WP 2020-18 A General Approach to Estimating Random Preference Models of Risk Attitudes

AUTHORS: Morten I. Lau & Hong Il Yoo

ABSTRACT: The random preference (RP) model provides an integral framework for modeling within-individual heterogeneity in choice behavior, by attributing this heterogeneity to preference parameters in the underlying theory of risk attitudes instead of an additive error term that is external to the theory. However, most empirical studies in structural estimation of risk attitudes turn to additive error specifications because the RP likelihood function is computationally unattractive. We propose a general approach to estimating the RP model that facilitates empirical applications in this alternative modeling framework. Our estimation approach illustrates that the RP model is just as flexible as other stochastic choice models. By applying a kernel smoothing procedure, we can construct a versatile likelihood evaluator of the RP model that can accommodate any decision theory, types of lottery pairs, and parametric distribution of unobserved heterogeneity.