Does Granting Driver’s Licenses to Undocumented Immigrants Increase Fatal Car Crashes?

This study asks whether granting driver’s licenses to undocumented immigrants affects the number of fatal car crashes in the United States. Using county-level data from 2007 to 2018 from the U.S. Department of Transportation’s Fatality and Injury Reporting System, the researcher analyzed how license reforms influenced crash rates. The results show that allowing undocumented immigrants to obtain licenses increased total fatal crashes by about 5%, or roughly 0.46 more deaths per county per year. The increase was strongest in states with larger undocumented populations and was mainly linked to risky driving behaviors rather than higher traffic volumes.

The Policy Scientist's Perspective

This article has broad relevance because it connects immigration policy and transportation safety —areas central to how states balance social inclusion with risk management. The study’s data—county-level fatal crash records from 2007–2018—are comprehensive and well-constructed. Its quasi-experimental difference-in-differences design provides credible causal inference. Findings indicating a roughly 5% rise in fatal crashes are statistically and substantively meaningful. They refine earlier state-specific research by extending coverage across multiple jurisdictions, making the results more generalizable. The paper is timely given ongoing state-level debates over licensing access and road safety implications, and it stands as one of the most empirically consequential contributions published in the past month on this topic.

Full Citation and Link to Article

Zhao, R. (2025). "The impact of granting undocumented immigrants driver’s licenses on fatal crashes". Journal of Policy Analysis and Management, 1–25. https://doi.org/10.1002/pam.70053

Extended Summary

Central Research Question

This study investigates whether granting driver’s licenses to undocumented immigrants affects the frequency of fatal motor vehicle crashes in the United States. Specifically, the author seeks to estimate the causal impact of state-level driver’s license reforms (DLR) that, beginning in 2014, permitted undocumented residents to obtain driver’s licenses. The paper asks whether these reforms led to measurable changes in the number and type of fatal crashes and explores potential mechanisms underlying any observed effects, such as changes in risky driving behavior or moral hazard. The research is motivated by growing state-level policy variation and an ongoing national debate over the public safety and economic consequences of extending legal driving privileges to undocumented populations.

Previous Literature

Prior studies on undocumented immigrants and public safety have produced mixed findings. Earlier work emphasized that undocumented populations tend to exhibit lower felony and arrest rates than U.S.-born citizens or legal immigrants (Light et al., 2020; Light & Miller, 2017). Other studies focused on economic and social impacts, suggesting that undocumented immigration can contribute positively to local economies and tax revenues while potentially exerting modest downward pressure on low-skilled wages (Borjas, 2003; Mayda & Peri, 2017).

Research specifically examining driver’s license reforms is limited but instructive. Lueders, Hainmueller, and Lawrence (2017) evaluated California’s Assembly Bill 60 (AB60), which took effect in 2015, and found that extending licenses to undocumented residents reduced hit-and-run accidents without increasing overall crash rates. Their analysis, however, was restricted to a single state and a short time frame, potentially missing longer-term behavioral changes. Other related studies identified that such reforms increased insurance coverage and reduced uninsured driving (Churchill et al., 2021) and improved undocumented immigrants’ employment opportunities and car ownership (Cho, 2022; Amuedo-Dorantes et al., 2020).

This paper extends the literature in three ways. First, it examines multiple states and a longer time period (2007–2018), allowing for broader inference and assessment of policy effects that may emerge gradually. Second, it relies on national data from the U.S. Department of Transportation rather than state-level reporting systems, ensuring comparability and consistency across jurisdictions. Third, it explores behavioral mechanisms—particularly moral hazard and the Peltzman effect—to explain how new license eligibility might lead to increased risk-taking once legal penalties and financial liabilities are reduced. In doing so, the study links transportation safety, behavioral economics, and immigration policy in an integrated causal framework.

Data

The study uses county-level panel data from 2007 through 2018, drawn primarily from the U.S. Department of Transportation’s Fatality and Injury Reporting System Tool (FIRST). This database combines information from the Fatality Analysis Reporting System (FARS) and the Crash Report Sampling System, providing a census of all motor vehicle crashes in the United States that resulted in at least one fatality within 30 days of the incident.

The sample includes data for counties in 49 states and the District of Columbia. Vermont was excluded from the main analysis because of its very small undocumented population and partial compliance with federal immigration enforcement requests, which could reduce the number of undocumented residents affected by the policy. The study focuses on fatal crashes rather than all crashes, arguing that fatal incidents impose the greatest social and economic costs and are more consistently and reliably recorded across jurisdictions.

Additional covariates include county-level measures of population, unemployment, and road expenditures, as well as state-level indicators such as car fuel taxes, vehicle miles traveled (VMT), and licensed drivers per 1,000 residents. These variables were drawn from the Bureau of Economic Analysis, U.S. Census, and Bureau of Transportation Statistics. The identification of states that implemented DLR between 2007 and 2018 came from the National Conference of State Legislatures. By the end of the study period, 11 states and the District of Columbia had enacted reforms permitting undocumented immigrants to obtain licenses, while 38 states had not.

The average county experienced 11.35 fatal crashes per year during the sample period, 95% of which were non–hit-and-run events. Approximately 27% of these crashes involved speeding. Other summary statistics—such as road expenditures per capita ($372 on average), fuel taxes ($137 on average), and carpool usage (9.6%)—suggest significant variation across states and counties, supporting the need for fixed effects modeling.

Methods

The author employs a generalized difference-in-differences (DiD) design that leverages the staggered timing of DLR implementation across states. This quasi-experimental approach compares changes in fatal crash rates in states that adopted DLR (treated states) to those that did not (control states), before and after reform adoption. The model controls for county and year fixed effects, thereby accounting for unobserved time-invariant differences between counties and national time shocks affecting all areas.

The key estimating equation regresses the log of (fatal crashes + 1) on an indicator for DLR adoption and a vector of time-varying county and state-level covariates. Robust standard errors are clustered at the state level. The paper also estimates event-study models to test for pre-trends and assess dynamic effects over time, allowing the author to confirm the parallel trends assumption and evaluate how treatment effects evolve in the years following reform.

To strengthen causal claims, the author employs multiple robustness checks. These include alternative transformations (inverse hyperbolic sine), count models (Poisson and negative binomial), and specifications that incorporate state-specific linear pre-trends following Goodman-Bacon (2021). The study also adopts the interaction-weighted event study estimator of Sun and Abraham (2021) to address potential contamination in traditional two-way fixed effects models.

The author further tests for heterogeneous effects by the share of undocumented immigrants in a state, investigates whether increases in crashes were concentrated on interstate highways (to assess spillovers from other states), and examines whether population migration from control to treated states confounds results. Placebo tests include random assignment of reform dates and analysis of crash types unlikely to be affected by the policy (e.g., commercial truck crashes).

Findings/Size Effects

The study finds that granting driver’s licenses to undocumented immigrants increases total fatal crashes by approximately 4.9% after controlling for covariates, or about 0.46 additional deaths per county per year. The estimated effect is strongest in states with larger undocumented populations. The effect appears gradually, growing from statistically insignificant levels in the first year to roughly 6.5% by the third year following reform.

Importantly, the increase in fatal crashes is not driven by higher traffic volume. Although the number of licensed drivers rose by 2.5% after reform, vehicle miles traveled did not significantly change, suggesting that many undocumented immigrants were already driving before legalization. Instead, the increase appears linked to behavioral changes consistent with moral hazard and the Peltzman effect. Once licensed, undocumented immigrants face lower expected legal and financial costs from accidents. With valid licenses and auto insurance, they are less likely to risk arrest or deportation for minor infractions, reducing the perceived cost of risky driving behaviors. Supporting this interpretation, the study finds significant increases in speeding-related fatal crashes, which account for more than half of the total rise in fatalities.

Additional analyses show that the effect is concentrated in non-pedestrian and non–hit-and-run crashes. Pedestrian-involved fatalities do not change significantly. Nor is there evidence that cross-state travel or interjurisdictional migration biases the results. Placebo tests confirm that the observed effects are unlikely due to chance, and analyses of commercial-vehicle and motorcycle crashes show no corresponding increase, further suggesting that the impact is specific to newly licensed, non-commercial drivers.

The magnitude of the effect is economically meaningful. Given that there are over 3,000 counties nationwide, the estimated increase translates to roughly 1,400 additional fatal crashes per year if all states adopted similar policies. The author acknowledges that this cost must be weighed against potential benefits such as improved insurance coverage, labor market access, and reductions in unlicensed driving, though these are not quantified in the paper’s cost-benefit section.

Conclusion

The paper concludes that driver’s license reforms for undocumented immigrants led to a measurable increase in fatal crashes, particularly over time and in states with larger undocumented populations. The results are consistent with the hypothesis that reduced legal and financial risks alter driving behavior, leading to modest but statistically significant increases in risk-taking.

The findings stand in contrast to earlier research that reported no increase—or even a reduction—in crashes, suggesting that short-term evaluations may understate longer-run behavioral adjustments. By incorporating a longer timeframe and broader state coverage, the study provides a more complete picture of the policy’s effects.

The dataset is of high quality and national scope, enhancing both reliability and comparability. While the DiD approach offers credible causal inference, it cannot perfectly replicate the internal validity of a randomized controlled trial. Still, the author’s extensive robustness checks, event-study analyses, and treatment of heterogeneity provide strong support for causal interpretation. The generalizability of the results is plausible across U.S. jurisdictions with similar undocumented populations and driving conditions, though less certain internationally, given differences in immigration enforcement and road safety institutions.

Overall, the study represents a careful and technically strong contribution to the policy literature on immigration and public safety. It expands the empirical foundation for assessing how inclusive licensing policies affect not only labor market participation and insurance coverage but also broader social outcomes such as roadway safety.