What America's Users Spend on Illegal Drugs 1988–1998
December 2000
1. Consumption-based Estimate of Retail Expenditures
Cocaine and Heroin
Between 1989 and 1998, American users spent $39 billion to $77 billion yearly on cocaine and $10 billion to $22 billion yearly on heroin. To arrive at these estimates, we multiplied the number of users by their typical expenditures, and then converted the resulting estimates to 1998-dollar equivalents. Most of the downward trend results from changes in the consumer price index.
The Number of Cocaine and Heroin Users
The National Household Survey on Drug Abuse (NHSDA), the Nation's most comprehensive survey of drug use, measures drug use among the American household population age 12 and older, as well as among people living in group quarters and the homeless.3 The NHSDA misses a part of the population that may be a key to determining the extent of drug use: those hardcore drug users who, although not homeless, are too unstable to be considered as part of a household, or who, if part of the household, are unlikely to answer surveys.4
This less-stable population of hardcore drug users is, however, well-represented in data collected by the Drug Use Forecasting (DUF) program, which questions a sample of arrestees in 24 central city jails and lockups about their drug use.5 DUF also asks arrestees to voluntarily produce specimens for urinalysis. This helps to confirm whether the interviewees have used any of up to 10 types of drugs during the two to three days before the interview. Although urinalysis is subject to error and tells us nothing about the frequency of drug use, it adds credence to estimates of drug use when self-reports are unreliable.
The hardcore user is identified in the NHSDA as one who used cocaine at least one or two days a week every week during the year before the survey, or one who used heroin on more than 10 days during the month before the survey. In this analysis, hardcore users in the DUF data are defined as those who admitted using cocaine or heroin on more than 10 days during the month before being arrested.6 Occasional users are identified in the NHSDA as those whose drug use was less frequent than the hardcore drug use criteria described above. Occasional use cannot be estimated from DUF.7
Appendix A explains how we used data from the NHSDA and DUF, as well as other sources, to estimate the number of drug users in the United States. The rest of this section provides an overview and reports findings. According to one estimate, hardcore drug users seem to account for about three-quarters8 of all cocaine used in the United States, so understanding hardcore consumption patterns is crucial to estimating expenditures on cocaine. The concentration of heroin consumption is probably similar. Thus, estimating hardcore drug use is especially important. The calculations start by estimating the number of hardcore users who are arrested during the year. This number is then divided by the average number of arrests that hardcore users generate during the year. For example, if hardcore users account for 2 million arrests per year, and if hardcore users are arrested an average of 0.5 times per year, then there must be 2 million divided by 0.5, or 4 million, hardcore users in the nation. We then subtract estimates of hardcore users in jails and prisons, because they are unlikely to use heroin or cocaine heavily while incarcerated. The trick, of course, is to obtain reasonable estimates of both the number of hardcore users who are arrested during each year and the average number of arrests that they generate during the year (see Appendix A).
Once estimates of the number of hardcore users are available, the next step is to estimate how much they spend on cocaine and heroin. The best way to learn this information is to ask the users, and studies sponsored by ONDCP, the National Institute on Drug Abuse, and the National Institute of Justice provide data (see Appendix B). An estimate of the retail sales value of illicit drugs consumed by heavy users follows from multiplying estimates of typical expenditures by estimates of the number of hardcore users.
Estimates of expenditures by hardcore users are then converted to units measured in kilograms of heroin and cocaine, so that amount consumed can be compared with the amount of drugs trafficked into the country. This requires an estimate of the prevailing retail prices for illicit substances. Here, too, ONDCP and other agencies have sponsored research leading to estimates of what substance abusers pay for drugs on the streets (see Appendix C). Dividing the estimate of retail sales value by the prevailing price paid by users gives an estimate of the total amount of drugs purchased, and this amount can be converted readily into metric ton units.9
This explains the derivation of estimates of drugs used by hardcore users, but while hardcore users probably account for at least three-quarters of the cocaine and heroin used in this country, they do not account for all illicit drug consumption. One view is that the National Household Survey on Drug Abuse under-states the number of hardcore drug users and the amount that they spend, but that the NHSDA provides a reasonably accurate estimate of the amount of more casual drug use. Thus, this report comple-ments expenditures by hardcore users on cocaine and heroin based on DUF data with expenditures on these substances by more casual users based on the NHSDA.
This report provides preliminary estimates of methamphetamine use, based mostly on DUF data, and using estimation procedures similar to those used to estimate cocaine and heroin use. Finally, estimates for marijuana use and for other illicit drugs (excluding cocaine, heroin, marijuana, and methamphetamine) come from the NHSDA, with some adjustments for under reporting.
Table 3 provides estimates of the number of hardcore and occasional cocaine and heroin users derived from the NHSDA and the DUF data. (Users of other drugs will be discussed later.) Because the NHSDA was not administered in 1989, the 1989 NHSDA estimates used in this report are the average of 1988 and 1990 data; also, SAMHSA changed the survey in 1994, and statistics from earlier years were adjusted by SAMHSA to take these changes into account. Estimates for 1998 through 2000 are projections based on trends observed in earlier years.10
Excluding persons in custody, between 1988 and 1998, about 3.2 million to 3.9 million Americans were hardcore users of cocaine and approximately 2.9 million to 6.0 million were occasional users. Another 630,000 to 980,000 Americans were hardcore users of heroin, and 140,000 to 600,000 were occasional users. Considering the overlap between hardcore cocaine users and hardcore heroin users, the estimates suggest that there were about 3.3 million hardcore users of heroin or cocaine in 1998.11 Although imprecise, these estimates are consistent with reported estimates derived by others using different methodologies and data.
For example, Rhodes, Langenbahn, Kling and Scheiman12 provided one national estimate of 508,000 hardcore heroin users, and a second national estimate of 582,000 hardcore heroin users. The authors explain why both estimates probably understate the true number. We are aware of only one other national estimate of heroin addicts, by Hamill and Cooley,13 who concluded there were 640,000 to 1.1 million heroin addicts in 1987. These estimates are roughly consistent with our 1988 estimate of 920,000 hardcore heroin users.
Simeone, Rhodes and Hunt14 estimated that there were about 300,000 hardcore cocaine/heroin users in Cook County in 1995. Assuming a constant proportionality between the number of hardcore users in a population and the number of emergency room admissions attributed to them, an extension of the Simeone, Rhodes and Hunt estimates suggest there are about 4.0 to 4.5 million hardcore users in the nation. Although such an assumption of proportionality rests on shaky grounds, it nevertheless leads to estimates of a magnitude remarkably close to the 3.3 million estimate used in retail sales calculations.
The Substance Abuse Mental Health Services Administration estimated that about 3.6 million Americans have a severe need for substance abuse treatment exclusive of treatment for alcohol abuse.15 SAMHSA derived this estimate by identifying someone as needing treatment if he met one of four criteria and then inflating the estimates to account for undercounting in the NHSDA.16 Because the inflation factor is only 20 to 30 percent, it seems likely that SAMHSA's estimates of the number of cocaine and heroin users who need treatment would be smaller than the estimates given here for weekly heroin and cocaine users. SAMHSA does not report the need for treatment by type of drug, but we applied the SAMHSA algorithm to the NHSDA data as best we could and inflated the resulting estimate by 25 percent.17 The result was that 920 thousand cocaine users needed treatment, as did 130 thousand heroin users and 59 thousand people who used both heroin and cocaine. Thus, SAMHSA estimated that almost 1.2 million people need treatment for cocaine abuse, and almost 190,000 need treatment for heroin addiction.
Not all weekly users of cocaine need treatment, so an estimate of 3.4 million weekly users (1996) may conceivably be consistent with SAMHSA's estimate of 1.2 million who need treatment. Similarly, weekly heroin use may not indicate a need for treatment, so an estimate of 190 thousand heroin addicts could conceivably be consistent with our estimate of 900 thousand weekly heroin users. Although conceivable, these differences are so large that they tax credulity. There are three problems. The first is that, from the view of our calculations, a 20 to 30 percent inflation factor is insufficient to approximate the number of hardcore users not represented by the NHSDA. A second problem is that the SAMHSA estimates suggest that at a maximum, about 25 percent of all people who need treatment for substance abuse are current users of heroin or cocaine. In fact, all 17 CEWG (Community Epidemiological Work Group) sites18 report more than 25 percent of their treatment admissions are for cocaine or heroin, and 11 of 17 report that more than half their admissions are for cocaine or heroin. Although not all people who need treatment actually receive treatment, we would expect a closer correspondence between those who need treatment for cocaine and heroin, and those who receive treatment for those substances. Third, according to the Treatment Episode Data Sets (TEDS), roughly 200,000 heroin users and another 250,000 cocaine users received treatment per year between 1993 and 1997.19 SAMHSA's estimates are inconsistent with TEDS. Thus, even after attempts to inflate estimates based on the NHSDA, the estimates seem to understate the number of hardcore heroin and cocaine users, and consequently, the SAMHSA estimates cannot be reconciled with our estimates.
Table 3—Estimated Number of Hardcore and Occasional Users of Cocaine and Heroin (Thousands), 1988–2000
Columns may not add due to rounding
Sources: NHSDA 1988, 1990 through 1998; DUF 1988 through 1998; Uniform Crime Reports (UCR) 1988 through 1997.
1 The NHSDA was not administered in 1989. Estimates are the averages for 1988 and 1990.
2 Due to sample overlap, the estimated number of composite hardcore cocaine users is derived from the sum of DUF hardcore cocaine users and one half of NHSDA hardcore cocaine users.
Trends in Drug Use
If the prevalence estimates have some justification, what can be said about trends? Because the estimates presented in Table 3 are based on a consistent methodology from 1988 through 1997, they can be compared meaningfully from year to year. We do not know the standard errors for these estimates, however, so we lack a probability basis for judging whether or not changes are statistically significant. Our estimates seem to show a decrease in the number of hardcore cocaine users from 1988 to 1991. Thereafter, the estimated number of hardcore cocaine users fluctuates from year to year but follows no strong trend. Estimates of occasional use from the NHSDA show a consistent downward trend. Table 3 shows a decrease and then an increase in hardcore heroin use. This recent increase in hardcore heroin use has a counterpart in the NHSDA, which also reports a recent increase in heroin use among household members.
Because trends in drug use are often disputed, it may be helpful to discuss whether or not other evidence is consistent with our findings. Hardcore drug users are frequently in trouble with the law, so a temporal change in incarceration practices will necessarily have a large effect on them. Based on estimates explained in Appendix A, the increase in prison populations between 1988 and 1998 would have incapacitated an additional 200,000 hardcore cocaine users and an additional 72,000 hardcore heroin users. These are sizable yet conservative numbers, because they do not take into account inmates and detainees under the supervision of local correctional authorities.
The AIDS epidemic provides another reason for expecting a decrease in heavy drug use, especially by heroin users, but also for others who inject drugs. According to the Centers for Disease Control20 217,000 injection drug users had been diagnosed with AIDS as of 1998, and 87,000 had died of the disease. Having AIDS does not preclude substance abuse, of course, but advanced AIDS must make it all but impossible to support heavy use of heroin. Adding together hardcore heroin users who are incarcerated and hardcore heroin users who have died implies about 150,000 fewer hardcore heroin users at the end of the decade than at the beginning of the decade. The figure may be closer to 200,000 when we consider heroin users with advanced AIDS.
If no other factors affected hardcore drug use, we would expect a decline in hardcore cocaine users and, especially, hardcore heroin users, from 1988 to 1998. Offsetting these trends toward less use, however, is an apparent recent increase in heroin use by people who do not inject. This might result from the increased availability of higher purity heroin. Trends reported by SAMHSA in the 1998 Treatment Episode Data Set (Table 5.3) are consistent. Between 1993 and 1998, the proportion of admissions for heroin inhalation increased from 23 percent to 28 percent. Moreover, those admitted for heroin inhalation tend to be younger than those admitted for heroin injection; they are more likely to be experiencing a first treatment episode; and among heroin abusers experiencing a first treatment episode, those who inhale have typically used for a shorter time. Recent tabulations based on the National Household Survey on Drug Abuse and the Monitoring the Future Survey have suggested renewed drug use by youths.21 Nevertheless, this increase is a relatively recent phenomenon, and it followed a decrease in earlier years. It is difficult to believe that these youth could have progressed to heavy use as of 1998, and certainly they could not account for much of the increase in treatment episodes for heroin—where fewer than 5 percent of patients are under twenty years old.22
Finally, according to the Substance Abuse and Mental Health Services Administration, emergency room mentions for cocaine use have increased from about 80,000 in 1990 to about 161,000 in 1997. Emergency room mentions for heroin grew from about 34,000 in 1990 to 72,000 in 1997. A naïve observer might infer that cocaine and heroin use doubled between 1990 and 1997, but this is almost certainly wrong. Little is known about the dynamics of emergency room use by hardcore cocaine and heroin users, but some speculation might be helpful. According to the 1997 DAWN (Drug Awareness Warning Network) report, dependence is the dominant drug use motive for heroin and cocaine users seeking emergency room assistance—86 percent for heroin mentions and 68 percent for cocaine mentions. Either chronic effects, withdrawal or seeking detoxification are the typical reasons for going to the emergency room—62 percent for heroin mentions and 50 percent for cocaine mentions.23 Addicts are more likely to seek treatment as they age, and treatment episodes seem to become more frequent over time.24 For this reason alone, we would expect to see emergency room mentions increase even if the number of hardcore heroin and cocaine users did not change. Furthermore, we suspect that hardcore heroin and cocaine users will develop an increasing number of chronic health conditions as their addictions advance and as they age. This, too, can account for an increase in emergency room mentions. While DAWN can be very valuable for detecting short-term changes in specific jurisdictions—such as a spike in overdose deaths—it would seem to have little or no value as a tool for monitoring long-term trends in the prevalence of substance abuse.
Average Amount Spent on Cocaine and Heroin
DUF interviews from 1989 and later asked respondents how much they spent on drugs during a week. The question did not separate cocaine from heroin spending or exclude other drugs, so we must infer how much was spent on cocaine and how much was spent on heroin. Also, some respondents gave answers that were implausibly large, so based on the methodology explained in Appendix B, we adjusted estimates to moderate the effect of extreme values. Because of a change in questionnaire design, DUF does not provide comparable estimates after 1995. Estimates for 1996–2000 are just the 1995 estimates adjusted for inflation.
Table 4 provides estimates of the median expenditure on cocaine and heroin. Based on evidence presented in Appendix B, using the median expenditure in retail sales calculations has a greater justification than using a mean expenditure. All estimates were converted to 1998 dollar equivalents based on the consumer price index.25
In 1998, hardcore cocaine users spent $191 a week on cocaine, and hardcore heroin users spent $214 a week on heroin (Table 4). These DUF estimates lack precision, but they are reasonable considering other data about expenditures on illicit drugs. For example, an analysis of data from a special addendum26 to the 1998 DUF instrument in 1995 gives some information for the heroin numbers.27 Based on the median, hardcore heroin users spent $140 per week; based on the mean, they spent $330 per week. The mean is probably too high, because it likely includes purchases by some users who intend to resell part of the lot.28 Appendix B provides a review of expenditure patterns reported by other researchers.
Table 4—Weekly Median Cocaine and Heroin Expenditures Reported by Arrestee Hardcore Users, 1989–2000
Sources: DUF 1989 through 1994
Of course, occasional users spend less per week than do hardcore users. Based on NHSDA data, occasional cocaine users spent $19 per week in 1988, $23 in 1989, $27 in 1990, $30 in 1991, $34 in 1992, and $35 in 1993. More recent estimates are unavailable. No such estimates are available from the NHSDA for occasional heroin users. For them, we assumed a weekly expenditure of $50 per week.
Total Expenditures on Cocaine and Heroin
Between 1988 and 1998 American users spent $39 billion to $77 billion yearly on cocaine and $10 billion to $22 billion yearly on heroin (Table 5). We derived these estimates by multiplying the number of hardcore and occasional users in Table 3 by the median expenditures in Table 4 (and the figures cited earlier for occasional users) and adding the results.
How the Estimates are Affected by Varying the Assumptions
The estimates of expenditures may vary due to assumptions made about the number of hardcore and occasional users and about their average expenditures. Because hardcore users account for the bulk of drug spending, estimates of total expenditures are especially sensitive to the accuracy of estimates of expenditures by hardcore users. Consequently, we tested how sensitive our expenditure estimates are to assumptions made about the number of hardcore users and their typical expenditures. Because the factors that entered the calculations were not derived from probability samples, it is impractical to develop a statistically based margin of error.
First, we determined how the expenditure estimates would be affected if we used lower or higher estimates of the number of users than were reported in Table 3. Because the retail sales estimates are roughly propor-tional to the number of hardcore users, if the estimate of hardcore users is off by plus or minus 25 percent, then the retail sales estimates would be off by the same proportion.
Second, we determined how the expenditure estimates would be affected if we varied our assumption about typical drug expenditures. Some studies reported in Appendix B are based on reported expenditures by cocaine users entering treatment, and those users have much higher expenditure patterns than are assumed in the retail sales calculations. If these expenditures were considered typical, the retail sales value of cocaine would be two to four times the amount reported here. This seems an implausibly large expenditure that would exceed not only available income for most users,29 but the value of the supply of the drugs as well. (For a further discussion of this topic, see Appendix B.)
Although an average expenditure figure based on a treatment population is certainly too high, it might be realistic to adopt the average (rather than the median) drug spending numbers reported by DUF as a high estimate. Then, the composite totals on both cocaine and heroin use would be 60 to 80 percent greater than estimates based on the median expenditure patterns. For the reasons we cited above, it is doubtful that expenditures in the United States approach this high estimate.
At the opposite extreme, hardcore users who report their use in the NHSDA appear to consume less than half as much cocaine as hardcore users represented in the DUF data. Their expenditures might be considered a low estimate of typical cocaine spending by hardcore users. Giving more weight to the NHSDA expenditure figures would reduce the amount reported in Table 5 by half. However, it is difficult to reconcile estimates that are half as large with the amount of heroin and cocaine that enters the country.
Other analysts have made clever use of available data to derive their own estimates of retail expenditures on cocaine and heroin. Even after adjusting for the limitations of these other studies, our estimates are higher than theirs, perhaps suggesting that—if anything—we might adjust our estimates downward.30 But, for reasons noted above, a large downward adjustment seems unwarranted.
Table 5—Total Expenditures on Cocaine and Heroin, 1988–2000 ($ in billions, 1998 dollar equivalents)
Since weekly expenditures from DUF data were not available for 1988, we used the 1989 amounts as proxies for 1988 in calculating total expenditures.
Accounting for Income in Kind
Our expenditure estimates reflect money that actually changed hands at the retail level. But drugs are often obtained as "income in kind," sometimes as payment for serving a role in the distribution chain and sometimes as payment for sex. For reasons explained in Appendix B, we assume that hardcore users of heroin received 22 percent of their drugs as in-kind payment in 1988, but that this percentage fell linearly to 11 percent as of 1995 because of changes in the way that heroin was distributed.31 We assumed that users of cocaine received 11 percent of their cocaine as income in kind throughout the period.
If we monetize in-kind payments at street prices, then the 1998 dollar expenditure on cocaine would increase by about $4 billion, and the 1998 dollar expenditure on heroin would increase by about $1.5 billion. These totals are not reflected in Table 5, but we do take them into account later when we estimate the bulk amounts of cocaine and heroin used in America.
How Much Cocaine and Heroin is Consumed?
To estimate how much cocaine and heroin Americans consume, we used data from the System to Retrieve Drug Evidence (STRIDE) to estimate the street prices paid for cocaine and heroin. These data come from laboratory analyses of purchases by Drug Enforcement Administration agents, other Federal agents, and some State and local agents. The price varies with the size of the purchase lot. Cocaine is much less expensive when bought as a large lot than when purchased as a smaller lot. This is also true of heroin. Therefore, to estimate the average street price of illicit drugs, it is necessary to know how much a typical buyer purchases each time he makes a purchase. The larger the quantity of drugs purchased, the lower the per unit price. There is scant evidence on this topic. Appendix C details our assumptions.
The price of cocaine fell sharply throughout the early 1980s (not reflected in the table), increased during 1990, and then declined again into 1998 (Table 6). Most of the decline after 1990 is caused by an increase in the consumer price index. The price of heroin also fell throughout most of the 1980s and the mid 1990s. It has remained relatively constant as of 1995.
Table 7 shows estimates of the amount of cocaine and heroin that was consumed based on the expenditures reported in Table 7 (adjusted to account for drugs earned as income in kind) and the retail prices reported in Table 6. According to the data for the 1988 to 1998 period, cocaine users consumed somewhere between 270 and 400 metric tons of pure cocaine each year. The level of consumption has stayed close to 300 metric tons throughout the 1990s. Heroin users consumed between 7 and 13 metric tons of pure heroin each year during the same period. Consumption has been close to 13 metric tons during the latter part of the decade.
Because estimates are not totally accurate, trends are uncertain. However, it appears that the amount of cocaine consumed in the United States has changed very little over the last eight years. The estimates are somewhat higher in 1988 and 1989 than in later years, but given the margin of error in these estimates, no strong trend is apparent. Total expenditure on cocaine has fallen over time, but this is attributable almost exclusively to using the consumer price index to inflate past expenditures.32
Trends in heroin use may be different. The amount of heroin used seems to have decreased from 1988 and 1989 into the early 1990s. Thereafter, heroin consumption may have increased. As already noted, there seem to be fewer heroin addicts in the middle 1990s than there were at the end of the 1980s. The HIV virus and AIDS have taken a toll, and many users have been incarcerated. Yet, prices have fallen so much that remaining users have been able to purchase much more than they did in the past, and these lower prices may have attracted new users into the market.33
Other studies provide comparable estimates. Using a much different estimation methodology, Rand researchers estimated that about 451 metric tons of cocaine entered the United States in 1989.34 This compares with our estimates of 394 metric tons. The Rand researchers estimate that 7.8 metric tons of heroin entered the States in 1991.35 Our estimate is 6.8 metric tons.
Table 6—Retail Prices Per Pure Gram for Cocaine and Heroin, 1988–2000 (dollars, 1998 dollar equivalents)
Source: STRIDE 1981 through 1998
Table 7—Total Amount of Cocaine and Heroin Used, 1988–2000 (in metric tons)
Sources: See Tables 3 through 6.
Methamphetamines
We applied the computing algorithms used to derive estimate for cocaine and heroin to the problem of getting estimates for methamphetamines. When applied to methamphetamines, the approach does not work as well, for reasons that are discussed in this section. Nevertheless, the calculations are sufficiently accurate to provide rough measures of the number of heavy users as well as of the scale of expenditures and amount used. Calculations are summarized in Table 8.
According to our calculations, there are probably between 300,000 and 400,000 hardcore users of amphetamines. As before, a hardcore user is someone who uses a drug on more than ten days per month. The estimate is technically about amphetamines, because that is the question posed in the DUF interview. Hereafter, however, amphetamine users are assumed to be methamphetamine users. This assumption is justified by the observation that in 1997, more than 96 percent of those who tested positive for amphetamines were confirmed by a second test to be positive for methamphetamine.
This estimate is tentative for two reasons. The first is that methamphetamine use is rare among arrestees in many cities, so the estimates are really based on the experiences of a few cities, and those experiences are then prorated across the nation. The fact that so few cities account for the estimates may impart additional uncertainty to the calculation. The second reason for skepticism is that the estimates vary markedly from year to year. Most of that year to year variation is hidden in Table 8 because a three-year moving average was applied to smooth the data.
Combining the DUF data from all years, hardcore amphetamine users spend about $90 per week on their use of methamphetamines. The table shows the $90 after adjustment by the consumer price index from 1989 to 2000. Because the sample size is relatively small, we did not attempt to determine a trend in expenditures, but rather, we assumed the $90 estimate applied to all years.
The estimate of total revenue comes from multiplying the number of hardcore users by their weekly expenditure, and then multiplying by 52 to determine a yearly expenditure. The result was multiplied by 4/3 (the reciprocal of 0.75) to account for occasional users. Methamphetamine users currently spend somewhat more than $2 billion per year on methamphetamine use. The next step was to estimate the price of methamphetamine. Appendix C explains the price derivation, and that the price estimate is probably too high or too low over the entire reporting period. It is difficult to know which. The final step is to divide total revenue by the price per pure gram. If casual users account for roughly 25 percent of consumption, the estimate is 9 to 16 metric tons. As noted, seeking precision would be quixotic; these estimates are best treated as matters of scale with a wide (but unknowable) confidence interval.
There is scant evidence to support any secondary check on these calculations. According to the TEDS data, 15 to 18 percent of treatment admissions between 1992 and 1997 identified cocaine as the primary drug of abuse. Methamphetamine was the primary drug for between 1.0 percent (1992) and 3.6 percent (1997) of admissions. If we take the 1997 numbers to imply that there were 5 hardcore cocaine users for every 1 hardcore methamphetamine user, and if we accept the estimates of the number of hardcore cocaine users from earlier, then there would be about 700,000 hardcore methamphetamine users. That is about double the estimate reported in Table 8. If we take the 1992 numbers to imply that there were roughly 15 hardcore cocaine users for every hardcore methamphetamine user, and if we again use the earlier estimates of hardcore cocaine users, we would say there are about 230,000 hardcore methamphetamine users, somewhat more than half of the number that we actually estimate. Perhaps there is some comfort here that the scale is about right, but precision is elusive.
Assuming the scale is about right, what can be said about the trend? The TEDS data show an increase in admissions with methamphetamine named as the primary drug of abuse. Just 1.0 percent of admissions in 1992 and 1.3 percent of admissions in 1993 were for methamphetamines. This compares with 2.6 percent in 1996 and 3.6 percent in 1997. We do not see those trends reflected in Table 8. This may be because hardcore users can take years to enter treatment for the first time, but after their first admission, subsequent admissions happen more frequently. Thus, a relatively constant number of hardcore methamphetamine users between 1989 and 1999 could be consistent with an increase in treatment admissions.
Drug prices might be considered a barometer of the availability of an illicit substance, which in turn partly determines the number of hardcore users. Rhodes, Johnson and McMullen report that the proportion of hardcore methamphetamine users in five jails, which had an appreciable number of methamphetamine users, showed cyclical behavior between 1989 and 1998. The proportion fell through 1991, and it then increased to a new peak in 1994. Thereafter, the proportion decreased. Rhodes, Johnson and McMullen36 show that prices moved in the opposite direction (up when use was down, and down when use was up) throughout this period, reinforcing the inference that prices are a barometer of methamphetamines' availability.
Table 8—Calculation of Total Methamphetamine Consumption, 1989–2000
Marijuana
In this section, we estimate the dollar value of marijuana consumption by multiplying the following factors: number of users in the past month, by the average number of joints used in the past month, by the average weight per joint, by the cost per ounce. Calculations are summarized in Table 9.
Number of Marijuana Users
More Americans use marijuana than either cocaine or heroin. During 1998, for example, about 11 million Americans used marijuana or hashish at least once in the month before the NHSDA. This number is about the same as it was in 1988: 11.6 million. The trend was for decreasing use into the early 1990s and then increasing use into the late 1990s.
Average Number of Joints Used Each Month
We calculated an individual's total number of joints used each month by multiplying the number of days of marijuana use in the past month by the number of joints used per occasion. For those without valid answers for these questions, we imputed the total monthly use (see Appendix D). In 1995 the NHSDA stopped asking respondents about the number of joints and amount of marijuana used in the last month. Because marijuana users reported using an estimated 18.7 joints per month in 1994, we assumed the same was true for years after 1994.
Average Amount of Marijuana Used
The average amount of marijuana used in the past month was calculated from several questions in the survey (see Appendix D). This number has changed little over time—about 0.014 ounces per joint. However, the average number and weight of joints used by those who smoke marijuana cannot tell the entire story about trends in marijuana use because marijuana's THC content has changed over time. Delta-9 tetrahydrocannabinol (THC) is marijuana's primary psychoactive chemical. According to a study conducted at the University of Mississippi,37 the average THC content of sinsemilla was at a relative peak in 1990 and 1991. That average fell from 10.5 percent in 1991 to 8.6 percent in 1992, and to 6.0 percent in 1993. The THC content of commercial-grade marijuana remained fairly constant at less that 4.0 percent from 1985 to 1992, but jumped to about 5.4 percent in 1993. According to the 1995 National Narcotics Intelligence Consumers Committee (NNICC) report, the THC content of commercial grade marijuana averaged 3.3 percent, and the THC content of sinsemilla averaged 6.7 percent, in 1995; according to the 1997 NNICC report, the commercial grade content was 5.0 percent, and the sinsemilla content was 12.2 percent. Because we do not know the mix of sinsemilla and commercial-grade marijuana used by the typical user, we cannot know, for certain, whether users are smoking more or less marijuana as measured by THC content.
Price
Price is the final factor in calculating the total value of marijuana consumption (see Appendix D). Marijuana prices were roughly $350 per ounce in the late 1980s. These prices are for a one-third ounce purchase, which appears to be a typical purchase size by frequent users. They jumped to closer to $450 per ounce during the early 1990s. Throughout the rest of the decade, prices were considerably lower. The price trends appear to be roughly consistent with trends in THC content. That is, marijuana prices were relatively low in the late 1980s when sinsemilla's THC content was comparatively high. Excluding 1990, prices were comparatively high in the early 1990s when THC content was low. Low prices toward the end of the 1990s correspond to high THC content. Taken together, these two trends suggest that marijuana was more difficult to buy in the early 1990s than it was before and than it has been since the early 1990s.
Table 9—Calculation of Total Marijuana Consumption, 1988–2000
Total Consumption Estimates
The factors required to calculate total marijuana consumption are shown in Table 9. In 1998, we estimate that average users consumed 18.7 joints a month. The average amount of marijuana used per joint equaled 0.0136 ounces.38 At a retail price of $320 an ounce, these users spent an average of $81 each month ($980 a year) on marijuana. This number, multiplied by the 11 million monthly users, yields a consumption estimate of $11 billion for the year.
These estimates may be low. Users are likely to under report socially disapproved behaviors, even when those behaviors are legal.39 They would seem to have even more incentive to under report illegal behaviors.40 Given under reporting rates for tobacco and alcohol use, it might be reasonable to inflate marijuana estimates by about one-third. On the other hand these estimates could be too high. Joints are frequently shared, and it seems plausible that these calculations double count some consumption. At any rate, our estimates of total spending are in line with estimates by others.41
Other Drugs
Most of the money spent on illicit drugs in America is spent on cocaine, heroin, marijuana, and methamphetamine. However, expenditures on other illicit substances (inhalants and hallucinogens) and on licit substances consumed illegally (other stimulants, sedatives, tranquilizers, and analgesics) is not small. Much of this drug use appears to be reported to the NHSDA.42 We do note, however, that the NHSDA undoubtedly misses some users, and those who are reached probably have an incentive to misrepresent their consumption.
Table 10 shows the number of respondents who, according to the NHSDA, used these other drugs between 1988 and 1998. To complete the table, estimates for 1999 and 2000 were set to the 1998 estimate. Those respondents who admitted use during the year were asked how frequently they used the drug.43 We then used these data to compute an average number of days a year that the respondents used a drug.44 Since the survey lacks information about the number of doses taken on days that the drug was used, we assumed that each day of use resulted in a single dose. This is most certainly an underestimate.
It is difficult to determine prices per dose. Both the Drug Enforcement Administration's (DEA) Illegal Drug Price/Purity Report and the National Institute on Drug Abuse's Community Epidemiological Working Group (CEWG) provided wide ranges.45 For current purposes, we assumed that each dose costs $5, a price that was consistent with those reported by the DEA and the CEWG. These street prices may be too high, however, because many of the legal drugs were likely to have been purchased at prescription prices and diverted to illegal use.
To estimate the yearly expenditures on these drugs, we multiplied three factors: the number of users, by the average number of doses per year, by the price per dose. Our best estimate is that Americans spent between $1.5 billion and $3.3 billion on other drugs during each of the last eleven years (Table 10).
These estimates are imprecise for the reasons noted above. However, even if we halve or double the estimates to reflect uncertainty, drugs other than cocaine, heroin, marijuana and methamphetamines must be a relatively small part of the total expenditure that Americans make on illicit substances and on legal substances consumed illegally.
Conclusion about Consumption
According to the consumption-based procedure, Americans spent about $66 billion on heroin, cocaine, methamphetamine, marijuana, and other illegal drugs in 1998: $39 billion on cocaine, $12 billion on heroin, $11 billion on marijuana, $2.2 billion on methamphetamine, and $2.3 billion on other illegal drugs (Table 11). Table 11 appears to show a substantial decrease in expenditures on illicit drugs between 1988 and 1998. Most of this change is attributable to inflation as reflected in the consumer price index. This decrease may not be apparent to hardcore users, because illicit drug consumption is a predominant part of their market basket (illicit drugs are not part of the market basket used to compute the CPI), while the nominal price of heroin and cocaine have fallen or remained about the same since 1988, and the price of marijuana has fallen since 1992. On the other hand, these decreased expenditures may have very real consequences for dealers, who probably have market baskets that are much more like that of typical American consumers.
In this section of the report we examined the use of drugs, that is, the demand for illicit drugs and for licit drugs used illegally. In the next section, we examine the availability of illegal drugs in the domestic market. Comparing the amount of drugs consumed (from this section) with the amount of drugs available for consumption (the next section) provides additional confirmation that consumption-based estimates are credible.
Table 10—Other Drugs: Total Yearly Users (thousands) and Expenditures ($ in billions, 1998 dollar equivalents), 1988–1998
Table 11—Total Expenditures on Illicit Drugs, 1989–2000 ($ in billions, 1998 dollar equivalents)
