Efficiency of Queensland public hospitals via spatial panel stochastic frontier models

Abstract We adapt a range of spatial econometric models to the four-component error term panel stochastic frontier framework to estimate the inefficiency of public hospitals in Queensland, Australia, and to investigate different channels of spatial effects in hospital performance. Our results demonstrate a statistically significant presence of the spatial dependence from the autoregressive dependent variable and the autocorrelated error term. Additionally, we observe a positive spillover effect of input factors, as well as some impacts from accounting for the spatial dependence on the inefficiency estimation. Specifically, the resulting inefficiency estimates from the spatial models turned out to be higher than those from the non-spatial model, yet the magnitude of difference is relatively modest, confirming the approximate validity of the non-spatial stochastic frontier approach for this dataset.