Multi-output efficiency and operational safety: An analysis of railway traffic control centre performance

Transportation service providers are under increasing pressure to raise cost efficiency without sacrificing safety. We show the usefulness of a nonparametric multi-output framework to both monitor staff efficiency and detect operational safety concerns. To realistically model input-output relations at an hourly rate of Belgian computerized railway traffic control centres, we apply a Data Envelopment Analysis (DEA)-based framework with proportional cost allocation restrictions and introduce output-specific metafrontiers. Our analysis (covering each single hour of the complete year 2015) shows that production tasks with a highly variable work load, when characterized by binding cost allocation restrictions and high within-traffic-control-centre efficiency, are more prone to human error. Further, we show how DEA combined with disaggregated data can be used to a priori assess staff schedule changes.