What if every state had acted in 1994?
The Brady Act passed in 1993. The Assault Weapons Ban in 1994. But universal background checks never became law. We ask: what if they had? Using the observed gap between strong-law and weak-law states, we generate 500 plausible alternative histories for the national firearm homicide rate.
The accumulating toll of delay.
Each bar represents the median estimate of lives that would not have been lost in that year under the counterfactual scenario. The uncertainty grows with time, but so does the cost.
1. Natural experiment. We exploit the cross-sectional variation in state gun law strength as a proxy for what national adoption might achieve. States with law scores ≥ 70 (13 states including California, New York, and Massachusetts) serve as the treatment group. States with scores < 40 (23 states including Mississippi, Alabama, and Texas) serve as the control.
2. Treatment effect. The average firearm homicide rate in treatment states is approximately 23% lower than in control states across 2019–2024. This reduction is the basis for the counterfactual.
3. Ensemble generation. 500 trajectories are generated by (a) varying the treatment effect size around the observed mean using Normal(μ, 0.15μ), (b) varying implementation lag from 0–5 years post-1994, and (c) adding noise proportional to the actual rate. Each trajectory applies the treatment effect to the observed national homicide rate.
4. Lives saved. The gap between actual and counterfactual rates is multiplied by estimated US population for each year to yield lives saved. Cumulative totals are reported with 90% confidence intervals derived from the 5th and 95th percentile trajectories.
Limitations. This is an observational cross-sectional analysis, not a causal inference. State-level differences reflect many factors beyond gun laws (urbanization, poverty, policing). The treatment effect assumes weak-law states would respond to legislation the same way strong-law states did — a strong assumption. Confidence is rated CANDIDATE, not HIGH.