固定效应部分线性可加面板数据模型的WCQR估计及变量选择
首发时间:2022-03-11
摘要:本文考虑固定效应部分线性可加面板数据模型,先通过固定效应变换复合分位数回归法消除固定效应的影响,再利用加权复合分位数回归(WCQR)估计方法对模型进行估计,其中非参数分量采用B样条基函数逼近,所提出的方法可以同时估计参数回归系数和非参数分量。为了识别线性部分显著变量,利用自适应LASSO的方法同时实现参数的变量选择和非参数函数的估计。在一定的正则性条件下,证明了所提估计的渐近性质与Oracle性质。最后通过随机模拟验证该方法的表现,证实了所提出方法的有效性。
关键词: 固定效应部分线性可加面板数据模型 加权复合分位数回归 自适应LASSO
For information in English, please click here
WCQR estimation and variable selection for fixed effects partially linear additive panel data model
Abstract:This paper considers the fixed effects partially linear additive panel data model. Firstly, the influence of the fixed effects is eliminated by the fixed effects transformation composite quantile regression method, and then use the weighted composite quantile regression to estimate the model. The nonparametric components are approximated by B-spline basis functions, and the proposed method can estimate both parametric regression coefficients and nonparametric components. In order to identify the significant linear variables, the adaptive Lasso method is used to select the parameters and estimate the nonparametric functions simultaneously. Under certain regularity conditions, the asymptotic properties and Oracle properties of the proposed estimator are proved. Finally, the performance of the method is verified by random simulation, and the effectiveness of the proposed method is verified.
基金:
引用
No.****
同行评议
勘误表
固定效应部分线性可加面板数据模型的WCQR估计及变量选择
评论
全部评论