Nonparametric Efficiency Estimation In Stochastic Environments

This paper develops a new nonparametric model for efficiency estimation. In contrast to Data Envelopment Analysis (DEA), it does not impose debatable production assumptions like free disposability and convexity, and it does not assume that the data are measured without error. The estimators are asymptotically unbiased and have an asymptotic variance that is comparable to that of stochastic frontier estimators (provided the latter use a correct specification of the functional form for the production relationships). In addition, the estimators can be computed using a simple enumeration algorithm.