Point and Interval Estimation in Cross-Sectional Stochastic Frontier Models: The Effects of Sample Size

Abstract

This study utilizes Monte Carlo experiments on simulated data to study the effects of sample size on the empirical accuracy of both point and interval estimates of technical efficiency in cross-sectional stochastic frontier models. Also considered is the robustness of Coelli's (1995) test statistic for the presence of skewness. It is found that sensitivity to sample size varies by model as well as the relative amount of inefficiency present in the data. Furthermore, large amounts of inefficiency are not optimal in terms of interval estimation accuracy. Finally, results indicate that Coelli's asymptotic test statistic is robust in moderately small samples, though performance varies with the underlying distribution of inefficiency.