Will the Nasdaq default spur CVA for CCPs?

Quant proposes a model to calculate bank credit risk exposure to a CCP

For a few years now, quants have been debating whether banks should take a credit valuation adjustment (CVA) charge to account for their exposure to a central counterparty’s (CCP) default.

While some argue CCP defaults are rare, and the loss given a default generally low, others say that, as long as that exposure exists, dealers should try to quantify it. One thing both sides agree on: calculating CVA for CCPs is not straightforward.

In this month’s first technical, Central counterparty CVA, Matthias Arnsdorf, global head of the counterparty credit risk quantitative research team at JP Morgan, demonstrates a model for computing CVA exposure for a CCP using the limited information that is available to a bank facing it as a clearing member.

Typically, dealers calculate CVA on a counterparty using the probabilities of default implied by credit default swap spreads (CDS). Stating the obvious, this information is not available for CCPs. Modelling an individual bank’s exposure to losses from the default of a single member of a CCP, wherein the CCP itself doesn’t default, is still more complicated.

Such losses can still be painful, especially for the larger constituents of a CCP’s guarantee fund: the default of Norwegian power trader Einar Aas at Nasdaq Clearing, which saw members handed a €107 million ($119.7 million) replenishment bill, is a recent example.

To date, most research on modelling CVA exposure to a clearing house has rested on the assumption that at least some confidential information will be made available for the purpose – for instance, the initial margin contribution of other clearing members, which acts as a credit risk mitigant – even if the positions members hold at the CCP are kept confidential. Arnsdorf argues this isn’t realistic.

“There are a number of papers that show, assuming you have all of the information, what you do [to compute CVA] in a huge amount detail, what the precise calculation is,” says Arnsdorf. But, in reality, “there is very little information you have about CCPs and CCP portfolios. So the challenge is what you do to overcome that and still get a reasonable estimate out”.

The most likely source of loss CCP members face is a hit to their default fund contribution stemming from the default of another clearing member. When a member defaults and both its initial margin and default fund contribution are exhausted, other members find themselves on the hook. This doesn’t necessarily cause the CCP to default, which is quite rare, but still generates losses to members.

“When you think about exposure to a CCP, or CVA to a CCP, the first thing is it’s not necessarily about CCP default itself. The risk comes from loss mutualisation,” says Arnsdorf.

In order to calculate a CVA charge, a bank first needs to know the distribution of potential losses from all other clearing members, which aren’t known. Banks do, however, know two pieces of related information: the size of the CCP’s total initial margin pool; and how much of this their own margin contribution accounts for.

When you think about exposure to a CCP, or CVA to a CCP, the first thing is it’s not necessarily about CCP default itself. The risk comes from loss mutualisation

Matthias Arnsdorf, JP Morgan

The initial margin any member has to post is directly linked to its distribution of potential losses – and can thus be used to estimate one’s losses, Arnsdorf’s research suggests.

When a clearing member defaults, its losses need to be allocated to the surviving members. In his paper, Arnsdorf assumes this allocation will be a function of the proportion of the member’s own default fund contribution to the total default fund at the CCP – again, both of which are known.

“You know what the total potential loss is – so you [can] calculate what your portion of that loss is and allocate that back to your firm, in proportion to your default fund contribution,” says Arnsdorf. “Basically, the larger you are, the more losses you are going to be allocated.”

Once a member has calculated its expected loss, it then needs to be weighted according to the individual default probabilities of all the CCP’s other members – available via their publicly traded CDS spreads – to arrive at the CVA number.

As an example of how his model works, Arnsdorf applies a stressed scenario – borrowed from the US’s Comprehensive Capital Analysis and Review (CCAR) tests – that gives a typical expected loss for a bank facing a CCP of between 10 basis points and 40bp of the total initial margin pool.

Although the result in this example is based on a stressed scenario, losses to individual banks arising from member defaults may be likely even when there is no market-wide stress. One example is the Nasdaq episode, which raised many questions about the quality of the CCP’s risk management, including the margin methodology used, the concentration of positions and the auction process carried out post-default.

Banks have been swift and vociferous in their calls for all CCPs to review their risk management standards in the wake of the incident. But one thing dealers could do in the meantime is at least be aware of their exposure to CCPs by calculating a CVA charge. Some US banks are already required to do this under CCAR, it is understood – but banks elsewhere would be wise to do so, too, without needing a regulatory push to do so.