How to Bootstrap DEA Estimators: A Monte Carlo Comparison

This paper evaluates the performance of three bootstrap algorithms for the data envelopment analysis (DEA) estimator using a Monte Carlo simulation study. The Löthgren and Tambour (1997) (LT) algorithm; the Simar and Wilson (1997b) (SW) algorithm; and a combination of the LT and SW algorithms (the LSW-algorithm) are considered in the study. The empirical coverage accuracy of bootstrap confidence intervals are simulated under both variable returns to scale (VRS) and constant return-to-scale (CRS) restricted DEA estimators. The results indicate that the LSW-algorithm performs slightly better than the LT-algorithm, which in turn performs better than the SW-algorithm.