评价综合效率的RAM-DEA模型具有非径向性、客观性、指标多样性的特点,但模型中一维权重参数未充分考虑决策单元与评价指标的相互影响,导致测得的效率值可能存在偏差。在以往研究的基础上,将决策单元与多种投入、产出指标的差异纳入模型之中,建立优化权重RAM-DEA模型并进行实证研究。为全面衡量工业行业效率,构建了以计算机数、网站数为信息化投入,能源、劳动力、资本为自然投入,工业增加值为期望产出,温室气体与总颗粒物排放量为非期望产出的综合效率评价指标体系,并对中国39个工业行业综合效率进行评价。通过比较模型改进前后测算的综合效率值,发现优化权重RAM-DEA模型可以涵盖多类投入、产出指标且考虑到各指标与决策单元之间的影响,测得的效率值能反映出工业行业间差异。

The RAM-DEA model for evaluation of comprehensive efficiency has the characteristics of non-radiality, objectivity and index diversity. However, the parameters designed in the model does not fully consider the interaction between decision making units(DMU) and various indices, that may cause bias. On the basis of previous research, under the consider of differences between input/output indicators and decision making unit, we establish a weight optimized RAM-DEA model. In order to measure the efficiency of the industrial industry comprehensively, the number of computers and websites are included in information input, while energy, labor, capital as natural inputs, industrial profit as expected output and greenhouse gas and total particulate emissions as undesirable output. The empirical study of China"s 39 industrial sectors shows the comprehensive efficiencies resulted from the previous model and weight optimized RAM-DEA model. We can find that the weight optimizing RAM-DEA model can contain many kinds of input and output indices under considering effects of DMUs with indices on evaluation, thus reflecting differences of various industrial sectors through accounting.

F124.3

国家自然科学基金项目（面上项目）“面向小样本多属性决策的软集合理论及其应用研究”（71171209）;国家自然科学基金项目（面上项目）“面向不确定性混频数据的软集合预测模型与方法研究”（7167010609）