Is Ellsberg Behavior Evidence of Ambiguity Aversion?

Abstract

We perform two types of lab experiments to assess the normative and positive appeal of preference models exhibiting ambiguity aversion. Our first experiment is a simple extension of the Ellsberg [1961] two-color urn experiment in which there is an option that hedges ambiguity away completely and that dominates the options that correspond to Ellsberg behavior. 63% of subjects choose the dominated Ellsberg options, which compares similarly to the proportion of subjects choosing the risky urn in the standard two-color experiment. While subjective expected utility cannot explain this choice, also none of the classical models of ambiguity aversion can explain this choice. Our second experiment is also based on the Ellsberg two-color urn experiment. In this experiment, in various treatments, we provide advice in the form of short video clips in favor of, as well as against, the Ellsberg choice. We find suggestive, but not conclusive, evidence that subjects choices are influenced by advice and, in these cases, mostly in the direction of abandoning the Ellsberg option.