral requirements is substantially more difficult for them and is less efficient.17

A second problem with variation margin is the inherent delay caused by non-immediate posting (which can be days or at least hours). This interrupts the flow of variation margin through the system and causes funding liquidity risk. This effect will be particularly strong in volatile markets and even more so when there is a large asset price shock. CCPs have a privileged position with respect to margin exchange and may, therefore, interrupt margin flow significantly for their own benefit (if not that of the market in general). Here is a statement that seems to support the view that variation margin has high liquidity costs (at least in a crisis):

The following discussion of CME cash flows emphasizes variation margin payments because, as will be discussed, these payments placed the greatest stress on the financial system during the week of October 19 (Brady 1988).

Heller and Vause (2012) quantify the potential impact of variation margin calls in stressed markets. One of their conclusions is: ‘Variation margin calls on G14 dealers from CCPs that cleared all of their IRS [interest rate swap] or CDS [credit default swap] positions could cumulate over a few weeks to a substantial proportion of their current cash holdings, especially under high market volatility.’ Pirrong (2013) argues that variation margin has the potential to contribute to financial stress. The mechanism of MTM and (sometimes very large) variation margin posting on a daily (and sometimes intradaily) basis leads to a very tight coupling between parties. During stressed periods, rigid variation margin requirements can lead to substantial spikes in short-term liquidity needs. The size and timescale of these requirements could be billions of dollars during a period of hours. This arises from the non-perfect velocity of variation margining in the financial system. Furthermore, this velocity may slow in a crisis period as institutions may attempt to hold on to margins for longer, due to liquidity needs and worries about credit quality. At the same time, the ability to obtain short-term credit is likely to be at its most difficult. Variation margin requirements can also lead to feedback effects. Large price moves and their associated variation margin requirements may lead to fire sales that in turn may lead to further price moves in other markets. Hence, the very mechanism that is intended to reduce counterparty risk can potentially create systemic risk.

In the event of a clearing member default, the short-term liquidity requirements of variation margin on a CCP could be severe, as CCPs will need to pay out variation margin against losses incurred by the defaulter. Heller and Vause (2012) note that this would require a CCP to have access to short-notice liquidity backstops worth several billion dollars. Without this, a CCP may struggle to meet the variation margin requirements given the time horizons (hours) involved.

One important feature of variation margin is that it exposes market participants to MTM volatility, which in turn may increase their likelihood of default (as in the case of monolines and AIG, Section 2.4.4). This is also shown by Kenyon and Green (2013), who refer to a ‘virtual default’ as one which is created by the forced MTM effect of margining which otherwise would be avoided. Whether virtual defaults are correct or not is another matter. However, the likelihood is that tighter coupling to MTM via methods such as variation margining would create more, not fewer, defaults.18

It is important not to lose sight of a key benefit of variation margin, which is that it prevents excessive exposures and leverage from building up in the first place. With respect to the aforementioned cases of AIG and monoline insurers, the resulting losses would likely have never happened since it would not have been possible for the institutions to have built up such large losses.

10.3.3 Initial Margin

Even if variation margin is zero cost and should, therefore, be a natural consequence of OTC derivatives trading, the same is not true of initial margin. Initial margin represents overcollateralisation and moves away from a zero-sum-game market. This is especially true since initial margin requires segregation not to create counterparty risk of its own.

Additionally, initial margin calculations are subjective and complex due to their need to assess future (rather than current) exposure at some arbitrarily-defined confidence level and time horizon. Initial margins can also be procyclical.

Unlike variation margin, initial margin can at least be posted in non-cash securities. However, the admissibility of such securities represents a difficult balance. On the one hand, if a party allows a wide range of securities to be posted, then this relieves the liquidity strain of posting. On the other hand, having liquid and high credit-quality margin that retains its value even in a crisis period implies a more narrow range of eligible securities. Competition between CCPs may also lead to more relaxed margin restrictions, which ultimately increases the risk of failure.

10.3.4 Cost and xVA

With respect to variation margin, there is generally no xVA cost component. Variation margin is close to settlement and therefore simply offsets a valuation (or equivalently offsets cash flow payments). For this reason, a case of ‘perfect collateralisation’, which relates to frictionless variation margin exchange, does not lead to any xVA components, as discussed in more detail in Chapter 16. It is rather the lack of variation margin that leads to a valuation adjustment, normally known as funding valuation adjustment (FVA), as shown previously in Figure 2.12 and discussed in more detail in Chapter 18.

This does not mean that variation margin has no cost, but rather that this cost is not easy to quantify in line with some of the risks discussed in Section 10.3.2. Put another way, the costs are not related to the absolute amount, but rather are second order and relate to the underlying volatility of the variation margin requirements. From a bank's perspective, variation margin costs are contingent liquidity costs and may relate to the requirements under liquidity regulations such as liquidity coverage ratio and net stable funding ratio (Sections 4.3.3 and 4.3.4). For end users, variation margin costs may be more obvious due to the fact that their derivative positions are directional (see Figure 2.9). This can be seen in the behaviour of entities such as multilateral development banks (MDBs) who – thanks to their strong credit quality – have typically enjoyed the benefits of one-way credit support annexes (CSAs) (Section 7.3.2) when trading with banks. However, this leads to charges for credit, funding,19 and capital (CVA, FVA, and KVA) that in recent years have increased substantially due to the tighter funding and regulatory environments in which such banks are operating. An obvious way to reduce these large charges would be for MDBs to post variation margin via entering two-way CSAs. Whilst there is some evidence of this,20 there has not been a wholesale change in market behaviour, which suggests that the inherent costs (liquidity) outweigh the benefits (more favourable pricing). As with banks, these costs are probably considered largely to be contingent, such as the potential need to fund a large amount of variation margin requirements after a large price move in stressed market conditions.

Since initial margin represents overcollateralisation and must be segregated, its costs are more direct. The term margin value adjustment (MVA) is generally used to define initial margin costs, whether these may arise in bilateral or cleared environments. The volatility of initial margin can also be thought of as representing an additional cost on top of the average initial margin requirements.

From an xVA perspective, the reduction of counterparty risk and increase in funding liquidity risk discussed in Section 7.6.2 can be seen via a reduction in counterparty risk (CVA) and related capital costs (KVA), and an increase in initial margin costs (MVA), as shown in Table 10.2.

Table 10.2 Indicative comparison of counterparty risk, funding, and capital costs under different margin terms.

Bilateral
(no margin)		 Bilateral
(variation margin)		 Bilateral
(initial margin)		 Centrally-cleared
Counterparty risk (CVA and KVA) High Medium Low Low
Initial margin funding (MVA) None None High High

When posting and receiving margin, institutions are becoming increasingly aware of the need to optimise their margin management as, during the GFC, funding efficiencies emerged as an important driver of margin usage. Margin management is no longer a back-office cost centre, but can be an important asset optimisation tool, delivering the most cost-effective margin. An institution must consider the ‘cheapest-to-deliver’ cash margin and account for the impact of haircuts and the ability to rehypothecate non-cash margin. For example, different currencies of cash will pay different overnight indexed spread rates, and non-cash margin, if rehypothecated, will earn different rates on repo. The optimisation of margins posted across both bilateral and centrally-cleared trades (where there is optionality) will also become increasingly important (e.g. increasing bilateral margin usage of non-CCP-eligible securities). This is discussed later in Section 16.2.3.

10.3.5 Seniority

As shown in Sections 6.4.4 and 7.5.1, netting and margining do not eliminate risk. What they actually achieve is the redistribution of risk by changing the seniority of various creditors: the creditors benefiting from netting and margining (e.g. derivatives counterparties) become more senior, whilst other creditors are effectively demoted. As noted by Pirrong (2013), it is also important to look beyond the impact of clearing on OTC derivatives markets only. OTC derivatives make up only a subset of an institution's balance sheet.

It may be realistic to make OTC derivatives more senior and demote others since these other creditors may be less systemically important. However, it is important to foresee the potential changes brought about by clearing. Faced with a deleveraging in derivatives markets, market participants will likely find other ways to create leverage and obtain alternative forms of credit to support the increase in the required margin. These changes in capital structures, in turn, may create risks in areas of financial markets that have previously been viewed as relatively benign. It could be that regulation is too focused on OTC derivatives and is therefore blinkered to other possible dangers.

10.3.6 Bilateral and Cleared Markets

Table 10.3 contrasts some high-level differences between traditional bilateral clearing and central clearing (and to some extent bilateral clearing with initial margin). Traditional bilateral clearing follows a survivors-pay approach, where parties hold capital against possible losses when their counterparties default. Such capital is typically calculated based on a one-year time horizon and is sensitive to credit quality (e.g. a bank would need to hold more capital against a weaker-rated counterparty). As a result, the risk sensitivity and potential procyclicality of this capital is small. In theory, incentives are strong as losses are borne in general by those taking risks, although the process in the event of default is uncoordinated, with each party closing out transactions individually. In a bilateral market, variation margin may be used, but typically not initial margins (historically).

Table 10.3 Comparison of bilateral versus central clearing. Note that bilateral initial margins correspond more closely to the central clearing characteristics.

Traditional bilateral clearing (no initial margin) Central clearing or bilateral clearing with initial margin21
Margining Variation margin or none Variation and initial margin
Model Survivors pay Defaulter pays
Primary loss absorbency Capital Initial margin (and default funds)
Risk horizon ∼ 1 year ∼ 5 days
Risk view Long term (e.g. based on fundamental credit analysis and ratings) Short term (e.g. dependent on short-term market risk)
Credit quality sensitivity Strong Weak/none
Market risk sensitivity/procyclicality Small Potentially large (although reduced by using stressed data, for example)
Incentive Losses aligned to risks Loss mutualisation and potential moral hazard
Default close-out Uncoordinated bilateral close-out Coordinated auctions
Segregation None Initial margin segregation required

Central clearing (and in terms of many characteristics, bilateral initial margining) is very different and follows a defaulter-pays approach. The main loss absorbency is provided by initial margins. These are based on a short time horizon (e.g. five days for an OTC CCP or 10 days for bilateral margining requirements) and are usually relatively insensitive to credit quality. This can potentially make initial margins much more sensitive to market factors, which in the extreme can lead to procyclicality (which in turn can be mitigated by aspects such as using long time horizons and stressed data periods). Loss mutualisation via default funds (and capital for default funds) is used to absorb large losses, but potentially creates adverse incentives. There are centralised auctions for closing out a defaulter's portfolio, which may be more efficient than close-outs in bilateral markets.

The above table should illustrate that central clearing and initial margining has both advantages and disadvantages which may be difficult to define precisely (e.g. risk sensitivity is probably a good thing, but in the extreme can lead to procyclicality, which is clearly not). What is also clear is that central clearing changes many aspects of OTC derivatives trading and the underlying risks.

Comparison of loss allocation in bilateral (no margins), bilateral with margin, and centrally-cleared markets.

Figure 10.6 Comparison of loss allocation in bilateral (no margins), bilateral with margin, and centrally-cleared markets.

It is also interesting to compare the loss absorbency in different setups (Figure 10.6). In a bilateral market with no margining, this is based only on capital held by each party, whereas in a bilateral market with margins,22 loss absorbency is provided jointly between capital and margins. However, in a centrally-cleared market, a single capital amount is replaced by margins, mutualised default funds, and the associated capital requirements for both these components.23 The important point is that regulation must try to create the correct incentives. For example, the capital requirements in a bilateral market should be smaller as initial margin is posted, or a CCP with a larger default fund should impose lower capital requirements on its members. This is particularly difficult to do, especially with a relatively simple methodology for determining capital requirements. This is covered in more detail in Chapter 13.

NOTES

  1. 1 Out of a total of well over 100, although some of these may have been unable to bid on the underlying portfolio.
  2. 2 Chasan, E. and A. Saphira (2010). Firms reaped windfalls in Lehman auction: examiner. Reuters (15 April). www.reuters.com.
  3. 3 Bunge, J. (2010). CME, Lehman Book Bidders Likely Protected From Lawsuits. Wall Street Journal (15 April). www.wsj.com.
  4. 4 CPSS-IOSCO (2012) states that ‘An OTC derivatives CCP may need to consider requiring participants to agree in advance to bid on the defaulting participant's portfolio and, should the auction fail, accept an allocation of the portfolio. Where used, such procedures should include consideration of the risk profile and portfolio of each receiving participant before allocating positions so as to minimise additional risk for the non-defaulting participants.’
  5. 5 LCH Ltd Default Rules Section 2.5. www.lch.com.
  6. 6 Strictly speaking, the last point at which the defaulted member paid variation margin.
  7. 7 This assumes linearity but, in reality, the close-out of a smaller portfolio may be easier and the resulting risk premium smaller.
  8. 8 EMIR article 45(4): ‘A CCP shall not use the margins posted by non-defaulting members to cover the losses resulting from the default of another clearing member.’
  9. 9 For example, under the EU's Capital Requirement Regulation article 306(1)(c), a clearing member will have to pass on the impact of actions such as VMGH to get appropriate capital relief.
  10. 10 Sourbes, C. (2013). BoE's Carney: liquidity support for CCPs is a ‘last-resort’ option. Risk (28 November). www.risk.net.
  11. 11 Kelly, K. (2010). Collateral Demands Growing for BP. CNBC (28 June). www.cnbc.com.
  12. 12 It could be argued that the credit risk could at least be appropriately priced with respect to the creditworthiness of BP at the time. However, this illustrates that a margin agreement can expose an institution to volatility arising from its own credit risk, which is a risk that may be more effectively managed by banks.
  13. 13 Hirschler, B. (1999). Ashanti wins three-year gold margin reprieve. GhanaWeb (2 November). www.ghanaweb.com.
  14. 14 ‘Mr. Sam Jonah, Chief Executive of Ashanti, referring to their financial problems, stated “I am prepared to concede that we were reckless. We took a bet on the price of gold. We thought that it would go down and we took a position.”’ From O'Connor, G. (1999). Ashanti left exposed by ‘exotics’. Financial Times (8 November). www.ft.com.
  15. 15 ISDA Margin Survey (2013b, Table 4.5).
  16. 16 For example, consider an airline hedging its exposure to the price of aviation fuel via an oil swap or forward contract. The airline will make (lose) money on those hedges in the event of high (low) fuel prices, which will be balanced economically by higher (lower) fuel costs over the period of the hedge. However, when margin is brought into this picture, cash margin against losses on a derivative cannot be funded by cheap fuel price benefits, which accrue over time. The result of this is that the airline may have liquidity problems in relation to using derivatives with associated variation margins.
  17. 17 European Association of Corporate Treasurers (2011). Open Letter to the Commissioners of the European Union (7 November).
  18. 18 Carney, J. (2012). Here's What That Deutsche Bank Trade Was Really All About. CNBC (7 December). www.cnbc.com. This example is of an alleged avoidance of MTM losses that may have otherwise led to a default or government bailout.
  19. 19 Including charges for any initial margins (independent amount) that the MDB may also hold.
  20. 20 Pollack, L. (2012). The Bank of England gets economical with its derivatives. Financial Times (22 June). www.ft.com.
  21. 21 This shares many of the characteristics of central clearing but does not involve loss mutualisation or coordinated auctions.
  22. 22 Variation margin and potentially also initial margin.
  23. 23 There are two distinct types of capital covering trade exposure and default fund exposures, as described in Section 13.6.