Separation of uncontrollable factors and time shift effects from DEA scores

It has been pointed out that DEA scores may be influenced by several external environmental factors, which are uncontrollable for DMUs. It implies that the DEA efficiency score without data adjustment might be biased and impractical for measuring genuine management efficiency. Therefore it is essential to eliminate uncontrollable effects from DEA scores and evaluate “pure” managerial efficiency for DMUs. In an effort to solve this problem, we employ a multi-stage data adjustment procedure using DEA and regression models, which is originally proposed by Fried et al. [1999] consisting of four stages. In this study, we further modify this procedure by introducing newly developed devices in each stage; Connected Slacks-Based Measure (CSBM) model at the first and fourth stages, the Tobit model with DMU dummies at the second stage, and a data tuning procedure at the third stage. Then we decompose the technical inefficiency into three factors, i.e. environmental effects, time shift effects and pure technical inefficiency. Lastly, we apply this procedure to the electric power utilities in Japan and the US and compare their pure technical efficiency and causes of inefficiency.