Market at Risk

ISDA-led analysis suggests the FRTB could cause a capital hike of as much as 140%, and participants believe a failure to tackle concerns over internal model approval may force them to exit certain businesses

AT A GLANCE
The final Fundamental Review of the Trading Book framework was published in January 2016.

Industry analysis conducted by ISDA and other trade associations suggests market risk capital requirements could increase between 1.5 and 2.4 times under the new rules.

Where banks fall on the spectrum will depend on the extent to which they obtain internal model approval.
The P&L attribution test is an important component in the internal models approval process, but there is uncertainty on how this process will work.

Rules regarding non-modellable risk factors are also a concern, prompting attempts to establish an industry utility.

Renovating regulation is a complicated process, but the overhaul of market risk capital requirements has been a particularly long and difficult road for regulators and industry alike. Nearly four years on from the first consultation on the Fundamental Review of the Trading Book (FRTB) in 2012, the final standards were published in January 2016. That should have created a clear path towards implementation, but concerns remain over key components of the new rule book.

“The FRTB makes it much more difficult to gain approval to use internal models for market risk”
— John Mitchell, Credit Suisse

While the Basel Committee on Banking Supervision addressed issues that had been raised in response to earlier drafts, industry participants believe the final framework could result in a much greater increase in capital than regulators expect. A lack of clarity around the process for gaining approval to use internal models is also raising concerns, as that could leave many desks with no option but to use the more capital-intensive standardised approach.

“The FRTB makes it much more difficult to gain approval to use internal models for market risk, and the approval process involves new tests that are not clearly defined, so we can’t yet be sure how many desks will pass or fail. That could mean banks will have to hold much more capital than has been estimated by the Basel Committee, as the decrease in internal model usage means significantly higher capital requirements,” says John Mitchell, director of market risk management at Credit Suisse.

“The cliff effect between the capital required under internal models and standardised models is a real concern”
— Eric Litvack, ISDA

Impact

The Basel Committee has not been blind to the potential effect on capital levels, and conducted impact testing during consultation on the FRTB. On publishing the final framework in January, it estimated that the standards would result in a weighted mean increase of roughly 40% in total market risk capital requirements. That estimate was based on data the banks had provided as part of an earlier consultation, and was recalibrated to take account of the alterations in the final rules.

A 40% increase represents a significant hike in capital requirements, but industry analysis that has been undertaken since January suggests the impact could be far greater. Using data submitted by 21 internationally active banks, ISDA joined forces with the Global Financial Markets Association and the Institute of International Finance, in conjunction with the Global Association of Risk Professionals, to run its own impact assessment.

The summary results, published in April, have raised eyebrows. The analysis found the FRTB would result in an increase in market risk capital of 50%, but this estimate is based on the assumption that all banks receive internal model approval for all desks – an unlikely scenario, as the rules deliberately raise the bar that banks must meet to use internal models.

If all desks were to fail the internal model tests, the industry analysis concluded market risk capital would increase by 2.4 times – a 140% hike. Of equal concern is the cliff effect between the standardised approach and the internal models approach, which could see capital requirements increase immediately and steeply if a desk were to lose model approval under the new framework – by as much 6.2 times for a foreign exchange desk and 4.1 times for an equity desk. That could lead to a sudden and destabilising reallocation of capital and business activity.

“The difference between the capital required under the standardised and internal models approaches is disturbing, and we are very concerned about the impact this could have on certain markets. We had been told previously that changes in the risk weights were intended to more closely align the two approaches, but it doesn’t seem to have worked. So if this is the intent, some revision is still necessary,” says Debbie Toennies, head of regulatory affairs for the corporate and investment bank at JP Morgan in Chicago.

As for the overall capital increase that could result from FRTB, ISDA believes the capital hit for most banks is likely to fall somewhere between 50% and 140%, depending on how many desks secure model approval.

Internal models

To what extent they can achieve that goal is likely to be determined largely by two key factors. First, the calibration of the capital floor framework, currently being considered by the Basel Committee, will determine whether the use of internal models merits the resources required to develop them.

Second, uncertainty about the process for gaining model approval needs to be ironed out before banks can estimate the number of desks that will be able to use internal models.

“There is still uncertainty over the exact nature of the model eligibility requirements, but if they are overly stringent, it could lead to a significant number of desks being relegated to the standardised approach. The cliff effect between the capital required under internal models and standardised models is a real concern, which we think could lead to much more significant increases in capital requirements than regulators intended,” says Eric Litvack, chairman of ISDA.

To some extent, the FRTB is more lenient on the use of internal models than other Basel Committee initiatives. While the option to use internal models for operational risk and the credit valuation adjustment is being removed altogether, the committee is retaining internal models for market risk but raising the standards banks must meet to use them. Internal model usage for market risk is dependent upon explicit approval from a bank’s supervisor, which will only be granted if a comprehensive set of qualitative and quantitative criteria has been met.

P&L attribution

These tests include profit and loss (P&L) attribution and back-testing, which aim to ensure internal models align properly with front-office systems. If both models produce similar results, then regulators should theoretically have reasonable confidence in the validity of a bank’s internal models. But if there is a discrepancy between the two, then it raises a red flag.

“P&L attribution is a sensible concept to test model validity, but it is complex to implement. The framework has not been properly tested and could ultimately fail models that are not actually flawed. We are working to reconcile these issues and hope the regulators will respond, because being forced to use the standardised approach would have serious implications,” says Panayiotis Dionysopoulos, director in the risk and capital team at ISDA.

The P&L attribution and back-testing framework is set out in detail in the appendix of the final standards, but a crucial component of the test is inconsistent with the way it is later described in the glossary.

The appendix explains that banks must calculate a risk-theoretical P&L (RTPL) for each trading desk, which is “the P&L that would be produced by the bank’s pricing models for the desk if they only included the risk factors used in the risk management model”. The FRTB glossary, however, states that the P&L attribution test compares “the hypothetical P&L predicted by risk management models with the actual P&L”.

There is a major difference between using front-office pricing models to calculate the RTPL, as per the appendix, and using risk management models, as per the glossary. The appendix approach would primarily focus on risk factor completeness between the two models, while the glossary approach would go further to assess the accuracy of the capital model. The mathematical construct of the test means it will be harder for banks to pass.

Given the final framework for the P&L attribution test had been adjusted from earlier versions of the FRTB, some believe the glossary definition may have been a simple mistake that wasn’t updated in the final standards. But until the Basel Committee makes a formal amendment, it is difficult for banks to prepare internally.

“P&L attribution is an entirely new and additional test, so in any form it will necessitate more work and there will be a higher chance of not getting approval for internal models,” says Credit Suisse’s Mitchell.

Non-modellable risk factors

Concerns over the framework do not stop at the P&L attribution test, however. The FRTB also dictates that for a risk factor to be modelled internally, it must have at least 24 observable ‘real’ prices per year, with a maximum of one month between two consecutive observations. Any risk factors that do not meet the criteria would be deemed non-modellable and would therefore have to be capitalised using the standardised approach.

“P&L attribution is a sensible concept to test model validity, but it is complex to implement”
— Panayiotis Dionysopoulos, ISDA

“In addition to P&L attribution, banks must determine if each risk factor can use an internal model by tracking how many times trades or quotes occurred in the market relevant to that risk factor,” Mitchell explains. “Large banks have tens of thousands of risk factors, which will require a huge amount of granular trade-level data to determine which of them are modellable.”

The industry analysis found that non-modellable risk factors (NMRFs) remain a major component (30%) of the internal models approach capital charge. Further, the requirement for 24 price observations to determine the eligibility of a risk factor for inclusion in an internal model relies on a pool of data that does not exist today, and may require some kind of industry utility to be developed.

“This is still at an early stage, but we expect NMRFs to be a key area of focus during the second half of this year. Banks may benefit from some kind of mutualised data solution that would bring together the required data and ensure there is no over capitalisation as a result of NMRFs,” explains Dionysopoulos.

Some technology vendors have already risen to the challenge, including Markit, which unveiled a modular offering for FRTB compliance on May 24. It will ultimately be up to the banks to determine what technology they need, but dealing with the data implications of NMRFs looks to be a heavy lift. When combined with the complexity of the P&L attribution and back-testing requirements, the framework could drive some banks to opt voluntarily for the standardised approach, despite the capital costs.

“With limited time and resources available, and these new and complex requirements for models, it seems likely that more banks will opt to use standardised models instead of internal models and will have to adapt to the additional capital by reducing their market-making activities. Banks are already scaling back or re-evaluating these activities, so the concern is the FRTB will accelerate this,” says Mitchell.

It remains to be seen how much appetite the Basel Committee has to make revisions to the final framework, and with a 2019 deadline for implementation, time is short. But if concerns over the P&L attribution test and NMRFs are not tackled, banks may struggle to use internal models and so end up on the higher end of the capital spectrum, participants say.

A capital increase of 140% was the worst-case scenario quantified in the industry’s impact assessment, but it could be closer to reality than some would like, which may ironically lead to increased risk-taking among some banks.

“By increasing the quantum of capital required, the trend away from internal model approaches clearly increases the cost of capital market financing by the banking sector, but by reducing risk sensitivity, they also implicitly increase the incentives to allocate capital to riskier activities that generate higher returns,” says Litvack.