Frank Elderson, Member of the Executive Board of the European Central Bank and Vice-Chair of the Supervisory Board of the ECB, delivered a keynote address at the joint ECB-Bank of England conference in Frankfurt am Main on 14 October 2026. The conference focused on supervising climate risk in a non-linear world, bringing together experts across disciplines from 18 countries and three continents.
Modern financial risk management has traditionally relied on the assumption that sufficient data, historical patterns, and robust models allow banks to estimate losses with precision. These estimates guide how institutions price risk, set aside provisions, and determine capital requirements. While this approach functions effectively when underlying relationships remain stable and is supplemented by stress testing and model overlays, the modern economic environment has become increasingly complex, interconnected, and uncertain.
Macro trends such as geopolitical fragmentation, demographic shifts, artificial intelligence, quantum computing, and the ongoing climate and nature crises differ in origin, scope, and speed. However, they share a common denominator by arising within complex systems that exhibit non-linear behaviour. Such systems allow small changes to trigger disproportionately large effects, amplify shocks through feedback loops, and produce abrupt transitions through tipping points. Addressing these non-linearities remains manageable within the realm of the measurable, but deeper challenges emerge in a state of radical uncertainty, where future events and probabilities cannot be measured with confidence.
Elderson applied a simple framework categorising challenges into known knowns, known unknowns, and unknown unknowns. Regarding the known knowns of climate and nature risks, banks and supervisors have made substantial progress over the past decade in identifying, quantifying, and managing these factors. Improved tools, data, and modelling now allow for better quantification of direct economic damage, such as extreme weather impacts on economic activity, labour productivity, and supply chains. Banks can also utilise geo-location data to model physical hazards on traditional financial metrics like loss given default.
Data indicates that the average share of interest income earned from carbon-intensive sectors by significant supervised banks has remained consistently above 70% since 2022. Furthermore, institutions possess metrics on the average energy performance of financed properties and exposures vulnerable to physical risks. Nevertheless, this progress only covers part of the spectrum, as many forward-looking climate risks differ entirely from historical precedents.
The second category encompasses known unknowns, representing risks whose existence is recognised but whose timing, magnitude, or probabilities remain uncertain. Economic forecasting and climate analysis both deal with complex adaptive systems where precise forecasts are impossible, though plausible scenarios can be explored. Specific examples include climate tipping points, such as the potential dieback of the Amazon rainforest or the collapse of the Atlantic Meridional Overturning Circulation, as well as compound events where severe droughts, heatwaves, water scarcity, and agricultural disruptions interact simultaneously.
Additionally, humanity has transgressed seven of the nine planetary boundaries regarding nature degradation. The interaction and financial system propagation of these forces remain uncertain because future impacts are not reflected in historical data. Scenario analysis serves as an indispensable tool in this environment, not as a forecast, but as a structured exploration of plausible futures. Climate scenario analysis has advanced considerably, with Network for Greening the Financial System scenarios acting as a global reference point.
