This paper evaluates Mexico City's voluntary gun buyback program, Sí al desarme, sí a la paz, launched in January 2019. Using state-month data for 2017¿2023, we estimate program effects with synthetic control and synthetic difference-in-differences designs. We find a statistically significant decline in our composite outcome, which combines recorded firearm-related crimes with unclassified deaths to conservatively guard against possible outcome reclassification. The estimated decline is equivalent to roughly 17% of the composite outcome's pre-treatment mean. Because unclassified deaths need not be firearm-related, this specification prioritizes robustness to misclassification over direct substantive interpretation. The more interpretable disaggregated estimates show that the effect is concentrated in intentional firearm injuries, while effects on firearm homicides are statistically indistinguishable from zero across specifications. An illustrative Phase 1 calculation implies approximately 11,354 MXN ($560 USD) in spending per estimated averted event in the composite outcome. We also find no detectable substitution into non-firearm injuries. Mechanism evidence indicates that the program primarily collected civilian handguns and is consistent with a situational opportunity-reduction channel in civilian settings: reducing immediate access to firearms can limit escalation and lower the likelihood that routine conflicts translate into gun injuries, rather than directly disrupting organized-crime arsenals. Robustness checks, including placebo timing tests around COVID lockdown and triangulation with INEGI vital statistics, suggest that the results are not driven by reporting or reclassification. The findings contribute the first causal evidence on a voluntary gun buyback in Mexico and, by focusing on non-lethal firearm injuries rather than solely on homicides, identify a margin of impact that prior Latin American evaluations have not examined (Cano et al., 2024).
