Automated Enforcement and Traffic Safety

Traffic safety has been a persistent societal and public policy challenge. Conventional law enforcement by human police is often cost-ineffective due to the information asymmetry and negative externality that arise from unsafe driving behaviors. Recent decades have seen a rise of automated enforcement (in the form of traffic cameras), yet there has been limited understanding of its impact on road safety. Using a longitudinal dataset of traffic camera installations and police reports of road accidents in a metropolitan city in China, we study accident trends near the road intersections where traffic cameras are present. We analyze two types of cameras: advanced ones (that detect various traffic violations via constant video capture and analytics) and conventional ones (that detect limited violations via temporary image capture). Employing an event study design, we observe a significant and persistent decline in accidents near advanced cameras, compared to the locations without cameras or only with conventional cameras. We examine the underlying mechanisms and obtain supporting evidence: First, advanced cameras augment violation detection by capturing more violations (enforcement augmentation). Second, they constantly monitor and record traffic conditions, enhancing accident cause identification (evidence gathering). Finally, both capabilities gradually give rise to technology-enabled deterrence, such that the presence of cameras can deter risky driving and reduce accidents, regardless of whether such cameras are capable of detecting relevant traffic violations. This work contributes to the intersection of information systems, transportation economics, and criminology; and it offers crucial policy insights into the effective implementation of automated enforcement for traffic safety