WP 2020-18 Primal and Dual Methods for Estimating Random Preference Models of Risk Attitudes

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

AUTHORS: Morten I. Lau & Hong Il Yoo

ABSTRACT: We present two methods for estimating Random Preference (RP) models of risk attitudes. Our first method is intended for any class of RP models, including those addressing multidimensional risk attitudes as per non-expected utility theories. Existing methods cannot be used to estimate such models without imposing severe constraints on the model and data. We present a fully versatile alternative. Our second method is intended for models of unidimensional risk attitudes that display single-crossing properties. We show that these models are dual to seemingly non-structural regression models. One can thus use standard regression models in structural estimation of unidimensional risk attitudes.