The standardised approach to counterparty credit risk regulation, dubbed SA-CCR, was finalised by the Basel Committee on Banking Supervision in 2014. So far, only a handful of jurisdictions have adopted it – this excludes the UK, US and Europe.
The main reason behind the delay seems to be that the regulation is riddled with inconsistencies, which would then feed through into many other Basel regulatory frameworks that use SA-CCR exposures as an input.
In Revisiting SA-CCR, published today on Risk.net, Mourad Berrahoui, head of counterparty credit risk modelling at Lloyds Banking Group, Othmane Islah, head of quantitative research at Quantuply and a senior consultant for capital and margin valuation adjustments at Lloyds Banking Group, and Chris Kenyon, head of derivatives valuation adjustment quant modelling at MUFG Securities EMEA in London, highlight some of these issues and propose ways to improve the framework.
SA-CCR is essentially a standardised way of calculating the exposure at default (EAD) of a counterparty to a trade, which is given by a scaled sum of the replacement cost of the derivative and the potential future exposure (PFE) add-on, which covers adverse moves in the future.
The SA-CCR is meant to replace the current exposure method and the standardised method, both of which were criticised for failing to treat margined and non-margined trades differently.
More sophisticated banks could also use their own modelled approach to compute the EAD, but the importance of SA-CCR stems from the fact that it could be an input to other regulations such as the leverage ratio and the net stable funding ratio.
Given the widespread impact of the SA-CCR, the board of governors of the Federal Reserve System, the Federal Deposit Insurance Corporation and the Office of the Comptroller of the Currency invited public comments on the framework in December 2018.
The quants’ paper, a version of which was submitted as part of the public consultation, flags three issues that highlight the ineffectiveness of SA-CCR in capturing risk appropriately.
We are deriving from the current standard itself to say how well the SA-CCR or our revisit performs relative to the ideal. So this gives us a standard to measure ourselves againstChris Kenyon, MUFG Securities
The first one is that, for linear trades such as swaps, economically equivalent positions can have different PFE add-ons.
“For example, if you have a 10-year swap, or a five-year swap and a forward-starting five-year swap, you don’t get the same capital for the two alternatives,” says MUFG’s Kenyon.
Risk sensitivity is another issue with the framework, as the same inconsistency applies to positions that are economically zero – in other words, positions that net to zero risk because the cashflows are equal and opposite. Such positions can get “material” add-ons, says Kenyon.
The framework is also not sensitive to moneyness of the derivative, or the price of the underlying relative to the strike price. In reality, however, the moneyness level can change the risk profile of trades, for example, if the general level of rates change.
To fix these issues the quants propose a cashflow decomposition method to break down various linear trades into their cashflows and then apply the SA-CCR add-ons based on those cashflows instead. The quants apply the commonly used Itô’s Lemma model to identify the sources of volatility of the cashflows, which then gives the effective notional of the trade, which signifies the risk of a cashflow. This is in turn used as an input into the SA-CCR formula that then computes the PFE add-on.
Doing it this way ensures consistent treatment across equivalent positions, ie, those with similar cashflows. The decomposition method already exists under the current SA-CCR framework for non-linear trades. The quants expand this to apply it to linear trades as well, where most of the inconsistencies exist.
The quants say the resulting model is to be used as a way to compare and study the issues with the SA-CCR, which can in turn help improve it.
“We are deriving from the current standard itself to say how well the SA-CCR or our revisit performs relative to the ideal. So this gives us a standard to measure ourselves against,” says Kenyon. “Since we are trying get rid of some problems in the standard we know where to look.”
Numerical studies carried out by the quants show that if one looks at swaps with varying moneyness, for instance, the range of SA-CCR accuracy is –19% to +37%, whereas their revision is within a few per cent across the different degrees of moneyness. The quants compare both against an exposure simulation identified from the SA-CCR framework itself to measure the accuracy.
The most publicised issue with the SA-CCR so far is its inability to provide a reasonable amount of offset for margined trades. In an industry study carried out in 2017 it was found that in some instances, banks’ counterparties would have to post 10 times more margin versus the internal modelling method to get an almost full offset.
However, the issues highlighted in the paper are even more fundamental and show the regulation needs a much closer look when being translated into legislation in various jurisdictions. Whether US regulators, who have asked for public feedback on the rules, will be open to a material tweak is unclear. But the evidence suggests they may have to at least consider it.