Betting on Machina's reflection example: an experiment on ambiguity

In a recent paper, Machina (2008) suggested choice problems in the spirit of Ellsberg (1961) which challenge tail-separability, an implication of Choquet Expected Utility (CEU) to a similar extent as the Ellsberg paradox challenged the sure-thing principle implied by Subjective Expected Utility (SEU). We have tested choice behavior for bets on one of Machina’s choice problems, the reflection example. Our results indicate that tail-separability is violated by a large majority of subjects (over 70% of the sample). These empirical findings complement the theoretical analysis of Machina (2008) and, together, they confirm the need for new approaches in the analysis of ambiguity for decision making.