A COMPARATIVE STUDY OF SOME ESTIMATORS IN ECONOMETRIC MODEL WITH MULTICOLLINEARITY

Abstract

Mostly, economic data are afflicted with the problems of multicollinearity. This leads to inaccurate parameter estimates in Ordinary Least Squares. Therefore, this paper examined the efficiency of three methods of parameter estimation in regression model (Ordinary Least Squares(OLS), Ridge Regression and Least Absolute Shrinkage and Selection Operator (LASSO)) under multicollinearity. Monte-Carlo experiment of 1000 trials was carried out for four sample sizes (20, 50, 100 and 150), each with three levels of collinearity( Low, Mild and Severe). The findings from this paper showed that when the collinearity level between the predictors is low, irrespective of the sample size, OLS is the most efficient estimator. However, under mild or severe collinearity condition, irrespective of the sample size, Lasso is the most efficient estimator.