A new stochastic frontier model with cross-sectional effects in both noise and inefficiency terms

This paper develops a new stochastic frontier model that allows for cross-sectional (spatial) correlation in both the noise and inefficiency terms. The model proposed is useful in efficiency analyses when there are omitted but spatially-correlated variables and firms benefit from best practices implemented by other (adjacent) firms. Our model is inspired by Wang and Ho (2010) as it also possesses the scaling property, though the transformations performed are cross-sectional (spatial) in nature. A nice feature of the proposed model is that it can be estimated by maximum likelihood using standard software. The model is illustrated with an application to the Norwegian electricity distribution sector. We first detect positive global spatial autocorrelation in the noise term across firms. In contrast, the spatial spillovers in firms’ inefficiency are local as they are mainly generated between bordering firms. Finally, we find that most of firms’ inefficiency relies on own factors and that the inefficient firms can better manage their improvements in efficiency than more efficient firms.