Estimating unobserved individual heterogeneity using pairwise comparisons

We propose a new method for studying environments with unobserved individual heterogeneity. Based on model-implied pairwise inequalities, the classifies individuals in sample into groups defined by discrete heterogeneity unknown support. establish conditions under which are identified and consistently estimated through our method. show that performs well finite samples Monte Carlo simulation. then apply to estimate model of lowest-price procurement auctions bidder heterogeneity, using data from California highway market.