Effects of Multicollinerity in Data Envelopment Analysis (DEA) Models of Product Market Efficiency

Although the effect of multicollinearity among independent variables and means of handling it have received wide attention in the context of multiple regressions, these issues are as yet unexamined as they apply to efficiency estimates. In this exploratory paper we examine the impact of multicollinearity among input variables on the stability of efficiency estimates using data envelopment analysis (DEA) and present an alternative methodology that provides relatively robust estimates. Our analysis with previously published data also shows that irrespective of the methodology used, the linear functional form is more sensitive to the presence of multicollinearity.