Efficiency in the Brazilian banking system using data envelopment analysis

The objective of this paper is to evaluate bank efficiency in the period from 2012 to 2016 by applying Data Envelopment Analysis (DEA) in a dataset of 37 Brazilian banks provided by the Brazilian Central Bank. We have explored three gaps in research conducted with Brazilian banks by using the intermediation approach to select variables, by analysing the main causes of bank inefficiency and by identifying how inefficient banks in scale can improve their efficiency. Brazilian banks presented an average efficiency of 51.4% for the Charnes, Cooper and Rhodes (CCR) model and 69.8% for the Banker, Charnes and Cooper (BCC) model. The largest banks have performed well in regards to Pure Technical Efficiency (PTE), but failure to operate at the optimal scale level has impaired Technical efficiency (TE), jeopardizing the position of these banks in the efficiency ranking. These banks, in the majority, presented decreasing returns to scale, while the smaller banks had increasing returns to scale. Inefficiency of Brazilian banks is slightly more related to technical and administrative issues than to the scale of operations, although the banks have many opportunities for improvement in this second aspect, especially the larger banks. Ribeirão Preto Bank was the most efficient bank in the group, followed by Cooperativa Sicredi Bank and Alfa Bank. All three banks can be considered small banks. The results indicate that the largest banks are not necessarily the most efficient ones. The efficiency of the sector could be increased if policies were adopted to increase the participation of the smallest banks in the sector, which is currently highly concentrated in the largest ten banks. Government could encourage a dilution in market share of larger banks either through fiscal stimuli among small banks or by fostering mergers and acquisitions.