Insurers explore new methods for climate risk modeling
A new WTW report argues that traditional insurance models are failing to keep pace with climate-driven events, urging actuaries to integrate new data sources and scenario-driven analytics. Separately, reinsurance firm SCOR is backing Lancaster University's research into using "prediction markets" to supplement actuarial models for forecasting climate-linked risks. These initiatives reflect a broader industry push for more granular, real-time data to manage perils like floods, wildfires, and windstorms.
- Traditional catastrophe models often fall short because they rely on historical data that no longer reflects the current climate reality, leading to potential underestimation of risks. This is particularly true for "secondary" perils like landslides or hail, which are becoming more frequent and severe due to climate change. - The SCOR-backed CRUCIAL initiative at Lancaster University uses a prediction market platform called AGORA, originally developed by Winton Capital Management, to aggregate expert forecasts on climate-related risks like Atlantic hurricane seasons. This approach aims to synthesize diverse expertise and data into a single, evolving collective forecast. - Parametric insurance is a growing alternative that pays out based on predefined triggers, such as wind speed or rainfall levels, rather than actual losses. This model, projected to become a $34.4 billion market by 2033, is being used for diverse applications from protecting outdoor workers' incomes during heatwaves to safeguarding coral reefs. - Insurers are increasingly leveraging AI and machine learning to analyze diverse datasets, including satellite imagery and weather station readings, to refine underwriting and pricing for climate risks. For example, Swiss Re uses machine learning to improve its flood risk and catastrophe models. - A recent ZestyAI survey of over 200 P&C insurance executives revealed varied adoption of new modeling techniques; traditional actuarial models are still most common for wildfire risk (54%), while stochastic models are preferred for severe convective storms (45%). - For actuaries, integrating new, non-traditional data sources presents significant challenges in data quality, validation, and governance, especially when dealing with "black box" AI models where the decision-making process is opaque. - The mismatch between the short-term focus of most regulatory financial disclosures and the long-term projections of climate models creates challenges in accurately stating risk. Climate models are often more reliable over decades, while regulatory and business planning cycles are much shorter.
