Extreme risk aversion: an economic approach based on the rank-dependent expected utility model.

Summary The thesis aims at characterizing the aversion of individuals to major risks. This characterization is based on the notion of willingness to pay (to avoid this type of risk) and mobilizes Quiggin's (1982) expected utility (EU) and rank-dependent expected utility (RDEU) models. These two models allow for an identical treatment of monetary consequences, but differ in their treatment of probabilities. In the context of major risks, this results under UEDR in a potentially very large overvaluation of very small probabilities. Based on a method of approximating willingness-to-pay that is well suited to the major hazard framework, we show that the willingness-to-pay of a UEDR individual can be substantially higher than that of an EU individual. In particular, the magnitude of this difference in willingness to pay is strictly equivalent to the magnitude of the subjective overvaluation of very small objective probabilities of loss. In addition to this theoretical result, the thesis conducts an experimental investigation, using the Tradeoff elicitation method (Deneffe and Wakker, 1996), which confirms the standard result of the UEDR model that individuals overweight very small probabilities. The experiment also shows that the smaller the objective probability, the greater the overweight. Finally, based on the theoretical and experimental results of the thesis, we evaluate the case of the subjective cost of a major risk, in particular of a serious nuclear accident. Our results show that this cost is reflected in consents to pay that are much higher under UEDR than under UE. These differences in willingness-to-pay between the two models clearly show the impact of the magnitude of the overweighting of very small probabilities on the characterization of individuals' behavior towards major risks.