A panel-based proxy for gun prevalence in US and Mexico

There is a consensus that the proportion of suicides committed with a firearm is the best proxy for gun ownership prevalence. Cerqueira et al. (2018) exploit socioeconomic characteristics of suicide victims in order to develop more refined prevalence indicators. They rest on location fixed effects of the victim's place of residence from a discrete-choice model for the likelihood of committing suicide with gun. We empirically assess this new indicator using gun ownership data from the Behavioral Risk Factor Surveillance System (BRFSS) and suicide registers of the US National Center for Health Statistics (NCHS) from 1995 to 2004. We show that the panel-based proxy variables correlate more with the percentage of households with firearms than the conventional proxy based on the proportion of suicides committed with a firearm does. We further estimate gun prevalence across Mexican states. This is a relevant application because there is no representative household survey in Mexico with information about gun ownership at the state level.