A New Biased Estimator in Linear Regression Model

摘要

In this paper, we propose a new ridge type estimator to overcome the multicollinearity problem, and we call the new biased estimator as the stochastic restricted ridge estimator (SRRE). In the mean squared errors matrix sense SRRE will be compared with several other biased estimators. The necessary and sufficient conditions for the superiority of the new estimators SRRE over the ridge estimator (RE) and the Mixed Regression Estimator (MRE) in the mean squared error matrix criterion are derived. A numerical example with Monte Carlo simulation is given to illustrate the theoretical results.