ABSTRACT

There are many common efficiency evaluation methods and DEA is just one of the commonest. However, as the research continues, more and more scholars question traditional DEA model (such as CCR). When DEA (CCR) is used to evaluate the efficiency, all decision-making units choose the weights which are the most beneficial to themselves, but the actual difference in the importance among these indexes is not taken into consideration. During the actual calculation process, different decision-making units are chosen and though the same input index and output index are adopted, the final weights are very inconsistent, which is contrary to actual situation. The results are that the evaluation discrimination degree is not high and the relatively effective sorting among these decision-making units is difficult. In order to resolve these problems, scholars at home and abroad propose many improvement methods, which are generally divided into three classes. The first class is cross evaluation method, with the cross efficiency evaluation model proposed by Sexton and others (1986) as the representative. Then, Doyle and Green (1994) proposed the two-step approach to resolve the non-uniqueness of cross efficiency. Wu Jie (2008) combined both of them and put forward DEA cross evaluation model which considered the interval mean cross efficiency of all weight information. Though these evaluation models resolve the sorting problem of decision-making units, the weight allocation problem is not resolved. The second class is the preference evaluation method, with the cone ratio C2 WH model proposed by A. Charnes and others

and others (2011) applied this idea in the study on public rental housing allocation and made certain achievements. This model doesn’t require value anticipation, which is one step closer to the resolution of weight allocation and decision-making unit sorting, but the possibility of non-unique solution still exists theoretically after introducing two virtual decision-making units.