Abstract
We examine the relationship between traffic fatalities and state marijuana laws using data from 1985 through 2019 and Poisson difference in difference models that allow effects to vary over time. We show numerous attributes of state marijuana laws are captured by a single underlying dimension, implying interdependence and potential inseparability of the effects of recreational marijuana laws (RMLs) and prior medical marijuana laws (MMLs). Controlling for MMLs, we find no statistically significant change in fatalities associated with RMLs, while, as in earlier work, MMLs are associated with lower fatalities. There is a statistically insignificant decline in fatalities in states bordering RML states. Type 1 error rate inflation is present in this context, but it is not strong enough to overturn the finding that lower fatalities followed MMLs.
Highlights
As of March 2021, thirty-five U.S states and the District of Columbia had legalized marijuana use for medical purposes
We find lower state traffic fatalities following the implementation of medical marijuana laws (MMLs), consistent with earlier work
This is true whether we employ a simple MML indicator or a continuous indicator of the permissiveness of state medical marijuana laws
Summary
As of March 2021, thirty-five U.S states and the District of Columbia (hereinafter 51 states) had legalized marijuana use for medical purposes. Anderson et al (2013), and Malivert and Hall (2013) show that the price of highquality extralegal marijuana fell substantially following implementation of MMLs in states that later implemented RMLs, presumably leading to corresponding increases in extralegal recreational consumption before the adoption of an RML. Effects of RMLs observed by comparing outcomes in the five years after the effective year to any given period before adoption may reflect this pre-existing upward drift and lack of control for the effects of prior MMLs. There is evidence of significant cross-border marijuana sales between RML and non-RML states (Hansen et al, 2020a). We find evidence of type 1 error rate inflation, but it is not strong enough to overturn our finding that reductions in traffic fatalities follow MMLs. The inflation of type 1 error rates only strengthens our finding of no statistically significant further effects of RMLs
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
