A Spatial Autoregressive Combination Stochastic Frontier Model Separating Spatial Dependence in the Frontier and in Technical Inefficiency

Abstract

We identify the attributes and limitations of spatial autoregressive stochastic frontier models, spatial inefficiency error stochastic frontier models, and straightforward integrated models. We also propose a separable spatial autoregressive combination stochastic frontier model that overcomes the limitations. The proposed model’s main features are as follows: (a) spatial dependence in the frontier and in technical inefficiency are separately identified; (b) a negative spatial autocorrelation in technical inefficiency is permitted; (c) technical inefficiency follows a truncated-normal distribution.