Road safety performance across local governments: a data envelopment analysis approach.

Road safety performance varies among local municipalities since road safety involves multiple inputs (e.g. facilities and programs) and multiple outputs (e.g. crashes and casualties). A benchmark analysis between local governments is a useful platform to improve their road safety performances because the analysis can specify which other municipalities' processes should be consulted to improve their practices. This study employs the data envelopment analysis approach for benchmark analysis. The input-oriented variable returns to scale model is applied to Korea. The outputs are the numbers of crashes and fatalities, and two sets of indexed inputs comprising institutional managements and governmental interventions are visited. In particular, standardized data formatting between, and controllability by, local governments are highlighted as the required properties for inputs. The overall efficiency scores and the nature of returns to scale reveal the under-investments in road safety in Korea. Efficiency scores by each municipality are also provided. This examination shows the projected values of inputs and outputs as well as benchmarks for each municipality. A sensitivity analysis based on the leave-one-out method generates stable efficiency scores across local governments, which shows the robustness of the model. Finally, the incorporation of environmental variables and dynamic analysis is suggested for further studies.