FMP 6: Central Clearing

In a bilaterally cleared market every pair of dealers signs a master agreement covering all transactions between them. Standard documentation comes from the International Swaps and Derivatives Association, and the terms set out netting arrangements, collateral arrangements and what happens on a default. Six participants need fifteen such relationships, each with its own credit judgement. Under central clearing those six join one clearing house, positions are transferred to the CCP, and fifteen credit relationships collapse into six.

Neither picture is complete. Non-standard transactions between financial institutions, and a portion of the business done with non-financial end users, stay bilateral, while standard interdealer transactions run through clearing houses. Several CCPs compete for the same flow, so even a fully cleared set of trades may sit at more than one.

Members, non-members and client clearing

Not everyone is a clearing member. Small financial institutions and non-financial companies clear through a member that stands between them and the CCP, posting initial margin and variation margin to it as customers of a futures broker do.

Why interest is paid on variation margin

Futures contracts settle daily, and variation margin belongs outright to whoever receives it. Over-the-counter contracts are different. Their cash flows arrive periodically, and for a European option all of them arrive at maturity. A clearing house still values portfolios daily and moves variation margin, but that money remains the property of the payer until the contract requires payment. So interest is paid: bilaterally from the receiver to the payer, and where a CCP sits in the middle from the paying member to the clearing house and onward. Interest is also paid on initial margin balances.

Source: the structural comparison set out in the chapter. The final column is added here as a study aid.

When a clearing member fails, the CCP replaces its positions and pays whatever that costs, drawing on its resources in a fixed sequence, each layer touched only once the one above it is exhausted.

The first claim is on the initial margin the defaulting member posted. If the close-out costs more, that member’s own default fund contribution goes next. Should a shortfall survive both, the contributions of surviving members are drawn down, and firms that never traded with the failed member begin paying. Only then is the equity of the clearing house at risk.

Auctions, loss allocation and tear ups

Before any of this settles, the clearing house tries to dispose of the position, usually by auction, inviting surviving members to bid for a package offsetting the defaulting member’s book. Members have a real incentive to bid, since a quick close-out protects their own contributions. A failed auction leaves harsher tools: the CCP may allocate losses to members that have recently made gains, clawing back variation margin they received, and it may tear up transactions, closing out contracts entered into with the defaulting party at prices that leave surviving members with a loss.

Initial margin is calculated from historical data, and the question it answers is narrow: how much could be lost if this member defaults and prices move against the position while it is closed out. The cleared convention is five days at 99% confidence, so margin is set such that if closing out takes five days, the clearing house is 99% certain the deposit covers the loss.

Margin is computed on the netted portfolio rather than trade by trade, one reason a member gains by concentrating its clearing in one place. Requirements are recalculated as volatility changes, since stale parameters sank the early clearing houses, and members called intraday must pay almost immediately.

The default fund and equal treatment of members

Default fund contributions are sized separately. Initial margin is calibrated to a member’s own portfolio and covers that member’s failure, while the default fund is a shared reserve for what the clearing house might face beyond individual margin, including several members failing at once.

One feature differs sharply from bilateral practice. A clearing house applies the same rules to every member when setting initial margin and default fund contributions, so credit quality does not enter the calculation as it does for a dealer negotiating collateral terms with a weak counterparty. Letting a CCP grade its own members would import a good deal of subjective judgement, which is why regulators accept the practice.

Before the crisis, variation margin was routine between dealers in the bilaterally cleared market but initial margin was not. A dealer might demand it from a much weaker counterparty, yet two dealers of similar standing posting initial margin to each other was almost unknown. Uncleared derivatives now attract both, wherever the two sides are financial institutions or one side is a systemically important non-financial firm trading in volume.

The uncleared calibration is more conservative

The uncleared requirement runs to a ten-day horizon at 99% confidence, measured in stressed market conditions rather than calm ones. The margin one party posts should cover the largest fall over that period in what the contracts are worth to it, which is the largest rise in what they are worth to the other side. If the poster defaults, the survivor may need up to ten days to replace the positions, and replacement turns expensive if prices run meanwhile. Firms calculate uncleared initial margin using SIMM, the Standard Initial Margin Model, developed by the International Swaps and Derivatives Association with market participants.

Why uncleared initial margin goes to a third party

Variation margin on uncleared trades usually passes straight from one counterparty to the other. Initial margin cannot. Imagine counterparty A sending USD 1 million of initial margin to counterparty B while B sends USD 1 million back to A. The transfers cancel, the net posted by each side is zero, and nothing absorbs a default loss. So uncleared initial margin goes to a third party and is held in trust, out of reach of the poster.

Margin is a cost, and firms price it

Interest is paid on margin, yet institutions still treat it as expensive because the cash could have been used elsewhere. The usual approach compares the interest earned with the firm’s average cost of borrowed funds and books the difference as a charge: for variation margin that charge is the funding value adjustment, or FVA, which can be negative, and for initial margin it is the margin value adjustment, or MVA. That benchmark is debatable, since margin is a fairly safe use of money. Against the cost sits the benefit: inside a functioning clearing ecosystem a financial institution is far less likely to lose money because another one failed, and that benefit arrives only if the whole industry is inside the arrangement.

Very little of the over-the-counter market was regulated before the 2007-2008 global financial crisis. Participants could execute and clear a trade however they liked, and nobody had to tell a central authority it existed. Complex derivatives built from portfolios of subprime mortgages, riskier than average, were widely blamed, and that judgement drove what came next.

The concern was systemic risk: a default by one derivatives dealer inflicts losses on the dealers facing it, those losses push further dealers into default, and the interconnectedness of the whole population eventually brings down the financial system. Meeting at Pittsburgh in September 2009, the G-20 leaders set out to break that chain. Their communique asked that standardized contracts be “cleared through central counterparties by end-2012 at the latest” and traded on exchanges or electronic platforms where appropriate. The same paragraph called for reporting to trade repositories, higher capital against contracts not centrally cleared, and continued assessment by the Financial Stability Board.

The three requirements that emerged

Three obligations came out of the meeting. Standardized derivatives must be cleared through CCPs, a category that covers credit default swaps on indices and plain vanilla interest rate swaps, which make up most traded volume. They must also trade on electronic platforms so prices are visible; these venues are swap execution facilities in the United States and organized trading facilities in Europe, and a product traded there passes straight to a clearing house. Every trade must be reported to a central trade repository, which is how supervisors see the risks being run.

The first two bite only where both sides are financial institutions, or where one side is a non-financial company judged systemically important for its derivatives volume, so dealers may still trade standardized contracts bilaterally with most end users. Even so, forcing interdealer business into clearing houses produced an enormous rise in cleared volume.

Closing the loophole on uncleared trades

Supervisors saw that a dealer could sidestep the clearing obligation by tweaking a contract until it was slightly non-standard. A further G-20 meeting in 2011 extended margin requirements to uncleared derivatives, and those rules were phased in between 2016 and 2020. Bank for International Settlements data show the result. By June 2019, CCPs handled 78% of interest rate derivatives, along with 54% of credit default swaps, and bilateral clearing has shrunk in exactly those two categories.

The whole regulatory structure hangs on the word standard. A standard transaction is one clearing houses are prepared to clear, which pushes the question back a step. Four conditions have to hold before a CCP takes a product on.

The legal and economic terms must be standard across the market, since a clearing house cannot run a book of bespoke contracts that each behave differently. Generally accepted valuation models must exist, because the CCP determines variation margin at least once a day. The product must trade actively, since an inactive market makes it hard to unwind a failed member’s position and hard to obtain up-to-date valuations, and a clearing house will not build systems for a product its members rarely touch. Extensive historical price data must be available, because initial margin is calibrated from history.

What is in the category and what is not

Two families currently qualify: index credit default swaps, and interest rate swaps. Others may join, with swaptions and single-name credit default swaps the obvious candidates, while most exotic derivatives look likely to stay outside indefinitely, since they fail the activity and data tests by construction. Anything cleared bilaterally rather than through a CCP is uncleared, now a regulatory category rather than a casual description.

What a CCP is not allowed to do

Clearing houses serving this market are heavily supervised. The Financial Conduct Authority in the United Kingdom, for instance, monitors the risks taken by LCH Clearnet closely. The governing principle is that a CCP should not take risk unconnected to clearing. Speculative trading on its own account would be inappropriate for an institution the market must face, and the standard formulation is that a clearing house should behave like a public utility.

CCPs cover their costs with a fee per trade, and they can earn a spread between what they make on invested initial margin and what they pay members. Where a clearing house is owned by its members, surplus profit returns to those owner members; where outside investors own it, commercial pressure to grow earnings follows. Competition gives users a choice and pushes each venue to improve its systems. The danger is undercutting on initial margin and default fund contributions to win business, which raises the probability of a failure.

Netting under a master agreement works only inside the pair of firms that signed it. A positive net value creates a credit exposure for the party it favours and a negative one creates none, so across several dealers those exposures add up: a firm cannot set what one counterparty owes it against what it owes another.

Take three parties clearing bilaterally. Transactions between Party A and Party C are worth USD 8 million to A and minus USD 8 million to C, those between A and Party B are worth USD 5 million to B and minus USD 5 million to A, and those between B and C are worth USD 2 million to C and minus USD 2 million to B. The positive positions total USD 15 million of credit exposure. Push it all through one clearing house, and each party nets across both relationships: USD 3 million owed to A, USD 3 million owed to B and USD 6 million owed by C. Exposure falls to USD 6 million.

When central clearing reduces netting instead

The gain is not automatic. Positions at one clearing house cannot usually be netted against positions at another, so a member spread across several CCPs posts initial margin at each. Netting across venues, known as interoperability, is possible in principle but not yet common practice. A subtler loss follows from the standard definition. Where two parties trade both standard and non-standard products, bilateral clearing nets the two under one master agreement; move the standard trades to a CCP and that netting set splits in two, and what is given up can outweigh what is gained.

The most practical advantage is how much easier it becomes to leave a position. A bilaterally cleared trade can be exited only by returning to the original counterparty and negotiating a close-out, which normally involves a payment one way or the other. If that counterparty refuses reasonable terms, the alternative is an offsetting trade elsewhere, and the firm then carries credit risk to two counterparties. Where both trades sit at the same clearing house they cancel and the credit risk disappears; at different clearing houses the benefit is lost.

Loss mutualization is the second advantage, and the one regulators care about. Bilaterally, a default loss lands entirely on the counterparties who traded with the failed firm. Centrally, it is shared among all members, including firms that never dealt with it, and dispersing a shock across many balance sheets is what reduces systemic risk.

Third, a clearing house performs the margining, netting, settlement and default resolution every participant would otherwise do for itself, with large resources and intense oversight behind it. Central clearing also improves liquidity and encourages standard documentation.

The costs on the other side of the ledger

Two problems familiar from insurance appear here. Moral hazard is the tendency of protection to encourage more risk taking, and it shows up because a market participant has less reason to scrutinise the firms it trades with once the clearing house carries that risk. Adverse selection is the tendency to attract worse risks, and it bites whenever a dealer can choose between clearing a trade and keeping it bilateral, as with a standard derivative traded against a non-financial end user.

Central clearing is also pro-cyclical, meaning it deepens bad episodes. Volatile markets are precisely when institutions run short of liquidity, and also when clearing houses raise initial margin requirements and call for larger default fund contributions. The demand for cash arrives at the worst possible moment.

A further disadvantage is opacity. A clearing member cannot easily judge the credit risk it runs, because its default fund contribution, and possibly some of its variation margin gains, depend on members whose trades it cannot see. Bilateral clearing concentrates exposure but at least the firm understands it. The clearing house has the opposite vantage point: it sees every member’s whole portfolio and can respond to excessive risk taking by limiting trading or raising initial margin. Against that sits concentration itself, since an operational problem at a CCP touches far more transactions than one at a dealer.

One reading of the reforms is that banks too big to fail were swapped for clearing houses too big to fail, and the scale is real. At the beginning of 2020, SwapClear at LCH Clearnet had a cleared interest rate swap book whose notional principal exceeded USD 300 trillion, and a failure there would require a rescue, most likely by the British government. The counterargument is that a CCP is a far simpler organisation. It admits members, values transactions, sets initial margin and default fund contributions, and runs the netting systems, while a bank does all that alongside lending, proprietary trading, funding decisions and a wide range of operational risk.

The record supports that. When Lehman Brothers filed for bankruptcy in September 2008, the largest in United States history, clearing houses closed out its positions within days, while disputes over its bilaterally cleared transactions ran on for years at great expense.

Correlated defaults, model risk and liquidity risk

The most serious problem is that member defaults are positively correlated. A member failing because economic conditions have deteriorated is unlikely to be the only one, which is why supervisors require clearing houses to model simultaneous failures and to run stress tests against imagined adverse events. Auction risk follows: a failed auction forces surviving members to share losses, which can trigger further defaults, damage the reputation of the venue and prompt resignations. A member must normally eliminate its exposures and give one month of notice before leaving.

Model risk weighs more heavily on an over-the-counter clearing house. Exchange contracts are priced from observable quotes and margined under standard rules. Contracts in this market run for longer, vary more in their terms, offer less price transparency and trade only now and then, so the CCP leans on valuation models for both variation margin and initial margin, and poorly behaved models invite member disputes.

Liquidity risk arises from investing margin. A clearing house sitting on tens of billions in initial margin trades off return against liquidity, since Treasury bills yield less than corporate bonds while corporate bonds are harder to sell. Because member defaults arrive with turbulent markets, any holding must be assessed under stressed conditions. Fraud, computer systems failure and hacking, litigation costs and losses on invested margin complete the list.

How the risks differ for members and non-members

A clearing member faces the clearing house directly. It posts initial margin, funds a default fund contribution that another firm’s failure can consume, faces loss allocation against its variation margin gains and possible tear ups, and must meet intraday calls almost immediately. It also carries the risk that the CCP itself fails. A non-member reaches the clearing house through an intermediary, posting margin to the clearing member it trades through, and if that member fails it depends on whether its positions and collateral can move elsewhere. So the member carries mutualised loss and CCP failure risk, the non-member the risk of its intermediary.