基于效用网络 DEA 的银行运营绩效评价

针对两阶段网络数据包络分析法(data envelopment analysis, DEA)中加法和乘法效率分解模型所存在的问题，文章提出期望效用理论分析框架下的效率评价方法。首先考虑子系统的内在关系，给出子系统期望效率的定义，并通过蒙特卡洛仿真算法进行求解；然后利用多属性效用函数和不含投入的DEA模型集成得到系统总效率；进而将该方法拓展到一般性的多阶段序列网络数据包络分析法结构中，扩大其应用范围；最后通过对中国上市商业银行运营效率的评价来验证方法的有效性与合理性。

According to the drawbacks of additive and multiplicative decomposition models in two-stage network data envelopment analysis (DEA), this paper proposes a novel approach to efficiency evaluation based on expected utility theory. Firstly, considering the internal relations between two stages, the expected efficiency for each stage is defined and calculated by Monte Carlo experiments. Then, by using multi-attribute utility function and DEA model without input, the overall efficiency of the system is obtained. Moreover, in order to expand the scope of application, the approach is extended to a generalized multi-stage network structure. Finally, the effectiveness and rationality of the proposed approach are verified by an empirical study about operational performance of listed commercial banks in China.