The coverage of extreme risks of natural disasters: theoretical and empirical analysis.

Summary The objective of this thesis is to reflect on the systems of coverage of the risks of natural disasters, with particular emphasis on the perception that economic agents have of the characteristics of these risks. The aim is to better understand the determinants of existing coverage systems in the world and of individual behavior, and to compare the different forms of public intervention in the face of these risks. Our work, which is essentially empirical, is based on statistical data from international organizations and risk research centers, but also on behavioral data from a controlled experiment. The typology of flood risk insurance systems in the world that we develop shows that the public system is globally dominant and mainly applied in low-income countries with high exposure to risk, while the mixed public/private system is mainly applied in rich countries with relatively low exposure. Regarding the demand for insurance, the willingness to pay for insurance against natural hazards is significantly lower than that for other hazards, regardless of the probability and amount of loss. By comparing different systems of public intervention, we show that the most effective measure seems to be a subsidy of the insurance premium, but that this can be very costly if the population is predominantly composed of individuals with strong biases in risk perception.