Answer: Not automatically. In order to be eligible for the 0% conversion factor such overdraft facilities should be unconditionally and immediately cancellable. For retail exposures banks always have to estimate EAD (or include usage of lines in their estimates of LGD). See paragraph 275 of the instructions for further guidance.
2. The Retail IRB framework makes reference to asset maturity being 'subsumed in the correlation assumption,' suggesting the risk weight functions have been calibrated for maturity. Does this have any impact on the way PDs should be calculated?
Answer: No. The fact that the maturity is subsumed in the correlation assumption just implies that for retail mortgages no explicit maturity adjustment is required. This decision was based on the consideration that the introduction of a separate maturity adjustment for retail mortgages would be too complicated since this would require a separate analysis of prepayment risk and transition behaviour of mortgage counterparties, etc.
3. A common method of deriving PD is to use monthly or quarterly data from a particular pricing segment within a portfolio. A potential issue arises if the observations are drawn from a growing portfolio. The loss information will be biased to the early portion of the loss vintage curve, and the early portion of the loss vintage curve is not fully ramped. Therefore, losses will be lower and PD will be understated. Is this acceptable?
Answer: In general, the Committee wants to look to strong internal bank practices as a guidepost. Where a bank believes the issue identified here is material, presumably the bank is considering whether additional steps (i.e. segmentation by vintage) are necessary to achieve appropriate estimates of risk and economic capital. The Committee has stepped back from explicit mandatory segmentation requirements in areas such as this because of concerns with excessive burden and complexity, but with the understanding that banks will take the appropriate steps to deliver unbiased estimates of PD. Over time, of course, approaches that do not deliver unbiased estimates will be shown to be inadequate through the results that they produce.
Answer: For the purpose of calculating correct capital requirements, banks can assign defaulted exposures to any maturity band they like since the maturity correction is a function of PD and when PD equals 1 (as is the case for defaulted loans) the maturity correction becomes nil, i.e. for defaulted assets capital requirements do not depend upon maturity. The QIS-templates automatically take account of this and include such exposures in the column 'maturities exempted from the explicit maturity adjustment'.
Answer: For purposes of QIS the bank should first determine whether the portfolio meets the retail definition. In that case it should be included in the retail portfolio using average PD, LGD and EAD figures for homogeneous buckets of this pool of assets (for purposes of QIS the bank may treat the whole portfolio as a single bucket if completing QIS otherwise would not be possible). If the portfolio does not satisfy the retail criteria, it should be included in the corporate portfolio. All eligible collateral, if any, should be taken into account for calculating (foundation) IRB LGDs. Although corporate exposures should be rated individually, we realise that such a requirement may not be realistic for QIS. Consequently, for this exercise some concept of an average PD may suffice. In addition, banks can use estimates.
Answer: A time weighted LGD is calculated by first calculating LGDs for individual years, then averaging these LGD estimates. It gives disproportionate weight to a default that occurs in a year when few other credits default. If LGD is correlated to the number of defaults, then a time weighted LGD is a biased estimate of the cycle average LGD. A simple illustration may help to clarify this issue. Assume we have the following loss history:
LGD is obtained by dividing total losses by the total amount of assets in default (or a process that results in that outcome), not by adding 10, 90 and 10 and dividing by 3 (or a similar procedure), i.e. we would obtain a number closer to 90 then to 10 (in this case 88.4%). This is what we call a default weighted LGD.
Ceteris paribus the same logic should be applied when calculating EAD.
Answer: Only advanced IRB banks are allowed to do so, although they should ensure that their estimates do not take into account any effect of double default (see question E.5).
Answer: Unfortunately, it is not possible for the Committee to say what PDs are associated with external ratings. Banks which seek to align or map their internal grades to the rating scale of an external credit assessment institution will need to be able to demonstrate the reasonableness of their mapping to the external benchmark. It will not be sufficient for a bank to simply assert that its internal grades align with an external agency's grades. Some guidance on how this might be achieved is provided in paragraphs 409 and 410 of the Technical Guidance document.
For QIS-purposes, we advise such banks to use appropriate long-range historical default rates of the relevant rating agency and apply adjustments where necessary, i.e. to make your own 'best-efforts' PD estimates using whatever information is available to you. Even limited internal data series may be helpful in deriving PD estimates that are more meaningful than estimates that would be based on information of the rating agencies alone (the process used should be included in the answers to the data quality questionnaire submitted by your supervisor).
Answer: Loan covenants do not affect maturity. Maturity refers to contractual payments. As long as a covenant has not been breached contractual maturity is not affected. Similarly status depends on a facility being unconditionally cancellable. If there is a covenant, cancellation of the facility is, by definition, conditional and consequently the facility cannot be considered uncommitted. In this case (provided that banks satisfy the criteria mention in paragraph 274 of the Technical Guidance), however, covenants may influence the amount to which the credit conversion factor is applied. Advanced IRB banks should incorporate the effect of covenants in their estimates of EAD.
10. There seems to be an inconsistency between the foundation IRB templates and paragraph 251 of the Technical Guidance. If I have a loan of € 100 with financial collateral worth € 88, paragraph 251 instructs me to apply the haircuts mentioned in paragraph 110 in the Technical Guidance and calculate LGD*. Assuming a collateral haircut of 10% applies, I calculate LGD* as follows: LGD* = 45% * (100-(100%-10%)*88)/100 = 45% * 20/100 = 9%. I do not see how I should input this LGD-value in your templates.
Answer: You are correct, the templates take an approach that is slightly different from the approach described in paragraph 251 but that arrives at the same answer. In order to input your example into the templates, you must do the following. You have a loan of € 100 with financial collateral of € 88. After application of haircuts (10%), the collateral is worth € 80. Following the approach taken in the templates, you enter an amount of € 80 under the heading financial collateral (LGD = 0%) and the remaining € 20 under the heading senior unsecured (LGD = 45%). As you will see this generates an average LGD for your exposure equal to (80*0%+20*45%)/100 = 9%. The presentation used in the templates gives us more data on the kind of collateral you are using.
Answer: The PD associated with a rating grade is meant to be a one-year PD. When assigning exposures to a rating grade, however, banks are expected to take into account the borrower's ability and willingness to contractually perform despite adverse economic conditions or the occurrence of unexpected events.
12. How should advanced IRB banks input their EAD estimates in the templates? Is there a difference between (undrawn) committed lines and other off-balance sheet items reported in panel a) of the AIRB templates?
Answer: The concept of EAD applies to both undrawn committed and other off-balance sheet exposures in AIRB. However, within the spreadsheets banks are only required to split committed undrawn exposures by EAD bands. In order to reflect their own estimates of EAD for undrawn commitments-including commitments that would receive a 0% Credit Conversion Factor under Foundation IRB-banks should use the PD/EAD matrix in panel b) of the AIRB templates.
PD/EAD matrices are not included for other off-balance sheet items in panel a). This is because banks must apply (their own) credit conversion factors to these exposures outside the spreadsheets, as is the case in the standardised and FIRB approaches for other off-balance sheet items. The resulting exposure amounts (which reflect internal estimates of EAD) should be entered in the appropriate yellow cell in panel a) within the IRB sheets. Further information on the EADs used for these off-balance sheet items should be included in the notes section.
Note that repos and OTC derivatives reported in panels c) and d) continue to be treated according to the existing rules (replacement cost plus potential future exposure) as set out in paragraph 278 of the Technical Guidance.
13. Under foundation IRB, what is the LGD that I should use for a subordinated loan against which an amount of eligible collateral in excess of the minimum collateralisation requirements mentioned in paragraph 256 of the Technical Guidance has been pledged?
Answer: Whether collateralised or not, the full amount of a subordinated loan would receive a 75% regulatory LGD under foundation IRB (see footnote 61 of the Technical Guidance).
14. Under IRB, if I have given a commitment to provide a counterparty with a short-term self-liquidating trade letter of credit (that would satisfy the criteria mentioned in paragraph 276 of the Technical Guidance), may I apply the relevant credit conversion factors successively? In other words, may I first apply the 50% conversion factor for the facility and next the 75% conversion factor in order to reflect that I only provided a commitment and not the facility itself?
Answer: Yes, the overall credit conversion factor for such a commitment would be 50%*75% = 37.5% under the IRB approach. Similarly, under the standardised approach the overall credit conversion factor would be 20%x20% (for an undrawn trade finance facility of less than 12 months) or 20%x50% (for an undrawn trade finance facility of 1 year or more).
15. The AIRB spreadsheets provide an EAD input grid only for (undrawn) commitments. What about other off-balance sheet items?
Answer: The concept of EAD applies to both committed and other off-balance sheet exposures in AIRB. However, within the spreadsheets banks are only required to split undrawn committed exposures by EAD bands. In order to reflect their own estimates of EAD for commitments-including commitments that would receive a 0% Credit Conversion Factor under Foundation IRB (refer FAQ A.11)-banks should use the PD/EAD grid in panel b) of the AIRB templates.
PD/EAD grids are not included for other off-balance sheet items. This is because banks must apply credit conversion factors to these exposures outside the spreadsheets (as is the case in the standardised and FIRB approaches for other off-balance sheet items). Except for OTC derivatives, banks should use their own credit conversion (i.e. EAD) factors. The resulting exposure amounts (reflecting internal estimates of EAD) should be entered in the appropriate yellow cell in panel a) within the IRB sheets. Further information on the EADs used for off-balance sheet items should be included in the notes section.
In the case of OTC derivatives, banks are not permitted to use their own internal assessments of credit equivalent amounts. Instead, the rules for the standardised approach continue to apply (refer paragraph 298 of the Technical Guidance).
Return to QIS3 FAQ Table of Contents