On the 3rd of June 2016, the Institute of International Finance (IIF) responded to the Basel Committee on Banking Supervision (BCBS) consultative document on reducing the variation in credit risk-weighted assets (RWA). The IIF has expressed several concerns with the BCBS proposals, primarily on the impact on RWAs and the loss of granularity in risk measurement. We broadly agree with the alternative proposals suggested by the IIF which is seeking to maintain a level of risk-sensitivity in regulatory capital calculations, to adjust the proposed parameter floors, as well as promoting cooperation across banks by centrally pooling data for observed performance of large corporates. The blunt nature of the BCBS proposals appears to unwind the progress made in embedding risk sensitive Regulatory Capital estimates since the financial crisis.
However, our primary concern with the BCBS proposals (not discussed in significant detail in the IIF response) is the further separation of risk estimates produced for Regulatory Capital and the Expected Credit Losses (ECL) calculated under the new IFRS 9 impairment measurement approach. This is despite the industry seeking to align the accounting and Regulatory Capital perspectives for many years.
The current, widely-adopted, approach to IFRS9 modelling utilises the IRB infrastructure as a foundation (with the removal of the Regulatory Capital constraints) upon which IFRS 9 compliance amendments are developed. This is encouraged by the IASB, but also by the BCBS itself when it says in article 28 of the Guidance on credit risk and accounting for expected credit losses published in December 2015 that
“Using common underlying processes (i.e. systems, tools and data) across a bank to the maximum extent feasible could reduce cost and potential bias and also encourage consistency in the measurement, management and reporting of credit risk and ECL.”
Consistency, however, will be harder to achieve for those asset classes transitioning to Standardised capital estimates, given that credit decisions, pricing and Risk Appetite will be based on the current Advanced IRB models, the impairment metrics will be calculated using a Point in Time (PiT), non-conservative version of those Advanced IRB models, and RWAs will be on a completely different basis. This goes against the principle behind the Basel IRB Use Test according to which banks need to use their AIRB models in their Risk Management – if the models are not good enough for RWA calculations, then why would they be good enough for pricing and credit decisioning?
It seems that, with such blunt censoring of internal risk estimates within the BCBS proposals, the most accurate measure of a bank’s true level of risk (albeit on a Point in Time basis) will emerge from the IFRS 9 Impairment estimates, not the RWAs. The IFRS 9 forward-looking Impairment approach may also serve as a more accurate tool for Stress Testing given the BCBS proposal for numerous floors to risk parameter estimates in the Regulatory Capital models.
We believe the comparability assessment conducted by the BCBS in July 2013 (used as evidence for the BCBS proposal) has highlighted some excellent areas for improvement in reducing variability between banks. However, the proposed solutions appear excessively blunt given that a middle ground exists where banks can maintain a certain level of risk sensitivity. Such comparability assessments are a great tool for the BCBS and the local supervisors, and a more regular, formal completion of such exercises could help maintain consistency within the industry. This would rapidly facilitate early identification of over or under estimation of RWAs in given asset classes which can then be subsequently adjusted accordingly by, for example, calibrating a model to a common level of risk.