Do U.S. Gun Buyback Programs Reduce Either Gun Crime or Firearm Deaths?

This study asked whether U.S. city gun buyback programs actually reduce gun crime, homicides, or suicides. Using data from the National Incident-Based Reporting System (1991–2015) and the National Vital Statistics System, the authors analyzed nearly 700 buyback events across more than 400 cities. They found no evidence that these programs lowered firearm-related crime or deaths. Statistically, they could rule out any decrease in gun crime larger than 1.2 percent in the year following a buyback and 4 percent after one year or more. In some cases, gun crime slightly increased within two months of the program.

The Policy Scientist’s Perspective

The policy question of whether gun buyback programs reduce firearm violence is important because it touches on the broader challenge of identifying evidence-based approaches to reducing gun deaths in a nation with over 390 million privately owned firearms. This study’s findings matter precisely because they question the efficacy of a widely promoted policy tool. Published in the Journal of Policy Analysis and Management—a leading outlet for applied policy research—and authored by well-established economists Anderson, Sabia, and Ferrazares, the article offers a technically rigorous evaluation. The datasets used, including NIBRS and NVSS, are among the best available for measuring firearm incidents and mortality. The authors employ quasi-experimental difference-in-differences and synthetic control methods. Their approach provides credible causal inference under real-world conditions. The large multi-city dataset enhances generalizability, showing consistent null effects across jurisdictions. As one of the few modern, large-scale empirical assessments of gun buybacks, this article constitutes an important and timely contribution to policy evaluation, particularly amid renewed public debate following recent mass shootings.

Full Citation and Link to Article

Ferrazares, T., Sabia, J. J., & Anderson, D. M. (2025). Have U.S. gun buyback programs misfired? Journal of Policy Analysis and Management, 44(4), 1211–1249. https://doi.org/10.1002/pam.70045

Extended Summary

Central Research Question

The article asks whether city-level gun buyback programs (GBPs) in the United States reduce gun-related crime, suicides, or homicides. While gun buybacks are a long-standing and politically popular strategy, their effectiveness as a violence-reduction tool remains uncertain. The authors—Ferrazares, Sabia, and Anderson—seek to provide the first credible causal estimates of these programs’ effects using national data. Their central inquiry is not only whether GBPs produce measurable declines in firearm-related violence but also whether the magnitude of any reduction is large enough to justify continued public investment. They investigate both short- and long-term outcomes, testing whether reductions in crime or death occur in the months immediately following a buyback and whether any effects persist over time.

Gun buybacks are intended to reduce the number of firearms in circulation by offering citizens cash or gift cards to voluntarily surrender their guns. The underlying logic is that fewer guns should translate into fewer opportunities for gun-related harm. However, critics argue that buybacks are unlikely to attract the individuals most likely to commit crimes and may instead collect older or inoperable weapons from low-risk owners. The authors frame their research within this tension, asking whether empirical evidence supports the policy’s intended goals.

Previous Literature

Prior studies of firearm availability and crime have found mixed results, and few have examined buybacks directly. Earlier research in the economics of crime literature consistently explored relationships between gun ownership, regulation, and violence. Duggan (2001) used gun magazine subscriptions and NRA membership as proxies for firearm ownership, finding that higher gun prevalence correlated with higher homicide rates. Cook and Ludwig (2006) and Chalak et al. (2022) used firearm suicide rates as a proxy for gun availability and similarly found positive associations with firearm homicide. Other studies of background checks, waiting periods, and safe-storage laws (Ludwig & Cook, 2000; Anderson & Sabia, 2018) suggest that tighter firearm regulations can reduce certain categories of gun deaths, particularly accidental or youth suicides.

Internationally, Australia’s 1996 National Firearms Agreement (NFA) has been the most studied large-scale buyback. The NFA led to the surrender of more than 640,000 firearms—approximately 20% of the nation’s privately owned guns. Leigh and Neill (2010) found significant declines in firearm suicides, while Chapman et al. (2016) questioned whether those results reflected causal effects, since non-firearm deaths also declined in similar patterns. U.S. research, by contrast, has been sparse and highly localized. Callahan et al. (1994) found no reduction in crime following Seattle’s 1992 buyback, and Braga and Wintemute (2013) reported a 30% decline in shootings after Boston’s 2006 Operation Ceasefire, though that initiative included multiple anti-violence interventions, making causal attribution impossible.

In summary, while prior evidence suggests that policies reducing access to guns can lower violence, buyback programs have lacked rigorous evaluation. Most U.S. studies have been descriptive, limited to single cities, and unable to separate the effects of buybacks from broader enforcement or cultural factors. This study fills that gap by providing a multi-jurisdiction, longitudinal, quasi-experimental analysis of all known city buybacks between 1991 and 2015.

Data

The authors use two primary data sources. The first is the Federal Bureau of Investigation’s National Incident-Based Reporting System (NIBRS), which records detailed monthly data on crimes—including whether a firearm was involved—for roughly one-third of U.S. law enforcement agencies. Their final sample covers 245 agencies serving populations of at least 50,000 people, encompassing 36,516 agency-by-month observations from 1991 to 2015. These data capture both violent and nonviolent gun crimes, distinguishing firearm-related offenses from those involving other weapons or no weapon at all.

The second source is the National Vital Statistics System (NVSS), which provides complete county-level death records by cause, including firearm-related homicides and suicides. The NVSS data extend across the same 1991–2015 period and allow the authors to assess whether buybacks correlate with changes in mortality. To construct a comprehensive inventory of buyback events, the authors compiled data from media archives, legislative histories, and local press reports. They identified 687 buyback programs across approximately 406 cities in 221 counties, including both single- and multi-year initiatives. The average program collected 436 firearms, though most gathered far fewer.

The dataset’s breadth allows for robust generalizability. Because the analysis includes hundreds of cities across multiple regions, it captures considerable diversity in socioeconomic conditions, policing intensity, and gun laws. While NIBRS coverage is incomplete nationally, it includes many large urban jurisdictions where buybacks are most common. The authors also use several control variables—such as unemployment, education, racial composition, policing resources, and other firearm laws—to account for confounding factors that could influence crime trends independently of buybacks.

Methods

The study employs several quasi-experimental techniques designed to approximate causal inference. The core approach is a difference-in-differences (DiD) Poisson regression model, comparing monthly changes in gun-related crime before and after buybacks in cities that conducted them against those that did not. The model includes fixed effects for both the law enforcement agency and time (month-by-year) to account for unobserved heterogeneity and common shocks. Additional controls for demographic and policy variables further strengthen internal validity.

To test the robustness of their assumptions, the authors conduct multiple supplementary analyses. Event-study models examine trends in gun crime up to one year before and after a buyback to test for parallel pre-treatment trends, a key requirement for causal interpretation. They also estimate longer-term (four-year) effects to assess persistence. A difference-in-difference-in-differences (DDD) specification compares changes in gun crimes relative to non-gun crimes, serving as a falsification test to rule out spurious correlations.

Recognizing that treatment effects may vary across cities, the authors incorporate synthetic control and synthetic DiD methods. These approaches construct weighted combinations of non-buyback cities that closely match pre-treatment trends in gun crime for each buyback city, improving causal inference under heterogeneous effects. They also control for potential spillover effects by testing whether nearby jurisdictions without buybacks experienced related changes in gun crime.

While the paper does not use randomized controlled trials—unfeasible for this type of intervention—the combination of DiD, event-study, and synthetic methods provides a strong quasi-experimental framework. The authors explicitly report confidence intervals and robustness checks, enabling readers to evaluate the precision of their estimates.

Findings/Size Effects

Across all specifications, the authors find no evidence that gun buyback programs reduce gun crime, suicides, or homicides. The main Poisson regression results show no statistically significant decline in firearm-related offenses within the first 12 months after a buyback. Their estimates are sufficiently precise to rule out, with 95% confidence, any reduction greater than 1.2% in the year following a buyback and greater than 4.0% one year or more afterward.

In fact, the only statistically discernible pattern is a modest short-term increase in gun crime—approximately 4% during the first two months following a buyback—though this result is small and sensitive to specification choices. The authors interpret this as potentially reflecting behavioral substitution: individuals turning in low-risk or nonfunctional firearms, while criminals face less deterrence if they believe others are disarming. Importantly, the increase does not extend beyond the initial two-month window, and there is no evidence of compensatory rises in non-gun crimes.

Analyses using NVSS death records similarly show null results for firearm-related homicides and suicides. The authors find no measurable decline in deaths involving guns, nor any corresponding increase in non-firearm deaths that might indicate method substitution. The null effects hold across numerous model variants, including synthetic control analyses and event studies that test for delayed or cumulative impacts.

Taken together, the size effects are substantively negligible. Even large-scale or repeated buybacks do not appear to influence firearm crime or mortality rates in a statistically or practically meaningful way. These findings remain consistent across geographic regions and are robust to inclusion of demographic and policy controls.

Conclusion

The authors conclude that U.S. gun buyback programs, as currently implemented, are largely ineffective as tools for reducing firearm-related violence or deaths. Their results suggest that most buybacks fail to meaningfully reduce the stock of functional firearms in circulation or to target individuals at high risk of committing gun crimes. Because many programs pay relatively low amounts—often $50 to $200 per gun—they attract voluntary participants with minimal criminal propensity, leading to negligible changes in firearm supply or accessibility.

The study’s implications are significant for both policy design and empirical research. First, it highlights the importance of rigorous evaluation before scaling or federally funding local buybacks. Second, it demonstrates the feasibility of applying modern quasi-experimental methods to long-standing policy questions previously addressed only through descriptive case studies. Finally, it underscores that well-intentioned but weakly targeted interventions can consume public resources without measurable impact.

From a methodological perspective, the authors’ use of multi-source administrative data, extensive robustness checks, and advanced DiD variants make this one of the most technically credible analyses of gun buybacks to date. Although the absence of randomization limits definitive causal proof, the study meets contemporary standards for observational causal inference. Its findings are broadly generalizable across U.S. jurisdictions with similar program structures.

In the broader policy context, the paper contributes an important corrective to public assumptions about gun control effectiveness. While it does not imply that firearm regulation is futile, it demonstrates that symbolic or poorly designed programs are unlikely to yield measurable safety benefits. In this sense, the article represents both a methodological advance and a substantive contribution to evidence-based policymaking on gun violence.