WP 2020-18 Random Preference Models of Unidimensional and Multidimensional Risk Attitudes

Posted On October 20, 2020
Categories Working Papers, WP 2020
Download Paper

AUTHORS: Morten I. Lau & Hong Il Yoo

ABSTRACT: Apesteguia and Ballester [2018] introduce the class of Ω-ordered lottery pairs that nests popular experimental designs such as multiple price lists. For this class, they show that the Random Preference (RP) model with Expected Utility Theory (EUT) generates choice probabilities which are monotone with respect to a unidimensional risk aversion parameter. We show that given Ω-ordered lottery pairs, RP-EUT models with unidimensional risk aversion parameters are dual to standard discrete choice models. This dual link allows one to use seemingly atheoretical regression models to recover the structural parameters of RP-EUT models that capture unobserved heterogeneity in risk aversion, as well as behavioral noise, across individuals. We also describe a general empirical method that does not require Ω-ordered pairs and EUT. This method allows one to analyze the RP model with theories that allow for multidimensional risk attitudes. We demonstrate its application to the RP model with Rank-Dependent Utility.