WORKING PAPERS
Driver Licensing Stringency and Fatal Crashes: Triple-Difference Evidence from U.S. States
Abstract: This paper examines the effect of driver licensing stringency on fatal road accidents using comprehensive policy data covering 50 U.S. states over 1994-2022. I develop a novel, holistic measure of licensing stringency using factor modeling and employ a triple-difference approach exploiting variation across states, time, and driver age. Results reveal a trade-off between reduced mortality and licensing access, with increased stringency reducing fatal crashes by 4.5-5.5 per 100,000 licensed drivers but also reducing young driver licensing rates by 7.7 percentage points. Comparing population-based and licensed driver-based rates suggests policies both screen out high-risk drivers and improve behavior among the licensed. Insurance premiums, however, remain largely unaffected by the licensing policies most effective in reducing fatal accidents.
Full paper available on request (PDF; slides also available).
WORKS IN PROGRESS
Age-Based Crash Affinity Analysis: A Classical Collision Theory Framework
Abstract: This paper employs a classical collision theory framework to analyze age-based crash formation patterns using comprehensive fatal crash data covering all 50 U.S. states from 1994 to 2022. I decompose crashes into age-specific crash rates and affinity effects that measure systematic deviations from random mixing patterns. The analysis reveals three key findings. First, all age groups have become substantially safer over time, with seniors experiencing the largest improvements (15.6% reduction by 2022) and young drivers showing notable improvements after 2006, coinciding with graduated driver licensing (GDL) policy implementation. Second, drivers exhibit strong same-age clustering in crashes, with teen-teen interactions showing the highest affinity (0.0110, p<0.01), followed by young adult-young adult (0.0092) and senior-senior (0.0072) interactions. Third, teen crash risk operates primarily through multi-vehicle interactions rather than single-vehicle incidents, with teens showing significantly lower single-vehicle crash rates than expected (-0.0013, p<0.05). These patterns have important policy implications: the strong within-group affinities indicate that age-targeted interventions can generate multiplicative safety benefits by disrupting high-frequency crash combinations. For example, GDL policies that reduce teen driving exposure disproportionately reduce teen-teen crashes—the combination with the highest affinity—yielding safety improvements that extend beyond direct effects on teen crash rates. The temporal stability of affinity coefficients supports their use as structural parameters for projecting the effects of demographic shifts and targeted interventions on aggregate crash outcomes.