Clearing houses exist for one reason: to mitigate the effect of a trading firm’s default on the rest of the market. And they have done this fairly well over the decades according to the consensus.

More recently, however, financial regulators have turned their attention to clearing houses’ approaches to the other risks they face, which have been grouped together under the catch-all term non-default losses.

Some of the main NDLs are: investment losses including those from the failure of a repo counterparty or downgrades to the credit ratings of a bond issuer; losses related to failures of custodians or settlement firms; losses from fraud, theft or dodgy staff; losses from cyber-attacks; and losses from operational or systems failures.

NDLs then are pretty dull but they have resurfaced the age-old argument around clearing: who is ultimately on the hook if things go badly wrong? The clearing house itself or its customers?

One thing everyone (including the regulators themselves) seems to agree on is the definitions of NDLs need more work.

IOSCO and the Basel Committee on Payments and Market Infrastructures (CPMI) published in August last year a report outlining areas where further guidance may be needed on central counterparties’ approaches to non-default losses.

This followed an August 2022 discussion paper from the international regulator which sought “to advance industry efforts and foster dialogue on central counterparties’ (CCPs) management of potential losses arising from non-default events”.

The regulators also said in August last year they will “in the near term” launch a further consultation on the topic as new guidance may be necessary beyond the current Principles for Financial Market Infrastructures (PFMI) rules.

“Given the wide range of practices reported and the different views expressed by industry, CPMI and IOSCO intend to undertake additional work on NDLs across all FMI types, including an assessment of the implementation of the PFMI and guidance on general business risks,” the report said.

“Further engagement with industry stakeholders will also be undertaken to inform a public consultation in the near term on further guidance or recommendations with respect to NDLs.”

Six months on, the issue of clarity about NDLs is not going away. Speaking at the WFE Conference in Madrid this week, Laura Bayley, the head of clearing services at SIX group, the Swiss exchange group, said: “The difficulty with non-default loss scenarios is they are so difficult to predict and define. This is where I would require more guidance and more definition by the various regulators on all the variances of non-default loss scenarios.

“Cyber is the classic one but a ransomware attack is very different to an attack that will delete all your data from your systems and leave you incapable of restarting your CCP. Those are very different scenarios which have very different consequences for the CCP, for which the CCP will need to prepare.”

Bayley added: “That’s where we would wish for more detailed work on the scenarios so we can then try to work out the consequences.”

Giles Ward, a senior policy adviser within the General Secretariat at the International Organisation of Securities Commissions (IOSCO), conceded on Tuesday that more work needs to be done.

Ward said: “We certainly heard those points which is why following on from the 2022 paper which set out some ideas but not standards, we are further developing our work as a matter of urgency. We have a group working on NDL which is looking to develop more detailed guidance or recommendations, and our IMSG (Implementation Monitoring Standards Group) in parallel is conducting its assessment which will be useful to feed into that.”

He continued: “In terms of scenarios, participant default we understand well and here the scenarios are quite easy to model – you can have one default, two defaults or three defaults for example – whereas with non-default losses the scenarios are infinite: there’s custody, cyber, fraud and so on.”

Ward said some NDLs are easier to model than others: “For some buckets, historical data would be useful such as investment and custody which are long-standing risks but the key issue is the risks such as fraud where we don’t have a lot of backward looking data.”

Richard Metcalfe, the head of regulatory affairs at the World Federation of Exchanges, told the WFE conference on Tuesday: “The thing about non-default losses is that, because they are manageable but not modellable, they are hard to quantify. They are pretty rare as far as we can tell, and they can’t be pooled as they are idiosyncratic to the way an individual organisation is managed.”

He added: “The really interesting bit comes with what sorts of tools you can have. What you can’t do is apply credit risk management tools such as variation margin or haircutting.”

Metcalfe concluded: “The practice can vary a little bit from CCP to CCP in as much that if you have your members involved in determining where investments should be made, then it reasonable to expect them to take some sort of liability.”

The FIA, another trade body which represents CCP members, responded to the IOSCO paper last year with its position. Jackie Mesa, chief operating officer at the FIA, said in August: “We have long argued that CCPs should be responsible for most NDLs and transparent about how they are allocated. As we said in our paper in 2020, the guiding principle for allocating these losses should be who manages the risk.”