摘要
针对面板数据个体固定效应复合分位数回归模型,研究回归系数估计的渐进相对效率.采用计算复合分位数回归估计和最小二乘法估计的协方差矩阵的迹的比值,计算结果表明复合分位数回归相对于最小二乘法的渐进相对效率的比值大于70%.还将Zou在2008年提出的适应性lasso的想法应用于此面板数据个体固定效应复合分位数回归模型,构造出适应性lasso惩罚复合分位数回归估计,并在适当条件下证明其估计的渐进性质.
In this paper,the asymptotic relative efficiency of regression coefficient estimation was studied for the composite quantile regression model of individual fixed effect in panel data.The ratio of trace of covariance matrix of composite quantile regression estimation and least squares estimation was calculated.The results showed that the asymptotic phase of composite quantile regression was relative to that of least squares method.The ratio of efficiency to efficiency was more than 70%.This paper also applied Zou s idea of adaptive lasso proposed in 2008 to the composite quantile regression model of individual fixed effect in panel data,constructed the adaptive lasso penalty composite quantile regression estimation,and proved the asymptotic nature of the estimation under appropriate conditions.
作者
刘燕
范永辉
LIU Yan;FAN Yong-hui(School of Mathematics Science,Tianjin Normal University,Tianjin 300087,China)
出处
《哈尔滨商业大学学报（自然科学版）》
CAS
2019年第1期105-108,共4页
Journal of Harbin University of Commerce：Natural Sciences Edition
关键词
面板数据
复合分位数回归
渐进相对效率
适应性lasso
变量选择
渐进正态
panel data
composite quantile regression
asymptotic relative efficiency
adaptive lasso
variable selection
asymptotic normality
