Research paper
School of criminology, University of Montreal, C.P. 6128, succursale Centre-ville, Montral, QC, H3C 3J7, Canada
International Centre for Comparative Criminology, C.P. 6128, succursale Centre-ville, Montral, QC, H3C 3J7, Canada
a r t i c l e
i n f o
Article history:
Received 17 December 2012
Received in revised form
29 November 2013
Accepted 9 December 2013
Keywords:
Drug prices
Drug trafcking
Risks and prices model
World-system perspective
Network analysis
Drug law enforcement
a b s t r a c t
Background: Illegal drug prices are extremely high, compared to similar goods. There is, however, considerable variation in value depending on place, market level and type of drugs. A prominent framework for
the study of illegal drugs is the risks and prices model (Reuter & Kleiman, 1986). Enforcement is seen as
a tax added to the regular price. In this paper, it is argued that such economic models are not sufcient
to explain price variations at country-level. Drug markets are analysed as global trade networks in which
a countrys position has an impact on various features, including illegal drug prices.
Methodology: This paper uses social network analysis (SNA) to explain price markups between pairs of
countries involved in the trafcking of illegal drugs between 1998 and 2007. It aims to explore a simple
question: why do prices increase between two countries? Using relational data from various international
organizations, separate trade networks were built for cocaine, heroin and cannabis. Wholesale price
markups are predicted with measures of supply, demand, risks of seizures, geographic distance and
global positioning within the networks. Reported prices (in $US) and purchasing power parity-adjusted
values are analysed.
Results: Drug prices increase more sharply when drugs are headed to countries where law enforcement
imposes higher costs on trafckers. The position and role of a country in global drug markets are also
closely associated with the value of drugs. Price markups are lower if the destination country is a transit to
large potential markets. Furthermore, price markups for cocaine and heroin are more pronounced when
drugs are exported to countries that are better positioned in the legitimate world-economy, suggesting
that relations in legal and illegal markets are directed in opposite directions.
Conclusion: Consistent with the world-system perspective, evidence is found of coherent world drug
markets driven by both local realities and international relations.
2013 Elsevier B.V. All rights reserved.
236
Trade networks
A major contribution of the world-system perspective was to
shift the focus of analysis from individual countries to the relations
between them. Empirical tests of the world-system perspective
then quickly used tools of social network analysis (SNA). A similar trend can be observed for drug trafcking: recent editions of
the World Drug Report, a widely-cited annual publication by the
United Nations Ofce on Drugs and Crime (UNODC), include a discussion of drug ows and routes between countries. In a recent
publication, Paoli, Greeneld, and Reuter (2009) used the network
terminology to describe the world heroin market as a trade network in which distant regions can affect aspects of local markets,
but grounded their analysis in traditional economics.
However, Paoli et al.s work is a notable exception: drug trafcking is usually not analysed in relational terms. Farrell, Mansur, and
Tullis (1996), who analysed cocaine and heroin trafcking in Europe
during the 1980s and 1990s, still provide the most comprehensive examination of wholesale prices. Their analysis is interesting
because it shows how European drug markets evolved over a 10year period (19831993). It also introduces the idea that countries
had steady roles in the market throughout the period and that a
wide array of factors explain that situation. Farrell et al. observed
that wholesale prices were lower in the countries that serve as gateways to the European markets Spain, Portugal, and Turkey. In
neighbouring countries, drugs were a little more expensive, but
still cheaper than in most other countries, which is consistent with
the idea that prices increase with distance. They also observed
that drugs were expensive in Switzerland and concluded that it
reected a more general pattern: everything was more expensive in
Switzerland. Finally, Farrell et al. suggested that the level of risk for
importers was associated to wholesale prices, citing the example
of the Netherlands, a country that was thought to be more lenient
about drugs and where lower than expected prices were observed.
Despite its descriptive nature, Farrell et al. (1996) offer an important contribution to the eld for at least two reasons. First, they
provide one of the few analyses of transnational drug trafcking
based on empirical data gathered by the UNODC. The UNODC data
has limitations but provides conveniently accessed information
unavailable anywhere else (Caulkins, 2007). Second, their study
was an attempt to build a general model of high-level trafcking that tries to account for price variations without focusing on
local and anecdotal explanations. Their tentative results demonstrated that country-level analyses are instructive and deserve
further research. Furthermore, while their results are consistent
with economic principles, Farrell et al. suggested that prices reect
the role of a country within the global drug trade. They nd evidence of a coherent European market driven by both local realities
and international relations.
Since the exploratory analysis of European drug prices by Farrell
et al. (1996), there has been virtually no attempt to formulate a
general model of country-level price variations. Costa Storti and De
Grauwe (2009a, 2009b) did use a theoretical model to explore the
impact of globalization on drug prices, but their analysis was aimed
at explaining variations of retail prices over time. This paper offers
an empirical test of the risks and prices model (Reuter & Kleiman,
1986) combined with propositions from the world-system perspective to explain illegal drug price increases at country-level.
Methodology
Data and network construction
The primary data for this study was gathered together by the
UNODC and covers a 10-year period from 1998 to 2007. The
UNODC releases an overview of various indicators of drug trafcking and consumption in its annual report, World Drug Report.
Most indicators are collected through an annual survey, the Annual
Questionnaire Reports (ARQ), which is lled out by ofcials in different countries. In instances in which ARQ data are not available, the
UNODC complements with data from other sources, such as INTERPOL (Chandra, Barkell, & Steffen, 2011). In most cases, the data is
released without further test of its validity. The average value for
the whole period is used, except when there is a specic mention.
Other data sources are detailed below.
An almost undisputed feature of illegal drugs is that they are
exchanged in separate markets, although some convergence is
found (UNODC, 2011). For example, a single region (the Andes) produces all the cocaine consumed in the world, while three regions
produce heroin (the Golden Crescent and Golden Triangle in Asia
and South/Central America) and cannabis is grown virtually everywhere. The number of sources is strongly associated with the
availability of drugs; it also seems plausible to expect different price
variations and determinants in different markets.
The rst step in this analysis was to build drug trade networks.
Two types of relational data were used to build separate trade
networks for plant-based drugs (cocaine, heroin, and cannabis)
at country-level. The rst data is a collection of seizures of signicant quantities of drugs that occurred between 1998 and
2007.1 This dataset provides detailed information on a large
number of cases, including origin and/or destination countries
(n = 20 527 dyads). When Spanish authorities seize drugs coming
from Venezuela and headed to France, they collect information on
a network of three nodes (Spain, Venezuela, and France) and two
relations (VenezuelaSpain, SpainFrance). The accumulation of
1
The dened thresholds of signicant quantities used by UNODC are as follows:
1 kg or more for cannabis and 100 g or more for cocaine and heroin.
237
GNP PPPx
median value of GNP PPP
where GNP PPP stands for a countrys gross nation product per
capita adjusted for purchasing power parity in 2002 (IMF, 2002).
Algeria had the median value, with $5900/capita, which means
that the wholesale price of drugs in Algeria is equal to its adjusted
value. Luxembourg had the highest factor of adjustment (10.19) and
Malawi had the lowest (0.09); if drug prices were the same worldwide, the adjusted value of drugs worth $1000 in Algeria would be
$10 190 in Luxembourg and $90 in Malawi.
2
A systematic review of information contained in 48 annual reports and country
overviews published by the UNODC, the Bureau of International Narcotics and Law
Enforcement Affairs (BINLEA), the International Narcotics Control Board (INCB) and the
European Monitoring Center for Drugs and Drug Addiction (EMCDDA) was conducted.
3
Prices are not adjusted for purity. Some authors (e.g. Caulkins, 2007) have
suggested that quality-adjusted prices should be analysed instead of raw prices
because price changes may manifest through quality changes rather than selling
price changes. However, the purity of drugs requires chemical tests that are not
routinely done in most countries of the world. Furthermore, there are a number of
important issues (e.g. selection bias) related to purity tests that are well-beyond
the scope of this paper. Finally, substantial quality variations have been observed at
retail-level (Caulkins et al., 2004; Darke, Topp, Kaye, & Hall, 2002), but the situation
is less clear at higher levels of drug trafcking.
238
4
Bahrain was excluded from the analysis because it had a value more than eight
times higher than the average.
are higher if nodes are located on every path between two countries (de Nooy, Mrvar, & Batagelj, 2005; Freeman, Borgatti, & White,
1991; White & Borgatti, 1994). Flow betweeness is different from
more traditional measures of centrality (e.g. betweeness centrality) because it includes every possible path, while other measures
are based solely on the shortest paths (Borgatti, 2005). Standardized values are used for cocaine and heroin. Because cannabis is
grown almost everywhere in the world it is almost impossible to
trace it back to source; consequently, ow betweeness could not
be calculated for cannabis.
An estimate of the number of potential consumers is also
included in the analysis, to test the proposition that large markets
are more competitive and prices could therefore be lower. It is given
by the product of active population size (age 1564) multiplied
by the prevalence of consumption reported in the 2009 edition
of the World Drug Report (UNODC, 2009). A measure of potential markets is also included to better account for transit positions.
For example, Spain is a well-known turntable for moving South
American cocaine into Europe. Spain itself is not a very large drug
market compared to more populated countries, but it allows drugs
to enter a multi-million user market. Therefore, Spain plays a crucial role: not only are drugs exported from Spain to a large number
of countries but those destination countries are populated with a
large number of drug users. Potential buyers is a measure of the
number of users in countries connected to destination countries.
In the case of heroin, the number of potential buyers (from other
countries) and potential consumers (from the country of destination) is highly correlated (r = 0.625; p < 0.001) and related to price
markups in a similar way. For cocaine, both measures of transit
are slightly less correlated (r = 0.473; p < 0.001) and have opposite
effects on price markups. Therefore, both measures were included
in models predicting cocaine price markups.
Relational variables
It has been recognized for decades that geographic distance is
an important determinant of the pattern of legal trade (Beckerman,
1956; Disdier & Head, 2008). The volume of trade between distant
countries is expected to be lower because of higher transport costs,
among other costs, and prices, of course, are expected to be higher
where costs are higher (Deardroff, 1998). Consequently, the geographic distance between countries was included in the models
predicting price markups.
5
The non-parametric correlation between Mahutgas classication and exports
(in US$) is nearly perfect (Spearmans rho = 0.967; p < 0.001; n = 53). The core of
the world-economy is constituted by the top exporting countries in the world; the
less exports, the further from the core.
239
Table 1
Bivariate correlations between reported cocaine, heroin and cannabis wholesale
prices (US$), 19982007 (coefcient: Pearsons R).
Cocaine (ln)
Heroin (ln)
Cannabis (ln)
*
**
Cocaine (ln)
Heroin (ln)
Cannabis (ln)
0.28* (n = 69)
0.49** (n = 77)
0.46** (n = 81)
p < 0.05.
p < 0.01.
breaking a key assumption of OLS regression analysis (no multicollinearity). While economic contributions are acknowledged,
corruption was preferred over GNP because the paper is aimed at
providing a criminological and sociological explanation for price
variations rather than an economic analysis. However, these measures are expected to have opposite effects on drug prices: prices
should be higher in richer countries, while richer countries have
lower levels of perceived corruption. At the same time, the possibility of avoiding seizures and arrests by buying a way out is
a signicant source of savings. Lower risks mean lower compensation, which in turn should be associated with lower prices. Adjusted
prices are expected to better capture the effect of corruption on
the value of illegal drugs because the effect of relative wealth is
included directly in the dependent variable.
An important assumption should also be mentioned. In many
cases, the situation of a whole country and all the trafckers
operating within its borders is summarized by a single value.
Therefore, throughout the analysis, it is assumed that wholesale
prices are indicative of the value of drugs in a country after importation and before exportation to another country; it was not possible
to assess possible price increases within countries.6 For example,
according to the 2008 edition of the World Drug Report, a kilo of
cocaine sold for $31 580 in Canada (UNODC, 2008). During the same
period, the Royal Canadian Mounted Police the federal police
reported that a kilo of cocaine could be bought in British Columbia
(the far western part of Canada) for $20 000, while in Saskatchewan
(a more central province) it cost $70 000 (RCMP, 2007). The value
reported by the UNODC is somewhere in-between but not perfectly representative of the situation in either British Columbia or
Saskatchewan.
Results and discussion
Consistent with the observations of Farrell et al. (1996) there
is a statistically signicant and positive correlation between prices
for cocaine, heroin, and cannabis in a country; as a general rule,
when one drug is high-priced in a given country, other drugs are
also high-priced, suggesting that there is a set of common factors
that explain drug prices. However, the correlations are sufciently
low to require further examination. Consequently, analyses are provided for three types of plant-based illegal drugs: cocaine, heroin,
and cannabis (Table 1).
Preliminary analyses revealed that sending country features
were unrelated to price markups but that destination country features were. Thus corruption, police per capita, potential consumers,
6
Prices are expected to increase within a country for various reasons. Most
notably, there are costs associated with stocking drugs for future transactions
(Caulkins & Reuter, 1998). Drugs that are not moving can nevertheless be detected
and seized by local enforcement agencies; stocking drugs are thus likely to be associated with additional risk compensation, even if it is only a moderately risky activity.
Caulkins (1995) and Caulkins and Padman (1993) also demonstrated that distance
from the entry point is strongly related to drug prices within a country. Drugs tend
to be least expensive near their source and more expensive farther away. Local trafckers are also in business to make money and also pass their costs on to the next
buyers.
240
Table 2
Descriptive statistics, price markups.
Unadjusted prices (US$)
Adjusted values
Cocaine
Heroin
Cannabis
Cocaine
Heroin
Cannabis
N
Average
Median
Standard
deviation
Minimum
226
37 431
30 598
33 950
212
31 986
18 471
40 575
110
2756
1691
3570
214
185 660
172 578
166 070
220
144 622
74 220
204 250
111
12 149
4258
18 761
300
(Dom. Republic Haiti)
8
(Colombia Panama)
8
(Croatia Slovenia)
154
(Colombia Ecuador)
1
(Guinea Mali)
Maximum
134
(Afghanistan
Tajikistan)
1 025 456 (Pakistan
Australia)
97 017 (South
Africa Ireland)
Table 3
OLS regression models, price markups, 19982007 (coefcient: beta).
Cocaine
N pairs of countries
Destination: corruption
Destination: police per 1000 inhabitants
Destination/transit: ow betweeness
Destination/transit: potential buyers (ln)
Destination: number of users (ln)
Geographic distance (kms; ln)
Towards core
R squared
*
**
Heroin
Cannabis
Markups ($US)
Markups (adjusted
prices)
Markups ($US)
Markups (adjusted
prices)
Markups ($US)
Markups (adjusted
prices)
225
0.204**
0.007
0.235**
0.325**
0.101
0.518**
0.080
0.692**
213
0.450**
0.122**
0.220**
0.246**
0.024
0.292**
0.198**
0.758**
212
0.240**
0.153*
0.251**
0.085
0.407**
0.003
0.275**
220
0.660**
0.265**
0.112*
0.102
0.212**
0.104*
0.616**
102
0.379**
0.115
0.269**
0.452**
0.133
0.463**
103
0.731**
0.212**
0.063
0.135
0.087
0.616**
p < 0.1.
p < 0.05.
p < 0.01
7
An anonymous reviewer pointed out that there could be an issue with the
assumption of statistical independence because countries could be repeated in the
dataset; for example, the fact that the U.S. appears four times as a receiving country
means that its features appear four times in the models. Consistent with the literature, supplementary analyses using robust errors and cross-random effect terms did
show differences in standard errors but the estimates remained roughly the same.
Conclusion
Drug trafcking is an illegal activity that consists of multilateral
exchanges of prohibited goods between producers, distributors,
and consumers in a market-like context (Naylor, 2003). At the
global level, drug trafcking is best conceived as a series of relations
between countries. Network analysis naturally ts this relational
denition; knowledge about drug trafcking is gained from the
combination of cross-national and relational analyses.
This paper aimed to test propositions derived from two apparently unrelated theoretical perspectives. The risks and prices model
(Caulkins & Reuter, 2010; Reuter & Kleiman, 1986) states that law
enforcement imposes additional costs on trafckers that are passed
on to the next buyers of the drugs in the form of higher prices. The
model predicts that prices will be higher where costs are higher.
Most costs are due to risk compensation for possible prison sentences and violence (Caulkins & Reuter, 1998).
241
242
Table A1
OLS regression models, wholesale prices, 19982007 (coefcient: beta).
Cocaine
N countries
Corruption
Police per 1000 inhabitants
Geodesic distance from source
Networks: ow betweeness
Networks: import
Networks: export
America
Africa
Europe
R squared
*
**
Heroin
Cannabis
58
0.347**
0.011
0.508**
0.229*
0.639**
58
0.684**
0.095
0.300**
0.140
0.743**
82
0.317**
0.192*
0.118
0.318**
0.173
0.389**
82
0.670**
0.198**
0.100
0.214**
0.178*
0.718**
104
0.402**
0.064
0.112
0.191**
0.234**
0.310**
0.716**
104
0.514**
0.079
0.110*
0.160**
0.229**
0.268**
0.817**
p < 0.1.
p < 0.05.
p < 0.01.
Acknowledgements
This research was partly funded by a grant from the Social Sciences and Humanities Research Council (Government of Canada).
The author would like to thank Pierre Tremblay and Carlo Morselli
for their support, comments and suggestions throughout the
research. The author would also like to acknowledge the contribution of anonymous reviewers.
Conict of interest statement
No conict of interest is declared.
Appendix A.
Table A1.
References
Beckerman, W. (1956). Distance and the pattern of intra-European trade. The Review
of Economics and Statistics, 38(1), 3140.
Benson, J. S., & Decker, S. H. (2010). The organizational structure of international
drug smuggling. Journal of Criminal Justice, 38, 130138.
Boivin, R. (2011). Le monde lenvers: analyse de la structure du trac transnational de
drogues illicites. School of Criminology, University of Montreal. PhD dissertation.
Boivin, R. (2013). Drug trafcking networks in the world-economy. In C. Morselli
(Ed.), Crime and networks. London, New York: Routledge.
Borgatti, S. P. (2005). Centrality and network ow. Social Networks, 27, 5571.
Bouchard, M., & Ouellet, F. (2011). A survival analysis of failure in the drug trade.
Global Crime, 12(1), 7086.
Bouchard, M., Potter, G. R., & Decorte, T. (2011). Emerging trends in cannabis cultivation and the way forward. In T. Decorte, G. R. Potter, & M. Bouchard (Eds.),
World wide weed: Global trends in cannabis cultivation and its control. Burlington:
Ashgate Publishing Company.
Caulkins, J. (1994). Developing price series for cocaine. Santa Monica, CA: RAND.
Caulkins, J. (1995). Domestic geographic variation in illicit drug prices. Journal of
Urban Economics, 37, 3856.
Caulkins, J. P. (2007). Price and purity analysis for illicit drug: Data and conceptual
issues. Drug and Alcohol Dependence, 90, S61S68.
Caulkins, J. P., Pacula, R. L., Arkes, J., Reuter, P., Paddock, S., Iguchi, M., et al. (2004).
The price and purity of illicit drugs: 1981 trough the second quarter of 2003. Washington: Ofce of National Drug Control Policy.
Caulkins, J., & Padman, R. (1993). Quantity discounts and quality premia for illicit
drugs. Journal of the American Statistical Association, 88(423), 748757.
Caulkins, J., & Reuter, P. (1998). What price data tells us about drug markets. Journal
of Drug Issues, 28(3), 593612.
Caulkins, J., & Reuter, P. (2007). Illicit drug markets and economic irregularities.
Socio-Economic Planning Sciences, 40(1), 114.
Caulkins, J. P., & Reuter, P. (2010). How drug enforcement affects drug prices. In M.
Tonry (Ed.), Crime and justice: An annual review of research (vol. 39) (pp. 213271).
Chicago: University of Chicago Press.
Chandra, S., Barkell, M., & Steffen, K. (2011). Inferring cocaine ows across Europe:
Evidence from price data. Journal of Drug Policy Analysis, 4(1), 118.
Chase-Dunn, C. (1989). Global formation. Cambridge: Basil Blackwell.
243