The impact of medical marijuana laws on marijuana use and other risky health behaviors

The goal of this paper is to estimate the impact of state laws that permit medical use of marijuana on other risky health behaviors, such as smoking, binge drinking, and use of other illicit drugs. Marijuana is the most commonly used drug in the U.S. and emergency room visits and substance abuse treatment related to marijuana use has increased over time. Although marijuana is a Schedule I drug under the Controlled Substance Act of 1970, many states in recent decades have legalized marijuana use for certain medical conditions. (Proponents of medical marijuana argue that marijuana is effective in relieving chronic pains, nausea, and improving appetite, but this is controversial).

There are several mechanisms by which legalized medical marijuana use might affect other risky health behaviors.  Marijuana could act as a “gateway drug”, leading to subsequent consumption of hard drugs such as heroin or cocaine (e.g. Mills and Noyes, 1984; Newcomb and Bentler, 1986; Deza, 2012).  Furthermore, marijuana may be a complement to, or substitute for, other substances (e.g. Pacula, 1998; Farrelly et al., 1999, 2001; Williams et al., 2004; Chaloupka et al., 1999; Dinardo and Lemiuex, 2001; Yoruk and Yoruk, 2011).  Recent studies have examined the impact of medical marijuana policies on youth marijuana consumption, and tend to find a positive association (Cerda et al., 2012; Wall et al., 2011; Lynne-Landsman et al., 2013). A contribution of my study is that it examines the adult population.

In this paper, I examine the data from the restricted-use National Survey of Drug Use and Health (NSDUH), which is a repeated cross sectional dataset that spans 2004 to 2011.  It collects information from roughly 70,000 randomly selected individuals in the US of age 12 or older. Using the restricted-use geographic identifiers I am able to merge in state marijuana laws.  From 2004-11, 8 states have allowed marijuana use for medical purpose.  Exploiting this variation across states over time, I estimate difference-in-differences models to examine the impacts of these policy changes on risky health behaviors.  I also estimate regressions by relevant subgroups such as adolescents, adults with chronic medical conditions, race, education, and gender. In order to address policy endogeneity, which is a potential threat to my identification strategy, I estimate an auxiliary model that estimates the probability a state passes medical marijuana legislation as a function of the political environment of the state, socioeconomic characteristics of the state, and other relevant characteristics such as drug related arrests per capita, percent of population with excellent health or cancer. 

This study provides early estimates of the impact of recent medical marijuana laws on risky health behaviors.  This is useful for both understanding possible unintended consequences of such laws and for better understanding the determinants of risky health behaviors.