Modified Two Parameter Regression Estimator for Solving the Multicollinearity

Abstract

When the multicollinearity problem appears in the multiple linear regression model, the performance of the unbiased estimator which is the ordinary least squares (OLS) is inefficient. To solve the above-mentioned problem, several biased and almost unbiased regression estimators are introduced. In this study, as an alternative to the OLS estimator, a modified two-parameter regression estimator called the Dawoud biased regression (DBR) estimator is proposed. Moreover, we theoretically compare the performance of the DBR estimator with the OLS and some existing estimators by the criterion of the mean squares error. Furthermore, a Monte Carlo simulation study and real-life data are given to evaluate the performance of the DBR estimator. The main finding is that the DBR estimator performs better than other regression estimators under determined conditions.