Independent Randomization Devices and the Elicitation of Ambiguity Averse Preferences

Random incentive mechanisms are an important tool of experimental economics. According to such mechanisms subjects are asked to choose from a variety of problems, but are paid according to only one randomly drawn problem. I show that subjects’ choices in random incentive mechanisms and single choice experiments are identical if they view the randomization device that determines the choice problem as “stochastically independent”. This behavioral notion of stochastic independence is shown to be compatible with the maxmin expected utility model as well as with the smooth ambiguity averse model. This stands in contrast with one of the main streaks of the debate surrounding these two models, which claims that stochastic independence is compatible with the former model but not with the latter. These results can be viewed as a counterpoint to the observation that random incentive mechanisms can only be used to elicit preferences over objective lotteries when these preferences satisfy the independence axiom, the present article shows that this observation does not extend to an environment of subjective uncertainty. Random Incentive (RI-) Mechanisms are widely used in Experimental Economics. In such mechanisms subjects are presented with a variety of choice problems. They are asked ∗Max Planck Institute for Research on Collective Goods, Kurt-Schumacher-Str. 10, 53113 Bonn, Germany. [email protected] †I would like to thank Simone Cerreira-Vioglio, Jürgen Eichberger, Paolo Ghirardato, Christian Kellner, Asen Kochov, Matthias Lang, Massimo Marinacci, and the seminar participants ar the MPI for their comments.