QIS3 FAQ: E. Credit Risk Mitigation

1. When there is more than one type of collateral or when there is both collateral and guarantee covering an exposure, how should banks sub-divide the exposure in calculating the risk mitigation effect?

Answer: When there is a difference in the risk-weighted assets depending on how the exposure is sub-divided, banks should calculate the effect of mitigation in a way that maximises the capital benefit of risk mitigation (i.e. the way that minimises the amount of risk weighted assets). Generally, for different types of collateral, this would mean calculating the effect of the risk mitigant in the same order as in the table presented in paragraph 256.

2. Footnote 63 states that '(a) lower LGD may be substituted ... when the guarantee is supported by eligible collateral. Are there any conditions that such collateral must meet?

Answer: The lending bank must have clear rights over the collateral, and must be able to liquidate or take legal possession of it, in a timely manner, in the event of default, insolvency or bankruptcy (or otherwise-defined credit event set out in the transaction document) of the borrower, even if the guarantor is not in default. All the minimum requirements set out in paragraphs 80-81 and 86-89 need to be met.

3. How should the maturity for repos governed by master-netting agreements be calculated under the IRB approach when there is an explicit maturity adjustment?

Answer: The weighted average maturity of the transactions under the master-netting agreement should be used, with a 5-day floor applied to the average. The nominal value of each transaction should be used for weighting the maturity.

4. The rules say that no transaction using CRM techniques should obtain a higher capital charge than the same transaction without such techniques would receive. In some cases, however, using the substitution treatment for guarantees may lead to higher capital charges (e.g. if a retail exposure is guaranteed by a bank with a relatively high PD). How should we proceed in such cases?

Answer: In such cases you should ignore the guarantee. If using substitution treatment would result in higher capital charges, capital charges must be calculated as if the guarantee were not available.

5. At multiple places the technical guidance indicates that banks must ensure that their adjustments to PD and/or LGD estimates do not reflect double default effects. What exactly does that mean?

Answer: If a bank has an exposure to a counterparty with a PD of 1% that is guaranteed by a counterparty with a PD of 0.5% the risk mitigating effect of this guarantee is recognised by allowing the bank to treat this exposure as if it were an exposure to the guarantor rather than the original obligor. A truly risk sensitive model would also recognise the effect that default of the guarantor is only an issue when the original obligor is also in default. In an ideal case-when defaults of the obligor and the guarantor are fully independent-this would imply that capital requirements could be based on a PD that equals PDobligor X PDguarantor, a number which would be considerably smaller than either the PD of the obligor or that of the guarantor and consequently there would be a considerable difference in capital requirements. The difference between both numbers is called the double default effect. The true double default effect is highly dependent upon the correlation between obligor and guarantor at the moment of default of the obligor. Estimation of this correlation is beyond the scope of the new capital accord, and consequently any double default effects should be ignored for purposes of calculating capital requirements.

6. If a bank is using a credit risk mitigant, like insurance, that effectively functions like a guarantee is it allowed to treat such risk mitigants as an ordinary guarantee?

Answer: Yes, provided that such a products meets the operational requirements for guarantees laid down in paragraph 154 to 165 of the Technical Guidance any product may be treated as a guarantee.

7. Paragraph 171 of the Technical Guidance indicates for unrated first-to-default derivatives 'the risk weights of the assets included in the basket will be aggregated and multiplied by the nominal amount of the protection provided by the credit derivative to obtain the risk weighted asset amount.' What exactly does aggregation mean in this case?

Answer: The term aggregation implies that you will have to use the sum of the risk weights on the individual assets. If there were three assets in the basket and each had a PD of 1%, the resulting risk weight would be 3 times 97.44% which equals approximately 292%.

8. For the IRB approach: what exactly are the dividing lines between residential mortgages, real estate as collateral in the IRB approach, income producing real estate and high volatility commercial real estate?

Answer: Unfortunately there is no clear answer to this question, borderline cases are unavoidable. Generally speaking a bank would first look whether the loan could be assigned to the retail portfolio. In order to be eligible (paragraph 194), the exposure should be secured by residential properties. In such cases the size of the loan is irrelevant, both first and subsequent liens qualify; the loan, however, has to be extended to an individual that is the owner-occupier of the property (paragraph 192). If the latter condition is not met, the loan should be treated as corporate. Supervisors, however, have some discretion regarding the inclusion of buildings containing only a few rental units or loans secured by a single or small number of condominium or co-operative housing units in a single building or complex (paragraph 192).2


      When a loan is not eligible as a residential mortgage it should be treated as a corporate loan. Whether the real estate collateral reduces LGD depends on the extent of collateralisation (as described in paragraph 256 of the Technical Guidance) and on whether the set of eligibility criteria described in paragraphs 455-458 of the Technical Guidance has been met.

      When the loan is collateralised by income producing real estate banks should use their own loss experience and supervisory prescription in order to determine whether the real estate involved should be treated as high volatility commercial real estate (HVCRE). If this is the case banks will be required to map their internal risk grades to five supervisory categories, each of which is associated with a specific risk weight. If this is the case, application of the special HVCRE is obligatory for both foundation and advanced IRB banks (see paragraph 215). If the exposure is not included in the HVCRE exposure class the exposure probably should be classified as income producing commercial real estate. For this category of exposures (see paragraphs 212-214), banks that do not meet the requirements for the estimation of PD under the corporate foundation approach will be required to map their internal risk grades to five supervisory categories, each of which is associated with a specific risk weight.3 Banks that meet the requirements for the estimation of PD will be able to use the foundation approach to corporate exposures to derive risk weights for these exposures. Banks that meet the requirements for the estimation of PD and LGD and EAD will be able to use the advanced approach to corporate exposures to derive risk weights for these exposures.

9. I have an exposure of € 100 against which the obligors pledges € 20 of eligible real estate and € 20 of eligible other collateral. My reading of the second bullet of paragraph 257 is that this collateral is eligible, since, although the individual amounts do not meet the 30% threshold, the sum (€ 40) does meet it. Is this interpretation correct?

Answer: Yes, you are correct. This exposure does meet the minimum required degree of collateralisation and consequently you can assign an LGD of 35% to € 20/140% (for the real estate collateral and an LGD of 40% to € 20/140% (for the other collateral). The remainder of the loan (€ 71.43) falls in the 45% LGD-band.

10. In reviewing the table for the standard supervisory haircuts in paragraph 114 of the Technical Guidance, I've noticed what appear to be two different sets of ratings. Are both the Standard and Poor's and Moody's ratings used in this table?

Answer: The ratings used in paragraph 114 are Standard and Poor's long-term and short-term rating grades. For example, the reference to 'AAA-AA-/A1' refers to a long-term rating range of 'AAA to AA-' and a short-term rating of 'A1.' Both long-term and short-term ratings are presented because the haircut is applied to collateral with varying residual maturities.

11. A n-th-to-default swap on a pool of assets (e.g. 100 corporate loans, the nominal amount of each loan is € 10 million) fulfils the criteria of the definition in paragraph 486 of the Technical Guidance (specifically a) and c)). Should this swap be treated under the securitisation framework?

Answer: No, a n-th-to-default swap should be treated according to the rules set for credit risk mitigation (paragraphs 170-173).

2 For QIS-purposes, national supervisors may give more detailed prescriptions to their banks.

3 When one of the supervisory slotting approaches has been applied, eligibility of collateral is no longer an issue, since the availability of collateral is reflected in the slotting criteria themselves.

Return to QIS3 FAQ Table of Contents