【主讲】魏杰 (华中科技大学)
【主题】Censored Quantile Regression with Endogeneity in Functional Coefficient Models
【时间】2015年4月3日 (周五) 15：30-17：00
【地点】上海财经大学经济学院楼602室
【语言】英文
【摘要】This paper considers estimation of coefficients which are functions of some covariates in a model subject to endogeneity and censoring. We employ a conditional distribution function as a control function to purge dependence of endogenous variables to random errors. We select uncensored quantiles via a smooth selector function. The control function and the input of the selector function both come from nonparametric quantile regression. We construct the estimator of functional coefficients in a fashion of pairwise difference. We prove that our estimator is consistent and asymptotically normal. This property helps us develop a specification test for constancy of coefficients. The test proves to be asymptotically consistent and possess nontrivial local power against local alternatives.
