CLIMATE CHANGE IMPACT ON NON-LIFE INSURANCE LIABILITIES: THE RIVER FLOOD CASE

The climate change effects on the frequency and severity of extreme weather events are proving to be a challenge to insurance companies. Since August 2022, the European Insurance and Occupational Pensions Authority (EIOPA) demands the inclusion of the prospective climate risks that the insurers expect to impact their businesses in the Own Risk and Solvency Assessment Report (ORSA) under the Solvency II directive. This study focuses on the assessment of the effects of climate change on the specific event of river flooding and the consequent impact on the property insurance portfolio of an insurance company in mainland Portugal. The approach is divided into two moments: (i) the determination of the probability of occurrence of flooding events under three climate scenarios (RCP 2.6, 4.5 and 8.5) and for three time periods (2022-2032, 2032-2050, 2050-2100), using the public data from the Copernicus program; (ii) and the estimation of the vulnerability to floods for the different geographical areas, using data from the Environmental, Planning, Investigation and Cartography WebGIS freely available spatial data infrastructure. The probability of occurrence is given by the variation of the joint return period of precipitation and river discharge for future scenarios based on historical data. The joint probability is investigated by fitting the Clayton, Frank and Gumbel copulas to the data where the margins follow mainly Gamma, Weibull and Generalized Pareto distributions. The analysis shows an expected increase in the probability of occurrence of floods under the RCP 2.6 scenario while decreasing for the other scenarios. The classification random forest algorithm is applied to explain the vulnerability of an area to floods based on the historically flooded areas and their geographical characteristics. According to the computed Gini Importance, elevation and slope are the most important characteristics. The product of the probabilities of occurrence, vulnerability to floods and the sum insured of the property portfolio constitutes the measure of risk to which each area is exposed. Each area is identified by the first two digits of the zip code. The zip codes in proximity to the main Portuguese rivers are the ones that experience a greater risk of losses.