存在中间产品退出的混合型多阶段系统DEA效率评价

DEA efficiency evaluation of hybrid multi-stage system with intermediate measure exits

具有复杂结构的系统经常会存在初始投入在不同子阶段分配和中间产品非完全再投入并存的现象,在其效率评价中忽略中间过程的传统Data Envelopment Analysis(DEA)模型往往会高估系统效率,而不考虑中间产品退出情形的网络DEA.模型对系统整体效率评价又可能偏低.为更好测算复杂结构系统的运行效率,基于企业技术创新的特征,建立同时考虑投入分配与中间产出分配问题的混合型多阶段系统DEA效率评价模型.依据分配比例测算各阶段效率,采用阶段效率的投入加权平均评价的系统整体效率,并计算模型的最优解.所提出的模型能够相对准确地反映复杂结构系统的特征,并深入系统"黑箱"内部,更大程度地为优化系统效率提供信息.对我国33个大中型工业企业的行业技术创新效率评价验证了模型的有效性.

The assignment of initial input at different sub-stages and the incomplete reentry of intermediate output are often two co-existent phenomena for a system with complex structure. The traditional data envelopment analysis （DEA） models often overvalue system efficiency due to sub-process ignorance; while the network DEA models with the negligence of intermediate outputs exits probably underestimate the efficiency. For a better efficiency evaluation, this paper develops a DEA model for hybrid multi-stage system in consideration of input and intermediate output assignment issues and the system complex structure. The efficiency of sub-stage is evaluated in terms of the allocation proportion, the system overall efficiency is calculated by the input weighted average of sub-stages efficiency, and the optimum solution of the model is given at last. The proposed model can precisely reflect the structural feature of system and provide more valuable information for the system efficiency optimization by opening the ＂black box＂. A model application to the technological innovation efficiency evaluation of China＇s 33 large and medium-sized industrial enterprises by sector is introduced to verify the availability of the model.