Brown Bag Seminar: Jau-er Chen

Dynamic Instrumental Variable Quantile Regression with Unbalanced Panel Data and Interactive Effects

Jau-er Chen, National Taiwan University and MIT

This paper studies a dynamic panel instrumental variable quantile regression with interactive effects and an unbalanced panel dataset. The proposed econometric procedure controls for dynamic bias and endogeneity bias, and then consistently estimates the parameters without suffering the incidental parameters problem in the context of instrumental variable quantile regressions. The proposed estimator controls for individual time-varying heterogeneity via a factor structure on the unobserved heterogeneity, offering a more flexible approach to the panel data least squares. Our econometric procedure is justified empirically by a partial-adjustment model of firm leverage and DeAngelo and Roll (2015, Journal of Finance). Extensive Monte Carlo results are provided. The procedure is theoretically based mainly on Chernozhukov and Hansen (2006 & 2008, Journal of Econometrics) and Bai (2009, Econometrica). Currently, we are working on deriving the asymptotic properties of the proposed estimator, which suffices conducting hypothesis tests. As to an empirical application, we implement the estimation procedure using an unbalanced firm-level data set and conduct bootstrap-based inference in R. All the related functionalities of our econometric procedure such as plotting are wrapped into a R package compiled by ourselves, which provides a user-friendly interface for empirical researchers.