学习效应嵌入下动态决策单元DEA效率评价与管理目标设定的研究及应用

数据包络分析（DEA）作为评价和改进决策单元效率的一种有效方法和工具，在很多领域得到广泛应用。传统DEA模型将决策单元视为静态的投入产出系统，决策单元自身的动态性特征（如学习能力、经验累积因素、投入的离散特征等）常被忽视，实际评价中决策单元动态信息的完备表述和学习效应参数的引入亟待相关理论和方法来完善。为此，本项目的研究对象是存在动态特性的决策单元；主要工作是通过动态决策单元的时序结构分解，以学习瞬时效应传递矩阵为外生变量构建DEA评价模型并以之为基础进行“干中学”效应下决策单元的全要素生产力指数模型的设计和效率分解，提出科学的动态决策单元效率评价方法；并应用于资源优化配置和产出水平选择等管理目标设定问题；研究成果最终运用于国际、区域和企业的环境治理效率评价和优化决策实践。本研究项目将扩大DEA研究领域，丰富和完善DEA理论体系，并为宏微观层次上的环境治理管理问题提供一定的政策和决策支持。

As an effective tool for evaluating and improving the performance of decision-making units (DMUs), DEA has been widely used in various areas. The traditional DEA models take DMUs as a static input-output system, and the dynamic features of decision-making units (such as learning capability, experience accumulation factors, discrete features of inputs) have been regularly ignored. Besides, some relevant theories and methods are in need for the complete expression of dynamic information of DMUs in practical evaluation as well as the introduction of learning effect parameters. In this case, this study focuses on dynamic DMUs and the main task is to design DEA evaluation models based on the external variables of instantaneous learning effect transfer matrix by decomposing DMU sequence structure, and to re-construct Total Factor Productivity index models and decompose the efficiencies considering the “Learning By Doing” effect. Meanwhile, the scientific evaluation methods of dynamic DMU efficiency are put forward and applied in managerial target-setting problems such as resource optimizing allocation and output level design. The study findings are applied in the real practice of environmental management performance evaluation and decision-making optimization in countries, regions and manufacturing enterprises. This study enlarges DEA research areas, enriches DEA theoretical systems and offers certain policy-making and decision-making support for the environmental management both in the macro and in micro level.