含时间因子的删失面板模型的估计与推断

Estimation and inference of censored panel data model with timefactors

本文研究了随机相关下的含时间因子的删失面板数据的估计与推断问题.该模型允许了不可观测的个体效应对潜在结果变量有着时变的影响.在假设误差项是对称且相互独立的条件下,给出了模型中的斜率参数与时间因子的识别与估计.文中给出的估计量是√N一致且渐近正态的.此外,文中提出的秩得分检验统计量无需对密度函数进行估计,在相对小的样本下也有着较好的表现.蒙特卡罗模拟和关于女性劳动时间的实证应用的结果验证了估计量的可行性.

This paper develops estimation and inference of censored panel data model with time factors under random correlated models.The model includes an interactive term to allow the unobserved individual characteristics having a time-varying effect.By assuming both error terms symmetric and independent with each other,the common slope coefficients and time factors can be identified.The proposed estimators are root-N consistent and asymptotically normal.In addition,we come up with a quantile rank score test statistics.By avoiding to estimate density functions,the test statistics perform well in relatively small samples.Simulations and an application to women’s working hours are carried out to prove the feasibility of our methods.