Mean square error matrix comparisons of estimators in linear regression |
| |
Authors: | ötz Trenkler |
| |
Affiliation: | Statistics Department , University of Dortmund , Dortmund 50 , 4600 , West Germany |
| |
Abstract: | The aim of this paper is to provide criteria which allow to compare two estimators of the parameter vector in the linear regression model with respect to their mean square error matrices, where the main interest is focussed on the case when the difference of the covariance matrices is singular. The results obtained are applied to equality restricted and pretest estimators. |
| |
Keywords: | Mean Square error matrix Linear regression model Biased estimation Restricted least squares estimator Pre-tesl estimator |
|
|