Measuring regret theory in the health and financial domain

This paper applies an experimental design developed by Bleichrodt et al. (2010) to test the key assumption of original regret theory (Loomes and Sugden, 1982): convexity of the regret function. We elicit preferences for financial and health outcomes for about 1,000 subjects, yielding some evidence of minor dierences between financial domain and health domain. While aggregate results seem to support regret theory at first sight, individual-level analyses show that the majority of subjects violate the predictions of regret theory with a convex regret function. Our results thus challenge the predictive accuracy of regret theory as a descriptive theory of decision-making under risk.