CR 18 – Central Clearing

Most derivatives that change hands every day are traded on an exchange, a central marketplace where participants trade contracts with fixed, uniform terms, futures and options being the obvious examples, at quoted prices. Gathering all the buyers and sellers in one place makes trading more liquid and prices more efficient. A contract rarely starts life on an exchange, though. It usually trades privately first, and only after it has built up enough volume, enough standard terms and enough liquidity does it make the move to a listed venue.

Futures exchanges go back at least to the nineteenth century. A future is an agreement between two parties to buy or sell a set quantity of an asset on a later date at a price fixed today, and it was invented so that merchants and producers could lock in prices and protect themselves against adverse moves. The early exchanges were little more than meeting places. They did not settle trades or manage counterparty risk, so every deal remained a bilateral obligation between the two traders, and membership of the exchange served mainly as a badge of reliability. A member who failed to honour a contract was declared in default and could be fined or thrown out.

What an exchange provides

Three services sit at the core of any exchange. First, it standardises products: maturity dates, the minimum price tick, the grade of the deliverable asset and the delivery location and process are all fixed in the contract design, leaving price and quantity as the only things to negotiate. Second, it supplies a trading venue, physical or electronic, open only to approved firms and individuals who agree to follow its rules, and this concentration of orders is what produces price discovery, the process by which the interaction of buyers and sellers sets the price of an asset. Third, it reports transaction prices to participants, data vendors and subscribers, which makes prices visible to the whole market.

From direct clearing to complete clearing

Risk reduction on exchanges developed in three stages. The first stage was direct clearing: when a party held positions with the exchange that offset each other, the two were simply cancelled, and only the price difference changed hands. Paying the difference instead of delivering the asset became standard in futures markets precisely because it shrinks the amount that any one party can fail to pay, and standard contract terms are what make two positions interchangeable (fungible) enough to cancel. Direct clearing is the exchange equivalent of netting in over-the-counter (OTC) derivatives.

The second stage was the clearing ring, in which three or more members could ring out positions that offset across the group. The catch was that each participant had to accept a substitute counterparty in place of the one it had originally dealt with. Portfolio compression in today’s OTC market performs a similar job. Even with direct clearing and rings, however, members still owed obligations to one another, so the failure of one member would land losses on everyone who had traded with it, even though trading on the exchange was anonymous.

The third stage, complete clearing, began in 1891. Here a central counterparty (CCP) is inserted between all the trading parties and takes over every contractual right and obligation, so that each member faces the CCP alone. Adoption was slow. The Chicago Board of Trade (CBOT) ran without a CCP for around 30 years, finally adopting one in 1925, partly because the government pushed it. The London Metal Exchange was among the last futures exchanges to adopt one, in 1986, again with regulators pressing. One explanation for the reluctance is that clearing makes counterparty risk uniform: strong members lose the advantage of their better credit and effectively subsidise weaker ones. That voluntary adoption was so slow suggests that, in some markets at least, the costs of clearing may have outweighed its benefits. Even so, every exchange-traded contract is now centrally cleared, whether the CCP is run by the exchange or provided to it as a service by a separate company.

Exchange-traded markets adopted central clearing long ago, but bilateral OTC derivatives stayed with a bilateral model until the late 1990s, when SwapClear was launched within LCH.Clearnet (now LCH Ltd) to clear interest rate swaps between the major dealers. Regulatory reform after the global financial crisis has since made OTC clearing far more common. A centrally cleared OTC trade keeps some OTC features, since it is still agreed privately between two parties, but it then passes through the same central clearing machinery that serves listed contracts. A product does not need to be liquid enough to list on an exchange to be cleared, yet it does need a reasonable degree of standardisation and liquidity and must not be too complex. So, many OTC products will never be clearable, and some participants will stay outside clearing because of obstacles such as funding liquidity risk. Even so, a large share of OTC volume is in clearable products traded by participants who can clear.

Five conditions for a product to be clearable

Standardisation. The CCP becomes legally responsible for every cash flow of a cleared product, so the legal and economic terms must be standard. Standard terms also allow identical contracts to be offset and compressed, and they make replacement easier if a member defaults. Before 2009, for instance, CDS contracts were given standard coupons and maturity dates, and that change had to happen before any CDS could move to a CCP.

Complexity. Only vanilla products can be cleared, because the CCP must be able to value them quickly and reliably every day to set margin and to run a default. Exotic products are harder: valuation for variation margin becomes subjective and initial margin is very hard to calibrate, so the CCP ends up either under-margining, which leaves too much risk, or over-margining, which makes clearing expensive. A product can have perfectly standard terms and still be too complex to value robustly.

Liquidity. A liquid product gives the CCP reliable prices for variation margin and a long history of price data for calibrating initial margin. It can also be replaced cheaply after a default, whereas illiquid trades carry wide bid-offer spreads and can attract large risk premiums when markets are stressed. Liquidity in OTC markets is often temporary and tends to fade as a product ages, yet the CCP guarantees each position until it matures.

Wrong-way risk. Wrong-way risk (WWR) arises when a counterparty’s default is adversely linked to the value of the contract, so that the contract loses value just as the counterparty fails. That combination is particularly harmful in default management. CDSs illustrate the tension: the clearing mandate was largely prompted by them, yet their WWR arguably makes them the hardest products to clear.

Market volume. Building the models, processes and systems to clear a product is expensive, so the market must be large enough for the CCP to earn back its development cost.

A central counterparty is a financial market infrastructure (FMI): a body of rules and operating arrangements built to spread, control and shrink counterparty risk in a market that would otherwise be bilateral. It does this by changing the shape of the market. In a purely bilateral market of, say, six large dealers, each dealer may hold trades with every other dealer, producing a dense web of connections. When a CCP stands between buyers and sellers, that web collapses into a hub and spokes: each dealer has one connection, to the CCP.

Two benefits follow at once from this picture. The market becomes less interconnected, which can soften the shock when one participant becomes insolvent, and the CCP, sitting at the centre, can see members’ positions and so make the market more transparent.

Why the simple picture is incomplete

The hub-and-spoke diagram is broadly right but leaves out three features of real markets. Client clearing comes first: firms that cannot become members of a particular CCP must clear through a clearing member, which complicates both the movement of margin and the handling of any default. Next, bilateral trades persist, because not every OTC derivative is suitable for clearing and a sizeable population will stay bilateral. Finally, there are many CCPs around the world, and they may be linked implicitly, for example through members they share. A realistic map therefore shows dealers connected to the CCP, some dealers still trading with one another bilaterally, and clients reaching the CCP only through the dealers that clear for them.

The hub structure has an obvious weakness as well. A CCP at the centre is a single point of failure, and concentrating everything in one place removes the diversification that comes from many independent decision makers. A serious mistake or failure at the CCP itself, as distinct from the failure of one of its members, could be catastrophic for the whole market.

Clearing costs money, and CCPs recover it directly through per-trade fees and indirectly through sources such as interest earned on the assets they hold. Because CCPs are nodes of the financial system, they must stay resilient above all during severe market stress, which argues for a utility model focused on long-term stability rather than short-term profit. The opposite case is also strong: a CCP needs excellent staff and systems to run advanced risk management, and competition between CCPs gives users choice. Both points favour profit-making CCPs, but competition for business can turn into a race to the bottom on practices such as margin calculation, and that would make CCPs riskier.

The number of CCPs involves a similar trade-off. Many CCPs maximise competition but risk the same race to the bottom on cost and split the benefits of multilateral offset across venues. A few CCPs deliver more offset and economies of scale but concentrate systemic risk. One global CCP would be best on several counts, yet the realistic outcome is a fairly large number, because the market fragments in two ways. Regions want their own local CCPs, either for trades in their currency or for all trades of institutions based there, and some regulators insist that supervised institutions use the regional CCP for certain products. Products also fragment the market, since OTC CCPs have tended to specialise vertically, in interest rate swaps or CDSs for example, so no single CCP clears every clearable product, and regulation does nothing to encourage one that would.

Bifurcation and the limits of mandatory clearing

A clearing mandate can add to systemic risk rather than remove it. Large banks must clear their standard trades but keep their non-standard trades bilateral, which breaks up the netting that used to apply across the two groups and can increase net exposure. Exemptions for end users and for foreign exchange can produce further distortions and openings for regulatory arbitrage. How far this bifurcation goes depends on how much of the market can be cleared in practice; the International Swaps and Derivatives Association (ISDA) puts the clearable share at about 80% of OTC derivatives. ISDA’s own 2013 figure for the non-clearable share is 13%, but that number uses the total market size as its base, including the double counting created when trades are novated to CCPs, which is why the two figures do not add to 100%.

Vertical and horizontal structures

Ownership tends to follow one of two patterns. In a vertical structure the CCP is a division of an exchange, owned by it, and clears only products traded on that exchange; futures markets developed this way as exchanges built in their own clearing function. In a horizontal structure the CCP is a separate company, typically owned by its clearing members and financially independent, so it can clear across many markets and asset classes. Horizontal clearing suits bilateral OTC derivatives naturally, since no exchange is involved. Regulators lean towards the competition that horizontal clearing permits and ask CCPs to be open. Under Article 29 of MiFID II, for example, the European Commission proposes giving CCPs access rights to trading venues, so that participants could clear with a CCP other than the one an exchange or platform designates, in the hope of lower costs and better service.

Much of the OTC market will be centrally cleared over the coming years, and a large part already is, which is practical because a few clearable products, interest rate swaps above all, account for a very large share of outstanding notional. Extending clearing to new products is slow, however, because each must meet the conditions described earlier.

LCH Ltd was the first CCP to clear OTC trades in significant volume, through SwapClear, which dominates clearing of interbank interest rate swaps. Other major OTC clearers include ICE (Intercontinental Exchange), whose ICE Clear Credit arm handles some CDS clearing, along with Eurex and the Chicago Mercantile Exchange (CME).

Novation is the legal step that puts the CCP in the middle of a trade. In general terms it means replacing one contract with one or more new contracts. When a trade between a buyer and a seller is novated to a CCP, the original contract is extinguished and two new ones take its place: one between the seller and the CCP and one between the CCP and the buyer. In effect, the CCP insures each side against the default of the other.

Novation only works if the new contracts are legally enforceable and if it is certain that the original parties no longer owe each other anything once it is complete. Provided those conditions hold, the two original traders have no counterparty risk to each other at all. The CCP usually guarantees a trade from the moment it is matched, becoming the buyer to every seller and the seller to every buyer. One practical result is that a clearing member need not assess the credit of whoever it originally traded with, and on electronic platforms the parties may never learn each other’s identity, so clearing can make trading anonymous.

Matched book and conditional market risk

Because every position the CCP takes on with one member is offset by the opposite position with another, the CCP runs a matched book and carries no net market risk. What it does carry is the counterparty risk of all its members, now gathered in one place. Its market risk is conditional: if a member defaults, the trades the CCP had with that member disappear from the hedge side, and the CCP is suddenly left holding the other half of those trades with an open market position. A CCP has several ways to restore a matched book, the main one being an auction of the defaulter’s positions among the surviving members. It also protects itself in advance by requiring members to post financial resources designed to cover the losses it would face if one or more of them failed.

Bilateral trading tends to pile up overlapping and often redundant contracts, because many banks and end users trade with one another for different reasons and with different business models. That redundancy raises counterparty risk and ties institutions more tightly together. Central clearing attacks it through multilateral offset: once every trade faces the CCP, a member’s obligations to and from all other members can be set against each other, whether the quantity being offset is cash flows or margin requirements.

Multilateral offset compared with compression

Portfolio compression can also produce netting-like effects across several counterparties, but it works only by actually terminating trades or cash flows. Multilateral offset at a CCP is broader. Suppose two swaps almost cancel but have slightly different reference rates or maturities. Compression may be unable to remove them, yet the CCP will still recognise that the combined market risk is small and set a correspondingly low initial margin. In short, a CCP can offset risk itself, whereas bilateral compression can only remove objective items such as cash flows. The simple three-member picture also ignores multiple CCPs, trades that remain uncleared and non-derivative positions, all of which can erode the offset in practice.

Compression services still have a role once trades are cleared. At first sight that seems odd: compression is designed to leave a portfolio’s market risk unchanged, so it would not normally be expected to lower margin. The gain comes from cutting the number of trades and the gross notional held at a CCP, which brings several efficiencies:

• initial margin, because some CCPs apply multipliers to large portfolios;
• regulatory capital, notably the leverage ratio, which is driven by gross exposure;
• operations, through fewer cash flow payments and less reporting;
• computation, since margin calculations run on fewer trades;
• default management, because an auction of a smaller portfolio has fewer trades to close out.

Compression is therefore complementary to clearing rather than a substitute. Published data for interest rate swaps cleared through SwapClear show the pattern clearly: the amount of notional that has passed through the service rises steadily, while the notional actually outstanding stays roughly flat, because a large and growing volume is compressed away. Compression is spreading to more products and counterparties, and newer techniques are appearing, such as blended-rate compression, which can compress any number of positions with different fixed rates provided their remaining cash flow dates are the same.

Unlike a bank, which leans mainly on capital to absorb losses, a CCP relies first on margin. Since complete clearing arrived, CCPs have collected both variation margin and initial margin as their main protection against a clearing member becoming insolvent. Centrally cleared markets have traditionally imposed standardised and stricter margin than bilateral markets, although the BCBS-IOSCO margin rules for uncleared trades are closing that gap.

Variation margin

Variation margin pays over each day’s change in value, so the CCP needs timely and reliable prices for every cleared product and, where prices cannot be observed, valuation methods that the market accepts as standard. Those methods change with practice; the move from Libor discounting to overnight indexed swap (OIS) discounting for interest rate swaps is a well-known case, and in 2010 LCH.Clearnet switched its $218 trillion swap portfolio to OIS discounting. Variation margin is called daily and sometimes intraday, particularly in volatile markets or after a large move, and it is almost always paid in cash because it functions, directly or indirectly, as settlement. A member’s variation margin and settlement amounts across its clearing activity in one currency can be netted into a single daily payment or receipt, but different currencies are usually not netted against each other.

Initial margin

Initial margin protects against the losses the CCP might suffer in closing out a defaulted member’s portfolio. It is calculated from scenarios of price moves over an assumed close-out horizon, called the margin period of risk (MPoR) or liquidation period and often taken as five days, and is set at a high confidence level of at least 99%. It stays in place for the life of the trade and moves up or down as market conditions and remaining risk change. Initial margin need not be cash: eligible collateral can include major-currency cash as well as treasury and agency securities issued by governments. After haircuts, the collateral should leave the CCP with little credit, market or liquidity risk, which means securities must be priced frequently, easy to sell and of low credit risk.

Interest on margin

CCPs hold very large amounts of margin, and most pay members interest on excess cash posted as initial margin, typically linked to a short-term benchmark, for example FedFunds minus 10 basis points (bps). To fund those payments the CCP must earn something on the cash, but it must invest with great care so as not to add risk: deposits at central banks or commercial banks, reverse repos, or other holdings that carry minimal credit and liquidity risk.

When margin is large enough to cover a member’s default, clearing follows the defaulter-pays principle: the failing member has already funded the cost of its own failure. Making that true in every conceivable scenario would demand enormous contributions from each member, far too costly to be practical. So, a member’s own resources are sized to cover its default to a high level of confidence, which leaves a small probability that losses exceed them and fall on the others.

That residual loss is handled by loss mutualisation, the second basic principle of clearing. Every member pays into a default fund, also known as a guarantee or reserve fund, which is normally drawn on after the defaulter’s own resources are exhausted. Because all members contribute, all of them share in absorbing an extreme loss. A loss that consumes a large part of the default fund is meant to be very unlikely.

One consequence surprises people new to clearing. The loss a member bears from another member’s default has nothing to do with whether the two ever traded. A member can lose money through the default fund even if it never dealt with the defaulter, has no net position at the CCP, or holds a net position in the same direction as the defaulter. Some loss allocation rules, discussed later, can tilt the outcome in favour of such members, but the default fund itself is shared.

Only clearing members deal directly with a CCP, and the requirements for becoming one fall into three groups. Admission criteria come first, such as a strong credit rating and a sufficiently large capital base. A financial commitment follows, because members must pay into the default fund; contributions partly track trading activity but often have a minimum, so only firms expecting substantial volume will find membership worthwhile. Operational demands complete the list: members must post liquid margin frequently, take part in fire drills that rehearse a member default and bid in auctions if a member actually fails.

As a result, clearing members are mostly large global banks and a few other very large financial institutions. Smaller banks, buy-side firms and non-financial end users generally do not join, while large regional banks may join only a local CCP in order to offer domestic clearing to their customers.

How clients clear

Firms that are not members, known as non-clearing members or clients, clear through a clearing member. Arrangements vary, but the usual rule is that the client has a bilateral relationship with its clearing member rather than with the CCP. The member typically mirrors the CCP’s terms in that relationship, especially on margin. Clients post margin but do not contribute to the default fund. The member charges them, explicitly and implicitly, for the service, including its cost of supporting the default fund, and members often ask clients for more margin than the CCP requires. Members can earn a return on margin held above the CCP’s requirement, which compensates them for the service and the added risk, but they must also account for the capital and leverage ratio cost of client clearing.

A client may use more than one clearing member. One reason is portability: the client wants the option to move its portfolio to another member, and in a default that is precisely the goal, so that the client’s trading can continue.

Whether a client’s portfolio can actually be ported, and what risks the client bears, depends on how its margin travels from client to clearing member to CCP and how it is segregated along the way. Depending on the arrangement, a client may be exposed to the CCP, to its clearing member, or to its clearing member together with that member’s other clients. Two broad account structures are used.

Fellow customer risk, the risk of losing margin because another client of the same member failed, only bites when the clearing member defaults together with one or more of its other clients. While the member survives, client defaults are the member’s problem and it must make them good. Other details also matter: whether margin is segregated by value or the actual assets are kept apart, whether separation is physical or only legal, whether omnibus margin is posted gross or net, and what happens to any excess. In the US, individual segregation is delivered through the model known as legally segregated, operationally commingled (LSOC), under which client assets are protected legally but are not held physically apart.

A CCP is legally bound to perform on every contract it clears, so counterparty risk in a cleared market is concentrated in the CCP. Members no longer need to watch each other’s credit quality; what they are exposed to is the combined credit quality of the membership as a whole, which shifts the spotlight onto how well the CCP itself is run and how resilient it is. (Some loss allocation methods do leave members sensitive to the credit of particular counterparties, but only at the margin.)

To mitigate counterparty risk the CCP carries out three linked functions. Through multilateral offset, it nets or compresses trades that began with different counterparties because they all now face one centre. For loss absorbency, it collects variation and initial margin to cover default losses and holds default fund contributions for extreme losses beyond margin, with both depending on volatility, correlation and how large the cleared portfolio is. In default management, it handles a member’s failure by hedging and auctioning the positions, allocating any loss above the margin held, and porting client trades to solvent members.

Membership requirements

Membership standards are the first line of defence, keeping out firms that would bring undue risk, though they must not become so strict that they are anti-competitive. They focus on three things: creditworthiness, meaning the member’s probability of default as judged by external ratings or internal assessment; liquidity, meaning its ability to meet margin calls at short notice; and operational capacity, meaning its ability to follow CCP rules such as those governing auctions. Members also face minimum capital, default fund contributions and a duty to take part in default management.

A member default is the main risk a CCP faces, because it breaks the matched book. The CCP must keep paying variation margin to surviving members whose positions have gained, but the variation margin that would have funded those payments was due from the member that has just failed. A CCP has far more authority over this situation than a bilateral counterparty would. Its rules let it declare a member in default and then manage that member’s portfolio, so it can halt the member’s trading, close out or transfer its positions and those of its clients, and sell the securities it holds as margin. It neutralises its exposure in two steps.

Macro-hedging. The CCP first cuts the portfolio’s sensitivity to the main risk factors, such as interest rates, using liquid hedging trades, which reduces the market risk it is carrying. A default management group decides which hedges to apply and how.

Auctions. The CCP then auctions the defaulter’s portfolio to the surviving members, usually split into sub-portfolios, for instance by currency, in whatever way the group judges best. Surviving members submit two-way prices and the best bid wins each sub-portfolio. Members have a shared incentive to bid well, because a successful auction avoids losses reaching the default fund, and the rules may add individual incentives, for example by allocating losses to losing bidders before winning ones. Because each sub-portfolio is auctioned together with its macro-hedges, it has little directional risk, which makes it easier to price sensibly even in volatile markets.

The default management group combines CCP staff with senior traders seconded from member firms on a rotating basis, with a firewall between those traders and the rest of their firms. To keep the process sharp, the CCP runs fire drills, perhaps twice a year, in which members submit prices on test portfolios, and new members must pass a driving test proving they can process, price and bid on large portfolios within a few hours.

Porting client positions

Client positions are best moved to a surviving clearing member, and many clients arrange backup members in advance for exactly this purpose. Porting only works if the receiving member accepts both the portfolio and its margin, which depends on how client margin was charged and segregated. If porting fails, the client trades are handled in the default process alongside the defaulter’s own portfolio, and the client’s initial margin is used to cover losses to the extent its segregation allows.

Lehman Brothers at SwapClear

Lehman Brothers defaulted on Monday 15 September 2008 with an OTC interest rate swap portfolio of $9 trillion at SwapClear. A default management group of seconded traders from member firms was formed and worked with LCH.Clearnet’s risk team under strict confidentiality. Macro-hedges were applied and reviewed daily, and 90% of the risk was neutralised within one week. Competitive auctions of five sub-portfolios, one per currency, then ran between Wednesday 24 September and Friday 3 October, which meant the full set of 66,000 trades had been sold within three weeks. Only around a third of Lehman’s initial margin was used, and the remainder was returned to the administrators. A later default at Nasdaq went less well and consumed a substantial part of the default fund.

The margin period of risk (MPoR) is the time between the last margin a defaulter posted and the moment its risk has been fully removed. A CCP can make this period shorter than in bilateral markets for several reasons. It calls margin daily and possibly intraday, in cash only, so there are no settlement delays. It has full control over all calculations, so disputes cannot hold things up. It makes sure members can meet the operational demands of posting margin, including posting on behalf of their clients if necessary. Its privileged position under bankruptcy law lets it declare a default without outside obstruction and move straight into its default process, whereas in a bilateral market the bankruptcy administrators may challenge how trades were closed out or hedged. For example, if a party hedges a defaulter’s portfolio and the hedges lose money while the portfolio gains, the defaulter is likely to argue that it should not bear the hedge losses.

The CCP’s MPoR splits into three periods. During the pre-default period of no action, from the last margin posting until the CCP acts, the clock runs for less time than in bilateral markets because a CCP can declare default easily and with little risk of legal challenge. During macro-hedging, liquid hedges neutralise the key sensitivities quickly and overall risk drops sharply, although basis risk and bid-offer costs remain. During the auctions, risk falls gradually to zero as each sub-portfolio is sold, though the CCP may pay a cost to get positions taken off its hands.

In the Lehman case the whole sequence took three weeks, but most of the risk was hedged long before that. Like its bilateral counterpart, the MPoR used in risk models is best treated as a stylised parameter rather than a literal date, since risk decays progressively. For OTC derivatives, CCPs generally assume five business days, against a minimum of 10 days in bilateral markets, and this five-day period underlies CCP initial margin models. For exchange-traded derivatives, which are more liquid, one or two days is more common.

When a member defaults, its initial margin and default fund contribution are available to pay for macro-hedging and the auction. If they fall short, or the auction fails, the CCP has further resources and loss allocation tools, and the loss waterfall sets the order in which they are used. (Variation margin is not a layer because it already offsets the portfolio’s value, although any extra variation margin the CCP holds counts as a defaulter-pays resource.)

The defaulter-pays layers come first: the initial margin and default fund contribution of the member or members that failed. Next there is usually a tranche of the CCP’s own equity, known as skin in the game, which gives the CCP a direct reason to avoid losses that run past the defaulter’s resources. Not every CCP’s rules include it. When an algorithmic trading error caused a futures broker on the Korea Exchange (KRX) to default in 2014, KRX charged a $4.3m loss to the members’ default fund without losing anything itself, because its rules put its own capital at risk only after the whole default fund of around $200m was gone.

Once those layers are used up, losses hit the rest of the default fund. This is survivors pay, since the money belongs to the surviving members. The loss can be shared pro rata or by rules designed to encourage active bidding in the auction, for example by allocating losses to winning bidders last.

Losses large enough to wipe out much of the default fund are meant to be exceptionally rare, but if one occurs, a CCP has further tools that vary from one CCP to another.

Rights of assessment oblige members to make additional, unfunded contributions to a depleted default fund. Recapitalisation is typically triggered once a significant fraction of the fund, such as 25%, has been consumed. Rights without any limit rights would make CCP failure almost impossible short of every member failing, but they would create severe moral hazard, give members unlimited exposure and could destabilise markets by demanding cash at the worst moment. So, rights of assessment are usually capped at a maximum amount over a given period.

Probably the most common alternative tool is variation margin gains haircutting (VMGH). It relies on the matched book: because the book was matched, the defaulter’s losses must correspond to gains on the opposite positions held by other members. Under VMGH, members whose positions have risen in value since the default receive only part of the variation margin owed to them, while members who owe money must still pay in full. This mimics the economics of a bilateral market, where a counterparty’s default would also cost those who were owed money. A member being haircut may not be in profit overall, since its cleared gain may offset losses on bilateral trades, positions at another CCP or client trades. If the CCP can auction the defaulter’s portfolio at mid-market prices, VMGH is guaranteed to work; it fails mainly when close-out happens at a large premium to mid-market.

Tear-up is used when the exposure cannot be neutralised by auction: the CCP terminates contracts with surviving members that offset the defaulter’s unmatched trades, returning itself to a matched book. The trades torn up may be the original opposite trades or any suitable ones, and all other contracts, possibly the vast majority, are left alone. Unlike VMGH, tear-up spares the CCP any risk premium in an auction. It can be partial or full; a full tear-up ensures the CCP survives, but members are left with losses and with the market risk the CCP could not hedge.

Forced allocation requires members to take on portfolios at prices set by the CCP. The effect resembles tear-up, since imposing the reverse trade on a member is economically close to tearing up an existing one. It can fall more randomly, because any member can be allocated positions, whereas tear-up needs the member to hold suitable trades already, and unlike tear-up a member cannot pass the impact on to its clients.

After loss allocation

Some methods, tear-up and forced allocation among them, are potentially unlimited: the CCP never fails because it can push any loss onto its members, which is severe enough to bring members down. Where allocation is finite, the CCP’s remaining capital is used next, and if losses continue, the CCP fails unless it obtains outside liquidity support such as a central bank bailout. Some CCPs add credit lines from banks or guarantees from insurers; such liquidity is costly to keep on tap but may be justified given the exposure to a severe price move or to a default followed by members withdrawing. A CCP can also impose losses on members through the default fund long before it is near failure itself, as the Nasdaq case showed.

None of these methods lets a CCP use the initial margin of non-defaulting members. Haircutting initial margin has been proposed (Elliott 2013) but never used and may not be permitted, and Article 45(4) of the European Market Infrastructure Regulation forbids a CCP from using survivors’ margin to cover another member’s default. In theory, then, initial margin is risk free because it never enters the waterfall, but with tear-up potentially unlimited, since the cost of rehedging torn-up positions has no upper bound, that is cold comfort for a surviving member.

Initial margin is normally set to cover a member’s default with high confidence, such as 99%, yet breaches do happen, and when they do the losses can be very large. Bates and Craine (1999) estimated that after the 1987 crash, expected losses given that a margin call had been breached rose by an order of magnitude. Default losses are heavy-tailed, so covering the region beyond initial margin needs far more money than the margin itself. That is the classic insurance problem, and pooling is the only practical answer: the default fund is shared among members and spreads the extreme loss from one failure across all of them. Mutualisation can dampen systemic problems, but it brings risks of its own.

Because it is mutualised, a default fund gives far more coverage per unit contributed than initial margin, and that is what keeps clearing affordable, since paying for one’s own default in every scenario is not realistic. A default fund contribution can look small next to initial margin and still provide much more loss absorbency.

Smaller initial margin with a correspondingly larger default fund is cheaper but raises moral hazard, because default fund losses become more likely and the defaulter-pays principle holds less often. Clients matter here too. They do not contribute to the default fund, so they pay for their own risk only through the initial margin that the CCP sets for their clearing member and the member passes on. A large default fund therefore mutualises client defaults as well as member defaults. It also discourages portability, because without enough initial margin attached to client positions, the member receiving them may have to pay more into the default fund.

So, weighting contributions towards initial margin encourages better behaviour, while weighting them towards the default fund gives more total loss absorbency and cheaper clearing at the price of moral hazard.

Losses reaching the default fund should be rare, consistent with the confidence level of initial margin, and history largely bears this out. LCH reports seven defaults over its history, all handled within the defaulter’s initial margin with no effect on other members or on markets. When default funds are needed, though, losses can be large. In 1987 the Hong Kong Futures Exchange lost far more than its default fund, and Nasdaq more recently suffered default fund losses. Such losses may become more likely as CCPs grow and clear more complex and risky products.

Sizing the fund is hard precisely because it covers extreme events. Estimating exposure beyond initial margin runs into fat tails, complex dependencies and wrong-way risk, and the probability that a CCP exhausts its fund cannot be measured with any accuracy, since it depends on member defaults coinciding with extreme price moves and illiquidity. So, CCPs are generally required to size the total fund through predefined stress tests and then allocate it fairly simply, for example pro rata to initial margin, perhaps averaged over time, or to total position size, possibly with a floor. Contributions are updated less often than margin, monthly for instance.

The total is usually expressed as the number of member defaults the CCP can survive, typically one or two. The CPSS-IOSCO (2012) principles ask a CCP to hold enough additional resources to survive, under extreme but plausible conditions, the failure of whichever participant, together with its affiliates, would generate the biggest aggregate credit exposure; this is Cover 1. A CCP with a more complex risk profile, or one that is systemically important in several jurisdictions, should be able to survive the default of the two such participants with their affiliates; this is Cover 2. SwapClear, for example, sizes its fund to cover the default of its two largest clearing members using extreme but plausible historical and theoretical stress scenarios built for interest rate markets.

The step from one to two defaults matters. Heller and Vause (2012) looked at dealers and concluded that covering the joint failure of the two most important interest rate swap or CDS dealers, instead of only the single most important one, could require default funds about 50% larger. It is unsurprising that OTC default funds are expected to be robust: they would be hit only in a serious crisis in which large, systemically important members had failed, markets were in turmoil and the failed firms were members of several CCPs at once. Only very large default funds can secure a CCP’s integrity in that situation.

A CCP accepts only products that are liquid, standardised and free of exotic features, whereas any product can be traded bilaterally. Multilateral netting at a CCP can be more efficient than bilateral netting and compression, CCPs impose strict margin (although bilateral margin rules are tightening), and a CCP runs a single coordinated auction where bilateral markets replace positions piecemeal. The main cost of loss absorbency also differs: bilaterally it has mostly been capital, though new initial margin rules are changing that, while in cleared markets it is initial margin plus default fund contributions, with capital requirements generally low.

A CCP can make a market more transparent and safer, with more fungible contracts and better liquidity. Its main advantages are these.

• Transparency. Facing a member on a large share of its trades, the CCP can see concentrations hidden in bilateral markets and act on them, for instance by limiting further trading. That visibility can calm the panic that arises when nobody knows who is exposed to whom. The benefit should not be overstated: exchange-traded transparency and margin did not stop Leeson’s speculative positions from bringing down Barings, and the later rogue trading losses at Societe Generale (2008) and UBS (2011) also arose on listed products.
• Multilateral offset. Trades with different counterparties offset at the CCP, which makes it easier to enter and exit positions and lowers costs.
• Loss mutualisation. Losses beyond the defaulter’s resources are spread across members, reducing the hit to any one firm and the chance of systemic contagion.
• Legal and operational efficiency. Centralised margining, netting and settlement can cut costs, and common rules can reduce legal risk.
• Liquidity. Easier trading and multilateral netting can deepen liquidity and lower barriers to entry.
• Default management. A well-run central auction can disrupt prices less than many firms scrambling to replace positions during a crisis.

Clearing was confined to listed derivatives for more than a century, while bilateral OTC markets outgrew exchanges over the last two decades. OTC products are harder to clear because they are less liquid, longer dated and more complex. Mandatory clearing therefore draws criticism.

• Moral hazard. As in insurance, members have little reason to monitor one another’s credit or manage counterparty risk carefully when the CCP bears most of the risk.
• Adverse selection. Large banks that specialise in OTC derivatives may understand the risks better than the CCP and send it the products it underprices; anecdotally, participants already know which CCPs are cheaper for pay-fixed or receive-fixed swaps.
• Bifurcation. Clearing standard products while others stay bilateral splits netting sets, undoing multilateral offset benefits and raising costs.
• Procyclicality. Raising margin or haircuts in volatile markets drains liquidity when it is scarcest, and the more frequent, uniform margining of a CCP can amplify this compared with more flexible bilateral practice.

Several CCP functions can be delivered by other means: trade compression gives netting benefits, bilateral markets can and do use margin, and trade repositories can supply transparency.

Since CCPs mitigate counterparty risk, a natural question is whether the clearing mandate, together with bilateral margin rules, will gradually shrink credit value adjustment (CVA), along with the related funding (FVA) and capital (KVA) value adjustments, until they matter much less than before. The answer is largely no. These adjustments arise mainly from uncollateralised OTC derivatives with end users, who are exempt from the mandate and would clear only voluntarily, which is unlikely since most of them struggle to post margin.

A typical breakdown of a bank’s CVA shows that only about 15% comes from banks and other financial institutions, the part most likely to fall as clearing and bilateral initial margin spread. The other 85% comes from corporates and governments, exempt counterparties that mostly cannot or will not clear. The uncollateralised trades that dominate xVA are here to stay. Clearing and new margin rules may lower CVA in some cases while raising other components, notably margin value adjustment, which comes from funding initial margin, so xVA has to be assessed as a whole, including, increasingly, the CVA of exposure to a CCP itself.