ECB’s Elderson urges banks to prepare for unmodellable shocks

Frank Elderson urged banks on October 14 to strengthen their capacity to withstand unexpected shocks, warning that climate and nature-related risks can exceed what historical data, financial models and standard scenarios are able to capture.

Opening a joint ECB–Bank of England conference, he highlighted a gap in banks’ capital planning: climate stress tests and scenario exercises used in internal capital adequacy assessments covered all the risk drivers that banks themselves considered material in only around 15% of cases.

The European Central Bank published his remarks as “Ensuring resilience in the face of uncertainty” on October 14, 2026. The speech addressed current risk-management practices and the 2026–28 supervisory priorities, rather than announcing a new numerical capital requirement. This account is based on that published address: https://www.ecb.europa.eu//press/key/date/2026/html/ecb.sp261014~e4ca33588e.en.html

Better data, but incomplete coverage

Elderson acknowledged substantial progress over the past decade in identifying and measuring climate and nature risks. More detailed hazard and location data, he said, now help banks estimate how flooding and drought could affect financial measures such as loss given default—the portion of an exposure a bank loses when a borrower defaults.

Banks also have better information on the energy performance of financed properties, lending in areas exposed to physical hazards and deteriorating credit quality within those exposures. However, he cautioned that many assessments still depend on estimates.

One measure underscored the continuing importance of transition risk: according to Elderson, the average share of interest income earned from carbon-intensive sectors by significant supervised banks has remained above 70% since 2022. That is an income measure, not a statement that those sectors account for the same share of bank assets or expected losses.

The speech’s supporting notes described uneven coverage across portfolios and risk categories. Some banks address transition risks without adequately covering physical risks, or the reverse. Operational and market-risk channels also receive less attention than credit risk.

Scenarios are explorations, not predictions

Elderson organised the challenge into three categories: risks that can already be identified and quantified; recognised threats whose timing or scale remains uncertain; and outcomes or interactions that are not yet understood.

The second category includes climate tipping points and overlapping shocks. Drought can coincide with extreme heat, water shortages can disrupt agriculture, and physical damage can occur alongside stress on energy systems. Historical observations alone cannot fully describe how these combinations might affect borrowers and banks.

He described scenarios as structured ways to examine plausible futures, rather than forecasts. Their purpose is to expose vulnerabilities, challenge assumptions and trace how shocks could travel through economies and financial institutions.

The Network for Greening the Financial System’s publicly accessible scenarios have improved consistency and comparability across jurisdictions, he said. But banks should not allow a small set of standard narratives to narrow their search for threats. They need to examine extreme but plausible conditions and adapt global scenarios to local exposures.

What models leave out matters

Elderson pointed to thawing permafrost as an example of the gap between scientific understanding and model coverage. Citing the Intergovernmental Panel on Climate Change, he said further warming would release additional carbon through thawing, but most global climate models do not directly represent that feedback.

He added that the separate adjustment used in the latest assessment still excluded abrupt thaw and interactions with Arctic wildfires. Research cited in the speech suggested that natural emissions from processes including permafrost thaw or wildfires could add as much as 0.6°C in an adverse scenario. This was a conditional research estimate, not a forecast of an inevitable temperature increase.

His broader warning was that a model’s inability to quantify a mechanism does not establish that the mechanism is financially insignificant. Local effects can also differ sharply from global averages, making expert judgement and exposure-specific analysis important complements to modelling.

“Scenarios must be applied, not just adopted,” Elderson said. He called for regular updates and closer cooperation among supervisors, banks and researchers.

Preparing balance sheets and decisions before a shock

For genuinely unforeseen outcomes, Elderson argued that adding more scenarios is not enough. Banks need sufficient capital and liquidity, resilient operations and the flexibility to change strategy when conditions move beyond their assumptions.

He also urged banks to examine the wider tools available to them: underwriting standards, adjustments to model outputs, risk appetite, portfolio choices and pricing. Insurance protection must be assessed alongside governments’ more limited fiscal capacity to act as insurers of last resort.

Adaptation and mitigation investment formed another part of his proposed response. Banks should test not only potential losses but also whether their governance and decision-making arrangements can respond quickly when risks change. He linked that preparedness to prudential transition planning.

The warning extends beyond climate

Elderson said the same approach applies to demographic change, geopolitical risks and technological disruption. He linked those challenges to European banking supervision’s 2026–28 priorities, which treat heightened uncertainty as a concern across risk areas.

Artificial intelligence is already changing cyber and operational risks, he said, even though the eventual scale is uncertain. Quantum computing presents a different timing problem: encrypted information can be collected now in anticipation that future technology will make it readable.

The conference brought together specialists from 18 countries and three continents. Elderson’s central request was for banks to retain rigorous measurement while building safeguards for shocks that measurement cannot yet adequately describe.