The effect of precommitment and past-experience on insurance choices : an experimental study

Thomas Papon, EUREQua

Résumé. Ce papier présente les résultats d'une étude expérimentale consacrée à l'analyse des comportements d'assurance face à des risques de petites probabilités occasionnant de lourdes pertes potentielles. Cette étude révèle que les comportements d'assurance individuels sont susceptibles de dépendre d'une part du vécu des individus face au risque, d'autre part de la durée de la période d'engagement pendant laquelle les individus s'engagent à conserver la même décision d'assurance. Cette étude indique que les modèles de décisions non-additifs tels que Dual Theory ou la Cumulative Prospect Theory disposent d'un pouvoir descriptif supérieur au modèle classique d'espérance d'utilité. Cette étude permet également d'apporter un éclairage nouveau sur le fait stylisé selon lequel les individus se montrent relativement réticents à souscrire une assurance contre des risques de type catastrophique telles que les catastrophes naturelles, alors même que cette assurance leur est proposée à des conditions tarifaires avantageuses.
Mots clés : Demande d'assurance, heuristiques et biais de perception, méthodologie expérimentale, Cumulative Prospect Theory, Dual Theory.

Abstract. This paper reports results from an experimental study that investigates insurance behaviors in low-probability high-loss risk situations. This study reveals that insurance behaviors may depend on the individual prior experience towards risk. It may also depend on the duration of the commitment period, namely the period during which individuals commit themselves to maintain the same insurance decision. Non-additive decision models such as Dual Theory and Cumulative Prospect Theory seem to have a higher descriptive power than Expected Utility Theory when explaining subjects' behaviors. This paper presents a direct experimental test of the prediction of Myopic Prospect Theory relative to insurance demand. This study is also designed to test the significance of gambler's fallacy and availability bias in the insurance decision process. These theoretical concepts help to understand many behaviors commonly observed in reality but which remain unexplained within the E.U framework. In particular, this paper provides new explanations about the puzzling fact that people usually fail to obtain insurance against disaster-type risks such as natural disasters, even when premiums are close to actuarially fair levels. According to our experimental results, the deficiency of insurance demand for natural disasters may be due to the lack of individual prior experience towards such risks ; as well as the relatively short commitment period of insurance policies (usually one fiscal year) compared with the empirical frequency of major natural hazards (centennial and even more).
Keywords : Insurance demand, Low-probability high-consequence risks, heuristics and bias in risk perception, experimental methodology, Cumulative Prospect Theory, Dual Theory.

JEL Classification : C90, C91, D1, D81, D84, G22, M31.