Response to consultation paper on draft regulatory technical standards on the permanent and temporary uses of the IRB approach
Go back
The text in article 150(1)(a) and (b) of CRR refers to exposures where:
(i) the number of material counterparties is limited; and
(ii) it would be unduly burdensome for the institution to implement a rating system for these counterparties.
The former is addressed in article 2(1) point (c) of the RTS. The latter is dealt with in the qualitative criteria section.
An 8% threshold could impose undue pressure on both competent authorities and banks to implement models that may not fully comply with the applicable minimum standards.
In conclusion, we believe that the introduction of a quantitative threshold in the RTS (as the proposed 8%):
goes beyond the mandate of CRR;
distorts the spirit of the regulation; and
brings in unnecessary complications.
If co-legislators had meant to include a quantitative threshold on the exposure they would have done so and requested the EBA to calibrate it. But the fact is that level-1 legislation only refers to the number of counterparties and the unduly burdensome implementation.
For the sake of consistency, the definition of counterparty, which is missing in this RTS, should follow that of the large exposure regime (connected clients) as it is a concept already implemented within banks.
The limit of number of counterparties should be raised to 100 which is still a small number for statistical purposes. We really think that 20 as a threshold is too small.
As for the definition of “unduly burdensome” the EBF suggests removing article 2(2) and instead indicate that reliance is placed upon an institution to prove that it cannot comply with Title II, Chapter 3 (Credit Risk), Section 6 (IRB approach) Articles 169 – 191.
Both technical options seem complicated to implement in practice and both leave significant room for interpretation for competent authorities.
The EBF is of the view that comparisons between results of the standardised approach and the IRB, like the one put forward in proposal number 1, should be avoided. Proposal number 1 presumes that the likelihood of underestimation of risk is necessarily a function of the difference between an approach that pursues to measure risk with a high degree of precision (i.e. IRB) and a method that is much blunter (i.e. SA). This is fundamentally flawed. As a matter of principle, the SA is no sound reference to ascertain whether a risk is correctly estimated.
As the proposed quantitative threshold does not tackle the root problems associated with the development of IRB models, we would firstly recommend to look at each portfolio of risk on a standalone basis and to assess whether it can be IRB compliant in accordance with CRR section 6 (articles 169 to 191) and pass the implementation tests (articles 143 to 145).
If a threshold is finally used, the level of 8% is considered extremely stringent. It should be set at 15% or it would otherwise override the qualitative criteria.
The EBA should accept the permanent use if any of the two proposed metrics, i.e. either total exposure or risk weighted exposure, is below the threshold. Portfolios of non-significant business units are presumed to be discontinued or at least not to be renewed. Therefore, the proof of one of the metrics being below the threshold should be sufficient evidence of immateriality. As regards the risk-weighted exposure value, it is not a true risk sensitive measure but only the result of the standardised approach.
Article 2(1)(c) sets a condition that the number of material counterparties does not exceed 20. We believe that the threshold should be 100 instead. This number is still statistically insignificant.
Regarding article 3, we suggest a limit of 15% because it would strike a balance between depth of rollout and cost/benefits with sufficient consideration given to the principle of proportionality.
Finally, we would like to draw the attention of the EBA to paragraph 1 of article 1 of the draft RTS where the term “exposure value” is referred to. For the sake of clarity, there should be references to the precise provisions of the capital requirements regulation (CRR). We understand that the intention is to use the exposure value as defined in articles 111 and 166 to 168 of CRR for the standardised approach and the IRB approach, respectively.
We support the concept that rollout is not used as a means of distorting capital requirements. Nonetheless, we feel that what is proposed in article 4 is unnecessarily over engineered. The evaluation of capital impacts based on limited data could be prone to error and may provide unnecessary delay in agreement of a rollout plan.
Q2: Do you agree with the proposed draft RTS regarding permanent partial use of the Standardised Approach (SA) for the exposures specified in Article 150(1)(a) and (b) of the CRR?
We believe that point (a) and (b) of Article 2 (1) of the draft RTS, which sets an immateriality limit in the RTS in relation to the intuitions’ total exposure value of the set of relevant exposures and the total risk-weighted exposure amount of the set of relevant exposures, go beyond the scope of the provision laid down in point (a) and (b) of Article 150 of the CRR.. The relevant criteria in point (a) and (b) of Article 150 of the CRR are the number of material counterparties and the burden for institutions to implement rating systems for these counterparties. Accordingly, we believe that point (a) and (b) of Article 2 (1) should be deleted from the RTS.The text in article 150(1)(a) and (b) of CRR refers to exposures where:
(i) the number of material counterparties is limited; and
(ii) it would be unduly burdensome for the institution to implement a rating system for these counterparties.
The former is addressed in article 2(1) point (c) of the RTS. The latter is dealt with in the qualitative criteria section.
An 8% threshold could impose undue pressure on both competent authorities and banks to implement models that may not fully comply with the applicable minimum standards.
In conclusion, we believe that the introduction of a quantitative threshold in the RTS (as the proposed 8%):
goes beyond the mandate of CRR;
distorts the spirit of the regulation; and
brings in unnecessary complications.
If co-legislators had meant to include a quantitative threshold on the exposure they would have done so and requested the EBA to calibrate it. But the fact is that level-1 legislation only refers to the number of counterparties and the unduly burdensome implementation.
For the sake of consistency, the definition of counterparty, which is missing in this RTS, should follow that of the large exposure regime (connected clients) as it is a concept already implemented within banks.
The limit of number of counterparties should be raised to 100 which is still a small number for statistical purposes. We really think that 20 as a threshold is too small.
As for the definition of “unduly burdensome” the EBF suggests removing article 2(2) and instead indicate that reliance is placed upon an institution to prove that it cannot comply with Title II, Chapter 3 (Credit Risk), Section 6 (IRB approach) Articles 169 – 191.
Q3: Do you agree with the proposed draft RTS regarding permanent partial use of the SA for the exposures specified in Article 150(1)(c) of the CRR? Which of the two alternative proposals presented in the impact assessment section under ‘Technical options considered’ do you prefer?
The fixed upper limit of 8% of total exposure value and total risk weighted exposure amount associated with exposures to insignificant business units and/or products seems restrictive and unnecessarily rigid, cf point (a) and (b) of Article 3 of the draft RTS. A higher threshold, for example of 15%, would better address the purpose of identifying non-significant business units.Both technical options seem complicated to implement in practice and both leave significant room for interpretation for competent authorities.
The EBF is of the view that comparisons between results of the standardised approach and the IRB, like the one put forward in proposal number 1, should be avoided. Proposal number 1 presumes that the likelihood of underestimation of risk is necessarily a function of the difference between an approach that pursues to measure risk with a high degree of precision (i.e. IRB) and a method that is much blunter (i.e. SA). This is fundamentally flawed. As a matter of principle, the SA is no sound reference to ascertain whether a risk is correctly estimated.
Q4: Do you agree with the quantitative thresholds proposed in Articles 2(1), 3 and 4(2) of these draft RTS? If not, what thresholds do you consider more suitable?
As regards article 2(1), points (a) and (b), the EBF believes that the 8% threshold should be removed for the reasons mentioned in our response to question 2 above.As the proposed quantitative threshold does not tackle the root problems associated with the development of IRB models, we would firstly recommend to look at each portfolio of risk on a standalone basis and to assess whether it can be IRB compliant in accordance with CRR section 6 (articles 169 to 191) and pass the implementation tests (articles 143 to 145).
If a threshold is finally used, the level of 8% is considered extremely stringent. It should be set at 15% or it would otherwise override the qualitative criteria.
The EBA should accept the permanent use if any of the two proposed metrics, i.e. either total exposure or risk weighted exposure, is below the threshold. Portfolios of non-significant business units are presumed to be discontinued or at least not to be renewed. Therefore, the proof of one of the metrics being below the threshold should be sufficient evidence of immateriality. As regards the risk-weighted exposure value, it is not a true risk sensitive measure but only the result of the standardised approach.
Article 2(1)(c) sets a condition that the number of material counterparties does not exceed 20. We believe that the threshold should be 100 instead. This number is still statistically insignificant.
Regarding article 3, we suggest a limit of 15% because it would strike a balance between depth of rollout and cost/benefits with sufficient consideration given to the principle of proportionality.
Finally, we would like to draw the attention of the EBA to paragraph 1 of article 1 of the draft RTS where the term “exposure value” is referred to. For the sake of clarity, there should be references to the precise provisions of the capital requirements regulation (CRR). We understand that the intention is to use the exposure value as defined in articles 111 and 166 to 168 of CRR for the standardised approach and the IRB approach, respectively.
Q5: Do you think that separate quantitative thresholds should apply for application of these draft RTS on an individual and on a consolidated basis? Which of the two alternative proposals presented in the impact assessment section under ‘Technical options considered’ do you prefer?
The EBF supports flexibility in the application of the thresholds at solo level.Q6. Do you agree with our analysis of the impact of the proposals in this Consultation Paper? If not, can you provide any evidence or data that would explain why you disagree or which might assist our analysis of the possible impact of the proposals?
The analysis should take into account the fact that each bank has its own asset mix. The cost-benefit of rollout varies, with diminishing returns, depending on the structure and size of the organisation. Proportionality is implied in CRR article 150(1)(c) and should be carried through this RTS. Compliance costs increase as the size of the portfolio reduces. A sense of proportionality and flexibility is required for individual banks.We support the concept that rollout is not used as a means of distorting capital requirements. Nonetheless, we feel that what is proposed in article 4 is unnecessarily over engineered. The evaluation of capital impacts based on limited data could be prone to error and may provide unnecessary delay in agreement of a rollout plan.