Sharp Bounds in the Binary Roy Model

We derive the empirical content of an instrumental variables model of sectoral choice with discrete outcomes. The partial identification results extend existing work on sharp bounds in binary choice threshold crossing models in allowing sector specific unobserved heterogeneity. Assumptions on selection include the simple, extended and generalized Roy models. The derived bounds are nonparametric intersection bounds and are simple enough to lend themselves to existing inference methods. Identification implications of exclusion restrictions are also derived. The derived bounds are applied to the analysis of the effect of Swan-Ganz catheterization and the robustness of previous findings to the introduction of procedure-specific unobserved heterogeneity is examined.