关键词: 综合评价, 数据包络分析, 超效率, 强平凡性公理, 稳健型生产前沿

Abstract:

The super-efficiency DEA model is widely used because it can measure the efficiency of effective decision-making units （DMUs）. However， there are still several key problems that have not been well addressed. （1） The uniform evaluation criteria. The super-efficiency DEA model assumes the DMU being evaluated should be excluded from the production possibility set， which indicates that the evaluation standard （Production Frontier） is different for each DMU， so whether the measurement results under different standards are comparable requires new theoretical support. （2） The lack of theoretical basis in economics. Due to the DMU being evaluated is excluded from the production possibility set， the super-efficiency production frontier constructed by other DMUs is an incomplete “production function”， and such a frontier lacks economic implications. If these two problems can't be solved well， then the rationality of the super-efficiency DEA model will never be guaranteed. （3） DEA production frontiers can be achieved in theory， but the difficulty of achieving these production activities is very different. From the existing literature of DEA methods， there are still few methods to research the difficulty of implementing production frontiers. In order to solve the above problems， the basic ideas of the generalized DEA method is refered to. Firstly， a strong ordinary axiom is proposed to give the construction method of robust production frontiers. Next， a modified DEA model （RSD） based on the robust production frontiers is proposed， and it is proven that the efficiency value given by the model （RSD） is equal to the super-efficiency of DMUs， and a new explanation is given for the economic meaning of DEA super-efficiency by the production theory. Furthermore， the concept of robust projection based on the model （RSD） is proposed， and a calculation formula for the difficulty of implementing DEA projection is provided. Finally， the proposed method and the original method are applied to compare and analyze the technological innovation efficiency of China's biomedical listed enterprises.According to empirical analysis， it is found that the proposed method explains the economic meaning of the radial super-efficiency DEA model， clarifies its axiom system， and makes the efficiency measurement results obtained based on the modified model more reliable and the projection improvement goal more feasible. In addition， the proposed method in this paper not only further improves the super-efficiency DEA effectiveness measurement method， but also provides theoretical support for the application of this method in economic management problems. All the data are derived from CSMAR， CNRDS， WIND database and the annual report of listed companies from CNINFO.

Key words: comprehensive evaluation, data envelopment analysis （DEA）, super-efficiency, strong ordinary axiom, robust production frontiers