EBA considers bottom-up stress testing with top-down elements

EBA considers bottom-up stress testing with top-down elements

Fri 25 Nov 2022

The European Banking Authority (EBA) is tasked, in cooperation with the European Systematic Risk Board (ESRB), to initiate and coordinate biennial EU-wide stress testing exercises to assess the resilience of institutions to adverse market developments. The objective is to provide supervisors, banks, and other market participants with a common analytical framework to consistently compare and assess the resilience of the EU banking system to shocks, be they adverse or not, and ultimately challenge the solvency of EU banks.

The forthcoming 2023 exercise builds upon discussions initiated in 2019 when the EBA proposed in a discussion paper to make the stress testing framework evolve for future rounds, after the publication of a special report of the European Court of Auditors (ECA). In its report, the ECA recommended that the EBA should test the resilience of institutions by introducing a top-down approach to complement the current bottom-up approach in order to ensure greater consistency and more control over the process. At the same time, such an approach would provide a benchmark for the stress tests conducted by the competent authorities and individual banks.

The EBA, therefore, discussed whether the framework could evolve from a bottom-up approach to a top-down one – such as the comprehensive capital analysis and review (CCAR) in the US – or to what extent a mixed or hybrid approach could be envisaged. The main difference is that a bottom-up approach refers to banks using their own models and means to perform the stress test. In contrast, a top-down test mostly relies on supervisors’ capabilities, such as the use of a challenger model and data from supervisory reporting.

Previous exercises in 2016, 2018 and 2021 were quite similar in terms of methodology and scope of risks covered, with the exception of the 2018 exercise, when banks had to cope with the introduction of International Financial Reporting Standard (IFRS) 9 and the new staging and provisioning model.

Main elements and scope of the 2023 package

The stress testing exercise is still based on a common EBA methodology associated with a set of templates aiming at capturing starting point data and stress test results.

The package contains:

  • A methodological note that describes the common methodology on how banks should calculate the stress impact of the common scenarios and sets constraints for their bottom-up calculations;
  • The templates used for collecting data from the banks, as well as for publicly disclosing the outcome of the exercise;
  • Additional guidance provided by the EBA on templates.

The 2023 exercise is carried out on a sample of banks covering broadly 75% of the European Economic Area (EEA) banking sector, which is 5% more than in 2021, and is still running at the highest level of consolidation, as per the Capital Requirements Regulation (CRR) scope and rules. As this 5% corresponds to about 50% of newly subjected institutions, the methodology contains some derogations for smaller institutions as per the proportionality principle. Notably, for the first time there are several subsidiaries of US banks subject to the stress tests, which have likely relocated to the EU after Brexit.

Moreover, it should be noted that beyond this sample, supervisors could decide on extending the scope to other banks that have a minimum of €30bn of total assets. To that respect, the ECB should once again submit all banking union area banks under its direct supervision to this exercise.[1] The list of participating banks is provided in Annex I of the methodology.

As regards macroeconomic scenarios and risk type specific shocks, a common ECB baseline scenario that all participating banks will have to implement will be published just before the official launch of the exercise, as well as a common adverse scenario provided by the ESRB. The scenarios will cover the period of 2023-2025 with a static balance sheet assumption.[2]

A hybrid approach

After discussion with the industry, the contemplated approach of the exercise is still a bottom-up stress test, but there will be some top-down elements. For the EBA, this mix-and-match approach is actually a “constrained bottom-up” exercise, since banks will still be required to project the impact of the different prescribed scenarios using their own models. However, they will be subject to strict constraints and to a thorough review by supervisors.

Like previous exercises, risks to be covered are:

  • Credit risk, including securitisations;
  • Market risk, counterparty and credit valuation adjustment risks (CCR and CVA);
  • Operational risk, including conduct risk;
  • Net interest income (NII);
  • P&L and capital items not covered by other risk types.

In particular, banks will be required to make use of prescribed parameters for risk weights of securitisations, the credit loss path of sovereign exposures[3] and for the projections of net fee and commission income (NFCI). For the latter, practices such as loss-transfer agreements and transfer pricing with entities outside the scope of consolidation should not be considered.

Then, banks will have to consider the regulatory framework brought into force and applicable as of 31 December 2022, as well as any decisions taken by competent authorities regarding the application of the Capital Requirements Regulation/Capital Requirements Directive (CRR/CRD), including national measures, but they will not have to anticipate other changes to the regulatory framework. The impact of the exercise will be reported in terms of capital (CET, T1, total, LR) and from a transitional to a fully loaded perspective. For global systemically important institutions (GSIIs), if the projected leverage ratio (LR) for a given year of the stress test horizon falls below the LR maximum distributable amount (MDA) trigger point[4], banks will be required to project reductions of distributions for the same year following some simplifying assumptions.

Finally, the use of new internal models and modifications of existing ones is mandatory where they were decided before 31 December 2022. The reference data will be those of year-end 2022 figures.

The stress testing outcome will feed the 2023 supervisors’ SREP

First, the ECB will undertake the heavy lifting of ensuring the quality and comparability of the results submitted by banking union area banks. Then it will consider each bank’s individual result for the supervisory review and evaluation process (SREP). Indeed, stress testing exercises have become an efficient tool for supervisors to help them calibrate additional capital requirements or expectations in the context of Pillar 2, yet the exercise is a no pass or fail test.

Key milestones of the 2023 cycle :

  • Launch of the exercise at the end of January;
  • First submission of results to the EBA at the beginning of April;
  • Second submission to the EBA in mid-May;
  • Third submission to the EBA at the end of June;
  • Final submission to the EBA in mid-July;
  • Publication of results by end-July.

The disclosure of granular data on a bank-by-bank basis is supposed to facilitate market discipline and serves as a common ground on which competent authorities base their assessments.

What to expect from current and future exercises

Pending forthcoming baseline and adverse scenarios, and contrary to the revolution expected when the EBA discussed the framework in 2019, the 2023 methodology is not so different to 2021.

During the consultative period, banks complained that the contemplated dual-track approach could increase their workload and put them in the difficult position of having to explain differences between the two sets of results – banks’ leg vs supervisors’ leg – to stakeholders and investors. Moreover, a pure top-down approach is unlikely to materialise over time since any changes to the stress testing framework that would generate more work for institutions would mean an extra burden for the supervisors in charge of the quality assurance on the outputs generated by bank models. However, the constraints – in the form of projections modelled by the supervisor – would likely become much more important over time.

Finally, as this exercise increases the sample of banks, in particular those that are using standardised approaches to compute capital requirements, we might see the stress test delivering a wider range of outcomes due to the different risk sensitivity in the stress test across both more and less sophisticated banks. Plus, the process of stress-testing banks on standardised approaches could become even more pertinent once the Basel III prudential rules enter into force in the EU because, for many banks, the cost and operational difficulties of maintaining internal models could outweigh the capital efficiency gains.

[1] Banks subject to mandatory restructuring plans agreed by the European Commission could be included in the sample by competent authorities if they were assessed to be near the completion of the plans. On the contrary, banks with specific business models or banks taking part in a merger or are acquired by another bank could be excluded, subject to approval.

[2] It means that assets and liabilities that mature or amortise within the time horizon of the exercise should be replaced with similar financial instruments in terms of type, currency, credit quality at date of maturity, and original maturity as at the start of the exercise. No workout or cure of S3 assets is assumed in the exercise. Moreover, it is assumed that banks maintain the same business mix and


[3] It has to be noted that risks arising from sovereign exposures are covered in credit risk and in market risk depending on their accounting treatment.

[4] As per newly CRD5 introduced articles 141b and 141c.