错误先验假定下回归系数Bayes估计的小样本性质

The Small-sample Properties for the Bayes Estimator of Regression Coefficients under Misspecified Prior Assumption

摘要: 本文基于错误指定的先验假定获得了回归系数的Bayes估计（BE），并在均方误差矩阵准则下对其与最小二乘（LS）估计进行了比较．导出了它们的相对效率的界．讨论了在后验Pitman Closeness准则下BE相对于LS估计的优良性。

Abstract: In this artical, under misspecified prior assumption the Bayes estimator （BE） of regrtession coefficients are obtained, based on the mean square error matrix criterion, it is compared with least square estimator （LSE）. The bounds of the relative efficiencies for the considered estimators are derived. Filially, under the posterior Pitman closeness （PPC） criterion, the superiority of the BE over LSE is discussed.