Do Transportation Network Companies (TNCs) improve or worsen road safety? Assessing crash impacts in San Francisco

Abstract

Introduction: Over the last decade, Transportation Network Companies (TNCs) have emerged as a popular mobility option. Yet past research has not conclusively established the linkage between their service operations and road crash risk. Our analysis compares road crash outcomes in San Francisco between 2010, when TNCs were negligible, and 2016, when they comprised 15% of vehicle trips within San Francisco. Method: We estimated a fixed-effects Poisson regression model for four crash outcomes, controlling for the change in background traffic, speed, and time-invariant factors including roadway geometry. Results: Our results show that TNCs have three competing influences on crashes: (a) they increase total vehicle miles traveled (VMT) and thus crashes; (b) VMT on TNCs themselves is associated with fewer crashes, potentially due to a newer vehicle fleet or the selection of drivers; and (c) TNC pick-ups and drop-offs are associated with more crashes, potentially due to more conflicts at the curb. We apply these models to estimate the number of crashes expected from a counterfactual 2016 scenario without TNCs and compare that to the with-TNC scenario to isolate the total effect of TNCs. This model application reveals that TNCs are associated with a 4% decrease in total crashes. We also find that TNCs are associated with a slight reduction in fatal and injury crashes, crashes involving a bicyclist or pedestrian, and alcohol-involved crashes, but the results are not statistically significant for these crash types. Practical applications: By disaggregating TNC services into distinct operational components, our research provides valuable insights for urban transport planners and policymakers seeking to maximize the safety benefits of emerging transportation technology while mitigating their associated risks.