How Do Recreational Cannabis Legalization Laws Affect Racial Disparities in the Criminal Legal System?

The authors ask whether state recreational cannabis legalization reduces long-standing racial disparities in the criminal legal system. Using national data from 2007–2019 on arrests, prison admissions, hospitalizations, crimes, and police staffing, they track outcomes for White and Black adults before and after legalization. They find large drops in cannabis possession arrests (down 62% for White adults and 51% for Black adults) and cannabis sales arrests (down 44% and 49%). Sales arrests for other illegal drugs also fall modestly. Serious violent and property crimes do not increase, but racial disparities in overall criminal justice outcomes remain even after legalization.

Why This Article Was Selected for The Policy Scientist

The article examines how recreational cannabis legalization affects racial disparities in arrests, incarcerations, and related criminal-justice outcomes. The use of large administrative datasets from multiple systems enhances measurement quality, and although generalizability is strongest for early-adopting states, the data cover substantial geographic and temporal variation. The study employs contemporary difference-in-differences methods. The findings—large reductions in cannabis arrests for all groups, limited spillovers on other crimes, and only partial narrowing of racial disparities—represent a meaningful incremental extension of prior literature.

Full Citation and Link to Article

Meinhofer, A., Rubli, A., & Cunningham, J. P. (2025). State recreational cannabis laws and racial disparities in the criminal legal system. American Economic Journal: Economic Policy. Advance online publication. https://doi.org/10.1257/pol.20230670

Central Research Question

The central research question examines whether state recreational cannabis legalization reduces or reshapes longstanding racial disparities—especially Black–White disparities—in arrests, incarcerations, and related criminal legal system outcomes. The authors also ask whether legalization generates broader spillover effects, including shifts in violent and property crime, other drug offenses, hospitalizations involving cannabis or other drugs, and the size and composition of local police forces. A second set of questions concerns mechanisms: whether changes are attributable to reduced criminalization of cannabis, changes in illegal drug markets, changes in patterns of drug use, or adjustments in law enforcement behavior. Together, the study seeks to identify the causal effects of legalization on both direct cannabis-related outcomes and spillover consequences that may widen, narrow, or leave unchanged racial disparities across multiple domains of the criminal legal system.

Previous Literature

Existing work provides substantial evidence that recreational cannabis laws reduce cannabis possession arrests overall, increase legal-market activity, and have limited effects on violent or property crime. Prior studies, however, generally focus on the aggregate population and rarely evaluate differential effects by race beyond cannabis possession. A small number of papers show reductions in cannabis possession arrests for both Black and White individuals following legalization or decriminalization, but the literature has not systematically explored spillover effects into other offenses—drug-related or otherwise. The authors situate their study as filling three gaps: (1) the need for a comprehensive evaluation of racial differences across a wide spectrum of criminal justice outcomes; (2) the need to examine spillovers on hospitalizations, violent and property crime, and illegal drug market activity; and (3) the need to bring higher-quality administrative data and more states into the analysis. The study also builds on prior work using staggered difference-in-differences approaches to evaluate cannabis reforms, referencing widely cited studies of crime, policing, and drug use that have shaped the policy debate. The authors frame their contribution as extending and generalizing existing findings with better data and broader outcomes, including outcomes not previously analyzed at scale.

Data

The analysis draws on several large administrative datasets widely used in empirical criminology and health policy. Arrest data come from the FBI Uniform Crime Reports (UCR) for 2007–2019, capturing arrests by race, offense type, and county. The data are aggregated to county-year cells and include cannabis possession, cannabis sales, other drug possession, other drug sales, violent offenses, property offenses, and Part 2 offenses. Incarceration data come from the National Corrections Reporting Program (NCRP) and the National Prisoner Statistics (NPS), both providing race-specific counts of admissions and year-end prison populations by offense category. To examine the incidence of criminal activity separately from enforcement behavior, the authors collect call-for-service and reported crime data from nine major RCL-state cities that provide geo-coded incident-level public data. These data allow neighborhood-level comparisons between predominantly Black and predominantly non-Black census tracts. Hospitalization data come from the Healthcare Cost and Utilization Project (HCUP-SID) and state-shared inpatient discharge records, providing state-by-race counts of hospitalizations involving cannabis use disorder, cannabis poisoning, and illegal drug use or overdose. Mortality data come from the National Vital Statistics System, enabling measurement of homicides—including those involving firearms—by race. Finally, law enforcement staffing counts come from the LEOKA dataset, which provides annual data on sworn officers and civilian personnel. Across these sources, the authors assemble a large multi-state panel with race-specific, offense-specific, and time-specific coverage appropriate for causal inference using variation in legalization timing.

Methods

The empirical strategy uses two-way fixed-effects difference-in-differences (DID) models, exploiting staggered adoption of recreational cannabis laws across eleven states between 2012 and 2019. The authors estimate static and dynamic effects using canonical specifications with jurisdiction and time fixed effects, population weighting, and controls for cannabis decriminalization laws. Event-study models assess parallel pre-trends and the timing of post-legalization effects. Because DID estimators can be biased when treatment timing varies, the authors perform tests for negative weighting and report robustness checks using heterogeneity-robust DID estimators (Sun–Abraham, de Chaisemartin–d’Haultfoeuille, and Borusyak–Jaravel–Spiess). They also estimate models using dispensary opening dates instead of legalization dates, impose stricter UCR reporting-coverage thresholds, and test for spatial spillovers from neighboring RCL states. In city-level analyses, they use tract-level event-study models comparing minority and non-minority neighborhoods within treated states. Across all outcomes, standard errors are clustered at the state level (the treatment level). The methods do not employ randomized controlled trials or natural experiments stronger than standard DID, but the authors attempt to approximate causal inference using the best available national data and widely accepted quasi-experimental techniques.

Findings/Size Effects

Recreational cannabis legalization produces sharp, consistent reductions in cannabis-related arrests: possession arrests decline by 62 percent for White individuals and 51 percent for Black individuals; sales arrests decline by 44 percent and 49 percent, respectively. Absolute Black-White disparities shrink substantially, but relative disparities change little because arrest rates fall proportionally. Spillovers to other drug categories vary: arrests for other drug possession remain unchanged, but arrests for other drug sales fall by 22 percent for White individuals and 17 percent for Black individuals, reducing absolute disparities in drug enforcement. No increases are observed in arrests for violent or property offenses, and point estimates rule out meaningful crime-increasing effects. Evidence for low-level Part 2 offenses is mixed: estimates suggest an 11 percent increase in Part 2 arrests for Black individuals, but pre-trends in California complicate interpretation. Incarceration results show declines in drug-related admissions only for White individuals (a 34 percent decline relative to baseline), with no change for Black individuals. On the behavioral side, hospitalizations involving cannabis use disorder and cannabis poisoning increase for both racial groups, while hospitalizations involving other illegal drugs rise significantly for Black individuals. Homicides show no increases and possibly decline modestly for Black individuals, driven by reductions in gun-related homicides in a subset of states. Calls-for-service and city-level crime measures suggest that drug-related incidents fall in minority neighborhoods after legalization, with little change in violent or property crime. Police staffing increases slightly—about 7 percent in total officers—with civilian personnel rising by 16 percent, consistent with reallocation of resources after legalization.

Conclusion

The study concludes that recreational cannabis legalization substantially reduces arrests for cannabis-related offenses across racial groups and modestly narrows absolute racial disparities in the criminal legal system without causing increases in serious crime. Some spillovers—such as reductions in other drug sales arrests—extend beyond cannabis and appear consistent with changes in drug markets or policing priorities. Other spillovers, including increases in cannabis-related hospitalizations or mixed evidence on low-level offenses, highlight areas in which legalization alone does not resolve racial disparities and may interact with pre-existing inequities in complex ways. Because relative disparities remain largely unchanged and incarceration declines accrue disproportionately to White individuals, the authors argue that legalization is insufficient to eliminate longstanding racial inequities without complementary policies addressing criminal history, record clearing, and structural conditions that shape law enforcement behavior. The data support generalizability to states with similar demographic and policy environments, although the authors caution that most early-adopting RCL states differ from later adopters in political climate and baseline racial composition. The study’s contribution is to generate the broadest, most systematic evidence to date on racialized impacts of legalization, using extensive administrative datasets and modern DID methods.