Association for Public Policy Analysis & Management

Marijuana policy has recently been a hot topic both in the public sphere and in the research field. As of this writing, 23 states plus the District of Columbia have passed laws that allow for at least some sort of protection for using marijuana for medicinal purposes, and four more states have passed laws legalizing recreational marijuana use. Variation in policies across local governments is a great tool for policy analysts to learn about the effects from marijuana laws on different outcomes, such as crime. So far, most studies have aggregated state data to study the effects of marijuana policies on crime, ignoring the rich variation that exists within each state law. This paper attempts to shed some light on the importance of the heterogeneity in implementation within states, by examining the variability in medical marijuana laws in California. I estimate the relationship between county medical marijuana laws pertaining to supply (dispensary allowances/bans and home cultivation limits) and various types of arrests (drug arrests, violent crime arrests, property crime arrests, drugged driving, and other nuance crimes).  I also attempt to isolate the effects on crime that are associated to changes in use and to changes in suppliers (i.e. dispensaries) by focusing on specific types of arrests.

California became the first state to legalize medical marijuana in 1996, and since then counties have established their own laws regarding medical marijuana use and production. This provides a rich source of data from 58 counties that vary in terms of types of cultivation allowed for residents and producers, the number of dispensaries allowed, among many other factors. The RAND Drug Policy Research Center has created a dataset that codes the ordinances for the various types of marijuana-related laws in all 58 counties in California over the period 1997-2014. We have also put together count data on the number of arrests by categories of crime in each county by year, in addition to other economic and demographic variables of these counties. I estimate difference-in-difference models that allow me to identify the extent to which more liberal supply provisions (allowance of dispensaries and large scale cultivation) are associated with different types of crimes.  I find no direct evidence of an impact on violent crime arrests, but preliminary analyses show interesting results for some property crimes. Additionally, I find that types of arrests related to marijuana use, such as DUI and misdemeanor drug offenses, can help isolate the effect of changes in supply to changes in use.