banks' evaluation variables identification and reduction in the data envelopment analysis context using structural equation modeling

Data envelopment analysis method is one of the most widely used methods to performance evaluation of banks. According to extensive studies in the field of bank performance evaluation, various data envelopment analysis models such as basic CCR, BCC models, multi-stage and network models have been presented. According to the models and developments, have been used various and numerous variables. In this research, first, the types of variables in the bank’s evaluation were identified and categorized. On the other hand, the presence of many variables for evaluation in the context of data envelopment analysis would violate the general rule that confirmed relation between decision maker units and evaluation variables. Therefore, an a priori approach was used to reduce the variable before implementing the data envelopment analysis model. The approach was used based on structural equation modeling and confirmatory factor analysis, which was analyzed using AMOS and SPSS software. Elementary analyze result show that model fit indices not confirmed and some of variables non- significance at 90% confidence level and have low factor loading. Therefore, Due to the non-significance of a number of variables, the variable was adjusted. The results of the modified model analysis confirmed it. Using this approach, extracted the effective and key inputs and outputs that capable to make distinguish between banks performance. Finally, model variable reduced from 22 to 8. As a result, this approach can be applied for identification key variables from variable lake in the banks evaluation and other evaluation system with many indicators.