Guns do kill people: Novel global evidence on the cross-national relationship between gun ownership and (gun) homicide

Despite recent advances, the existing macro-social, cross-national research on whether gun ownership exacerbates (gun) homicide remains unsettled. Studies report positive, negative, and null results. This is in part due to pervasive methodological issues like small sample sizes, inappropriate modelling and sparse controls for confounding, crude measures of gun ownership, and not differentiating between gun homicide and total homicide rates. This paper presents a novel estimation strategy performed on a new cross-national dataset covering more than 100 countries and spanning 2000–2019, which is by far the largest global sample to date. Using the validated proxy of gun ownership (percentage of gun suicides), both simple cross-sectional as well as Mundlak-corrected, correlated random-effects models – which are robust to time-invariant, country-specific heterogeneity – consistently show evidence of a statistically significant, positive, and sizable effect on gun homicide. This result survives a battery of robustness tests, different controls, and an alternative measure of gun ownership. However, I find no support for the existence of a significant relationship between gun ownership and the total homicide rate.