Drug Policing in the 21st Century Concepts and Strategies for Policing the New Drug Crisis

Drug policing faces two simultaneous crises. One is the drug crisis itself, which the public and policymakers expect police to play a role in suppressing. The other is a crisis of public confidence, in policing in general but especially in the efficacy of the enforcement-driven “War on Drugs.” This report frames an approach to policing drugs meant to address both crises—a 21st-century approach to drug policing.

Routine drug policing is widely perceived to be an ineffective approach to controlling drug problems. This is because it is relatively hard for policing to cut off the supply of drugs altogether. Routine enforcement can increase the price of drugs, making the drugs scarcer, but these effects are likely small, and the effects of price on demand are, in turn, quite small in addicted subpopulations. These concerns are particularly relevant in the current crisis, in which novel methods of drug production have driven prices to rock bottom.

Faced with this evidence, some argue that policing and drug problems should be totally disconnected, through a policy of decriminalization. This, advocates argue, would reduce both the health harms of drug use and the harms of the criminal-justice system. However, the evidence indicates that the former claim is probably wrong—the expected average effect of decriminalization on figures like the overdose death rate is probably close to zero. Moreover, policymakers are capable of addressing the fact that the criminal-justice system can harm drug users, and it would be foolhardy to drop the good aspects of drug enforcement as a method to avoid these “bads.”

What are the “goods” of drug policing, and how can they be bolstered? This report culminates in three strategies for a modern, evidence-based approach to drug policing:

• Drug-Market Crackdowns: Rather than do routine, haphazard enforcement, police can focus all their resources on particular drug markets or drug problems and enforce against them simultaneously. This has the effect of crippling the market, circumventing the limited effects that policing has on price by substantially reducing supply altogether.
• Responding to Emerging Threats: Emerging drug markets—especially novel synthetic substances—represent a particular opportunity for police effectiveness. By differentially targeting small markets, police can keep them small, having an outsize impact relative to targeting large but hard-to-control markets.
• Policing as Public Health: Police officers are not just enforcers; they are also first responders, and they should see themselves as frontline actors in getting people the treatment they need. Equipping police with naloxone is a widely taken first step. But treatment referral following overdose, or diversion to treatment in lieu of arrest or prosecution, is a promising way for police to take the lead on controlling drug problems.

These strategies collectively imply a more strategic approach to drug policing than the historical norm. Routine buy-busts are likely no longer effective strategies for fighting the drug crisis—if they ever were. But police can still play a substantial role in combating the drug crisis. And if they do so intelligently, they can regain the public’s trust.

Drug policing currently faces two crises.

One is the ever-rising tide of drug overdose deaths. There were an estimated 112,000 drug overdose deaths in 2022, and about the same number likely occurred in 2023.[1] These figures represent an exponential increase in the deadliness of drugs from even two decades ago. That increase in deadliness is, in turn, driven by the proliferation of highly potent synthetic drugs.[2] Controlling the new synthetic crisis is arguably the greatest challenge that drug enforcement has ever faced.

At the same time, drug enforcement faces a profound crisis of public confidence. This extends beyond the general decline in confidence in the police, which has recovered from the record lows of 2020.[3] Drug policing, specifically, is widely perceived as pointless, racist, and a major driver of mass incarceration and the destruction of minority communities. In a 2021 poll commissioned by the ACLU, 83% of Americans said that they believed that the War on Drugs was a failure, 65% supported ending it, and 66% supported ending criminal penalties for drug possession.[4] There is a great deal of public sentiment, in other words, in support of dramatically curtailing, if not altogether eliminating, drug enforcement.

Against this backdrop, the question that this report seeks to answer is how stakeholders—police and civilian leaders, police officers, and the general public—should think about drug enforcement. What should drug enforcement look like against the drug crisis? How can drug enforcement prove its effectiveness to a skeptical public? How can it help people suffering from addiction, rather than hurt them? What, in other words, should drug policing look like in the 21st century?

To answer these questions, this report first investigates how drug enforcement works. It argues that conventional drug enforcement—arresting dealers and seizing drugs—operates by increasing the costs of drug sellers, in turn reducing supply and thereby increasing the price and reducing the quantity of drugs sold. This effect is strongly limited by the elasticity of the price of drugs with regard to enforcement, and the elasticity of demand with regard to price—both of which are, by best measures, quite small. This means that very large increases in policing are required to effect even small changes in the level of drug consumption. This constraint is particularly important in the context of the synthetics crisis, which has—by virtue of the innovation that synthetics represent—driven street prices to near-zero.

This pessimistic argument does not mean that policing has no role to play in controlling the drug problem, or that decriminalization is a prudent choice. First, this report reviews the evidence on decriminalization, arguing that there is little reason to believe that it achieves its stated goals. Notwithstanding underinformed claims about Portugal, which this report interrogates, the balance of the evidence suggests that decriminalization has little effect on public health, and it might have only a marginal effect on drug users’ involvement in the criminal-justice system. More important, decriminalization is just one among many directions that policy can take. And it is simply less efficacious than a counterfactual effective policing strategy.

What does that effective policing strategy look like? The last three sections of this report lay out several approaches. One, on drug-market “crackdowns,” focuses on attacking markets at many points simultaneously, shrinking them and making them much easier to control. Such crackdowns have a proven track record of success. A second approach is targeting small or “emerging” markets, using comprehensive surveillance and deterrent warnings. Such a strategy helps overcome many of the constraints imposed by large markets and might reduce net harm overall. The third approach is “public-health policing.” It takes advantage of police officers’ status as first responders not only to administer lifesaving naloxone but to serve as a key point for diversion to treatment for users who need it.

None of these tactics is novel to this report—each represents a previously studied approach tried by several police departments. But collectively, they suggest a different drug policing for the 21st century: policing that is more strategic, more outcomes-focused, and more oriented toward public health. Rather than policing less, drug enforcement should meet the 21st-century challenge by policing smarter. This report explains how.

How does drug policing affect drug problems? At first glance, the relationship might seem so intuitive as not to be worth articulating. But a discussion might help elaborate some of the challenges faced by cops in contemporary drug policing.

First, some definitions. “Drug enforcement” should be taken to mean activities implicating a primary role for a local police department. It should not be taken to include operations involving countries that are the drugs’ source (which can involve law enforcement, e.g., the Drug Enforcement Administration), nor does it include interdiction in transit countries, at the border, or short of the local municipality. Rather, enforcement here means the range of approaches available to local law enforcement: stop and search, civil enforcement (e.g., ticketing), seizure of drugs, arrest, incapacitation in prison or jail, and the deterrent threat of any of these consequences. In this section,

specifically, we are interested in what we might call “routine” enforcement—enforcement carried out in the course of everyday police work. Nonroutine enforcement—deliberate operations that try to work within the dynamics of drug markets—is discussed later.

“Drug problems” should be taken as expansively as possible, i.e., any problem associated with the production, distribution, sale, or consumption of illicit substances.[5] That could mean violence between market participants; disorder generated by the operation of drug markets; crimes committed by people trying to purchase drugs; individual harms of drug use, chronic (heart disease, mental illness, infection by needle sharing) or acute (overdose, death through intoxicated behavior); and social harms of drug use (harming of others through intoxicated behavior, damage to interpersonal relationships, public disorderliness).

What is the relationship between drug enforcement and drug problems?[6] Intuitively, the former should affect the latter primarily through its effect on supply. Police may play a role in demand reduction, as discussed below, but this is not the primary focus of this section. This enforcement-problem relationship can take the form of enforcement actions targeting individual dealers or drugs. If a police officer seizes drugs, either directly or during arrest, those drugs are no longer available for consumption. If a drug dealer is arrested and incapacitated, the number of dealers is reduced—both directly and through deterrence. Consequently, would-be buyers must spend more time looking for fewer drugs—both effects that reduce drug consumption. This style of enforcement is the bread and butter of drug policing: buy-busts, rounding up dealers, and turning out users’ pockets.

If the drug supply were static—if manufacturers produced drugs at a constant rate unresponsive to enforcement efforts—this kind of enforcement could just escalate until all drugs entering a given jurisdiction were seized, reducing availability to zero. But drugs are commodities, sold in markets, which respond dynamically to regulatory effects on the supply chain. Consequently, in large drug markets with many buyers and sellers, and with a variety of options to source product, consumers can replace their drugs relatively easily. For example, in a study of Vancouver injection-drug users, 70% of those who had their drugs seized said that they “immediately” acquired more drugs.[7]

Enforcement and Price

Does this mean that reactive enforcement has no effect? No; it means that loss of product becomes a component of retail price, in much the same way that “shrinkage” (i.e., theft) is a component of legitimate businesses’ operating costs, and therefore prices. This is true in terms of having to replace lost product, but it is also true insofar as arrest and incarceration risk affects how much dealers need to be paid to work.[8] In other words, drug enforcement does reduce supply by shifting the supply curve, both increasing the market-clearing price and reducing the quantity supplied.

By how much does a dose of enforcement shift prices? Estimates are few and ambiguous. A 1998 study found no relationship between changes in DEA drug enforcement and subsequent cocaine and heroin prices.[9] A 2003 study found that the price that drug users reported paying did not change after a large heroin bust in Vancouver.[10] Using state-level data, researchers in a 2004 study estimated that a 30% increase in arrests yielded an increase in cocaine prices of 6%–11%.[11] In 2006, a College of William and Mary economist reported that increasing arrest rate is negatively associated with cocaine price, which she attributes to increased enforcement encouraging sellers to move product more quickly.[12] A 2016 study showed that methamphetamine precursor regulation led to large but temporary increases in the price of meth, reverting within a year.[13]

None of these estimates is ideal. Most rely on the Drug Enforcement Administration’s price data, the validity of which is hotly debated in the scholarly literature.[14] Researchers vary in their preferred measure of policing intensity. And none of the studies is designed to isolate the causal effect of enforcement on price, net of all sources of bias. For decades, simultaneity bias plagued the literature on the effect of policing on crime generally, with papers regularly failing to disentangle the way the level of policing affected the level of crime from the way the level of crime affected the level of policing.[15] Similarly, it is easy to see how falling drug prices could induce more enforcement, producing an illusory negative association between enforcement and price.

Still, the limited evidence suggests that the marginal arrest has a relatively small effect on the level of drug consumption. Even the most positive estimated effect, the 2004 study from Ilyana Kuziemko and Steven D. Levitt, implies that a 1% increase in arrests increases prices by roughly 0.2%–0.33%.[16] There were 766,000 drug arrests in 2022; increasing prices by 10% (assuming, probably incorrectly, a constant ratio) would entail an additional 230,000–380,000 arrests. Specific supply controls—like making pseudoephedrine, a widely available meth precursor, a behind-the-counter drug—can have large effects on price. But, at least in that case, the effects appear to be only temporary, as markets adjust and source new precursors.[17]

Price and Demand

Even if we assume that drug arrests are unit-elastic with regard to drug price—that a 1% change in arrests corresponds to a proportional 1% change in price—arrests will still have a limited effect on actual consumption. This is because drug demand itself is not unit-elastic with regard to price. Estimates consistently find that the demand for drugs is inelastic, meaning that consumers reduce their consumption by less than 1% for each 1% increase in price.[18] This implies that the effect of enforcement on the market will do more to increase prices—and thus drug dealers’ earnings—than it will to reduce the actual quantity consumed.

Research often bears out the finding that routine patrol or other drug-control tactics do relatively little to reduce drug offenses (offenses are plausibly a proxy for the level of drug-related activity, and therefore demand). A 2000 study of Oakland, California, found that a civil remedy program targeting areas of heavy drug use was substantially more effective at reducing drug-related calls for service than “haphazard, business-as-usual patrolling” in matched control areas; in the latter, police calls actually rose.[19] A 2003 study found that police raids on nuisance bars reduced drug dealing, but the effects largely disappeared after the raids were over.[20] A 2005 study found that stationing Philadelphia police on high-intensity drug corners had no effect on rates of drug crime citywide.[21] A 2015 study found that CCTV cameras paired with patrol officers reduced violence and disorder but had no effect on drug offending.[22] A study published last year found that bicycle patrol in Philadelphia’s Kensington neighborhood might actually have increased drug crimes.[23] 

Assume, for the sake of argument, that the effect of enforcement on price is far more elastic than the low-quality estimates cited above. Even then, police enforcement’s effectiveness for drug problems is constrained by the price elasticity of demand for drugs—how much a change in the price of drugs affects the consumption of drugs. On average, the most recent meta-analysis finds, the demand for drugs is slightly inelastic: a 10% increase in price results in about a 9% decrease in consumption. But that overall figure hides important heterogeneity. Among regular users—i.e., addicts—a 10% increase probably yields only a 3%–4% decrease in consumption.[24] Some estimates go lower, with one analysis using data on arrested drug users—who underwent urinalysis to independently corroborate their reported levels of use—finding price elasticities closer to a 1% decrease in consumption, given a 10% increase in price.[25] In other words, enforcement can have a much larger effect on casual users who, by dint of the infrequency and character of their use, are not those with the most drug problems.

The survey of the literature so far does not mean that enforcement fails to “work.” That is, the effect of enforcement on price is probably not zero, nor is the effect of price on consumption. This is especially true in the long run, insofar as higher prices can make it harder for people to initiate drug use and thereby reduce the number of people who become addicted.

Rather, the preceding survey means that insofar as enforcement tries to reduce drug problems by affecting the price of drugs, it faces an uphill battle. All the available evidence suggests that it takes a substantial increase in enforcement to have even a small impact on consumption. Additional marginal busts or arrests can have some effect, but it probably takes a fairly large number to have that effect.

The previous section is critical to understanding the drug challenges to law enforcement in the current moment. In short: a dramatic change in the drug supply has driven prices to near-zero. This, in turn, means that truly heroic levels of additional enforcement are necessary to obtain meaningful reductions in drug consumption.

In the U.S. today, the major drug problem is the now-startlingly high rate of overdose death. There were an estimated 112,000 drug overdose deaths in 2022, the most recent year with complete data.[26] That represents a 17% increase over 2020, a 157% increase over 2012, and a 332% increase over 2002. The primary driver of this increase is the widespread availability and use of “synthetic” drugs—principally synthetic opioids like fentanyl, but also methamphetamine. Synthetic drugs are differentiated from organic drugs like marijuana and opium, or refined drugs like heroin or cocaine. While the latter are grown in a field or produced from plant matter, the former are produced in a lab using simple precursor chemicals. Synthetic drugs can be, and—as currently produced for sale—are, far more potent. This makes them a more attractive product for traffickers and consumers, and it makes them easier to overdose.[27]

Crucially, the economies of synthetic drug production and distribution make them much, much cheaper than their organic counterparts. The raw materials, labor, and processes needed to produce synthetics are less expensive and—relevant to the fact that risks help determine price—much easier to hide from counter-narcotics agencies. A clandestine lab is much harder to spot from the air than a poppy field, for example. High potency per dose means that consumers need relatively little actual volume to satisfy demand—one estimate suggests that Americans consume less than 10 tons of fentanyl per year, compared with 145 tons of cocaine and 47 tons of heroin.[28] The lower total weight needed to supply the market means that traffickers can absorb a much higher rate of shrinkage (especially given how cheap it is to replace lost product) and therefore pass fewer costs on to the consumer. Finally, because synthetic drugs are cheap to make and cheap to smuggle, they are easy to acquire for retail, creating a robust (and therefore price-minimizing) market at the street level.

Confirming this model, synthetic drug prices are shockingly low. On the street, fentanyl pills can retail for as little as 50 cents per dose.[29] In one review of darknet prices (combining retail and wholesale volumes), fentanyl sold at about 12 cents per dose, on average, compared with heroin, at 90 cents per dose.[30] Mirroring the evolution of the market, fentanyl prices fell 50% between 2016 and 2021, according to one analysis.[31] Prices of meth have similarly fallen, by as much as 80% between 2016 and 2022.[32]

The declining price probably reflects in part a large reduction in the level of enforcement. Police departments reported about 540,000 non-marijuana[33] drug-enforcement arrests to the FBI in 2022, a remarkable decline from the 800,000 reported in 2019.[34] This reduction in intensity was part of a larger drawdown of police activity in the wake of the Covid-19 pandemic and protests instigated by the murder of George Floyd.[35] Insofar as price is even somewhat responsive to enforcement, reducing the latter’s intensity by roughly a third almost certainly contributed to current rock-bottom prices.

That said, prices were falling before 2020, and the fundamentals of synthetic drug production suggest that they would have done so even if the level of enforcement had remained constant. Lower levels of enforcement, in other words, have only hastened the inevitable. The emergence of synthetic drugs in the illicit market represents a profound increase in productivity for drug producers and—like such increases in productivity in licit markets—it means that consumers are getting a better (i.e., more potent) product at a better price.[36]

The dramatic secular decline in prices means that traditional—routine, reactive—drug enforcement is particularly ill-suited to addressing current drug problems. Assume again the Kuziemko and Levitt elasticities from the previous section, i.e., a 1% increase in enforcement is associated with a 0.2%–0.33% increase in price. If fentanyl prices have fallen 50%, returning them to 2016 levels entails increasing prices by 100%, implying a tripling or quintupling of enforcement levels. Much larger increases would be necessary to, for example, undo the 80% decline in methamphetamine prices, or octuple the price of fentanyl to match the current price of heroin.

Of course, price may not be linearly responsive to enforcement in this way, particularly at very low levels—increasing price from $0.50 to $1 and from $10 to $20 is the same percentage-wise, but very different in absolute terms. Similarly, enforcement may theoretically have impact on quantity consumed beyond price, e.g., by increasing search times. That said, the basic problem persists. Drugs are extremely cheap, such that extremely high levels of enforcement are necessary to increase prices to levels that have anything like a real effect on the quantity consumed.

Some readers, faced with the high levels of enforcement necessary to increase prices, might argue that the cost is worth it. In some sense, this is a reasonable position—the drug crisis probably costs, expressed in monetary terms, trillions of dollars per year.[37] At the same time, readers should give more weight to cost-benefit concerns. Given a fixed—or, in reality, declining[38]—population of officers with finite time, it is not obvious that dedicating millions of police man-hours to drug enforcement is the best use of police resources, especially compared with the substantial social costs of nondrug crime.[39] This is especially so because the estimated elasticities of policing regarding other types of crime, particularly violent crime, give much more reason for optimism about impact compared with drug crime.[40] Similarly, fixed government dollars dedicated to the drug crisis may not necessarily be the best-used increases in enforcement. Other spending categories, such as treatment and prevention, may be a better use of the marginal dollar in terms of returns.[41] 

At this point, readers might have a sense of profound pessimism about the usefulness of policing in combating the drug crisis. Put simply, routine policing is not likely to have an appreciable impact on the consumption of synthetic drugs, particularly among the population most at risk for adverse effects. Does that mean, though, that police should have nothing to do with combating the drug problem? As the next section argues, the simple answer is: not at all.

Against this backdrop, one increasingly fashionable view argues that policing and drug problems should have nothing to do with each other. The costs of drug enforcement, in both money and liberty, are nontrivial. The federal government spent $11 billion in FY 2023 on domestic drug law enforcement, and states and municipalities doubtless spent much more collectively.[42] Police departments reported half a million non-marijuana drug-crime arrests to the FBI in 2022, as well as another 200,000 marijuana arrests.[43] Roughly 190,000 people were incarcerated nationwide on drug charges in 2022.[44] While both figures have declined substantially in recent years, these are still substantial costs imposed by enforcement. What if, instead of paying these costs, we stopped treating drugs as a criminal-justice issue?

At the libertarian fringes, this view is usually expressed as support for full legalization and commercialization, with hard drugs retailed in a lightly regulated market, as cigarettes and alcohol currently are. With the exception of marijuana, though, this position commands relatively little public support.[45] More popular is the notion of “decriminalization,” removing criminal penalties for drug crimes and replacing them with civil penalties. In a 2019 poll from the Cato Institute, 55% of respondents favored “recategorizing drug offenses from felonies to civil offenses.”[46] And in the previously mentioned ACLU poll, 66% “support eliminating criminal penalties for drug-possession statute and reinvesting drug enforcement resources into treatment and addiction services.”[47]

Decriminalization, particularly of possession offenses, is not a fantasy. Small possession is decriminalized—meaning that it carries no jail time, but entails only a fine or other civil consequences—in Czechia, Portugal, Italy, Spain, Mexico, and Uruguay, as well as in the Canadian province of British Columbia and the Australian Capital Territory.[48] In the U.S., voters in Oregon decriminalized small possession under Measure 110 in November 2020, although the state reversed course in 2024. In February 2021, the Washington State Supreme Court ruled the state’s drug-possession statute unconstitutional, leaving possession de facto decriminalized for a number of months.[49]

Why decriminalize? Advocates generally emphasize the harms imposed by the criminal-justice system on people who use drugs. They also argue that decriminalization reduces the stigma associated with drug use, thereby inducing people to seek treatment or harm-reduction services. They assert that decriminalization “opens the way” to a “health approach” to drug problems. This implies a hard distinction between the criminal-justice and public-health systems, with drug decriminalization encouraging—even if only symbolically—a transition from one to the other.[50] 

These arguments raise several empirical questions. Does decriminalization meaningfully advance a “health approach” to drug problems, by reducing stigma or by symbolic reassociation? And does it meaningfully reduce criminal-justice involvement?

Health Impacts

Most research on the effect of decriminalization on health and social outcomes focuses on Portuguese decriminalization. Motivated by a nationwide heroin crisis in the 1990s, Portugal decriminalized “consumption, acquisition, and the possession for personal use of narcotic drugs and psychotropic substances” in 2001, replacing prison penalties with administrative fines for offending.[51] Overdose deaths subsequently fell sharply. While they later climbed, Portugal’s overdose death rate remains well below the European average, which advocates of the Portuguese approach take as evidence of success.[52]

Of course, identifying a decline following decriminalization does not establish causality—the drop may represent some exogenous shift in drug supply or demand, or it may just be mean reversion as the crisis crested and receded. The high-quality literature on the topic is surprisingly ambivalent. A 2011 study compared Portugal with other European countries using a difference-in-difference approach and found that decriminalization significantly increased overdose deaths, although the author raised questions about the significance of the finding.[53] A 2017 study compared Portugal to a synthetic “control Portugal” constructed of a weighted average of other European countries.[54] The study finds that decriminalization reduced overdose deaths relative to the synthetic counterfactual, although the results are similarly not significant.[55] A 2022 study constructed and parameterized a market model that predicted that decriminalization slightly reduced OD deaths in the short run but had no long-run impact.[56] Lastly, a 2023 study exploited variation in the level of a large immigrant group with high rates of drug use to find that decriminalization lowered arrests but had no effect on drug-related hospital admission rates.[57]

In other words, the research suggests that there may have been increases or decreases in overdose deaths, but the strongest signal is that it had no effect on that and other measures of health outcomes. These papers are limited by the fact that they identify the effect of decriminalization as the difference between Portugal and some control after 2001. But Portuguese decriminalization did not happen in a vacuum; it was combined, at the policy level, with a substantial increase in the availability of drug-treatment services. Between 2000 and 2009, the number of outpatient treatment units in Portugal increased 58%.[58] The number of people actually in treatment rose from 24,000 in 1998 to 39,000 in 2008.[59]

It is not obvious that Portugal’s decriminalization represented a meaningful change in the actual intensity of enforcement. In 2015, one researcher showed that in the immediate lead-up to decriminalization, less than 1% of Portuguese people incarcerated on drug offenses were convicted of use charges. Following decriminalization, while police could no longer arrest drug users, they could issue citations. However, the number of citations was basically the same as the number of arrests before reform.[60]

In other words, Portugal’s decriminalization might not—despite the attention often paid to it—tell us that much about the actual effects of decriminalization. Other European decriminalizations are even less well studied. A 2001 book reviewed the evidence on Italy’s de- and re-criminalizations, calling it ambiguous. As the authors note, drug overdose death rates followed the same trend in Italy during a period of re-criminalization as they did in Germany (where drugs were criminalized) and Spain (decriminalized). This suggests a lack of effect.[61]

There is some high-quality research on the effects of drug decriminalization in Oregon and Washington. Pivotally, both jurisdictions decriminalized without, at least for a period, expanding treatment services, allowing the identification of the effect of decriminalization absent Portugal-style treatment expansion. Two 2023 studies both used synthetic control methods to identify the effects of decriminalization in both states.[62] However, the two papers used slightly different definitions of overdose death and different time periods. Probably as a result, they returned slightly different estimates of the effects of decriminalization. Researcher Noah Spencer found that decriminalization increased overdose deaths by 0.39 per 100,000 people (p = 0.02) and in Washington by 0.17 per 100,000 (p = 0.02). Spruha Joshi et al. found that Oregon’s decriminalization increased overdose deaths by a nonsignificant 0.27 per 100,000 (p = 0.26) and in Washington by a marginally significant 0.112 per 100,000 (p =0.06).[63]

Which of these results to find convincing comes down to which data choices one thinks are best, as well as how much importance to ascribe to statistical significance, particularly in the context of synthetic control methods.[64] Broadly, though, the U.S. results are consistent with decriminalization having no effect on overdose deaths (similar to the Portuguese results) or very slightly increasing them relative to counterfactual. The imprecision of all these results, and their sensitivity to methodological choices, suggests that it is possible, albeit far from guaranteed, that a jurisdiction can accomplish decriminalization without dramatically increasing overdose deaths. However, almost none of the foregoing research suggests that decriminalization reduces overdose deaths. It is thus hard to support the proposition that decriminalization per se reduces drug-related health harms.

Criminal-Justice Impacts

What about the argument that decriminalization reduces the harms of exposure to the criminal-justice system? Here, the case is theoretically stronger. Pretrial detention has measurable and significant negative effects, particularly on future employment and re-offense risk.[65] Among people who use drugs, jail booking is associated with greater risk for overdose death. This association is often explained by detention lowering a user’s tolerance and thus increasing overdose risk upon release.[66] A similar risk is often cited when police officers seize users’ drugs, although that claim is belied by the observation (mentioned previously) that most users replace their seized drugs almost immediately. At least one study finds that drug busts are chronologically and geographically associated with an increase in overdose deaths.[67]

Complicating this argument is the observation that decriminalization of small possession does not necessarily have a large effect on drug users’ risk of exposure to the criminal-justice system. One reason for this is that jurisdictions that decriminalize often already have low levels of drug-possession enforcement. Immediately before drug decriminalization in Portugal, for example, just 25 people were in Portuguese prisons for drug-use charges.[68] In Oregon, drug possession accounted for just 7% of arrests in the year before Measure 110 passed,[69] and none of those arrestees went to prison; as the Oregon secretary of state put it: “Prior to M110, Oregon had no adults in custody serving time in prison for solely drug possession-related offenses.”[70]

At the same time, people who use drugs, particularly people who are severely addicted to them, frequently cycle through prisons and jails for nondrug offenses. For example, 24%–36% of people with a heroin-use disorder pass through correctional facilities each year.[71] Some 60%–80% of arrestees in one five-city survey tested positive for drugs at arrest, and roughly 66% of sentenced jail inmates meet the criteria for drug dependence or abuse.[72] Many of these people are not incarcerated because of drug possession offenses. For example, 40% of state inmates convicted of robbery, burglary, or car theft said that they were under the influence at the time of their offense, and 25%–33% of state burglary and robbery offenders said that their crime was motivated by the need to buy drugs.[73]

In other words, decriminalization may not do that much to reduce the exposure of drug users to the criminal-justice system. Time and energy might be better expended on asking how to improve the quality of drug-treatment services provided within the system, rather than pushing for decriminalization.

This last point suggests that the question is not whether decriminalization reduces criminal-justice harm but whether decriminalization reduces harm relative to a given level of drug-enforcement intensity. For example, decriminalization is probably harm-reducing relative to a regime in which anyone who uses drugs is shot in the head without trial. But is it harm-reducing relative to a low-intensity regime of drug policing? Relative to a high-intensity regime? What’s the indifference point?

At the least, there is an important difference between no enforcement and some enforcement. As previously discussed, casual users’ demand for drugs is almost unit-elastic, suggesting that even low levels of enforcement can substantially reduce or deter casual use, which can indeed reduce the acute harms of even nonaddictive intoxication. Furthermore, deterring casual use should, over the long run, reduce initiation—in turn, reducing the size of the addicted population. What “low levels” of enforcement look like is a tricky question; but generically, if casual use is growing in prevalence—as it did in the late 1970s, for example—enforcement may be too low.

In addition, drug sales and use generate harmful externalities. Public intoxication, whether with licit or illicit substances, is unsightly and potentially dangerous to others. Concentrations of drug-dealing activity—drug markets—are associated with violent crime alongside increased risk of overdose death.[74] Even if drug enforcement does not reduce the level of drug use, it may still be justified insofar as it reduces these other drug-associated problems.

Many of the harms that decriminalization is meant to prevent, furthermore, are associated with lapsed access to drugs. But several innovative programs have sought to expand access to drug treatment in jails, combined with continuing care after release. Similarly (as discussed below), effective policing can combine drug busts with access to overdose prevention and treatment services. If people with substance-use disorders pass through these programs, are they better or worse off than if decriminalization had kept them out of jail or let them hold on to their drugs?

The point here is that even insofar as decriminalization reduces criminal-justice harm, the level of criminal-justice harm is not an inalterable policy variable. Insofar as policy creativity is on the table, it is a deliberate choice—often an ideological choice—to prefer decriminalization to expanding opportunities for care within the criminal-justice system.

Such expanded opportunities are mostly outside the scope of this report. But the next sections examine one particular institution of the criminal-justice system: policing. After all, if police are making drug problems worse, depolicing drugs is preferable to the status quo. But if policing can make drug problems better—by minimizing the availability of drugs, controlling the externalities of drug markets, and providing services and support to struggling users—we should prefer to move drug policing in that direction, rather than simply policing less.

The next three sections discuss three evidence-based approaches to drug policing that are responsive to the 21st-century problem: strategic drug-market disruption, emerging-threat suppression, and public-health policing. Each of these is a component of a positive, 21st-century drug policing strategy that offers a smarter, more humane alternative to the abdication of responsibility symbolized by decriminalization.

One of the great frustrations of routine drug policing is that it never seems to make the market smaller.[75] In Cop in the Hood, his ethnography-cum-memoir of his year in the Baltimore Police Department, CUNY criminologist Peter Moskos captures this frustration:

The drug-dealing block is a buzz of constant activity. Dealers hawk their wares, customers come and go, and addicts roam the street hustling for their next hit. Occasionally a police car will appear and the street crowd will disperse, slowly walking away from the police car.… Soon after the appearance of a police car, the street will be deserted. When the police car leaves, the crowd returns. The Eastern District’s 45,000 residents account for over 20,000 arrests every year. Most arrests are drug-related. Police officers patrol in their cars, respond to 911 calls, and clear corners. These officers, who by and large hate the ghetto, are frustrated to see those arrested go free in the revolving door of the criminal justice system: “justice for criminals,” goes the well-worn police cliché. The cycle repeats. Police earn court overtime pay while residents get rap sheets. It’s a horrible equilibrium, and police are the fulcrum.[76] 

Moskos’s choice of the word “equilibrium” is apt, for what he and his fellow officers experienced was an equilibrium: a neutral or balanced state of a system—in this case, a drug market. There are very few barriers to entry into selling drugs: as long as you have a supplier, highly addictive products basically sell themselves.[77] Consequently, the number of sellers in the market is set by the risk-adjusted rewards of selling—how much the dealer can expect to make, adjusted for how likely he is to end up dead or in prison.[78] If officers arrest and otherwise impose risk at a constant rate, the equilibrium level of employment remains constant.[79]

Policing, as previously discussed, operates on drug markets by raising these risks. But the effectiveness of this approach is highly contingent on the size of markets. In relatively large drug markets—like the Eastern District market patrolled by Moskos—a large number of dealers means that the per-dealer arrest risk is low, and a very large increase in enforcement is necessary to appreciably alter the incentives in the system.[80] To understand this point, consider a market in which there are 10 dealers and police make five arrests a week. The weekly arrest risk in the market is 50%, which is extremely high. By contrast, in a market in which there are 200 dealers and police arrest five a week, arrest risk is 2.5%—much smaller. More important: in the big market, doubling the number of arrests has a much smaller impact on the arrest risk—and therefore drug prices/dealer participation—than in the small market. A 1986 study already cited, for example, estimated that as of 1984, dealers in the relatively small heroin market faced an annual probability of incarceration of about 14%, while dealers in the much larger marijuana market faced an incarceration risk of 0.1%.[81] A later estimate suggests that the arrest risk per cocaine transaction is about 1 in 15,000.[82] 

Can a large drug market be made small—and thus easier to control? In theory, massively increasing enforcement through arrests could have this effect. But doing so is costly, often politically unfeasible, and not an optimal use of police resources.

Alternatively, police could engage in a “crackdown”—a burst of police resources, targeting as many dealers as possible simultaneously, to shrink the market down to a more manageable size. Rather than continuously dedicating more cops to marginally increase the arrest risk in a market, a crackdown dedicates many more cops to dramatically increase arrest risk for a short period of time. If successful, the market is substantially shrunk—meaning that following the crackdown, far fewer arrests are necessary to impose the same arrest risk. If routine drug arrests are reactive and tactical, crackdowns are proactive and strategic—they try not to raise the risks of dealing, but cripple the market altogether.

There is a long-running empirical literature supporting the efficacy of crackdowns in suppressing drug-market activity. The earliest is probably the 1983 crackdown on the heroin market in Lynn, Massachusetts, where the Essex County district attorney focused his detail of the state narcotics unit. The detailed unit did not strike all at once but focused its resources, targeting known dealing corners and stepping up arrests. In its first 10 months, the Lynn task force made 140 arrests, 5%–10% of all arrest activity in Lynn in that period.[83]

This crackdown seemed to yield results. Crime dropped precipitously. In the year following the crackdown, robberies fell by 18.5% over the previous period, burglaries by 37.5%, and crimes against the person by 66.7%. The public reported greater satisfaction with police antidrug efforts. Qualitative interviews with drug-treatment counselors and users in treatment suggested that heroin users moved into treatment in response.[84]

A similar early success was obtained in a controlled analysis of a crackdown in Jersey City, New Jersey. In this study, drug-crime “hot spots” were randomized to receive routine enforcement (the “control” strategy) or an approach that combined engaging local community members, initiating crackdowns—lasting a few hours or days—and then assigning officers for a weeklong follow-up period. Comparing calls for service in treated and untreated areas before and after the intervention, the authors found that the crackdown strategy significantly reduced disorder (but not violence or property calls). In the areas around the hot spots, they find a diffusion of benefits, with a particularly large decrease in narcotics calls.[85] In short, the Jersey City strategy seems to have suppressed drug-dealing activity, even if it did not reduce other kinds of crime.

A third variation was first tried in High Point, North Carolina, in 2003, where four “overt” drug markets were targeted with a novel crackdown strategy over three years. First, police assembled a list of all the dealers in the targeted market. These lists were surprisingly short—between 16 and 32 people. Police then built extensive cases against each of these individuals. Finally, dealers were called in, presented with the evidence against them, and told that they could either get out of the market—with the support of services provided by the city—or go to prison.[86]

Almost immediately, the drug markets collapsed. In the city’s West End, narcotics officers went from making “multiple purchases from 11 people at 17 locations” to being unable to make even a single purchase in 16 attempts. “Street corner and drug house activity, drive-through buyers, and prostitutes were simply not in evidence.” Crime also declined dramatically, with violent crime dropping 75% in the 100 days after implementation.[87]

The High Point strategy, today called a Drug Market Intervention (DMI), has been successfully replicated in several other jurisdictions. A DMI was found to have “a statistically significant and substantive reduction in crime, drug, and nuisance offenses in the target neighborhood.”[88] A DMI in Nashville saw significant and large declines in property and narcotics offenses, as well as in calls for police assistance.[89] In another analysis, the researchers recalculated the impact of the High Point intervention using a synthetic control methodology. They found that previous evaluations using lower-quality methods had understated the effectiveness of the High Point bust.[90]

Crackdowns are not always successful. While the Lynn crackdown was effective, a larger market in Lawrence, Massachusetts, was not successfully suppressed by a similar strategy.[91] A DMI in Peoria, Illinois, also failed.[92] More alarmingly, in a seven-site replication study of the DMI strategy, only four sites actually conducted any call-ins, and only one actually reduced crime.[93] Crackdowns are also not cheap: one estimate pegged the cost of a DMI at $100,000–$150,000 (in 2011 dollars).[94] 

Two inferences should be drawn from these caveats. One is that the success of a crackdown is probably highly contingent on the size and other characteristics of a market. There need to be few enough dealers in the market so that simultaneous arrest or call-in is feasible—otherwise, the effect on arrest risk will not be significant enough to affect the market. The other inference is that success depends on the ability of officers to carry out a coordinated exercise. Reactive drug policing is easy: see a dealer, lock him up. Crackdowns require planning, surveillance, interagency coordination, and coordination with community stakeholders. In other words, they are easy to get wrong.

That said, on a theory level, crackdowns offer an effective alternative to the reactive model. And on an empirical level, there is strong reason to believe that they can at least suppress “flagrant” drug markets and might make drugs harder to obtain more generally. They are almost certainly a better use of the marginal officer man-hour than undirected, nonstrategic drug enforcement.

Crackdowns work by making drug markets small, and thus easy to control. But ideally, police can intervene in a market while it is still small and keep it that way. Strategies to do so may be an effective way to minimize the development of new drug threats. It might also minimize the resources needed in the future.

There are reasons to think enforcement in emerging markets might be particularly effective.[95] One is the simple logic of risk ratios articulated earlier: fewer dealers means that the risk per dealer is higher, such that the same level of enforcement can have a bigger impact. Another reason is that small markets can be fragile, often depending on a single source of supply that, if interrupted, can cut off the whole market. For example, enforcement probably successfully suppressed the market for fentanyl in the 1990s, when it was a relatively rare substance restricted to local markets.[96]

The idea of intervening in small markets might be counterintuitive. After all, if a market is small, it is less harmful than a big market. Shouldn’t we prioritize enforcement resources for the most harmful and biggest markets? In some sense, yes. But enforcement should be prioritized not just based on how bad the problem is but on how much impact enforcement can have per effort expended. Furthermore, it is arguably better—in public policy generally—to solve a problem before it becomes big, rather than to let it become big and then attempt to solve it.

To motivate this view, consider the case of xylazine, which, in combination with fentanyl, the Biden administration recently designated as an emerging drug threat.[97] Xylazine is conventionally used as an animal tranquilizer.[98] But it is also abused in combination with opioids—it has been a problem in Philadelphia’s Kensington drug market for some years, and in Puerto Rico even longer.[99] Part of the appeal of xylazine in combination with fentanyl is that the high induced by the latter is strong but fast; adding a tranquilizer helps extend the high.[100] On the other hand, xylazine is not an opioid, so it does not respond to naloxone.[101] It also can cause necrosis—the rotting off of skin, with wounds that don’t heal and require amputation.[102]

Xylazine is not yet prevalent in many U.S. drug markets. June 2022 data suggest that it is moving east to west (much like fentanyl did).[103] That means that many markets that have undergone the fentanyl transition, especially those on the West Coast, now have the opportunity to deter the spread of xylazine before it hits their communities. To return to the theme of this section, if xylazine-adulterated fentanyl becomes widely available—as it has in Philadelphia—it will become much harder to suppress it than if enforcement efforts are directed at it on the front end.

No evidence-based interventions are specifically designed to target emerging markets, although, in theory, crackdown strategies should apply. Other tactics, though, should supplement effective enforcement against emerging markets.

One is effective drug surveillance. This is a particular problem in the case of emerging drug threats—by definition, they are cases where enforcers do not know what they are looking for. Federal and academic guidance can be helpful; but fundamentally, it is the responsibility of drug-enforcement officials to be aware of what problems could be emerging in their communities.

Surveillance can take many forms. Death records are a reliable but trailing indicator—a postmortem toxicology screen will pick up a new drug, but by the time people are dying from the drug, it might already be widespread. Screening arrestees might provide more timely information, as might coordinating with treatment-service providers receiving public funds. Also helpful might be coordination with frontline public-health workers and drug users themselves, both of whom are particularly informed about when the supply changes. Wastewater surveillance, used across Europe and in Australia, is a cheap and efficient way to track emerging drug threats.[104] There is no reason to prefer one of these methods over another—contingent on available resources, a unit monitoring emerging drug threats should incorporate every piece of information it can.

What to do when an emerging drug threat is identified? Identifying and arresting distributors is paramount but challenging. An additional proactive strategy might be to publicly communicate that enforcement resources will be focused on suppressing the emerging substance. Issuing a credible threat to dealers—that the police will be especially focused on the emerging threat, and that those found selling it will face the full extent of the law—might help deter the growth of the novel substance’s market share, in turn keeping the market small and easier to control.[105]

The effectiveness of such strategies is not well studied. More research is needed in how to identify and control small markets. But in theory, targeting them might do more to reduce total drug harm than attempting to take down a large and well-entrenched market.

Police officers are not merely enforcers; they are also first responders. This is true in the case of drug overdoses: in one study of police officers’ reports from New York, officers arrived at the scene of a suspected overdose before emergency medical services 86% of the time.[106] Consequently, effectively policing the new drug crisis requires officers to see their job as having a public-health function in addition to its supply-reduction function. Officers’ goal should not just be to “get the bad guys” but to save users’ lives and get them help when possible.

The most straightforward form that this takes is training officers in administering the opioid-overdose-reversing drug naloxone and equipping them with it in the field. Easy to carry and cheap—a single kit costs $22–$60—naloxone is a low-effort commitment for a department.[107] One survey finds police officers broadly supportive of the practice, too.[108] Consequently, it is already a widely taken step. In a recent national survey of police departments, 82% reported that their officers carry naloxone.[109]

Naloxone administration, however, is of only limited efficacy in improving public health. The balance of the evidence shows no effect of naloxone access, as measured by laws expanding its availability, on overdose death rates.[110] This is because reversing an overdose does not address the underlying causes of the overdose: the use, often compulsive, of deadly drugs. To be sure, reversing overdose is good in itself, especially because even nonfatal opioid overdose can still cause a range of lasting medical complications.[111] But just carrying naloxone is not nearly enough where public health is concerned. What is needed is a strategy that goes beyond overdose reversal.

Specifically, effective public-health policing entails officers seeing themselves as playing an active role in the treatment pipeline. Insofar as they are often present at moments of personal crisis for people dealing with addiction—overdose, accidents, domestic conflict, etc.—police have the opportunity to encourage those same people to make positive changes in their lives. Yet, while many police officers carry naloxone, only about 25% of agencies reported that they follow up after an overdose.[112] In other words, many agencies are not taking advantage of the opportunity to help drug users improve their lives.

One model for this sort of follow-up is a “Quick Response Team” (QRT), also sometimes called a “Post-Overdose Response Team” (PORT). The first QRT was probably set up in Colerain, Ohio, a suburb of Cincinnati, in 2015. Teams consist of an officer, an “addiction expert,” and a firefighter or paramedic. The team makes a follow-up visit several days after someone has overdosed, talking with him or her about substance use and offering access to treatment. About 75% of QRT visits result in face-to-face contact, with 80% of those accepting treatment (although what that treatment consists of is unclear).[113] Since 2015, the intervention has been replicated across Ohio and the country.[114] 

Evaluations generally offer suggestive evidence that QRTs or QRT-like models improve outcomes for people dealing with addiction. A study that compared Massachusetts municipalities with and without QRTs found that overdose deaths grew significantly more slowly in municipalities that implemented the programs.[115] A separate QRT-like program, which involved referring overdose survivors from emergency rooms to follow-up treatment, resulted in roughly a third of referrals accepting some treatment.[116] Neither analysis represents a true randomized evaluation, however, and jurisdictions that do pursue the QRT model should formally evaluate their implementations.

Another model for public-health policing is “law-enforcement assisted diversion,” or LEAD. First piloted in Seattle in 2011, LEAD is a diversion program that offers people arrested for low-level drug crimes the opportunity to enter treatment instead of being booked and charged for their offense. The specifics of eligible offenses and the kind of treatment—from “low-barrier,” harm-reduction-focused counseling to intensive treatment—can vary. But in general, LEAD interventions work by offering people an opportunity to enter treatment, rather than the criminal-justice system.

LEAD interventions have been evaluated across several jurisdictions, and the results are generally supportive. In a non-randomized, propensity-score-matched study, Seattle’s LEAD program reduced rearrest risk by 60% within six months, and by 58% over the longer term.[117] It also reduced the risk of felony charges.[118] Participants in Madison, Wisconsin’s diversion program, MARI, were marginally significantly (p = 0.11 in an N = 315 study) less likely to be rearrested within six months, especially if they completed the mandated treatment.[119] At the 12-month follow-up, those who completed treatment were significantly less likely to be arrested or to have died from overdose than those who did not participate or who did not complete treatment.[120] An evaluation of San Francisco’s LEAD implementation found that, compared with a propensity-score-matched control group, LEAD participants were significantly less likely to be rearrested for felonies or misdemeanors.[121] And in a multisite study in North Carolina, LEAD exposure and participation significantly reduced arrest and incarceration relative to control.[122]

None of these studies adequately overcomes problems of selection bias—people who opt for LEAD likely differ from those who opt against it, and people who complete LEAD-mandated treatment differ from those who drop out, in ways correlated with outcomes. True effects of LEAD are likely much smaller than those reported here. At the same time, the causal model—that treatment reduces re-offense and overdose risk—is ex ante plausible, such that we should be willing to believe that at least some of the effect is causal. Implementing jurisdictions should also consider conducting true randomized controlled trials on their LEAD implementations.

Jurisdictions interested in implementing QRT/LEAD/policing-facilitated treatment diversion approaches must seriously consider certain questions. Do the police have an effective working relationship with treatment providers? Does the jurisdiction have enough treatment capacity? Is the treatment provided evidence-based, medication-assisted, and long-term as necessary? Patients who enter detoxification but then receive no formal aftercare are significantly more likely to relapse following treatment.[123] Is the jurisdiction prepared to provide people with not just a bed for detox but the long-term care needed to overcome addiction severe enough to induce criminal behavior?

Those caveats acknowledged, LEAD and QRT models address concerns raised by stakeholders across the board. They help treat substance use disorder as a medical issue, not a criminal issue. They reduce the burden on police officers (who are not doctors and do not like pretending to be doctors) to address addiction and its consequences. They offer addicts both a carrot (access to treatment and staying out of jail) and a stick (the threat of jail if they do not comply). Most important, these models seem to reduce rates of harmful drug use, which should be everyone’s goal.

Drug enforcement was not always as unpopular as it is today. When then-president George H. W. Bush declared an all-out offensive against drugs in 1989, most Americans supported him.[124] Two decades ago, Americans still listed drug interdiction and arresting dealers as the best strategies for combating the drug problem, above providing treatment and education.[125]

Times have changed—and with them, Americans’ opinion of drug enforcement. No longer is there much appetite for indiscriminately sweeping corners or locking up thousands of drug users. While the claims that critics of the Drug War make are often exaggerated or entirely fabricated, the reality is that the American public is much warier of drug enforcement than it was in the heyday of the War on Drugs.

At the same time, the drug threat is very different from what it used to be. Crack cocaine markets were uniquely violent, and the benefits of shutting them down were most obviously reflected in its effects on violent crime.[126] Contemporary drug markets are substantially less violent—because of professionalization, technology reducing street dealing, and other factors—but they are also much deadlier. And they are much more lethal for the same reasons that they are much harder for crack-era enforcement tactics to address: a radical change in the production of drugs has driven availability and potency sky-high and prices to near-zero.

The solution to these two problems is not to separate police and the drug problem. It is not to decriminalize—a policy proposal that attracts far more attention than its likely impact merits. Rather, the solution is to try to adapt enforcement to the novel issues that it faces.

This report has discussed three approaches: market crackdowns, emerging-market enforcement, and public-health policing. Together, these imply a drug enforcement that is smarter, more strategic, and more focused. That is a drug enforcement that can have a real impact on current problems—before it’s too late.

Charles Fain Lehman is a fellow at the Manhattan Institute, working primarily on the Policing and Public Safety Initiative, and a contributing editor of City Journal. His work has appeared in outlets including The Atlantic, the Wall Street Journal, National Affairs, and National Review. He has discussed public safety policy before the House of Representatives and the U.S. Commission on Civil Rights, and at colleges including Carnegie Mellon and Cornell. He is a 2023–24 Robert Novak fellow with the Fund for American Studies. Prior to joining the Manhattan Institute in 2021, Lehman was a staff writer at the Washington Free Beacon. He is originally from Pittsburgh, and now lives outside Washington, DC, with his wife and sons.

Please see Endnotes in PDF

Photo: kali9/E+ via Getty Images