Instrumental Variables for Binary Treatments with Heterogeneous Treatment Effects: A Simple Exposition

This note provides a simple exposition of what IV can and cannot estimate in a model with abinary treatment variable and heterogeneous treatment effects. It shows how linear IV is amisspecification of functional form and the reason why linear IV estimates for this model willalways depend on the instrument used is because of this misspecification. It shows that if onecan estimate the correct functional form (non-linear IV) then the treatment effects areindependent of the instrument used. However, the data may not be rich enough in practice tobe able to identify these treatment effects without strong distributional assumptions. In thiscase, one will have to settle for estimates of treatment effects that are instrument-dependent.