报告人：韩青博士

时间：5月24日（周四）上午9:00--10:30

报告题目：Nonparametric Censored Estimation with Endogeneity

摘要： This paper deals with the problem of nonparametric estimation using censored data in a model that features endogeneity. Nonparametrics with endogenous variables is difficult to handle because of ill-posed inverse problem. Nonparametrics with censoring does not attract the attention as it deserves because people are inclined to shift to quantile estimation when data is censored. We stick to the nonparametric estimation under these two conditions and claim that endogeneity shapes the model to be additive, and censoring delivers a (nonparametric) LAD estimation under the assumption of conditional zero median of the error term. So this paper transforms the problem into a Nonparametric Additive Least Absolute Deviation estimation which is saliently robust than L₂ norm estimation. We establish the asymptotic normality of the estimated unknown functions, the estimation and inference are easy to carry out.