(1) General comments/Application date and scope
The German Banking Industry Committee (GBIC) concurs that it makes sense to integrate the requirements for supervisory stress testing in the SREP-Guidelines. We also understand the incorporation of requirements governing P2G, not least because many competent authorities already make a distinction in practice between P2R and P2G. Both the IRRBB guidelines and the stress testing and SREP guidelines have a considerable impact on the competent authorities’ requirements as regards risk management and control processes in the institutions. Implementation in organisational policies is normally regarded as a complete package because of the complex interaction between the requirements. The GBIC advocates a harmonised, later implementation date in this respect to avoid a situation where existing risk management requirements have to be adapted continuously.
Since the guideline was subject to a general review, a new implementation date should be given (paragraph 592).
(2) Summary of findings and scoring (section 4.10 inter alia)
In paragraphs 87, 136, 410 and 493, the findings on which the derivation of the individual viability ratings will be based should be formally classified – similar to supervisory assessments – and the relationship between the findings and the viability scores should be documented. This would allow banks to deploy their resources in a more targeted manner so that they can achieve a better score. Paragraphs 87, 136, 410 and 493 should therefore be supplemented as follows: “In the summary, the individual findings should be classified on a scale from 1-4 (1 = low severity, 4 = critical severity) and the methodology for deriving the viability score from the individual finding classifications [and the risk scores for sub-categories] should be communicated to the institution.”
(3) Assessment of institutions’ stress testing (section 5.7.3)
With regard to the requirements for the management of IRRBB and internal stress tests, please refer to the separate comments prepared by the GBIC. To the extent that they are relevant, amendments to these guidelines should also be reflected in chapters 5.7.3 and 6.5 of the revised SREP guidelines.
(4) Assessment of inherent credit risk (paragraph 200)
Under point a. of paragraph 200, competent authorities are to use institutions’ internal stress testing to validate the quantitative risk measurement results produced by internal models. Such an analysis does not appear to make sense. These internal models were not usually developed on the assumption of a stress situation, i.e. under fundamentally different environmental conditions. Validation of these internal models based on stress test results is not possible. The final sentence of point a. in paragraph 200 should be deleted.
(5) Summary of findings and scoring (paragraph 218)
According to paragraph 218, competent authorities should consider all known future regulatory changes when assessing stress test results (if these affect institutions within the time horizon of the stress test exercise). This means, conversely, that institutions would also have to consider all known future regulatory changes in their stress tests. We wish to point out that this was frequently not possible to a sufficient extent in the past due to uncertainty about the final regulatory requirements.
(6) Assessment of interest rate risk from non-trading activities (section 6.5)
With regard to the supervisory review of IRRBB, the GBIC welcomes the fact that internal methodologies can also be considered as appropriate, in accordance with paragraph 325.
From the GBIC’s perspective, section 6.5 lacks a reference to the assessment of IRRBB in the trading book in accordance with section 6.3 of the SREP guidelines. In particular, a case in which large parts of the IRRBB in the trading book have already been hedged by non-trading book positions should be analysed. How shall such a case be considered? Viewed in isolation, the risk estimate would be too high and hence also the own funds requirements.
In accordance with paragraph 311 b., an immateriality threshold should be applied to basis risk. For many institutions, this risk is almost meaningless when compared to interest rate risk.
In paragraph 311 c., options that are not interest rate-sensitive but nonetheless have an impact on IRRBB are not mentioned. In German institutions, these statistical options are commonly present in private housing finance, specifically the option allowing unscheduled repayments and the option in case of interest lock-ins of more than 10 years. Only some customers exercise these options based on purely rational criteria, as is the case with capital market products. Customers mainly have other reasons for exercising an option.
(7) Assessment of inherent IRRBB (section 6.5.2)
Paragraph 313: As an additional point d., the assessment of the institution’s overall situation (total own funds requirements, ratio of IRRBB to total risk, quality of interest rate risk management, etc.) should be listed.
In paragraph 315b., paragraphs 321 and 324, the outlier test systematically disadvantages institutions that have accumulated a high level of assets in their interest rate book thanks to a stable investment policy. As a rule, these assets do not lead to a 1:1 increase in capital. The higher the hidden reserves generated by the investments based on national accounting principles, the greater the interest rate coefficient will be. This high present value basis of the interest rate book means that particularly stable banks (with a high net asset value) are especially impacted by relatively high changes in present value. As a result, the outlier criterion cannot serve as an objective measure for assessing IRRBB compared with other countries. This indicator does not become more meaningful if the test is based on six scenarios and the denominator is still an accounting measure (whether own funds or Tier 1 capital). This does not change the effect described. If these indicators continue to be used to classify interest rate risk, the GBIC believes that no automatic mechanism can be based on them from which an own funds requirement can be derived.
The separate examination of the credit portfolio required in paragraph 319 is not critical for understanding interest rate risk. What is decisive is the total cash flow before and after steering the interest rate portfolio. The complexity arises when interest rate options are added, regardless of whether they are generated by the customer business or the interbank business. Nor does paragraph 319 clarify why less liquid instruments are particularly emphasised.
Paragraph 323: It should be clarified that this separate analysis of fair value instruments, etc. only makes sense for the NII perspective.
Paragraphs 319-328: The requirements are far too detailed overall and too technical, so there are concerns that the competent authorities could lose sight of an institution’s overall situation during examinations. As it can be presumed that the required analyses will be performed by the institutions themselves and reported to the competent authority, the principle of proportionality should be explicitly addressed here.
Paragraph 329: Total cash flow should be considered as a single figure and not just selected time buckets. Different cash flow structures can result in exactly the same risk/return profile. The GBIC believes that the supervisor should not imposed any additional requirements on the institutions based on this.
(8) IRRBB strategy and appetite (section 6.5.3)
Paragraph 332 e.: Interest rate risk is not automatically linked to cyclical changes in the economy. Entering into IRRBB is a conscious management decision. The IRRBB strategy is oriented on benchmarks or rules, but these must be dependent on the level and steepness of the yield curve, and not on macroeconomic indicators. The GBIC therefore advocates deleting this requirement.
Paragraph 332 k.: Using derivatives does not necessarily mean incurring a risk. Derivatives can also be used solely for risk mitigation/closing out. The special significance attributed to derivatives here is not clear. We are requesting the deletion of this point.
(9) Overall SREP assessment (paragraph 502 and following paragraphs)
The categorisation is not clear and offers scope for interpretation. For example, a bank that has a viability score of 1 (VS 1) in two categories and VS 3 in two other categories does not formally meet any of the five descriptions. For a total score of 2, all risk types and the recovery capacity must be rated at 1–2, and for category 3 all of them would have to score 2–3. The description should be amended to avoid an inconsistent decision here. Because a purely maximum approach would be excessively conservative, we are proposing using commercial rounding of the average of the individual categories to derive the score.
Assessment of the recovery capacity would be added as a new criterion for awarding the overall SREP score. The GBIC does not believe that it is appropriate to place recovery capacity on the same level as the business model or overall internal governance, among others. Rather, it is a subcategory of the risk assessment for capitalisation and liquidity. The criteria for recovery capacity in table 13 should therefore be deleted.
Alternatively, there should be at least a transparent explanation of the criteria to be used for assessing recovery capacity as adequate (or not), or a quantitative scale for assessing when recovery capacity is estimated to be low, medium or high.
It should also be considered that recovery capacity is based on a large number of assumptions. Examples would be assumptions regarding haircuts, the underlying scenario or the observation period. It must therefore be ensured that banks apply the same/similar conservative assumptions and periods so that recovery potential can be made comparable for all banks, and to avoid disadvantaging banks with more conservative assumptions in the assessment.
(10) Annex 2.4 (page 230)
The references should be updated (e.g. “CEBS Guideline on stress testing”).
(11) Additional GBIC comments on the current version of the SREP guidelines (EBA/GL/2014/13)
Regardless of the amendments currently planned, the GBIC believes that the revision of the SREP guidelines should also be used to revise some of the requirements in the 2014 version of the guidelines. In our comments on the earlier consultation (EBA/CP/2014/14), we expressed a number of criticisms that were not incorporated into the final guidelines (EBA/GL/2014/13). One example is the overestimating effect of the risk-by-risk approach in Title 7 for assessing the SREP capital requirements (P2R). In addition to that we would like to highlight the following aspects:
- Paragraphs 151, 187, inter alia of the draft guideline: In cases where peers are used to assess institutions, they should be explicitly notified to the institutions, together with the methodology used to select them. Transparency is necessary in this step so that institutions are given an opportunity to comment on their assignment and on any relevant differences compared with peer banks. This approach would also enable institutions to gain a better understanding of the supervisory perspective and promotes learning best practices within the institution’s own peer group. The following wording should therefore be added to the concerned paragraphs: “Competent authorities should inform institutions about the peers and the rationale behind building homogeneous peer classes used for comparison in their decision.”
- Paragraph 359 et seqq.: In the same way as for paragraph 151, the benchmarks and peers used for the assessment should be made transparent to the institutions so that the institutions can comment on them and best practices are promoted in the peer group. The following wording should therefore be added to paragraph 362 of the draft guideline: “…they should explain to the institution the underlying rationale and details of these benchmarks”.
(1) The “Background and rationale” section cites the “post-recovery strategy” as part of the recovery planning. Recovery planning merely comprises strategic analyses that are relevant before or for the recovery event, but does not include any strategic direction of the banks for the time after recovery has been completed. Because of the large number of ways in which a recovery event can be structured, it is not possible to rationally define a concrete post-recovery strategy. The GBIC’s view is that this term is therefore misleading and should be deleted.
(2) A note that additional areas of regulation need to be considered makes sense as a general principle. However, proportionality needs to be considered. A detailed comparison should not constitute a mandatory requirement for the supervision of smaller and medium-sized institutions. The assessment of recovery plans by competent authorities as an element of SREP should be removed from the scope of the guidelines for smaller and medium-sized institutions for reasons of proportionality (also to take account of the limited resources available to the supervisors).
(3) The GBIC expressly welcomes the fact that competent authorities should also explicitly incorporate the assessments, information and outcomes of other supervisory activities in the course of the SREP, so as to avoid redundant information requests in future. At the same time, SREP outcomes should certainly not mechanistically inform other supervisory processes and thereby be linked to them. In the GBIC’s view, for example, the possibility of an IPS/DGS intervening at an early stage should be included in the wording of numbers 12, 13, 16 and 17 (Background and rationale) and paragraph 395.
(4) The greater interlinking of coordinated management approaches and risk management systems further increases the complexity of bank management. From the GBIC’s perspective, smaller institutions should be explicitly exempted from the requirement to closely compare coordinated management approaches such as recovery planning, ICAAP/ILAAP and RAF, as the EBA guidelines would otherwise counteract the intended reduction in complexity for small institutions (see CRR/CRD review).
(1) Overall, this results in a considerably more comprehensive assessment of internal governance (EBA/GL/2017/11) as a part of the SREP. Application in full would disproportionately burden less significant institutions in particular. The GBIC is therefore calling for a stronger emphasis to be placed on the proportionate and risk-based approach as part of the assessment of internal governance.
(2) Overall internal governance framework (section 5.2)
The requirement in paragraph 91 k. for the internal governance framework to be transparent to all stakeholders goes beyond the requirements of the internal governance guidelines (EBA/GL/2017/11). Title VII of the governance guidelines stipulates internal communication with all relevant staff. For interested members of the public and the owners of the institution, the disclosure requirements and annual reporting already provide sufficient information. Paragraph 91 k. should therefore be deleted.
(3) Internal control framework (section 5.6)
The encouragement of a positive attitude towards risk management mentioned in paragraph 96 is an overarching aspect of an institution’s risk culture. However, this cannot be defined specifically or reviewed as part of the internal control framework. We suggest deleting that particular clause.
(4) Assessment of institutions’ stress testing (section 5.7.3)
Paragraph 112 should explicitly exclude the use of quantitative outcomes of internal stress tests for informing the SREP assessment. Analysing the qualitative suitability of stress tests makes sense, especially in the context of assessing risk management, but if internal quantitative outcomes are factored into the supervisory assessment, this can lead to misguided incentives. This is diametrically opposed to the fundamental concept of internal stress tests – identifying negative developments clearly and at an early stage – and should therefore be avoided by the supervisors. Subparagraph a. of paragraph 112 should therefore be deleted.
(5) Quantitative assessment of institutions’ stress tests done for ICAAP and ILAAP purposes (section 5.7.3)
It can make sense to analyse the impact of stress tests on an institution’s MREL requirement. Paragraph 126 should clarify that this will not be used as a basis for calibrating the MREL requirement. That would unnecessarily complicate the process of determining the requirement and restrict the level playing field between various institutions.
(1) Overall, the requirements enable a certain degree of flexibility and a proportionate, incremental implementation by the competent authorities. The GBIC welcomes the clarification that P2G will be offset against the capital conservation buffer (paragraph 397). By contrast, certain other requirements are too restrictive and call into question the fundamental idea of subdividing into P2R and P2G. The stipulations in the SREP guidelines on P2G should generally be guided by the current drafts of the CRD review that are being discussed. The GBIC has the following additional specific comments:
(2) Determining and setting P2G (section 7.7.2)
The EBA should specify in greater detail in paragraph 388 how the outcomes of supervisory stress tests will be used to determine P2G. Quantitative explanations regarding the incorporation of stress test outcomes into the determination of P2G should be provided. This is indispensable in light of the need for a consistent application by NCAs.
We would like to additionally note with regard to the use of supervisory stress test outcomes for determining P2G that a lack of standardisation and potential variability in stress testing methods will render reliable capital planning impossible. Changes in the configuration of the parameterisation or the alignment of stress tests will thus prevent the institutions from performing necessary and required forward-looking ability to plan and ensure capital adequacy.
Paragraph 393 should therefore explain in greater detail which of the “certain aspects” of P2R should additionally be included in P2G.
There should be no minimum level whatsoever for P2G, but rather an individual level that covers microprudential stress effects should apply to each institution, provided that those effects exceed the capital conservation buffer.
Paragraph 397: From our perspective, offsetting P2G against the capital conservation buffer necessarily results from the aligned objectives of both capital requirements: to cover stress effects. We are therefore asking for a correspondingly unambiguous formulation in paragraph 397 (“Competent authorities should fully offset P2G against the capital conservation buffer”).
(3) Communication and composition of P2G (section 7.7.2)
Paragraph 398: The GBIC is seeking considerably greater transparency about the determination of individual P2G. As a general principle, institutions should be able to base their capital planning and actions on the supervisory dialogue regarding P2R/P2G, which demands sufficient transparency about the reasons for defining supervisory actions and P2R/P2G. Paragraph 398 should be clarified accordingly. Additionally, the GBIC believes that benchmarking stress test outcomes (in both qualitative and quantitative terms) would be useful for the institutions in order to classify individual P2G. The GBIC therefore suggests the regular publication/notification of aggregate (benchmarking) information on stress test outcomes and/or P2G (but not at the level of individual institutions) by competent authorities.
The requirement in paragraph 399 and Annex 5 for P2G to be met in full by CET1 is inappropriate in our view. Article 104b of the Commission’s draft of CRD V stipulates the use of own funds and – in contrast to P2R in accordance with Article 104a(4) – no Tier 1 sub-ratios are specified. We believe that this is entirely adequate. The EBA does not have a mandate to specify a tightening of the expected Level 1 requirements by means of guidelines. In the opinion of the GBIC, stipulating the use of CET1 only would mean an inappropriate restriction, especially because other components of own funds can be used for loss absorption as well. The P2G notified by the supervisor constitutes an expectation as regards capital adequacy that does not have to be met by the institution at all times. We cannot understand in this respect that stricter standards are being imposed here than in the Pillar I own funds requirements and the P2R to be complied with at all times. We can only understand the EBA’s arguments at the hearing on 16 January 2018 to a limited extent. According to the EBA, it is necessary to meet P2G in the form of CET1 because P2G “protects” compliance with the combined buffer requirement.
According to paragraph 384 et seqq., P2G must be set sufficiently high that enough capital is available so that the institution does not fall below the TSCR even in an adverse stress scenario over a period of at least two years. On the one hand, it is possible to use buffers as a general principle. On the other, AT1 capital instruments are also suitable for absorbing losses. This means that an adverse scenario that materialises could be also be covered by AT1 capital without affecting the buffer requirement. But compliance with the TSCR would be assured even if the buffers start to be used for the losses. In our view, the requirement to hold P2G exclusively in CET1 is far too conservative, which is why we are urging other capital instruments to be allowed similar to Article 92 of the CRR.
According to paragraph 399, P2G must also be incorporated into the recovery plan together with the capital planning and risk management. This passage could be interpreted in such a way that P2G must also be included in the determination of the capital indicators in accordance with Title III and Annex II of EBA/GL/2015/02. However, we are of the opinion that P2G should not be included in the recovery plan for the following reasons:
- P2G sits above the combined capital buffer requirement and does not entail any automatic restriction on distribution (MDA) in accordance with Article 141 of EU Directive 2013/36/EU.
- Including P2G in the capital indicators in accordance with EBA/GL/2015/02 would run counter to the buffer concept of the combined capital buffer requirement – which specifically provides for a temporary shortfall in a crisis.
- Including P2G in the capital indicators in accordance with EBA/GL/2015/02 could trigger an obligation to issue an ad hoc release (at the latest after the shortfall) and thus result in the (unintended) publication of the P2G requirement.
The institution can temporarily fall below the P2G requirement in a stress or crisis event without automatically having to decide about a recovery status. P2G should not therefore have any impact on the recovery plan. We therefore suggest deleting the term “recovery planning” in paragraph 399 and Annex 5.
Paragraphs 400 and 401:
We believe that the wording of paragraph 400 contradicts the stacking order shown in Figure 6 and envisaged in the current discussions on CRD V. Breaching the stacking order cannot be excluded in the event of an AT1/T2 shortfall. For example, institutions could be forced to hold CET1 for P2G instead of using it to cover the AT1/T2 shortfall. This would contradict the stacking order. In our opinion, capital components that are originally designated for P2G purposes may generally be used in the case of an AT1/T2 shortfall.
By analogy, we believe that the requirements of paragraph 401 and in the example for communicating prudential requirements in section 7.9 go too far. If it were necessary to include P2G in all capital ratios, this would be the equivalent of a further increase in the capital requirements and hence an additional burden on the institutions. Optionally, the Pillar 1 requirements for AT1 and T2 should be disregarded in the interaction of P2G with other capital requirements. The ideal composition of the entire capital requirement (including P2G) can best be assessed by the competent authority, which is why it should be given the necessary freedom of action here.
We are therefore requesting the deletion of paragraphs 400 and 401 as well as the example for communicating prudential requirements in section 7.9. To avoid misunderstandings, we are also seeking clarification that P2G will also not have any effect in future on the Maximum Distributable Amount (MDA) and must not therefore be included in restrictions on distribution.
(4) Capital planning and other supervisory measures to address capital adequacy in stressed conditions (section 7.7.3)
According to paragraphs 404 – 407 and paragraph 547 a., as well as Annex 5, the draft guidelines require institutions whose own funds are sufficient to meet the OCR, but that fall below P2G, will have to submit a capital plan to the competent authority. Under Article 142 of CRD IV, this measure is only necessary if the institution does not meet the combined buffer requirement (i.e. OCR). The requirement thus constitutes an inappropriate tightening of the CRD requirements and should be deleted. This is also necessary in light of the fact that institutions prepare internal capital planning as part of ICAAP (see normative perspective of ECB’s SSM ICAAP guide) and the relevant information about regular data capture for SREP must be provided to the competent authorities.
(1) We believe that the disclosure of P2G – as proposed – is not reasonable. We wish to additionally point out that current drafts on CRD V (Article 104b) discuss the incorporation of an exclusion regarding the use of insider information in accordance with Article 17(1) of EU Regulation No 596/2014.
(1) Unfortunately, the core request to dispense with target capital ratios contained in the GBIC comments on the consultation paper on supervisory stress testing in 2016 (EBA/CP/2016/28) was not taken up. The proposed changes to the wording do not resolve our concerns.
According to paragraph 580, competent authorities should consider whether to set pre-defined target capital ratios (hurdle ratios, pass mark ratios). We would like to point out that, wherever the pass mark is set, this will ultimately be an artificial, arbitrary level. As a consequence, reaching or failing to reach the minimum ratio constituting the pass mark will be an arbitrary result. The absolute level of a capital ratio can never serve as the only indicator of a bank’s likelihood of becoming insolvent. In addition, the appropriate level of capital that a bank should hold will depend on its business model. A one-size-fits-all pass mark therefore makes no sense whatsoever.
What is more, such an approach suggests – wrongly – to the general public that a bank which exceeds the pass mark is “safe” while a bank which fails to reach it is on the brink of insolvency. Yet the subsequent financial difficulties experienced by banks which easily “passed” EBA’s stress test was a major reason for that test’s bad reputation. It is vitally important to make clear that the information value delivered by a stress test will necessarily be limited.
For these reasons, we would still recommend dropping the idea of a pass mark, given the potential for misinterpretation by the general public. A sounder approach would be to focus, when analysing and communicating the results, on the sensitivity of the capital ratios as a measure of participating banks’ vulnerability to the adverse scenario. The published results could focus on the absolute and percentage reduction in capital ratios on the day at the end of the simulation as a measure of the severity of the scenario’s impact.
We still recommend deleting paragraph 580.
(2) Supervisory stress tests are already a familiar instrument in Germany, including for LSIs. We are asking for the following aspects to be take into consideration, based on this experience.
(3) As far as possible, stress tests should be performed on the basis of existing information. In light of the extensive, ambitious goals in paragraph 572 et seqq., however, there are considerable doubts about whether this can be achieved with the existing supervisory reporting data. The goals with respect to challenging assumptions and risk drivers (paragraph 572 b.) will probably require substantial additional data from the banks’ internal management. Simplified approaches should be added for LSIs and small banks that only require manageable additional inputs.
(4) The outcomes and methodology of the stress tests should be sufficiently coordinated and transparent to allow the intended gaols to be achieved in a comprehensible and suitably practicable manner. Ultimately, this is the only way to secure acceptance and enable targeted discussions on conclusions at an objective level. The proposed use for planning the primary focus of supervisory activities (paragraph 573) can also only be implemented to reflect practice and risks following sufficient consultation. Last but not least, the transparency and simplicity of the methodology with regard to its use for determining P2G should also not prevent the ability to plan own funds requirements. Amendments to stress testing methods that are relevant for determining P2G must be published promptly so that they can be incorporated by the institutions.
(5) We do not think that the goal of achieving a similar level of detail and scope of instruments with regard to the scope and organisation of supervisory stress testing as for internal stress testing is realistic. We also do not think it is realistic to expect the supervisor to perform sensitivity analyses, scenario analyses and reverse stress testing (paragraph 578 b.). The adequacy of institutions’ own stress tests is already assessed, and we believe that this is a more efficient approach that can be realistically implemented. These results also already offer supervisors a starting point for planning the main focus of their assessments.
(6) The use of supervisory stress tests should also be limited to special, definable situations and issues that require uniform data resources that can be aggregated. The requirements governing supervisory stress testing should therefore concentrate on these special cases for obtaining comparable data on concrete issues and can then result in a valuable information gain, as demonstrated by experience in Germany.
(7) Paragraph 588 states that competent authorities can also develop institution-specific scenarios. This contradicts the goals, set out above, of obtaining comparable data on concrete issues. We therefore suggest deleting the passage “or to develop institution-specific scenarios for individual institutions (the latter should not be seen as relieving institutions from the responsibility of designing own scenarios for the purposes of ICAAP and ILAAP stress testing), or a combination of the two.”
(8) The draft guidelines give the impression that a parallel standard set of ongoing stress test calculations should be established by the competent authorities, though this would not provide additional benefit compared to information that are already available in institutions. There is no justification in this respect for the massive increase in further information and reporting requirements that can be expected from the “effective dialogue” with the institutions (paragraph 584).