Response to consultation on Draft Guidelines on the methodology to estimate and apply credit conversion factors under the Capital Requirements Regulation
1. Question 1: How material are the cases for your institution where you would have to assign an SA-CCF to exposures arising from undrawn revolving commitments and thus restrict the use of own estimates of LGDs within the scope of application for IRB-CCF in the CRR3? For which cases would you not have enough data to estimate CCFs but have enough data to es-timate own estimates of LGDs?
There might be material exposures in the application portfolio corresponding to not material exposures in the development sample, for example in case of newly launched type of exposures within asset classes already validated.
We don’t have evidence of cases where the availability of data allows for the internal estimation of LGD but not of CCF.
2. Question 2: Do you have any comments related to guidance on the identification of a relat-ed set of contracts which are connected such that they constitute a facility?
Some institutions face difficulties in linking the credit limits of individual contracts within a facility; therefore, the possibility of identifying a related set of contracts which are connected such that they constitute a facility could be a facilitation in this regard.
On the other hand, it would be required to insistutions to precisely define the linking structure between contracts and facilities, which could be not straightforward.
In cases where multiple obligors are present, the need for defining precedence rules for assigning unique obligor-level attributes (e.g., segment) would arise.
3. Question 3: Do these GL cover all relevant aspects related to the definition of revolving commitments that you consider relevant for the scope of the IRB-CCF? Have you identified any product that should be in the scope of the IRB-CCF that is currently excluded in the GL? In terms of off-balance sheet exposures, how material are the exposures that fall within the defined scope of the IRB-CCF for your institution?
The Consultation Paper does not clarify whether certain products fall into the scope of the revolving committments as defined under article 166 CRR3. For instance, the granted amount of a self-liquidating loan, is subject to a credit assessment in the origination phase. Then, the borrower’s outstanding balance is permitted to fluctuate based on its decisions to borrow and repay, up to the granted amount defined during the origination phase, and subject to a light credit evaluation of the single request.
Similarly, it is unclear why installment products that include a margin (e.g., mortgages disbursed in tranches based on project progress) are excluded from the scope of CRR and, consequently, from these guidelines.
Finally, based on our experience, off-balance sheet exposures are material for Non Retail segments. Moreover, some Institutions specialize in these types of products, with material exposures.
4. Question 4: Are there products that have an advised limit of zero but a nonzero unadvised limit that should be included in the scope of the IRB-CCF GL? How material are these cases for your institution?
Materiality depends on the banks; we have seen institutions for which this case is material (e.g., because they granted this type of product to all employees).
5. Question 5: Do you think that dynamic limits (e.g. limits the extent of which is dependent on the market value of financial collateral pledged by the obligor in relation to the revolving loan) warrant a specific treatment in the IRB-CCF GL? How material are these cases for your institution?
Yes, the specific features of this type of products (e.g., Lombard loans) mean that they require dedicated guidelines.
The materiality of such exposures depends on the banks: since there are institutions specialized in this kind of products, their level of exposure can be significant.
6. Question 6: Have you identified any unwarranted consequences of including fully drawn revolving commitments in the scope of the IRB-CCF. How material are these cases for your institution?
No, historically these are cases that have always been present in the portfolios of institutions across several European countries (e.g. in Italy), often with material exposures, and they have always been managed within the CCF estimation. In the current consultation paper, it is requested that these cases be managed using the committed amount, but we do not identify any unwarranted consequence in this regard.
8. Question 8: Are there cases for your institution where the calibration samples should be shorter than the sample used to calculate the long run average (LRA) CCF
In our experience, there are the following types of situations that could require a calibration sample shorter than the sample used to calculate the LRA CCF: (a) the full calibration sample is not representative of the current application portfolio (e.g. because of changes in the management strategies, changes in the definition of default, etc.); (b) the historical data lacks the information/drivers needed to differentiate the model calibration clusters; (c) other situations as defined in par. 28 of the EBA GL 2017/16.
9. Question 9: Do you have any concerns with the requirements introduced to analyse and mitigate a lack of representativeness for CCF? Do the requirements on the different data samples when observing a lack of representativeness impede your ability to model CCF portfolios?
We do not see any obstacles to the ability to model CCF portfolios, also because, for the development sample, the requirements are even more relaxed compared to PD and LGD.
10. Question 10: Do you have any concerns with linking the fixed CCF to the lack of historical data available to the institution in relation to the coverage by the RDS of material subseg-ments of the application portfolio? How is your institution currently treating these cases?
Yes, as mentioned in Question 1, we do identify significant concerns: for example, in case of the introduction of new products that might be material in the application portfolio, but not material in the development sample.
This effect (i.e. the imbalance between the size of the application portfolio vs. the development sample) could be amplified in the case of low-risk new products, considering that low-risk products require more time to enter the estimation sample with a sufficient number of observations.
In such cases, it would be more appropriate to allow for the estimation of a CCF consistent with the specific characteristics of the product, and not necessarily equal to 100%.
Our concern, therefore, is that this treatment could be excessively penalizing, even for potentially relevant portfolio segments (also considering that it is not possible to maintain the FIRB approach for individual products).
Finally, it is worth mentioning that there could be an issue regarding lack of historical data also for fully drawn facilities requiring a fixed coefficient (i.e. it might be necessary to define not only a fixed CCF, but also a fixed limit factor).
11. Question 11: Are there any concerns with requiring consistency in the analysis of changes in the product mix with the institution’s definition of facility? Are institutions able to identify and link contracts (partially) replacing other contracts where the closing or repayment of one contract is related to the origination of a new contract? Are institutions able to link new contracts that are originated after the reference date to related contracts existing at refer-ence date? In particular, is it possible in the case contracts that are revolving commitments are replaced by contracts that are non-revolving commitments (e.g. by a term loan)?
Our concern is that, technically, this is something difficult to achieve. Often, the available data do not allow this reconciliation analysis to be performed in a simple and clean way, and approximations and assumptions are required.
12. Question 12: Do institutions consider it proportionate to the risks of underestimation of CCF to perform the identification analysis and allocation procedure? If it is deemed not propor-tional, what would be an alternative approach that is still compliant with Article 182(1b) CRR?
No, we believe it would be disproportionate if an exact and detailed reconciliation is expected (i.e., at single-contract level). Such analysis would require a significant effort for institutions in particular for the historical backcast of the allocation procedure and will inevitably need for appropriate adjustments which will in turn lead to a MoC definition.
Instead, in our opinion, a portfolio-level impact analysis could be feasible with a limited risk of underestimation of CCF.
13. Question 13: Do you have any concerns on the proposed approach for the treatment of so-called ‘fast defaults’? In case you already apply a 12-month fixed-horizon approach, do you apply a different treatment for ‘fast defaults’ in practice, (and if so, which one)? Is the ‘fast default’ phenomenon material according to your experience? If yes, for which exposures, exposure classes or types of facilities?
We have no concerns regarding the proposed approach for fast defaults. Based on our experience, this type of case is generally not material.
However, a potential concern arises for facilities that remain inactive for a long time and then default within a few months after becoming active borrowers (e.g. credit cards migrating from transactor status to active borrowing and defaulting within less than one year).
In our opinion, from a credit risk prediction perspective, using a 12-month period prior to the default date where there is zero utilization and zero drawing activity may introduce bias into the CCF risk drivers. In such cases, it could be appropriate to introduce a flag to mark these dormant facilities (with a proper “dormant” criterion to be defined) and possibly shorten their estimation horizon, similarly to what is proposed for fast defaults.
14. Question 14: Do you have any concerns on the multiple default treatment? To what extent are your current models impacted by the application of a multiple default treatment?
We have no concerns regarding the proposed approach, as it is also consistent with what is applied for LGD.
15. Question 15: Do you agree with the three principles for the calculation for realised CCF in the context of consumer product mix, and their implications for the cases mentioned as ex-amples? In case of disagreement, what is the materiality of the cases with unwarranted re-sults, in particular in relation with the definition of facility applied in your institution? In case of material unwarranted results, can you describe your alternative practice to this CP?
Since different contracts could follow different internal processes (e.g., different actors managing different type of contracts), the proposed rule might not be consistent with the institutions' monitoring processes in place.
As an alternative, work could be done at the macro-facility level to avoid technical issues related to splitting movements, while still preserving the accuracy of results at portfolio level.
16. Question 16: Are there any concerns related to the allocation mechanism described in these GL?
As for questions 12 and 15, we believe it would be disproportionate if an exact and detailed reconciliation is expected (i.e., at single-contract level) and it might not be consistent with the institution’s monitoring processes.
Conversely, a portfolio-level impact analysis could be feasible with a limited risk of underestimation of CCF.
17. Question 17: Where credit lines are kept open even if the facility is in default, the alterna-tive option described in this consultation box could lead to high realised CCF values. Is this a relevant element for your institution and if yes, why and how material are these cases with-in the scope of IRB-CCF models?
Applying the alternative option described in the consultation box, would make the estimates less meaningful (artificially high realised CCF values) and would reduce their usefulness for business processes.
18. Question 18: In case of multiple defaults, the CCF might also be driven by drawings while the obligor was in its default probation period or in the dependence period between the merged defaults. Do you expect this to be material for your CCF models?
We believe this aspect could be material, especially for less recent dates and in cases where the default event was reconstructed historically, as in such circumstances there tends to be greater volatility in entries/exits from past-due status.
In addition, it is worth noting that such behaviour (drawings while the obligor was in its probation period or in the dependence period between merged defaults) is more comparable to standard pre-default drawings than to actual drawings after default, and therefore could potentially bias the estimates.
19. Question 19: Do you see any unwarranted consequences of the proposed approach for in-corporating additional drawings after default? In particular, in order to maintain consistency between the realised CCF calculation and the calculation of the denominator of the realised LGD as described in paragraph 140 of the GL PD and LGD, would this require a redevelop-ment of your LGD models?
We believe it is more appropriate to discount only the additional drawn amount after default and not the entire exposure (“maximum of the drawn amounts after the moment of default”). Moreover, it should be specified that the interest component should be excluded from this calculation (only the principal amount to be considered) to ensure consistency with the LGD framework.
For banks applying a different treatment from the one proposed, this would require re-developing the LGD based on the updated denominator. In some cases, we have also seen banks considering the calculation of a calibration factor only — one for EAD and one for LGD.
20. Question 20: Do you think that the relative threshold is an appropriate approach to restrict the use of the alternative CCF approach for those facilities in the region of instability? Do you think it is appropriate to define a single relative threshold per rating system or are there circumstances where multiple relative thresholds would be warranted? Do you see a need to use an absolute threshold in addition to the relative thresholds?
We consider the relative threshold to be appropriate but not sufficient on its own, as it should be complemented by an absolute threshold (see point 3 below).
We have no concerns regarding the definition of a single threshold for each rating system.
Yes, as for the DoD, an absolute threshold would be needed to handle cases of facilities with small limits. A fixed regulatory value for such thresholds could also be used (differentiated by segment as in the DoD): imposing thresholds might indeed speed up the estimation and validation process, but in some banks inconsistencies could arise between the proposed regulatory values and the observed data.
21. Question 21: Do you consider the guidance sufficiently clear in relation to the requirement for institutions to set up a policy to define a threshold value?
The need for institutions to define a policy within which values are determined based on a reasonable trade-off principle that considers the variance of the estimator (and not other criteria commonly used in practice, such as outliers) is sufficiently clear.
Our understanding is that the practical implementation of this policy remains the responsibility of each individual institution (for example, when applying a criterion based on the variance of the estimator, it is up to the institution to define which variance value(s) should be considered).
22. Question 22: Do you consider it appropriate to set a prescribed level or range for the de-fined threshold, and if so, what would be an appropriate level for the threshold? In case an absolute threshold is warranted, what would be an appropriate prescribed level for an abso-lute threshold?
We are aware that prescribed regulatory values/ranges might indeed speed up the estimation and validation process. Nevertheless, we believe it could be particularly useful to analyse the behaviour of the variance across the various specific portfolios. In light of these considerations, we do not consider appropriate to set a prescribed level or range for the defined threshold.
See the previous point. However, if it is decided to define regulatory values for the absolute threshold, such values could be taken from those used in the DoD for defining past-due status.
23. Question 23: Do you think that, for the facilities in the region of instability, and/or for fully drawn revolving commitments, a single approach should be prescribed (e.g. one of the ap-proaches above defined in the Basel III framework), or that more flexibility is necessary for institutions to use different approaches they deem most appropriate for these facilities?
Yes, we believe it is preferable to prescribe a single approach (generally, the first approach proposed in the CP works very well in practice - it is simple and covers several underlying cases).
However, for fully drawn commitments, this solution does not cover situations where a drawn amount exists in the absence of a limit. In such cases, the use of the drawn amount instead of the limit could be considered, also to ensure comparability of values.
24. Question 24: If such flexibility is indeed warranted, what is the technical argumentation why prescribing a single alternative approach for these facilities is not suitable? Which cases or which types of revolving commitments could not be modelled under the approaches pre-scribed? Are there types of revolving commitments that could not be modelled by any of the approaches described in the Basel III framework?
As stated in the response to Question 23, we believe a single approach would be preferable; however, the limit-based approach does not cover situations where a drawn amount exists in the absence of a limit. In light of this, the use of the drawn amount instead of the limit could be considered.
We do not identify any relevant types of revolving commitments that could not be modelled using one of the approaches described in the Basel III framework.
25. Question 25: Which of the three approaches described in the Basel III framework is pre-ferred in case a single approach would be prescribed?
As stated in the response to Question 23, we believe that the first approach proposed in the CP works very well in practice as it is simple and covers several underlying cases.
26. Question 26: For the purpose of the long run average calculation, are there any situations where such intermediate exposure weighted averaging at obligor level would lead to a dif-ferent outcome (that is unbiased) with regard to the CCF estimation? How material is this for your portfolio?
We do not see any situations in which the two-step approach would be preferable. Moreover, in the first of the two proposed steps (the averaging at obligor level), we find no consistency with the processes that are in place within a bank, which are generally at obligor or facility level, but not at CCF's single-pool/single-grade level.
On the other hand, the two-step approach is the one required for LGD; therefore, if a one-step approach were prescribed, this would create an inconsistency between CCF and LGD.
27. Question 27: Do you have any comments on the condition set to use the simple approach to estimate additional drawings after default. Do you consider that the simple approach is also relevant for retail portfolios?
We believe that the proposed option (simple approach) is reasonable and useful, also to simplify the estimation process.
Yes, we do. Indeed, the reasons why the approach should differ between Retail and Non-Retail segments are not clear (in fact, also in Retail, drawings after default could affect estimation results), as this would create an inconsistency between the two segments.
28. Question 28: It was considered that requiring institutions to exclude unresolved cases from the long run average CCF, if their realised CCF is lower than the LRA of the corresponding fa-cility grade, could be seen as too conservative. Do you have any comments on this treat-ment introduced in the simple approach? Do you have specific examples when this treat-ment would not be appropriate?
Yes, it would be overly conservative, since the lower CCFs of incomplete recovery processes could be observed not only due to a low number of days in default, but also due to changes in institutions’ policies. If the latter is true, the institutions could be penalized twice for the same reason — both in the LRA CCF and in the MOC for weak representativeness.
29. Question 29: Do you have any comments on the modelling approach to estimate additional drawings after default for unresolved cases?
The first comment is that, in some institutions, we have noticed that the weight of unresolved cases is limited (it generally depends on the duration of the pre-litigation phase, i.e., if Past Due and, especially, UTP exposures last for a short time because the case is quickly either cured or turns into default, then the number of unresolved cases at each reference date will tend to be low). Therefore, the modeling effort could be excessive considering the materiality of the phenomenon.
The second comment is that it is not clear why the approach should be differentiated by segment (Retail vs Non-Retail).
30. Question 30: Do you have any concerns with the requirement to use as a maximum drawing period the maximum recovery period set for LGD?
Yes, we have some concerns, because these are not related phenomena: credit lines are generally revoked within a relatively short period, whereas recoveries can occur much later in time.
Therefore, it does not seem reasonable to propose an alignment between the two phenomena.
31. Question 31: For CCF estimation, do you use estimation methods that incorporate portfolio-level-calibration of the estimates? What are the main reasons to use a calibration at a level that is higher than the grade-level calibration?
Yes, we have seen institutions performing calibrations at a level that is higher than the grade-level calibration.
The main reasons are the following: (a) if the LRAs of two or more grades are statistically equal, it makes sense to simplify the calibration process, as long as it is always possible to verify ex post the calibration of each grade; (b) in cases where the LRA does not follow the expected monotonic trend at the single-grade level, but instead shows inversions.
32. Question 32: Do you have any comments on the guidance for the CCF estimation of default-ed exposures?
Based on the structure of section 3.8 of the CP, and in particular paragraphs 127, 128, and 129, it is not clear whether the need for institution to "apply the non-default grade level estimates to facilities based on their latest available grade assignment" etc. (par. 129) is required also in the modeling approach.
33. Question 33: Do you have any comments on the determination of the low share of ob-served additional drawings after default in the historical observation period relative to the observed undrawn amount at default date? Do you consider it appropriate to set a pre-scribed threshold to determine what constitutes this low share? If so, what would be an ap-propriate value for such a materiality threshold?
(1) We find it unclear why, in order to qualify for a simplified approach, institutions are required to perform complex analyses. It might be more proportionate - and still acceptable - to perform analyses at a portfolio or asset class level (e.g., share of drawable margins at default over total drawable margins, including performing exposures, or share of drawable margins over outstanding committed amounts for defaults).
Furthermore, it could be useful to reverse the logic of the requirement and, instead of setting a minimum low share, establish a high backstop threshold beyond which access to the simplified approach would no longer be allowed. This way, the goal of limiting access to the simplified approach would still be achieved, while the determination of regulatory thresholds would likely be easier.
(2) Yes, a regulatory threshold would be useful in order to avoid complex (and not always consistent) bank-specific materiality assessments.
(3) We agree that defining a regulatory threshold is not easy; however, we believe it is easier to reason in terms of high thresholds, such as “share of drawable margins at default over total drawable margins (including performing exposures) = 10%” or “share of drawable margins over outstanding committed amounts for defaults = 20%”. Beyond these limits, access to the simplified approach would be precluded.
34. Question 34: Are there examples where the haircut approach should be considered the most appropriate approach for estimating the downturn CCF?
No, we have not seen such cases.
It is true that an haircut could be applied to the margin in cases where there is an inverse relationship between the CCF and the margin; however, we do not consider this to be the most appropriate solution.
35. Question 35: Do you think the add-on of 15 percentage points is adequately calibrated when the downturn impact cannot be observed nor estimated? Could you provide clear examples or reasons why this add-on should be higher or lower than 15 percentage points?
From a practical perspective, the CCF is countercyclical, so a value of 15% is already sufficiently conservative.
Applying a higher value would make the estimates less meaningful and would reduce their usefulness for business processes. In other words, if a CCF of X% is estimated and the application of an add-on makes it significantly different (for example, doubles it), the resulting post–add-on CCF becomes merely a regulatory value with no real economic or business meaning, and therefore useless for internal processes.
36. Question 36: Have you observed, or do you expect a (statistically significant) correlation be-tween economic indicators and realised CCFs? If so, do you expect higher or lower levels of CCFs observed in the downturn periods compared to the rest of the cycle? Do you have pol-icies in place that restrict or, on the other hand, relax the drawing possibilities in the down-turn periods?
Parts (1) and (2): See response 35.
Part (3): We have never seen policies explicitly prescribing a restriction of drawing possibilities during downturn periods; however, this is something that is normally observed for revocable or amendable margins during downturn periods.
37. Question 37: The possibility to have no downturn effect on CCF estimates is restricted to the case where observations are available during a downturn period. Which alternative meth-odologies could be used to prove the non-existence of a downturn effect on CCF estimates, in the case where no observation is available during a downturn period?
In the case where no observation is available during a downturn period, we consider the application of a regulatory floor to be understandable.
Indeed, while it is always possible to extrapolate a relationship between CCF and macroeconomic factors over the observed time horizon, if this relationship is defined based only on non-recessionary phases, we do not see how the absence of a downturn effect could be justified.