Randomization Devices and the Elicitation of Ambiguity Averse Preferences

Economics

Speaker:	Sophie Bade, Max Planck Institute for Collective Goods
Date:	Friday 14 December 2012
Time:	16.15
Location:	Matrix Lecture Theatre, Building One

Further details

Random Incentive (RI) mechanisms are widely used to obtain the responses of experimental subjects from a large variety of choice sets. Subjects are paid only according to one choice problem that is randomly selected after the subjects have reported their choices for all problems. If subjects choices in the separate problems are identical to their choices in the RI-mechanism at large, then the RI-mechanism satises isolation. When agents are expected utility maximizers, isolation holds if and only if independent randomization devices select which problem leads to actual payment. This paper investigates whether the RI-mechanisms that have been used in the literature to elicit ambiguity averse preferences satisfy isolation and thereby uncover true preferences. I show that subjects can use the randomization device of an RI-mechanism as a hedging device, which will contaminate the data. Under suf- ciently strong independence conditions RI-mechanisms will satisfy isolation. But these conditions are so stringent that the randomization device can no longer be mutually independent of the preexisting uncertainty facing the agent.