Beyond the single bank: macroprudential insights from the 2025 EU-wide stress test and its extensions

Prepared by Costanza Rodriguez d’Acri and Frances Shaw

Published as part of the Macroprudential Bulletin 32, November 2025.

This overview article provides an introduction to the 2025 Macroprudential Stress Test Extension Report (MaSTER), released as the 32nd edition of the Macroprudential Bulletin, which investigates how the EU-wide stress test and its extensions provide a broader assessment of the systemic vulnerabilities of euro area banks. The 2025 EU-wide stress test results are expanded via a top-down model-based toolkit to assess additional risks, perform policy simulation exercises, and present novel approaches to gauging the severity of the adverse scenario. In this article, lessons are drawn from these extended stress tests through a comparative analysis of capital depletion. These exercises, conditioned on the EBA’s 2025 adverse scenario, depend on the selected transmission channels and explore a relevant albeit incomplete set of risks. Overall, available results support authorities’ cautious approach to capital buffers. They suggest that, while most banks are resilient to the tested shocks, considering risks not monitored under the current EBA methodology (such as climate risk, liquidity risk and contagion risk) may uncover new vulnerabilities.

1 Introduction

Bottom-up stress tests, including the EU-wide exercise, provide insights and data to support macroprudential risk identification and policy design. This edition of the Macroprudential Bulletin uses these to expand the microprudential findings of the 2025 EU-wide stress test to a macroprudential perspective.^[1] Each article and box relies on the results from the 2025 stress test to look beyond the perspective of individual banks and assess additional macroprudential risks using a more comprehensive diagnostic toolkit. These include climate risk, liquidity risk and contagion risk, as well as feedback effects to and from the macroeconomy and other financial sectors.

Stress tests constitute an important forward-looking tool to support macroprudential policymakers. By simulating banks’ reactions to severe but plausible adverse scenarios, stress tests help uncover hidden vulnerabilities. They are instrumental in estimating the potential impact of emerging systemic risks and in informing prudential policy choices, offering useful information and insights beyond their microprudential use.

2 Scenario severity indicators enhance stress test credibility and policy relevance

Authorities need well-calibrated, relevant scenarios to identify financial stability risks and inform prudential policies. Adverse scenarios for stress tests differ from traditional downside scenarios in that they outline hypothetical yet plausible paths for macro-financial variables which represent a sufficiently severe and relevant stress event. The interpretation of stress test results, as well as their credibility, has therefore been increasingly associated with the notion of scenario severity.

Figueres et al. (2025) introduce a framework for evaluating scenario severity, showing that stress test scenarios have become increasingly severe over time. Indicator-based methods, macroeconomic modelling and plausibility metrics are employed to summarise scenario severity in a tractable way. Severity is also assessed from the starting point of the macroeconomy, taking into account how plausible scenario paths vary with the prevailing levels of systemic and financial risk.

3 Building on the EU-wide bottom-up stress tests for risk assessment

The 2025 EU-wide stress test results can be used as a building block for a wider range of top-down risk assessments. Under this year’s adverse scenario^[2], an aggravation of geopolitical tensions disrupts trade channels and lowers global economic growth. Although banks projected greater losses due to deteriorating credit, market and operational risk compared with the 2023 stress test, solid profitability helped cushion them against these losses. However, the banking sector is exposed to additional risks that are not reflected in the methodology underlying the 2025 EU-wide stress test. These are therefore analysed in this edition of the Macroprudential Bulletin.

Transition and physical climate risks are assessed using top-down models that leverage the data and results submitted by banks. By combining firm-level information from the AnaCredit dataset and starting point information on banks’ probabilities of default from the EU-wide stress test, Abbondanza et al. (2025) investigate the impact on the probability of default relating to firms’ transition risk arising from green investments aimed at reducing carbon emissions. They find a moderate Common Equity Tier 1 (CET1) impact in addition to the EU-wide depletion of around 74 basis points, mostly driven by exposure to energy-intensive sectors. Focusing on flood risk, Abbondanza et al. (2025) integrate physical risk from the Nationally Determined Contributions scenario outlined by the Network for Greening the Financial System (NGFS)^[3] into the EBA’s adverse scenario and estimate that banks’ credit risk losses increase by an additional 77 basis points.

Liquidity and contagion risks are investigated by combining findings from the EU-wide market risk stress test with models that capture sectoral interconnections. Using granular data from banks, insurance corporations and investment funds, the ECB Interconnected System-wide stress test Analytics (ISA) tool models how shocks propagate across sectors (Grassi et al., 2025). This approach maps out how financial entities adjust their balance sheets to changes in asset prices and liquidity conditions. It shows that, across sectors, investment funds are more exposed to market swings and contagion losses due to the composition of their portfolios. Overall, banks’ second-round contagion losses amount to 29 basis points over the initial quarter of the scenario.

Contagion effects are further evaluated by linking bottom-up stress tests for different entities or by looking in more detail at overlapping portfolios. Baena and Molitor (2025) use the published results from ESMA’s annual Central Counterparty (CCP) stress test to explore the potential impact of a CCP default on banks’ solvency. The findings suggest that bank solvency is not severely affected in aggregate, but some banks suffer larger losses. In parallel, Baena and Grassi (2025) investigate how correlated defaults can amplify systemic risks during tail events and document how they vary over time and across sectors.

4 Using stress tests to support macroprudential policy

Stress tests are increasingly used to support the setting of macroprudential buffers in the EU. A recent joint report by the ECB and the European Systemic Risk Board (ESRB) listed 17 European Economic Area (EEA) national authorities that use stress-testing approaches to calibrate buffers.^[4] In addition, the Bank of England has for many years combined stress test results with broader macro-financial indicators to guide countercyclical capital buffer (CCyB) decisions, ensuring that the banking system can absorb losses while continuing to support the real economy.^[5]

The constant balance sheet assumption applied in the EU-wide stress tests provides valuable information for macroprudential policy. Under this approach, banks are assumed to keep the size and composition of their balance sheets unchanged over the stress test horizon. This assumption aligns with the macroprudential policy objective of ensuring the continued flow of credit to the real economy during times of stress while preserving the resilience of the banking system.^[6] On this basis, Caccavaio et al. (2025) use the data collected under the last four EU-wide stress tests to complement existing ECB methodologies used to measure the positive neutral CCyB rate.

Dynamic balance sheet stress tests offer a complementary setup to assess risks, back-test buffers and inform prudential policy decisions during crises. A dynamic balance sheet model anticipates how banks will react to a crisis and the consequences of this, allowing policymakers to better quantify banks’ actions and their impact on the economy. It also facilitates the evaluation of policy actions during crises, such as releasing macroprudential buffers or restricting dividend payouts.^[7] In this edition of the Macroprudential Bulletin, Couaillier et al. (2025) show that releasing the CCyB and systemic risk buffer (SyRB) would enable banks to absorb greater losses and provide more credit to the economy.

5 Lessons from the 2025 EU-wide stress test and its extensions

Adding risks beyond those explored in the EU-wide stress test increases capital depletion marginally, which supports authorities’ cautious approach to bank capital buffers. Shocks stemming from climate risk increase losses by up to 0.8 percentage points relative to the 2025 EU-wide results (Chart 1, panel a, yellow bars).^[8] By contrast, relaxing the EBA’s methodological assumptions (Couaillier et al., 2025) results in less adverse impacts, even when accounting for feedback loops from the banking sector to the economy (Chart 1, panel a, last column in left graph and first column in right graph). Total losses from the ISA simulation are not directly comparable because of the different time horizon assumed (Chart 1, panel a, second column in right graph); however, the second-round impact can be added to the official results (Chart 1, panel b). Accounting for additional losses from these extensions, and factoring in offsetting impacts from the relaxation of constraints, results in a slightly higher net depletion that remains below the final outcomes of the 2023 stress test overall (Chart 1, panel b). Many risks (e.g. cyber risks) are not tested by these extensions and the results are based on the same adverse scenario, depending on various models and their underlying assumptions. Overall, these additional results corroborate authorities’ cautious approach to capital buffers.

Incorporating new risks, such as climate risk, can shed light on hidden vulnerabilities. Examining the correlation between bank-level depletion in the EU-wide exercise and the macroprudential stress test simulations (Chart 2, panel a) produces some interesting patterns. Bank losses from models that include risks already captured in the EBA methodology – such as a dynamic balance sheet model – are positively correlated with the losses reported in the EU-wide stress test. Allowing for additional counterparty credit risk or second-round contagion effects confirms patterns of vulnerability aligned with those of the EU-wide stress test (Chart 2, panel a, “CCR” and “ISA 2nd round effects” columns). The weak relationship between climate-related losses and EU-wide losses suggests that the vulnerable banks identified in the EU-wide exercise are not those facing the largest climate-related losses.

While depletion is comparable across exercises, there are differences in the most vulnerable banks and the number of banks breaching the maximum distributable amount (MDA) threshold. Comparing the share of risk-weighted assets held by banks in the bottom 20th percentile of the depletion distribution reveals that smaller banks are more severely affected under the EU-wide stress test (Chart 2, panel b, first bar). In the ISA simulations, it is instead larger banks that are more affected, owing to their larger exposure to NBFI entities, second-round effects and the different time horizons characterising each simulation.^[9] The number of banks breaching the MDA threshold also varies across simulations. It is highest in exercises that incorporate climate losses in addition to the EU-wide losses (Chart 2, panel c). Under the dynamic balance sheet simulations, however, fewer banks breach the MDA threshold, with further reductions observed when available buffers are released (yellow bar).

Chart 1

System-level depletion and net additional losses across different stress test simulations

a) System-level depletion across additional stress test simulations	b) Net losses from additional stress test simulations
(percentages)	(percentages)

Chart 2

Correlations, tail banks and breaches of the MDA threshold

a) Correlations between 2025 EU-wide and macroprudential exercises	b) System-level REA	c) Number of banks breaching the MDA threshold
(correlation coefficient)	(percentages)	(total)

6 Conclusion

The 2025 macroprudential stress test extensions show that adding risks beyond those explored in the EU-wide stress test increases capital depletion marginally, supporting authorities’ cautious approach to bank capital buffers. Through a modular approach, combining top-down and bottom-up inputs, this edition of the Macroprudential Bulletin leverages the extensive, bottom-up data of the EU-wide stress tests to conduct more comprehensive analysis of systemic risk, taking into account new risks such as the interconnections between financial actors. Compared with the 2025 EU-wide results, the extensions lead to a marginal increase in net losses for euro area banks. Differences also emerge regarding the most vulnerable banks, reflecting the underlying transmission channels at play. Incorporating new risks, such as climate risk, brings additional information to the stress test exercise. Furthermore, the extensions only assess a single scenario and a relevant albeit incomplete set of risks, which supports authorities’ cautious approach to capital buffers.

Looking ahead, EU-wide stress tests hold significant potential to support macroprudential policy. The EU-wide scope of these stress tests ensures a consistent framework across jurisdictions, which, in combination with country-specific scenarios, makes them a valuable tool for addressing systemic risks at the national level and enhancing policy coordination at the European level.^[10]