Bayesian Stochastic Frontier Models under the Skew Normal Settings

Abstract

Recently, a skew normal-based stochastic frontier model has emerged as a promising tool for efficiency analysis. In this paper, a Bayesian framework for statistical inference is presented, incorporating both informative and non-informative prior knowledge. The efficacy of the Bayesian approach is evaluated through rigorous examination using both simulation data and real data from a manufacturing productivity study. A comprehensive comparison with the conventional maximum likelihood approach is conducted. Results from both simulated and empirical investigations unequivocally demonstrate the superior performance of the Bayesian methodology.