Résumé en anglais

This article models a situation in which a monopolistic insurer evaluates risk better than its customers. The resulting equilibrium allocations are compared to the consequences of the standard adverse selection hypothesis. On the positive side, they exhibit the property that low-risk people are better covered than higher-risk people. On the normative side, the article shows that there are two reasons for avoiding excessive risk classification: one is the classical destruction of insurance possibilities, and the other comes from the distrustful atmosphere generated by new asymmetric information.