Damage modelling, the Catastrophe risk modelling structure for insured damages

Description

A catastrophe model is the instrument the (re)insurance industry uses to relate a hazard to the losses it causes. It has three units. The hazard unit models the physical event — for CCR, a two-dimensional distributed overflow and runoff model at 25 metre resolution, calibrated at watershed scale across the whole of France and adapted to the demonstration catchments by refining the terrain slopes so that water spreads realistically along short watercourses. The vulnerability unit holds the insured contracts, geolocated to the address, with their type of risk and insured value. The damage unit merges the two through damage functions that relate water depth or flow to a destruction rate, obtained by dividing the amount of claims by the insured value.

The destruction-rate approach is what allows the model to be transferred. Where no locally calibrated damage curve exists for runoff or overflow, the observed claims are used to derive one; where they exist, the model is validated against floodmarks and observed flood extent from the events concerned. Insured losses can then be estimated for the event set, and — coupling the flood impact model to a global climate model — for a changed climate.

Applied within NAIAD by CCR with researchers from INRAE, BRGM and the Université Côte d'Azur, the model was used to quantify the damage avoided by preventive measures, including Nature Based Solutions, in the French demonstration catchments.

Applications to decision making in the insurance sector

This is the sector's own appraisal instrument, used for the question the sector actually asks: does the measure reduce the losses we would otherwise pay? Quantifying avoided damage in the same framework that prices the risk means the answer is expressed in a form the industry can act on, rather than in a parallel set of units that then has to be argued across.

The model also supports the analytical work a carrier or reinsurer may wish to do on its own portfolio: which of its exposures would be reduced by an intervention, and by how much.

Added value of this modelling approach

It starts from observed claims rather than from assumed values, so the result is anchored to losses that actually occurred. And because the same structure serves both the current climate and a projected one, the effect of climate change on the benefit of a measure can be quantified rather than described.

What decision can be improved with this modelling approach

Whether a preventive measure warrants recognition in pricing or in the terms of cover; how much of the risk would remain after the measure; and how the balance between risk reduction and risk transfer should shift as the hazard changes.