Assessing the significance of factors effects in Output oriented DEA measures of efficiency:an application to Brazilian Banks

This paper uses output oriented Data Envelopment Analysis (DEA) measures of technical efficiency to assess the significance of technical effects for Brazilian banks. Efficiency measurements are computed for a multiple output (securities, loans, and demand deposits) and for a combined measure of output. The factors or technical effects of interest in the analysis are bank nature (multiple and commercial), bank type (credit, business, bursary and retail), bank size (large, medium, small and micro), bank control (private and public), bank origin (domestic and foreign), and non-performing loans. The latter is a measure of bank risk. Maximum likelihood in the context of the truncated normal distribution, the exponential distribution, and general Tobit models, as well as a nonparametric analysis of covariance, are the statistical tools considered to make inferences on technical effects. The final model chosen is of the Tobit type and is defined by a gamma distribution. The response providing both the most informative content regarding the significance of factors and the best fit to the data is an efficiency measure derived from a multiple input - multiple output product oriented DEA computed under the assumption of variable returns to scale. Bank origin and bank type are the only significant effects