Efficiency spillovers in Bayesian stochastic frontier models: application to electricity distribution in New Zealand

This paper proposes a spatial Bayesian random effects stochastic frontier model that allows for unobserved heterogeneity and spillovers between firms' efficiencies with an exogenous spatial weights matrix. Proposals for efficiency measurement in the spatial context add to the debate in the literature. The approach shows good small-sample performance, which is very relevant for applied researchers, and explores guided walk metropolis as a simple and computationally efficient alternative to classic rejection techniques. The approach is applied to a sample of 28 New Zealand electricity distribution firms between 1996 and 2010, finding spatial dependence with a second-order contiguity matrix.