Stress testing at the ECB: mastering risk analyses and impact assessments with top-down models

19 November 2025

By Luis de Guindos, Vice-President of the ECB

Broadening the 2025 EU-wide bottom-up stress test

This year’s EU-wide stress test^[1] showed that banks are resilient under the adverse scenario,^[2] thanks to strong profitability and capital. Nevertheless, the results were obtained using a specific methodology^[3] that does not, by design, capture the full range of shock transmission mechanisms and risks faced by the financial system. The methodology assumes, for example, that banks would keep their lending unchanged over the scenario horizon. As a result, the stress test cannot assess the impact of banks’ deleveraging on the economy as a whole. However, when faced with stress, banks would tend to reduce the size of their loan books and dispose of their riskiest exposures. Similarly, the EU-wide stress test is not designed to capture contagion effects between banks and non-bank financial institutions or the structural impact of climate risks.

This is where top-down stress testing comes in. Building on the results of the EU-wide bottom-up stress test and the granular data collected in that context, top-down stress tests cover a broader scope and add further insights to the original findings. Our 2025 Macroprudential stress test extension report (MaSTER), which we published today, complements the EU-wide stress test results^[4]. By examining how banks and non-banks react to shocks, and by adding new risks, MaSTER quantifies the importance of climate, fire sale and contagion risks. The report also accounts for the additional strains on the macroeconomy caused by deleveraging and de-risking by banks.

Reduced lending and contagion from non-banks

The ECB’s top-down banking sector stress test model finds that banks would noticeably reduce their lending to the economy under the hypothetical adverse scenario, which envisages a long recession, combined with a market slump and sticky inflation triggered by global fragmentation. The contraction in lending would improve banks’ capital ratios to a certain extent compared with a simulation in which they keep their balance sheets unchanged. However, this comes at a cost: the resulting credit squeeze would lead to an additional cumulative real GDP contraction by around 2 percentage points after three years (see the red bar in Chart 1, panel a).

Our analysis also assesses how macroprudential policy can mitigate these effects by releasing available macroprudential buffers (see the green bar in Chart 1, panel a). This offsets the pressure on credit supply from banks’ deleveraging to a certain extent and reduces the negative impact on economic growth. The effect is modest, though, reflecting the fact that releasable buffers are currently relatively limited in size.

Chart 1

Allowing banks to adjust their balance sheets slightly improves capital ratios but triggers credit supply shocks, worsening the downturn

a) Real GDP, deviation from starting point	b) Net losses from additional stress test simulations
(percentage deviation from initial level)	(percentages)

Using the ECB’s top-down system-wide stress test model, MaSTER also builds on the 2025 EU-wide stress test results with regard to contagion risks between investment funds, insurance corporations and banks. It examines the potential short-term contagion effects under the adverse scenario when non-banks adjust their balance sheets in response to stress. This would exacerbate declines in asset prices and market losses.

Among institutional sectors, investment funds – in particular equity funds – face the greatest losses, followed by insurance corporations. For banks, these spillover effects result in limited additional depletion of their Common Equity Tier 1 (CET1) ratios, on average, as most are well hedged against market risks. Therefore, if contagion effects were to materialise, banks’ capital buffers would shrink to a limited extent on average. Banks with less sophisticated hedging capabilities, however, can be affected more significantly. These findings emphasise the importance of a system-wide perspective to capture spillover effects both within and across financial sectors. Furthermore, the results show how solvency-driven liquidity shocks can trigger additional market reactions, which in turn spread through the financial system and amplify the losses stemming from the initial shocks.

Climate-related risks

Top-down models can also be used to assess emerging risks such as transition and physical climate risks not yet captured in the EU-wide stress test. To assess the banking sector’s financial risk from the transition to a green economy, the adverse scenario is expanded using the assumptions underpinning the relevant scenario produced by the Network for Greening the Financial System. This additional source of stress leads to an increase in banks’ projected credit risk losses by around 70 basis points, mostly driven by exposure to energy-intensive sectors. When physical risk caused by flooding is incorporated into the adverse scenario, we calculate an impact on credit risk losses of a similar size.

Overall assessment of banks’ resilience

When we consider the additional losses from all of these stress test extensions and factor in the offsetting impact from banks’ deleveraging and the easing of the European Banking Authority’s methodological constraints, we find that the overall net capital depletion for the EU banking sector is slightly higher. The additional depletion caused under the assumptions in the MaSTER is on aggregate modest, as the total depletion remains below the final outcome of the 2023 stress test (Chart 1, panel b). However, the findings also reveal significant vulnerabilities for specific business models and banks, which experience considerably larger losses when climate or contagion risks are included. The results therefore support the authorities’ cautious approach to capital buffers.

Balancing costs and benefits in stress testing

Bottom-up solvency stress tests for banks are granular and precise. This makes them a crucial element of the overall risk assessment frameworks of banks and authorities alike. At the same time, top-down stress testing is also becoming increasingly important. Central banks are improving their methodologies and use of these granular exercises to support financial stability and prudential policies more efficiently.^[5] One of the strengths of a top-down approach is the ability to deliver broader insights without imposing additional costs on banks or financial entities, beyond those associated with submitting the data required for stress test modelling. This approach also minimises potential biases stemming from tactical behaviour by banks (as Frank Elderson and I discussed in a previous blog post). And it reduces the effort needed to assess the quality of banks’ own bottom-up projections.

Top-down models may come at the expense of some precision compared with bottom-up exercises. Yet, overall, they offer significant advantages in terms of flexibility and coverage. They allow policymakers to test multiple risks and scenarios, including those not yet part of standard stress-testing methodologies. This flexibility is crucial in a rapidly evolving risk landscape, where new vulnerabilities can emerge unexpectedly.

Check out The ECB Blog and subscribe for future posts.

For topics relating to banking supervision, why not have a look at The Supervision Blog?