Medical Marijuana Laws and Illegal Marijuana Use Yu-Wei Chu * Michigan State University 110 Marshall Adams Hall East Lansing, MI 48823. [email protected] September 30, 2012

Abstract Seventeen states and the District of Columbia have passed laws that allow individuals to use marijuana for medical purposes. In this paper, I use marijuana possession arrests and treatment referrals by medical professionals to estimate the impact of medical marijuana laws on marijuana usage among non-patients. I find that these laws increase marijuana arrests among adult males by about 20%. The effect is strongest among young adults and decreases with age. I also find that marijuana treatment referrals increase by more than 10% after the passage of medical marijuana laws. In contrast to previous studies, my analysis also shows some evidence that these laws affect juveniles’ marijuana use. JEL Classification: I10 I18 H75 K42 Keywords: medical marijuana laws, marijuana use

*

The author is grateful to Jeff Biddle, Todd Elder, and Gary Solon for their guidance and suggestions. Thanks also go to Michael Conlin, Steven Haider, Sheila Royo Maxwell, Stacey Lynn Miller, and participants at the Empirical Micro Lunch Seminar at Michigan State University for helpful discussions and comments.

"By characterizing the use of illegal drugs as quasi-legal, state-sanctioned, Saturday afternoon fun, legalizers destabilize the societal norm that drug use is dangerous . . . Children entering drug abuse treatment routinely report that they heard that 'pot is medicine' and, therefore, believed it to be good for them.” Andrea Barthwell, M.D., Former Deputy Director of the White House Office of National Drug Control Policy, in an editorial in The Chicago Tribune, Feb. 17, 2004

1.

Introduction Medical marijuana legislation represents a major change in U.S. policy towards

marijuana in recent years. Advocacy groups such as NORML consider such legislation the first step towards full legalization. As of July 2012, seventeen states and the District of Columbia had passed laws that allowed individuals with designated symptoms to use marijuana for medical purposes (ProCon.Org, 2012a). Six more states have legislation pending, while twelve other states introduced legislation that failed to pass. Although the Obama administration takes a relatively liberal attitude towards medical marijuana, federal agencies such as the Drug Enforcement Administration (DEA) and the Office of National Drug Control Policy (ONDCP) remain firmly opposed, and marijuana is still listed as a Schedule I drug with no accepted medical value. Because the number of legal patients was relatively small at least before 2009, the direct impact of medical legalization should be limited. The main reason that the DEA and ONDCP oppose such laws is based on the notion that they would increase marijuana use among non-patients (Drug Enforcement Administration, 2011). There is a strong correlation between medical marijuana legislation, the perceived risk of marijuana and marijuana use. According to the 2008 National Survey on Drug Use and Health (NSDUH), among the states with the highest rate of marijuana use and the lowest perceived risk, ten out of fifteen have adopted medical marijuana legislation. 1 Despite the strong correlation, only a small number of studies have assessed the causal link between medical marijuana laws and usage, with mixed results. Importantly, most of the studies only cover a short time period, leading to imprecise estimates based on a small number of state-level law changes. Existing 1

The fifteen states with the highest use rate are Alaska, California, Colorado, Delaware, District of Columbia, Maine, Massachusetts, Michigan, Montana, New Hampshire, New York, Oregon, Rhode Island, Vermont and Washington; the states with the lowest perceived risk are the above states with California, Delaware and New York replaced by Minnesota, Virginia and Wisconsin.

1

studies often focus on juveniles, while only adults are qualified patients under medical marijuana laws, and the marijuana prevalence rate is actually higher among young adults than among juveniles. In addition, these studies only examine the marijuana prevalence rate and ignore potential changes at the intensive margin. To examine whether medical marijuana laws increased illegal marijuana use among nonpatients, especially for adults, I use data on marijuana possession arrests in cities from the Uniform Crime Reports (UCR) as a proxy for illegal use, for the years 1988 through 2008. As shown in Table A1 in the Appendix, these data cover a period when 12 states legalized medical marijuana. As arrest depends on not only the number of users but also their use or transaction frequencies, it is a mix of measure for both the intensive and extensive margins. Another advantage of arrest data is that it represents an objective measure. It does not suffer from the selfreporting bias that is common in survey data (Golub et al., 2005; Harrison and Hughes, 1997). Since medical marijuana laws are expected to change social acceptance and perception of marijuana, changes in reporting behavior are of particular concern in the current context. For example, Miller and Kuhns (2011) find that arrestees report use more honestly after the passage of medical marijuana laws. I estimate reduced-form models for the effect of medical marijuana laws on male arrests, controlling for city and year fixed effects as well as city time trends. I find that these laws increase illegal marijuana usage, particularly for young adults. For adult males, on average, medical marijuana laws are associated with around a 20% increase in arrest rate, or an increase of about 25 arrestees per 100,000 persons in cities. The effect is stronger among males aged 1829, while there is no obvious effect in age groups over 40. The estimates also show a 10% increase in arrest rates for male juveniles, although the estimates are relatively imprecise in this age group. At first glance, a 20% increase might seem implausible large. Conceptually, arrest data capture changes in both the intensive and extensive margins, and a significant part of the 20% increase can be viewed as a change at the intensive margin. 2 Empirically, marijuana arrests are 2

We can model arrest as follows: 𝐴 = ∑𝑁 𝑖=1 𝑃(𝑋𝑖 ) ∗ 𝐹𝑖 , where Fi is individual i's transaction or use frequencies, N is the number of marijuana users; P (X) is the probability of being arrested per transaction, a function of Xi, factors such as local law enforcement. Assume P(X) to be the same for everyone, then log(A) = log(P(X)) + log(𝐹� ) + log(N), where 𝐹� is the average of Fi. So, arrest reflect effects on both the extensive and intensive margins. Empirically, P(X) may depend on Fi and possibly look like a step function, i.e., only users with Fi greater than some amount have a positive probability of being arrested.

2

likely to be concentrated on heavy users, as the average probability of being arrested is low, and a 20% increase in heavy users is not particularly large for policy changes regarding substance use. For example, zero tolerance laws that lower the legally allowable BAC from 0.08 to 0.02 for drivers reduce heavy drinking for targeted population by 13-18% (Carpenter, 2004). Because heavy marijuana users are often associated with potential negative outcomes, such as developing dependence and need for treatment, I use data on admissions to rehabilitation facilities from the Treatment Episode Data Set (TEDS) to provide further evidence on the effects of medical marijuana laws. The TEDS data allows me to obtain estimates that are not biased by law enforcement from using treatments referred by professional medical providers. The results are consistent with the findings from the arrest data: on average, medical marijuana laws increase the professional referred treatments by around 10% for adult males. In contrast to previous studies that do not find any effect of medical marijuana laws on juveniles, I find a positive effect of 5-15% for juvenile treatments. I also use only first-time treatments to create a measure representing individuals to exclude potential effects on recidivism. Somewhat surprisingly, instead of a smaller effect, the estimates indicate a 20% increase in first-time treatments for adult males. This research addresses the heated policy debate on medical marijuana laws by presenting evidence for an increase in illegal use among non-patients. By using data reflecting effects on heavy users, this research is more relevant to the design of policy because heavy usage is associated with negative health and social outcomes. In particular, the results from the marijuana treatment data provide empirical support for such concerns and suggest a direct medical cost from these laws.

2. Medical Marijuana Laws In the late 1980s and the early 1990s, smokable marijuana was discovered to have a positive effect on patients suffering from nausea, a common symptom among cancer patients and the increasing number of AIDS patients (Pacula et al., 2002). With growing evidence of positive medical effects and lobbying by marijuana legalization advocacy groups such as NORML, many states have joined in passing a new wave of medical marijuana legislation since 1996. Table A1 in the Appendix provides an overview of each state’s medical marijuana laws (For legal documents, see ProCon.Org, 2012a).

3

These laws permit patients with legally designated diseases and syndromes to use marijuana as a treatment. Patients can legally possess marijuana up to a fixed amount. In many states, they can cultivate marijuana on their own. These laws also allow “caregivers” (most of whom are patients as well) to grow and provide marijuana to patients on a not-for-profit basis. In most states, it is mandatory to register as a qualified medical marijuana patient or caregiver and to renew this registration every year. 3 The designated symptoms and conditions typically include AIDS, anorexia, arthritis, cachexia, cancer, chronic pain, glaucoma, migraine, persistent muscle spasms, severe nausea, seizures, and sclerosis. Some laws, such as those in California and Colorado, even allow use for “any other illness for which marijuana provides relief” (Cohen, 2010). Because even designated syndromes such as chronic pain can be defined subjectively, such legislation actually provides a way for all marijuana users to become legal patients. However, before 2009, the number of legal patients remained relatively small except in California. 4 A very imprecise estimate from ProCon.org (2012b) indicates that, as of January 2009, the total number of legal patients was about 0.27 million people, or 0.19% of the population in medical marijuana states. One reason might have been that becoming a registered patient did not greatly increase the ease of acquiring marijuana because the number of dispensaries was limited. Some marijuana dispensaries with grey legal status did exist under the name of caregiver, but how prevalent they were depended on the attitude of the local government (often at the city level) and the actions of law enforcement, which could change from time to time. This was because the state medical marijuana laws did not directly allow marijuana dispensaries in order to conform to federal regulations in which marijuana remained a Schedule I drug. The legal environment has changed since 2009. The Obama administration has stated that the federal government will no longer seek to arrest medical marijuana users and suppliers so long as they conform to state laws. 5 This statement largely resolved the legal dispute between state and federal governments, and it motivated several states to initiate and pass medical 3

California created a registration program in 2004 but registration was voluntary. Colorado allows patients who do not join the registry to use the "affirmative defense of medical necessity" if they are arrested on marijuana charges. Maine passed an amendment in November 2009 that created a registration program and required mandatory registration starting January 1, 2011. Washington does not have a registration program. 4 There is no official number of patients for states without registration. However, based on the large number of dispensaries, it is believed that California has many more patients than other medical marijuana states. 5 The Obama administration's medical marijuana policy is swinging recently. In 2012, there have been cases of DEA raids on medical marijuana providers which arguably conform to state/local laws.

4

marijuana laws. The number of registered patients and dispensaries has increased significantly since 2009.

3. Previous Literature There is little doubt that medical marijuana legalization increases marijuana usage at the intensive margin, because some existing users will become legal patients and they will be able to increase their consumption safely and easily. On the other hand, since the number of legal patients is limited, the major policy debate is whether an indirect effect of laws increases use among non-patients. It is a popular belief among public media that legalization has increased illegal marijuana use, especially among juveniles (O'Connor, 2011). Illegal marijuana use could increase, in terms of both the intensive and extensive margins, because of the greater ease of obtaining marijuana, either from a legal patient or from a dispensary. More significantly, however, these laws could send a “wrong message” to the public that increases illegal use. If marijuana were defined by law as a beneficial medicine rather than a harmful drug, this would increase social acceptance for recreational use. People may also expect more lenient law enforcement and broader legalization in the future. Unlike changes in legal penalty of marijuana possession that ordinary people may not be aware of, the process of referendum and the setup of medical marijuana program make the public much more aware of the legislation (MacCoun et al., 2009; MacCoun, 2010; Pacula et al., 2005). As Becker and Murphy’s seminal paper (1988) shows, lowering the perceived health and legal risk will lower the price and increase use. Empirically, Johnston et al. (2011) and Pacula et al. (2001) show that the perception of risk is a good predictor of marijuana use among high school students. Empirically, there is a strong correlation between medical marijuana legislation, perceived risk and marijuana use; nevertheless, the causal link is not well supported by existing studies. Based on Drug Abuse Warning Network data for the years 1994–2002, Gorman and Huber (2007) do not find any significant change of marijuana use among arrestees using a time series framework. Their data are limited to a small portion of arrestees with available urine test samples from only four cities in California, Colorado and Oregon in a short time span. Moreover, criminals are probably a demographic group that does not respond to the change of laws due to the existing high use rate and low perceived risk. On the other hand, based on the same dataset, Pacula et al. (2010) find that medical marijuana laws increase marijuana price (as reported by

5

arrestees), which they interpret as an increase in demand for marijuana along with an upward sloping supply curve. Drawing on public-use data from the NSDUH for the years 2002 through 2008, Wall et al. (2011) find that legalization was associated with a higher prevalence rate of marijuana use among 12- through 17-year-olds, while Harper et al. (2012) show that these results are quite sensitive to including state fixed effects. Using a number of datasets that cover a longer period, a recent working paper from Anderson et al. (2012a) finds no evidence of an increase in marijuana use among teenagers. In fact, the estimates for juveniles from Anderson et al. (2012a) and Harper et al. (2012) are often negative. The estimate for young adults aged 18-25 from Harper et al. (2012) are positive but insignificant in their extended sample through 2009. There are many limitations on the public-use data of NSDUH. First of all, it is only available since 2002, and there were only five states that changes their laws within the sample period. The problem of a shorter time period is further aggravated by the fact that the data is smoothed across years. The NSDUH provides the state level measures as two-year moving averages estimated from a logistic model. For example, the measure in the 2008 report is a predicted probability using both 2008 and 2007 data. The standard fixed-effects estimates may not be very reliable because there is little much within-state variation in these data. The two-year moving averages also make coding for the first-year of legalization arbitrary, which could present problems in a short sample. As mentioned above, a direct effect of medical marijuana laws is to increase marijuana use at the intensive margin for legal patients. Potential indirect effects, such as

lowering

perceived risk or increasing availability, also suggest an effect on the intensive margin as well. However, existing studies estimate the effect of laws exclusively in terms of the extensive margin. Ignoring the intensive margin could seriously underestimate the effect especially for adults due to their low initiation rate (Gfroerer et al., 2002). 6 For policy concerns, the intensive margin is at least as important as the extensive margin. It is common for a policy to affect only the intensive margin. For example, Carpenter (2004) finds that zero tolerance laws only decrease heavy drinking while having no effect on participating in drinking.

6

The measure of initiation rate in Gfroerer et al. (2002) is first-time use, which should be viewed only as a lower bound for the change at the extensive margin especially for adults. A current user could have tried marijuana early in his life, but starts to use regularly since recently.

6

4. Analysis of the Uniform Crime Reports 4.1.

The data The primary data for this study is arrest for marijuana possession from the FBI's Uniform

Crime Reports (UCR). 7 The UCR arrest data is an administrative series of monthly police records from state and local police agencies across the U.S. It provides information on arrest counts by age, gender and race in each crime category along with agency populations (estimated from the Census). I use the yearly aggregated arrest data provided by the Inter-university Consortium for Political and Social Research (ICPSR), as the FBI also reviews and checks the data using annual arrest totals (Akiyama and Propheter, 2005). Previous studies like Conlin et al. (2005) have used these data as measures for usage of illegal substances. The UCR arrest data has a hierarchy rule which only records arrests according to the most serious offense. As a result, arrestees classified under marijuana possession do not simultaneously commit other serious crimes (such as cocaine possession or other violent or property crimes). Because a person may be arrested many times, each arrest count does not necessarily represent a single individual. My sample covers the period 1988 through 2008. 8 I use data starting from 1988 to avoid potential influences from decriminalization and the crack epidemic. Eleven states decriminalized marijuana in the 1970s, though there are only minor differences across non-decriminalized and decriminalized states in the late 1980s (Pacula et al., 2003; Pacula et al., 2010). 9 There was also a strong declining trend in marijuana use/arrest in the 1980s, which could be related to the crack epidemic and President Reagan’s “war on drugs.” I drop California and Colorado due to possible lower levels of enforcement which will directly affect the likelihood of marijuana arrest. As Cohen (2010) points out, the loosely worded medical marijuana legislation in those states have created a “regulatory vacuum.” In fact, the

7

There is also arrest for marijuana sale/manufacture, which is small and relatively constant across years. To be recorded as a sale arrest, the amount must exceed some minimum with intention to sell. Because marijuana transactions often involve small quantities, and sale intention is hard to prove, sale arrest is often due to large scale transactions. In fact, a proportion of marijuana possession arrestees are probably low level sellers. 8 2008 was the latest data available when I began this study. Although data through 2010 became available recently, looking at the period prior to 2009 has an advantage that the number of legal patients was relatively small, and the federal policy was fairly uniform prior to the Obama administration. In addition, strong economic recession may affect drug use. As of July 2012, most states that passed laws after 2008 have not yet accepted patient applications. (Only Arizona began to accept patient application since April 2011). 9 Decriminalization is better termed as depenalization, since marijuana possession is still legally a crime and it results in being arrested. Empirically, depenalization has either no or very small effect on marijuana use, and most citizens do not even know such legal change (MacCoun et al., 2009; MacCoun, 2010; Pacula et al., 2005).

7

Colorado attorney general, John W. Suthers, has said, “In Colorado it’s not clear what state law is.” (Johnson, 2009) It is even said, although perhaps with some exaggeration, that in Los Angeles, San Francisco and Denver there are more marijuana dispensaries than Starbucks coffee shops or CVS pharmacies (Coté et al., 2008; Osher, 2010). A large number of dispensaries could reflect a lenient attitude on the part of law enforcement and local government, because most of them are not strictly legal. For example, California law requires dispensaries to be non-profit and to have a consistent care-giving relationship with their patients. This is not generally the case. Moreover, both California and Colorado attempted to legalize marijuana; although these attempts were unsuccessful, some cities like Denver did pass laws that legalized marijuana. Even such laws are legally ineffective because they violate state laws, they can still influence local law enforcement. In 2010 California became the second state to decriminalize marijuana possession, making it a civil infraction rather than a crime, and the penalty for this infraction became one of the lowest in the U.S. 10 Since participation in the UCR program is generally voluntary, many agencies do not report every month or every year; and even when an agency reports, it may not report data in all categories. Empirically, most missing data is from agencies with small populations and those that do not report for a whole year (Lynch and Jarvis, 2008). Since it is not possible to distinguish a true zero from missing data in the UCR, I only use police agencies located in cities with populations greater than 50,000, as the FBI also checks and communicates regularly with these agencies to ensure data quality (Akiyama and Propheter, 2005). I restrict the sample to cities because marijuana transactions are closely related to social networks and the probability of arrest depends on population density. 11 Since population size is generally increasing over time, I include earlier observations of the above cities to make the panel more balanced. (I exclude 215 city-year observations that have populations less than 25,000). Similar to Carpenter (2007), and as is common in the criminology literature, I focus on adult male arrest, and I use observations only if the agencies report arrests for marijuana 10

In September 2010 California enacted a new law that effectively decriminalizes marijuana possession (S.B. 1449) by making possession of 28.5 grams or less of marijuana only a civil infraction. Massachusetts passed a similar law making marijuana possession (less than 1 oz.) a civil offense in November 2008. Although marijuana possession (1 oz. or less) is still legally a crime (class 2 petty offense) in Colorado, it only results in maximum fine of $100 and no jail time. 11 For agencies in MSAs with more than 50,000, about 70% of populations live in cities, and about 70% of all MSA agencies are city agencies. On average, marijuana arrest rate in cities is about twice as large as in non-cities.

8

possession for at least six months in that year. 12 The final panel consists of 562 cities and 8722 city-year observations in which about 40% of the cities are observed in at least 20 years. The sample covers 9 medical marijuana states that passed laws before July 2008, including Alaska, Hawaii, Maine, Montana, Nevada, New Mexico, Oregon, Rhode Island, and Washington. In addition to California and Colorado, Vermont is also not included in the sample due to sample construction. (Michigan passed its law in November 2008 and is coded as a non-medical marijuana state.) Table 1 lists the means and standard deviations of marijuana possession arrest rates per 100,000 among each age group for states with and without medical marijuana. States with medical marijuana laws have lower arrest rates in all age groups, likely due to a lower level of law enforcement, but the distributions of arrests among age groups are similar. The arrest rate is highest among those aged 18-24 and declines with age. In fact, the age distribution and the trending over time of the UCR arrest data are consistent with underlying marijuana use from survey data such as NSDUH and Monitoring the Future.

4.2.

The Results My primary empirical strategy involves estimating city- and year-specific marijuana

arrest rates as a function of whether the state has an effective medical marijuana law in place in that year. I begin by estimating the following model by OLS:

(1)

Yist = β Lawst + City fixed effectsi + Year fixed effectst + City linear time trendsit + City squared time trendsit + εist,

where Yist is the marijuana possession arrest rate among adult males or each male age group per 100,000 (or its log) for city i in state s and year t. Lawst is a dummy variable indicating whether a

12

I include 201 agency-year observations that report only in December since agencies may report annually; their mean and standard deviation are similar to observations that report for least six months. I drop 10 observations that have zero adult male marijuana arrest. I only consider males both to be consistent with the existing literature and because males are much more likely to be in the criminal justice system than are females. For example, the average arrest rate for adult males in my sample is seven times that for adult females.

9

state s had a medical marijuana law during year t. 13 In addition to city and year fixed effects, I include city-specific time trends to capture the time-varying unobservables within a city such as law enforcement. As a robustness check on functional form and a solution to zero-value problems of logarithms, I also estimate a fixed-effect Poisson model with the same specifications. In the main specification, I do not include any control variables because city-specific time trends and fixed effects have already accounted for any smooth-trending variables. 14 Throughout this paper, the estimated standard errors are clustered at the state level and therefore are robust to serial correlation, within-state spatial correlation, and heteroskedasticity. Table 2 shows the estimates of β based on Equation (1). The dependent variable is the adult male arrest rate in the upper panel and the logarithm of arrest rate in the middle panel; the lower panel shows results from a fixed-effect Poisson model in which the dependent variable is arrest rate. Columns (1) and (2) show the estimates from Equation (1), and they are positive and highly significant in all specifications. For example, based on Column (2), medical marijuana laws, on average, result in an annual increase of 26.94 adult male arrestees per 100,000 city residents; if we interpret the log points as percentage change, this is about a 22.5% increase in the adult male arrest rate. The estimates from the fixed-effect Poisson model are quantitatively similar. The point estimate (the partial effect on the logarithm of the conditional mean), 0.198, is very close to the estimate from the log specification (the partial effect on the conditional mean of the log arrest rate). It also implies an average partial effect of an increase of about 31 arrestees per 100,000 city residents (= 0.198×158.5, where 158.5 is the mean of adult arrest rate). One possible concern is that city-specific time trends over-fit the data. In Columns (3) and (4), I use state-specific time trends instead, and the results are nearly identical. The last two columns, (5) and (6), show qualitatively similar results estimated from a specification with only a separate group time trend for all medical marijuana states that passed laws before July 2008. Figures 1 provides graphical evidence of the effect of medical marijuana laws on arrests. The graph shows the average adult male marijuana arrest rate (in logarithms) before and after 13

For the first year, Lawst equals 1 if the law is effective before July 1st, and equals 0 otherwise. I code the law based on the effective date rather than the passing date (it only matters for Nevada).Note that there could be a huge time lag between the law being effective and the marijuana program starting to accept application. 14 Potential non-smooth control variables include legal change in marijuana penalty, unemployment rate, and 0.08% blood alcohol concentration (BAC) laws. As mentioned in Note 10, most studies do not support that change in penalty affect marijuana use. Unemployment rate and 0.08% BAC laws are included along with other control variables in Table 5 as robustness check.

10

the passage of medical marijuana laws, where the X-axis measures the year relative to the state's law change, with 0 denoting the year of passing the law, 1 denoting the following year, and so on. To create a synthetic control group, I compute an average of the log arrest rates in non-medical marijuana states for each year, and then take a weighted average of these yearly averages, in which the weights come from the relative composition of years in the treatment group (medical marijuana states). For instance, in “Year 0,” 58% of observations in the treatment group are from Oregon and Washington, which passed the laws in 1998 (coded as 1999); 2% of observations are from Maine, which passed the law in 1999 (coded as 2000); and so forth. So the weight put on the year 1999 average arrest rate in the control group is 0.58; the weight put on the year 2000 average arrest rate is 0.02, and so on. In other words, in “Year 0,” 58% of the observations in the control group are selected from year 1999, 2% are from 2000, and etc. The treatment group shows a persistent jump in the arrest rate from “Year -1” to “Year 0.” The magnitude of the jump in the log arrest rate is about 0.15, which is similar to the regression results above. Table 3 reports the effects of medical marijuana laws for each age group, in which I control for city quadratic time trends along with city and year fixed effects. 15 The upper panel includes the estimates from the level specification. Because there are many zero values in each age group, in the lower panel the results are estimated from a fixed-effect Poisson model. The relative magnitudes of these estimates are consistent with the age distribution in Table 1. They are larger among younger age groups and decrease with age, and as expected, the estimates for the oldest age groups (age 40-44 and age 45+) are small and insignificant. In the upper panel, the sum of the estimates in all age groups under 40 is about 27, which is nearly identical to the estimate in Table 2, and the majority of increase comes from people under age 30. In the lower panel, young adults also exhibit a greater effect than older age groups. The average partial effects from the fixed-effect Poisson model are quantitatively similar to the results in the upper panel. For example, the estimate implies 10.2 (= 0.215×47.54) additional male arrestees aged 18-20. Table A2 reports descriptive statistics and the estimates of β for juveniles aged 12-17 and 15-17. Because there are many zeros, I only use a linear model with level specification and a fixed-effect Poisson model. For all juveniles aged 12-17, the results are positive but never significant. For ages 15-17, the estimates are often significant for the level specification, but 15

In the lower panel, age 45+ is estimated only with linear time trends. Because there are too many zeros (nearly 20%), there is not enough variation after controlling for linear trends and fixed effects. The result is qualitatively similar if I control for state quadratic time trends.

11

somewhat noisy for the Poisson model. Because the data quality on juveniles is not as good as adults, the interpretation on these estimates should be cautious. 16 Although not reported here, the estimates also indicate a positive and significant effect of around 17% on adult females. In Table 4, I investigate the dynamic responses of the adult male arrest rate to adoption of medical marijuana laws. The left panel is a level specification and the right panel is a log specification. In the first two columns in each panel, I replace Lawst by a set of dummy variables, Years 0-1 through Years 8-9 (the maximum lag), that indicate each two-year interval after the medical marijuana laws are enacted. Note that the estimates for later years are driven mostly by Oregon and Washington. The estimated standard errors become larger when squared city time trends are included, but the magnitudes stay similar. Although the estimated effect on marijuana arrest seems to be increasing over time, a Wald test cannot reject the null hypothesis that the estimates for Years 0-1 through Years 6-7 are identical. (It is able to reject the null hypothesis when the estimate for Year 8-9 is included.) Therefore, the restriction of a constant effect on Lawst should be reasonable. The latter two columns in each panel include an additional dummy, Years (neg.1-2), that indicates the two-year interval before the passage of the laws. The estimates for this dummy are small and insignificantly different from zero, while the estimates remain similar for post-law dummies, which indicates that policy endogeneity is not a serious concern in this context. The results are similar if I include another dummy that indicates years three and four before the passage of laws (not reported). In Table 5, I check the robustness of the main results using different specifications and samples. In Column (1), as in Carpenter (2007), I scale arrest counts by a factor that equals the fraction reported of a year (12 divided by the number of months reported) using agencies that report at least six months (agencies that only report in December are excluded). In Column (2), I include city agencies that report any number of months without scaling. In Column (3) and (4), I include city police officer rate (from the UCR) and other state-level controls such as black male rate, unemployment rate, per capita local and state expenditures on police protection, per capita local and state expenditures on health and hospital expenditures, and 0.08 BAC laws. The results 16

The data on juvenile crime and custody rates are much less complete than the associated data for adults (Carpenter, 2007; Levitt, 1998). Also, the juvenile justice system is very different from the adult system such as its procedures, incentives, and sanctions. Note that the adult arrest rate is much lower in medical marijuana states, while their juvenile arrest rate is similar to non-medical marijuana states. Actually, it is possible that, after the passage of medical marijuana laws, the law enforcement towards juvenile becomes stronger due to reallocation the resources from adults to juveniles.

12

are nearly identical to these in Table 2. The sample size is smaller because 2001 and 2003 government expenditures were not developed by the Census Bureau due to sample redesigning. Because most state level controls are poorly estimated, it seems that fixed effects and city specific time trends have accounted for most of the variations from these controls. I prefer the specification without any controls as it includes more years in the sample. In Column (5) and (6), the dependent variable is adult male arrest rate aggregating from all available agencies in the UCR to the state level, including all cities and non-cities with positive populations. 17 The estimates are similar to Table 2, though they are a little smaller and insignificant when squared state time trends are included. The estimates from Tables 2–5 will be biased if law enforcement responds to medical marijuana laws. For example, if law enforcement becomes stricter, then the estimates are upward biased. In Table 6, I indirectly test some implications for potential changes in law enforcement. The upper panel is level specification and the lower panel is log specification (except for Column (2)). There is one state, Arizona, which did pass a referendum to legalize medical marijuana in 1996 that did not lead to an effective law. 18 If the increase in arrest in medical marijuana states are driven by unobservables common to states that initiate the legalization process rather than by the laws themselves, the arrest rate in Arizona would increase after passing the referendum. However, Column (1) shows that the arrest rate in Arizona actually decreases after the referendum, which is perhaps a result of that law enforcement becomes more lenient. In Column (2), the dependent variable is the marijuana possession arrest ratio of adult African Americans to all adults. 19 Because there are many zero values, I use a fixed-effect Poisson model in the lower panel. It is well documented that African Americans are much more likely (about 2.5 times) to be arrested for marijuana possession offences due to racial profiling, even though their prevalence rate is similar to whites (Golub et al., 2007; Ramchand et al., 2006; Reuter et al., 2001). 20 Therefore, if there were an increase in the proportion of African American arrestees, this would be a “smoking gun” that police are using racial profiling to increase 17

I exclude zero population agencies such as university or national park police departments. Arizona passed a referendum in 1996 (Proposition 200), which legalized medical marijuana under doctors’ prescription. However, the state legislature dismantled it through the terminology — "prescribe." Because marijuana is a schedule I drug, the DEA prohibits physicians from "prescribing" it. This made Proposition 200 ineffective. 19 The ratio of black arrestees includes females since the UCR does not separate gender within races. 20 Most of the literature only accounts for racial differences at the extensive margin. On the other hand, based on the TEDS, blacks have a larger proportion of high-frequency users. 18

13

marijuana arrests. However, the estimate is very close to zero and there is no evidence of change in racial composition after the passage of medical marijuana laws. In Column (3), I use the ratio of adult male marijuana possession arrests to total drug possession arrests as a measure of prevalence of marijuana use, which could be approximately interpreted as a change in relative demand for marijuana. 21 In Column (4), I use the ratio of adult male marijuana possession arrests to total crime arrests. One advantage of these measures is that they partially control for the level of law enforcement and eliminates the measurement error from estimated population (Fryer et al., 2005). If the increase in marijuana arrests is simply due to a stricter level of law enforcement, then medical marijuana laws would not have an effect on these ratios. The ratio in Column (3) is increased by 0.05 or a 16% increase, and the ratio in Column (4) is increased by 0.005 or a 25% increase. As for percentage change, the results are very similar to these in Table 2. A similar measure used in the literature for law enforcement toward illicit drug is the ratio of total drug arrests (for both sale and possession) to total arrests (Pacula et al., 2003; Resignato, 2000). For consistency, I only use adult male arrests to create this ratio. Column (5) does not show evidence of change in law enforcement towards overall drug offenses. Alternatively, I directly test whether medical marijuana laws increase arrests for nonmarijuana drug possession. In Columns (6), the dependent variables are adult male arrest rate for possession of all other drugs. The estimates are small and negative, so there is no evidence of an increase in arrests for non-marijuana drug possession. Cocaine is still the second most popular drug in the U.S., and I estimate the effect of laws on cocaine possession arrests (adult male) separately in Column (7). Contrary to the common belief that marijuana use will motivate hard drug use, the estimates indicate a decrease in cocaine use (although it could be a result of lower level of law enforcement). Interestingly, Anderson et al. (2012a) also find a huge reduction in teen cocaine use. As noted in the previous sections, the number of legal patients and dispensaries was relatively small in the sample period, and it is more so because California and Colorado are not in my sample. Therefore, the increase in marijuana use is most likely caused by changes in perceived legal and health risks. Because of the different legal process, it is plausible that referendum states are more liberal than lawmaker states, and a higher referendum passage rate 21

The UCR provides the category of total drug possession arrest as well as three subcategories other than marijuana: 1. Opium, cocaine, and their derivatives; 2. Truly addicting synthetic narcotics; 3. Other dangerous nonnarcotic drugs.

14

may also imply a more liberal attitude. In Table 7, I compare states passing laws through referenda to states in which legalization was enacted by lawmakers. If the existing public attitude is very liberal, so the perceived risk of marijuana is already low, then the implied effect on marijuana use will be smaller. Therefore, the arrest rate in referendum states with a high passage rate would increase less than in states with a low passage rate or lawmaker states. (See Table A1 for passage rate in each state.) I create Law×Referendum, the interaction term of Lawst and a dummy denoting referendum states, and Law×Pro rate, the interaction term of Lawst and passage rate (%) of referenda (0 for lawmaker states). The left panel is level specification and the right panel is log specification. The above prediction is supported in log specification: the estimates on Law×Pro rate and Law×Referendum are negative and significant. 22 In the level specification, however, the estimates are quite noisy and show a positive sign. Drug arrests often occur during or after transactions. However, unlike cocaine or heroin, marijuana transactions are closely related to social networks. About 60-80 % of people acquire marijuana from a friend and often for free; when there are monetary transactions, they are often exchanged indoors (Caulkins and Pacula, 2006; Pacula et al., 2010; Substance Abuse and Mental Health Services Administration, 2004; Taylor et al., 2001). The probability of being arrested is actually low, and therefore most arrestees are probably heavy users who make regular transactions. 23 Because heavy users are more likely to be associated with dependence and need for treatment; in the next section, I use marijuana treatment patients to provide direct evidence that medical marijuana laws increase use among heavy users.

5. The Analysis of Treatment Episode Data Set The treatment data is from the Substance Abuse and Mental Health Services Administration's (SAMHSA) Treatment Episode Data Set (TEDS) for the years 1992 through 2008. The TEDS collects admission data from all substance-abuse treatment facilities that receive public funding in each state. (Some states only collect data on public funded patients.) For each admission, the data identifies the primary, secondary, and tertiary substance abuse 22

Because people in referendum states should be more aware of the laws, the larger effect from lawmaker states suggests that the effect of legalization is stronger than the potential effect coming from the publicity surrounding the laws. Because the publicity effect is likely to be decreasing over time, the dynamic in Table 4 also implies that the publicity effect is not a major cause for the increase in marijuana use. 23 Although heavy users might prefer to buy more in each transaction and therefore make less transactions, most marijuana transactions involve small quantities because legal penalty directly depends on the weight of possession.

15

problem of the patient, his/her demographics such as gender and age, referral sources, and the number of prior treatment the patient ever received. 24 Similar to the UCR, each admission does not represent an individual, but it is possible to create a measure representing individuals by using only admissions without any prior treatment. This measure can avoid bias from recidivism that is particularly a potential problem for using treatment data (Anderson, 2010). I focus on two professional referral sources, alcohol or drug abuse care providers and health care providers that reflect professional criteria of marijuana abuse. These medical professionals are unlikely to be directly affected by general public's perception of risk and law enforcement. 25 Because the number of admissions in the TEDS greatly fluctuates in some state-years (see also Note 26), as commonly used by the SAMHSA, I create ratios of marijuana treatments to all substances treatments within professional referrals for each state. To be consistent with the UCR arrest, I only use adult (above age 18) male admissions and exclude California and Colorado. The sample has all medical marijuana state that passed laws before July 2008; except for Alaska that is missing for most of years, they have data in every year. 26 Table 8 presents descriptive statistics on the marijuana-related and marijuana-primary treatment ratios. I define marijuana-related treatment admissions if marijuana is identified as either primary, secondary or tertiary abuse problem, and marijuana-primary treatment admissions if marijuana is recorded as the primary abuse substance. The denominators of these ratios are professional referred treatment admissions for all substances. In the upper panel, the numerators of these ratios are marijuana treatment admissions with any number of previous treatment episodes. About a third of patients have marijuana abuse problem, but only less than 8% of patients have marijuana as their primary problem. To obtain a measure representing individuals, in the lower panel, I construct the ratios using only first-time marijuana treatment admissions. 27

24

About half of the adult male marijuana-related treatment are referred by criminal justice system, a quarter are individual referrals, and around 15% are professional referrals. The rest 10% are referred by community or religious organizations, and self-help groups such as Alcoholics Anonymous. 25 However, changes in perceived risk and law enforcement could affect people's behavior of seeking medical professionals. In fact, as in the UCR data, the estimates for California and Colorado laws are still negative and significantly different from other medical marijuana states. 26 Alaska does not report referral source for years 1999-2003, and it does not report any data for years 2004-2007. (Because the data quality in Alaska is so low, I also drop its 2008 data.) Moreover, since year 1999, there was probably a change in available funding or reporting process in Alaska and Washington: the total numbers of treatment reported were only about half to their previous levels. 27 It is not possible to observe whether a patient has had prior treatment episodes for a particular substance; only the number of previous treatment episodes a patient has had for any drug or alcohol problem is available.

16

On average, only 10% of admissions lack the information on previous treatments, although it is largely missing in some state-years. I restrict the sample to state-years which are missing less than 50% of this information, and scale the treatment ratios by the proportion of reporting data in each state-year. 28 Except for Washington that does not report the number of prior treatment for years 1992-1999, the information is very complete in medical marijuana states, with an average missing rate of 1.7%. It shows that about half of the patients are first-time patients. To examine the effect of medical marijuana laws on marijuana treatments, I estimate the following model by OLS:

(2)

Yst = β Lawst + State fixed effectss + Year fixed effectst + State time trendsst + εst ,

where Yst is the treatment rate or its log in state s and year t. As in the UCR analysis, I do not include any controls to keep a larger sample size. The results in Table 9 below are nearly identical when the same set of state-level controls is included (not reported). Table 9 shows the estimates of β from Equation (2). The first two columns are from the level specification, the second two columns are from the log specification, and the last two columns are from a fixed-effect Poisson model. The upper panel shows the results for all treatment ratios (with any number of previous treatment episodes). In terms of percentage change, the estimated effects are a little larger but noisier for marijuana-primary ratio. Specifically, on average, medical marijuana laws are associated with a 9-12% increase in marijuana-related treatment ratio, or 10-17% for marijuana-primary treatment ratio. Because there may be capacity constraints for treatment facilities, the smaller results are expected and consistent with the results from the UCR. Because a proportion of patients will repeatedly enter treatments due to addiction, we would expect the estimates based on patients with previous treatment episodes to be smaller than estimates from first-time patients. However, in the lower panel in Table 9, the estimates are actually greater. The estimates are similar across marijuana-related or -primary treatment ratios. Specifically, on average, medical marijuana laws are associated with a 17-25% increase in first28

I also exclude 14 state-years (including Rhode Island in 2003 and 2004) with zero first-time treatment. These observations are clearly missing data because they differ a lot from data in the previous or later years. The regression results in Table 8 are slightly greater without scaling or excluding zeros. (The estimated standard errors are 10-20% larger without scaling).

17

time treatments. In fact, the estimates in the upper panel are entirely driven by first-time treatments as that first-time treatments account for around 40-50% of all marijuana treatments. Although not reported, I estimate the effect of laws on patients with at least one previous treatment episodes, and the results are nearly zero. 29 It is straightforward to see graphically that the estimates in Table 9 are driven by firsttime treatments. Figure 2, constructed in the same way as Figure 1, shows the effect of laws on marijuana-related treatment ratios (in logarithm). The upper graph is from all treatments, and the lower graph is from first-time treatments. Both graphs show similar patterns of an increase in marijuana-related treatment ratios after the passage of laws in medical marijuana states; however, the magnitude from all treatments is much smaller than the magnitude from first-time treatments. In the Appendix, Table A3 shows the descriptive statistics and regression results from all referral sources. The data contain patients referred by criminal justice system and common individuals (or self referrals) that are probably affected by law enforcement or perceived risk among public. In general, the results are qualitatively similar for marijuana-primary treatments, although they are more sensitive to time trends specification especially for marijuana-related treatments. The estimates based on first-time patients are still larger than patients with previous treatment episodes. Figure A1 is the graphical analysis, and the graphs look very similar to Figure 2. In Table A4, I estimate the effect on male juveniles (age 12-17) for all source referrals. Because the number of juvenile patients is very small in some state-years, I drop 27 state-years in which the total number of patients (for any substances and from any referral sources) is less than 20. As 75% of juveniles are first-time patients, I do not separately estimate effects for firsttime treatments. Consistent with the popularity of marijuana among juveniles, 80% of treatment juvenile patients report marijuana abuse, and it is the primary abuse problem for nearly 60% of them. In contrast to previous studies that do not find any effect of laws on juvenile, the results indicate around a 5-15% increase in marijuana-related or -primary treatments. 30 I do not report 29

I also try to restrict the sample to be the same across the two panels, but the results are nearly identical. So the smaller estimates in the upper panel are not due to sample difference. 30 In Anderson et al. (2012a), they use state populations as the denominators to create treatment rates for teenagers aged 15-17 and 18-20, and their estimates are often negative. However, using population as denominator may be inappropriate as the TEDS data greatly fluctuates in some states. On the other hand, the estimates based on treatment rates using populations as the denominators are similar to results reported here if Washington are excluded (see also Note 26).

18

the results separately for professional referrals as the number of juvenile patients are quite small; the estimates for professional referrals are a little greater (6-19% increase) but much more noisier.

6.

Discussion of results and conclusion In this paper, I estimate the effect of medical marijuana laws on illegal marijuana use

based on marijuana possession arrests. My estimates show a positive effect that is, as expected, strongest among young adults. The results suggest that usage increased by roughly 20% among adult males. There is no evidence that these results are driven by stronger law enforcement. Based on the relatively small number of legal patients and dispensaries, and the smaller effect of laws from potentially more liberal states (Table 7), an important potential channel through which these laws increase marijuana use is through a lower perceived risk. Those who are arrested for marijuana possession are likely to be heavy users, who may be most prone to dependence and in need of professional treatment. I use marijuana treatment referrals by medical professionals as a proxy for heavy usage, and I find that this treatment increases by around 10% after the passage of laws. Somewhat surprisingly, the estimates on firsttime treatments indicate a greater effect of around 20%. In contrast to previous studies that use measures for general use rate among juveniles and do not find any effect of these laws, the estimates indicate a 5-15% increase in juvenile treatments. A 10-20% increase is a large effect but it is plausible for heavy users. Based on existing studies, MacCoun (2010) suggests that the non-price effect of marijuana decriminalization is around 35% increase in general use rate (use in past month). 31 Although medical marijuana laws represent a less dramatic change than decriminalization, a 10-20% increase is not particular large for heavy users as previous research suggests that heavy users are disproportionately responsive to legal changes (Becker and Murphy, 1988). This magnitude also seems reasonable in comparison to policy changes regarding other substances. For example, Carpenter (2004) finds that zero tolerance drunk driving laws are associated with a 13-18% reduction in heavy drinking. Similarly, DiNardo and Lemieux (2001) find that marijuana use rate among high school seniors increases by around 10% after the legal drinking age increased to 21.

31

Based on the 2002-2008 NSDUH, I find medical marijuana laws increase general use by around 5% for adults, although it is only significant for ages 18-25.

19

The difference between the estimated effects based on first-time treatments and all treatments has interesting implications. First, it implies that, on average, marijuana is not strongly addictive, which is consistent with existing medical evidence. Second, it suggests that medical marijuana laws do not have a significant effect on strongly addictive patients who repeatedly enter treatments. These patients could be "always-takers" who would be heavy marijuana users regardless of marijuana's legal status. Finally, consistent with the negative estimates on cocaine arrest from Table 6, it does not support the popular belief that use of marijuana increases abuse of hard drugs. Strongly addictive patients who repeatedly enter treatment facilities are often users of hard drugs such as cocaine and heroin. For first-time marijuana-related treatment patients, 37% also report cocaine abuse, and 6% report heroin abuse. On the other hand, for patients with at least one previous treatment episode, the proportion that reports cocaine and heroin abuse increases to 49% and 11%, respectively. Moreover, this pattern is monotonically increasing with the number of previous treatments. Although the estimates in this paper are only appropriate for inference on heavy users, they may be more relevant to policy concerns because heavy marijuana users are often associated with negative health and social outcomes, such as developing dependence, the need for treatment and future use of hard drugs (Chen et al., 1997; Fergusson et al., 2006; Gruber et al., 2003). A 20% increase in heavy users, as indicated by both arrest and first-time treatments, represents a nontrivial cost to the society. On the other hand, based on the estimates from on all treatments, the net effect on treatment is only 10%, which may be due to substitution between marijuana and other substances. This substitution can be viewed as a benefit of medical marijuana laws, and there could be additional benefits; for example, Anderson and Rees (2011) and Anderson et al. (2012b) show evidence for a decrease in drunk driving and suicide. Therefore, evaluating the net effect of medical marijuana laws requires a more careful benefit and cost analysis that is beyond the scope of this study.

Reference: Akiyama Y, Propheter SK. 2005. Methods of Data Quality Control: For Uniform Crime Reporting Programs. Federal Bureau of Investigation, Criminal Justice Information Services Division; 2005.

Anderson DM. Does information matter? The effect of the Meth Project on meth use among youths. Journal of Health Economics 2010;29; 732-742. 20

Anderson DM, Hansen B, Rees DI. Medical Marijuana Laws and Teen Marijuana Use. IZA Discussion Paper No. 6592 2012a.

Anderson DM, Rees DI. Medical Marijuana Laws, Traffic Fatalities, and Alcohol Consumption. IZA Discussion Paper No. 6112 2011.

Anderson DM, Rees DI, Sabia JJ. High on Life? Medical Marijuana Laws and Suicide. IZA Discussion Paper No. 6280 2012b.

Becker GS, Murphy KM. A Theory of Rational Addiction. The Journal of Political Economy 1988;96; 675-700.

Carpenter C. How do Zero Tolerance Drunk Driving Laws work? Journal of Health Economics 2004;23; 61-83.

Carpenter C. Heavy Alcohol Use and Crime: Evidence from Underage Drunk‐Driving Laws. Journal of Law and Economics 2007;50; 539-557.

Caulkins JP, Pacula RL. Marijuana markets: Inferences from reports by the household population. Journal of Drug Issues 2006;36; 173-200.

Chen K, Kandel DB, Davies M. Relationships between frequency and quantity of marijuana use and last year proxy dependence among adolescents and adults in the United States. Drug and Alcohol Dependence 1997;46; 53-67.

Cohen PJ. Medical Marijuana 2010: It's Time to Fix the Regulatory Vacuum. The Journal of Law, Medicine & Ethics 2010;38; 654-666.

Conlin M, Dickert‐Conlin S, Pepper J. The Effect of Alcohol Prohibition on Illicit‐Drug‐Related Crimes. Journal of Law and Economics 2005;48; 215-234.

Coté J, Allday E, Buchanan W, Lagos M. 2008. Feds say S.F. has more pot clubs than Starbucks, but it might not add up. San Francisco Chronicle. 2008.

21

DiNardo J, Lemieux T. Alcohol, marijuana, and American youth: the unintended consequences of government regulation. Journal of Health Economics 2001; 991-1010.

Drug Enforcement Administration D. 2011. The DEA Position On Marijuana. 2011.

Fergusson DM, Boden JM, Horwood LJ. Cannabis use and other illicit drug use: testing the cannabis gateway hypothesis. Addiction 2006;101; 556-569.

Fryer RG, Heaton PS, Levitt SD, Murphy KM. Measuring the Impact of Crack Cocaine. National Bureau of Economic Research Working Paper Series 2005;No. 11318.

Gfroerer JC, Wu L-T, Penne MA. Initiation of Marijuana Use: Trends, Patterns, and Implications. 2002.

Golub A, Johnson BD, Dunlap E. The Race/Ethnicity Disparity in Misdemeanor Marijuana Arrests in New YorK City. Criminology & Public Policy 2007;6; 131-164.

Golub A, Liberty HJ, Johnson BD. Inaccuracies in self-reports and urinalysis tests: Impacts on monitoring marijuana use trends among arrestees. Journal of Drug Issues 2005;35; 941-966.

Gorman DM, Huber CJ. Do medical cannabis laws encourage cannabis use? International Journal of Drug Policy 2007;18; 160-167.

Gruber AJ, Pope HG, Hudson JI, Yurgelun-Todd D. Attributes of long-term heavy cannabis users: a case-control study. Psychological Medicine 2003;33; 1415-1422.

Harper S, Strumpf EC, Kaufman JS. Do Medical Marijuana Laws Increase Marijuana Use? Replication Study and Extension. Annals of epidemiology 2012.

Harrison L, Hughes A. 1997. The Validity of Self-Reported Drug Use: Improving the Accuracy of Survey Estimates. In: Abuse NIoD, NIH Publication No. 97-4147. 1997.

Johnson K. 2009. States pressed into new role on medical marijuana. The New York Times. 2009.

22

Johnston LD, O'Malley PM, Bachman JG, Schulenberg JE. 2011. Monitoring the Future: national results on adolescent drug use: Overview of key findings, 2010. Institute for Social Research,The University of Michigan Ann Arbor; 2011.

Levitt Steven D. Juvenile Crime and Punishment. The Journal of Political Economy 1998;106; 1156-1185.

Lynch JP, Jarvis JP. Missing Data and Imputation in the Uniform Crime Reports and the Effects on National Estimates. Journal of Contemporary Criminal Justice 2008;24; 69-85.

MacCoun R, Pacula RL, Chriqui J, Harris K, Reuter P. Do Citizens Know Whether Their State Has Decriminalized Marijuana? Assessing the Perceptual Component of Deterrence Theory. Review of Law & Economics 2009;5; 347-371.

MacCoun RJ. Estimating the non-pricee effect of legalization on cannabis consumption. Rand Corporation 2010;Working paper WR-767-RC.

Miller RN, Kuhns JB. Exploring the Impact of Medical Marijuana Laws on the Validity of Self-Reported Marijuana Use Among Juvenile Arrestees Over Time. Criminal Justice Policy Review 2011.

O'Connor A. 2011. Marijuana Use Growing Among Teenagers. The New York Times. 2011.

Osher CN. 2010. As dispensaries pop up, Denver may be Pot Capital, U.S.A., The Denver Post. 2010.

Pacula RL, Chriqui JF, King J. 2003. Marijuana Decriminalization: What does it mean in the United States? , NBER Working Paper No. 9690. 2003.

Pacula RL, Chriqui JF, Reichmann DA, Terry-McElrath YM. State Medical Marijuana Laws: Understanding the Laws and their Limitations. Journal of Public Health Policy 2002;23; 413-439.

Pacula RL, Grossman M, Chaloupka FJ, O'Malley PM, Lloyd D. Johnston, Farrelly MC. 2001. Marijuana and Youth. In: Gruber J, Risky Behavior among Youths: An Economic Analysis. University of Chicago Press; 2001.

23

Pacula RL, Kilmery B, Grossmanz M, Chaloupka FJ. Risks and Prices: The Role of User Sanctions in Marijuana Markets. The B.E. Journal of Economic Analysis & Policy 2010;10; 1-36.

Pacula RL, MacCoun R, Reuter P, Chriqui J, Beau Kilmer KH, Paoli L, Schäfer C. 2005. What does it mean to decriminalize marijuana? A cross-national empirical examination. In: Grossman M, Lindgren B, Substance Abuse: Individual Behaviour, Social Interactions, Markets and Politics, vol. 16. Elsevier: Amsterdam; 2005.

ProCon.Org. 2012a. 17 Legal Medical Marijuana States and D.C. 2012a.

ProCon.org. 2012b. How many people in the US use medical marijuana? 2012b.

Ramchand R, Pacula RL, Iguchi MY. Racial differences in marijuana-users’ risk of arrest in the United States. Drug and Alcohol Dependence 2006;84; 264-272.

Resignato AJ. Violent crime: a function of drug use or drug enforcement? Applied Economics 2000;32; 681-688.

Reuter P, Hirschfield P, Davies C. 2001. Assessing the Crack-down on Marijuana in Maryland. The Abell Foundation: Baltimore, MD.; 2001.

Substance Abuse and Mental Health Services Administration S. 2004. How youth obtain marijuana. Rockville, MD; 2004.

Taylor BG, Fitzgerald N, Hunt D, Reardon JA, Brownstein HH. 2001. ADAM Preliminary 2000 Findings on Drug Use and Drug Markets -- Adult Male Arrestees. United States Department of Justice, National Institute of Justice: Washington, DC; 2001.

Wall MM, Poh E, Cerdá M, Keyes KM, Galea S, Hasin DS. Adolescent Marijuana Use from 2002 to 2008: Higher in States with Medical Marijuana Laws, Cause Still Unclear. Annals of epidemiology 2011;21; 714-716.

24

Table 1: UCR Descriptive Statistics in Each Male Age Group (1988-2008) All states

Medical marijuana states

Other states

Mean

Std. Dev.

Mean

Std. Dev.

Mean

Std. Dev.

All adult (age 18+)

158.46

130.99

118.12

84.64

162.06

133.77

Age 18 - 20

47.54

38.50

34.70

29.71

48.69

38.99

Age 21 - 24

39.41

34.66

27.64

21.13

40.47

35.43

Age 25 - 29

28.29

25.90

20.15

15.41

29.01

26.52

Age 30 - 34

17.51

16.39

13.26

10.37

17.89

16.77

Age 35 - 39

11.62

11.73

9.53

7.69

11.81

12.01

Age 40 - 44

7.29

8.00

6.31

6.09

7.38

8.15

Age 45 +

6.79

8.61

6.54

7.56

6.81

8.70

City-year obs.

8,722 (562 cities)

715 (48 cities)

8,007 (514 cities)

Note.― Medical marijuana states include only states that passed laws before July 2008; states that passed laws afterward are in "other states." California, Colorado and Vermont are not in the sample.

25

Table 2: Effects of Medical Marijuana Laws on Adult Male Arrests (1)

(2)

(3)

(4)

(5)

(6)

24.06** (10.83)

19.71* (11.62)

0.303*** (0.084)

0.253*** (0.061)

0.150** (0.074)

0.129* (0.078)

Linear (Group)

Quadratic (Group)

Linear Model 28.22*** (6.23)

26.94*** (6.58)

29.83*** (5.09)

25.59*** (5.16)

Log-Linear Model 0.282*** (0.095)

0.225*** (0.076)

0.288*** (0.090)

0.187*** (0.061)

FE Poisson 0.168*** (0.060)

0.198*** (0.042)

0.177*** (0.040)

0.168*** (0.036

Time trend specification Linear (City)

Quadratic (City)

Linear (State)

Quadratic (State)

Obs. 8722 Note.― All specifications include city and year fixed effects. Robust standard errors are reported in parentheses, and they are clustered at the state level. *** p