Screening effects of China's driver's license test reform: evidence from regression discontinuity design

Driver's license testing determines who may legally operate a motor vehicle and is therefore a central screening instrument for road safety. This study develops a screening-theoretic framework for driver licensing and uses microscopic traffic accident data from Guangdong Province, China, to evaluate the 2004 reform of China's driver's license test system. We apply a regression discontinuity design based on license issuance dates, using January 1, 2005 as the empirical cutoff that reflects the transition between policy implementation and license issuance. The results show that drivers licensed under the new test regime received, on average, 0.36 fewer demerit points for traffic violations in accidents. Mechanism analyses show that the reduction is concentrated in Type A (risk-judgment-related) violations, whereas Type B (skill-related) violations do not show a comparable reduction. This pattern is consistent with a screening-based interpretation in which the reform affected traits related to risk judgment more strongly than operational driving skills. The effect is most pronounced among drivers with no more than six years of driving experience and among younger and urban drivers. Heterogeneity analyses by vehicle, road, weather, and lighting conditions further suggest that the estimated reform effect varies across driving contexts, while estimates for small subgroups should be interpreted cautiously. These findings provide evidence on how test-based licensing reforms may affect driver violation behavior in comparable institutional settings.