Mean-preserving capacities: A tractable class of Choquet capacities

Abstract This paper introduces mean-preserving capacities for Choquet expected utility. A capacity is defined as mean-preserving when the value assigned to an event falls within an interval centered around its subjective probability, with the interval’s size reflecting the perceived level of ambiguity. These capacities are based on the complementary independence axiom and feature a centrally symmetric core for decision-makers who are averse to ambiguity. To illustrate their practical value, we explore a series of economic applications including portfolio choice, self-insurance and self-protection, the value of a statistical life, and precautionary saving. Mean-preserving capacities offer intuitive predictions for both ambiguity-averse and ambiguity-loving individuals, often requiring fewer or no ancillary assumptions than alternative models. Moreover, they simplify the analysis of greater ambiguity considerably.