JACKSTRAPPING DEA SCORES FOR ROBUST EFFICIENCY MEASUREMENT

A new approach for Robust DEA technical efficiency measurements is presented, based on a combination of Jackknife and Bootstrap resampling schemes. First, an algorithm implementing jackknife is used to extract leverage for all data points, that is, the impact of the removal of the observed point on DEA calculations performed on the rest of the data set. Next, Bootstrap resampling is implemented, taking into account leverage information. This approach proved to be robust to the presence of outliers and/or errors in the data set, and as it is completely automatic, it is particularly suitable for inspecting huge data sets