Research On Performance Evaluation Methods Of Province R&D Based On Improved Network DEA Model And Interval Efficiency

Performance evaluation is the focus of research and development(R&D) management. The main work of performance evaluation of R&D is to select an appropriate evaluation model. While traditional R&D efficiency evaluation index determines that its decision-making unit has the character of complex multistage structure. Using traditional DEA(Data Envelopment Analysis) model(such as CCR and BCC) can only evaluate efficiency under the conditions of "black box", ignoring the impact of internal structure and index. While Network DEA method takes intermediate variable and the own input and output of each stage into account, it’s able to research the efficiency of each DMU under complex structure. The existing centralized model and non-cooperative game model are nonlinear, and these models can’t access to overall efficiency of evaluation system directly. While Traditional NDEA method is short of the symmetry from input and output oriented value, and the consistency of overall conclusion when soving performance problems.This paper chooses Network DEA method as basic model of evaluation. Analysing it’s insufficient in application of performance evaluation of provincial R&D, then improving and optimizing this model. In order to eliminate the inconsistency of output-oriented and input-oriented efficiencies, virtual variables insteads actual data in constraint condition of intermediate variable. Considering the time lag of R&D input to output, this paper uses input index range from 2012 to 2013 instead of fixed value. Combining interval efficiency method for model extension and set up the NDEA interval model, as a new method of performance evaluation of China’s provincial R&D.By applying this model to study the performance evaluation of provincial R&D, it shows that the newly promoted method is effective. Compared with existing models, this model is linear, and it can obtain the overall efficiency directly. Thus it has significant advantages, not only can solve the concrete problems of performance appraisal of R&D activity, but also can be applied to other network structure for the evaluation of the efficiency.