Author: Songnian Chen; Shakeeb Khan; Xun Tang

Abstract:

We identify and estimate treatment effects when potential outcomes are weakly separable with a  binary endogenous treatment. Vytlacil and Yildiz (2007) proposed an identification strategy that  exploits the mean of observed outcomes, but their approach requires a monotonicity condition. In  comparison, we exploit full information in the entire outcome distribution, instead of just its  mean. As a result, our method does not require monotonicity and is also applicable to general  settings with multiple indices. We provide examples where our approach can identify treatment  effect parameters of interest whereas existing methods would fail. These include models where  potential outcomes depend on multiple unobserved disturbance terms, such as a Roy model, a  multinomial choice model, as well as a model with endogenous random coefficients. We establish  consistency and asymptotic normality of our estimators.

详情链接：https://doi.org/10.1016/j.jeconom.2023.105567