Binary Response Model with Many Weak Instrumental Variables
2020/01/21
研討會日期 : 2020-01-21
時間 : 10:30
主講人 : Ms. Dakyung Seong
主持人 : Professor Tzu-Ting Yang
地點 : Conference Room B110
演講者簡介 : Ms. Seong will receive her Ph.D. in Economics from University of California, Davis in 2020. Her research interests are Econometrics, and High-dimensional/Functional Data analysis. She is applying for a position of the Institute of Economics, Academia Sinica now.
演講摘要 : This paper considers an endogenous binary response model with many weak instruments. Unlike linear simultaneous equation models, binary response models with endogenous regressors and many weak instruments have received very limited attention from the literature despite its practical importance. This paper provides two consistent and asymptotically normally distributed estimators, a ridge-regularized conditional maximum likelihood estimator (RCMLE) and a ridge-regularized nonlinear least square estimator (RNLSE), based on first stage ridge-regularization. Consistent estimators of the asymptotic variances are also provided. Monte Carlo simulations show that both the RCMLE and the RNLSE outperform existing estimators when many weak instruments are present. We apply the proposed estimators to two empirical examples. The first example examines the effect of education on being employed for a full calendar year. The second example illustrates the effect of children’s development on maternal employment status using various instrumental variables.
