Probabilistic simulation tool estimating the potential losses of rare catastrophic events from simulated scenarios, combining a hazard module, an exposure module, a vulnerability module and a financial module.
A catastrophe model is a probabilistic tool that estimates the potential losses linked to rare, high-severity events, where the loss history is too sparse to serve as a pricing basis. Rather than relying on past claims, the model simulates a large number of possible events, combining four sequential modules. The hazard module generates a catalog of thousands of simulated events (earthquakes, hurricanes, floods) with their frequency and spatial intensity. The exposure module locates insured properties with their characteristics (construction, value, footprint). The vulnerability module estimates the expected damage ratio for each property type as a function of hazard intensity, using fragility functions. The financial module applies the policy terms (deductibles, limits, co-insurance) to translate damage into insured loss. The main output is a loss exceedance curve (OEP/AEP), which associates with each loss level an annual probability of being exceeded, from which indicators such as the average annual loss (AAL), the 100- or 200-year loss, or the probable maximum loss (PML) are derived. The three dominant vendors are AIR Worldwide (Verisk), RMS (Moody's) and PERILS. These models are indispensable for pricing high catastrophe reinsurance layers, managing accumulation, structuring cat bonds and computing the catastrophe module under Solvency II. Their essential limitation is that they are only as good as their assumptions, competing models being able to diverge materially, and their calibration on stationary historical data is increasingly inadequate in the face of climate non-stationarity. In cyber, catastrophe modeling is in its infancy, because the hazard is dynamic and vulnerability depends on human and technical behavior in constant evolution.
A reinsurer submits its global property portfolio to an AIR cat model for the European windstorm peril. The model produces an average annual loss of 180 million euros and a 200-year PML of 1.4 billion, figures that form the basis for pricing its treaties, sizing its capital charge and structuring a cat bond.
cat model, modèle cat, modélisation catastrophe, AIR, RMS, Verisk, modèle probabiliste catastrophe