The Benchmarking Comparison Analysis And Its Applications

Multi-Attribute Utility Theory(MAUT) is the most widely used methodology in performance evaluation.However,the MAUT employs the same function form and weights to all the alternatives,which can reflect the common features of alternatives very well,but it is often restrictive into considering the specific characteristics of each alternative.In addition,the MAUT can only provide limited infonrmtion to tell the managers how to improve the performance of the alternatives.As one of the effective managerial tools,Benchmarking has been widely applied to performance management in private and public sectors.Nevertheless,after lots of literature reviews on Benchmarking,we found that only part of academic papers employed mathematic methods to identify the benchmarks.There has been no unified and tight-logic mathematical foundation to guide the Benchmarking practice.In this thesis,we established a general theoretical framework for Benchmarking based on the practice and developed a methodology of performance evaluation called "Benchmarking Comparison Analysis"(BCA).Furthermore,we proposed three principle components of BCA:Preference,Measure and Benchmark set.The principle of BCA is as follows:firstly,to establish the initial preference of decision makers on the alternatives;secondly,to search and construct the benchmark set;thirdly,to measure the performance of alternatives by their relative or absolute distances to benchmark set.We first conducted systematic research on the foundations of BCA,which includes(1) summarize the often used preferences and measures in practice and academic research;(2) define the relevant concepts of Benchmarking;(3) study the extensions of the attainable set based on the axiom assumptions;(4) study the BCA models based on the ideal points,best practice,best and worst practice,and attainable set.We also pointed out that some classical multi-criteria decision methods such as TOPSIS,DEA,ELECTRE,PROMETHEE etc can be neatly fitted into the framework of BCA.Thereafter,based on the framework of BCA,we discussed the following four issues in performance evaluation:1.In order to incorporate the preference of decision makers into the evaluation model,we discussed the two specific cases of BCA,namely DEA and DEA-WEI models,and concluded the necessary conditions required in the preferences which were implied in the models.Then we proposed the Lexicographic DEA model.2.With respect to the undesirable variables in performance evaluation,we proposed the "extended free disposal" assumption,base on which we studied the DEA model with slacks based measure,radial measure and Russell measure respectively, and their relations with the models in previous literatures.3.We proposed the combined performance indexes of BCA in order to use the information of best and worst practice.And the results of empirical examples showed that the BCA-indexes have greatly enhanced the discrimination power of DEA models. Meanwhile,the results of simulation experiments presented that the BCA-indexes perform much better the BCC DEA model especially when the sample size is small.4.Considering the dynamics of benchmark sets with time,we defined a general dynamic performance index,and proposed a competitiveness index to measure the dynamic integrate capacity of the organization.In the end,we employed some BCA models to conduct two case studies on performance evaluation of institutes of CAS from the perspectives of CAS' internal benchmarks and international benchmarks.The results of the case studies show BCA models could have extensive applications in the performance evaluation of the institutes of CAS.