基于半自由处置变量的链型系统DEA模型

对网络数据包络分析(DEA)模型的一个特例即链型系统的DEA模型进行研究。传统DEA模型总是假设非有效的决策单元的所有输入(或输出)可以同比例改变以提高性能,但这并不符合所有的情形。在两阶段生产系统DEA模型中引入半自由处置变量,建立新模型。新模型将使非有效的决策单元和子决策单元可以根据管理需要有目的地分比例控制各输入和输出变量,一个简单的例子证明了模型比之传统方法更能发现非有效的决策单元并指明其性能改善的方向。最后,该方法被运用到商业银行效率评价的实践中。

DEA models for the chain-like systems with semi-discretionary variables

A type of simplified network data envelopment analysis(DEA) model,namely,DEA model for chain-like systems is studied.The traditional DEA models are based on the assumption that all the inputs(or outputs) of the inefficient decision-making units should be reduced(or increased) in the same proportion,which is not always true.The semi-discretionary variables are introduced into the DEA models for two-stage production systems.In this scenario,the decision makers of the units and subunits can control the input reduction or output expansion in different proportions to achieve the best performance.A simple numerical example proves that the proposed approach is more helpful to find inefficient decision-making units and the improvement directions than the traditional approaches.Finally the model is illustrated by an empirical application in the commercial bank evaluation.