Response to consultation on Guidelines on the treatment of CVA risk under SREP
Go back
However, to meet this objective, we would recommend EBA to remove the Threshold 1 because (i) the size of an institution derivatives business is not in general a relevant criterion to assess CVA Risk, especially for those small non-complex EU institutions calculating EAD according to the non-risk based standard method (CVA risk depends more broadly on the credit quality of counterparties and on the portfolio risk factors), and (ii) the threshold seems to be useless as those small non-complex EU institutions are required to calculate Threshold 2 in all cases, this second threshold being a fairly better indicator to determine relevance of CVA risk, notwithstanding the limitations exposed in the answer to Question 3.
If however the EBA decided to maintain Threshold 1, it would be more appropriate to consider that the CVA risk is relevant when both Threshold 1 and Threshold 2 are exceeded (instead of Threshold 1 or Threshold 2).
However, we consider that the proposed ratio suffers a major weakness as it is based on a measure disconnected from true economic risk. The relevance of CVA risk that could be assessed as part of a SREP process should rely on the accounting CVA and not the regulatory CVA.
Additionally, we are in the opinion that setting a unique x% value is inappropriate, moreover for both thresholds. A “one-fit-all” value is likely to jeopardize the relevance of supervisory reviews because it will fail to capture the specificities of the institution which should be part of the SREP process, as defined in Article 107(3) of EU Directive 2013/36, “in a manner that is appropriate to the size, the structure and the internal organisation of institutions and the nature, scope and complexity of their activities”. We believe that supervisors must keep the ability to appreciate specific cases and take customized views different from those enclosed in the guidelines if deemed necessary.
If EBA continues imposing this automatic measure despite the negative consequences, we argue that
- The sovereign counterparties continue to be exempted until at least the Commission has reviewed the prudential treatment to sovereign counterparties as a whole.
- The NFC- counterparties continue to be exempted until the definition and thresholds have been reviewed by ESMA in the EMIR context.
- In both cases, a precise assessment of the impact of the removal of the exemption on the derivative market is an important prerequisite, as shortage of supply of derivative may impact badly the EU economy.
Indeed, CVA losses can occur either because of an increase in the institution’s counterparties credit spreads or because of an increase in the institution’s exposure to its counterparties. But on a consolidation basis, in accordance with the IFRS principle, intra-group transactions are cancelled. Hence, there is no volatility to institutions’ own funds neither CVA loss to institutions themselves.
Moreover, according to the CRR, changes in the value of own liabilities (DVA) shall be filtered from own funds. And, actually, the credit spreads of consolidated entities will be based on the CDS of its parent undertaking. Consequently, even if intra-groups transactions were included for the CVA charge purpose, this charge will be filtered in accordance with the DVA treatment.
From our point of view, intra-group transactions shall only cover transactions between entities in the same consolidation perimeter (as define in art. 382(4)(b)) for the prudential purpose. Similarly it is quite acceptable to include intra-group transactions these transactions do not fulfil the criteria listed in the article 382 (4)(b) of the CRR. Besides, transactions with structurally separated subsidiaries shall not be considered as intra-group transaction.
Reversely transactions with insurance subsidiaries or with entities recorded in accordance with the equity method shall not be considered as intra-group transactions. As a consequence, the CVA charge shall be computed as usual for these transactions.
Moreover, we consider the proposed supervisory benchmark is arbitrary and flawed.
As already mentioned above, the Pillar 2 process allows institutions to use appropriate models to measure CVA risk and in particular within ICAAP. By stating that the competent authorities should use the value of the hypothetical own funds requirements for CVA risk as specified in these Guidelines, the EBA removes this opportunity of using appropriate models, which is contrary to this fundamentally sound principle..
Question 1: Do you agree with determining relevance of CVA risk by means of assessing the size of an institution’s derivative business using the exposure value for non-QCCP cleared derivatives transactions?
We understand the EBA’s objective to exclude from the step 3 small non-complex EU institutions which do not exhibit excessive CVA risk.However, to meet this objective, we would recommend EBA to remove the Threshold 1 because (i) the size of an institution derivatives business is not in general a relevant criterion to assess CVA Risk, especially for those small non-complex EU institutions calculating EAD according to the non-risk based standard method (CVA risk depends more broadly on the credit quality of counterparties and on the portfolio risk factors), and (ii) the threshold seems to be useless as those small non-complex EU institutions are required to calculate Threshold 2 in all cases, this second threshold being a fairly better indicator to determine relevance of CVA risk, notwithstanding the limitations exposed in the answer to Question 3.
Question 2: What are your views on how Threshold 1 should be calibrated?
Consistently with the answer to Question 1, we would rather recommend to remove the Threshold 1.If however the EBA decided to maintain Threshold 1, it would be more appropriate to consider that the CVA risk is relevant when both Threshold 1 and Threshold 2 are exceeded (instead of Threshold 1 or Threshold 2).
Question 3: Do you agree with determining relevance of CVA risk by means of assessing the share of own funds requirements for CVA risk to the total risk exposure amount?
Notwithstanding the arguments provided in the general comments, we are in the opinion that the share of own funds requirements for CVA risk to the total risk exposure amount is a better indicator to determine relevance of CVA risk compared to a volume based indicator.However, we consider that the proposed ratio suffers a major weakness as it is based on a measure disconnected from true economic risk. The relevance of CVA risk that could be assessed as part of a SREP process should rely on the accounting CVA and not the regulatory CVA.
Question 4: Do you agree with the approach provided for the determination of materiality of CVA risk?
BCBS has designed and calibrated at the global level the CVA risk framework for the consolidated level. Notwithstanding the arguments provided in the general comments, we are in the opinion that the determination of materiality of CVA risk should be assessed exclusively at the consolidated group level. This recommendation is consistent with the answers provided to questions 7 regarding intragroup transactions.Question 5: What are your views on how ‘x%’ (Thresholds 2 and 3) should be calibrated?
We believe the proposed threshold is flawed whatever the calibration of the “x” parameter because it rely on the regulatory CVA while the economic risk that needs to be monitored and adequately capitalized as part of ICAAP arises from accounting CVA variability instead.Additionally, we are in the opinion that setting a unique x% value is inappropriate, moreover for both thresholds. A “one-fit-all” value is likely to jeopardize the relevance of supervisory reviews because it will fail to capture the specificities of the institution which should be part of the SREP process, as defined in Article 107(3) of EU Directive 2013/36, “in a manner that is appropriate to the size, the structure and the internal organisation of institutions and the nature, scope and complexity of their activities”. We believe that supervisors must keep the ability to appreciate specific cases and take customized views different from those enclosed in the guidelines if deemed necessary.
Question 6: Do you agree with the scope of derivative transactions to be included into the calculation of hypothetical own funds requirements for CVA risk?
As mentioned in our general response, we are not trying to challenge the evidence that the exempted CVA perimeter is not risk free. On the contrary, we think that CVA risk shall be measured adequately. But, we do not agree with the approach proposed in the guidelines, which is based on the Basel 3 framework.If EBA continues imposing this automatic measure despite the negative consequences, we argue that
- The sovereign counterparties continue to be exempted until at least the Commission has reviewed the prudential treatment to sovereign counterparties as a whole.
- The NFC- counterparties continue to be exempted until the definition and thresholds have been reviewed by ESMA in the EMIR context.
- In both cases, a precise assessment of the impact of the removal of the exemption on the derivative market is an important prerequisite, as shortage of supply of derivative may impact badly the EU economy.
Question 7: Do you agree that intra-group derivatives transactions should be explicitly included into the scope of calculation? If not, what do you think could be a credible alternative treatment of the CVA risk of intragroup transactions?
No, we do not agree with this principle. We think that the intra-group transactions shall be exempted from the CVA charge at group level as soon as institutions are supervised on a consolidated basis. This exemption is consistent with the counterparty credit risk treatment for intra-group transactions on a consolidated basis.Indeed, CVA losses can occur either because of an increase in the institution’s counterparties credit spreads or because of an increase in the institution’s exposure to its counterparties. But on a consolidation basis, in accordance with the IFRS principle, intra-group transactions are cancelled. Hence, there is no volatility to institutions’ own funds neither CVA loss to institutions themselves.
Moreover, according to the CRR, changes in the value of own liabilities (DVA) shall be filtered from own funds. And, actually, the credit spreads of consolidated entities will be based on the CDS of its parent undertaking. Consequently, even if intra-groups transactions were included for the CVA charge purpose, this charge will be filtered in accordance with the DVA treatment.
From our point of view, intra-group transactions shall only cover transactions between entities in the same consolidation perimeter (as define in art. 382(4)(b)) for the prudential purpose. Similarly it is quite acceptable to include intra-group transactions these transactions do not fulfil the criteria listed in the article 382 (4)(b) of the CRR. Besides, transactions with structurally separated subsidiaries shall not be considered as intra-group transaction.
Reversely transactions with insurance subsidiaries or with entities recorded in accordance with the equity method shall not be considered as intra-group transactions. As a consequence, the CVA charge shall be computed as usual for these transactions.
Question 8: Do you agree with the approach provided for the determination of supervisory benchmark for material CVA risk?
As outlined in section 2.1, we believe that providing a quantitative measure of what should be the appropriate coverage of CVA risk as part of SREP is unsuitable.Moreover, we consider the proposed supervisory benchmark is arbitrary and flawed.
Question 9: What are your views on how ‘y%’ (Threshold 4) should be calibrated?
Whatever the calibration of “y%”, we consider the proposed approach is inappropriate. No calibration can address the shortcoming of the contemplated approach.Question 10: Do you agree with the approach provided monitoring of CVA risk by competent authorities and EBA and data to be provided to competent authorities for this monitoring?
We agree with the objective of enhanced supervision of CVA risk. However, as stated previously we consider the proposed approach will fail this objective (See sections 2 and 3 of the general comments). As a consequence, we believe the requested data will be of no use for the sake of monitoring CVA risk.As already mentioned above, the Pillar 2 process allows institutions to use appropriate models to measure CVA risk and in particular within ICAAP. By stating that the competent authorities should use the value of the hypothetical own funds requirements for CVA risk as specified in these Guidelines, the EBA removes this opportunity of using appropriate models, which is contrary to this fundamentally sound principle..