两个半相依模型回归系数的改进估计  

The Improved Estimates of Regression Coefficients in Seemingly Unrelated Regression Model with Two Equations

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,北京100022

出　　处：《应用概率统计》2009年第6期619-631,共13页Chinese Journal of Applied Probability and Statistics

基　　金：西南财经大学"211工程三期"统计学重点学科建设项目资助

摘　　要：对于两个半相依回归系统的未知回归系数,本文首先借鉴文献中给出的两步协方差改进估计的方法给出两种两步协方差改进估计序列,并给出其与两步估计等价的条件和均方误差意义下的优良性;其次,我们对文献中给出的一种两步估计作简单改进,使得改进后的估计在更大的参数空间内优于最小二乘估计.再次,本文另辟蹊径,构造了一种新的估计,同样地,此估计也具有更好的小样本性质.本文最后一节讨论了Pitman准则下两步估计的优良性.For regression coefficients in seemingly unrelated regression model with two equations, this paper obtains a two-stage covariance improved estimator sequence by using the covaxiance improved idea for reference. The condition under which this improved estimator is equivalent to the classic two-stage estimator and the optimal property on the term of mean square error are also obtained. And then, by improving the two-stage estimator in the literature, this paper derives a new improved estimator which shows small sample property relative to least square estimator in more wide parameters space. Through a ultimate way, this paper provides a new estimator which have better small sample property. Finally, this paper discusses the optimal property of two-stage estimator under the Pitman criterion.

关 键 词：半相依模型 协方差改进估计 两步估计 PITMAN准则 

分 类 号：O212.1[理学—概率论与数理统计]