Research On Cross Efficiency Evaluation Approaches And Applications Of Network Systems

With the increasingly fierce market competition and wave of global economy, theeconomic agents are becoming more networked, systematic and refinement. Theinternal structures and operation mechanisms are becoming more and more importantfor evaluating the performance of these economic agents, which will affect the socialand economic development level. Therefore, performance evaluation of theseeconomic agents is quite important. As a new performance evaluation approach,network DEA can asses the efficiencies of the whole system and its subsystems.However, there still exist some limitations, such as relying too much on self-appraise,which will affect the objectiveness and justness of evaluation. Therefore, this thesisintroduces the concept of cross efficiency into performance evaluation of networksystems, and develops several models to evaluate the performance of serial systems,parallel systems, and complex network systems by combining self-appraisal withpeer-appraisal. The proposed models are then used to evaluate the performance offinancial organizations.This thesis firstly reviews the two classic DEA models, and introduces networkDEA model for series systems, parallel systems and complex network systems. Then,several cross efficiency models for three kinds of network systems by using theaggressive, benevolent and neutral strategies from the perspective of competition andcooperation relationships among decision making units. Finally, the empiricalefficiency analysis of banking and insurance organizations are carried out to illustratethe effectiveness of proposed approaches. The empirical results show that theaggressive, benevolent and neutral cross efficiency models can cure some limitationsof traditional network DEA models, such as extreme weights, too many zerocoefficients. Meanwhile, the difference of efficiency scores under different strategiesis small, which indicate that they are very consistent in evaluating the performance ofnetwork systems..In a word, by combining self-appraisal with peer-appraisal, network crossefficiency DEA models can cure some limitations of network DEA models. Theproposed models can reflect the performance of network systems more justness andobjectivness and provide some information on the performance improvement ofnetwork systems, which is helpful for social and economic development.