Automated Enforcement and Traffic Safety

Abstract

Traffic safety remains a critical challenge for society and public policy. Conventional law enforcement by human police is often cost-ineffective due to information asymmetry and negative externalities of unsafe driving behaviors. Although automated enforcement, in the form of traffic cameras, has been increasingly implemented in recent decades, its impact on road safety and the underlying mechanisms are still debated in the literature. Using a longitudinal dataset of traffic camera installations and police reports of road accidents in a metropolitan city in China, we investigate accident trends near road intersections with and without traffic cameras. We differentiate between advanced cameras, which constantly record video and use machine learning techniques to detect various traffic violations, and conventional cameras, which identify limited violations through temporary image capture triggered by nearby electromagnetic devices. Using an event study design that exploits the staggered installation of traffic cameras, we observe a significant and persistent reduction in total accidents near advanced cameras, compared to locations without cameras or with only conventional cameras. Our analysis reveals three underlying mechanisms: (i) advanced cameras exhibit enhanced technical capabilities in automating violation detection (automated detection effect); (ii) advanced cameras constantly monitor and record road conditions, augmenting accident cause identification (real-time recording effect); and (iii) both capabilities collectively establish technology-enabled deterrence, leading to driver awareness of these cameras and behavioral adjustments that mitigate accident risks (driver learning effect). This work contributes to the fields of information systems, transportation economics, and criminology, providing policy insights into the effective implementation of automated enforcement to improve traffic safety.

Keywords: Automated Enforcement, Traffic Safety, Deterrence, Event Study

JEL Classification: M15, R41, O18, I18