research paper english about (legalization of marijuana)

art1_page_1.pdf

Home >English homework help >research paper english about (legalization of marijuana)

CHAPTER THREE

Consequences of Marijuana Use

Introduction

The debate about marijuana legalization has been ongoing for many decades, and each point of view offers a variety of arguments (see MacCoun and Reuter, 2001; Caulkins, Coulson, et al., 2012). Some of the arguments involve what philosophers call deontological moral concerns— roughly, concerns about the inherent rights and wrongs of using a mind-altering substance (other than under medical direction) on the one hand and concerns about the propriety of paternalistic restrictions on personal liberty on the other. However, much of the debate revolves around what is known and unknown about the practical consequences of marijuana use and about marijuana production and distribution. Although we acknowledge the deontological set of issues, this chapter focuses on these practical consequences.

It is important to distinguish two types of questions, each of which is important for thinking about the effects of marijuana legalization on health and other outcomes:

1. What effects does marijuana use have on health and other outcomes? 2. What effects would a change in marijuana laws have on patterns of use and, hence, on

health and other outcomes?

The effects of marijuana use have been studied extensively, and, because there are several excellent recent reviews of this literature (e.g., Danovitch, 2012; Fischer et al., 2011; Hall, 2014; Hall and Degenhardt, 2009; Volkow et al., 2014; J. Williams and Skeels, 2006), we are selec- tive rather than exhaustive in our coverage. As we explain in the next section, the existence of this research does not mean that the question has been answered; often, there are inherent limits on what can be known or on the types of research that have been used. The question of how changing the laws would affect health and safety has received less attention. Indeed, the debate often fails to recognize that it is a distinct question. But we argue that, even if all the uncertainty about the health and safety consequences of marijuana use were somehow resolved, one would not necessarily then know how marijuana legalization would affect those harms for at least four reasons.

First, the effects of marijuana legalization will depend on whether and how it influences the prevalence of use—meaning the total number of users—and the age distribution of those users. For example, is an increase in use by 40- to 44-year-olds as important as a comparable increase in use by 16- to 20-year-olds? The answer is “not likely,” given that the risk of engag- ing in other risky behaviors is higher for youth and the existence of concerns about marijuana’s effect on brain development. And the composition of any new using population might be important. For example, MacCoun and Reuter (2001) argued that people who would use

Co py ri gh t © 2 01 5. R AN

D Co rp or at io n. A ll r ig ht s re se rv ed . Ma y no t be r ep ro du ce d in a ny f or m wi th ou t pe rm is si on f ro m th e pu bl is he r, e xc ep t fa ir u se s pe rm it te d un de r U. S. o r ap pl ic ab le

co py ri gh t la w.

EBSCO Publishing : eBook Collection (EBSCOhost) - printed on 4/18/2017 8:38 PM via UNIV OF SOUTH CAROLINA - AIKEN AN: 1020749 ; Caulkins, Jonathan P., Kilmer, Beau, Rand Corporation, Kleiman, Mark, MacCoun, Robert J., Midgette, Gregory, Oglesby, Pat, Pacula, Rosalie Liccardo, Re.; Considering Marijuana Legalization : Insights for Vermont and Other Jurisdictions Account: usocar

28 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

under a legalization regime but not a prohibition regime are likely to be more-cautious people, implying that, other things equal, the average user might be a more careful user under legal- ization. We examine potential changes in the prevalence of marijuana use directly in Chapter Seven (also see MacCoun and Reuter, 2001; Kilmer, Caulkins, Pacula, et al., 2010).

Second, any effects on the frequency and intensity of marijuana use are at least as important as effects on the prevalence of use—and probably more important. This issue has received rela- tively little attention in the marijuana literature, but it is a major focus of theory and research on alcohol policies and outcomes, in which changes in the median of the consumption distri- bution can have implications that differ from those for changes among heavy users (e.g., Skog, 2006). Marijuana legalization might also influence the length of a using career; however, what little empirical evidence exists on this is mixed. MacCoun (2011b) found evidence against this in the Netherlands; however, van Ours and Williams (2007) used data from Australia to show that lower marijuana prices are statistically associated with earlier ages of initiation and that earlier ages of initiation are associated with longer use careers. Similarly, Jenny Williams and Anne Line Bretteville-Jensen (2014) showed that marijuana decriminalization in Australia led to a shift toward earlier age of first use.

Third, marijuana legalization is likely to change the modalities of both use (e.g., smoking, vaping, eating, drinking) and sales (who provides the marijuana and how it is promoted to the customer).1

Finally, marijuana regulation might change some of the risks associated with marijuana use, via THC potency, quality control, time and place regulations, education opportunities, or differential pricing and taxation. The extent to which these factors will change is unknown because it will depend on the regulations and taxes adopted (see Chapters Five and Six); we simply raise the point here that these sorts of factors could matter when thinking about poten- tial harms associated with use.

All these considerations make the analysis of the public-health and safety impacts of marijuana legalization quite complex. To make decisions about marijuana laws, we would like to know the total net consequences across users, doses, and outcome dimensions (e.g., respira- tory health, mental health, road safety). But as MacCoun (1998) and MacCoun and Reuter (2001) argued,

= × ×

total harm average harm per dose number of doses per user number of users,

or, more simply, total harm = harmfulness × intensity × prevalence. For any given outcome dimension, marijuana could be a significant source of harm per dose, and yet there might be little aggregate effect of legalization if use did not increase. Alternatively, legalization might sig- nificantly reduce average harm on a given outcome dimension yet raise total harm by increas- ing the amount of marijuana used or the age at which use begins. And total harm might decline along some outcome dimensions but increase along others. Finally, total harm across all citizens might look different from total harm for some subgroups, e.g., teenagers and young

1 Vaping is inhaling vapor from material that is heated but not burned like it is in conventional cigarettes. The distinction is elaborated in Chapter Four.

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 29

adults, people in poorer neighborhoods, polydrug users (that is, users of multiple drugs), people susceptible to mental health problems.

With these complexities and limitations in mind, the rest of this chapter discusses some of the most-relevant public-health and safety issues that Vermont should consider when decid- ing whether to pursue legalization. We begin with a careful discussion about why, even when there is lots of research showing a strong positive association between marijuana use and some outcome variable (e.g., crime or psychosis), that research might not indicate that we can inter- pret the association as causal. It also can help policymakers understand why, in some areas of research, the findings from a few methodologically strong studies outweigh the evidence from dozens of other, less rigorous studies. We then discuss in greater detail the general scientific consensus of the literature on the consequences of marijuana use.

Limitations of Existing Studies

There are inherent methodological challenges in studying the consequences of an illicit psy- choactive substance. Even for licit psychoactive substances (e.g., alcohol), dose–response effects are usually nonlinear, making extrapolation beyond the dose range in a given study difficult.2 And ethical barriers (appropriately) limit the doses and settings that we can study in con- trolled experiments involving true randomization of people to treatment and control condi- tions. When human experiments are possible, they are often of questionable generalizability; animal experiments are less restricted but even more limited in their generalizability. When the psychoactive substance is illegal, there are even greater barriers to research—ethical, legal, political, and practical.

Ethical considerations often preclude conducting truly randomized experiments, so sci- entists must often interpret evidence from purely observational data. Causal inference, how- ever, can only be confidently drawn when the people in the so-called treatment group (in this case, the marijuana users) are similar in all important ways to people who are untreated (the nonusing control group), both in terms of easily observed differences (e.g., educational attain- ment, gender, prior drug use) and less easily measured differences (e.g., maturity, self-control, susceptibility to mental health problems) before marijuana use begins. This is very difficult to ascertain when making use of real-world observational data because marijuana users differ from nonusers in myriad ways. Thus, the vast majority of studies (regardless of techniques applied) rely on correlational methods because they account for differences only in observable factors between users and nonusers that the data capture. Important differences between users and nonusers could exist that are unmeasured in the data but correlated with the decision to use or maintain use of marijuana. Such problems are common in social science and epidemio- logical research. This is the primary reason correlational methods cannot clearly establish true causal connection.

Statistical techniques have evolved such that stronger statements of associations can be made, or at least some of the bias caused by the nonrandom assignment can be identified

2 Traditionally, medical researchers have viewed dose–response relationships as a valuable cue for inferring causation in purely correlational designs, but this approach is highly problematic because dosages are often correlated with confounding risk propensities.

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

30 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

and adjusted for. Correlational studies involving individual-level data are stronger than other observational studies when the following are true:

• They study people across both settings and time periods so they can adjust for any stable but unobservable individual differences (so-called fixed-effect methods).

• They include control variables that are likely to be highly correlated with hypothesized important unobservable factors, such as deviance, maturity, and self-control.

• They use matching methods that try to make the user and nonuser groups as similar as possible on all the observable variables before the risk factor (marijuana use) is introduced (so-called propensity-score methods).

But even these sorts of rigorous statistical methods cannot completely eliminate the inherent causal ambiguity caused by nonrandom assignment; at best, such techniques reduce or identify some of the biases that self-selection creates (Manski, 2007).

A persistent inferential problem, even in strong observational studies, involves the classic difficulty of differentiating among three interpretations of any observed association:

• A can cause B (e.g., marijuana use can cause some health outcome). • B can cause A (e.g., people with the health problem seek out more marijuana, e.g., to self-

medicate) • Some third variable C can cause both A and B (e.g., the children of neglectful parents

are more likely both to start using cannabis at younger ages and to do badly in school), making the A–B correlation spurious.

The “B can cause A” account is plausible given that people seek out marijuana as medicine for a variety of conditions, but it is usually at least partially testable by examining whether the condition preceded the marijuana use. The “C can cause A and B” third-variable possibility is much more difficult to rule out. For example, marijuana use is strongly associated with drop- ping out of high school, but low attachment to conventional norms and roles might cause both marijuana use and dropping out. Multivariate studies can control for situational and personal characteristics that are measured, but it is not possible to rule the effect of unobserved factors in the absence of random assignment. There is very strong evidence for a shared risk propen- sity for substance use, delinquency, school problems, accidents, and mental health problems (see Elliott, Huizinga, and Menard, 1989; Kessler et al., 2005). So almost certainly some of the observed associations between marijuana use and poor outcomes are not causal. But it is much harder to judge whether that means that 90 percent of the association is causal, or only 10 percent, or even none at all.

To raise these concerns is not to diminish the seriousness of the potential harms associ- ated with marijuana. Readers will differ in their sense of where the burden of proof lies; should we err on the side of assuming that marijuana causes the associations until proven otherwise (a sort of precautionary principle)? Should we assume the opposite? Or should we split the differ- ence and assume expected harms that have been discounted for uncertainty?

In the abstract, the situation is similar to that posed by correlational evidence linking tobacco and cancer. Health experts were willing to act on a causal interpretation (tobacco causes cancer) well before it could be established rigorously—a decision that seems wise and prudent when considered today. But the analogy to the current marijuana debate is prob-

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 31

lematic. In the case of tobacco, there were not compelling reasons to believe that cancer (or some genetic propensity to cancer) somehow caused people to smoke tobacco. However, for marijuana, there is a plausible hypothesis that its use is a form of self-medication for people coping with other health and behavioral problems. (As we note below, researchers are begin- ning to test this hypothesis; so far, it does not appear to be a major part of the associations, but more research is needed.) More important is the fact that the tobacco industry’s accounts of how smoking and cancer might both be attributable to various third variables were not sci- entifically persuasive (see Michaels and Monforton, 2005). In contrast, there are very strong reasons to believe that marijuana use and various health and behavioral problems might share common causal antecedents—personality traits, emotional problems, poor coping and self- control mechanisms, bad peer influences, and disordered families and neighborhoods.

But observing that the evidence is not as strong as it was for tobacco is not, in some sense, terribly reassuring; tobacco use was killing hundreds of thousands of people per year. Harms can reach a threshold for alarm far before reaching that level.

Consequences of Marijuana Consumption on Health

Several recent comprehensive reviews have assessed the current scientific evidence regarding the acute and chronic health effects associated with marijuana use (Hall, 2014; Volkow et al., 2014; Gordon, Conley, and Gordon, 2013; Hall and Degenhardt, 2009). The upshot of these reviews is that persistent, frequent use is unhealthy, mostly because of modest elevation in the risk of a diffuse range of adverse outcomes, but also because of a risk of marijuana dependence that is much greater (at least one in nine users) than most risk levels reported in the health and safety literature.

The literature focuses primarily on traditional use modalities (mainly smoking) so does not necessarily address the full range of marijuana products that are becoming increasingly popular. Furthermore, very few and only the more-recent studies provide explicit informa- tion regarding the specific doses of cannabinoids (in particular, THC and CBD) consumed.3 Because we expect that legal markets will shift the types of products, the average potency, and the ratios of specific cannabinoids contained within the products consumed, the relation- ships identified from these studies might no longer hold. The relationships could get stronger or weaker, depending on how the presence or absence of particular cannabinoids contributes to the relationship. Indeed, evidence suggests that the average potency of seized marijuana has been increasing in states that provide explicit legal protection for marijuana dispensaries (Sevigny, Pacula, and Heaton, 2014). If this trend continues with legalization, insights from the previous health literature could understate future health impacts. Similarly, to the extent that contaminants (mold, bacteria, pesticides, or other additives) contribute to some of the observed health effects associated with marijuana use, government product testing and label- ing rules could reduce some of the health hazards (see Daley, Lampach, and Sguerra, 2013).

3 Ironically, even if outcomes became more severe, the correlation between dose and outcome could get weaker. Rather than rising linearly, dose–response relationships tend to be S-shaped, jumping up at a threshold-like inflection point and then leveling off. Thus, if the number of observations drops at the low end of the potency range, correlations between dose and various outcomes might actually become hard to detect.

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

32 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

The remainder of this section addresses some of the acute and chronic health risks associ- ated with marijuana, as well as the medical benefits.

Acute Health Risks Fatal Overdose

Unlike many other psychoactive substances, marijuana does not pose any serious risk of a fatal overdose. Hall (2014) noted that the dose range that is fatal in animal studies is “far greater . . . than even a very heavy cannabis user could use in a day” and that there are “no reports of fatal overdoses in the epidemiological literature.” Hall (2014) categorized descriptions that Hartung et al. (2014) gave of the deaths of two young men that were plausibly attributable to “acute cardiovascular complications evoked by smoking cannabis” separately as cardiovascular events, not overdoses. But even under a broader notion of fatal overdose that includes such car- diovascular events, given the vast number of incidents of consumption in recent decades, the risk of acute death is extremely low.

Accidental Poisoning

As noted in Chapter Two, some marijuana users in Vermont do end up in EDs and being admitted to the hospital because of marijuana. Twenty years ago, one might have assumed that ED mentions of marijuana were ancillary; the patient went to the ED because of the effects of some other drug but just happened to also be using marijuana. However, ED mentions of mar- ijuana have grown enormously since the early 1990s, including episodes in which marijuana is the only drug mentioned (SAMHSA, 2002, 2013a). So it is of interest to ask how legalization might affect those trends.

ED episodes involving children are a particular concern. Colorado has experienced an increase in young children admitted to EDs because of accidental ingestion of marijuana- infused edibles (e.g., Ingold, 2014b; Wang, Roosevelt, and Heard, 2013). According to an article in The Denver Post published in May 2014, nine children went to just one hospital ED (Children’s Hospital Colorado) between January and May 2014, which was more than it saw the entire year before. Seven of the nine were admitted to the hospital’s intensive care unit to be watched due to extreme sedation and agitation, and one required a respirator. A more formal analysis focused on medical marijuana published in The Journal of the American Medical Association found a “new appearance of unintentional marijuana ingestions by young children after modification of drug enforcement laws for marijuana possession in Colorado” in 2009 (Wang, Roosevelt, and Heard, 2013). It is important to be careful about drawing conclusions from simple pre–post analyses without adequate control variables. For example, any increase in reported pediatric incidents might be at least partially attributable to rising levels of THC potency, which could increase the likelihood that parents seek medical help. But it seems clear that brightly packaged edibles, designed to mimic regular sweets, pose a very plausible risk of accidental ingestion.

The Rocky Mountain High Intensity Drug Trafficking Area (HIDTA) reports that ED visits related to marijuana (for those of any age) increased in Colorado by 29 percent in the first year after legalization of personal possession and use (12,888 in 2013 versus 9,982 in 2012), and that was before stores selling recreational marijuana opened.

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 33

Anxiety

An important acute health risk of using marijuana is anxiety. In 2011, marijuana accounted for 38 percent of ED visits in which illicit drugs were mentioned (about 450,000 visits, or one for every 40 past-month marijuana users; SAMHSA, 2013a, Table 4); this is a 62-percent increase since 2004 (SAMHSA, 2013a, Table 9). The most common stated reason for these visits is “unexpected reaction” (Kissin and Ball, 2003), which is usually a transient panic attack brought on by extreme intoxication.

Cognitive Impairment

The acute (short-term) cognitive effects of marijuana intoxication are easier to establish than the long-term effects. Studies show that being under the influence of marijuana can impair short-term memory, attention, reaction time, and psychomotor performance (Hall and Pacula, 2003). Attempting to estimate the aggregate costs associated with such performance losses is tricky; one would need to separate the direct effects of the substance from personality traits and other factors that are correlated with the decision to use in these settings and that might themselves be causes of poor performance. But given the sheer number of aggregate hours of intoxication (surely in the billions nationally), the costs could be considerable. Next, we take up two closely related questions—marijuana’s effects on accident risks and the possible chronic effects of heavy marijuana use.

Accident Risks

Another important health concern associated with marijuana consumption is the risk of acci- dents (e.g., falls, motor-vehicle accidents, and workplace accidents). There is clear evidence from strictly controlled laboratory trials that marijuana use reduces psychomotor performance in ways that increase overall risk of accidents and, in particular, impairs driving (Ramaekers, Berghaus, et al., 2004; Ramaekers, Moeller et al., 2006). Of course, not unlike with alcohol, the degree of impairment is a function of the dose, as well as individual-level factors, includ- ing age, body mass, and length of time using the drug (Degenhardt, Hall, and Lynskey, 2001; Ramaekers, Berghaus, et al., 2004). Although early evidence from simulator and epidemiologi- cal studies was far less conclusive (see Ramaekers, Berghaus, et al., 2004; Blows et al., 2005), Room et al. (2010) argued that the more-recent better-controlled epidemiological studies do, in fact, provide credible evidence that marijuana users who drive while intoxicated are at greater risk of motor-vehicle crashes. Meta-analyses conducted since Room et al.’s evaluation, account- ing for differences in study design and use of case controls, conclude that recent marijuana use (indicated by THC in blood or self-reported use near the time of the accident) more than doubles the risk of a car crash (Asbridge, Hayden, and Cartwright, 2012; Li et al., 2012).

There is understandably a strong desire to have quantitative metrics that would allow one to conclude that some particular concentration of THC or its metabolites produced an increase in risk equivalent to that of some familiar measure of alcohol intoxication. For example, after a review of the evidence on the impairment of driving-related skills by alcohol or cannabis, one international group of experts concluded that a THC concentration of 7–10 nanograms (ng) per milliliter in serum is sufficient to produce impairment equivalent to 0.05-percent blood alcohol content (Grotenhermen et al., 2007). Some argue that the pharmacokinetics of THC suggest that any serum concentration of THC could be indicative of intoxication sufficient to impair driving, because THC concentrations are measurable in blood only within the first two hours of smoking marijuana while the psychomotor effects can last for eight hours or more

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

34 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

(Asbridge, Hayden, and Cartwright, 2012; Neavyn et al., 2014). Thus, considerable debate con- tinues about the ideal (from a policy perspective) blood or serum levels to indicate marijuana intoxication while driving. We view this quest as being as of yet unfulfilled, though innova- tions are in development in various jurisdictions in the United States and abroad, including defined per se levels for impaired driving and saliva testing.

This measurement issue also has serious implications for assessing the public-health and safety consequences of marijuana legalization. If legalization increases marijuana use, hold- ing everything else constant, we would expect to see an increase in the share of people whom police test for DUI who test positive for THC; however, that does not necessarily mean that legalization will lead to a net reduction in traffic safety. Much will depend on how legalization influences the use of other substances, especially alcohol (further discussed in “Substitution and Complementarity with Other Substances”). Thus, those trying to evaluate how legaliza- tion influences traffic safety should focus on the overall accident or fatality rate, not just the number of cases involving marijuana or other substances.

Chronic Health Risks Dependence

The 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) (National Institute on Alcohol Abuse and Alcoholism, 2002; see also Compton, Thomas, et al., 2007, Table 1) found that about 2.6 percent of adults would meet Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) (American Psychiatric Association, 1994) criteria for cannabis dependence in their lifetimes. About one in ten lifetime users will, at some point, meet the dependence criteria: one in six for those who started using in adolescence and about half of all daily users (Hall, 2014). The probability of remission from dependence is 97 percent—66 percent within a decade and half within six years (Lopez-Quintero et al., 2011)—but the impact on the user’s life can be significant. Drug-dependence diagnoses are made using a checklist of criteria that allows people with different configurations of symptoms to receive the same diagnosis, and those symptoms might not match popular stereotypes of what an addict is like.

The diagnostic criteria for the marijuana-dependence diagnosis have been controversial (see MacCoun, 2013b), in part because some of the symptoms are arguably due to drug laws and social context in addition to any purely psychopharmacological properties of the drug. But there is fairly compelling evidence that heavy marijuana use can produce many of the hall- marks we associate with addiction. For example, evidence for both marijuana tolerance and a marijuana-withdrawal syndrome continues to accumulate and is clearer now than in the late 20th century (Hall, 2014). Perhaps the most-important evidence that marijuana dependence is a serious problem is that nearly one in six past-year users report that they are trying to quit using the drug (Compton, Dawson, et al., 2013). About half of marijuana treatment admis- sions in the United States are due to criminal justice system referrals, but the other half are not, and marijuana accounts for a large and growing treatment population in the Netherlands, where criminal justice referrals are much less common (MacCoun, 2011b). Rising treatment rates in the Netherlands coincide with increases in marijuana potency (MacCoun, 2011b; Hall, 2014), an issue to which we return in Chapter Six.

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 35

Respiratory Health, Cardiovascular Health, and Cancer

There are plausible reasons that marijuana smoking might increase risk for respiratory diseases. For example, marijuana smoke contains many of the same carcinogens as tobacco smoke (Moir et al., 2008). But the epidemiological evidence has been inconsistent and difficult to interpret because so many users also smoke tobacco (Hall, 2014). The most-comprehensive research to date suggests that chronic marijuana smoking is significantly less risky than tobacco smoking. Marijuana smoking probably creates greater pulmonary harm at high doses, but heavy users are less common in epidemiological studies, making inferences more difficult. Marijuana’s chronic effects on cardiovascular functioning are similarly inconsistent across studies, but, as noted earlier, there is some evidence that serious acute effects are possible. Hall (2014) reviewed several lines of suggestive evidence for a possible association of chronic marijuana smoking and male testicular and prostate cancers. Although more research is needed, the available evidence suggests that long-term marijuana smoking is less risky than long-term tobacco smoking, in part because a typical marijuana smoker will ingest far less than a typical tobacco smoker and is far less likely to use on a daily basis.4

Brain Development

Volkow et al. (2014) reviewed the emerging body of evidence on possible adverse effects of marijuana on brain development. Although the studies on humans are strictly correlational, controlled experiments on animals suggest that a causal role for marijuana in these associations is plausible. But because several different factors influence brain development, the question remains: If marijuana changes human brain development, does this lead to long-term negative consequences?

A recent neurological study comparing recreational users and nonusers found signifi- cant differences in gray-matter density in several brain regions involved in reward processing (Gilman et al., 2014), and these varied with frequency of use. As the authors recognize, these associations might not be caused by marijuana use and might even cause marijuana use, and their user and nonuser samples were matched on only a very limited set of variables. Still, the results are very troubling, and longitudinal brain-imaging studies might resolve some of the uncertainty.

Long-Term Cognitive Impairment

Scientists and activists alike vigorously debate the claim that marijuana use produces cogni- tive impairment, much like earlier arguments about an amotivational syndrome. Even if one is convinced about the acute effects, establishing that there are chronic, cumulative effects— possibly even irreversible effects—is much more challenging. Science is working hard to try to answer this question but cannot definitively answer it at this time. Even if any impairment is limited to the period of heaviest use (usually late adolescence), however, there might be long- term consequences due to processes that are social or developmental rather than neurological. For example, even a few years of poor academic performance during high school can have cas- cading effects on college and career prospects.

Meta-analyses of the literature suggest that associations between marijuana use and cog- nitive functioning are fairly weak and somewhat inconsistent (Grant, Gonzalez, et al., 2003;

4 We remind readers that (1) smoking is only one way to consume marijuana and (2) marijuana is also used to address some physical ailments (further discussed in “Consequences of Marijuana Consumption on Health”).

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

36 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

Schreiner and Dunn, 2012), and they appear to be limited to a subset of very heavy users who began using very early. But the tremendous heterogeneity of types and measures of cognitive functioning, study populations, and available control variables plague these meta-analyses. Further, much of the literature was developed when average age at first use was higher and average potency was lower.

Perhaps the most rigorous and ambitious study of chronic effects is the 38-year Dunedin Multidisciplinary Health and Development Study analysis of 1,038 New Zealand residents born in 1972 and 1973 (Meier et al., 2012). We focus on this study more than some others in this chapter given its prominence in current legalization debates and because it nicely illus- trates the complexity involved in drawing causal conclusions from nonexperimental data.

Meier et al. (2012) reported significant declines in neuropsychological functioning and intelligence quotient (IQ ) among persistent daily or near-daily users who began using in ado- lescence. The study has many important methodological strengths, including its longitudinal design and sophisticated statistical methods. For example, the study includes IQ measurements from before any cannabis use had occurred, allowing the researchers to control for the possibil- ity that lower IQ is a cause of the decision to use cannabis. Still, the authors could not rule out the role of unobserved confounding variables (especially time-varying ones) that might influ- ence both marijuana use and cognitive functioning. For example, Rogeberg (2013) presented a simulation study showing that the Dunedin results are consistent with a model in which differences are due to time-varying socioeconomic confounders rather than any direct effects of marijuana. Daly (2013) presented a hierarchical regression analysis of another longitudinal data set to show that the Dunedin results are potentially attributable to the lack of controls for stable personality traits that correlate with both marijuana use and cognitive functioning.

Meier and colleagues (Moffitt et al., 2013) responded to these criticisms. In response to Rogeberg (2013), they directly showed that socioeconomic status did not account for their results, while conceding Rogeberg’s broader point that, because of unobserved time-varying factors, their observational design could not conclusively establish causation or rule out spuri- ous associations. In response to Daly, they show that controlling for a measure of self-control did not significantly alter their findings. But although self-control is similar to conscientious- ness, one of the five main personality factors that Daly did mention, it is not the same, and, at any rate, Daly’s demonstration involved openness to experience—a completely different big- five factor.5

We hope that readers now understand why findings like this are not as straightforward as often discussed, not because of any failings by the researchers but because of the inherent ambiguity that accompanies nonexperimental findings on complex human phenomena involv- ing many potential causal pathways. It is premature to argue that long-term cognitive impair- ment has been clearly established, but just as premature to argue that the risks are nonexistent.

The National Institute on Drug Abuse (NIDA) is planning a decade-long longitudinal study of 10,000 teens (Reardon, 2014), which will feature periodic brain imaging, but even that approach cannot solve the problem of unobserved confounders. This will complement the Dunedin design and surely enhance our understanding of the neuropsychology of marijuana users. But the only really decisive way to rule out confounding effects is by using a controlled experiment, which is ethically precluded. Thus, we believe that the balance of the evidence is

5 For a brief summary of the big five personality factors, see Barrick and Mount (1991).

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 37

consistent with the proposition that persistent regular cannabis use is associated with small but meaningful long-term impairments in cognitive functioning, but the possibility remains that the associations are partly or wholly spurious.

Mental Illness

Many of the potential health risks of marijuana have been studied and debated for more than four decades. But perhaps the most serious concern to have emerged in recent years involves a growing body of evidence linking marijuana use to psychotic symptoms and possibly even schizophrenia. There are several interrelated claims about this link, and some are better estab- lished than others.

First, is there a statistical association between marijuana and experiencing psychotic symp- toms? Yes, and it has been replicated many times in many different populations, using many different ways of operationalizing both drug use and mental illness (see Donoghue et al., 2014; Hall, 2006, 2014; Large et al., 2011; McLaren et al., 2010; Proal et al., 2014; Radhakrishnan, Wilkinson, and D’Souza, 2014).

Second, in numerous longitudinal studies, the temporal pattern of the association is usu- ally more consistent with the marijuana-use-leads-to-mental-illness model than with a self- medication (i.e., mental illness leads to marijuana use) account (see Hall, 2006, 2014).

Third, experimental studies using double-blind methods establish that controlled exposure to marijuana increases the likelihood of measures of negative affect, dissociation, and disordered thought during intoxication that are common symptoms of psychosis (see Radhakrishnan, Wilkinson, and D’Souza, 2014). We have to interpret these findings with some care because the fact that these are symptoms observed in schizophrenia does not neces- sarily imply that their occurrence in the experiments is evidence of psychosis. The ability to experience some of these psychotomimetic symptoms in a voluntary and time-limited fashion is, for some people, part of the attraction of the recreational use of the drug.

Fourth, despite considerable effort, researchers have been unable to rule out the possibility that the association between marijuana use and psychotic symptoms is due to some common risk factor. In some studies, controlling for plausible common risk factors eliminates the asso- ciation between marijuana and psychosis, and, in other studies, these control variables weaken but do not eliminate the association (see Hall, 2006, 2014; McLaren et al., 2010; Proal et al., 2014; van Ours and Williams, 2012). Lewis, Heron, and Zammit (2013) observed that “[t]hose investigators who adjusted more comprehensively for confounders also found that the associa- tion was reduced somewhat more,” raising the possibility that additional control variables, if assessed, might further attenuate the association.

Fifth, if cannabis use does have a causal impact on psychosis, it appears to be highly con- tingent on the timing and intensity of cannabis use and possibly on a genetic propensity or other existing personal and environmental risk factors (Hall, 2006, 2014; Proal et al., 2014). Hall (2006) noted that these contingencies might explain

first, why the risk of psychosis in cannabis users is only increased 2–3 times; second, why there have not been large increases in the incidence of psychoses in line with the rise in rates of cannabis use in young adults in recent decades; and third, why the age of onset of schizophreniform disorders might be earlier in cannabis users. (p. 194)

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

38 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

Finally, the best available evidence is that any population effects of marijuana on the psychoses are likely to be small (Hall, 2014). Hickman and colleagues (2009) estimated that

if cannabis is related causally then the risk of schizophrenia in 1997–99 for men aged 20–24 was approximately 1 in 1500 for heavy cannabis users and 1 in 2400 for light users. For women aged 20–24 the risk of schizophrenia was 1 in 4000 for heavy cannabis users and 1 in 6600 for light cannabis users. (p. 1858)

This implies that thousands of users would have to be prevented from use for a year to prevent one case of schizophrenia. Still, this does not mean that the risk is inconsequential. That legal- ization could increase the number of frequent marijuana users in a jurisdiction by many thou- sands is not at all implausible. And even if marijuana simply accelerated the onset of schizo- phrenia, rather than causing it, an increase in person-years of psychosis can be enormously costly for patients, families, and taxpayers.

Medical Benefits

Different cultures have used marijuana for therapeutic purposes for thousands of years for various ailments (O’Shaughnessy, 1843; Grinspoon and Bakalar, 1993; Robson, 2014). As described in Chapter Two, Vermont allows physicians to recommend marijuana for debilitat- ing medical conditions, where debilitating includes

Cancer, acquired immune deficiency syndrome (AIDS), positive status for human immu- nodeficiency virus (HIV), multiple sclerosis (MS), or the treatment of these conditions if the disease or the treatment results in severe, persistent, and intractable symptoms; or a dis- ease, medical condition, or its treatment that is chronic, debilitating and produces severe, persistent, and one or more of the following intractable symptoms: cachexia or wasting syndrome, severe pain or nausea or seizures. (VCIC, undated [b])

There is a growing scientific literature documenting the medical benefits of THC and increasing interest in assessing the efficacy of CBD and other constituents of the cannabis plant (e.g., terpenes). In 1999, the Institute of Medicine (IOM) assessed the scientific literature on medical marijuana and concluded,

Scientific data indicate the potential therapeutic value of cannabinoid drugs, primarily THC, for pain relief, control of nausea and vomiting, and appetite stimulation; smoked marijuana, however, is a crude THC delivery system that also delivers harmful substances. (Joy, Watson, and Benson, 1999, p. 4)

The IOM noted that five other reviews of the medical-marijuana literature were published between 1996 and 1999 and all but one concluded that “marijuana can be moderately effective in treating a variety of symptoms”; however, they all raised similar concerns about the “uncer- tain composition of plant material” and called for more research (Joy, Watson, and Benson, 1999, p. 180).

Since 1999, dozens of human clinical trials involving cannabinoids have been con ducted.6 Some involved cannabinoid extracts produced by pharmaceutical companies, while others

6 Hundreds—if not thousands—of cannabis studies involving lab experiments with animals and petri dishes have also been published since 1999. The lab studies examining the antitumor effects of some cannabinoids are especially promising

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 39

have involved raw cannabis material that is either smoked (combusted) or vaporized. Sativex® is a plant-derived oral spray that is roughly 50 percent THC and 50 percent CBD. It is used to treat spasticity from MS in 11 countries, and some studies support the producer’s claim that

Exploratory trials in several hundred [MS] patients consistently showed significant advan- tages for Sativex® over placebo in the relief of spasticity, chronic pain, muscle spasms, bladder-related problems and sleep quality which appeared to be maintained over long- term treatment, and the medicine was generally well tolerated. (Robson, 2014, p. 26)

Randomized, double-blind trials with non-MS populations suggest that Sativex could reduce unilateral peripheral neuropathic pain and allodynia (Nurmikko et al., 2007) and inflamma- tory pain (Blake et al., 2006).7

Although medical-marijuana dispensaries sell tinctures, extracts, and nonpsychoactive topical preparations with varying levels of THC, CBD, and other cannabinoids, these particu- lar products have not been the subject of clinical trials. However, the anecdotal evidence about high-CBD extracts reducing seizures—especially among children—has been convincing and likely explains why more than ten states enacted CBD laws in 2014 (Ingold, 2014c).8 Clinical trials of a plant-derived CBD spray (Epidiolex) to treat children suffering from severe seizures started in the United States in early 2014 (Bunim, 2014).

The systematic reviews of the Cochrane Collaboration provide what are arguably the most- rigorous assessments of the evidence for medical applications of marijuana and the chemicals it contains. Cochrane reviews examining the medical benefits of cannabis for epilepsy (Gloss and Vickrey, 2014), HIV or AIDS symptoms (Lutge, Gray, and Siegfried, 2013), dementia (Krishnan, Cairns, and Howard, 2009), and schizophrenia (McLoughlin et al., 2014) found that few available trials meet their methodological standards for inclusion and that the avail- able evidence is insufficient to establish medical benefits. No doubt, this paucity of strong stud- ies partly reflects the enormous obstacles of doing marijuana research. Given the tremendous proliferation of medical-marijuana systems in the United States, more and better clinical trials are clearly needed.

(National Cancer Institute, 2014). Although clinical trials of antitumor effects have not been conducted on humans, the National Cancer Institute notes that “a single, small study of intratumoral injection of delta-9-THC in patients with recur- rent glioblastoma multiforme reported potential antitumoral activity” (see Guzmán et al., 2006, and Velasco, Sánchez, and Guzmán, 2012). 7 As this document was going to press, a clinical trial of Sativex for cancer pain was reported to yield “no statistically sig- nificant difference between subjects using [Sativex] and those given a placebo” (Hirschler, 2015). Results from two other phase III trials are expected in 2015. (A phase III trial is one in which the “drug or treatment is given to large groups to confirm its effectiveness, monitor side effects, compare it to commonly used treatments, and collect information that will allow the drug or treatment to be used safely” [U.S. National Library of Medicine, 2008]). 8 One might refer to this as the Gupta effect because the CNN documentary Weed: Sanjay Gupta Reports that first aired in August 2013 spurred much of the discussion about CBD. Among others, it featured a three-year-old girl with severe myoclonic epilepsy of infancy (also known as Dravet syndrome) who suffered from nearly 300 grand mal seizures per week. After twice-daily use of a high-CBD, low-THC extract produced by farmers in Colorado, she reportedly experienced only two to three seizures per month, typically in her sleep. For more on this story, see Young (2013).

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

40 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

Consequences of Marijuana Consumption on Other Outcomes

Like being under the influence of alcohol, being under the influence of marijuana at school or work can impede development for some people by making it harder to learn and concentrate. And if intoxication leads users to skip school or work, or, in the case of school, not complete their homework, this could have a negative effect on future performance. Because THC is fat- soluble and can stay in the system long after a use episode, there is concern about the residual effects of marijuana on cognitive functioning.

The challenge for researchers is trying to figure out whether regular or heavy marijuana use has any lasting effects on education or employment outcomes, or what economists often refer to as human capital development. There are strong correlations between frequency of marijuana use and many education and employment outcomes, but here we run into the same selection issues that we have mentioned throughout this chapter. Here we highlight three out- comes that have received a relatively large amount of attention in the marijuana research litera- ture: dropping out of high school, labor outcomes, and criminal activity.

Deciding to Drop Out of High School

The correlation between marijuana use and dropping out of high school is positive, but it is unclear whether the relationship can be attributable to cognitive effects, peer effects, both, nei- ther, or some other factors. Largely based on a series of longitudinal studies, the Hall (2014) review of the research concludes that “[r]egular adolescent cannabis users have lower educa- tional attainment than non-using peers”; however, his review raises important questions about how much of the relationship is causal. For example, the review highlights a recent Australian twin study that attributes the association to genetic and environmental risk factors instead of cannabis (Verweij et al., 2013) and suggests that this is consistent with two other U.S. twin studies. The review also notes that “the adverse effects of cannabis use on educational outcomes may be amplified by school policies that exclude students who are caught using cannabis from secondary school.”

A recent article that combined samples from three longitudinal studies in Australia and New Zealand found that those who used marijuana daily before age 17 had significantly lower odds of completing high school and earning a college degree (Silins et al., 2014). Although the study found a statistically significant dose–response relationship, this is neither necessary nor sufficient for causation. The authors argue that the relationship between marijuana use and completing high school “probably does not arise from a reverse causal association,” but they note that “this possibility remains plausible” (p. 291).

Another longitudinal study based in the United States by McCaffrey et al. (2010) found that the apparent effect of marijuana use on dropping out of high school disappeared once they controlled for cigarette use. This led the authors to conclude that the effect of marijuana use on decisions to drop out was not attributable to a reduction in cognition (because cigarette smok- ing does not seriously impair cognition). Additional analyses revealed that parental and peer influences could explain the marijuana–dropout relationship.

Labor Outcomes

We previously noted the acute effects that marijuana use can have on cognitive skills and psychomotor performance. With respect to the research about the effect of illegal drug use on employment and worker productivity, the results are all over the place. Kilmer and Pacula

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 41

(2010) reviewed the econometric literature of the topic and concluded that the evidence is mixed, attributing the variation to three factors:

• These studies often assess earning labor-market outcomes for populations of varying ages. • There is “inconsistent treatment of indirect mechanisms through which substance abuse

could impact earnings, for example through educational attainment, health, fertility and occupational choice.”

• Studies often use different definitions of substance use (e.g., past year, past month, daily).

Criminal Activity

There is a long-established positive correlation between marijuana and crime (Dembo et al., 1987; Dawkins, 1997; Baker, 1998). People who commit criminal acts are more likely than those who do not to use marijuana (Taylor and Bennett, 1999; Makkai and Fitzgerald, 2000), and people who use more marijuana commit more criminal acts than those who use less mari- juana (McRostie, Castle, and Marshall, 2001). Still, most experts believe that the correlation could be due to common risk-seeking or delinquency factors that generate a spurious associa- tion between marijuana use and crime.

Although there have been numerous studies that have attempted to examine the contem- poraneous relationship between marijuana use and violent or property crime, rigorous exami- nations using techniques that enable causal interpretation of results are rare (Pacula, Lundberg, et al., 2013). Instrumental-variable approaches and longitudinal analyses of prospective cohorts provide the clearest evidence (Arseneault et al., 2000; Markowitz, 2005). However, even when a statistically significant association remains in these studies, the studies have never been able to demonstrate that the people being studied were actually under the influence of marijuana at the time of the offense (suggesting that intoxication was a direct cause). Thus, they never completely eliminate the possibility that a third unobserved factor is causing the statistical association.

For example, a paper by Green et al. (2010) identified a link between marijuana use and property crime by applying propensity-score matching to a longitudinal community cohort of blacks in Chicago. Heavy adolescent marijuana users (defined as those who had used 20 or more times in their lifetimes) were matched with nonheavy marijuana users based on vari- ables that would confound the relationship between drugs and crime. After matching, the two groups were similar in important observable characteristics (measured from the survey), including personality traits (such as aggression), family situation (such as mother’s use of dis- cipline), and elementary school adaption and achievement (such as teacher’s rating of conduct problems). Using these matched groups, heavy marijuana use in adolescence was associated with crime generally, but, when the authors decomposed their findings by different types of crimes, they found that the association held for drug-related crime and property crime, not violent crime. The association with drug-related crime is not surprising but is more a reflection of marijuana being illegal than of any psychopharmacological effects on criminality. The odds ratio associated with property crime based on heavy marijuana use was 1.5:1.

Although the Green et al. study is clearly suggestive, it cannot rule out that marijuana use is simply a correlate of a long-term trajectory of criminality because use is not measured in close temporal proximity to engagement in crime (which is needed to conclude that the substance itself has a criminogenic effect). Indeed, in supplemental analyses that attempted to explore the extent to which marijuana use was the true cause, the authors found that dropping out of high

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

42 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

school was a key mediating pathway through which heavy adolescent marijuana use negatively influenced crime over the life course. Other social scientists have also found evidence suggest- ing that marijuana use is simply a mediating influence of a delinquent life course (Pedersen and Skardhamar, 2010; J. Ford, 2005; Windle and Wiesner, 2004; Fergusson, Horwood, and Swain-Campbell, 2002).

The crime rate in Colorado has been a topic of discussion postlegalization, with advocates on all sides of the debate arguing that the evidence does not support their opponents’ positions. For example, comparing Denver crime data from the first six months of 2013 and those from the first six months of 2014, The Huffington Post noted that rates of property crime, as well as homicide, sexual assault, and robbery, were all down; only aggravated assault was up (by 2.2 percent). The author noted,

Correlation does not imply causation, regardless of which way the crime data move, and after just six months, it may be too early to identify any strong social trends. But evidence of a crime wave simply has not materialized—despite numerous dire warnings prior to legalization. (Ferner, 2014)

Similarly, a Washington Post blog post argued, “So perhaps we should hold off on the panicky stories about pot-fueled crime waves for a bit—especially since the early data show that crime has actually dropped” (Balko, 2014).

On the other side, opponents of legalization argued that evidence of a crime drop in Col- orado comes from those who wanted to “cook up numbers they wished to see” (Sabet, 2014). They cite sources suggesting that total crime in Denver actually went up 7 percent in the first six months of 2014 from rates in 2013 (see, e.g., Thurstone, 2014). The story depends on what offenses are being considered, what data sources are used, and whether crime reports are being combined with arrest statistics. After commissioning a similar study of crime in Denver, the head of the National Association of Drug Court Professionals noted that “we are promot- ing the position that the question remains open, and at best we can say there is contradictory evidence when trying to draw conclusions about the effect marijuana legalization has had on crime” (as quoted in Thurstone, 2014).

To make any sense of crime statistics in Colorado or Washington State, a more com- prehensive multivariate analysis is needed, including crime trend data from nonlegalization states. The implementation of legalization in Alaska, Oregon, and Washington, D.C., will pro- vide additional evidence. Unfortunately, a convincing study will require a longer postchange period, so we might not know more for several years.

Consequences of Marijuana Prohibition

Enforcement of laws against marijuana use and distribution imposes costs. It not only requires public expenditures as discussed in Chapter Two; it can also reduce the well-being of people and neighborhoods in a variety of ways. Having an arrest record—let alone being incarcerated, even briefly in jail—can disrupt legitimate careers and impair future job prospects (Raphael, 2014). Fines can also be significant: For someone who works close to the minimum wage in Vermont, paying $200 for possessing less than 1 oz. could consume the take-home pay from the better part of a full week of work.

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 43

Collateral consequences associated with drug convictions can negatively affect users and their families. For example, depending on the state of residence and level of offense, a drug conviction can entail loss of food assistance, access to public housing, and federal financial aid for postsecondary education (Federal Statutes Imposing Collateral Consequences upon Con- viction, undated; Mauer and McCalmont, 2013). On the other hand, according to a report funded by the Marijuana Policy Project, Vermont is already one of the ten states with the least severe collateral sanctions for marijuana offenses (Boire, undated).

Marijuana enforcement in the United States disproportionately affects blacks (ACLU, 2013; Glaser, 2015). Especially in the context of stop and frisk, these police encounters are correlated with self-reported trauma, anxiety, and other mental health problems (Geller et al., 2014). In addition, enforcing laws that have limited popular support can erode the legitimacy of authorities and might reduce compliance with the law (Nadler, 2005; Tyler, 1990) and coop- eration with the police (Tyler and Fagan, 2008).

As mentioned in Chapter Two, legalizing marijuana will not eliminate the costs associ- ated with marijuana prohibition. If a minimum legal purchase age is imposed, there will be penalties for underage possession and those who supply minors. There may also be penalties for public consumption, which could especially affect out-of-state visitors who might not be allowed to consume in their hotel rooms.

Substitution for or Complementarity with Other Substances

The previous sections identified various marijuana-related outcomes and reviewed literature on their magnitudes. However, even if such a catalog of direct effects were both exact and exhaustive, it could still be a terribly incomplete and misleading basis for a benefit–cost analy- sis because legalization’s indirect effects, via changes in the use of other substances, could well outweigh the importance of the marijuana-related outcomes themselves (at least as they are currently understood today). Unfortunately, the uncertainty about these indirect effects is, in many instances, even greater than the uncertainty about legalization’s effect on marijuana- related outcomes. This inconvenient ambiguity places severe limits on the confidence anyone should have in prior predictions about whether legalization will be a net win or a net loss for society.

This story of indirect effects mediated through other substances is easiest to tell, and most often raised, in terms of alcohol. Suppose that legalization led to a doubling of marijuana con- sumption of all sorts, including not only a doubling of controlled recreational use but also a doubling of compulsive abuse and dependence. One might well view this as a net bad because of all of the marijuana-related harms discussed above.

However, the total social cost associated with alcohol abuse is very much larger than all costs and outcomes related directly to marijuana use. So if the doubling of marijuana use came about because all these new marijuana users switched from drinking alcohol, that could be a net win from a public-health perspective, particularly if these people would otherwise have been binge drinking (Caulkins, Hawken, Kilmer, and Kleiman, 2012). Indeed, Caulkins,

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

44 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

Hawken, Kilmer, and Kleiman (2012) found that even a 10-percent reduction in alcohol abuse accompanying the doubling in marijuana use could be a net win for society.9

Alas, that story of increased marijuana use being a substitute for alcohol use is not the only possibility. It is also possible for two consumer goods to be complements, such that, when market conditions change in ways that promote greater use (and abuse) of one, that might lead to greater—not lesser—use (and abuse) of the other.

If marijuana and alcohol proved to be complements, and legalization led to any sizable increase in alcohol use and abuse, then legalization would be a net loss. Even if all marijuana- related costs magically disappeared, that could not offset the harm caused by a 10-percent increase in alcohol-related problems.

Not surprisingly, this prospect of marijuana legalization affecting alcohol-related prob- lems becomes something of a Rorschach test for an observer’s position concerning legalization. When interaction with alcohol use is raised, those who favor legalizing marijuana are often certain that any increase in marijuana use would replace and so reduce alcohol use; after all, the two substances are alternative ways of achieving intoxication. Conversely, opponents of legalization are often just as certain that greater marijuana use will bring greater alcohol use; after all, polydrug abuse is the norm (Mohler-Kuo, Lee, and Wechsler, 2003).

The descriptive statistics concerning overlap in use are clear. Marijuana users are much more likely than are nonusers to drink and to abuse alcohol. For example, current marijuana users are five times as likely as nonusers to meet DSM-IV criteria for alcohol abuse or depen- dence (26 percent versus 5 percent); that is, one in four current marijuana users is a problem drinker (calculated using 2012 NSDUH data using the SAMHSA online tool).

Indeed, simultaneous use is common. The national household survey asks people what, if any, other substances they used the last time they drank alcohol.10 Among the 15.4 million people who used both alcohol and marijuana at some time in the past 30 days, 54 percent reported using marijuana along with alcohol the last time they drank, a proportion that rises to 83 percent among daily or near-daily marijuana users.

As was discussed earlier in this chapter, correlation need not, however, imply causality. Alcohol use might cause marijuana use, not the other way round. Or there could be third variables—such as coming from a broken home or having a risk-taking personality—that lead people to consume more intoxicants generally, including both alcohol and marijuana. As we discuss in Appendix A, some scientific literature makes a serious attempt to tease out these complicated overlapping causal pathways, but, unfortunately, that literature is inconclusive.

Nor is alcohol the only substance with which there could be important interactions. Consider the one substance that could cause even greater social harms than alcohol—namely, tobacco. The overlap between marijuana and tobacco use is at least as strong as the overlap between marijuana and alcohol use. Past-month marijuana users are three times as likely as

9 Doubling of marijuana use would not lead to even a halving of all drinkers, because there are nearly ten times more drinkers than people who use marijuana. According to the 2013 NSDUH, there were 136.9 million past-month alcohol users and only 19.8 million past-month marijuana users (unadjusted for underreporting). Indeed, there were three times more binge drinkers in 2013 (60.1 million) than there were marijuana users. So a doubling of marijuana users—even if all the new users had been binge drinkers and became teetotalers—would reduce the social cost of binge drinking by only about one-third. 10 The question’s wording: “Think again about this last time you drank any alcoholic beverages, when you had [number] [drink/drinks]. Did you also use [drug] while you were drinking or within a couple of hours of drinking?” and “What other drug or drugs did you use while you were drinking or within a couple of hours of drinking?”

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

Consequences of Marijuana Use 45

nonusers to smoke cigarettes (59 percent versus 19 percent), a ratio that rises to 6:1 for those under the age of 21 (53 percent to 9 percent). Ninety-five percent of marijuana users report using tobacco at some point in their lives.11 And again, concurrent use is common. More than half of marijuana users under the age of 35 report smoking a blunt (a hollowed-out cigar filled with marijuana) within the past 30 days (more than three-quarters among blacks). In Europe, mixing tobacco with marijuana or hashish in the same cigarette or joint has long been the norm (Leggett, 2006); it is not hard to imagine tobacco companies wanting to promote that practice in the United States after legalization.

Suppose that legalizing marijuana caused even a 1-percent increase in tobacco smok- ing. Because tobacco kills well over 400,000 people in the United States every year, then, in that hypothetical, legalizing marijuana might—in the long run—cause 4,000 additional pre- mature deaths per year, an outcome that could outweigh any plausible benefits of marijuana legalization.

On the other hand, legalizing marijuana could shift marijuana consumption away from combusting plant material (as in smoking a standard joint) and toward heating and vaporizing extracts in vaping pens. The latter has become increasingly popular in places that have eased their marijuana prohibitions and is akin to using e-cigarettes to consume nicotine. Suppose this spilled over to make vaporization the norm among nicotine users (i.e., converted cigarette smokers into e-cigarette users). Some argue that that could be a public-health win if the health harm per person-year of nicotine dependence turns out to be lower when nicotine is delivered via e-cigarettes than with smoking traditional cigarettes (Cahn and Siegel, 2011).

Or it could be a public-health disaster; some in the tobacco-control community worry that e-cigarettes will prove to be not so much a substitute for traditional cigarettes as a path- way or stepping stone through which nonsmokers could get pulled into smoking (Grana, Benowitz, and Glantz, 2014). E-cigarette cartridges (like marijuana vaping-pen cartridges) come in all sorts of child-friendly fruit flavors, and, even if e-cigarettes create no risk of lung cancer directly, they do unambiguously deliver nicotine, which is highly dependence-inducing. So if the tobacco-control community’s fears prove true, then a legalization-induced bump in vaping of all kinds, including e-cigarettes, might end up increasing, rather than reducing, the amount of tobacco smoking and all its many attendant harms.

At this point, even though the literature reviewed in Appendix A suggests that marijuana and tobacco are complements today, no empirical evidence can truly assess the likelihood of these very different future scenarios concerning the impact marijuana legalization might have on tobacco smoking.

Legalization could also affect the use and abuse of other illegal drugs. Long ago, there was great concern that trying marijuana could be a gateway that caused users to go seek stron- ger and stronger highs. Those fears arose from the combination of conditional probabilities (children who use marijuana are much more likely to progress to harder drugs) and sequential order (marijuana usually predates use of harder drugs). But those facts together do not imply causality. Various observers (e.g., Morral, McCaffrey, and Paddock, 2002) have shown that the same patterns could emerge if third variables (e.g., broken homes, risk-seeking personalities)

11 Oddly, most of the remaining 5 percent report consuming marijuana in the form of blunts (cigars stuffed with mari- juana), which they apparently do not view as using tobacco products even though the shell of a cigar is made of tobacco leaves.

Co py ri gh t © 2 01 5. R AN

co py ri gh t la w.

46 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

cause use of both marijuana and hard drugs, and marijuana gets used first simply because it becomes available to children first.

But showing that the data do not imply that a causal version of the gateway hypothesis holds is not the same as showing that there is no causal effect. Third variables could account for some but not all of the correlation. Furthermore, the connection need not be purely bio- chemical. For example, use of marijuana could lead teens to spend more time with others who use marijuana (birds of feather flock together or, more formally, homophily), and those mari- juana-using peers might have more-positive attitudes toward use of other drugs or know how to obtain those other drugs. Likewise, marijuana use might lead the individual to self-identify and to be identified by others (labeling) as being the sort of person who uses drugs of all kinds. And so on. So there could be a causal path from greater marijuana use to use of hard drugs that is social or psychological, even if there is no biochemical link. Hence, although confidence in the old-fashioned version of the gateway hypothesis went beyond the empirical evidence, con- fidence in the irrelevance of the gateway hypothesis might be equally naïve.

Marijuana could also be both a complement and a substitute for other drugs but on dif- ferent time scales—e.g., they might be substitutes in the short run and complements in the long run. Or marijuana might also be a complement for some drugs but a substitute for others.

And even if marijuana were a complement to other illegal drugs under today’s circum- stances, that does not mean that legalizing marijuana would necessarily increase use of hard drugs. One of the primary motivations for the Netherlands’ de facto legalization of retail can- nabis sales was the desire to separate the soft- and hard-drug markets, so people could obtain cannabis without coming into contact with sellers of hard drugs, and there is some evidence of success in that regard (MacCoun, 2011b).

It is also worth noting that the vast majority of initiation into the use of any of these drugs, including marijuana but also other illegal drugs, alcohol, and tobacco, occurs before the age of 21, so the legal status of most potential initiates would not change. What would change is the supply, variety, price, and availability of marijuana products.

One goal of the either–or discussion in the preceding paragraphs is to shake readers from any strong prior convictions that they can just know that legalizing marijuana will increase or that it will decrease use or abuse of this or that other substance. It is easy to assemble deductive arguments in either direction, as advocates on both sides of the marijuana-legalization debate routinely do.

An honest analysis should view the issue as an empirical question to be resolved by look- ing at evidence from well-run studies. Hence, Appendix A briefly reviews empirical evidence from the scientific literature concerning whether and under what circumstances marijuana appears to be a substitute or a complement for use of other drugs.

Unfortunately, the punch line remains enormous uncertainty; the nature and size of the totality of legalization’s indirect effects mediated through changes in the use of all other substances—legal and illegal—remains a huge unknown.

One can be slightly more definitive when considering each substance or category of sub- stances one at a time. Table 3.1 captures the gist of this complicated literature in the simplest possible terms, distinguishing along just two dimensions: (1) size of the literature underpin- ning the estimates and (2) degree of consensus among those studies.

For alcohol and for the other illegal drugs, the existing literature is ambiguous, with some studies pointing in one direction and others pointing in the opposite direction. For opioids, there is consistency, but perhaps only because there are so few studies. The one substance for

co py ri gh t la w.

Consequences of Marijuana Use 47

which the literature provides clearer indication is the interaction with tobacco use, for which there is considerable evidence and it tilts strongly toward complementarity (i.e., greater mari- juana use would be expected to lead to greater tobacco use). Complementarity can arise with respect to initiation (marijuana initiation often predates tobacco initiation, creating the pos- sibility of a reverse gateway), concurrent use (some users think that nicotine enhances the marijuana high), or cessation (marijuana use predicts lower success rates when trying to quit tobacco use).

Note that, even if the literature assessing historical patterns were unanimous, though, none of it speaks directly to Vermont’s situation. Legalization is new; before 2012, no major modern jurisdiction had ever legalized commercial production of marijuana, and it is only in 2014 that stores have opened in the two early-adopting jurisdictions (Colorado and Washing- ton State). So all of the historical evidence comes from changes in the price, availability, or sanctions for marijuana under regimes that prohibited production and commercial distribu- tion of marijuana (or from changes in supply of the other drug on use of marijuana). Drawing inferences from that backward-looking literature about what would happen if Vermont legal- ized involves substantial extrapolation beyond the support of the data. So the literature can, at best, be suggestive; it is in no way definitive.

Concluding Thoughts

We began this chapter by asking two very basic questions: (1) What effect does marijuana use have on health and safety? and (2) What effects would a change in marijuana laws have on patterns of use and, hence, health and safety? By this time, the reader is keenly aware that what the literature has found about the effects of marijuana use in the past might not be the effects of marijuana use in the future because the findings to date have been based largely on observational data that reflect use of a substance containing largely unmeasured amounts of cannabinoids that can differentially influence marijuana’s impact on health and safety. We also do not know precisely what will happen to marijuana use with legalization or, specifically, the extent to which heavy or harmful use will rise, which is directly relevant for understanding the public-health and safety consequences. Nor do we know how the product might change (e.g., potency, mode of use) or how these changes might differentially influence relationships identi- fied here. Thus, it is difficult to say with any certainty whether the associations identified in the past will be maintained in the future.

Table 3.1 Summary of Literature on the Extent to Which Marijuana Is a Substitute for or Complement with Other Drugs

Substance Studies Agreement Among

Studies Finding

Alcohol Many No consensus Unknown

Tobacco Many High consensus Complementarity

Prescription opioids Few Consensus Substitution

Illegal drugs Few No consensus Unknown

co py ri gh t la w.

48 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

With those caveats in mind, the current status of the literature does identify some clear acute and chronic health effects, especially of frequent, high-dose marijuana use. Acute risks include accidents and impaired cognitive functioning while intoxicated, as well as anxiety, dys- phoria, and panic. Longer-term risks of persistent heavy use include dependence and bronchi- tis. There is some suggestive evidence of other serious risks for heavy marijuana users, including psychotic symptoms (which is different from being diagnosed with schizophrenia), cardiovas- cular disease, and testicular cancers. More research is needed before firm conclusions can be drawn. Although the literature showing a relationship between marijuana use and crime is extensive, there is little evidence that use itself increases criminal behavior, so one would not expect legalization to have important effects on nondrug crime, either favorably or unfavorably. Finally, the literature that persistently identifies a negative association between marijuana use and school attendance and achievement has not definitively determined whether the associa- tion is causal or not.

So, perhaps some of the greatest health and safety consequences of marijuana use reside in the implications of its use in lieu of or in addition to other intoxicating substances. There the clearest evidence pertains to interactions with tobacco, and it suggests that greater marijuana use would be expected to also lead to greater tobacco use.

It is important to emphasize that we have not attempted to offer a cost–benefit analysis. We have reviewed evidence for the health and safety consequences associated with marijuana but not its nonmedical or aesthetic benefits. We believe that such benefits are real and that they should matter, but they are far more difficult to quantify, and they have received far less research attention than the harms of marijuana use (see Caulkins, Hawken, Kilmer, and Klei- man, 2012, Chapter Six).

Indeed, if Vermont consumes 20 t of marijuana each year and each gram produces four hours of intoxication, this means that marijuana produces about 80 million hours of intoxica- tion in Vermont each year. If legalization increases intoxication by 50 percent, which is not an unreasonable projection under certain assumptions (see Chapter Seven), those 40 million hours could figure prominently in a benefit–cost analysis—but it is entirely unclear how they should be scored. As Caulkins, Hawken, Kilmer, and Kleiman (2012) noted, “some of those hours are intensely pleasurable and cause no harm to the user or anyone else; others contribute to dys- function and personal failure” (p. 181).

We emphasize that the relevant policy question is not whether marijuana’s current harms outweigh its benefits but whether and how legalization might change those harms and benefits and in which direction. At least at present, answering that question is more a matter of judg- ment than of calculation, and different readers will reach different conclusions.

co py ri gh t la w.

CHAPTER FOUR

Supply Architectures

Introduction

One of the key elements of a plan for implementing legalization is determining how marijuana will be supplied. Specifically, what kinds of organizations will be allowed to produce and distribute marijuana? Historically, the answer under prohibition has been “none.” Now some states that allow the use of marijuana under medical recommendation allow production and sale for that purpose, while others allow production only for personal use. In November 2012, Colorado and Washington State made a dramatic break with past practice by voting to allow a for-profit commercial marijuana industry that is licensed and regulated somewhat like the alcohol industry.

But there is a lot of policy space between traditional prohibition and such commercial legalization (see MacCoun, 2013a). Figure 4.1 presents 12 alternatives regarding who is allowed to supply marijuana, arrayed from most to least restrictive.1 Voters in Colorado and Washing- ton skipped over many intermediate options, such as barring for-profit corporations but allow- ing co-ops or nonprofits whose boards include public-health and child-welfare advocates.

Each of these supply strategies is really a broad category of options encompassing consid- erable scope for fine-tuning. And a bad implementation of a good strategy might underperform relative to a wise implementation of an inferior one. Therefore, picking a strategy is more the beginning of a discussion and design process than an answer to the question of what should be done. However, some consequences inevitably flow from selecting who is allowed to produce and supply marijuana that no amount of fine-tuning can overcome.

The chapter describes 12 approaches, breaking them down into three groups:

• the two options most commonly discussed in the United States – Retain prohibition but decrease sanctions. – Implement an alcohol-style commercial model

• eight options that find a middle ground between those commonly discussed – Allow adults to grow their own. – Allow distribution only within small co-ops or buyers’ clubs. – Permit locally controlled retail sales (the Dutch coffee-shop model). – Have the government operate the supply chain (government monopoly). – Have a public authority operate the supply chain.

1 There could be exceptions to this general ordering. For example, it is possible to conceive of a highly taxed and tightly regulated private market that was actually more restrictive than a state monopoly with low prices and aggressive marketing along the lines of many state lotteries.

co py ri gh t la w.

50 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

– Permit only nonprofi t organizations to sell. – Permit only for-benefi t companies to sell. – Have very few closely monitored for-profi t licensees.

• two extreme options – Increase sanctions. – Repeal the state’s prohibition without creating any new, product-specifi c regulations.

We then compare these strategies across a series of criteria likely to be important to policy- makers and voters, such as production costs, product quality assurance, government ability to restrain suppliers’ promotion of harmful use, cost of government control eff orts, ability to generate state revenue, and confl ict with the Controlled Substances Act (CSA) (Pub. L. 91-513, 1970, Title II).

Figure 4.1 Twelve Supply Alternatives to Status Quo Prohibition

co py ri gh t la w.

Supply Architectures 51

The Commonly Discussed Options

Th is section describes two commonly discussed supply options (also illustrated in Figure 4.2).

Retain Prohibition but Decrease Sanctions (Mend It; Don’t End It)

Many people support legalizing marijuana not so much because they want access to a broader range of higher-quality and lower-priced marijuana products as because they dislike marijuana prohibition (Galston and Dionne, 2013). Th ey are troubled by the large number of marijuana arrests and their racially disparate character (Nguyen and Reuter, 2012) or by how damaging a youthful indiscretion with marijuana can be to access to student loans, public housing, and jobs (Blumstein and Beck, 2005; University of California Los Angeles Integrated Substance Abuse Programs, 2007; Raphael and Stoll, 2009). See Chapter Th ree for additional discussion.

For those who dislike the excesses of prohibition, one option is simply to eliminate those excesses or, to borrow President Bill Clinton’s slogan for welfare reform, to “mend prohibition but not end it.” Vermont has already taken two substantial steps toward liberalizing its prohi- bition by decriminalizing possession of up to 1 oz., reducing the sanction to just a $200 fi ne (maximum) and $147 in surcharges for the fi rst off ense (eff ective July 2013), and providing access to medical marijuana for those with serious debilitating medical conditions (see Chapter Two).

Figure 4.2 The Commonly Discussed Options

co py ri gh t la w.

52 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

Additional steps in this direction that Vermont could take include the following:

• further reducing the fines for possession, e.g., to the $25 level established for Washington, D.C., or fully legalizing possession for personal use but not sales

• reducing the number of investigations and prosecutions of marijuana-market offenses or reducing sentence lengths for those convicted of producing or distributing marijuana

• eliminating incarceration for those convicted of producing or distributing marijuana • expunging criminal records of those who have been convicted of minor marijuana viola-

tions • after a certain lag, expunging criminal records of those convicted of distribution offenses.

(Vermont does not have mandatory minimum sentences for marijuana-distribution offenses, so eliminating them is not a relevant option. See 18 V.S.A. 4230.)

The level of enforcement effort deserves more attention than it usually receives. Exis- tence of prohibition does not imply any necessity to enforce the prohibition aggressively. Many jurisdictions traditionally put little energy into enforcing prohibitions against prostitution and gambling. Indeed, a number of U.S. cities have adopted local policies making marijuana enforcement the lowest law enforcement priority (e.g., Berkeley, San Francisco, and Seattle). On the other hand, enforcement and prosecution are primarily local decisions, harder to influ- ence at the state level.2

Implement an Alcohol-Style Commercial Model (Regulate Like Alcohol)

When people use the term legalize without further elaboration, they might often have in mind what might be called the standard commercial model, leaving production, distribution, and sale to the competitive private market, subject both to the standard laws and regulations that apply to all economic activity and to some additional rules specific to that product. For mari- juana, these additional rules mostly pertain to the following:

• who can use (e.g., anyone over 21) • quality control (e.g., testing requirements) • packaging (e.g., requiring certain labeling) • industry structure (e.g., requiring or banning vertical integration between producers, dis-

tributors, and retailers) • product selection (e.g., whether to allow the sale of concentrates and edibles, whether to

restrict potency) • retail operations (e.g., rules that keep a minimum distance between stores and sensitive

locations, such as schools; require vendor training; ban special sales and volume dis- counts; limit amount either per transaction or per user per day or month).

2 People commonly opine that the United States should be more like European nations on drug policy (e.g., Steves, undated). What such observers might have in mind is that the United States avoid excessive enforcement or expand funding for treatment, prevention, and harm reduction. In 2010, the United States had a somewhat higher marijuana arrest rate per capita than the average for 24 European countries: 275 versus 177 per 100,000 (authors’ calculations). But U.S. prevalence is also higher: The number of marijuana arrests per past-month user was actually higher in those European countries than in the United States (6.6 versus 4.9 per 100 past-month users). Likewise, the marijuana arrest rates in Canada in 2010 were about 221 per 100,000 people and 4.5 per 100 past-month users, only 10 to 20 percent lower than in the United States.

co py ri gh t la w.

Supply Architectures 53

To reduce the risk of exports to other states, Washington State has also sought to limit the amount produced to that needed to serve its own domestic market.

In other states, this approach is referred to as “regulate like alcohol.” But Vermont and the other so-called control states take a more restrictive approach to distilled spirits. We describe those systems, with their direct state involvement in one or more tiers of the distribution system (e.g., with state-controlled package or alcoholic beverage control [ABC] stores), under the heading “Have the Government Operate the Supply Chain.” Rather, “regulate like alcohol” for most of the country means just requiring market operators to obtain a license and follow a larger-than-usual set of product-specific regulations.3

Washington State and Colorado (and soon Oregon and Alaska) limit marijuana sales to dedicated stores, as opposed to allowing other retail stores to stock marijuana products along- side other items. That helps keep minors away from retail selling and renders impractical some cross-price subsidization tactics. (If convenience stores could sell marijuana, one might expect them to use marijuana as a loss leader, selling below cost in order to increase customer volumes and recoup forgone marijuana revenue through greater sales of other items.) It also keeps mari- juana sales separate from alcohol and tobacco sales.

The guiding spirit of this approach is to let the market evolve to maximize the efficiency of production, the appeal of products to consumers, and the size, scale, and scope of the market—subject only to remaining within the regulatory parameters. This spirit is reflected in which types of agencies are charged with implementing the rules: a revenue-raising agency in Colorado and the Liquor Control Board in Washington, rather than a public-health agency. A public-health agency might instead view the problem as one of killing off the black market while suffering the smallest possible increase in use. However, creating a public-health agency with the requisite capacity to issue licenses, monitor production, issue and enforce regulations, and collect taxes would constitute a challenge, while those activities are among the core com- petencies of revenue departments and liquor boards.

No one knows how a dynamic competitive market in marijuana would evolve,4 but we think that some plausible conjectures include the following:

• Production costs of usable marijuana will be much lower at production scales above 10,000 to 100,000 sq. ft. than they have been in traditional (indoor) operations that were smaller than a few thousand square feet, so average firm size will grow.

• Production will shift from indoors to greenhouses (greenhouses are cheaper, artificial lighting is expensive, and densely packed plants are more prone to pests and disease) and possibly to outdoor production (cheaper still but subject to some caveats explored below).

• The marijuana industry will want to shift consumption patterns away from traditional loose marijuana (for pipes and roll-your-own joints) that could require labor-intensive hand trimming via some combination of the following:

3 Further confusing matters is the fact that Colorado’s legalization explicitly made reference to “regulate like alcohol.” It now allows vertical integration between producers and retailers even though the central purpose of the “three-tier system” of postprohibition alcohol regulation in the United States is preventing such vertical integration and associated practices, such as producers selling directly to or otherwise controlling the retailers. 4 Indeed, this description ducks some important issues, such as the evolution of typical potencies, however typical gets defined, and whether additional cannabinoids beyond THC and CBD will rise to prominence.

co py ri gh t la w.

54 Considering Marijuana Legalization: Insights for Vermont and Other Jurisdictions

– mechanization of trimming (viewed skeptically by many in the industry5) – production and sale of extract-based products – sale of rolled joints (akin to conventional manufactured versus roll-your-own tobacco cigarettes).

• Declining costs of production per unit of intoxicant will shift profit opportunities toward – firms that promote brand identification via mass marketing – boutique brands for affluent consumers – on-premise consumption (compare cost of a muffin at a café with that of one from a grocery store)

– high-touch retailers – bundling (marijuana in candy bars might yield higher margins than just marijuana) – cross-subsidization (e.g., promoting ski vacations with complimentary marijuana).

The great unknown is the direction of product innovation:

• toward (or away from?) very-high-potency products • toward (but how far toward?) concentrates and edibles • the biggest wild card of all: the potential development of new products based on sys-

tematic extraction and blending of the hundreds of possibly psychoactive chemicals in the cannabis plant, of which today only two (THC and CBD) have reasonably well- characterized activity.

The contrast between extracts and conventional usable marijuana is important, so it bears elaboration—albeit with a reemphasis on the caveat that these projections are all speculative. These considerations are also not necessarily germane only to a commercial market model. A gov- ernment monopoly could promote the production and sale of extracts despite the associated public-health concerns, just as government monopolies promote lotteries despite various con- cerns they raise. However, as Rolles and Murkin (2013) observed, profit-oriented companies might be expected to behave that way, whereas regimes that balance profit with broader public interest might wish to restrict legalization to usable marijuana, at least initially. And, more generally, one might expect the industry to be more innovative under a (regulated) free-market model than with some of the other options, particularly, say, if production were limited to small noncommercial co-ops.

Usable marijuana generally refers to the flowering tops of cannabis plants and sometimes the leaves, suitably prepared.6 The flowering tops have the highest concentrations of THC but constitute only a modest proportion of the plant’s weight. So a considerable proportion of the THC produced by the plant resides in leaves and trim, material that, in the past, has often been discarded because machines that can efficiently extract that THC were too expensive

5 The majority view is that mechanized trimming destroys the aesthetics of the product. Another view is that some variet- ies are already amenable to mechanized trimming because of the shape of their buds and leaves, and growers could selec- tively breed this desirable trait into other strains. 6 Washington State’s Initiative Measure 502 (I-502) restricts the definition to flowering tops; by contrast, 18 V.S.A. Chap- ter 86 Subchapter 2 governing Vermont’s medical-marijuana program defines usable marijuana as “the dried leaves and flowers of marijuana, and any mixture or preparation thereof, and does not include the seeds, stalks, and roots of the plant.”

co py ri gh t la w.