Response to consultation Paper on draft ITS amending the benchmarking Regulation
Go back
We agree with the necessity to complement the metrics in relation with IFRS 9 models with qualitative information as it allows to contextualize the results and differences between banks and as it will help avoid any misleading conclusions. We do not believe that a predefined qualitative questionnaire would be the most effective and meaningful to understand the quantitative data submitted.
On the contrary, we advocate that the qualitative information should be provided through bilateral exchanges between regulators / supervisors and banks, as it is already proceeded in the case of similar benchmark exercises. Such interactive discussions are crucial to ensure that regulators and supervisors have a fair and accurate understanding of the data / information reported by banks and that detailed qualitative explanation closely linked to the quantitative data submitted is provided.
Indeed, drawing up a qualitative questionnaire that addresses all aspects of IFRS 9 modelling is a complex exercise. As the qualitative questions asked may be not specific enough, they may be subject to interpretation by banks and they would hardly match to the operational reality of IFRS 9 models. Therefore, the expected responses to any pre-defined qualitative questionnaire should be taken with the greatest caution to avoid any misunderstanding between supervisors and banks which would be detrimental to the benchmark exercise. For that reason, we believe that the explanation of the quantitative results of the benchmark exercise for IFRS 9 modelling will require close discussions between supervisors and banks to provide accurate explanation on IFRS 9 modelling.
In our view, it is essential that regulators and the industry engage in proactive discussions to ensure a high level of confidence in the output of such benchmarking exercises.
The annual notification of the reference date of the exercise avoids any misunderstanding
Moreover, the EBA acknowledges the difficulties to fully decompose the changes of variability between the IRB and the IFRS 9 models but however requests the “PD – IRB without conservative adjustments”. We would like to remind that the credit risk template (c105.01) already contains information such as “EAD weighted average default rate for calibration”, “Case weighted average default rate for calibration” and “Long-run PD” which provide information to the EBA in relation to the “intermediate steps“ for IRBA models. Moreover, the requirements which concern the Margins of Conservatism (MoC) and the downturn estimation have been harmonised in the EBA IRB Repair Program. Such harmonisation of the estimation features should be implemented by banks according to the deadlines communicated in July 2019 : the main deadline is by end 2021 with an exception for LGD models for financial institutions and very large corporates which changes can wait until 2023. Therefore, in the objective of comparability, we consider that it is premature even redundant with other templates to request risk parameters without conservative adjustment – including PDs - for the benchmarking exercise of April 2021. Moreover, in future exercises, we would like to question the relevancy of submitting such metrics for LDP portfolios which are partly concerned by the shift from IRBA to IRBF, implying that some of the internal models could be maintained not its “pure regulatory” form or even decommissioned by banks.
In this context, comparisons between banks applying different types of economic scenarios will reveal irrelevant discrepancies, because comparing individualised scenarios for material exposures with common scenarios for non-material exposures.
A notion of materiality for the application of economic scenarios should be introduced to reflect the rationale used by banks on this issue.
In addition, to better understand banks’ practices, the template deserves interactive discussions between the EBA and banks.
Indeed, the EBA should be aware that some institutions may use the qualitative SICR indicators with different degrees of priority to determine the shift to exposures to the stage 2 (e.g. one indicator directly stage 2) conditioning the scheme of IT algorithms . Therefore, there might be institutions which may not fill in all of the modalities of the variable 600 Qualitative Stage 2 Trigger set. Such structural conditions should be taken into account in the EBA analysis.
However, if the criteria for the SICR assessment were to be maintained, the banks will implement these criteria for the purposes of the benchmark exercise, with increased burden data collection and without necessarily being linked to the operational implementation of the IFRS 9 modelling.
Q1. Do you agree with the necessity to complement the quantitative data collection with qualitative templates?
We are welcoming this question asked by the EBA, implying that the proposal of submitting a qualitative template is still under discussion and not acted as definitive.We agree with the necessity to complement the metrics in relation with IFRS 9 models with qualitative information as it allows to contextualize the results and differences between banks and as it will help avoid any misleading conclusions. We do not believe that a predefined qualitative questionnaire would be the most effective and meaningful to understand the quantitative data submitted.
On the contrary, we advocate that the qualitative information should be provided through bilateral exchanges between regulators / supervisors and banks, as it is already proceeded in the case of similar benchmark exercises. Such interactive discussions are crucial to ensure that regulators and supervisors have a fair and accurate understanding of the data / information reported by banks and that detailed qualitative explanation closely linked to the quantitative data submitted is provided.
Indeed, drawing up a qualitative questionnaire that addresses all aspects of IFRS 9 modelling is a complex exercise. As the qualitative questions asked may be not specific enough, they may be subject to interpretation by banks and they would hardly match to the operational reality of IFRS 9 models. Therefore, the expected responses to any pre-defined qualitative questionnaire should be taken with the greatest caution to avoid any misunderstanding between supervisors and banks which would be detrimental to the benchmark exercise. For that reason, we believe that the explanation of the quantitative results of the benchmark exercise for IFRS 9 modelling will require close discussions between supervisors and banks to provide accurate explanation on IFRS 9 modelling.
In our view, it is essential that regulators and the industry engage in proactive discussions to ensure a high level of confidence in the output of such benchmarking exercises.
Q2. In your view, which aspects, from a LDP perspective, are relevant to investigate from a qualitative perspective, where there might be different practices leading to different impact across institutions?
As explained in our answer to question 1, only bilateral discussions between banks and the EBA on the quantitative data submitted would provide relevant qualitative information and aspects on the different practices that are inherent to the banks’s LDP portfolios where practices may vary from other asset classes.Q3 Do you have any concerns on the three changes applied to the credit risk IRB templates? In particular, do you believe the extension of the data collection for hypothetical RWA will add a significant burden to the exercise?
Generally, we advocate for a stability of requirements in relation to the calculation of several metrics in the credit risk IRB templates. In this perspective, we suggest that the EBA sticks to the conclusions made in the “EBA Report – Results from the 2018 low and high default portfolios exercise” which advocate for stabilising reporting definitions which will ease the comparisons across time in a consistent way (“for future exercises, and with the benefit of a stable sample of institutions and clearer reporting definitions, more emphasis on comparisons across time will help to gain additional insight as to whether convergence in modelling practices is taking place as a result of the EBA review of internal models.”). We have noted that some indicators such as the ones related to the calculation of RWAs have been modified in several exercises, which seems to be in opposition to the EBA’s initial intention.Q4: Stakeholders are invited to express their view on the new reference date specification with respect the precedent method to specify the reference dates for the exercise.
With respect to the new reference dates specification, detailed in paragraph 12 of the consultation paper EBA/CP/2019/15, institutions support to keep the applicable method as it is. Banks are not supporting the introduction of a standardised method avoiding to reset every year the market risk references date.The annual notification of the reference date of the exercise avoids any misunderstanding
Q5: Stakeholders are invited to express their view on the implementation of the Benchmarks Regulation, in terms of which rate to apply in the instruments in the market risk exercise.
Institutions support to consider EU Benchmarks Rate Regulation within the Benchmark exercise, through the alternative drafting proposal of annex V, point (aa). Nevertheless, as regards to the underlined part of the paragraph, institutions call the EBA to confirm the specification of the alternative rate will consist on the notification of the list of the alternative rates.Q6. Do you see any issues or lack of clarity in the definition of the data points of template 111.01?
Concerning template 111.01, the EBA should clarify the reporting of only pre-CRM PDs in line with the asumptions of IRB credit risk templates.Moreover, the EBA acknowledges the difficulties to fully decompose the changes of variability between the IRB and the IFRS 9 models but however requests the “PD – IRB without conservative adjustments”. We would like to remind that the credit risk template (c105.01) already contains information such as “EAD weighted average default rate for calibration”, “Case weighted average default rate for calibration” and “Long-run PD” which provide information to the EBA in relation to the “intermediate steps“ for IRBA models. Moreover, the requirements which concern the Margins of Conservatism (MoC) and the downturn estimation have been harmonised in the EBA IRB Repair Program. Such harmonisation of the estimation features should be implemented by banks according to the deadlines communicated in July 2019 : the main deadline is by end 2021 with an exception for LGD models for financial institutions and very large corporates which changes can wait until 2023. Therefore, in the objective of comparability, we consider that it is premature even redundant with other templates to request risk parameters without conservative adjustment – including PDs - for the benchmarking exercise of April 2021. Moreover, in future exercises, we would like to question the relevancy of submitting such metrics for LDP portfolios which are partly concerned by the shift from IRBA to IRBF, implying that some of the internal models could be maintained not its “pure regulatory” form or even decommissioned by banks.
Q7. Do you see the need to adjust or add any variable for the intended outcome? Please specify.
No commentsQ8. Would you see any particular problem in filling some of the data requested? For which reasons? Please give your comments related to the PD 12month and the economic scenario in question 7.
No commentsQ9. Do you see any issues or lack of clarity in the definition of the data points of template 111.02? Please explain
For countries where banks' exposures are non-material or the number of counterparties is limited, in most cases, average economic scenarios are applied. Average economic scenarios are thus established on the basis of averages of aggregates by country groups or major geographical regions. Conversely, for countries where banks' exposures are material, economic scenarios are individualised by countries.In this context, comparisons between banks applying different types of economic scenarios will reveal irrelevant discrepancies, because comparing individualised scenarios for material exposures with common scenarios for non-material exposures.
A notion of materiality for the application of economic scenarios should be introduced to reflect the rationale used by banks on this issue.
In addition, to better understand banks’ practices, the template deserves interactive discussions between the EBA and banks.
Q10. Do the categorisations reported above reflect the approach applied by your bank, in incorporating forward looking information? If not, please explain what are the main differences.
The 3 approaches proposed by the EBA are the mainly explored approaches. If there are specificities (for instance slightly different version of one of the approaches), the EBA should concede that it is not possible to account for all specific situations in only 3 approaches and therefore allow for some flexibility in the metrics to be reported or notQ11. For banks applying Approach 3, what will be, in your view, an appropriate approach for reporting the data related to the PD in scenario 0 (i.e. the PD considered in the application of the impairment requirements under IFRS 9)? Do you think that, (if available) a probability weighted average PD represents an appropriate proxy? Do you think that the PD used for the SICR assessment represents an appropriate proxy? If not, what other approach do you suggest to report this data?
No commentsQ12. Do you believe that additional macro-economic variables should be tested in future exercises and if yes, which ones would be appropriate in the context of a benchmarking exercise?
We think there are various practices between banks in terms of macro-economic variables to be tested which will make it difficult to suggest alternative list. Therefore, we suggest to keep the list of macroeconomic variables to its simplest formQ13. Do you see any issues or lack of clarity in the definition of the data points of template 111.03?
The SICR assessment criteria proposed by the EBA for the benchmarking exercise do not always correspond to those used by banks in their management systems in application of IFRS 9. Thus, we propose that the interactive discussions between the EBA and the banks should allow for an understanding of the native data in the systems used to analyse the SICR.Indeed, the EBA should be aware that some institutions may use the qualitative SICR indicators with different degrees of priority to determine the shift to exposures to the stage 2 (e.g. one indicator directly stage 2) conditioning the scheme of IT algorithms . Therefore, there might be institutions which may not fill in all of the modalities of the variable 600 Qualitative Stage 2 Trigger set. Such structural conditions should be taken into account in the EBA analysis.
However, if the criteria for the SICR assessment were to be maintained, the banks will implement these criteria for the purposes of the benchmark exercise, with increased burden data collection and without necessarily being linked to the operational implementation of the IFRS 9 modelling.