How different are we? Identifying the degree of revealed preference heterogeneity

We propose a nonparametric method to identify the degree of heterogeneity in individual preferences. Using preference information revealed by observed behavior, the method estimates interpersonal preference heterogeneity as the Kemeny distance between individual preference rankings. Using data from the U.S. Panel Study of Income Dynamics (PSID), we derive bounds on the distance-based heterogeneity measure, which we then use to group individuals with similar preferences. We demonstrate that constructing such preference types can substantially strengthen empirical analysis by (i) producing more accurate demand predictions, (ii) improving welfare analysis, and (iii) detecting functional misspecifications in parametric methodologies.