Too much choice: The problems with Europe's plethora of segregation models

In the US, segregation of client assets is a simple matter – only one approach is allowed for over-the-counter trades. But in Europe, where there are more clearing houses, and no prescribed approach, things are messier. And it could be dealers, not clearing houses, that ultimately decide what is offered. By Tom Osborn

When Europe's over-the-counter derivatives clearing rules take effect – probably during the first quarter next year – users may be able to choose from no fewer than 15 different segregation models, offered by the market’s four major central counterparties (CCPs). That could be bewildering, but for the clearing members that act as gateways to a CCP, the problem is less abstract – there will be too many different account types to support.

“We’re concerned about the multiplicity of margin segregation models on offer. Building to all models simultaneously is a complex and inefficient undertaking and doesn’t make sense, particularly at a time when the industry has the opportunity to establish a common architecture that minimises risk,” says Hester Serafini, global co-head of OTC clearing at JP Morgan in London.

As a result, banks may not offer all these segregation models to their clients. Serafini says the time needed to address all the legal questions related to each model, as well as building the architecture to support them – which includes connections to the CCPs, reporting functionality and the infrastructure to manage different collateral arrangements – means dealers may not be able to fully test each model before the start of mandatory clearing. “As we don’t want to compromise the stability and reliability of the service we offer our clients, the short timescales in front of us could force us to make a choice rather than connect to all, at least in the short to medium term,” Serafini says.

That view is echoed by Andy Ross, London-based European head of OTC clearing at Morgan Stanley: “In terms of which segregation models we support, we are driven by client demand. But it is unlikely any members will support, potentially, more than 10 different segregation models across all European CCPs. It is easy for CCPs to suggest a wide variety of options to clients, but a sound legal basis and awareness of the costs associated with these segregation models needs to be established.”

The same argument is being put to the CCPs themselves. Dealers say a working group set up by the Futures and Options Association (FOA) is encouraging clearing houses to harmonise their efforts. It meets weekly, and although the FOA declines to comment on its agenda, sources close to the group say it is channelling dealers’ concerns to CCPs, and asking them to address specific issues relating to operational complexity. With a dialogue established, some hope a standardised approach could emerge naturally. As one senior executive at a CCP points out, the single segregation approach allowed in the US – known as legal segregation with operational commingling (Lsoc) – emerged from a CCP-led working group and was later embraced by the US regulator (Risk March 2012, pages 38–43).

It is unlikely any members will support, potentially, more than 10 different segregation models across all European CCPs

Some futures commission merchants hope to see the fruit of the FOA group’s work before the European deadline arrives. “We’re talking to all the CCPs about their plans to offer individually segregated accounts to clients, and what form they will take. Our hope is that we can build a standardised infrastructure required to support different models, before applying minor adjustments to meet the individual demands of the CCP,” says Mark Croxon, global product manager for OTC derivatives clearing at Nomura.

So, what’s currently on offer? At a minimum, each CCP will have to provide two forms of segregation – a requirement of the European Market Infrastructure Regulation (Emir). The first is an individual account, in which client assets are held separately from those of its clearing member and the member firm’s other clients. It means the cleared trades within the account will have to be margined in full by the clearing member. Unless that margin is posted by a client in advance of a CCP calling for it, the clearing member could face intra-day and potentially overnight funding obligations (Risk November 2012, pages 28–30). This is likely to make it an expensive form of segregation. Set against that, collateral should be better protected.

Omnibus account
The second form of segregation required by Emir is an omnibus account, in which collateral can be commingled with that of a clearing member’s other clients, and a clearing member can post the net amount of margin required for the clients in the account. It’s a riskier approach, but cheaper (see box, Pros and cons of the three main segregation types).

Nothing prevents a CCP offering a broader menu of options, but it has to ensure all segregation approaches are also available in an indirect clearing context – where the clients of a clearing member take on clients of their own (Risk October 2012, pages 28–30). And, in contrast to the blanket application of Lsoc in the US, Emir does not spell out how the different forms of segregation must work in practice. As a result, the two mandatory forms of segregation have given rise to a range of different interpretations, and hybrid approaches.

SwapClear, for example, currently offers the minimum two account types. Its omnibus approach is similar to Lsoc – meaning client collateral can be commingled, but a CCP is not allowed to use the collateral of non-defaulting clients to cure the default of another firm within the account. The risk not addressed in this model is that of a double default in which a collapsing client also takes down the clearing member and the account does not contain enough collateral – the assets of other clients could then be consumed.

The CCP also offers a form of individual segregation. But it plans to roll out two further variants of the segregated model in the second half of this year.

Eurex launched client clearing for interest rate swaps in November. It currently offers three segregation approaches, and is planning a fourth. It has a fully segregated account; a net-margined omnibus model tailored to UK clients, which is already live; and its pre-existing, futures-style elementary clearing model, wherein clients opt out of segregating their collateral altogether. During the second or third quarter, it intends to add an Lsoc-style model.

CME Clearing Europe, which began offering client clearing for rates in March, will provide clients with three models: a physically segregated account, an Lsoc-style model and an unsegregated omnibus account. Meanwhile, Ice plans to offer users of its soon-to-be launched European client clearing services the option of a segregated omnibus account or a fully segregated account. It is also considering bringing an additional Lsoc-style solution to market.

All told, that amounts to 14 proposed models. But even where the models share the same broad approach, clearers have their own distinct versions, giving clients more options and making further operational demands of dealers. CME, for instance, plans to introduce changes to its Lsoc-style model that will allow clients to choose whether margin payments in excess of the amount required to cover a position are held at the CCP level, under so-called ‘Lsoc-with-excess’ models. Both Eurex’s omnibus models offer protections comparable to Lsoc-with-excess – while client funds are commingled, positions and collateral are segregated at the CCP level, with any excess value allocated among the omnibus clients accordingly. Holding excess margin at the CCP level requires clearing members to give daily reports to the CCP on the value of collateral attributable to each client, and some dealers say this is effectively an entirely separate model to Lsoc, taking the tally of segregation models requiring their support to 15.

And there is an ongoing debate about how best to deliver full segregation. In its basic form, it involves three parties – the client, clearing member and CCP – with assets being passed from one end of that chain to the other, as required. That involves transit risk – the possibility that the clearing member intermediary will collapse while temporarily holding client assets – and settlement risk. As a result, so-called quad-party segregation – with the fourth player being a custodian or central securities depository – is being discussed.

In this model, client assets are only ever held at the custodian, and it is just the legal status of the assets that changes – being made available to either the clearing member if the client defaults, or to the CCP if the client and clearing member default. It eliminates settlement risk, and some clients are said to be happier with an approach in which their assets remain in a custodial account at all times. The as-yet-unsolved challenge is to perfect an arrangement in which the clearing member is certain client assets are available to mitigate its counterparty risk. This is essential if capital requirements for the clearing business are not to shoot up.

CCPs say offering full physical segregation will be more expensive for them because of the need to create, maintain and manage so many individual accounts. But the overall increase in costs will be determined by the clearing member, they say, and that may boil down to relationship issues.

For their part, dealers say many clients don’t yet appreciate the costs facing a clearing member in full physical segregation. “Unless a client has pre-funded their trade, then, by definition, we’re putting up the margin for them. If the client offers us gilts later in the day as margin, then great – but a CCP’s funding window may well have closed by that point, leaving us with an overnight funding cost. I’m happy to pass that cost on – or wear it, if it’s a good client – but it becomes an overall cost of the business,” says one senior clearing specialist at a global bank.

Another agrees, pointing out that physical segregation models have not taken off in the futures market. “My suspicion is that, right now, clients like the idea of segregation because it sounds like better protection. But once they start seeing the price tag associated with it – and some of this will be expensive, especially if there are many different sub-accounts that all need to be segregated separately – then clients may decide that, for that price, they may well go with the omnibus after all. The industry could spend a lot of time developing something very sophisticated and then find no clients use it,” warns a senior clearing specialist at one US bank. “Eurex already has a fully segregated model for the futures market, but nobody uses it. That either tells you that clients don’t want to pay the costs associated with it, or the model doesn’t fully work.”

In response, Eurex says its fully segregated model, dubbed an individual clearing member account, has garnered increasing interest from clients in recent months. The CCP claims to be in the process of onboarding multiple firms.

But some fund managers say the choice of which model to use will effectively be made for them, whatever the cost, because the protection available in an Lsoc-type approach is not enough. Dutch pension fund PGGM Investments, for example, is currently using LCH.Clearnet’s Lsoc model, the only one available when it began to implement rates clearing last year. But the fund is unlikely to continue using that account type, says Ido de Geus, PGGM’s head of treasury: “There are many aspects of the Lsoc model that make it unsuitable for long-only investors. For instance, under Lsoc clearing members cannot guarantee to return the same assets lodged by us as initial margin in the event of a default. Portability of positions cannot be guaranteed either, and it is doubtful if the concept works in practice.”

The lack of certainty about assets arises because collateral is collected from multiple clients in an account, commingled, and posted to the CCP on a net basis. In the event of a clearing member default, non-defaulting clients should get the full value of their assets back, but there is no way of ensuring the return of the specific assets posted. For many managers, says De Geus, particularly those pursuing liability-driven investment strategies, getting back the value of the assets is not enough.

In the stressed markets that would follow a clearing member default, the value of assets lodged as margin – highly rated sovereign debt, for example – is likely to rise sharply, amid a general flight to safe-haven assets. A fund manager given back the mark-to-market value of its assets, rather than the bonds themselves, would therefore face the expensive task of rebuilding its portfolio, rematching assets with liabilities. A senior executive at a clearing house estimates anywhere from 30–50% of the industry, as measured by assets under management, could see this as a sticking point.

Portability is also a key requirement for many firms (Risk August 2011, pages 16–20). In normal market conditions – if a single client has its own reasons for switching to another provider – it should be easy enough for an Lsoc client to move, clearing members say. Porting from an ailing clearing member, though, would be more difficult. Clients would probably be fleeing the stricken firm en masse, and may struggle to find another clearing member to take on a portfolio of large, directional positions in volatile market conditions. If a portfolio cannot be ported, the client could be closed out by the CCP – again receiving the value of its collateral rather than the original assets themselves. Individual segregation would avoid the latter problem – and may make it less operationally burdensome to move – although the funding obligations arising from the inability to net client positions could make it less likely that a clearing member is willing to take on the business.

Heavy lifting
CCPs add, however, that getting fully segregated models up and running requires some heavy lifting. One of the stickiest issues is ensuring the mechanics of the approach are compatible with local bankruptcy codes in every European state – in a cleared environment, the default of a client is supposed to mean that its clearing members are able to use its margin to cover the close-out costs, potentially jumping the bankruptcy queue. The text of Emir places the onus on CCPs to ensure this is legally permissible, and it is painstaking work, says William Knottenbelt, CME’s managing director for Europe, the Middle East and Africa.

“In the event of a client default under full segregation, the clearing member has first right to the collateral in the fully segregated account. We’re currently preparing documentation for regulators on how that will work under local bankruptcy laws within each different jurisdiction. There’s a lot of legal work that needs to be done. We would be doing well if we have legal certainty on how full segregation will work in every jurisdiction, for every bespoke contractual combination, by September or October. That would give all parties involved a six-month run-in towards the European clearing mandate.”

Even after all proposed models are finalised and approved by local regulators, however, every CCP faces the hurdle of a lengthy reauthorisation process from the European Securities and Markets Authority. This could take up to six months, clearers estimate privately, during which time they are likely to face an operational lockdown.

For now, clients are left in the awkward position of planning for clearing on the basis of models that have yet to gain regulatory approval. Barry Hadingham, head of derivatives and counterparty risk at Aviva Investors, says: “We need further clarity from everyone – clearing members, custodians and CCPs – on how physical segregation will work in practice. The CCPs can propose what they like, but if their models aren’t scalable operationally, then it’s not going to work for anybody.”

 

BOX: Custody in jeopardy?

The idea of a document published by the European Securities and Markets Authority (Esma) on March 20 was to tackle some common queries on the implementation of the European Market Infrastructure Regulation (Emir). Questions were reprinted, along with Esma’s answers, in the hope of “providing clarity on Emir’s requirements”.

One section has accomplished precisely the opposite. At first sight, what reads like a relatively innocuous paragraph on the topic of central counterparty (CCP) investment policy may prohibit the use of custodians in so-called quad-party segregation arrangements, dealers say. Instead, only a central securities depository (CSD), such as Euroclear or Clearstream, could be used. Under quad-party solutions – four-way agreements between a client, dealer, CCP and collateral agent – a client’s collateral never leaves its secure account with the agent; only legal rights over it change hands.

“One reading of Esma’s recently issued Q&A on Emir implementation suggests CCPs need to hold all collateral at a CSD. So certain models – such as the quad-party segregation model, or the models where the CCP opens an account at a custodian and collateral is moved there on behalf of clients with the clearing member having a first lien – look like they might not be compliant,” says one senior clearing executive at a US bank.

The key part of Emir is Article 47(3). One CCP says it had interpreted the text as allowing client collateral to be held at a custodian, and has based its proposed physical segregation model on that belief.

Article 47(3) of Emir reads: “Financial instruments posted as margins or as default fund contributions shall, where available, be deposited with operators of securities settlement systems that ensure the full protection of those financial instruments. Alternatively, other highly secure arrangements with authorised financial institutions may be used.”

But CCP answer 4(a) in the Esma Q&A says: “Depositing financial instruments with an operator of a securities settlement system via a custodian does not constitute a deposit with an operator of a securities settlement system for the purposes of Article 47(3) of Emir. Such a structure would instead amount to a deposit with an authorised financial institution.”

That has prompted CCPs and others to seek a clarification. “CME Group is actively working with regulators to seek clarification on the meaning of Article 47(3) of Emir, which is subject to different legal interpretations. Regardless of the outcome, CME Group is able to work with both custodians and operators of securities settlement systems within the full segregation model,” says William Knottenbelt, CME’s managing director for Europe, the Middle East and Africa.

Esma declined to say whether it had received requests to clarify its stance on CCP investment policy.

 

BOX: Pros and cons of the three main segregation types

Individual client segregation
Pros:
■ Asset-level protection that promises a client the safe return of instruments lodged as collateral in the event its clearing member defaults.
Cons:
■ Likely to be expensive, dealers warn, particularly if an asset manager has multiple sub-accounts that each need to be segregated individually.
■ Legal rights over assets – and rules governing their safe return – could be subject to the vagaries of local bankruptcy codes, and therefore not available in every jurisdiction.
■ Clients (and clearing members) stand to lose netting benefits between offsetting positions that are preserved via legal segregation with operational commingling (Lsoc).
■ No clearing house has an approach that has been tested in a default situation.

Lsoc
Pros:
■ Clients’ assets are not used to meet the margin obligations of another defaulting client.
■ Under the Lsoc-with-excess model, client collateral is further protected by being held at a central counterparty (CCP), rather than the clearing member.
Cons:
■ Clients are guaranteed to get back the value of assets lodged as collateral, but not the instruments themselves. A large proportion of asset managers say this is insufficient protection.
■ Clients worry that other clearing members will not be willing to take on ported positions amid the stressed conditions created by the default of a major clearing member, making it more important to some firms that the specific, posted collateral is returned.

Unsegregated omnibus
Pros:
■ Operationally simpler to implement for dealers, CCPs and clients, and therefore likely to be cheaper overall than other models.
■ Under net omnibus models, FCMs can net off margin calls against other clients, lowering their cost of funding.
Cons:
■ Client assets are commingled, mutualising risk and exposing them to losses in the event of one defaulting.