The Selection of Evaluation Indicators in Data Envelopment Analysis using Logistic Regression

Banks play substantial role in the national economy and its growth and prosperity. In this regard، recent researches have focused on performance evaluation of banks using “data envelopment analysis” (DEA). However، most of these studies has paid less attention to the selection of input and output variables. Obviously، the change in the variables set makes the efficiency scores and assessments of the decision-making units very different. Hence، in this paper، a logistic regression model is used in order to select the input and output variables. Applying this method indicates that the main variables of model are main source of financing as «input variable» and the bank facilities، resource absorption rate and number of bills as «output variables». These are of the greatest impact on forecasting of units efficiency (inefficiency). Then، we dealt with this set of variables to determine technical، allocative، and overall efficiency of 15 branches of Sepah Bank in Tehran during 2011. The results show that only 27 percent of units are 100% efficient، 20% of the units are 100% inefficient، 20% of units are allocatively inefficient and 34% of them are technically inefficient.