EBA consultation on Future of Stress Test: old wine in new bottles?

On 22 January 2020, the European Banking Authority (EBA) published a discussion paper on possible future changes to the EU-wide stress test. The proposed new stress test framework changes the current approach significantly and will challenge both banks and supervisors. In the new framework, two calculation legs are introduced. First, the supervisory leg mirrors the current framework enhanced by additional top-down components. Second, within the so-called bank leg the EBA proposed a revised methodology that finally breaks with the inflexibility of the constrained bottom-up approach. This change in paradigm will lead to changes regarding governance, processes, implementation and execution of the stress test exercise as well as model development resulting in an additional effort for banks.

Shortfalls of the current framework

Following the financial crisis, the EBA established an EU-wide stress test program to improve banks‘ resilience, enhance transparency, and restore trust in the EU banking sector. After four stress test exercises already conducted, both banks and the EBA itself are calling for a reform of the stress test program. The reasons for this are obvious:

  • the lack of clarity and prioritization of the stress test objectives,
  • the usage of results and their link to the supervisory process,
  • the application of methodological constraints for some risks,
  • the ownership of results, and finally
  • the resource-intensive nature of the stress test exercise.

The new framework should tackle these concerns and tries to unite two worlds: On the one hand, comparability and conservatism, on the other hand, flexibility to identify banks’ idiosyncratic risks.

Proposed new framework

The new framework remains a microprudential exercise, whose main objectives are the assessment of banks’ capital adequacy and the identification of risks. Based on these objectives, EBA defined four success criteria for changes: increasing relevance, comparability, transparency, and cost-efficiency.

The proposed new framework of the EU-wide stress test is based on two legs: the supervisory leg and the bank leg. Both legs would use the same macroeconomic stress scenarios and starting points. The main result will still be a stressed CET1-ratio. But while the supervisory leg would continue to be mainly a constrained bottom-up exercise (with the possibility of top-down elements to phased in over time), the banking leg would be less constrained and more flexible regarding assumptions and models. The results of both legs will be used for different purposes: the supervisory leg results would be used for the banks’ Pillar 2 capital guidance (P2G), while the bank leg results would serve market discipline and transparency. (please click to enlarge)

The supervisory leg is very similar to the current approach. While certain risk categories would be projected bottom-up, others would be projected top-down using challenger models and benchmarking tools. Affected positions that could be modeled by supervisors using top-down approaches are net interest income, credit risk and net fee and commission income. These top-down approaches should be introduced gradually.

Since the result is clearly the responsibility of the supervisor, the QA phase is much less time-consuming. The EBA proposes two data submissions from the bank side and one iteration with supervisors. Furthermore due to the top-down approach, deviations from the current static balance-sheet assumption would be possible. Indeed, this assumption is fundamental as it assumes that balance sheet positions stay constant over the stress test horizon. An example of relaxation to that methodological approach that is named in the discussion paper would be the sale of a subsidiary that was confirmed after the start of the exercise. However, whether management actions to mitigate the impact of the stress can be taken into account as well is not discussed in the paper.

Due to critics regarding the traditional EBA stress test-framework lacking flexibility and being not risk-sensitive, supervisors have rethought the current constrained bottom-up methodology. As a result, the proposed bank leg represents a new “fresh air” building block that sprinkles unconstrained bottom-up elements into the revised, general setting of the EBA stress test. Within that leg, banks will use the common methodology – and hence therein proposed scenarios and officially published templates – but enjoy individual freedom of adhering to the constraints included. More precisely, it is discussed that institutions will receive a list of constraints that can be relaxed or even disregarded.

In order to connect pillar 1-driven exercises with the pillar 2-environment, it is considered to allow banks to use ICAAP models for estimation purposes. Nevertheless, scenarios and starting points would then be maintained to be prescribed across participants. But guidance is envisaged to be limited, as, for example, a list of deployable management actions for scenario projections is to be released.

Supervisory leg Bank leg
Ownership of results Competent authorities Banks
Extent of disclosure Aggregate + Bank level Bank level
Granularity of disclosure Uniform disclosure of a limited set of information (mainly capital ratios and key drivers) for each scenario Uniform disclosure of a comprehensive set of information (similar to current transparency) for each scenario

Table 1: Disclosure requirements of proposed EU framework, based on Box 3 of the EBA discussion paper

Furthermore, disclosure requirements are of central interest to the EBA. Table 1 compares the envisaged disclosure properties of the respective legs. A fundamental distinction is that for the bank leg, banks have full ownership of their data. Hence, there no strict quality assurance of the regulator is planned. Results can be withheld by institutions in order to be only subject to the overall QA process. On the other hand, results are owned by competent authorities within the supervisory leg.

With regard to extent and granularity of disclosures, bank leg-data is planned to be published comprehensively on a bank-by-bank level whereby supervisory leg-data will be released on a limited basis for both, aggregate and bank level.


The proposed changes to the current EU-wide stress test program is very welcome particularly with regards to the success criteria. The discussion paper highlights their importance to the successful implementation of the new stress test framework. Relevance of the results is argued to be enhanced by the introduction of the bank leg, allowing institutions to concentrate on idiosyncratic risks and specificities of their business model. Second, compatibility of the banks‘ specificities and individual risk profiles will increase in the supervisory leg, while in the bank leg the level playing field across banks seems tough. Transparency remains high but the published results are to be adapted to the different stakeholders. Finally, cost efficiency should increase significantly due to the reduced QA phase. However, in the short term, there will be of course a significant need for resources to implement the new framework.

In addition to the above criteria, there is another important success factor for the new EU-wide stress test framework. It will be crucial to what extent the results of the supervisory leg and the bank leg can be reconciled. The markets, in particular, will demand such reconciliation of the different results and will require a corresponding explanation in the event of deviations. However, in its discussion paper, the EBA makes no statement on the possibility of reconciliation (or whether it is at all desired).

There is still much work to be done on the details. In particular, the methodology for the supervisory leg needs to be clarified, e.g. with regard to the impact on the SREP. Nevertheless, EBA’s proposal on possible changes to the EU-wide stress framework is a step in the right direction and can therefore be considered positive.



Frank Marinkovic

Telefon: +49 40 6378 1668





Sophie Kroll

Telefon: +49 89 5790 7097




Hinterlassen Sie einen Kommentar

Ihre E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.

/* */