Not in my backyard? Not so fast. The effect of marijuana legalization on neighborhood crime

Abstract

This paper studies the effects of marijuana legalization on neighborhood crime and documents the patterns in retail dispensary locations over time using detailed micro-level data from Denver, Colorado. To account for endogenous retail dispensary locations, we use a novel identification strategy that exploits exogenous changes in demand across different locations arising from the increased importance of external markets after the legalization of recreational marijuana sales. The results imply that an additional dispensary in a neighborhood leads to a reduction of 17 crimes per month per 10,000 residents, which corresponds to roughly a 19 percent decline relative to the average crime rate over the sample period. Reductions in crime are highly localized, with no evidence of spillover benefits to adjacent neighborhoods. Analysis of detailed crime categories provides insights into the mechanisms underlying the reductions.

Introduction

After Colorado and Washington became the first U.S. states to legalize recreational marijuana in 2012, the number of states legalizing or decriminalizing the sale and use of marijuana quickly expanded. After a wave of ballot initiatives in 2016, the sale and use of marijuana for recreational purposes was legal in 7 states, and another 22 states had legalized medical use. As states legalize the manufacturing, distribution, and sale of marijuana, the local health, economic, crime, and safety effects of marijuana dispensaries have become an important public policy issue. In addition to the aggregate effects of legalization, understanding the local effects of marijuana dispensaries on neighborhoods is important for designing policies to address concerns of residents who are broadly open to legalization but have a “not in my backyard” attitude toward dispensaries near their homes.¹

The economic welfare and public policy implications of marijuana legalization are broad in scope and stem from several primary sources. First, given that legalization improves access to marijuana and presumably reduces prices, in the long run, legalization could affect local health, economic, crime, and safety outcomes due to increased marijuana use.² Second, legalization may displace illicit markets affecting neighborhood outcomes, including crime or access to other illegal drugs.³ Third, marijuana dispensaries may have social or economic spillover effects that may affect welfare. Finally, there are direct implications for public finance through increased tax revenue and decreased enforcement costs.⁴

Our paper focuses on the short-run causal effects of marijuana legalization on neighborhood crime.⁵ To date, we are unaware of any research that studies the effects of full marijuana legalization on local crime, although several papers have analyzed local effects of decriminalization polices or legalization of medical marijuana.⁶ Two papers study aggregate effects of crime from decriminalization and legalization policies. Adda et al. (2014) study the effects of a depenalization policy in a borough of London, exploiting time variation from a policy change. The authors found that the decriminalization policy led to an aggregate decrease in crime arising from reallocation of police resources but also a decrease in home values, suggesting a welfare loss. Huber et al. (2016) use cross-sectional variation in state policies and panel data between 1970 and 2012 and find evidence that the legalization of medical marijuana reduces robberies, larcenies, and burglaries, although they find that decriminalization has no effect on crime. We expand on these aggregate studies by considering local variation in crime outcomes within a jurisdiction that has legalized marijuana.

Our approach is related to that of Freishtler et al. (2016) who study the effect of medical marijuana on neighborhood crime in Long Beach, California, and exploit a change in policy that led to the closing of dispensaries. They show that there was no change in crime locally, but found positive correlations between increased crime and the dispensary density in adjacent neighborhoods.

Chang and Jacobson (2017) also exploit the unexpected closing of dispensaries in California to identify the effect on crime. They find the somewhat different result; specifically that there is a temporary increase in crime very near the dispensaries after they have closed that dissipates over time.⁷ Our research expands on this work in an important way by accounting for the endogenous location of dispensaries in neighborhoods. In addition, we study both recreational and medical marijuana dispensaries and utilize panel data that capture both dispensary openings and closings.

In this paper, we investigate the local effects of marijuana legalization on neighborhood crime in Denver, Colorado, which was the first state to fully legalize marijuana use, sales, and production for medical and recreational purposes. The baseline analysis compares year-over-year changes in dispensaries to year-over-year changes in crime rates using monthly tract-level data from Denver. To account for the endogenous location of dispensaries, we exploit the change in policy that allowed for recreational sales starting in January 2014. This policy changed the spatial demand for marijuana by allowing for sales to out-of-state residents and increased demand from residents from neighboring municipalities that only allowed medical sales. Proxies for market access are used to instrument for changes in dispensary density around the time of the policy change.

One contribution of our research is the construction of a unique and rich geospatial data set.⁸ To measure dispensary locations, we use panel data from the State of Colorado that provide exact locations of dispensary licenses at monthly frequencies. Our primary measure of crime comes from the city of Denver and includes the location, date, and type of crime committed. We construct a detailed location-specific measure of available land using data on zoning, geographic features, and legal restrictions on dispensary locations, which we augment with demographic and employment data provided by the U.S. Census.

Initial analysis of the data shows that the locations of dispensaries are not randomly allocated across space or neighborhood characteristics. Dispensaries are more concentrated in areas with higher poverty, higher minority populations, and higher initial employment density. Correlations between the growth of new dispensaries and neighborhood characteristics strengthened over time.

It is also likely that dispensary locations and crime are correlated with unobservable neighborhood characteristics, which creates a challenge for causal analysis. For example, after the legalization of recreational marijuana sales in 2014, new dispensaries were subject to public hearings. Neighborhoods with more social cohesion, could potentially prevent the opening of new dispensaries. Previous studies on local crime effects have not directly addressed the endogeneity of dispensary locations.

An important contribution of our paper is that we employ a novel identification strategy that exploits shifts in demand across locations over time to analyze causal effects of marijuana legalization on crime. While the legalization of recreational marijuana in 2014 applied to the entire state, many municipalities within Colorado prohibit sales within their own jurisdictions. Residents living in municipalities near Denver that prohibit recreational sales often travel to Denver to purchase marijuana. Therefore, locations within Denver that have more access to demand from neighboring municipalities show more growth in their dispensary density, ceteris paribus. In addition, out-of-state tourists could purchase marijuana starting in 2014, further increasing the demand for dispensaries in locations with access to broader outside markets. In the empirical analysis, we use two geospatial variables to proxy for access to outside demand: a neighborhood's proximity to municipal borders and proximity to major roads or highways. These variables are then used to instrument for changes in locations of dispensaries over time.⁹ We combine our instrumental variables (IV) with our panel data to compare year-over-year changes in crime with year-over-year changes in dispensary density and use time fixed effects to control for aggregate crime trends.

Note that a particular advantage of our identification strategy relative to others is that it relies on variation within a single jurisdiction. Studies that use differences in policies across jurisdictions suffer from endogeneity if the policy decision is correlated with unobserved characteristics of a municipality. In our setting, the policy change is the same for all locations in the study, and the variation in treatment is due to an exogenous shift in external demand.

Our main IV results imply that receiving a dispensary in a neighborhood causes a reduction in crime; specifically, an additional dispensary per 10,000 residents is associated with a reduction of 17 crimes per 10,000 residents per month. The average number of crimes per 10,000 residents in Denver is 90 per month, so an additional dispensary is associated with roughly a 19 percent decline in crime. These IV results are robust across a number of specifications. The results from the ordinary least squares (OLS) specification, on the other hand, are positive, reflecting that dispensaries are on average selected into neighborhoods with increasing crime.¹⁰

In addition to finding an overall reduction in crime when a dispensary is added to a neighborhood, we also find that there is some variation across crime categories. The effect is generally strongest for nonviolent crimes; specific crimes most affected include criminal trespassing, public-order crimes, criminal mischief, and simple assault. There are also reductions in violent crimes driven by a decrease in aggravated assault, although results are not statistically significant. Reductions in these crimes are consistent with disruption of illicit markets and with a substitution away from alcohol use. However, we do not find strong evidence that legalization disrupts the sale of other illicit drugs. While our point estimates suggest that sales of other drugs decline, the estimates are not statistically significant. In addition, we do not find significant increases in marijuana-related crimes that are tracked separately by the City of Denver, which implies that there are not large crime effects from increased marijuana use itself.

Lastly, we find that the reduction in crime is very localized and contained within the census tract of the dispensary's location. We test for spillover effects by regressing changes in crime on the predicted change in dispensary density both within that tract and from neighboring tracts (from our first-stage regression). We find no significant effects of neighboring dispensary density on crime.

Overall, our results suggest that dispensaries cause an overall reduction in crime in neighborhoods, with no evidence of spillovers to surrounding neighborhoods. The local nature of these effects is consistent with increased policing or private security response near the dispensaries. These findings may point to an aggregate reduction in crime due to legalization, but further investigation would be needed to rule out a reshuffling of crime to other neighborhoods. Effects on specific crimes vary and are weakly consistent with the theory that legalization could disrupt illicit markets and also support evidence that marijuana could be a substitute for alcohol consumption. Lastly, there is no evidence that increased marijuana use itself results in additional crime. More generally, there is potential for further research to understand the underlying mechanisms that lead to the change in crime after legalization.

The rest of the paper is organized as follows. Section 2 discusses the policy implications of legalization and related literature. Section 3 provides background and descriptive data analysis of legalization in Colorado. Section 4 outlines the empirical methodology and identification strategy. Section 5 gives a detailed description of data collection and construction. Section 6 presents the main results. Section 7 provides some additional analysis and discussion. Finally, Section 8 concludes.