5th April 2017

Making the PRA’s proposals on modelling LGD work in practice for banks on the standardised approach

Written by Simon Hills, Executive Director, Prudential Regulation and Risk

One of the most interesting proposals in the PRA’s recent consultation paper on its requirements for banks wishing to move from the standardised approach (SA) to calculating credit risk using an Internal Ratings Based (IRB) approach, is the option to allow the use of external data to model Loss Given Default (LGD), perhaps by using the Council for Mortgage Lenders data on long run average default and recovery data.

This is an important and welcome development as there is a general perception that banks using the SA to calculate regulatory capital for credit risk, hold significantly more capital for certain portfolios, such as low loan to value (LTV) mortgages, compared to IRB banks. This potentially conflicts with the PRA’s secondary objective to support competition in financial services.

The PRA is proposing that a bank can apply for IRB before it has had a fully compliant rating system in place for at least three years, as long as the framework has been reviewed in one full annual cycle since the board gave internal approval for the use of internal rating systems for credit decisions, and lending and risk appetite policies.

It is important to understand that this proposed approach is compliant with exciting Capital Requirements Regulations about the use of modelling approaches.

Many banks seeking to move from a standardised approach may not have statistically adequate internal data on either Probability of Default or Loss Given Default (LGD) which are key inputs into IRB modelling. This lack of default data is exacerbated in a low interest rate environment in which there will be fewer stressed mortgage borrowers and thus lower rates of default. So the PRA is proposing that UK-based external data may be used to supplement internal data the bank already has.

But this ability to use external data is constrained. In particular the PRA is proposing that ‘reference points’ should be used in calculating the Probability of Possession Given Default (PPGD) one of the key components that drives LGD and is one of the most complex metrics to model. The impact of this is that the capital benefits of moving from a standardised approach to an internal modelling approach may be limited for loan-to- values in excess of 70%.

Although PRA’s proposals are a significant step in the right direction, it remains to be seen whether a standardised bank decides to move to a modelled approach, with the attendant costs of approval and ongoing support.

A parallel Basel Committee on Banking Supervision (BCBS) work stream is (still) considering the recalibration of the standardised approach and may recommend credit risk weights that are less than currently specified, particularly for lower LTV residential mortgages.

How many banks decide to dip their toe in the world of credit modelling is likely to be as dependent on the outcome of the BCBS deliberations as on the PRA’s clarified approach to transitioning to an IRB approach.

Please register or login to add this to your interests.