Healthy scepticism as an expected-utility explanation of the phenomena of Allais and Ellsberg

Abstract

Expected utility theory is a normative approach to the question of how people should rationally make choice under uncertainty. Its admirers (the present author among them) would use it in every such situation. Consequently alledged counter examples, in which it is arguably unreasonable to act as expected utility theory suggests, threaten to undermine the theory as a normative theory.

The phenomena of Allais and Ellsberg were proposed as just such examples. They differ from the later work of Kahneman, Tversky and others in that the latter exposes ways in which people's behavior, descriptively, differs from the expected utility theory norm. Those examples do not threaten the normative status of expected utility theory.

This paper reviews the Allais and Ellsberg examples, and proposes an explanation of them within expected utility theory. The key to the explanation is that the expectation in expected utility theory is to be taken with respect to the subjective beliefs of the decision maker. If I have positive probability of strategic behavior by the person making the offers, neither phenomenon is a paradox.