Robust DEA for banks' efficiency under uncertainty

The Data Envelopment Analysis (DEA) has been the benchmarked model for measuring the efficiency of banks over the years. However, inherent noise and uncertainties in the data are hardly considered for robust efficiency scores. The disadvantage is that a small perturbation in the uncertain parameters can lead to high infeasibility of the efficient solutions. This paper introduces a robust DEA into the measurement of banks efficiency. The proposed robust approach is based on the robust counterpart optimization of Ben-Tal & Nemirovski (2000), and it is implemented in the traditional DEA models germane to the performance measurement of banks. A preliminary result from data on banks in the Czech Republic indicates that efficiency scores measured with the robust DEA model provides a true and stable performance measure than the normal DEA model.