Modeling Preference Heterogeneity Within and Across Behavioral Types: Evidence from a Real-world Betting Market

Abstract

While it is commonly accepted that risk preferences differ across individuals, studies that estimate them typically allow for limited heterogeneity. We develop a methodology that allows for richer representation of heterogeneity both within and across utility types characterized by different behavioral features. This enables us to improve individual- and population-level estimates, and to assess the relative importance of loss aversion and probability weighting, and their prevalence in the population. Applying our model to individual sports-betting choices, we find that utility curvature alone does not explain observed choices and, while two-thirds of individuals exhibit loss aversion, all exhibit probability weighting.