Two-Stage DEA: An Application in Banking

Abstract

Measuring and improving the efficiency of the European banks have recently faced increasing interest. Utilising the recent development in data envelopment analysis (DEA), this study examines the efficiency of global systematically important banks via a two-stage production process. The two-stage DEA separates efficiency of the own sources use phase and the phase of deposits transformation into the earning assets. As we can divide the entire operational process into two sub-processes (deposit producing and earning assets producing), we can reveal substantial inefficiency in both dimensions, and assist in identifying the inefficiency sources. Methodology: In this study, we utilise the dynamic network DEA approach to disaggregate, evaluate, and test the efficiency of 25 European global systematically important banks during the period 2010–2017 with the variable returns to scale setting. We also use Li test to determine whether there exist significant differences between results. Results: 1. We have found out that in identifying the bank’s inefficiency the dynamic network DEA model is more effective than the conventional black-box DEA model as it can optimise both sub-processes simultaneously. The DNDEA allows resources to be allocated over time to maximise the production of desirable outputs and simultaneously minimise the production of undesirable outputs. 2. The results pointed to the fact that the inefficiency of the European global systematically important banks primarily results from the inefficiency of their earning assets producing sub-process. While at the beginning of the analysed period the difference in efficiency was significant, from 2014 the efficiency of the earning assets producing sub-process is gradually approaching the efficiency of the deposit producing sub-process. The overall efficiency of analysed banks was relatively stable with slight decrease in 2014; that was affected with the decrease in both sub-stages, mainly due to banks which in 2014 increased volume fixed assets and decreased the value of the deposit, while increasing their non-performing loans in the previous year. The results also show that within the group of banks with the lowest efficiencies in both sub-processes were banks whit the highest average number of employees, the highest average value of fixed assets and the highest average value of non-performing loans. On the other hand, within the group of banks with the highest efficiencies in both sub-processes, we can find banks with a below-average number of employees and fixed assets and with the highest average value of deposits, loans and other earning assets.