Multi‐stage stochastic frontier analysis for simple networks

Abstract

We develop a method for modelling multi‐stage production using stochastic frontier analysis. This approach is suitable for the analysis of costs or output where intermediate outputs become inputs into a subsequent stage of the production process, either within an organisation or in the form of a supply chain. Our focus is on higher education institutions in England, and the purpose is to assess the performance of our novel methods using Markov Chain Monte Carlo methods. Without taking into full account of the complexity of the ‘network’, key decisions cannot be made regarding intake quality, student/staff ratios, per‐student spending or academic reputation (the last of which involves costly decisions in terms of academic openings and the profile of candidates desired for any given university).