We estimate means and distributions of ex-ante treatment effects for obtaining university education relative to high school. To achieve this, we conducted a survey which elicited earnings expectations associated with counterfactual educational choices for a sample of highschool students in...