一种DEA效率概率占优方法研究

决策单元间的效率关系研究是效率研究中的重要领域.传统DEA模型通过计算决策单元各自最优权重情景下的效率最大值,来判断决策单元间的效率排序关系.而这种最优权重往往是极端权重,没有考虑其它权重情景.采用其它权重时,决策单元的效率值和排序往往会发生较大的变化.本文考虑了所有权重情景,提出了DEA效率概率占优模型,定义了决策单元的占优效率概率和被占优效率概率.该模型能够判断任意两个决策单元的效率概率占优关系;并将效率概率占优关系从两个决策单元扩展到决策单元与决策单元集.最后将其应用于决策单元效率排序.通过实证分析发现了两条重要结论:1)考虑所有权重情景的排序与采用极端权重的CCR效率值排序不一致;2)考虑所有权重情景时,一些无效单元的效率排序比有效单元的效率排序更高.

A DEA efficiency probability dominance method

The study of efficiency relationship among DMUs (decision making units, DMUs) is an important area in efficiency research. Traditional DEA model computes DMUs' maximum efficiency scores by optimal weights and uses the scores to rank DMUs. However the optimal weights always are extreme weights and the DEA model do not consider the other weights. When the other weights are used, DMUs' efficiency scores and ranking always are changed. This paper proposes a DEA efficiency probability dominance model and defines dominance efficiency probability considering all weights. This model can determine efficiency relationship between two DMUs and efficiency relationship between a DMU and a DMU set. Finally, we apply it to DMUs' efficiency ranking. Based on the application, we have two significant conclusions. One is that the rankings considering all weights is different from the rankings by CCR efficiency scores. Another is when considering all weights the rankings of some inefficient DMUs may be higher than the rankings of efficient DMUs.