The influence curve approach in data envelopment analysis

The behavior of efficiency in the stochastic DEA model is examined here in terms of the influence curve approach which quantifies the influence of observed data on the empirical fit of the production frontier. Data influence is analyzed here through robustness. Two types of robustness and sensitivity issues are analyzed here in terms of (a) a new class of minimax measures, and (b) a set nonlinear efficiency measures and it is shown through empirical applications that in suitable cases these measures outperform the conventional ones.