NBER Publications by Matthew Miller

Of the readily computed proxies for the prevalence of gun ownership, one, the percentage of suicides committed with a gun, performs consistently better than the others in cross-section comparisons. It is readily computed for states and counties and has a high degree of validity when tested against survey-based estimates. It also appears valid as a proxy for changes over time in gun prevalence, at least at the regional level. Our analysis of this proxy measure for the period 1979-1997 demonstrates that the geographic structure of gun ownership has been highly stable. That structure is closely linked to rural tradition. There is, however, some tendency toward homogenization over this period, with high-prevalence states trending down and low-prevalence states trending up.