BCBS has recently released consultative document,
Reducing variation in credit risk-weighted assets – constraints on the use of internal model approaches. It proposed that banks be barred from using internal models when they calculate how risky certain assets are. Lending to other financial institutions and large companies, and holding equities are the areas most affected.
The proposed changes to the IRB approaches set out in this consultative document include a number of
complementary measures that aim to:
1. Reduce the complexity of the regulatory framework and improve
comparability
2. Address
excessive variability in the capital requirements for credit risk.
Specifically,
the Basel Committee proposes to:
- Remove the option to use the IRB approaches for certain exposures, where it is judged that the model parameters cannot be estimated sufficiently reliably for regulatory capital purposes
- Adopt exposure-level, model-parameter floors to ensure a minimum level of conservatism for portfolios where the IRB approaches remain available
- Provide greater specification of parameter estimation practices to reduce variability in risk weighted assets (RWA) for portfolios where the IRB approaches remain available.
Under the proposal, banks will have to use a standardised method of calculating the riskiness of loans to financial institutions and to large corporates with assets of more than €50bn. The Basel group believes there is so much publicly available information on the credit risk of such institutions, that banks are rarely able to provide a better estimate than an approach standardised by regulators.
These proposed regulation are already dubbed as
BASEL IV by street. Banks have already spent billions of dollars to comply with the plethora of regulatory changes after the 2008 financial crisis.
Argument is if banks will not be able to use the IMM models for most of the derivative portfolio, as most of the derivative trades happens among financial firms, then how much benefit will banks have in developing highly complex IMM models.