Preference for randomization and validity of random incentive system under ambiguity : an experiment

The random Incentive System (RIS) is a standard method to incentivize participants in economic experiments. However, recent theoretical studies point out the possibility of its failure under ambiguity. We propose a modification of RIS, named independent RIS (I-RIS), to improve its reliability. We conducted an experiment to evaluate the performances of the standard RIS and I-RIS in direct and indirect manners. Whereas a nonnegligible fraction of participants are not consistent with the reversal-of-order axiom, a majority of ambiguity averse and seeking participants are. This implies that participants with nonneutral ambiguity attitudes may not report truthful preferences when RIS is used. However, randomization attitudes do not explain inconsistent choices under RIS. In addition, we did not find significant differences in performance between RIS and I-RIS. These results suggest that preferences for randomization, which is driven by nonneutral ambiguity attitudes, do not cause the failure of RIS.