基于双惩罚分位回归的面板数据模型理论与实证研究  

Theoretical and empirical study on panel data models based on double penalized quantile regression

作　　者：罗幼喜[1,2] 李翰芳[1,3] 田茂再[4] 郑列[1,2] Luo Youxi Li Hanfang Tian Maozai Zheng Lie(School of Science, Hubei University of Technology, Wuhan 430068, China Institute of Product Quality, Hubei University of Technology, Wuhan, 430068, China School of Mathematics and Statistics, Central China Normal University, Wuhan 430079, China School of Statistics, Renmin Univiesity of China, Beijing 100872, China)

机构地区：[1]湖北工业大学理学院,湖北武汉430068 [2]湖北工业大学产品质量工程研究院,湖北武汉430068 [3]华中师范大学数学与统计学学院,湖北武汉430079 [4]中国人民大学统计学院,北京100872

出　　处：《武汉科技大学学报》2016年第6期462-467,共6页Journal of Wuhan University of Science and Technology

基　　金：国家自然科学基金资助项目(11271368);教育部人文社会科学研究青年基金资助项目(13YJC790105);湖北工业大学博士科研启动基金资助项目(BSQD13050)

摘　　要：固定效应和随机效应同时选择是面板数据模型研究中的重要问题之一。本文通过分别对固定效应和随机效应引入条件Laplace先验,提出了一种新的贝叶斯双惩罚分位回归法。该方法不仅能对模型中重要解释变量进行自动选择,而且充分考虑到个体随机波动对解释变量系数估计带来的偏差。通过对方差分量的惩罚压缩,减少了模型中未知参数的个数,提高了模型自由度。Monte Carlo模拟及实证分析显示,所提出的方法不仅能准确估计出固定效应系数,而且能精确地捕捉到个体随机效应的波动。It is an important issue to select fixed and random effects simultaneously for panel data mod- els. This paper proposes a new Bayesian double penalized quantile regression method by introducing the conditional Laplace prior both for fixed and random effect parameters. This method can not only select the important explanatory variables in the model automatically but also give a full consideration to the biases of parameter estimation for explanatory variables which are produced by individual ran- dom fluctuations. By applying shrinkage to the variance components, the number of unknown param- eters in the model is reduced, thus the model＇s freedom degree is enhanced greatly. Monte Carlo sim- ulation and empirical study indicate that the proposed method can accurately estimate the fixed effect parameters and catch the exact fluctuation of individual random effects.

关 键 词：面板数据 分位回归 贝叶斯分析 固定效应 随机效应 变量选择 Laplace先验 

分 类 号：O212[理学—概率论与数理统计] F064.1[理学—数学]