Vulnerability curves for clusters of storms - A case study for Generali France

Winter windstorms are among the costliest natural hazards in Europe, with average annual insured losses estimated at €1.4 billion. In France, they consistently represent the most damaging peril. Estimating windstorm losses remains challenging because they are dominated by rare extreme events and due to the compounded nature of storm activity.

Windstorm losses are typically estimated using vulnerability curves that relate storm intensity to the probability and magnitude of damage. However, windstorms frequently occur in close temporal succession, forming storm clusters. The impacts of such compound events can accumulate, leading to cumulative losses that exceed those associated with isolated storms. While wind-impact vulnerability curves generally perform well, they do not account for the role of storm clustering in shaping damage occurrence and intensity. Improving the representation of clustered storm impacts could therefore refine risk characterisation, enhance loss estimation for both individual and compound events, and increase flexibility in reinsurance design.

Using the portfolio of Generali France as a case study, we investigate the role of storm clustering in wind-related insurance losses. Losses are first associated with individual storm tracks, and storm clusters are defined as sequences of damaging events separated by less than 96 hours. Our results indicate that approximately 85% of insured windstorm losses in France are attributable to clustered storms.

Building on these findings, we develop vulnerability curves for residential properties that explicitly account for temporally compounded storm events. These curves provide a more realistic representation of windstorm risk than traditional approaches, which typically assess losses either at the scale of individual storms or over an entire winter season. Our results highlight the importance of treating storm clusters as combinations of interdependent events.