The key points from these recent discussions, where the regulatory response may include some combination of:
- Greater disclosure by banks to explain how their risk weightings correlate with the underlying risks, and to justify the differences between model-based and standardised risk weightings;
- Tighter regulatory constraints on the use and specification of internal models; and
- Greater convergence in supervisory practices.
Implications for firms…
- The most important implication for banks may be in the form of higher capital requirements. For example, if a bank had to increase its risk weighting on mortgage lending from 10% to 15% this would require a 50% increase in the minimum capital the bank would need to hold against this business.
- Enhanced disclosure, tougher and tighter model specification requirements, and other operational requirements would increase the cost of internal modelling at the same time that the capital benefits were being restricted. This might drive some banks to switch away from internal model-based approaches to the use of the standardised approach for some or all of their business lines.
- These capital and other costs will have implications for pricing and for the attractiveness of different business activities.
- Tighter restrictions on model specification may drive a wedge between models used to calculate risk weightings and the models that banks would prefer to use for other purposes such as pricing, capital allocation and the calculation of risk-adjusted returns. This would be inefficient and costly, and could mean that banks would no longer pass the regulatory "use test" for internal model-based approaches to calculating risk weightings.
- Disclosure requirements may provide an opportunity to link public explanations of model results with enhancing senior management and board understanding of internal models and their input to risk management decisions.
In the detail…
Six pressures can be identified in this area, the combined impact of which is likely to be significant.
First, some national regulators, the IMF, market analysts, investors and academics have highlighted differences in the risk weightings used by banks, and have struggled to explain the reasons for these differences. As a result, investors and other stakeholders have begun to lose confidence in the consistency and comparability of banks' risk weighted assets as determined through the use of banks' internal models.
For example, a survey undertaken last year by Barclays1 of 130 bank investors at nearly 100 institutions found that:
- 63% of investors said that their confidence in banks' risk weighted assets (RWAs) had gone down in the last year.
- 52% of investors did not trust risk weightings, and only 13% did have trust in them.
- 84% of investors wanted simpler risk weighting calculations, greater transparency and the removal of model discretion.
Second, although the post-crisis Basel 3 package of capital and liquidity requirements maintained the model-based approaches to credit, market and operational risk, some individual regulators (including Andrew Haldane in the UK and Thomas Hoenig in the US) have called for a move to a less complex approach. This might place greater emphasis on a bank's leverage ratio, which would not involve either standardised or model-based risk weightings. This in turn would (i) place a higher floor under the model-based approach, so limiting the extent to which a bank could benefit from using its own internal calculations of risk weightings, or at least (ii) increase the pressure on banks to explain and justify – to regulators and other stakeholders – the gap between their leverage ratios and their capital ratios based on risk weightings.
Third, after the financial crisis highlighted a number of failures in the trading book regime – including inadequate capital held against market risk and excessive latitude in determining which assets could be placed in the trading book – the Basel Committee on Banking Supervision issued in May 2012 its proposals for a 'Fundamental Review of the Trading Book (PDF 556 KB).' These proposals included:
- Changing the underlying model methodology from 'value at risk' to 'expected shortfall' – a fundamental change that would add modelling complexity, and increase capital requirements for many assets;
- Changing the basis of definition for trading book assets to limit the assets that a bank could include in its trading book;
- A more detailed assessment of illiquidity risk - with additional capital add-ons for instruments at greater risk of illiquidity under stress; and
- Narrowing the differences between internal models and the standardised approach – which could include requiring banks using a model-based approach to calculate in addition what their capital requirement would be under the standardised approach; a closer alignment of allowable approaches to hedging and diversification; limiting the potential capital benefits of internal models relative to the standardised method by introducing a floor or surcharge (based on capital requirements under the standardised method); and significantly enhancing the operational requirements around the approval and maintenance of internal models.
Fourth, the G20 and the Financial Stability Board welcomed and endorsed the October 2012 recommendations of the Enhanced Disclosure Task Force (PDF 12.5 MB) (EDTF - a private sector group comprising banks, investors and audit firms) to enhance the risk disclosures of banks. These recommendations included enhanced disclosure by banks of how risk weighted assets are calculated, the impact of the use of internal models on a bank's regulatory capital requirements, how banks' internal ratings grades map across to external credit ratings, and how internal models are back-tested and validated. The intention of these enhanced disclosures would be for a bank to justify why its model-based risk weightings differ from the standardised approach risk weightings and why they are a good reflection of the underlying risks faced by the bank.
Large international banks are expected to adopt voluntarily the EDTF recommendations, in some cases beginning with their end-2012 annual reports and Pillar 3 disclosure documents.
Fifth, the Basel Committee and the European Banking Authority (EBA) have released the preliminary findings of their analysis of differences in risk weightings across banks.
The Basel Committee published in January 2013 a report (PDF 419 KB) on divergences in risk weights for banks' trading book assets. The report showed that there is considerable variation across banks in average published RWAs for trading assets, and that it is not possible to tell from public disclosures how much of this variation reflects differing levels of actual risk as opposed to other factors. In addition, a test exercise based on a hypothetical portfolio of market risk positions also showed substantial differences (by a factor of almost three) between the highest and lowest RWAs reported by banks. This was found to be due to a combination of (i) supervisory decisions applied either to all banks in a jurisdiction, or to individual banks; and (ii) the modelling choices made by banks.
Although the report made no policy recommendations, it highlighted three potential policy options that could be considered in the future:
- improving public disclosure and regulatory data collection to aid the understanding of RWAs – which would be consistent with the recommendations in the EDTF report;
- narrowing down the modelling choices for banks; and
- further harmonisation of supervisory practices with regard to model approvals.
Meanwhile, the EBA published in February 2013 a report on RWA differences across European banks' banking book (credit risk) exposures. This showed that the most important differences across banks in credit RWAs emerge in their retail and corporate exposures (because the RWAs on sovereign and bank exposures are both lower and less dispersed). In the sample of nearly 100 banks used by the EBA, the RWAs on non-defaulted exposures ranged from 9% to 35% for retail and from 42% to 95% for corporate.
Taking each bank's overall credit RWA, the EBA study found that half of the differences across banks are due to a combination of (i) the structure of the balance sheet (the balance between retail, corporate, bank and sovereign lending) and (ii) whether a bank is on the standardised or internal model (IRB) approach to calculating RWAs.
The other half of the differences across banks are due primarily to differences in model parameters (probability of default, loss given default and exposure at default). These may reflect (i) genuine differences in risk (eg one bank's corporate lending is genuinely less risky than another bank's lending); (ii) different model specifications (eg length of data periods); or (iii) supervisory differences (different minimum parameter values imposed by national supervisors). Disentangling these "model parameter" factors will require a "bottom-up" assessment, on which the EBA will work next.
Finally, national regulators are beginning to introduce regulatory requirements that are consistent with this overall direction of travel. For example, the Hong Kong Monetary Authority has announced that it will impose a floor of 15% on the risk weighting of new retail mortgage lending, compared with risk weightings of around 10% that are understood to be generated by banks' internal models.
1 BYE BYE BASEL? Making Basel more relevant, Barclays 23 May 2012