Jump to ContentJump to Main Navigation
The World Heroin MarketCan Supply Be Cut?$

Letizia Paoli, Victoria A. Greenfield, and Peter Reuter

Print publication date: 2009

Print ISBN-13: 9780195322996

Published to Oxford Scholarship Online: May 2012

DOI: 10.1093/acprof:oso/9780195322996.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 24 February 2017

(p.261) Appendix B Average Consumption and Purity

(p.261) Appendix B Average Consumption and Purity

The World Heroin Market
Oxford University Press

The analysis of opiate markets is complicated by the lack of good estimates of the quantities consumed by an opiate user each year. The issue comes up at a number of points in our analysis. For example, we need the quantity estimate to compare total production and consumption, and to establish whether opiate inventories have been growing or shrinking. Similarly, to estimate the share of Afghanistan’s production that flows through Tajikistan, we must estimate total heroin consumption in Russia and other markets supplied through Central Asia. For each nation, we need figures on the number of users and on how much each consumes on average in a year. Official statistics provide only estimates of the number of users.

Too few data exist to permit nation-specific estimates of average consumption for most countries of interest, other than the United States. To fill in the gaps and to provide a basis for benchmarking the nation-specific estimates that we have constructed, we have developed a “default” rate for average annual consumption in countries outside the United States (i.e., 30 grams). For the United States, we use a 15-gram estimate that is both consistent with U.S. government estimates of U.S. heroin consumption and with the uniquely high price of heroin in that country. This appendix describes the basis for the default rate and the U.S. estimate, and compares our figure with other aggregate studies. It also summarizes the small number of articles that include data permitting estimation of consumption by heroin users in specific samples. The focus on heroin is a result of the fact that no contemporary studies except for Iran (Cultural Research Bureau, 2001) have attempted to estimate average opium consumption.1

(p.262) Methodological Issues

The difficulty in developing consumption estimates arises largely from that fact that a user cannot report how much of an illegal drug he or she purchases. Sales are often made hurriedly in clandestine settings with little or bad information. No sale comes with a meaningful guarantee regarding the quantity or purity of the drug; in some nations, retail purity varies a great deal. Buyers can report only how much they spent or, in the case of heroin, how frequently they injected, smoked, or inhaled.

To use data on spending, one needs purity-adjusted price data to calculate the quantity of heroin each user consumes; such price data are available on a regular basis only in the United States. Within Europe, the EMCDDA reports “typical” retail purity as lying between 20% and 45% (EMCDDA, 2002a:26) but does not report a purity-adjusted average price. The great variation in observed street-level purity and unadjusted prices, and the lack of data on the correlation between them, creates a great deal of uncertainty in an estimate that simply divides average price by average purity.

To convert reports of use frequency to estimates of quantities, it is necessary to have an estimate of the standard quantity in a dose. The size of a dose is also not available on a systematic basis. Consider, for example, what was available for Russia. A doctor specializing in drug treatment in Moscow, who was interviewed for a related project, believed that addicts in her clinic injected two to four times daily. Paoli (2001) reports that the usual selling unit is 100 milligrams; the two to four times injection per day would then suggest a figure of 200 to 400 milligrams per day. Unfortunately, no purity data are available.

Purity is a major problem for the analysis of opiate markets. We believe that purity declines as heroin moves along the distribution chain, reflecting cutting with diluents by successive dealers. For example, for Turkey, as shown in chapter 4, the average purity in multikilo shipments in 2002 was about 40% whereas in Germany, in 2002, 45% of retail seizures were less than 10% pure and only 9% were more than 30% pure (REITOX [Germany], 2003:54) However there is substantial variation at all levels of the market. Reuter and Caulkins (2004) reported that from 1987 to 1991, approximately one eighth of the U.S. DEA’s U.S. retail purchases had purity less than 5% whereas more than 10% had purity greater than 75%. In U.S. low-level wholesale markets (more than 10 grams, raw quantity), the reported interquartile range in 2002 was 34% to 62% (ONDCP, 2004).

Our Procedure

Absent national consumption estimates, we have developed a “default” rate for consumption outside the United States—an average, expressed per user in pure heroin equivalent grams—using U.S. and other evidence. We started with the United States, because it provides the most systematic evidence on (p.263) consumption quantities. The U.S. ONDCP (2001) reports that U.S. heroin addicts consume roughly 15 grams of pure heroin per year, or about 50 milligrams per day when actively using, which we assume to be about 300 days per year, allowing for sickness, a few days in a local jail or treatment program, and other short-lived breaks in use.

For our non-U.S. default rate we assume that opiate users in other countries consume twice as much per capita as those in the United States (i.e., 30 grams of pure heroin equivalent per user per year), reflecting the lower prices outside the United States. The increase gives nod to price responsiveness, absent a full cross-country analysis of prices and demand elasticities. We believe that an estimate of 100 pure milligrams per user per day—consistent with an annual estimate of about 30 pure grams—for countries with opiate prices that are, relative to average earnings, much lower than the United States, is reasonable and not inconsistent with judgments of experts.

Other Aggregate Studies

A few other studies have attempted to develop population-level estimates. The UNODC (2005d) reports a global average of 28 grams per annum and a European average of 58 grams. For validation of the higher figure, the UNODC cites the results of a U.K. study on people entering treatment in 1997, which it states implies 68 grams2 (Gossop, Marsden, and Stewart, 1997). However, treatment research (e.g., Anglin and Hser, 1990) has consistently found that users enter treatment at times of peak use; thus, reports of use in the period immediately before treatment entry will overstate average use rates. Moreover, treatment entry is itself not randomly distributed across dependent users; those with more severe problems have a higher probability of being referred to treatment as a consequence of arrest. Thus we believe that the figure is too high. The UNODC figures for other regions are based on estimated total consumption from an input/output model, divided by estimated prevalence. The regional figures vary from 10 grams in South America to 56.5 grams in Oceania; for most regions, the figure falls between 15 grams and 33 grams.

Bramley-Harker (2001) estimated total pure heroin consumption in 1999 in the United Kingdom at about 11 metric tons,3 which for a population of 275,000 heroin addicts amounts to about 40 grams per addict per annum, nearly three times the U.S. figure of about 15 grams and a third higher than our 30-gram non-U.S. default rate. Although heroin is unusually cheap in the United Kingdom, a number of assumptions used in the estimating procedure may have biased the numbers upward. For example, the study assumed that the number of days of active consumption was 52 times the number reported the previous week by users not in treatment or prison. In fact, heroin users spend a good deal of their careers in treatment or prison; thus, the number of days of use will be substantially lower.

(p.264) A later estimate for the United Kingdom (Singleton, Murray, and Tinsley, 2006) showed a lower total figure of about 8 pure metric tons for 2003, representing differences in methodology and data sources. On the basis of a relatively sophisticated analysis of a survey of arrestees, the study estimated that intensive users consumed the equivalent of 160 to 240 adulterated milligrams per use day (48–72 grams per annum) and non-intensive users consumed 100 to 185 milligrams per use day (30–55 grams per annum) (Singleton et al., 2006:67). The figures were lower if the user was in treatment. With an estimated heroin-using population of 280,000 opiate users (Singleton et al., 2006:28), the implied annual consumption per user was approximately 29 grams per annum, a figure remarkably close to our default rate.


We have found only four studies outside the United States that report enough data to permit even a rough estimate of annual consumption. Atha and Davis (2003) report purity-adjusted data from a list of customers in northern England showing a median of 280 milligrams of heroin per day but a skewed distribution with a substantially higher average (e.g., the 75th percentile was 515 milligrams). These data came from records in a prosecution and required the strong assumption that each customer had no other dealer.

The other three studies draw on samples of users in treatment. Jimenez-Lerma, Manuel, Landabaso, Iraurgi, Calle, Sanz, Gutiérrez-Fraile (2002) report in a study of 80 addicts in a treatment clinic in the Basque region of Spain that their patients consumed an average 512 milligrams of heroin per day. The study does not report purity but, according to the Observatorio Español sobre Drogas (2005:172), the purity of a dose of heroin has consistently remained above 20% since 1998. Similarly, Smolka and Schmidt (1999) report average daily consumption of 740 milligrams per day for a sample of 22 addicts in treatment in Berlin. They do not present purity data. For Germany, heroin retail purity has generally oscillated, from 1996 to 2004, between less than 10% and 20%, with substantial variation across cities (BKA, 2005b:40). Gossop et al. (1997) report data on 1,075 treatment admissions in 1995 in the United Kingdom. Average monthly heroin consumption at admission varied by treatment modality, ranging from 9.4 grams (inpatient) to 16.4 grams (methadone reduction). The weighted average was approximately 12 grams per month, using data from the 809 respondents who reported at the 6-month follow-up.

Table B.1 attempts to array the results in a consistent fashion. However, because each study reported the results in a different fashion, we have had to use different procedures to develop the annual pure gram estimates. For example, Gossop et al. (1997) included figures on total monthly consumption at street purity and the number of days used per month, whereas Jimenez-Lerma et al. (2002) reported directly the average quantity per day. We have (p.265)

Table B.1 Individual study estimates of heroin consumption.



Sample Description (size)

Quantity per Day (grams)

Estimated Purity (%)

Implied Annual Consumption Pure Heroin (grams)

Atha and Davis, 2003

Northern England

Customers of a single dealer (92)




Jimenez-Lerma et al., 2002

Basque country

Treatment sample (80)




Smolka and Schmidt, 1999


Treatment sample (22)




Gossop et al., 1997


Multisite treatment sample (809)




also had to make assumptions for some studies about the number of days used per annum and we corrected for inconsistencies in data presentation. The results are thus quite approximate.

Given the bias that we believe arises from use of treatment samples to estimate average use, we draw comfort from these studies. They suggest that the 30 grams of pure heroin per annum, at least for western Europe around the end of the 20th century, may be a reasonable approximation.


Note that national prevalence estimates are often not well defined in terms of the frequency of use of the included population; they might include not only those dependent on opiates, but also occasional users. The official U.S. estimate for 2000 was roughly 900,000 chronic users (using at least eight times per month) and approximately 250,000 occasional users (ONDCP, 2001:9, table 3).4 Other nations do not have consistent series on the numbers of occasional, dependent, or chronic users. Moreover, the national estimates also include users of opium and morphine, who may consume large quantities by weight because their consumption methods, such as smoking in the case of opium, are less efficient. Absent reliable data, we rarely make any adjustment for users of opiates other than heroin, but, except for a few countries, it is thought that they constitute a small share of the total outside Asia.

(p.266) Annual consumption per user might be expected to vary over time and across countries. For example, the inflation-adjusted price of heroin has fallen in many western nations during the past 20 years. This should lead heroin users, on average, to consume somewhat more of the drug in recent years, unless the price of other, substitute drugs has fallen even more. In some western nations, heroin addicts are eligible for income support as a result of their poverty; in others, they are not. It is plausible that addicts in the former, with higher incomes, would consume more heroin. We offer these observations not as predictive statements, but as indicative of the factors that might lead to variation in quantities consumed—a topic that has not yet been examined.

Moreover, it may well be that in some nations, where opiates are particularly cheap in absolute and relative terms, the correct figure is higher than 30 grams per annum. Occasionally, reports provide some measure of the relative cost of a dose of heroin. For example, in Dushanbe, the capital of Tajikistan, in 1999 a 150-milligram dose sold for $0.60, a kilogram of rice cost $0.33, and a kilogram of cooking oil cost $0.66 (DCA, 2004), which suggests that heroin is relatively cheap for addicts in that country. It is unclear what would serve as counterpart commodity measures for a U.S. heroin addict, but with heroin costing the average user about $30 per day (ONDCP, 1999:13), the cost of a heroin habit may be higher as a ratio of the goods needed for survival, making it relatively more expensive in the United States in real terms.

As a possible upper bound on the amount that a user might consume, there are data from heroin maintenance programs launched since the late 20th century in several western European countries. In Switzerland, for example, addicts could consume as much heroin as they desired for a fixed payment (i.e., the cost was not related to the size of the dose) and daily consumption averaged about 500 milligrams per day or 150 grams per year (Rehm, Gshwend, Steffen, Gutzwiller, Dobler-Mikola, and Uchtenhagen, 2001). No illegal market will generate figures close to that.


(1.) Very detailed estimates of opium consumption in China spanning the nineteenth and twentieth centuries have been provided by Newman (1995; see chapter 2 this volume).

(2.) There are some problems in reconciling the interpretation of the UNODC (2005d:132) with the published article. The UNODC states: “The study showed an average consumption of 0.6 grams per day, and a consumption of, on average, 22 days per month. Average consumption per month was thus 14.9 grams of heroin (at street purity), which amounts to 179 grams per year. Applying the average purity of around 38 percent reported by forensic laboratories in the UK in 1997 (The Forensic Science Service, Drug Abuse Trends, various issues), average annual consumption would be 68 grams of pure heroin per problem drug user.” The article, in fact, reports prior month quantity used by treatment modality and our estimate of the weighted average (p.322) monthly consumption of heroin (using procedures that might bias the figure upward) is 11.7 grams per month. At 38% purity, that is 4.45 grams per month, yielding 54 grams per annum. However, the data were collected in 1995, not 1997, so the purity may not be correct.

(3.) By comparison, in chapter 5, we estimate average annual consumption of almost 9 metric tons from 1996 to 2000 and about 8.2 metric tons from 2001 to 2003. The cross-period difference arises largely from a change in the UN’s basis for calculating prevalence.

(4.) This estimate, which the ONDCP describes in the table notes as a projection, was simply an extrapolation of the 1998 figure on the assumption that little had changed. Although the series for the number of chronic users was stable for 1992 to 1998, that for occasional users fluctuated sharply, probably representing the small number of observations in the household survey on which it was based.