Do Collective Bargaining Rights for Police Increase Civilian Deaths?

Cunningham and Gillezeau (2025) ask whether granting collective bargaining rights to police unions increases civilian deaths caused by law enforcement. They analyze county-level data from 1959–1988, combining historical records on police bargaining laws with Vital Statistics data on deaths due to legal intervention, disaggregated by race. Using an event-study design, they find that the adoption of “duty to bargain” laws leads to a substantial increase in civilian deaths over time, driven almost entirely by non-White civilians. Medium- and long-run effects imply increases of roughly 50–110 percent in non-White civilian deaths, with no corresponding improvement in officer safety.

Why This Article Was Selected for The Policy Scientist

This article confronts a core institutional question in policing: whether labor protections for armed state agents materially alter the risks faced by civilians. Cunningham and Gillezeau, who have built a substantial research agenda on policing and labor institutions, place police unions squarely within the analysis of state violence rather than treating them as neutral workplace actors. The topic is especially timely amid renewed scrutiny of police power and accountability. The study advances the literature by showing that collective bargaining rights are associated with large, persistent increases in civilian deaths, particularly among non-White populations. The historical administrative data are unusually comprehensive for this period, and the event-study design reflects strong causal inference.

Full Citation and Link to Article

Cunningham, J. P., & Gillezeau, R. (Forthcoming 2025). Collective bargaining rights, policing, and civilian deaths. American Economic Journal: Economic Policy. https://doi.org/10.1257/pol.20230384 

Central Research Question

This article asks whether granting collective bargaining rights to law enforcement officers—specifically through state-level “duty to bargain” provisions—causally affects the number of civilians killed by police. The authors focus not on union membership itself, but on the institutional power created when states legally require public employers to bargain in good faith with police unions. The central question is whether these legal changes altered incentives, accountability, and behavior in ways that increased or decreased police-related civilian deaths, and whether those effects differed by race. A secondary but important question is whether expanded bargaining rights improved officer safety or broader public safety outcomes, allowing the authors to assess tradeoffs rather than examining civilian harm in isolation.

Previous Literature

The study builds on several strands of research. One literature documents racial disparities in police use of lethal force and shows that non-White civilians are disproportionately killed by law enforcement, even after accounting for crime exposure and contextual factors. Another literature examines the institutional determinants of police behavior, including militarization, technology, oversight, and administrative structure. A smaller but growing body of work studies police unions and labor protections, often arguing that unions reduce accountability by shaping disciplinary rules, investigative procedures, and legal protections for officers.

Prior empirical work on police unions is limited by data constraints and identification challenges. Studies using contract provisions or cross-sectional comparisons often struggle with endogeneity. More recent quasi-experimental work, such as studies exploiting court rulings or close union elections, suggests that collective bargaining can increase misconduct, but typically focuses on modern periods and aggregate outcomes. This article contributes by examining the historical expansion of police bargaining rights during the mid-twentieth century, a period marked by rapid institutional change and rising police violence, and by distinguishing between bargaining rights and unionization itself.

Data

The authors assemble a county-level panel covering the lower 48 U.S. states from 1959 to 1988. The primary outcome data come from the Vital Statistics Multiple Cause of Death files, which record deaths due to legal intervention by law enforcement and allow disaggregation by race. Although these data are known to undercount police-related fatalities, they are the most comprehensive government source available for the period and are consistently measured over time. The authors also employ bias-corrected fatality estimates from recent work to address underreporting concerns.

Information on collective bargaining laws is drawn from the NBER Public Sector Collective Bargaining Law Data Set, which documents the timing and strength of bargaining provisions by state and occupation. The key treatment is the adoption of duty-to-bargain laws for law enforcement. Additional data include arrests, police employment, officer deaths in the line of duty, and public safety expenditures from FBI Uniform Crime Reports and the Annual Survey of Governments, as well as county demographic characteristics from decennial census sources.

Methods

The authors use an event-study difference-in-differences framework that exploits variation in the timing of duty-to-bargain adoption across states. Counties in states that eventually adopt duty-to-bargain provisions are compared to counties in states that do not, before and after adoption. County fixed effects control for time-invariant local characteristics, while region-by-year fixed effects absorb broad regional trends. This design allows the authors to test for pre-treatment trends and to estimate dynamic effects over time.

Because police-related fatalities are rare events with many zero observations, the main specification uses Poisson regression models, supplemented by ordinary least squares and population-weighted models as robustness checks. The authors conduct extensive sensitivity analyses, including alternative estimators designed to address concerns about staggered treatment timing, controls for racial uprisings and affirmative action litigation, and models accounting for Law Enforcement Officers’ Bills of Rights. While not randomized, the design reflects modern causal inference standards and goes well beyond simple cross-sectional or correlational approaches.

Findings/Size Effects

The results show no evidence of differential pre-trends in civilian deaths prior to the adoption of bargaining rights, supporting the identification strategy. In the years immediately following adoption, there is little change in police-related fatalities, consistent with the time required for union formation and contract negotiation. However, medium- and long-run effects are substantial. Five to ten years after adoption, counties in duty-to-bargain states experience large increases in civilian deaths caused by police.

These effects are almost entirely concentrated among non-White civilians. Depending on the specification, non-White civilian deaths increase by roughly 50 to 110 percent relative to pre-treatment baselines in the medium and long run. In contrast, deaths of White civilians show no consistent or statistically significant change. The authors estimate that duty-to-bargain laws account for approximately 14 percent of all non-White civilian deaths due to police intervention between 1959 and 1988.

Importantly, the study finds no meaningful reduction in officer deaths or clear improvements in officer safety, suggesting that increased civilian harm is not offset by gains for law enforcement. The authors also present evidence that the effects are strongest in counties that actually formed police unions, reinforcing the interpretation that bargaining power and institutional protections, rather than unrelated state trends, drive the results.

Conclusion

The article concludes that granting collective bargaining rights to police officers, particularly through duty-to-bargain laws, had significant and lasting consequences for civilian mortality, with disproportionate effects on non-White populations. By focusing on institutional legal changes rather than union status alone, the study clarifies how labor law can shape the behavior of coercive state actors. The findings suggest that the expansion of police bargaining power altered accountability structures in ways that increased lethal encounters without improving officer safety.

More broadly, the article demonstrates the importance of examining labor institutions in settings beyond traditional employment outcomes. Methodologically, it shows how historical administrative data and event-study designs can yield credible causal evidence when randomized experiments are infeasible. While future research could benefit from exploiting sharper quasi-random shocks or modern administrative data, this study makes a substantial contribution by establishing a clear empirical link between police bargaining rights and civilian deaths.