“Quantile Regression with Endogeneity” by Professor Songnian CHEN
Professor Songnian CHEN
Department of Economics
The Hong Kong University of Science and Technology
In empirical research the quantile regression technique (QR) developed by Koenker and Bassett (1978) has increasingly become a major rival to the least squares method (OLS). Unlike the OLS, quantile regression techniques are particularly well suited to the analysis of censored data due to its equivariance with respect to monotone transformation. For the quantile regression with endogeneity, the instrumental variables quantile regression estimator (IVQR) developed by Chernozhukov and Hansen (2006, 2008) is becoming increasingly popular among applied researchers. Censoring and endogeneity are both common problems in practice, but at present, no satisfactory and easily accessible instrumental variable censored regression techniques are available. In this paper, we develop the sequential instrumental variable censored quantile regression procedure (SIVCQR) for the family of structural quantile regression coefficients. Our SIVCQR estimator is simple to implement in typical applications. By effectively transforming the difficult instrumental variable censored quantile regression problem into more standard IVQR procedure, our estimation approach makes the instrumental variable quantile regression techniques for censored data easily accessible to applied researchers. We provide large sample properties of the SIVCQR. Simulation results show that our estimator performs well.