Estimation of Censored Regression Models with Endogeneity
报告人：Qian Wang (University of Nottingham Ningbo China)
地点：Tencent Meeting (280-219-727)
Abstract: It is usually difficult to deal with endogeneity in censored regression models. Existing methods have some drawbacks, such as the requirements of continuous endogenous regressors or the difficult implementation of the estimation procedures. In this paper we investigate the estimation of censored models with endogeneity under mean restrictions and quantile restrictions on the errors respectively. We propose a weighting scheme utilizing the conditional independence of a special regressor and the errors. The parameters of interest in the mean model can be estimated by weighted two-stage least squares, and we extend instrumental variable quantile regression to accommodate the censored model under quantile restrictions. Our estimators do not require the endogenous regressors to be continuous, and are computationally convenient. We show that our estimators are consistent and asymptotically normal. Monte Carlo studies demonstrate that our estimators perform well in finite samples.
About the Speaker:
Qian Wang is currently an Associate Professor in Economics at University of Nottingham Ningbo China. She obtained her Ph.D. degree from the Hong Kong University of Science and Technology. Her research interests include econometric theory and applied econometrics.