Elderson urges banks to prepare for risks models cannot capture
Frank Elderson urged banks to build resilience beyond climate risk models at the European Central Bank (ECB)-Bank of England conference in Frankfurt on October 14, 2026. The event brought together experts from 18 countries and three continents.
Capital planning leaves risks uncovered
Elderson, ECB Executive Board member and Supervisory Board Vice-Chair, said climate stress tests and scenario exercises cover all risk drivers banks themselves deem material in only around 15 percent of cases.
Banks often use these exercises to calibrate buffers in their internal capital adequacy assessment process while integration into core models remains incomplete.
Since 2022, significant supervised banks have consistently earned, on average, more than 70 percent of interest income from carbon-intensive sectors.
Elderson urged banks to challenge scenarios, tailor them locally and supplement them with expert judgement.
He called for sufficient capital, liquidity, operational capacity and strategic flexibility to absorb shocks that cannot be predicted.
Climate feedback escapes the models
Elderson distinguished measurable risks, known uncertainties and outcomes not yet identified.
Seven of nine planetary boundaries have been breached; the Arctic is warming almost four times and Europe roughly twice the global average.
Natural emissions from thawing permafrost or wildfires could add up to 0.6°C in an adverse scenario, he noted.
He cautioned against relying narrowly on Network for Greening the Financial System (NGFS) scenarios and extended the resilience approach to demographic, geopolitical, artificial intelligence and quantum-computing risks.
Resilience beyond model precision
Elderson's message is that better measurement alone cannot address radical uncertainty.
He called for banks to test governance and business-model resilience, alongside underwriting, pricing and portfolio decisions.
The ECB's 2026-28 supervisory priorities treat uncertainty as a challenge spanning multiple risk areas.