SMU SOE Seminar Series (November 4, 2025): Partial Identification of Quantile Selection Models

TOPIC:

ABSTRACT

Arellano and Bonhomme (2017) and Chen, Feng and Zhang (2024) considered semiparametric estimation of a binary quantile selection model. Both articles impose a parametric structure on the copula function that characterizes the extent of sample selection bias. However, misspecification of the parametric copula function is likely to result in biased estimates and misleading inference. In this article we study partial identification and estimation of the model without imposing any parametric structure on the copula function. We also propose inference procedures, and all of our methods can be implemented in a straightforward manner. Numerical experiments show that our procedures work well. In addition, we also study partial identification and estimation of a quantile regression model subject to a censored selection without imposing any parametric structure on the copula function. As in the case of binary selection, our estimation and inferences procedures for the censored selection case are easy to implement.

Click here to view the CV.