Author(s): Hauser, Andrea
Date: 2021
Persistent ID: http://hdl.handle.net/10400.5/23306
Origin: Repositório da Universidade de Lisboa
Subject(s): Property Insurance; Tropical Cyclones; Claims; Cost; Regression Tree; Random Forest; Logistic Regression; Multiple Linear Regression; Risk Map; Seguro Multi-Risco; Ciclones Tropicais; Sinistros; Custos; Arvore de Regressão; Floresta aleatória; Regressão Logística; Regressão Linear Multipla; Mapa de Risco
Tropical cyclones have enormous destructive potential. In 2018 continental Por- tugal has been affected by hurricane Leslie, the weather-related event having the highest impact ever on the property portfolio of the portuguese insurance company Fidelidade, causing several millions euros of losses. The fear is that, in the near future, the occurrence of this type of events increases in intensity and frequency, as a consequence of the climate change due to the warming of the planet. Quantifying the potential loss to which the property portfolio of Fidelidade could be subject to, helps in approximately determining premiums and capital reserves, as well as in defining the coverage to be provided. In this work, an approach to model the costs caused by a tropical cyclone extreme event is presented. The model is based on the losses incurred by the property port- folio of Fidelidade due to hurricane Leslie. By using the estimated models, it is possible to produce cost estimates for different scenarios of interest for the com- pany. The estimated models are also used to build a risk map for the councils of continental Portugal. The results obtained indicate that the councils with the estimated higher average cost ratio are all located along the coast of the country.
