Application of data envelopment regression model (DERM) in Indian banking sector

In regression analysis, weighting is one of the procedures practiced to minimize the effect of non-response and non-coverage and to have the resultant information closer to population information. However no stable or uniform method for weighting is in existence and hence a new model of using the efficiency scores derived by testing 50 models of Data Envelopment Analysis is used as weights. This attempt is a first time in the literature and the weighting with DEA in regression analysis is exercised to fit the Data envelopment Regression model (DERM). Based on Akaike Information criterion (AIC), it is evident that the derived information is closer to population information in many models showing the model CCRNO (Convex-Constant-Radial-Non Orientation) as the highly suited for this purpose. An empirical study is carried out with eight independent and eight dependent variables of 88 banks by applying these weighting models.