Report
Controlling Cocaine: Supply Versus Demand Programs
Jan 1, 1994
Each year, the United States spends large sums of money at all levels of government to battle cocaine use. By 1992, this annual sum had reached $13 billion.[1] Almost three-quarters of that was for domestic enforcement, most of the remainder for controlling the amount of cocaine coming into the United States, and about 7 percent for treatment. Is this the most effective way to spend the money? The answer is important, given the level of expenditures involved and the destructive effects of cocaine on individuals, families, communities, and society.
Unfortunately, arriving at an answer is also difficult, because it requires a quantitative understanding of the cocaine epidemic. This in turn entails understanding how many people start and stop using cocaine each year and how users' consumption patterns change. It also requires understanding how those trends are affected by individual choice, by the market for cocaine, and by various government strategies to control the use of cocaine.
One approach to achieving such an understanding is through integrative mathematical modeling. This is the approach taken in a recent DPRC study that provides a quantitative foundation for decisions on allocating cocaine control resources. On the basis of that analysis, it was possible to conclude that treatment is seven times as cost-effective (at the margin) as domestic enforcement—the best alternative to treatment that the study evaluated. Thus, if it were possible to vary the current allocation of funds over control strategies, cocaine consumption could be reduced substantially without increasing total spending.
Before the effects of different control strategies could be analyzed, it was necessary to make sense of some confusing data regarding cocaine use. In the late 1980s, the most frequently relied-upon indicators of cocaine consumption exhibited contradictory trends. Surveys of households and high school seniors showed steady and substantial declines in the number of cocaine users after about 1985. However, the number of times cocaine was mentioned by medical examiners and the number of cocaine-related emergency-room episodes were rising steadily. Eventually it was recognized that different users consume very different quantities of the drug, and even though the total number of users was decreasing, the minority of heavy users was growing. The relative contributions of light and heavy users to total cocaine consumption remained unknown, however, as did the implications for future control efforts, until DPRC researchers Susan Everingham and Peter Rydell built and ran their quantitative model of cocaine consumption.
Everingham and Rydell found that while the number of cocaine users has fallen since the mid-1980s, total consumption has only leveled off. The drop in consumption expected from fewer users was offset by an increase in average annual consumption per user, as reflected in a rising percentage of total consumption represented by heavy users—from 40 percent in 1980 to about 70 percent in 1991 (see Figure 1). This trend is important for two reasons: First, although many light users are able to remain employed, finance their habit with legitimate income, and avoid severe adverse health consequences, heavy users generally present serious long-term social problems. Second, cocaine dealers' revenues are affected by the total amount of cocaine sold, not the number of users. If those revenues fall, the incentive for engaging in turf battles and other violent drug-related activities could also fall. The failure of consumption to fall along with the number of users is thus not a good omen for public safety.
The model also revealed implications for the difficulty of reducing cocaine consumption in the future. The future interval examined was the 15-year period following 1991—the latest year for which data were available at the time of the analysis. The researchers' estimates are based on assumptions about future recruitment of drug users. Predicting future initiation is uncertain at best, so the researchers chose instead to assess the consequences of several scenarios bounding the possibilities.
First, they assumed that the number of new users—those initiating cocaine use—would remain at its early-1990s level of about 1 million persons per year. They estimated that, in that case, total annual cocaine consumption would go up 20 percent by the year 2007 (see Figure 2). This scenario sets a sort of upper bound and suggests that preventive programs aimed at restraining initiation are very important.
An alternative scenario is a continuation of the current decline in initiation to about a half million new users per year in 2007. In that scenario, total annual consumption would hardly be affected. Even if the number of new users decreased to zero by 2007, consumption would fall by only 20 percent. Thus, if a big reduction in cocaine use is desired, cutting the number of new users, though necessary, is not enough. Programs that reduce new use must be supplemented by others that encourage heavy users to quit, or at least reduce their consumption. The reason for this is that, untreated, heavy users quit at a low rate, so that they stay in the cocaine-using population for a long time. Thus, even if it turned out that no one took up cocaine use over the 15-year period modeled, consumption in 2007 by those who have already started would still be about half what it was in 1991. (Decreasing initiation would cut the number of users by substantially greater proportions than it would cut consumption.)
The research on trends suggests the importance of programs aimed at reducing consumption. But how effective are such programs? Rydell and Everingham analyzed the relative effectiveness of four kinds of cocaine control programs:
To aid in the analysis, Rydell and Everingham constructed an economic model of cocaine supply and demand, including the effects of control programs. The model is quite flexible and can be used to answer various questions. For example, the researchers compared the four types of programs on the basis of how much more would have to be spent for each in the current year to reduce current and future consumption by 1 percent. The first three kinds of programs raise the cost to smugglers or wholesalers of supplying cocaine. Wholesalers react to increased costs by raising prices. Those price increases are passed on to the retail level, where they discourage both heavier consumption by current users and future initiation.
The programs' cost-effectiveness was estimated as follows. First, it was necessary to establish the relationship between a price increase and a consumption decrease. On the basis of what is known about such relationships for alcohol and tobacco and some preliminary data for cocaine, the researchers estimated that cocaine consumption decreases by 0.5 percent for every 1 percent increase in price (i.e., the price elasticity of demand is 0.5). Thus, a 2 percent price increase is required to achieve the 1 percent consumption drop set as the criterion for comparison. Since U.S. consumers spend about $38 billion annually on cocaine, a $750 million increase in the total cost to consumers would be required. The analysis suggests that a $250 million increase in expenditures on domestic enforcement would be enough to achieve such a price increase. The reasoning behind the analysis of interdiction and source-country control is similar.
Treatment reduces consumption in two ways: Most of those being treated are not using cocaine, and some treated consumers do not return to cocaine use after treatment. Here, achieving a 1 percent reduction means decreasing the 300 metric tons annually consumed in the United States by 3 tons. By how much does the typical course of treatment reduce a heavy user's cocaine consumption? This amount varies widely. Only a small fraction of heavy users stay off cocaine permanently after treatment, but each such success amounts to hundreds of grams. During the treatment period itself, the average heavy user's consumption is reduced by about 30 grams. Overall, in light of behavior both during and after treatment, putting one heavy user through treatment can be expected to reduce total cocaine consumption by 160 grams, on average. Approximately 19,000 such treatments would be required to reduce cocaine consumption by 3 tons. At an average cost of $17,000 per treatment, a total of $34 million in treatment costs would be required to achieve the 1 percent consumption drop.
Treatment is thus far more cost-effective than domestic enforcement. The gap between treatment and the other supply control approaches is even greater (see Figure 3); the latter are less efficacious than domestic enforcement for various reasons, e.g., because it is cheaper for cocaine producers and distributors to replace product and other assets lost further "upstream" from the retail point.[2]
Obviously, the specific numbers are sensitive to input parameters such as the relationship between cocaine price increase and consumption decrease, and the effect of treatment on subsequent behavior. However, for domestic enforcement to be more cost-effective than treatment, the values of both of these parameters would have to vary greatly from those estimated (see Figure 4). For example, even if elasticity of demand were three times what the DPRC researchers estimated, domestic enforcement would be more cost-effective only if the fraction of treated heavy users permanently quitting were less than one-third of that estimated.
The model has also been used to examine what could be gained (or lost) by increasing or cutting cocaine control funds or shifting them from enforcement (or other supply control measures) to treatment or vice versa. For example, if supply control were cut by 25 percent, $3 billion per year could be saved, while cocaine use would increase by 8 percent (see Figure 5). If one-third of the savings were spent on treatment programs, the current number of users treated could be doubled, more than compensating for the 8 percent increase in use. If all $3 billion were diverted to treatment programs, all heavy users could be treated and total cocaine consumption in the United States could be cut by one-third. Restoring the supply control funds would result in an additional reduction in cocaine use.
This does not, however, mean that all money now spent on supply control could be more effectively spent on treatment. There are limits to how many heavy users can be induced to undergo treatment and how often they can be induced to do so. As shown in Figure 5, even if all heavy users could be treated once each year (which is unlikely), the reduction in cocaine consumption could not be more than one-third. Furthermore, because of diminishing returns, each successive dollar shifted to treatment would be less effective than the previous one. On the other hand, each successive dollar shifted out of enforcement would have been more effective if left in enforcement than the previous one. Also, the specter of enforcement no doubt induces some users to enter treatment. Thus, the DPRC research by no means suggests that deep cuts in enforcement are advisable. It does suggest, however, that if additional money were available for cocaine control, it would be better spent on treating heavy users than on expanding supply control programs.
This report is part of the RAND Corporation Research brief series. RAND research briefs present policy-oriented summaries of individual published, peer-reviewed documents or of a body of published work.
This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.
The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.