Causal Inference
Studying the effectiveness of firearm policies in reducing firearm harms using causal inference Roni Barak Ventura* Roni Barak Ventura Maurizio Porfiri James Macinko Manuel Ruiz Marín
Firearm violence is a significant threat to public health in the U.S., where more than 200 people sustain a nonfatal firearm injury and more than 100 people die from it every day. In spite of these unsettling figures, Americans still seek to hold guns for protection of their homes and families. In order to prevent firearm-related harms without limiting citizens’ right to bear arms, one must identify the policies that minimize firearm harms. To this end, we study the causal relationships between firearm laws, firearm ownership, and firearm deaths. Specifically, we investigate whether permissive and/or restrictive laws promote and/or discourage firearm ownership and deaths.
We collate data about firearm-related laws from RAND’s State Firearm Law Database, where each law is classified as restrictive or permissive. We also collect the monthly numbers of accidents, homicides, and suicides committed with firearms reported on CDC’s Wonder database. Lastly, we obtain monthly estimates of firearm ownership from an econometric spatiotemporal model we have previously developed. We generate two monthly time series: one that reflects the restrictiveness of the firearm-related legal environment, and another that measures “Deaths Per Gun”. We apply the information-theoretic notion of transfer entropy to infer a causal link from the former to the latter.
We uncover a causal link for delays ranging from zero to three months, suggesting that restrictive laws can effectively reduce “Deaths Per Gun” in the short term. This analysis is the first to demonstrate a causal effect of firearm laws on firearm harms and firearm ownership. It serves as a stepping stone for analyses that could guide future legislation to effectively reduce regional firearm harm rates.