For a decade there have been claims that methamphetamine, originally a West Coast biker drug, is spreading east, but if so, the diffusion has been fairly slow, and there is no evidence that use is significantly increasing. As seen in Figure 2, cigarette users tend to be daily users, but most recent users of cocaine and the hallucinogens did so only once or twice in the past month. About a third of marijuana users use almost every day; surprisingly, only half as many alcohol users do so. The prevalence of drug use provides a distorted picture of the actual health and safety harms posed by psychoactive drugs. There are several reasons for this. First, because drug use is only probabilistically related to drug harm, most harms to self and others are attributable to heavy users, and for most substances , a minority of users account for a majority of the quantity consumed. Second, school- and household-based surveys under represent hard core users, who are more likely to be truants, school dropouts, homeless, or institutionalized. Another major category of drug-related harm is the transmission of HIV. Injection drug use accounts for about a third of all AIDS cases in the US , and non-injection use is associated with an elevated risk of unsafe sexual practices. But there is no reliable way to attribute some fixed percentage of AIDS cases to cocaine use, to marijuana use, and so on. In keeping with our general analytical framework, recall that total harm is the product of prevalence x quantity x average harm. Thus, marijuana – by far the most prevalent illicit – accounts for a sizeable share of all three harm indicators. But relative to their much lower prevalence, it is clear that heroin, cocaine, and methamphetamine are disproportionately harmful substances. A 2007 Lancet article by David Nutt and colleagues offers what is to date the most sophisticated attempt to rate psychoactive substances by their “intrinsic” health and behavioral harms. In decreasing order of harmfulness,vertical farming rack the worst five drugs were heroin, cocaine, the barbituates, street methadone, and alcohol.
Tobacco was ninth, cannabis eleventh, LSD fourteenth, and MDMA eighteenth. If one were starting a society from scratch, it is unclear where one might paint a bright line separating licit from illicit substances, but it is difficult to see why alcohol and tobacco would be more accessible than cannabis or MDMA. But of course, we are not starting a society from scratch. The data in Table 2 are misleading if one wants to rank the intrinsic psychopharmacological harms of drugs – marijuana looming too large – but they are valid indicators if one wants to know the contributions of different substances to social harm under the current regime and with current patterns of use. These kinds of indicators tells us little about how harmful each substance might be in a regime of regulated legal access. Average harm per use might approximate the rating levels in Nutt et al. , though heroin use would become a lot safer per dose. Total harm might differ, if for example legal LSD were to become massively popular. But this might be a short term effect; societies seem to learn from experience and scale back on drugs that are obviously dysfunctional . The FY2008 national drug control budget allocates 36 percent of drug funding for interdiction and source-country controls, 28 percent for domestic drug law enforcement, 23 percent for treatment, and 12 percent for prevention . We can only offer a whirlwind tour of the empirical literature on these interventions. Readers can find comprehensive assessments in recent monographs by Boyum and Reuter and the National Academy of Sciences . The drug policy literature is enormous and yet remarkably thin, in that rigorous program evaluations are rare. There are some valuable cost-effectiveness analyses , but these are limited by the available descriptive data and some daunting problems of causal inference , and many of the available evaluations were conducted by program developers, raising concerns about intellectual and financial conflicts of interest.In a classic analysis, Reuter and colleagues explained why we should not expect big impacts of efforts to thwart drug production and trafficking. First, it is not possible to completely “seal the borders” against relatively small packages of chemicals that will be sold at very high prices; there are too many possible smuggling routes and tactics, and dealers are very adaptive. Second, the price structure of illicit markets is such that bulk drug products in source countries are “dirt cheap” compared to the high retail street prices in the US.
At the source, the economic value is low by US standards but high by local standards. But most of the markup in US prices occurs in the last few links of the distribution chain, within US borders. For example, Caulkins and Reuter note that the wholesale price of cocaine or heroin in a source country is only about 1 percent of its US retail street price. For example $1500 of cocaine in Colombia may be worth $15,000 at the US border, and $110,000 in the US retail street market. Thus even very large seizures in other countries are unlikely to have big effects on local prices. In recent years, defense analysts have used time-series data to argue that interdiction and source-country campaigns actually do have a significant impact on street prices and US demand. But these analyses have been debunked by a National Academy panel , arguing that the apparent correlations are spurious and amplified by selective focus on certain source-country interventions that happened to precede short-term price drops. It does not follow, however, that we could eliminate these programs entirely without a detectable effect. Most analysts believe that interdiction risks do raise prices; it is just that there are probably steeply declining marginal returns to such efforts. Presumably, these programs serve other US political, diplomatic, and economic goals beyond drug policy, laudable or otherwise. But we could probably cut back significantly on these efforts without seeing an increase in US drug consumption. What has this massive social experiment bought us? Early in the growth period, around 1992, one could argue that it was correlated with a considerable drop in drug use relative to the late 1970s . But this period of optimism was short-lived. By 1996, about half of the gains were gone, and levels of use have remained fairly stable since then, even as the drug prison population continued to rise. In fact, illicit drug prices have plummeted during a period when massive law enforcement sought vigorously to make drugs more expensive . This is troubling, because prices do matter; contrary to widespread belief, even addicts have been shown to be sensitive to drug prices . From the perspective of prevalence and quantity reduction, falling prices are a serious problem. But conceivably, falling prices may be beneficial from the perspective of harm reduction, because addicts might be expected to conduct fewer income-generating crimes to feed their habit. This is another illustration of the need to confront hard trade offs in thinking about drug policies.
The harshness of US marijuana enforcement has long received considerable criticism, and indeed it is difficult to defend . But Caulkins and Sevigny warn against exaggerated concerns about unlucky marijuana smokers rotting away in a prison cell. Although 38 percent of state and federal incarcerations for drug offenses involved simple possession, “for only 2 percent of imprisoned drug-law violators was there no reason whatsoever to suspect possible involvement in distribution…depending on how strict a definition one preferred, one might argue that anywhere from 5,380 to 41,047 people were in prison in the United States solely for their drug use.” On the other hand, many who avoid time in prison do spend time in jail – as much as a third of arrestees in a study of three counties in Maryland . More troubling is the disproportionate imprisonment of African American men. African Americans accounted for about half of all drug incarcerations . A major factor is the differential severity of mandatory minimum sentences for crack vs. powder cocaine. Under these laws, a dealer would have to sell 500 grams of powder cocaine but only 5 grams of crack cocaine to receive the same five-year sentence. Since crack is more likely to be sold in African American communities, this has greatly widened the racial gap in sentencing. Even putting aside the questionable pharmacological and moral aspects of this differential policy, there is no evidence whatsoever for its effectiveness in controlling crime. Caulkins and colleagues show that conventional sentencing is significantly more cost effective. Although the crack mandatory sentences were trimmed somewhat in 2007,commercial indoor vertical farming and the Supreme Court recently acted to restore some judicial discretion in these cases . Whether these changes will translate into a closing of the large racial differential remains to be seen. The optimal level of drug law enforcement is surely well above zero, but just as surely, well below current levels . Caulkins and Reuter argue that we could reduce the drug prisoner population by half without harmful consequences; they note that this would still leave us with system “a lot tougher than the Reagan administration ever was.” Kleiman suggests tactics for getting more mileage out of less punishment through the use of small, quick sanctions, strategically deployed. In 2005, there were about 1.8 million people in substance abuse treatment in the US, about 40 percent for alcohol, 17 percent for the opiates, 14 percent for cocaine, and 16 percent for marijuana . There are certainly many thousands of people who need treatment and are not receiving it. Whether expanding the available treatment capacity would bring them in is an open question. We should be wary of assuming that a purely “public health” approach to drugs can work; the police and courts play a crucial role in bringing people into treatment – increasingly so with the expansion of drug courts and initiatives like California’s Proposition 36, the 2001 law which permits treatment in lieu of incarceration for those convicted for the first or second time for nonviolent drug possession . For most primary drugs of abuse, criminal justice referrals are a major basis for treatment: in 2005, 57 percent of marijuana treatment, 49 percent of methamphetamine, and 27 percent of smoked cocaine. But 36 percent of clients in alcohol treatment were referred by the criminal justice system, so legal status may not be the crucial lever. In a sophisticated cost-effectiveness analysis, Rydell and Everingham estimate that the U.S. could reduce cocaine consumption by 1 percent by investing $34 million in additional treatment funds, considerably cheaper than achieving the same outcome with domestic drug law enforcement , interdiction , or source country controls .
But because treatment effects are usually estimated using pre-post change scores that are vulnerable to two potential biases . First, the posttreatment reduction could reflect a simple “regression to the mean” in which an unusually extreme period of binge use would be followed by a return to the user’s more typical levels, even in the absence of treatment. Second, treatment pre- and posttests are vulnerable to selection biases because clients who enter and remain in treatment until post-treatment measurement are a non-random and perhaps very unrepresentative sample of all users. Regression artifacts would inflate treatment estimates; selection biases could either inflate or deflate the estimates. We believe that the full weight of the evidence makes it clear that treatment is both effective and cost-effective, but until these problems are better addressed, we cannot be sure that the benefits of expanded treatment would be as large as Rydell and Everingham implied. Even its most passionate advocates recognize that treatment’s benefits are often fleeting. About three quarters of heroin clients and half of cocaine clients have had one or more prior treatment episodes . Forty to sixty percent of all clients will eventually relapse, though relapse rates are at least as high for hypertension and asthma treatment . Importantly, Rydell and Everingham recognized that treatment can provide considerable health and public safety benefits even if it only reduces drug use while the client is enrolled. Held up to a standard of pure prevalence reduction , treatment is unimpressive. But by the standards of quantity reduction and harm reduction, treatment looks pretty good.