Efficiency Ranking via Combining DEA Evaluation and Bayesian Prediction for Logistics Enterprises

摘要

Data envelopment analysis (DEA) has been widely used in economic development evaluation and enterprise performance analysis. When a conventional DEA is used directly for ranking, there are some issues. It requires the investment and income data, and the latter are after-the-fact data, so DEA is ex post analysis. In addition, coarse granular DEA results may cause parallel ranking at high frequencies. Therefore, combining DEA and Bayes, we propose an efficiency prediction approach that does not require income data. The approach can predict efficiency levels under different investment combinations, and help logistics enterprises to rationally allocate limited resources in their decision-making. Furthermore, an efficiency ranking algorithm is designed by incorporating the overall probability distribution of data set and then is applied to evaluate fourteen A-level logistics enterprises in Anhui, China. Empirical results show that the DEA-Bayes approach has good discrimination for efficiency ranking. Unlike expert scoring, our evaluation process is based on logistics enterprise data and easy to operate.