ESBG welcomes the initiative to revise and re-centre the EU-wide Stress Testing framework. We hope that any changes to the framework will result in increased transparency and simplicity of the overall process. The tremendous increase in granularity over the past cycles should be subject to a cost-benefit analysis.

Changing the current framework as described in the discussion paper may however not bring additional benefits and risks reducing some of the advantages of the existing process. In particular, banks’ involvement throughout the process gives incentives to invest in and develop internal risk management frameworks and models. The banks involvement also allows supervisors to better understand the mechanics of banks’ internal models and projections in relation to their business models and risk profiles.

The different results of the two legs, as proposed in the discussion paper, would in all probability cause an increase in overall quality assurance (QA) requirements. This would increase the complexity of communication of results and lead to higher resource allocation (both in the banks and with the supervisors). The focus of the exercise would likely shift towards investigation and explanation of differences between the two legs of the calculation. In addition, the increase in complexity may not lead to the desired in-crease in comparability (analysts and investors will be required to compare and contrast multiple bank legs from across the industry).

The costs for developing such two-leg stress test approach should be considered. The balance between relevance on the one hand and cost-efficiency on the other do not appear to be very balanced within the proposed two legs exercise. To mitigate costs and increase benefits from future EU-wide stress tests, the EBA should rather consider improvements to the existing and already-known processes of the established stress test governance, procedure and methodology. With a particular focus on improving process efficiency: examples include the starting points for each of the various calculation streams i.e. the prepopulation of starting point data with data delivered to either EBA and/or competent authorities within regular reporting process (COREP reporting).

ESBG would therefore encourage the EBA continuing future stress tests with only a single-leg. The current approach has the benefit that banks are tested against a common methodology and, as stated in the paper, the main benefit of undertaking stress tests in such a coordinated fashion is the production of transparent and comparable outcomes. Moreover cost-efficiency criterion in section 3.1.2 (page 13) of the paper would be achieved using only a single-leg, together with the right balance of the other criteria.

Maintaining a single-leg would be in line with the primary objective of EU macro prudential stress testing to assess capital adequacy and identify risks. This would also allow focus to remain on improving the existing methodology and removing or relaxing constraints that do not have sensible impacts (e.g. treatment of sight deposits methodology, Risk Weighted Assets (RWA) floors, stage 3 recoveries, administrative expense floors etc.).

The regulatory authorities should consider maintaining a single-leg stress test along with the implementation of simplified rules on banks’ communication of results and improvements to the existing methodology by removing or relaxing constraints that do not have sensible impacts. Allowing banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary. If this change were implemented in lieu of a second “bank leg” calculation the exercise would keep the existing advantages of the current regime, allow for cost efficiency and achieve increased transparency to market participants. These changes should lead to more sensible results which are better aligned with the real expectations and to broader acceptance by market participants and the public.

Where feasible, a hybrid approach with the use of top-down models to replace some bottom-up elements could be an improvement over the current framework. This includes the implementation of top-down models (e.g. NII, NFCI and Credit Risk).

However, the use and especially the outcome of the models under the applied methodology would need to be discussed and reviewed with the participating banks. In particular, banks would need more transparency on the challenger models and the peer bench-marks’ outcome (design and calibration should be disclosed). Certain aspects of the methodology have proven troublesome in that a one-size approach does not work for all types of participating institutions. It may be necessary to consider multiple versions of the top-down model, tailored in some respects to the different types of participating institutions.

Constraints with respect to profit and losses (P&L) should be relaxed to allow for reasonable changes to costs. For example, in the development of the current Covid-19 crisis travel expenses will decline substantially. Another example is RWA floor in the credit risk area, where removal of the starting point floor should ensure that there is no double counting of capital held for defaulted exposures. As default-ed exposures are already accounted for through risk costs in the P&L.

Internal capital adequacy assessment processes (ICAAP) are relatively heterogeneous across EU banks. Using the internal stress testing processes as the basis for the bank leg will highlight the differences in banks internal practices.

The idea of prescribing a flexible methodology can bring advantages as the banks know what is best for their business and therefore, can decide when a constraint is beneficial or not to be applied.

In ESBG’s opinion, the regulatory authorities should therefore consider maintaining a single-leg stress test along with the implementation of simplified rules on banks’ communication of results. Allowing banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary. If this change were implemented in lieu of a second “bank leg” calculation the exercise would keep the existing advantages of the current regime, allow for cost efficiency and achieve increased transparency to market participants.

In our opinion, the regulatory authorities should consider maintaining a single-leg stress test along with the implementation of simplified rules on banks’ communication of results. Allowing banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary. If this change were implemented in lieu of a second “bank leg” calculation the exercise would keep the existing advantages of the current regime, allow for cost efficiency and achieve increased transparency to market participants.

ESBG believes it is important that the bank’s own calculations are taken into account when generating the final supervisory outcome. However, duplicating the entire calculation seems overly resource intensive.

The regulatory authorities should consider maintaining a single-leg stress test along with the implementation of simplified rules on banks’ communication of results. Allowing banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary. If this change was implemented in lieu of a second “bank leg” calculation the exercise would keep the existing advantages of the current regime, allow for cost efficiency and achieve increased transparency to market participants.

ESBG believes that running a full parallel calculation (with relaxed constraints or mixed methodologies) would greatly increase overall workload. Authorities should consider instead simplified rules on banks’ communication of results and allow banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary.

If the efforts for submitting/calculating data and for the Q&A sessions will be reduced, then a cost reduction could be obtained. On the other hand, as there will be multiple results to consolidate (and up-date, re-update, duplicate versions) this is likely to counterbalance any effort reduction.

In our view, the top-down adjustments by supervisors within the supervisory leg without banks’ consent is a step backwards as banks’ challenging possibilities for the supervisors top-down changes will still be limited in the future.

Additional efforts on all sides will be required to effectively communicate two legs of results, especially to market participants, without even knowing whether these will be recognised by the market. We get the feedback that analysts are in fact more interested in the results of the supervisory leg, which would often happen to be more pessimistic and that allows comparison between banks. These efforts should not be underestimated and need to be considered in any cost-benefit analysis of the change.

In summary, we see increasing costs and less benefits from having to perform a two-leg stress test.

In ESBG’s view, the most unrealistic constraint in previous stress tests was the pricing below the -0- floor on retail deposits. In the experience of some of our member institutions, negative rates for retail depositors have not been feasible. Repricing from a non-existent negative deposit rate resulted in excessive degradation of their overall net interest income (NII). Compared to the ECB calculation, using internal assumptions (for example not full but 50% pass through of the benchmark rate in NII) results in a major lower impact.

Floor on other administrative expenses: banks should be allowed to decrease this position in adverse scenario. Given the static balance sheet approach and the usual GDP decrease in adverse scenario, there can be substantial evidence for lower costs.

Other constraints that should be reconsidered:
• Coverage ratio STAGE1 must be non-decreasing. This is not necessarily true on aggregated (portfolio) level, if there is enough movement between STAGE1 and STAGE2.
• Staging criterion - Increase of 300% in PD must trigger S2 (this causes additional S2 migrations for minor changes in upper rating grades).
• IFRS9 PD/LGD curves – mean reversion application does not reflect accurately the behaviour of curves across ratings. Alternative could be: 1) Drop mean reversion 2) include it
in starting point
• No recoveries from S3 – generates overly pessimistic outcomes
• Total REA values floored at level of starting point. For IRB approach migration to default effect leads to decrease of REA. The effect of deterioration of through the cycle
parameters does not offset the migration to default effect, thus REA might be decreasing also under adverse scenari-os.
• Constraint from the template: According to ECB guidance for SSM banks following should hold:
o If PDpit(t) > PDreg(t) then PDreg(t+1) > PDreg(t)
o LGDpitNew(t) > LGDreg(t) then LGDreg(t+1) > LGDreg(t)
o In case of improvement of PiT parameters (bottom of economic cycle) this does not neces-sarily hold.
• Cap on net fee and commission income: banks should be allowed to use their own model with-out a minimum haircut, if the model has been validated and outcome can be
confirmed by back testing.
• Rationale for usage of benchmarks for sovereign portfolio unclear, more transparency here would be appreciated.

ESBG believes that for the bank leg the discretion of relaxing or not the methodology should be with the banks if it leads to more plausible results.

The regulatory authorities should consider maintaining a single-leg stress test along with the implementation of simplified rules on banks’ communication of results. Allowing banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary. If this change were implemented in lieu of a second “bank leg” calculation the exercise would keep the existing advantages of the current regime, allow for cost efficiency and achieve increased transparency to market participants.

Significantly different, driven in large part by the use of growth assumptions, absence of supervisory flooring and pass-through constraints, leading to results more aligned with real-world developments. Relaxing of other constraints has a more limited impact on results but can in certain areas materially affect the effort in calculation.

Growth assumptions in particular require different methodology for calculating projections.

Outcomes also depend to a large extent on the chosen scenario. While the EU-wide stress test considers a macro-economic systemic shock suitable for a wide range of stakeholders; the internal stress test focuses on scenarios tailored to the specificities of the individual bank (e.g. crisis in a certain portfolios, industries or geographies).

Comparability between banks is best assured via the supervisory leg with top-down constraints on methodology and balance sheet development. General guidance is welcome, in so far as it assists the banks in determining best practice.

As already mentioned, ESBG sees little benefit in introducing the bank leg.

The regulatory authorities should consider maintaining a single-leg stress test along with the implementation of simplified rules on banks’ communication of results. Allowing banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary. If this change were implemented in lieu of a second “bank leg” calculation the exercise would keep the existing advantages of the current regime, allow for cost efficiency and achieve increased transparency to market participants.

We do not think an increase in cost-efficiency in point 74 of the discussion paper will result from the proposed changes. There might be a benefit of less communication and QA with the authority however the communication and QA efforts might increase towards other external stakeholders such as rating agencies and analysts. This statement is also confirmed within Box 5 with expected communication challenges to arise from the publication of two different sets of results and the high probability of intense discussions between banks and authorities for the supervisory leg results as a major input to setting the P2G level. Furthermore, the last bullet point in Box 5 questions the overall need for calculation of a bank leg with no connection to setting of regulatory requirements such as P2G.

As already mentioned, ESBG believes that the regulatory authorities should consider maintaining a single-leg stress test.

Supervisory QA would, in fact, seem to have limited benefits in our view; especially as publication of the bank leg and subsequent analyst interest should enforce market discipline. Since the QA process can be very long and time consuming, reducing the QA sessions for one of the legs would be beneficial.

Though dynamic effects are considered in the internal stress testing program of some of our members, it is not clear how any of these could be sensibly implemented in the context of an EU Wide Stress Test without compromising comparability and efficiency.

The granularity of disclosures is too high in our view. We would therefore invite the EBA to consider decreasing the granularity; where this reduction would be possible.

The current disclosures for the EBA stress test deem to sufficient and also acknowledged by all relevant stakeholders next to competent authorities and banks such as analysts, investment banks and rating agencies.

The banks legs will not be directly comparable across the sample. Banks should be allowed maximum flexibility in the communication of the bank leg in a way that is appropriate to their own circumstances. The supervisory leg is best suited for cross-industry transparency and therefore should have a ‘required’ granularity.

As outlined in our answers to other questions we do not support EBA’s suggestion of incorporation of a second bank leg calculation. In case a second leg will be implemented, this introduces the risk that relevant stakeholders might misinterpret the results publication and much more Q&A and explanation work will be needed from both banks and regulators. This will in all probability generate significant efforts in order to get proper reconciliation between the different legs. All involved and interested stakeholders are used to the current and established disclosure procedures. Benefits to an additional leg must be clearly material in order to change an accepted and acknowledged process.

Instead of a second “Bank Leg”, the EBA should consider simplified rules on banks’ communication of results and allow banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary.

Compressing the banks results to only focus on capital depletion; though market friendly does not do justice to the depth of information used to generate the outcome. However, we do not see any challenges to maintaining the overall comparability given lower granularity of disclosure.

We would encourage the EBA to refrain from increasing the complexity or granularity of the transparency templates. Granularity of disclosures is too high in our view. Please consider decreasing the granularity; reduction of granularity should be possible. In particular with regards to Credit Risk IRB and STD approach.

Capital and P&L publication is generally fine and similar enough to existing reporting structures.

ESBG would encourage the EBA to refrain from increasing the complexity or granularity of the transparency templates. Granularity of disclosures is too high and should be reduced; reduction of granularity should be possible. The added-value of the current granularity is questionable, also because a lot of resources both in system and personal are needed to populate such data with uncertain benefits.

As already mentioned, ESBG generally believes that the regulatory authorities should consider maintaining a single-leg stress test.

The bank legs will, in fact, not be directly comparable across the sample. In our view, banks should be allowed maximum flexibility in the communication of the bank leg in a way that is appropriate to their own circumstances, including information on the revised CET1. The supervisory leg best suited for cross-industry transparency and therefore should have a ‘required’ granularity regarding CET1 capital depletion (Please also refer to our answer to question 15). This approach will probably result in more question marks than in added value for any involved stakeholders.

ESBG believes that the regulatory authorities should consider maintaining a single-leg stress test.

Supervisory involvement in the “bank leg” would, in fact, lead to a duplication of the efforts. Any deviation of the starting points will be enforced via market discipline in the publication (explanations will be required by market analysts). We would therefore invite the EBA to consider simplified rules on banks’ communication of results and allow banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary (Please see also answer to question 15).

Explaining deviations from the proscribed methodology will require clear and concise communication by the banks when publishing the “bank leg”. The banks will need to have clear and compelling reasons for deviating from the supervisory methodology. This will also mean that banks are required to comment on the results of the “supervisory leg”. The limited disclosure of supervisory models poses however a challenge to banks that cannot entirely explaining eventual differences between the bank leg and the supervisory leg.

Overall increase of efforts will depend on the level of granularity in the final publication of the two legs. We would therefore stress again that the regulatory authorities should consider maintaining a single-leg stress test along with the implementation of simplified rules on banks’ communication of results. Allowing banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary. If this change were implemented in lieu of a second “bank leg” calculation the exercise would keep the existing advantages of the current regime, allow for cost efficiency and achieve increased transparency to market participants.

Explaining deviations from the proscribed methodology will require clear and concise communication by the banks when publishing the “bank leg”. The banks will need to have clear and compelling reasons for deviating from the supervisory methodology. This will also mean that banks are required to comment on the results of the “supervisory leg”. The limited disclosure of supervisory models poses however a challenge to banks that cannot entirely explaining eventual differences between the bank leg and the supervisory leg.

A primary challenge in this respect is finding the proper balance of transparency and granularity for the disclosures. The level must be appropriate for external stakeholders to interpret the results and enable a comparison between the banks for both legs. Predicting the appropriate level of detail in advance of the exercise is difficult. In any case, the supervisory leg should be the absolute limit on the level of granularity. It does not make sense then to provide more granular information for the bank leg, if this information cannot be compared to the supervisory leg.

We would invite the authorities to consider simplified rules on banks’ communication of results and allow banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary at a level of detail left to the banks’ discretion.

Disclosing the supervisory of banks’ capital needs by publishing the final P2G as outlined in option (i) may increase transparency may help fostering market discipline. In our view, the application of options (ii) and (iii) could, in fact, lead to potential misinterpretation by market participants and investors.

ESBG considers that the P2G is a recommendation made by the supervisor in a case by case basis and there is no need to publish it since it is not a requirement. Alleging transparency reasons, banks are being forced to publish increasingly more information to market participants and analysts and not always all the information can be absorbed and interpreted in the correct way.

If P2G is published, analysts may consider it as a requirement and push banks to comply with it earlier than expected. It would have therefore a counterproductive effect from the point of view of market valuations, particularly towards listed banks.

Publishing the banks’ final P2G may also have unintended consequences since not all banks may be in a very comfortable capital position and they will be prejudice for publishing information that even the supervisor has been reluctant to publish.

Before considering P2G publication, it should therefore be clear to banks and to the market how the stress-test outcome feeds into P2G. Then, it will be necessary to ensure that the market has a full under-standing that P2G is not a requirement and that its use does not result in a trigger of any minimum distributable amounts (MDA) restrictions

In ESBG’s view, communication of management actions on the capital position are rather critical as they might be perceived as credit-negative by analysts and rating agencies. Especially if such actions do not fully reflect external stakeholders’ expectations. Public communication of such actions should be done on a voluntary, case-by-case basis.

The authorities should therefore consider simplified rules on banks’ communication of results and allow banks to communicate potential management actions and future dividend capacity. This information should be structured along with supplementary information at the bank’s discretion.

We believe that the proposed framework resolves some of the issues with the old framework while creating some new challenges. In particular, the ‘bank leg’ as described will impose undue burdens on existing resources across the banking industry, without commensurate benefits for the participants.

As outlined in our answers to other questions, we do not see benefits for certain criteria with the pro-posed framework (e.g. cost-efficiency).

For the ‘bank leg’, authorities should rather consider simplified rules on banks’ communication of results and allow banks to communicate concerns regarding top-down methodologies and constraints along with supplementary information where the banks deem necessary.

Please find the filled matrix also in the attached ESBG's response.

Bank leg:
Relevance: Increased: more tailored to an individual bank‘s business model.
Comparability: Decreased.
Transparency: When constraints are relaxed the us-age of banks‘ internal models will increase amount information but may decrease transparency and comparability.
Cost-efficiency: Decreased.

Supervisory leg:
Relevance: Same or slightly decreased.
Comparability: Increased.
Transparency: Decreased: supervisory models might not be fully transparent.
Cost-efficiency: Slight decrease depending on granularity of risk type submission.

Overall exercise:
Relevance: Same or slight increase: we assume that the overall relevance is mostly dependent on the supervisory leg while the bank leg would become more relevant and
the bank leg is the one based on internal knowledge of the business
Comparability: Decreased as the bank leg would be the more important leg having more disclosures/more granular data, the overall comparability will decrease.
Transparency: Decreased overall due to difficulty of making comparisons between institutions.
Cost-efficiency: Decreased: necessity to calculate second leg as well as calculations needed to assess the impact of single drivers of the calculation (supervisory vs bank
outcome). The cost-efficiency will be different for each risk type. However, should this be accompanied by decreasing the time spent with QA sessions and
fewer iterations in the supervisory leg, an increase in efficiency can be achieved

ESBG sees limited benefits and many costs when implementing two adverse scenarios. Performing two adverse stress test calculations would lead to material increase in efforts both for the banks and for supervisors during the QA process. Changing to two scenarios would also increase the complexity of communication to explain the results towards external stakeholders.

In addition, having two adverse scenarios will still not be sufficient to fully cover the risk profiles of all the EU banks due their heterogeneity of business models, clients and geographies. If one looks at the current Covid-19 crisis it is visible that different countries and banks residing in these countries are being impacted quite differently, meaning there will be never one or two perfect scenarios which will be equally severe for the whole sample of EU banks.

Single well calibrated scenarios can fulfill the purposes of two separated scenarios, therefore we would suggest keeping one scenario only.

ESBG believes that two asymmetric adverse scenarios will almost certainly lead to a significant increase in the complexity of the exercise. Because of this, this option will yield less benefit than one well calibrated single scenario.

Please see in addition our answers to question 28.

We think there is a higher value here compared to introduction of an additional scenario.

Nevertheless, the benefits of a sensitivity analysis for the EU-wide stress test would seem to be limited benefit and will engender higher costs both for calculation and communication/explanation to market participants. Sensitivity analysis are already a regular part of the overall ICAAP and the budgeting and planning process of a bank and such sensitivity analysis are then reviewed within the regular SREP process.

ESBG believes that exploratory scenarios will increase costs and complexity of the overall stress test exercise. We rather suggest sticking with the established processes and focus on separate single stress test exercises.

Exploratory scenarios introduce useful insights in the banks risk management practice, and therefore might be used as a valuable tool for the internal stress test exercises. Nevertheless, considering specific individual features of participating banks, appropriate calibration of such a scenario for a wide sample of banks is likely to not be technically feasible. Unfortunately, exploratory scenarios are likely to increase costs and complexity of the overall stress test exercise without clear benefits should they be calibrated based on “one suits for all” principle.