European Central Bank reverse stress test exposes gap between capital and liquidity models

European Central Bank reverse stress test exposes gap between capital and liquidity models
Author
Written by Russell Pereira
  • Aug 13, 2026
  • 7 min

The European Central Bank's geopolitical reverse stress test exposed a weakness beyond scenario design: some banks' capital losses still do not translate convincingly into funding and liquidity stress.

The European Central Bank (ECB) asked 110 significant euro area banks to do something unusual in its 2026 geopolitical reverse stress test. Each bank had to start with a severe outcome, a loss of at least 300 basis points from its common equity tier one capital ratio, then work backwards to design the geopolitical crisis that could cause it. The test exposed a problem beyond geopolitics. Several banks could produce the required capital loss without showing a similarly severe deterioration in funding and liquidity, suggesting that some stress-testing frameworks still struggle to describe how one shock spreads across the balance sheet. That matters because real bank stress does not arrive in separate capital and liquidity compartments. A sharp capital loss can undermine confidence, raise wholesale funding costs, move deposits, weaken market access and force asset sales. The ECB's reverse stress test therefore examined whether banks could model the whole path from an initial shock to financial distress, rather than simply calculate a sufficiently large capital loss.

The ECB made banks work backwards from failure

A conventional stress test gives banks a common economic scenario and asks them to calculate the damage. A reverse stress test starts at the other end: the bank is given a damaging outcome and must identify the combination of events that could produce it. The ECB used that approach with a minimum target of 300 basis points of common equity tier one capital depletion, while around 20% of participating banks voluntarily chose a more severe target. Banks could assume changes in monetary policy but could not depend on fiscal rescue measures that had not already been approved.

Most banks produced institution-specific geopolitical scenarios. Roughly a quarter modelled a Middle East conflict that closed the Strait of Hormuz, while cyberattacks also featured heavily. Eighty-six banks identified cyberattacks as a relevant risk and 57 treated them as the main source of operational disruption. Reverse stress testing itself is not new. The United Kingdom's Prudential Regulation Authority has required it for more than a decade, while the ECB's 2018 Internal Capital Adequacy Assessment Process (ICAAP) Guide already identified it as part of banks' expected stress-testing capabilities. An ECB review in 2020 found that reverse stress testing was often generic and lacked analytical depth.

What changed in 2026 was the severity and reach of the ECB's test. All 110 participating banks had to identify their own route to a fixed and substantial capital loss, turning reverse stress testing into a direct examination of where each institution believed its own vulnerabilities would emerge. The test's clearest modelling gap appeared in whether that capital shock was carried through into funding and liquidity.

Capital and liquidity do not always tell the same story

Several banks modelled severe capital depletion while showing little corresponding deterioration in liquidity or foreign-currency funding metrics. The ECB concluded that interactions between solvency and liquidity were generally not well captured in many banks' stress-testing frameworks. The finding needs some qualification: many banks did model reasonable effects on liquidity and funding. The problem was not that the sector ignored liquidity altogether, but that some banks produced capital and liquidity outcomes that did not appear consistent with the severity of the same underlying shock.

That inconsistency matters because a bank suffering a large capital loss is unlikely to experience that loss in isolation. Investors, depositors and counterparties respond to deteriorating financial strength. Wholesale funding can become more expensive or harder to obtain, deposits can move, collateral needs can change and assets that management expected to sell gradually may have to be sold quickly into weaker markets. A credible stress model therefore has to answer more than how much capital disappears. It also has to show what happens to the bank's ability to fund itself as confidence deteriorates.

The ECB found another sign of the same problem. Some banks assumed that their balance sheets would continue growing and revenues would remain resilient while the scenario was destroying 300 basis points of capital. Those optimistic assumptions and the weak liquidity response may point to a common modelling weakness: different consequences of the same crisis are not always being carried through one internally consistent set of assumptions. The findings suggest banks may increasingly need to demonstrate that ICAAP and the Internal Liquidity Adequacy Assessment Process (ILAAP) produce consistent capital and liquidity outcomes when run against the same stress. That consistency test also extends to the actions banks assume they can take once the stress materialises.

Banks cannot all use the same escape route

The banks relied on several common measures to protect their capital positions. About 40% cited portfolio sales or asset disposals, 39% cited deposit or loan repricing and 59% cited suspended dividends or reduced distributions. Each action can make sense for one institution, but its credibility changes when many banks try to do the same thing at the same time. If a large part of the banking system tries to sell similar assets during the same crisis, there may not be enough buyers at the prices assumed in individual plans. If several banks compete aggressively for the same deposits, funding becomes more expensive. A response that appears reasonable in one institution's model can therefore become much less effective once the rest of the system is also under stress.

The ECB refers to this problem as a crowding effect. For capital planning teams, the implication is practical: a management action cannot be treated as reliable simply because the bank has the operational ability to execute it. The bank also has to consider whether the market capacity needed to support that action will still exist during a system-wide shock. An asset sale requires a buyer and a deposit campaign requires funding to move from somewhere else. Both become harder to rely on when many competitors are trying to protect their balance sheets at the same time. Those assumptions turn the credibility of a stress model into a governance issue.

Supervisors are testing the reasoning behind the number

The ECB has been clear that the reverse stress test will not mechanically recalibrate Pillar 2 Guidance, but the findings can still matter to supervision. They will feed qualitatively into the Supervisory Review and Evaluation Process (SREP), through which the ECB assesses a bank's governance, risk management and internal capital planning. The supervisory question is therefore broader than whether a bank's model produces an acceptable capital ratio. Supervisors also need to judge whether management can defend the assumptions that produced that ratio.

A July 2026 update to the ECB's ICAAP Guide reinforced that management buffers remain the responsibility of the bank rather than an additional supervisory buffer. Revised European Banking Authority SREP guidelines published in June also placed greater emphasis on forward-looking and risk-focused assessment. Taken together, those developments suggest the burden of proof is shifting from producing a capital number to showing why that number remains credible when funding, liquidity and management actions are stressed at the same time. The next supervisory test is therefore likely to focus less on whether banks can design a more severe scenario and more on whether their capital, liquidity and management-action assumptions remain consistent under the same shock.

The next constraint is model consistency

For banks, the execution challenge is whether capital, liquidity and management-action models can produce one coherent account of the same crisis. That requires more than stronger scenario design. It requires assumptions, data and governance processes that reconcile across ICAAP, ILAAP and the actions management expects to take under stress.

The decisive test is whether management can demonstrate one credible path from the initial shock through capital loss, funding pressure and liquidity deterioration to the actions used to restore resilience. A better scenario cannot solve an inconsistent model. The bank has to show that the whole failure pathway holds together.


Conference Insights

More
How ICAAP is changing the way ASEAN banks think about capital

How ICAAP is changing the way ASEAN banks think about capital

ICAAP is becoming the gatekeeper for ASEAN bank lending

ICAAP is becoming the gatekeeper for ASEAN bank lending

Banks are adopting AI before they can fully see the risks

Banks are adopting AI before they can fully see the risks

AI will expose banks that have not fixed the foundations

AI will expose banks that have not fixed the foundations

Banks must fix their operating model before agentic AI can scale

Banks must fix their operating model before agentic AI can scale

Comments (0)
Cancel
Suggested Users
Chat with us WhatsApp