Probability, Expected Utility, and the Ellsberg Paradox

Abstract

The Ellsberg paradox is often cited as evidence for unknowable "ambiguity" versus computable "risk", and a refutation of the Savage axioms regarding expected utility maximization and the program for revealing "subjective" or "belief-type" probabilities. This note argues that researchers have been too quick to embrace the Ellsberg critique as a refutation of standard expected utility theory. First, Ellsberg performed no actual experiments, and in fact recent empirical evidence on the Ellsberg paradox argues against ambiguity. Second, simple explanations for the paradox deserve as much attention as theories that introduce a new concept such as "ambiguity." One such simple explanation is to consider the Ellsberg thought experiment as part of a "meta-experiment" that includes repeated draws, in which case the choices described by Ellsberg are consistent with expected utility theory.

Keywords: Probability, Expected Utility, Risk, Ambiguity

JEL Classification: D8, D81