Scenario-based CO2 thresholds for strategic asset allocation and climate stress testing in insurance

Abstract Climate change poses a dual threat to the insurance industry, affecting both assets (transition risk) and liabilities (physical risk). This study advances the 2024 EU-wide climate stress test by addressing its exclusion of reactive management actions. To bridge this gap, it develops a framework for proactive asset allocation, identifying CO 2 concentration thresholds that trigger divestment from corporate bonds and equities exposed to transition risks. Using a model based on climate-induced financial risks for insurers, we optimize divestment thresholds to balance risk and profit until 2030. The model incorporates stochastic projections of flood and cyclone claims, linked to CO 2 concentration trends with scenario-dependent distributions and correlations from NGFS forecasts. Crucially, our analysis demonstrates that remaining invested in brown assets beyond a CO 2 concentration of 439 ppm is no longer favorable from a risk or return perspective. Consequently, the identified, scenario-specific thresholds serve as a forward-looking basis for strategic asset allocation, outperforming static strategies by mitigating extreme tail risks in disorderly transitions.