STATISTICAL-INFERENCE AND NONPARAMETRIC EFFICIENCY - A SELECTIVE SURVEY

The purpose of this paper is to provide a brief and selective survey o f statistical inference in nonparametric, deterministic, linear progra mming-based frontier models. The survey starts with nonparametric regu larity tests, sensitivity analysis, two-stage analysis with regression , and nonparametric statistical tests. It then turns to the more recen t literature which shows that DEA-type estimators are maximum likeliho od, and, more importantly the results concerning the asymptotic proper ties of these estimators. Also included is a discussion of recent atte mpts to employ resampling methods to derive empirical distributions fo r hypothesis testing.