Why Isda ditched ‘off-cycle’ updates for Simm

Ad hoc updates riled industry, while regulators pushed for predictability in model recalibrations

The International Swaps and Derivatives Association announced on September 7 that its standard initial margin model (Simm) will be recalibrated at fixed, semi-annual intervals. The move follows regulatory calls for a more frequent and predictable update schedule and industry gripes about the first ‘off-cycle’ recalibration earlier this year.

Simm, which is widely used to calculate regulatory initial margin for non-cleared derivatives, came under criticism for its performance in 2020. The model was previously recalibrated annually using the prior year’s data, with a new version released each December. This meant the Covid-related market volatility in March 2020 was not reflected in Simm until December 2021.

The 20-month delay did not sit well with regulators. Isda responded by updating its calibration framework for Simm last November, to include quarterly ad hoc revisions of delta risk weights.

One of the potential pros to making that transition was that there would be the predictability

Tara Kruse, Isda

The first of these off-cycle recalibrations, reflecting elevated interest rate volatility at the end of 2022, became effective in July. Just two months later, Isda is ditching the off-cycle updates in favour of fixed semi-annual recalibrations, starting in 2025.

The trade body says the change was made at the behest of some regulators, who preferred fixed dates. “In our dialogue with global regulators, they were comfortable and understood our off-cycle process, but a number of them preferred the predictability of semi-annual,” says Tara Kruse, Isda’s global head of data, infrastructure and non-cleared margin “We did a ton of analysis to figure out what that schedule would look like, and one of the potential pros to making that transition was that there would be the predictability.”

The move has been broadly welcomed by industry participants. “Everybody wants predictability,” says Erik Petri, head of triReduce and triBalance at the post-trade firm Osttra. “We want predictability, but more importantly the market wants predictability and being able to plan ahead. Having semi-annual recalibration allows regulators to make sure the model aligns over time as well as possibly with the underlying volatilities of the market.”

Simm users – some of which struggled to implement the first off-cycle recalibration – also back the shift to fixed, semi-annual updates. “I’d rather have semi-annual than off-cycle,” says Eduardo Epperlein, global head of risk methodology at Nomura. “Off cycle is ad hoc and anything that’s ad hoc is more prone to operational error, so I support the concept of semi-annual.”

‘Good in theory’

Simm is calibrated using data from the previous three years plus one additional stress year. The current annual cycle sees new versions of Simm become effective in December, using historic data up to the end of the previous year. This means market events can take up to 23 months to be reflected in the model.

Such delays can trigger a jump in exceedance rates, where Simm margin requirements fail to meet regulatory minimums. Non-cleared margin rules require collateral posted against bilateral trades to cover valuation changes over a 10-day period with a 99% confidence level.

In a June 2022 letter to bank chief risk officers, the Bank of England’s Prudential Regulatory Authority (PRA) warned that shortcomings in Simm’s methodology could leave counterparties ‘systemically under-margined’.

The move to optional quarterly updates shortens the response time to a maximum nine months, though these off-cycle revisions are only triggered in response to “widespread” and “material” exceedance rates.

The off-cycle calibration was a way to meet regulatory expectations without creating a great deal of extra work for Simm users

Tara Kruse, Isda

This approach was seen as a compromise for addressing regulators’ lag concerns without committing to additional recalibrations unless there is a pressing need. “We didn’t want to create more work for firms, especially now we have smaller firms using it, so the off-cycle calibration was a way to meet regulatory expectations without creating a great deal of extra work for Simm users,” says Kruse. “It also aligns with our evidence-based approach, where doing an off-cycle calibration is based on market evidence. Just doing it more frequently means you may or may not see a material difference for the effort involved.”

Sources say users may not have expected an ad hoc calibration to emerge so soon after the new framework was finalised, yet extreme volatility in the fourth quarter of 2022 triggered the threshold of exceedance rates, setting the new plan into action. Version 25A of Simm was quickly pulled together for publication in May and implementation two months later.

The update left some users struggling to cram an operationally intensive process into a short timeline. Users have to crunch vast reams of data to generate new versions of the model, which then had to be validated by regulators ahead of implementation. Some banks struggled to secure appropriate IT resources for the project at short notice.

“Implementation of an optional quarterly recalibration has different challenges, this has probably led to the pivot towards a bi-annual recalibration,” says Stuart Smith, co-head of business development at Acadia.

While many participants supported the quarterly updates in theory, the reality of having just three months to complete the recalibration left many under-prepared.

“Any process that involves that many banks is hard work. If you consider the IT infrastructure implications, there are very intricate delivery and release plans to contend with. Everything’s planned months in advance, and all of a sudden you had to cram in this extra release with relatively little notice,” says Smith.

Related

After coronavirus rout, concerns raised about Simm

PRA’s pot shots threaten Simm’s global ubiquity

“The question for banks was whether to plan for these every quarter, given you don’t know whether they’re coming or not. When a new version is released, there’s an awful lot of patches to push out and it's hard work to schedule them all, especially if you get a relatively important one with little notice.”

Two other margin vendors agree with that assessment. “Some firms who weren’t using a vendor found it quite hard because it wasn’t planned, and they weren’t as prepared as other clients who can lean on a vendor to do the implementation of the new model,” says Gemma Bailey, business manager, triCalculate at Osttra. “For our clients, it wasn’t much of a shock, but we did have to go out to clients and inform them to make sure they were aware. Especially since the December calibration last year, clients are more proactive in terms of judging the impact on IM for these calibrations, whether on-cycle or off-cycle.”

While the industry managed to get the first off-cycle update over the line for its July 15 implementation, Jo Burnham, risk and margining expert at OpenGamma, says some users were still unprepared for the impact on their margin requirements.

“People got caught on the hop and didn’t have the resources there to deal with it, as they didn’t have as much notice as they did with the annual calibration. It wasn’t anywhere near enough to be able to offer facilities like allowing people to have a test version so that they can see what the impact is going to be,” says Burnham.

She adds that while the vendor offers a full testing service for annual reviews, many clients only discovered the real impact of the off-cycle update on their portfolios on the implementation day.

“We could give them background information or run tests looking at the likely impact on particular types of sensitivities or products, but it’s not a full testing service.”

Back to the drawing board

Isda denies that problems with implementing the off-cycle update played a role in its decision to move to semi-annual recalibrations. The trade body maintains the work required was much lighter than for a full annual overhaul.

“It’s not the same kind of lift as a full cycle as it’s limited in scope, focusing on particular risk classes and, within that, only the main delta risk weights. We do source most of the data from public sources rather than member firms, which does make things more straightforward in terms of sourcing and validating the data,” says Nnamdi Okaeme, Isda’s head of Simm.

Isda’s new approach comprises a primary and secondary calibration of the model.

Primary recalibrations replicate the current annual exercise, with full-year data from the prior year used to update all model parameters including risk weights and correlations. The primary recalibration will take effect in August, effectively bringing forward the annual update by four months. A secondary calibration, which becomes effective in February, will include first-half data from the prior year and only consider delta risk weights. However, in contrast to the off-cycle approach, secondary updates will include all delta risk weights and not just those driving exceptions.

The change means market events are reflected in the model up to 13 months after they materialise.

Epperlein says this achieves a fair balance between ensuring margins are conservative enough to withstand shocks, while avoiding procyclicality – a common criticism of clearing house margin models – and minimising the operational burden on users.

There’s still an argument that the calibration frequency should be tuned to volatility in the market

A risk manager at a CCP

“First was the question of how often we recalibrate. Semi-annual is probably good enough to address the issue of market scenario shocks,” says Epperlein. “Unlike certain market risk models, which you have to recalibrate frequently to capture the risk near-term, Simm includes a stress period and unless that changes, the impact of recalibration is minimal.”

Osttra’s Petri agrees semi-annual recalibrations should be enough, at least in the short-term. “If we see much larger spikes in volatility during short periods of time there may, at some point in the future, be a need to do even more frequent calibrations, but we’ve just moved from annual to semi-annual and I’d be surprised to see that increase further in the near future,” says Petri.

But not everyone is convinced Isda has cracked the puzzle. A risk manager at a CCP believes ad-hoc updates may still be needed in response to major stress events.

“Increasing the frequency of calibration is a good step and semi-annual might be good when you don’t have anything happening, but there’s still an argument that the calibration frequency should be tuned to volatility in the market,” this person says.

Meanwhile, Acadia’s Smith says semi-annual recalibrations could still prove operationally taxing for users.

“It’s not without its own challenges. Some banks are still concerned about being able to produce all the data and do the back-testing they need to do on the new compressed timescale. They still have data quality challenges to overcome and now need to squeeze that process down to six months,” says Acadia’s Smith.

Ahead of the changeover, Isda will work with the industry to improve the efficiency of the calibration process. For example, the trade body will expand the common risk interchange format, or Crif files, which are used to capture the risk sensitivities of trades, to include data for quarterly backtesting. This means users will be able to use Isda’s Perun platform to validate submissions.

The picture will become clearer when the first semi-annual recalibration becomes effective in February 2026 – assuming a major market event doesn’t change the equation before then.