Estimating marginal treatment effects under unobserved group heterogeneity

Abstract This article studies the treatment effect models in which individuals are classified into unobserved groups based on heterogeneous rules. By using a finite mixture approach, we propose marginal (MTE) framework choice and outcome equations can be across groups. Under availability of instrumental variables specific to each group, show that MTE for group separately identified. On basis our identification result, two-step semiparametric procedure estimating group-wise MTE. We illustrate usefulness proposed method with an application economic returns college education.