摘要：This paper considers three different approaches of handling multicollinearity in regression analysis using economic data. These techniques were applied to study determinants of money supply of the sector of the economy. The goal is to determine which of the economic variables included in the factors that influence money supply (either in the broad ‘M1’ or the narrow ‘M2’sense) is not actually contributing to the effect of money supply by monetary authority of the sector of the economy. A comparative analysis of the three methods using the adjusted R2 , Mean Square Error and Root Mean Error, as the statistics criteria, revealed that the result obtained from the Ridge Regression gave the outstanding performance as compared with the other two techniques. On the overall, it was found that the Ridge Regression performed best, followed by Principal Component Regression, while Latent Root Regression performed least.
关键词：Multicollinearity; Ordinary Least Squares; Principal Component Regression; Latent Root Regression and Ridge Regression
