Internal Model Pushback

The Basel Committee’s efforts to increase the comparability of risk-weighted assets has gained momentum in 2016, but risk managers warn that simplifying the capital framework could have a negative impact on the banking sector

Regulators have moved further towards limiting the use of internal models for the calculation of regulatory capital.

Restrictions have recently been placed on the way banks assess operational and credit risk.

Market participants fear that the broad adoption of a one-size-fits-all standardised model, or capital floors, would leave banks unable to adapt to specific markets and risks and could result in a herd mentality in the industry.

Japanese regulators have raised concerns over ever more prescriptive approaches to supervision.

For some time after the financial crisis, the Basel Committee on Banking Supervision referred loosely to its comprehensive reform package as “strengthening the resilience of the banking sector”. It took until late 2010 for Basel III to enter into the lingua franca of regulators. Fast-forward six years, and while Basel III is still deep in its implementation phase, regulators are working on a programme some have already labelled Basel IV.

“One of our main goals this year is to address excessive variability in risk-weighted assets modelled by banks,” said William Coen, secretary general of the Basel Committee, in a speech in Sydney on April 5. “Some – not us – have already dubbed these reforms ‘Basel IV’. I do not think the title in itself is important, but I note that each moniker bestowed on the global regulatory framework was characterised by a substantial change from the earlier version.”

The implication of Coen’s rebuttal of the Basel IV label is that addressing variability in risk-weighted assets (RWAs) will not fundamentally reshape the existing regulatory capital framework. But many practitioners believe that assumption to be mistaken, particularly given the various initiatives to reduce the influence of internal models. Such is the central role played by risk-sensitive internal models in calculating capital requirements that bankers argue the changes might not only raise the overall quantum of capital, but could also force the recasting of business models.

Reducing variability in RWAs has been on the regulatory agenda ever since the Basel Committee first commissioned a taskforce on simplicity and comparability in 2012. Senior regulators, including the US Federal Reserve’s Daniel Tarullo and the Bank of England’s Andy Haldane, have complained that the capital framework has become excessively complex, making it difficult to accurately compare RWAs across institutional and jurisdictional lines.

Internal models have been widely cast as the source of that complexity, but industry participants continue to defend them as a key component of a framework that is predicated on capital levels reflecting risk. If the use of standardised models is enforced more widely, it could drive capital levels beyond what is required by the underlying risk.

“It’s understandable that inconsistency in the application of internal models may lead to difficulties in comparison between institutions, but it’s not clear that removing internal models altogether is the right way to respond. A very significant proportion of the differences observed reflect real differences in risk profiles,” says Eric Litvack, chairman of ISDA.

“Taking a one-size-fits-all approach to risk and capital would come at the cost of having a framework that does not adapt well to specific markets and risks, and, if insufficiently granular, would likely also lead to an increase in capital charges inconsistent with a Basel Committee and Financial Stability Board objective of not significantly increasing capital requirements,” he explains.

Model restrictions

Nevertheless, the Basel Committee’s direction of travel looks to be set, and risk managers are having to come to terms with a world in which internal models may be outlawed for some risk types and made much more difficult to use for others.

Momentum gathered in January 2016 when the Group of Central Bank Governors and Heads of Supervision (GHOS) that oversees the Basel Committee made a commitment that the work to address excessive variability in RWAs would be completed by year-end. Since then, the committee has published the final framework for the Fundamental Review of the Trading Book (FRTB), which includes much more stringent tests for the use of internal models for market risk capital (see pages 17-19), while separate initiatives look likely to remove the use of internal models for other risk types.

Specifically, proposed revisions to the operational risk capital framework were published on March 4, which centre on a new standardised measurement approach and the removal of the option to use internal models. The model approach was stripped out on the grounds it had become unduly complex, leading to excessive variability in RWAs and inadequate levels of capital being held by some banks.

On March 24, the Basel Committee unveiled further proposals to reduce variation in credit risk-weighted assets, removing the option to use internal ratings-based approaches for certain exposure categories. Crucially, the consultative document also included a decision to eliminate the internal models approach (IMA) for credit valuation adjustment (CVA) risk – a controversial position it had indicated it might adopt, but had not confirmed in its consultation on revisions to the CVA framework in July 2015.

“The removal of IMA-CVA will still cause capital to rise, increasing costs particularly for non-financial counterparties and sovereigns that are exempt from clearing and margining”
– Mark Gheerbrant, ISDA

On the basis that CVA may not be effectively captured by an internal model designed to capitalise market risks in the trading book, and with the implementation of central clearing and margin requirements for non-centrally cleared derivatives set to reduce CVA risk, the committee reasoned that the complexity associated with IMA-CVA was no longer justified. But it’s a position that has left the industry with some concerns.

“It is all too easy for regulators to consider CVA a small piece of overall capital requirements that is falling anyway as a result of clearing and margining and conclude that internal models are unnecessary. However, the removal of IMA-CVA will still cause capital to rise, increasing costs particularly for non-financial counterparties and sovereigns that are exempt from clearing and margining,” says Mark Gheerbrant, head of risk and capital at ISDA.

It was not just the CVA decision itself that caught the industry off-guard, but also the timing. The Basel Committee had only recently launched a second quantitative impact study (QIS) on the CVA framework, and banks were still submitting data when the decision to remove the IMA was revealed. That now leaves banks with just two options for CVA – the standardised approach and the basic approach.

Some have concluded that political pressure from the GHOS and above may have contributed to the sudden decision to remove a key plank of the framework in the middle of the QIS. Participants hope that with the IMA now excluded, regulators will focus on the calibration of the standardised approach to CVA.

“Despite the removal of the IMA, we have still been asked to submit the corresponding data for the QIS, so I am hopeful this means the Basel Committee will use it to calibrate the standardised approach to a more appropriate level. If they can address the outstanding issues, then we would be content to use the standardised approach for CVA as an alternative to an internal model,” says Debbie Toennies, head of regulatory affairs for the corporate and investment bank at JP Morgan in Chicago.

The consultations on revisions to the operational risk capital framework and credit risk-weighted assets were both open for comment until June, so final standards can be expected in the coming months. Beyond the explicit removal of internal models in those reviews, other initiatives are also under way as part of the broader move to increase the comparability of RWAs.

Capital floors

One widely anticipated development is the introduction of capital floors, which would ensure capital across the system does not drop below a certain level, mitigating so-called model risk and reducing variation in capital ratios.

Following a consultation on a conceptual framework for capital floors in early 2015, the Basel Committee has yet to publish anything further, but the GHOS statement in January confirmed it would review full proposals on the design and calibration of capital floors at or around year-end. The level at which the floors are set will be critical in determining the viability of internal models in the future.

“The detail of the capital floors will be very important, because even though internal models have been retained in theory for some risk types, either a very high calibration of the floor or more onerous tests for model approval could effectively still remove the option to use internal models in practice,” says Gheerbrant.

Weighing up all of the recent developments and anticipating the advent of capital floors, it is difficult to imagine a future scenario in which internal models play as significant a role in the capital framework as they have in the past. That’s a prospect some risk managers have found disheartening, as it means they could be forced to adopt industry standard metrics that don’t reflect risk as accurately as current practice.

Some regulators have responded to these concerns by suggesting that models should continue to be used for risk management and pricing purposes and only be removed when it comes to calculating regulatory capital (see pages 20-23). But such an approach would lead banks to manage risk and capital separately – a practice that would turn the tables on the Basel II framework, which attempted to align capital with internal risk management, industry participants say.

“Calculating separate metrics for risk and capital could lead to very different numbers that senior management would need to consider and reconcile when setting risk appetite and making business decisions. We believe regulators should be looking closely at the assets and risks held by banks to determine model eligibility rather than enforcing a simplified standardised methodology,” says Panayiotis Dionysopoulos, director in the risk and capital team at ISDA.

As well as increasing bank capital, enforcing the use of standardised rather than internal models could also pose broader systemic risks. If the vast majority of banks were to use standardised metrics to calculate capital, it would inevitably reduce the diversity of business models across the sector, which is recognised as a risk to the system at times of stress.

“We are concerned by a widespread move towards standardised approaches to capital requirements, because it creates a herd mentality where banks are using the same model to calculate their capital levels. That creates incentives for banks to pursue the same business model, which can be a source of systemic risk when things go wrong,” says Toennies.

“We are concerned by a widespread move towards standardised approaches to capital requirements, because it creates a herd mentality where banks are using the same model to calculate their capital levels”
– Debbie Toennies, JP Morgan

Regulatory support

The arguments against restrictions on internal models have found support among some in the regulatory community. Speaking at ISDA’s annual general meeting in Tokyo on April 13, Nobuchika Mori, commissioner of Japan’s Financial Services Agency (JFSA), raised the possibility that the pendulum may have swung too far.

Highlighting the numerous new regulations that are now constraining bank balance sheets, including multiple capital ratios, liquidity ratios and the leverage ratio, Mori suggested regulators may have intervened too deeply into banks’ internal practices, based on the assumption that innovation in risk management could be abused.

“It looks as if a bank’s safety and soundness were surrounded and protected by many layers of thick, defensive walls. Yet, we had better think carefully whether thick walls are enough to attain our dual goal of financial stability and growth,” said Mori. “Instead of blindly trusting the thickness of the walls, we need to assess and strengthen the entire framework of prudential regulatory and supervisory policy.”

The JFSA is in the process of moving from a framework of static regulation to what Mori terms “dynamic supervision”, whereby the impact of new rules are more closely monitored and modified where necessary, with supervisors paying particular attention to banks’ relationships with the capital markets and the real economy. A consultation paper is expected to be published soon, which will set out the regulator’s plans in greater detail.

While other jurisdictions may choose to adopt similar approaches in the future, it appears that international standards are moving towards a stricter, less risk-sensitive capital framework. As those standards are enforced in the coming years, it will create a need for more stringent supervision at the national level to ensure risks are still being appropriately managed.

“It is perhaps clear that the Basel Committee is looking to return to its core mandate: writing the standards for banking supervision, rather than direct regulation. However, the interplay between a simplistic, international standard and the reality of national legislation makes those standards of supervision key to preventing future banking crises,” says Henry Wayne, managing director, regulatory reform in Citi’s risk analytics division in London.

In that context, it is perhaps understandable that the Basel Committee sees its current work programme as more of an extension to Basel III than the foundation of Basel IV. But finding an effective balance between risk sensitivity and simplicity that ensures a resilient banking sector with greater consistency in RWAs may turn out to be an elusive goal. While the Basel Committee consulted on that balance in 2013, Wayne believes the industry may not have paid sufficient attention at the time.

“We are now starting to see more direct intervention in the level of risk sensitivity on multiple fronts. As a result, it’s not clear that current international standards are really adapted to the realities of today’s markets, following the structural reform brought about by other silos of regulation. This remains a concern,” he says.