基于参数设计的信噪比交叉效率评价方法

针对DEA交叉效率评价过程中没有考虑自评与互评效率的作用而主观赋予相同权重导致交叉效率评价值不准确的问题.文章基于参数设计的思想,依据试验设计中可控与不可控因素的作用机理区分自评权重和互评权重对所评价决策单元交叉效率的影响与作用,将其界定为可控与不可控因素的管理学属性,明确不同权重作用机理;引入信噪比作为衡量决策单元交叉效率评价时的性能指标,实施DEA交叉效率评价方法的改进,设计出DEA信噪比交叉效率集结方法,从而实现交叉效率的集结方式由单一考虑交叉效率波动的均值转化为综合考虑交叉效率波动情况(均值与方差),交叉效率评价值用信噪比交叉效率替代交叉效率平均值更具有统计学意义并可从管理学角度解释,评价结果也具有更高的可区分性;最后通过算例分析验证了交叉效率评价理论上的必要性和该方法的合理性与可行性,同时发现了交叉效率评价中存在CCR有效DMU序位超出了有效DMU范围现象,建议应实施同质DMU检验和评价值归一化.文章的研究也为提高DEA交叉效率测算的准确性提供一种新思路.

An Evaluation Method for the DEA Signal-to-Noise Ratio Cross Efficiency Based on Parameter Design

In conventional DEA cross efficiency models,there is no consideration of the effects of self-evaluation and peer-evaluation efficiency and the same weights are subjectively assigned,which leads to an inaccurate evaluation value of cross efficiency.In this study,a new DEA signal-to-noise ratio cross-efficiency model based on parameter design is developed.According to the mechanism of controllable and uncontrollable factors in the design of experiment,the influences of self-evaluation weight and peer-evaluation weight on the cross efficiency of decision-making units(DMUs) are distinguished in order to clarify the corresponding mechanisms.Moreover,a signal-to-noise ratio is adopted as a performance index to represent the cross efficiency evaluation of DMUs.The necessity,rationality and feasibility of the newly developed model is verified by an example.It is recommended that decision makers should implement the homogeneous DMU test and normalize the values in the actual assessment process as some efficient DMUs in CCR model exceeds the valid range.Compared with previous models only considering the mean value of cross efficiency fluctuation,the proposed model has taken both the mean value and variance into consideration.It can be interpreted in terms of statistics and management when adopting the SNR cross efficiency rather than the cross-efficiency average as the evaluation value of cross efficiency and the evaluation results become more distinguishable.This study brings new insights in improving the accuracy of DEA cross efficiency measurement.