半相依模型参数估计的改进  

The Adjusted of Parameter Estimation in Seemingly Unrelated Regression

作　　者：马铁丰[1] 王松桂[2] 

机构地区：[1]西南财经大学统计学院,成都610074 [2]北京工业大学应用数理学院,北京100124

出　　处：《数学物理学报（A辑）》2009年第5期1256-1264,共9页Acta Mathematica Scientia

基　　金：北京市属市管高等学校人才强教计划项目(05006011200702);国家自然科学基金青年基金(10801005)资助

摘　　要：对于m个半相依回归系统的未知回归系数,文献[7]提出一种利用信息逐次迭加的方法,该文首先在其基础上给出一种进一步改进形式,并得到了其相合性,同时作者借鉴文献[7]提出逐次迭加信息的构造估计思想给出一种具有小样本优良性的可行估计,模拟研究也表明作者的改进估计是有效的.文献[10]根据Rao的协方差改进思想,给出一种更为简洁的两步估计,该文在此估计基础上给出一种改进形式,新估计具有更好的可操作性和均方误差意义下的优良性.For a seemingly unrelated regression with m linear regression models,the literature propose a new method which based on iterative information step by step.Based on this method,this paper obtain a farther adjusted form which also has congruence.At the same time, a new estimator is proposed by using the idea in literature.It is a feasible estimator having small sample property.The simulation result also indicates that our estimator is effective.The literature gave a concise two-stage estimator by using the covariance adjusted approach of literature [8]. This paper improves on it which shows the more power of small sample optimality by comparing with least square estimator.

关 键 词：半相依模型 协方差改进估计 两步估计 

分 类 号：O212.1[理学—概率论与数理统计]