Bootstrap Estimation for Nonparametric Efficiency Estimates

This paper develops a consistent bootstrap estimation procedure to obtain confidence intervals for nonparametric measures of productive efficiency. Although the methodology is illustrated in terms of technical efficiency measured by output distance functions, the technique can be easily extented to consistent non parametric frontier models. Variation in estimated efficiency scores is assumed to result from variation in empirical approximations to the true boundary of the production set. The bootstrap procedure also corrects for bias inherent in non parametric efficiency estimators.