Weighted estimators in regression with multicollinearity

Abstract

When dealing with highly correlated data, the ordinary least squares estimator is often unsuitable because of its iarge mean square error. This paper introduces an estimator, WPC, which is the weighted sum of the ordinary least squares estimator and the principal component estimator. WPC is compared with the ordinary least squares estimator and another weighted estimator recently introduced for use with highly correlated data. WPC is shown to have smaller mean square error under conditions that commonly occur in econometric studies