Estimation in Triangular Models of Binary Response under Quantile Restrictions
2013/02/05
研討會日期 : 2013-02-05
時間 : 14:30
主講人 : Mr. Jen-Che Liao
地點 : Conference Room B110
演講者簡介 : Jen-Che Liao is going to receive his Ph.D. in Economics from University of Wisconsin-Madison in 2013. His research field is in econometric theory and applied econometrics. He is applying for the position of the Institute of Economics, AS right now.
演講摘要 : In this paper, we consider a triangular simultaneous equations model with a binary outcome that is identified under a weak nonparametric quantile restriction. The proposed two-step estimation procedure combines Horowitz's (1992) smoothed maximum score estimator for semiparametric binary response models with a control function approach to the endogeneity problem. The estimator relies on estimates of two types of infinite-dimensional parameters: a nonparametric conditional quantile function with a nonparametrically generated control variable as the argument. A series approximation is employed for the former to exploit the additive structure of the model. Rates of convergence and the asymptotic distribution are derived. A stochastic expansion of the series estimator further reveals the influence of the nonparametrically generated regressor on the asymptotic behavior of the final estimator. Specifically, we show that the bias and variance of the final estimator are augmented by an additive factor that is proportional to the bias and variance, respectively, of the nonparametrically estimated control variable. The model of partially linear binary regression quantiles is covered as a special case of our general setup. In a simulation study, we present the finite-sample performance of the estimator and illustrate advantages of the proposed approach by comparing with other alternatives. Finally, an empirical example estimating a female labor market participation model with endogenous non-labor income is given.
