Less Model Reliance Should Reduce Bank Ratio Variation – Fitch

The Basel Committee on Banking Supervision has proposed that banks should stop using models to calculate capital for some hard-to-model portfolios and face significant constraints on model usage for others. If adopted, this should reduce variation in capital adequacy ratios across banks, says Fitch Ratings. But this increases the need for the Committee to develop a more risk-sensitive standardised approach (SA).

The proposals, published earlier this month for consultation until 24 June, eliminate the use of Internal Ratings Based (IRB) models for low-default exposures to banks, financial institutions, and large corporates with assets in excess of EUR50bn. This means that all banks would use the revised SA for calculating risk weightings on these asset types.

Low-default portfolios are difficult to model because data is limited and historic default experience is low. Such data is needed for reliable IRB modelling. Although not included in these proposals, sovereign exposures are also regarded as low default, and we think it is likely that the Committee might adopt a similar approach for this portfolio once its sovereign risk-weight review is concluded.

The Committee’s proposals would still allow IRB modelling for exposures to smaller corporates and for retail customers, subject to restrictions to narrow the range of outcomes.

For example, proposals to increase minimum probability of default (PD) assumptions for retail mortgages to 5bp from 3bp could increase risk weights on these portfolios by 50% if all else remains equal. For corporates with revenues above EUR200m but assets below EUR50bn, models are still allowed, but loss-given default assumptions (LGD) on unsecured senior lending will be fixed at 45%. Currently, banks can use internal model estimates to calculate capital charges for these exposures that may be less conservative. The introduction of a minimum floor to the models will mean outputs are more comparable across banks.

For the first time, the Committee revealed its thoughts on the calibration of an aggregate permanent risk-weighted, asset-capital floor based on the revised SA, to be in the range of 60% to 90%. This will limit the benefit to banks from lower risk weights generated by their IRB models, compared with the revised SA weightings.

If the Committee’s proposals are adopted, the revised SA will become far more important than the IRB approach for calculating capital requirements. Reducing banks’ reliance on internal models could boost public confidence in regulatory capital ratios, and enable creditors to make better informed decisions.

Banks have developed IRB models at significant cost and we expect them to lobby hard to maintain incentives to continue to use the modelled approaches. The Committee does not intend to raise overall capital requirements for banks, but we think these proposals could lead to an increase in capital requirements for low-risk weight portfolios. Some banks might question their continued investment in internal models, although we think many might still use them for their own risk management. We think this would be useful because models can create more robust risk-management frameworks.

An earlier Basel Committee study signalled notable differences in how banks estimate key model parameters including PD and LGD for the same exposures. The review concluded that a significant source of risk-weighted asset variation was due to different modelling choices between banks, such as the definition of default, and adjustment for cyclical effects. In contrast, the SA reduces variability because it prescribes set risk weights for different risk categories.

Author: Martin North

Martin North is the Principal of Digital Finance Analytics

Leave a Reply