BCBS consultative document: Revisions to the standardised approach for credit risk
Joint comments from Finanstilsynet (Financial Supervisory Authority of Norway) and Norges Bank. Letter of 25 March 2015 to Basel Committee on Banking Supervision.
We welcome the opportunity to comment on the 22 December 2014 proposal on Revisions to the Standardised Approach for credit risk. Experiences during the financial crisis of 2007-2008 highlighted the need for enhanced prudential standards for the banking sector. In this context, the continued suitability of both standardised and internal ratings-based (IRB) approaches for calculating capital requirements has been a key priority of recent bank regulatory efforts. While risk-sensitive capital requirements may appropriately reflect the inherent riskiness of exposures on bank balance sheets, such requirements can also result in increased procyclicality in economies. This consideration presents a challenge, balancing transparency and prudence, in the appropriate design of effective banking standards.
Finanstilsynet and Norges Bank are generally supportive of the proposal to make the standardised approach more risk-sensitive and the risk weights less dependent on external credit ratings. In particular, we are in favour of revising the standardised approach as proposed for the purpose of imposing a permanent standardised approach floor on IRB-modelled capital requirements for credit risk. However, making the standardised approach more risk-sensitive will also make it more procyclical.
It is important that the standardised approach reflects to a reasonable extent the riskiness of exposures, in order to provide sound incentives for the banks employing this approach in their lending policies. In that manner, the standardised approach may also represent a better and more realistic alternative for banks that would otherwise consider applying for IRB approval. However, we agree that the standardised approach should not rely on internal-modelled approaches to set capital charges. Although it is essential to reduce the reliance on external ratings, this should not preclude a limited role for external ratings when setting capital requirements.
Specific comments on the proposal are noted below:
Exposures to banks
Finanstilsynet and Norges Bank support that the risk drivers for exposures to banks are based on capital adequacy and non-performing assets. As regards the proposed measure for capital adequacy, we believe the use of the CET 1 ratio is better than the Tier 1 ratio since to date there is little experience on the impact of a write-down or conversion of Additional Tier 1 instruments on a bank. We also believe CET 1 capital is best in the sense that it is the capital of those who own and determine the strategies and goals of the bank. However, basing the measure for capital adequacy on the CET 1 ratio alone will not provide an extensive and accurate risk measurement. Specifically, there are significant differences in IRB-model derived risk-weighted assets (RWAs) arising from varying banking practices and supervisory interpretations. Banks in some countries may therefore have high CET 1 ratios resulting from the use of lower risk weights relative to other banks for a given level of risk; this could be observed through relatively lower leverage ratios. A risk-based capital measure also might lead to some counterintuitive results (as mentioned in the proposal), as higher capital ratios may be imposed for example under pillar 2 due to higher risk; however, such counterparties would receive lower risk weights under the proposal. The leverage ratio is relatively simple to calculate for all banks. We suggest that the leverage ratio is used to refine the risk driver for capital, so that it distinguishes to a greater extent between actual solvency levels across different banks. This could be done in a complementary manner, such that the observed leverage ratio can result in a bank being moved up or down one risk class. Alternatively, the leverage ratio may only be used to move a bank to a higher risk class if the bank has a leverage ratio below a specified level.
With respect to asset quality considerations, we believe the measure of asset quality should be as clear and objective as possible, and explicitly set forth by the committee. Such a measure should limit subjectivity inherent in accounting, banking and/or supervisory interpretation, and therefore serve to reduce potential differences in comparability across various banks and jurisdictions.
Under the EU's Capital Requirements Regulation (CRR), covered bonds are eligible for preferential treatment under strict conditions. We believe covered bonds should continue to receive lower risk weights relative to the risk weights determined for the general treatment of bank exposures. In Norway, there is a fairly large volume of covered bonds issued by residential property mortgage companies within strict loan-to-value limits. These covered bonds are safer than regular exposures to banks as they have a preferential claim over the institution's cover pool.
Exposures to corporates
Finanstilsynet and Norges Bank support the proposal that specialised lending will be defined in line with the IRB approach and classified as a subcategory with higher risk weights. We are also in favour of risk categories for senior corporate debt exposures, but recommend that profitability considerations, if feasible, be used instead of turnover, and that free cash flow is included as a complementary factor. For example, weak free cash flow would result in the exposure being moved to a higher risk class. With regard to the calculation of the leverage ratio, we believe that material off balance sheet exposures should be included.
Subordinated debt, equity and other capital instruments
Concerning the proposal for risk weights of regulatory capital instruments issued by financial sector entities, there is a different treatment for significant and non-significant investments. Significant equity investments in financial sector entities that are not deducted are risk-weighted at 250%, while non-significant equity investments that are not deducted are risk-weighted at 300% or 400%. We believe the risk weights for non-significant equity investments in financial sector entities should not exceed the risk weights for significant equity investments in financial sector entities.
Finanstilsynet and Norges Bank have previously investigated a number of data analyses of losses and concluded that the lower capital requirements for enterprises in the retail portfolio are not well supported. Retail treatment of a small business will also require stronger validation for the standardised method compared to the IRB method because higher PD values for small businesses under IRB may offset the reduction in capital charges provided by the retail treatment. In addition, expected losses for small businesses are significantly higher than those for larger enterprises, and the expected losses are not always fully priced into a bank's interests and fees. We therefore believe that retail exposures to enterprises should not be eligible for the preferential risk weights.
If lower risk weights for regulatory retail exposures are maintained for some exposures, we agree that the 0.2% numerical limit should be a binding regulatory standard and that additional risk drivers should be applied. In this respect, consideration could be given as to whether risk drivers based on the percentage of borrower income available to service the loan and the extent to which the loan is secured by durable goods provide a better differentiation.
Exposures secured by residential real estate
Finanstilsynet and Norges Bank agree that the LTV (loan to value) ratio and DSC (debt service coverage) ratio are suitable ratios for determining the risk weights for exposures secured by residential real estate. An alternative to the DSC ratio, as defined, is the measurement of the borrower's real ability to service the mortgage including normal living expenses. However, it is essential that the ratio can be measured objectively.
With respect to the design of LTV buckets, the calculation of risk weights should be completed on a consistent basis such that a loan at a given LTV will result in the same total risk charge if maintained on one bank balance sheet or across several1. A solution could be to impose a risk weight for the part of the loan that has LTV < 40%, a higher risk weight for the part of the loan corresponding LTV in the interval between 40% and 60%, an even higher risk weight for the part of the loan corresponding LTV in the interval between 60% and 80%, etc. Consideration should be given to the appropriateness of the number of LTV buckets and whether it is practical to maintain several buckets with narrow LTV ranges.
Concerning the LTV ratio, we believe it can be too strict to keep the value of the property constant as measured at origination for purposes of calculating the LTV ratio. The mortgage interest rate will be linked to the underlying risk; therefore, the capital charge (and subsequently, interest rate) should shift according to a change in the LTV ratio, even if this results in increased procyclicality. If this mechanism does not function, a borrower could switch banks in order to obtain a new mortgage with a more favourable interest rate that is linked with the property's new LTV and corresponding lower capital charge. At the same time, it is also crucial that property values are adjusted down, if there is a fall in property prices. Institutions are required to monitor the value of residential real estate at a minimum of once every three years for residential real estate, and more frequently where the market is subject to significant changes in conditions. Moreover, in Norway we maintain comprehensive and reliable statistics on property prices, which certainly many other countries will also have.
With respect to the DSC ratio, we believe such a ratio will be too difficult to keep updated on an on-going basis. However, banks should check the borrower's DSC ratio if they suspect a significant impairment in the borrower's ability to service the mortgage. We also believe a bank should be able to update the DSC ratio if a borrower negotiates with the bank in order to obtain a lower interest rate on the basis of a clearly improved DSC ratio. Otherwise the borrower may have a financial incentive to obtain a new loan in another bank and replace the old loan. Some additional division of risks made to the DSC ratio, should be considered.
Exposures secured by commercial real estate
Finanstilsynet and Norges Bank view no recognition of the real estate collateral (option A) as an appropriate option for determining the risk-weight treatment for exposures secured on commercial real estate. The loss experience on commercial property lending in Norway indicates that the risk is extremely volatile. Whilst in normal years, losses on commercial property can be very low, in crisis periods such losses have represented a large proportion of total loan losses.
Risk-weight add-on for exposures with currency mismatch
Finanstilsynet and Norges Bank support a risk weight add-on for exposures with currency mismatch applied to exposures secured by residential real estate and retail exposures. A way to address currency mismatch in a simple manner is to move exposures with currency mismatch one risk class up.
Off balance sheet exposures
Finanstilsynet and Norges Bank have believed that a credit conversion factor (CCF) of 0% has not been appropriate in relation to the underlying risks. Therefore, we agree that commitments which may be cancelled unconditionally without notice should receive a CCF of at least 10%. Credit card loans are often the first credit type upon which a borrower defaults. As such, these facilities can often quickly result in losses prior to their closure, suggesting that a related credit risk capital charge is warranted.
|Erik Lind Iversen
|Ida Wolden Bache
General Executive Director
Norges Bank Financial Stability
1Suppose DSC < 35% and LTV = 79%. If Bank A grants the entire loan, the risk weight for the loan according to the proposal will be 40%. If Bank A grants an amount that corresponds to an LTV of 59% and the customer obtains the remainder of the loan from Bank B, the average risk weight for the two loans will be: (59/79) * 30% + (20/79) * 40% ≈ 32.5%. Banks can in this way cooperate to bring down the capital requirement. For the average risk weight of the whole loan to be 40%, Bank B should be charged a risk weight of: (40% - (59/79)*30%) / (20/79) ≈ 69.5% for its part of the loan.