The evolution of safety-adjusted transportation efficiency for the road system in China

Traffic accident is a grievous problem that costs more than one million lives worldwide every year, but remains understudied in transportation efficiency literature. This paper develops safety-adjusted transportation efficiency to account for the negative outcomes in transportation including accidents, fatalities, injuries and property loss. We model the transportation efficiency under the data envelopment analysis (DEA) framework by treating the accident-related negative outcomes as undesirable outputs. Two DEA models, based on radial and non-radial structures respectively, are proposed for panel data. We apply the methods to 31 provinces in China over a 20-year horizon 1998-2017. We find that the evolution of China's overall safety-adjusted transportation efficiency follows a U-shaped path: It deteriorated between 1998 and 2002, steadily improved from 2002 to 2012, and stabilized during 2012-2017. The majority of the provinces improved their safety-adjusted transportation efficiency from 1998 to 2017, except for one province that maintained the status quo and three provinces that experienced a decline in performance. Improvement analysis is carried out to identify gaps in accident-related factors that each province should close to attain best-practice. Further, we find strong evidence of unconditional β-convergence and σ-convergence in safety-adjusted transportation efficiency, indicating that the provinces with low initial efficiency generally grew more rapidly and the dispersion of provincial efficiency levels diminished. The main findings are substantially different from the regular transportation efficiency analysis that does not consider the accident-related undesirable factors. The safety-adjusted transportation efficiency can convey important information that the regular transportation efficiency fails to capture.