Content uploaded by Michelle Dion
Author content
All content in this area was uploaded by Michelle Dion
Content may be subject to copyright.
Eradication Efforts, the State,
Displacement and Poverty : Explaining
Coca Cultivation in Colombia during
Plan Colombia*
MICHELLE L. DION and CATHERINE RUSSLER
Abstract. This study models the sub-national pattern of coca cultivation in Colombia
following the implementation of Plan Colombia (2001–2005). The results suggest
that aerial eradication reduces coca cultivation primarily through creation of signifi-
cant displacement and that coca cultivation is less intense in areas with a significant
state presence. Further, coca cultivation appears to be more common in less de-
veloped, agricultural regions where access to legal markets precludes other forms
of agriculture. Poverty has a significant, non-linear effect on coca cultivation; culti-
vation is most intense in regions of moderate poverty. Based on the findings, efforts
to reduce coca cultivation should emphasise developing local public infrastructure
and market access in conjunction with poverty reduction efforts and investment
in alternative development.
Keywords: Colombia, coca cultivation, aerial eradication, poverty, public infra-
structure, market access, Plan Colombia
Introduction
Coca production begins in the valleys and upper jungle regions of the
Andean region, where the countries of Colombia, Peru and Bolivia are
host to more than 98 per cent of the global land area planted with coca.
1
During the 1990s, Colombia rapidly overtook its neighbours and became
the region’s leading coca producer – while in 1994 the country accounted
for 22 per cent of total regional cultivation, by 1990 that figure had risen to
Michelle Dion is Assistant Professor of International Affairs at the Georgia Institute
of Technology, Atlanta, GA 30332-0610. (404) 385-4081. Fax : (404) 894-1900. Email :
mdion@gatech.edu.
Catherine Russler earned an M.S. in International Affairs at the Georgia Institute of
Technology, Atlanta, GA 30332-0610. Email: caterus@gmail.com
* The authors thank the JLAS editors and anonymous reviewers for constructive suggestions
for improving this article.
1
Rocio Moreno-Sanchez, et al., ‘ An Econometric Analysis of Coca Eradication Policy in
Colombia’, World Development, vol. 31, no. 2 (2003), pp. 375–83.
J. Lat. Amer. Stud. 40, 399–421 f2008 Cambridge University Press 399
doi:10.1017/S0022216X08004380 Printed in the United Kingdom
73 per cent.
2
This Colombian expansion coincided with the implementation
of aggressive coca eradication and interdiction campaigns in Bolivia and
Peru, backed by the USA, as well as the dismantling of the Colombian cartels
that imported Andean coca for production and export from Colombia.
Despite aggressive interdiction and eradication measures within Colombia,
myriad smaller cartels and the cultivation of coca have since proliferated in
its countryside.
3
The United States government views the illicit drug industry
as a national welfare threat and allocates enormous resources to its destruc-
tion. The USA has accordingly provided Colombia with well over US$4
billion since 1999 in order to combat the production of coca, cocaine’s
primary ingredient. The rationale underlying the US-funded supply-side
strategies, collectively called Plan Colombia, is that they will curb availability,
drive the street price up, and ultimately reduce the demand and consumption
of cocaine.
4
Yet wholesale prices for cocaine within the United States have
actually decreased throughout the Plan Colombia era, suggesting an ever-
abundant supply.
5
Supply-reduction programmes, which under Plan Colombia rely heavily
upon aerially-sprayed herbicides, have been implemented without a clear
understanding of the factors that drive sub-national cultivation patterns.
Forced eradication efforts, as a result, have amounted to large-scale ‘ weed-
whacking’. This is evident in US State Department reports, which state that
attempted cultivation in Colombia continues to expand despite increasing
efforts to suppress cultivation. Attempted coca production – defined as
eradicated plus un-eradicated coca – has risen by 36 per cent since 2000.
6
The failure of eradication efforts and of coca’s apparent cultivation shifts
within the Andean region can be characterised as a balloon-effect, where
production squeezed off in one locale simply forces illicit coca production
2
United Nations Office on Drugs and Crime (UNODC), Colombia Coca Survey (New York
2005).
3
Angel Rabasa and Peter Chalk, Colombian Labyrinth: The Synergy of Drugs and Insurgency and Its
Implications for Regional Stability (Santa Monica 2001) ; Ricardo Vargas, ‘ The Anti-Drug
Policy, Aerial Spraying of Illicit Crops and Their Social, Environmental and Political
Impacts in Colombia’, The Journal of Drug Issues, vol. 22, no. 4 (2002), pp. 11–60 ; Marı
´a
Clemencia Ramı
´rez Lemus, et al., ‘Colombia : A Vicious Circle of Drugs and War’, in
Coletta A. Youngers and Eileen Rosin (eds.), Drugs and Democracy in Latin America : The
Impact of US Policy (London 2005), pp. 61–97.
4
Coletta A. Youngers and Eileen Rosin, ‘The US ‘‘War on Drugs ’’ : Its Impact in Latin
America and the Caribbean’, in Coletta A. Youngers and Eileen Rosin (eds.), Drugs and
Democracy in Latin America : The Impact of US Policy, (London 2005), pp. 1–13.
5
‘Battles won, a war still lost ; Drugs in Latin America ’, The Economist, vol. 374 (2005),
pp. 35–6.
6
Washington Office on Latin America (WOLA), ‘ Memorandum to Foreign Policy
Aids-Appropriations. Rethinking Plan Colombia : As Drug Control Policy, Plan
Colombia Doesn’t Measure Up. ’ (10 June 2005), http://www.wola.org/media/June%
20200520FY2006%Approps%20for%20Colombia.pdf
400 Michelle L. Dion and Catherine Russler
into new geographic areas. The balloon-effect can be observed since Plan
Colombia’s commencement, with both a reversed displacement of coca to
surrounding countries and a dispersion and reconstitution of coca-growing
areas within rural Colombia.
7
While 2005 saw an eight per cent increase in
overall cultivation in Colombia, 44 per cent of the fields where coca was
detected in 2005 had never before been identified as coca-producing fields.
8
While the balloon-effect indicates the apparent resilience and mobility of
the coca crop, it lacks explanatory power. What explains the geographic
expansion of coca cultivation within Colombia? What drives coca to be
cultivated in certain geographic areas and not others? What are the requisite
preceding conditions to facilitate its cultivation ? Diverse hypotheses address
the origin and causes of coca production in Colombia, but most are based on
much qualitative, and little quantitative, evidence.
9
In order to answer these research questions, this article reviews existing
explanations of coca cultivation and develops an explanation of coca culti-
vation in Colombia that emphasises the local conditions that promote its
proliferation. This study utilises the most recently published times-series
cross-section data for 32 sub-national departments between 2001 and 2005
to examine explanations of Colombian coca cultivation following the
inception of Plan Colombia in 1999. The results offer support for the re-
allocation of aerial eradication funds toward the development of localised
public and market infrastructures as a long-term strategy for reducing
coca production. The results also suggest that, despite official claims to the
contrary, aerial eradication does not significantly reduce coca cultivation in
Colombia, and that where it does have an impact, this is both temporary and
in part due to displacement associated with fumigation efforts, generating
significant human and economic costs. Also, coca cultivation in Colombia
tends to occur in agricultural departments characterised by limited market
access and state presence. The analysis modifies the general assumption
that extreme poverty is a determinant of coca cultivation in Colombia by
illustrating a parabolic relationship between poverty and coca cultivation
in Colombia, where both very low and extremely high levels of poverty are
associated with low rates of coca cultivation. Taken together, these results
7
Ramı
´rez Lemus, et al., ‘Colombia : A Vicious Circle of Drugs and War’.
8
United Nations Office on Drugs and Crime (UNODC), Colombia Coca Survey (New York
2006).
9
Exceptions are Moreno-Sanchez, et al., ‘ An Econometric Analysis of Coca Eradication
Policy in Colombia’, and Ana Marı
´aDı
´az and Fa´bio Sa´nchez, ‘ A Geography of Illicit
Crops (Coca Leaf) and Armed Conflict in Colombia’, Crisis States Programme Working
Paper no. 47, (London July 2004). The former is a time-series analysis of national culti-
vation rates, and the latter looks at diffusion of cultivation to nearby departments. Neither
study includes the wide range of control variables included here. UNODC reports use
bivariate analyses. See UNODC, Colombia Coca Survey, 2005 and 2006.
Coca Cultivation in Colombia 401
suggest that coca cultivation is most likely in areas with sufficient labour and
land resources for production but that lack sufficient public infrastructure
and state presence to facilitate production for legal markets. The following
section reviews the existing hypotheses regarding the factors that explain
coca cultivation. A description of the data used to describe each variable and
its hypothesised effect precedes a discussion of the results of the statistical
analysis, including the substantive and statistical significance of each variable.
The article concludes with a brief discussion of the policy implications of
our findings.
Explaining Coca Cultivation in Colombia
Challenges in the effective analysis of coca production abound, particularly
in Colombia, due to the labyrinthine and highly politicised nature of the
problem. Anti-coca policies in Colombia have been propelled not by analytic
study of coca cultivation itself, but rather by external political pressures and
counterinsurgency efforts. Debate surrounding the militarisation of forced
eradication through Plan Colombia is inflamed by persistent questions
regarding the overall effectiveness, as well as the unintended social, environ-
mental and health consequences, of herbicide-based eradication.
10
Statistical
testing of the effectiveness of different supply-side programmes has only
recently become possible due to the availability of sub-national data on
cultivation in Colombia, collected through aerial observation over the last
several years. This paper uses department-level cultivation data to model
the theoretical factors, including market, state and socio-economic charac-
teristics, hypothesised to explain sub-national patterns of coca cultivation in
Colombia.
Market theories point to Colombia as a strategic locale for coca pro-
duction. Not only is it located within the Andean coca-producing region near
convenient air and sea routes to North American and European markets, but
it also houses large, isolated internal regions where the presence of the state
is weak.
11
US officials approach coca cultivation in Colombia as the supply
side of a larger market cycle run by narco-traffickers increasingly associated
with Colombia’s illegal armed groups. This cycle, officials from US and
Colombian governments argue, can be forcibly disrupted through tougher
10
Adam Isacson, ‘The US Military in the War on Drugs ’, in Coletta A. Youngers and Eileen
Rosin (eds.), Drugs and Democracy in Latin America: The Impact of US Policy, (London 2005),
pp. 15–60 ; Rosin and Youngers, ‘ The US ‘‘ War on Drugs ’’ : Its Impact in Latin America
and the Caribbean’; Vargas, ‘ The Anti-Drug Policy, Aerial Spraying of Illicit Crops and
Their Social, Environmental and Political Impacts in Colombia’; Ramı
´rez Lemus, et al.,
‘Colombia: A Vicious Circle of Drugs and War’.
11
Francisco E. Thoumi, Illegal Drugs, Economy, and Society in the Andes (Washington DC 2003).
402 Michelle L. Dion and Catherine Russler
mechanisms of law and order.
12
Subscribing to market theories of supply
and demand, US officials assert that worsened profit margins, both real and
perceived, would cause coca growers to abandon the crop and seek alterna-
tive economic endeavours. For example, the US State Department con-
cludes, ‘Crop control is by far the most cost-effective means of cutting
supply. If we destroy crops or force them to remain unharvested, no drugs
will enter the system’.
13
The US government argues that aggressive fumi-
gation of land dedicated to coca should decrease both the net amount of
coca presently cultivated and future attempted cultivation.
Launched bilaterally in 2000, Plan Colombia’s principle strategy for re-
ducing coca cultivation involves aerial fumigation by Direccio
´n Antinarco
´ticos
(DIRAN), Colombia’s anti-narcotics police. In 2005, after five consecutive
years of increases, and with support from the US Embassy, State Depart-
ment and private contractors, the Colombian government sprayed a total
of 138,367 hectares in 18 of Colombia’s coca-growing departments. Not-
withstanding this, the total area under coca cultivation increased by eight
per cent between 2004 and 2005, representing a total of eight per cent of
Colombia’s national territory.
14
The funds and overall resources expended
on combating coca via forced eradication are similarly notable, particularly
those devoted to its security component. Up to 75 per cent of the US$ 4.7
billion in US Plan Colombia funding has been allocated to the Colombian
police and military forces, which in 2000 launched counter-narcotics brigades
to secure the flight paths of fumigation aircraft.
15
In response to eradication efforts, coca growers in Colombia have com-
pensated in unexpected ways. They either grow greater quantities in smaller
but more dispersed plots, or they clear plots in areas even further from
state infrastructure, a reflection of the balloon effect.
16
Furthermore, there
is mounting evidence that the socioeconomic disruption and intensified
violence produced by Colombia’s fumigation strategy displaces thousands
of rural Colombians each year and may indirectly contribute to the sub-
national proliferation of coca to new areas.
17
Displacement refers to the
forced migration of people from their place of residence. While much of
Colombia’s internal displacement – which averaged 281,230 people between
12
Isacson, ‘The US Military in the War on Drugs ’.
13
United States State Department, Bureau for International Narcotics Control and Law
Enforcement, ‘Policy and Program Developments ’, International Narcotics Control
Strategy Report, 1 March 2004, http://www.state.gov/p/inl/rls/nrcrpt/2003/vol1/html/
29829.htm
14
UNODC, Colombia Coca Survey (2006).
15
Ramı
´rez Lemus, et al., ‘Colombia : A Vicious Circle of Drugs and War ’.
16
UNODC, Colombia Coca Survey (2006).
17
Isacson, ‘The US Military in the War on Drugs ’ ; Ramı
´rez Lemus, et al., ‘Colombia : A
Vicious Circle of Drugs and War’; Vargas, ‘The Anti-Drug Policy, Aerial Spraying of Illicit
Crops and Their Social, Environmental and Political Impacts in Colombia.’
Coca Cultivation in Colombia 403
2000 and 2005 – is directly attributable to intimidation and violence initiated
by guerrilla and paramilitary groups, fumigation itself fosters displacement
because it destroys peasants’ subsistence patterns by wiping out both their
illicit and licit crops.
18
Colombia’s nongovernmental Council for Human
Rights and Displacement estimates that in 2001 and 2002 – the two years
following Plan Colombia’s inception – fumigation alone displaced more than
75,000 people nationwide.
19
In addition, Colombia’s militarised eradication
strategies appear to produce intensified displacement in areas targeted
for fumigation as state, guerrilla and paramilitary elements vie for military,
economic and popular control.
20
This suggests that while aerial eradication
may directly reduce coca cultivation in some areas through the eradication
of plants, it may also indirectly reduce cultivation by generating localised
displacement. This article estimates both the direct effects and indirect
effects, via displacement, of aerial spraying on local coca cultivation.
It should be noted that most of Colombia’s displaced do not grow coca
and that the displaced also tend to migrate to its urban centres. Nevertheless,
forced eradication does generate a dispossessed labour force pre-equipped
with coca production know-how. Anecdotal reports indicate that Colombia’s
rural displaced, as a means of survival in their new locales, often engage
in and proliferate the cultivation, harvesting and production of coca.
21
In
addition, there is evidence that elements of Colombia’s floating population
migrate to coca-producing regions specifically in search of the economic
opportunities afforded by the illicit drug industry.
22
These studies suggest
that rural areas receiving people displaced by eradication efforts experience
increased coca cultivation.
18
Estimates of displacement from, Ministerio de la Proteccio´ n Social, ‘ Consolidado de
Programas.’ (2005), http://mps.minproteccionsocial.gov.co/consolidado/buscar.php. On
violence of guerrilla and paramilitary groups contributing to displacement, see Ana Marı
´a
Iba´n
˜ez and Carlos Eduardo Velez, ‘Civil Conflict and Forced Migration : The Micro
Determinates and the Welfare Losses of Displacement in Colombia ’, Documento CEDE
2005-35, Universidad de los Andes (Bogota´, June 2005) and Stefanie Engel and Ana Marı
´a
Iba´n
˜ez, ‘Displacement Due to Violence in Colombia : A Household-Level Analysis ’,
Economic Development and Cultural Change, vol. 55, no. 2 (2007), 335–66. On the effects
of fumigation on displacement, see Ramı
´rez Lemus, et al., ‘Colombia : A Vicious Circle of
Drugs and War’.
19
UNODC, Colombia Coca Survey (2004).
20
Isacson, ‘The US Military in the War on Drugs ’ ; Ramı
´rez Lemus, et al., ‘Colombia : A
Vicious Circle of Drugs and War’. Jennifer S. Holmes et al. find that coca eradication, and
not coca cultivation, appears to contribute to leftist guerrilla violence. Jennifer S. Holmes
et al., ‘Drugs, Violence, and Development in Colombia: A Department-Level Analysis’,
Latin American Politics and Society vol. 48, no. 3 (2006) : pp. 157–84.
21
Ramı
´rez Lemus, et al., ‘Colombia : A Vicious Circle of Drugs and War’ ; UNODC Colombia
Coca Survey (2005); Thoumi, Illegal Drugs, Economy, and Society in the Andes.
22
Rabasa and Chalk, Colombian Labyrinth : The Synergy of Drugs and Insurgency and Its Implications
for Regional Stability.
404 Michelle L. Dion and Catherine Russler
Unfortunately, the aerial eradication strategy implies a failure to fully
appreciate coca’s several comparative advantages as a cash crop within the
regions where it is grown. First, coca is hardier than most other crops ; it
can grow on poor-quality or depleted soil and steeply-sloped terrain and is
relatively resistant to climate variations and pests. Coca plants can provide
initial harvests in as little as six months of planting and from three to six
times per year, which provides a year-round income that seasonal crops
cannot offer. Second, coca requires little in the way of processing infra-
structure and is well suited to low-cost, long-range transport that does not
depend on access to good roads.
23
Beyond Colombia, cross-continental
studies conclude that coca growing regions are some of the most under-
developed agricultural areas of the world, where growers suffer from lack
of market access.
24
In Colombia, coca grows in regions that lack access to
regional markets and cities, where rough terrain and civil conflict have im-
peded the construction of viable roadways.
25
Colombia’s traditional coca-
growing departments of Putumayo, Caqueta and Guaviare are located deep
within the Amazonian jungle, where villages can be several days travel away
from the nearest regional market.
26
Given these characteristics, it is not
surprising that participants of the UNODC’s Andean-wide survey ranked a
lack of viable, stable markets for alternative crops as their number one reason
for rejecting legal agronomic activities.
27
Similarly, econometric studies show
that prices of key alternative crops in Colombia are negatively related to coca
production,
28
and anecdotal evidence supports claims that coca traders, and
increasingly illegal armed groups, offer growers almost guaranteed markets
with stable farm-gate prices and access to credit and seeds.
29
Therefore,
though coca can be grown on marginal land, it is also particularly likely to
be grown in agricultural regions with underdeveloped roads and market
infrastructure where other less hardy and more perishable crops would be
23
Kevin J. Riley, ‘Snow Job? The Efficacy of Source Country Cocaine Policies ’, Graduate
School Dissertation Series RGSD-102, National Defense Research Institute, 1993.
24
UNODC, Alternative Development: A Global Thematic Evaluation – Final Synthesis Report (New
York 2005).
25
Thoumi, Illegal Drugs, Economy, and Society in the Andes.
26
Nancy McGuire, ‘Combating Coca in Bolivia and Colombia : A New Perspective on the
Forces that Drive Peasant Coca Farming ’, Report for Council For Emerging National
Security Affairs, Washington, D.C.
27
UNODC, Alternative Development: A Global Thematic Evaluation – Final Synthesis Report
(New York 2005).
28
Moreno-Sanchez, et al., ‘ An Econometric Analysis of Coca Eradication Policy in
Colombia’.
29
Ana Marı
´aDı
´az and Fa´bio Sa´nchez, ‘ A Geography of Illicit Crops (Coca Leaf) and Armed
Conflict in Colombia’; David Mansfield, ‘ Alternative development : the modern thrust
of supply-side policy’, Bulletin on Narcotics–Occasional Papers, vol. L1, nos. 1 & 2 (1999)
pp. 19–44.
Coca Cultivation in Colombia 405
less successful. Though much of the discussion surrounding coca cultivation
in Colombia emphasises its utility as a cash crop that enhances the ability of
rural Colombians to subsist,
30
quantitative, cross-Colombian studies that
relate the relative availability of licit agricultural markets to coca cultivation
levels are lacking.
In addition to underdeveloped market infrastructure, an underdeveloped
public infrastructure or lack of state presence can also directly contribute to
higher levels of coca cultivation. Up to 40 per cent of Colombia’s territory
remains effectively beyond the control of the state, which has not exercised
a consistent presence throughout its territory.
31
Lack of state presence and
infrastructure can contribute directly to increased coca cultivation by limiting
the range of legal economic activities available to residents.
32
Weak state presence can also contribute to coca cultivation by pro-
viding a vacuum in which illegal armed groups compete for or establish
dominance. In areas beyond effective state control, illegal armed groups
may operate. The principal groups are the leftist Revolutionary Armed
Forces of Colombia (FARC), the smaller National Liberation Army
(ELN), and paramilitary groups created by wealthy land-owners and drug-
traffickers, such as the United Self-Defence Groups of Colombia (AUC).
In areas with weak state presence, the groups fight federal forces and each
other not only for territorial control and access to strategic trafficking
corridors, but for control of the economic resources within their areas
of operation. Often the drug trade finances the activities of illegal armed
groups, and the one strengthens and perpetuates the existence of the other.
The result is a proliferation of illegal armed groups in many of Colombia’s
ungoverned rural regions.
33
Some argue that illegal armed groups should lead to higher rates of coca
cultivation.
34
This is consistent with reports that Colombia’s illegal groups
have, in some regions, substituted for an absent state and imposed their
own regimes, defining their own laws and regulations and providing basic
education, a police force, and civil justice system to solve conflicts among
the population. In exchange for public order, the groups demand popular
allegiance and impose a tax on peasants’ productive activities, including coca
30
UNODC, Colombia Coca Survey (2005).
31
Jennifer S. Holmes, et al., ‘A Subnational Study of Insurgency: FARC Violence in the
1990s’, Studies in Conflict and Terrorism, vol. 30, no. 1 (2007), pp. 249–65.
32
Moreno-Sanchez et al., ‘An Econometric Analysis of Coca Eradication Policy in
Colombia’.
33
Rabasa and Chalk, Colombian Labyrinth : The Synergy of Drugs and Insurgency and Its Implications
for Regional Stability.
34
Moreno-Sanchez et al., ‘An Econometric Analysis of Coca Eradication Policy in
Colombia’.
406 Michelle L. Dion and Catherine Russler
cultivation.
35
Though traditionally Colombia’s illegal armed groups have not
participated directly in the production and supply chain, there is increasing
evidence of their direct involvement in production and trafficking in some
regions.
36
Further, in municipalities with both illegal group presence and
coca cultivation, cultivation appears to be more intense.
37
Therefore, the
absence of a state presence creates an environment that lacks sufficient
opportunities for legal activities and is permissive of illicit activities, includ-
ing those of illegal armed groups. Weak state presence should be directly
associated with higher rates of coca cultivation.
At the same time, when weak state presence permits violence perpetrated
by illegal armed groups intent on establishing economic and political control
of a territory, then conflict-related displacement may interfere with coca
cultivation. For example, studies suggest that weak state presence leads to
higher levels of guerrilla violence.
38
In some areas where narco-traffickers
once paid taxes to guerrillas to maintain a secure operating environment,
the traffickers have instead created their own paramilitaries, leading to esca-
lations in violent conflicts between these and guerrilla forces.
39
Interestingly,
the same studies find that coca cultivation itself is not a significant cause
of guerrilla violence.
40
Consequently, we might expect weak state presence
to be associated with higher levels of displacement and thus lower levels
of coca cultivation.
In addition to such factors as market conditions, state presence, and dis-
placement, Colombia’s colonisation resulted in inefficient bureaucracies,
ill-equipped to govern Colombia’s swaths of rough frontier, where cutting
corners in pursuit of survival or individual advancement became, if not
institutionally encouraged, at least socially tolerated.
41
This environment
of unregulated individualism fostered practices of corruption that allowed
large areas of the country to function independently of centralised control,
and meant that ungoverned regions became, in effect, ungovernable. This
35
Ricardo Vargas, Drug Cultivation, Fumigation and the Conflict in Colombia (Bogota, 1999) ;
Rabasa and Chalk, Colombian Labyrinth : The Synergy of Drugs and Insurgency and Its Implications
for Regional Stability; Thoumi, Illegal Drugs, Economy, and Society in the Andes.
36
Rabasa and Chalk, Colombian Labyrinth : The Synergy of Drugs and Insurgency and Its Implications
for Regional Stability; Thoumi, Illegal Drugs, Economy, and Society in the Andes.
37
UNODC, Colombia Coca Survey (2005); Dı
´az and Sa´nchez, ‘ A Geography of Illicit Crops
(Coca Leaf) and Armed Conflict in Colombia ’.
38
Holmes et al., ‘ Drugs, Violence, and Development in Colombia : A Department-Level
Analysis’; Holmes, et al., ‘ A Subnational Study of Insurgency : FARC Violence in the
1990s’.
39
Holmes et al., ‘A Subnational Study of Insurgency: FARC Violence in the 1990s ’.
40
Holmes et al., ‘Drugs, Violence, and Development in Colombia : A Department-Level
Analysis’; Holmes, et al., ‘ A Subnational Study of Insurgency : FARC Violence in the
1990s’.
41
Thoumi, Illegal Drugs, Economy, and Society in the Andes.
Coca Cultivation in Colombia 407
culture of corruption prevails throughout Colombia’s coca-growing regions,
further facilitating the expansion of cultivation.
42
Finally, poverty and illicit crop production tend to overlap, and inter-
national studies describe illicit cultivation areas as characteristically poor,
where indicators of malnutrition, infant mortality and illiteracy are consist-
ently and substantially higher than national averages. Specifically, there tends
to be an Andean-wide association between the size of landholdings and
the proportion of land dedicated to coca cultivation, where those with the
least amount of land were found to cultivate the largest proportion of their
holdings with coca.
43
An estimated 60 per cent of Colombia’s coca crops
are on small landholdings of less than two hectares, indicating that coca
cultivation in Colombia is part of a wider cropping pattern aimed at self-
sufficiency.
44
The UNODC asserts that while growing coca does not sig-
nificantly increase the income of rural Colombians, of whom 85 per cent live
below the international poverty line, it can serve to improve their basic
subsistence in the absence of other income generating activities.
45
However, in a department-level analysis, the UNODC found no statisti-
cally significant correlation between poverty and coca cultivation within
Colombia.
46
The study notes that departments like Guainı
´a and Choco,
where poverty levels were high, were not the departments with high levels
of coca cultivation in 2004, whereas the department of Meta, with relatively
fewer people below the poverty line, accounted for a significant portion
of Colombia’s coca cultivation. Though the UNODC found no linear
relationship, its findings do not preclude a curvilinear relationship between
poverty and coca cultivation. Coca cultivation may be limited in areas with
very little poverty and become more widespread as poverty increases.
However, it is not likely that areas with very high concentrations of
poverty will have sufficient resources to sustain coca production. Therefore,
we might reasonably expect the relationship between poverty and coca
cultivation to flatten at the highest levels of poverty.
This section has reviewed a variety of factors that are likely to explain
the sub-national pattern of coca cultivation in Colombia, including aerial
eradication, displacement, corruption, state presence, economic develop-
ment, agricultural activity, access to agricultural markets, and poverty.
42
Francisco E. Thoumi, ‘ Why the Illegal Psychoactive Drugs Industry Grew in Colombia ’,
Journal of Interamerican Studies and World Affairs, vol. 34, no. 3 (1992), pp. 37–63 ;
Thoumi, Illegal Drugs, Economy, and Society in the Andes.
43
UNODC, Alternative Development: A Global Thematic Evaluation – Final Synthesis Report;
Mansfield, ‘Alternative development : the modern thrust of supply-side policy ’.
44
Dı
´az and Sa´nchez, ‘ A Geography of Illicit Crops (Coca Leaf) and Armed Conflict in
Colombia’.
45
UNODC, Colombia Coca Survey (2005).
46
UNODC, ibid.
408 Michelle L. Dion and Catherine Russler
The following section articulates the specific hypotheses to be tested and
describes the data used in the statistical analysis that follows.
Hypotheses and Data
Colombia is divided into 32 administrative territorial departments. The time-
series, cross-section dataset consists of department level, annual obser-
vations for each independent variable from 2000 to 2004, and the dependent
variable, coca cultivation, from 2001–2005.
47
Descriptions of the variables
and their hypothesised effects on cultivation can be found in Table 1.
The analysis begins in 2001 because this is the first year that full national
coverage was achieved from UNODC and the Colombian government’s
use of satellite imagery and verification flights over coca growing areas
to monitor the location and spread of coca cultivation. In 2005, the area
within each department with active coca cultivation was between 28 and
17,305 hectares, with nine departments having no reportable levels of coca
cultivation.
48
Because the urbanisation and territory of each department
Table 1. Variable Definitions and Hypothesised Effects
Dependent variable Description
Coca Cultivation Square kilometres of coca under cultivation in each
department, divided by the rural population
Independent variables Description
Impact
On DV
Aerial eradication Percentage of department land area aerially
fumigated
x
State presence Public administration and community services
output (in thousands of constant pesos) divided by
the population
+
Displaced Number of persons displaced from department of
origin divided by the population
x
Incoming displaced Number of displaced persons resettling in
destination department divided by the population
+
Corruption Department government corruption index, 0 to 100 +
Licit Agriculture Licit agricultural output (in thousands of constant
pesos) divided by rural population
+
GDP per capita Department gross domestic product per capita
(in thousands of constant pesos)
x
Market access Paved roads as percentage of all roads x
Poverty Percentage of population with unsatisfied basic
needs
+/x
47
Colombia’s federal district, Bogota´, is excluded from analysis. Annual data for roadways,
government corruption, and poverty are not collected or published.
48
See UNODC, Colombia Coca Survey (2005) for data and a discussion of measurement issues.
Coca Cultivation in Colombia 409
differ, we calculate per capita cultivation rates by dividing the annual square
kilometres under cultivation within each department by its rural popu-
lation.
49
Aerial fumigation by DIRAN is the Colombian government’s primary
method for reducing coca cultivation.
50
In 2000, it sprayed 58,498 hectares
in 11 of the country’s previously identified coca-growing departments.
Coverage has expanded annually under Plan Colombia, and in 2004 targeted
over 136,552 hectares across 16 departments. Aerial eradication is the per-
centage of department land area aerially fumigated by DIRAN.
51
The direct
effect of aerial fumigation on coca cultivation should be negative.
Coca cultivation appears to occur in communities beyond the effective
reach of the state because weak state presence creates permissive conditions
that enable illegal activity.
52
In Colombia, department level governments
have fiscal and operational responsibility for basic services and infra-
structure. State presence in each department, calculated by dividing public
administration and community services output (in thousands of constant
pesos) by population, is hypothesised to be negatively associated with coca
cultivation.
53
Aerial eradication efforts and lack of state presence are also expected to
affect coca cultivation indirectly, through their effects on displacement.
Eradication contributes to displacement through its disruption of both legal
and illegal cultivation and the violence associated with military efforts to
secure areas for fumigation. Weak state presence contributes to displacement
when illegal armed groups compete for control of remote areas. The
Colombian government works with non-governmental agencies and the
Catholic Church to compile the Sistema Unico de Registro (SUR), a database of
49
Colombia’s departments range from 24.1% to 93.7% urban. Joshua D. Angrist and
Adriana D. Kugler, ‘Rural Windfall or a New Resource Curse ? Coca, Income and Civil
Conflict in Colombia’, National Bureau of Economics Research Working Paper No.
11219, Cambridge, MA, March 2005. Population data come from the Colombian
Government’s statistical database, the Departamento Administrativo Nacional de Estadistica
(DANE), http://www.dane.gov.co. Unless otherwise stated, all data used in this study are
retrieved from DANE.
50
Accumulated sprayed area is the sum of areas sprayed during a given time period, calcu-
lated by multiplying the length of flight lines by their width. It does not take into account
effective sprayed area, which disregards the overlap between adjacent sprayed bands and
areas sprayed several times in the same calendar year. UNODC, Colombia Coca Survey (2005).
51
Aerial eradication data from UNODC, Colombia Coca Survey (2006).
52
Though weak state presence may facilitate coca cultivation by enabling illegal armed groups
to protect or promote cultivation, it is difficult to directly measure the magnitude of armed
group presence without conflating this with violence. Typically, human rights violations or
other estimates of violence are used to measure armed group presence. However, these
measures should actually be associated with less cultivation due to the displacement and
disruptions of cultivation associated with such violence.
53
DANE, http://www.dane.gov.co/files/investigaciones/pib/departamentales/
410 Michelle L. Dion and Catherine Russler
internally displaced persons.
54
Federal law defines database entrants as
migrants forced to abandon their physical residences and employment
activity by armed conflict, generalised violence, massive human rights viol-
ations or other circumstances that threaten or drastically alter public order.
In describing internal displacement, SUR distinguishes between departments
where original displacement occurs, and the departments where displaced
persons relocate. This distinction is analytically important, because dis-
placement reflects the existence of a migratory cause. In areas with high
displacement, we would expect cultivation to decline due to the disruption of
agricultural activities, both licit and illicit, and of the local labour market.
Reception rates, in contrast, reflect the presence of floating populations that
seek employment. Some argue that the presence of displaced populations
is positively associated with coca cultivation, although this correlation is
weakened since displaced populations often flee to urban destinations.
55
For
this study, ‘displaced ’ measures the ratio of the annual number of displaced
persons to the total population in the department of origin. ‘Incoming dis-
placed’ measures the ratio of annual displaced persons to the total popu-
lation in the department of settlement.
56
In addition, widespread public corruption in Colombia encourages un-
regulated individualism among the civilian population, which facilitates the
expansion of the coca industry within Colombia.
57
Transparency Inter-
national assessed Colombia’s department-level governments using a cor-
ruption index scaled 0–100 with weighted categorical variables that measure
internal and external transparency in fiscal management, adherence to op-
erational mandates, and the responsiveness of Colombia’s department-level
governments. Higher outcomes on the corruption index are hypothesised
to be positively associated with higher rates of coca cultivation.
54
Presidencia Repu
´blica de Colombia,‘Registro U
´nico de la Poblacio´ n Desplazada ’, http://
www.red.gov.co/programas/apoyo_integral_ desplazados/estadisticas.htm
55
See Ramı
´rez Lemus, et al., ‘Colombia : A Vicious Circle of Drugs and War ’ ; UNODC
Colombia Coca Survey (2005); Thoumi, Illegal Drugs, Economy, and Society in the Andes ; and
Rabasa and Chalk, Colombian Labyrinth: The Synergy of Drugs and Insurgency and Its Implications
for Regional Stability.
56
The SUR database probably underestimates the true number of displaced, in part because
the government uses a more restrictive definition of displacement than that of NGOs. On
the other hand, the government’s SUR database has the advantage of tracking, in addition
to the department in which the displaced settle, the department from which people are
displaced. This allows us to disaggregate the effects of displacement in both sending and
receiving departments. CODHES, a Colombian NGO, only publishes data on the de-
partments that receive the displaced. Using the government figures reflects the more
conservative approach because any effects we find would be more pronounced were more
accurate data available. See Internal Displacement Monitoring Centre, ‘Colombia:
government ‘peace process ’ cements injustices for IDPs’, Norwegian Refugee Council,
Geneva, Switzerland, 30 June 2006.
57
Thoumi, ibid.
Coca Cultivation in Colombia 411
Coca is also more likely to be cultivated in agricultural regions, especially
when controlling for a department’s level of economic development. Even
though coca can be cultivated on marginal land, agricultural areas have
the resources necessary for cultivation, especially in comparison to more
urbanised areas. Licit agricultural production is measured by the department-
level gross domestic product (GDP) in thousands of constant pesos derived
from agricultural activities, divided by the rural population. Level of econ-
omic development is measured as departmental per capita gross domestic
product (GDP) in thousands of constant pesos.
58
Even when controlling for
both economic development and agricultural production, coca cultivation
should be more likely in areas that lack agricultural market access. Market
access is measured in terms of paved roads as a percentage of all intra-
departmental roads.
59
While it would be useful to separate structural and income indicators
of poverty, economic indicators in terms of income levels, unemployment
and underemployment in Colombia are not published at the department
level.
60
DANE utilises the Unsatisfied Basic Needs (UBN) Index to measure
poverty at the department level. The indicator, ranging from zero to 100,
is the percentage of households for which one or more of poverty indi-
cators are present. The poverty indicators include the materials with which
the house is made, water and sewage services, degree of economic depen-
dence, number of people per square metre, and school attendance of at least
one child between the age of seven and eleven.
61
While poverty is generally
assumed to be positively associated with coca cultivation, sub-national
studies have experienced difficulty in establishing a statistically significant,
linear relationship between poverty and coca cultivation.
62
Because coca
cultivation may be less common at both very low and very high levels of
poverty, the statistical analysis will test whether an inverted-U relationship
exists between poverty and coca cultivation remains when controlling for
other variables.
58
For department level GDP and population statistics, see DANE, http://www.dane.
gov.co/
59
Ministerio de Transporte, ‘Transporte en cifras ’, http://www.mintransporte.gov.co/
Servicios/Estadisticas/TABLASYGRAFICOS2004.htm.
60
For a discussion of these data issues, see Holmes et al., ‘ Drugs, Violence, and
Development in Colombia: A Department-Level Analysis’. Data to test hypotheses that
focus on factors of inequality, including the potential association between coca cultivation
and land ownership, prove similarly elusive.
61
The Colombian government is in the process of adopting a new measure of poverty based
on data projections. The UPN and this new measure, the System for Selecting Beneficiaries
of Social Services (SISBEN), are highly correlated.
62
UNODC, Colombia Coca Survey (2005).
412 Michelle L. Dion and Catherine Russler
Model Results
The relationships between the variables described above and coca cultivation
in Colombia are analysed using a Prais-Winsten regression model estimated
with panel corrected standard errors and fixed effects. Pooling the data
increases estimation efficiency by increasing the observable cases from 32 to
160, but pooled data often violate the assumptions of ordinary least square
(OLS) regression, and as a result can exhibit panel heteroscedasticity and
autocorrelation. Estimating the model with heteroscedastic panel corrected
standard errors accounts for panel heteroscedasticity.
63
Autocorrelation of
the errors was addressed by estimating the model using a Prais-Winsten
transformation with a common autoregressive parameter.
64
Panel fixed
effects model were included to account for unobserved differences across the
Colombian departments.
65
63
Nathaniel Beck and Jonathan N. Katz, ‘ What to do (and not to do) with Time-Series
Cross-Section Data’, American Political Science Review, vol. 89, no. 3 (1995), pp. 634–47. Beck
and Katz’s panel corrected standard errors (PCSEs) typically perform better when the
number of time points exceed the number of cross-sections. In this instance, the PCSEs
are nearly identical to uncorrected standard errors using OLS. Limited data and a large
number of parameters preclude estimation of accurate generalized least squares models
with complex corrections for assumed error structures.
64
The other modelling alternative to deal with autocorrelation would entail estimating the
model using OLS and including a lagged dependent variable. In this case, the lagged
dependent variable approach would be problematic or unnecessary. The lagged dependent
variable approach would use up scarce degrees of freedom and convert the analysis into
one of short-term change, despite our theoretical interest in cross-department variation.
Further, the estimates of the autocorrelation parameter (rho) in the Prais-Winsten models
suggests that autocorrelation within panels is not severe. Further, a lagged dependent
variable is problematic with the inclusion of fixed effects, creating additional bias beyond
that normally expected in such models.
65
Two tests were used to determine whether fixed effects were appropriate, and both in-
dicated that estimating the model with fixed effects is necessary. First, a Chow or F test of
the joint significance of the coefficients for the fixed effects generated statistically signifi-
cant Chow scores (see Table 2). In addition, the modified jack-knife procedure rec-
ommended to measure the mean absolute prediction error for each department, also
confirmed the utility of estimating the model with fixed effects. The jack-knife procedure
estimates a separate prediction model for each department by leaving out one department
at a time and using the model to predict coca cultivation in the omitted department. The
mean absolute error is the difference between the predicted and observed coca cultivation
for each department. The results of the test are presented in the Appendix. See Nathaniel
Beck. ‘Time-Series-Cross-Section Data : What Have We Learned in the Past Few Years? ’,
Annual Review of Political Science, vol. 4 (2001) : pp. 271–93.Fixed effects are also preferable to
random effects because it is unlikely that the intercept shifts are due to random error
uncorrelated with the independent variables, an assumption required of random effects
models. Instead, the intercepts shifts are likely to reflect systematic differences in the
average level of coca cultivation across departments rather than random error. We per-
formed a Hausman test, which indicated that the random effects were not more efficient
than fixed effects. The estimates of the random effects models were inconsistent. We
accept the loss of efficiency in the fixed effects models in favour of greater consistency.
Coca Cultivation in Colombia 413
Table 2 presents the results for two models, one with poverty as a linear
predictor of cultivation (Model 1) and a second with a non-linear relationship
between poverty and cultivation (Model 2). The results for both models are
nearly identical, and the discussion below will focus on those for Model 2.
Standardised coefficients are also presented to facilitate comparisons of
explanatory power across different independent variables.
According to the results, aerial eradication has a small and statistically
insignificant direct effect on coca cultivation in Colombia. The small size
of the substantive direct effect on coca cultivation is clear because aerial
eradication has the smallest standardised coefficient of the model. In ad-
dition, both displacement and state presence have substantively and statisti-
cally significant dampening effects on coca cultivation, according to the
Table 2. Colombian department-level coca cultivation, 2001–2005
Model 1 Model 2
bStandardized bbStandardized b
(PCSE) (PCSE)
Aerial eradication
tx
1
x0.048 x0.001 x0.048 x0.001
(2.049) (2.049)
State presence
tx
1
x4.956*** x0.172 x4.956*** x0.172
(1.706) (1.706)
Displaced
tx
1
x133.647** x0.270 x133.647** x0.270
(57.195) (57.195)
Incoming displaced
tx
1
7.785 0.062 7.785 0.062
(9.255) (9.255)
Corruption
tx
1
0.108 0.128 0.120 0.143
(0.086) (0.085)
Licit agriculture
tx
1
4.736*** 0.610 4.736*** 0.610
(1.027) (1.027)
Economic x0.001 x0.143 x0.001 x0.143
development
tx
1
(0.000) (0.000)
Market access
tx
1
x0.203*** x0.650 x0.302*** x0.964
(0.047) (0.085)
Poverty
tx
1
0.188*** 0.390 0.756** 1.569
(0.071) (0.312)
Poverty
2
tx
1
x0.007** x1.815
(0.003)
Rho 0.111 0.111
Wald x
2
(df) 492.82 (38)*** 492.82 (38)***
Chow test for f.e. (df) 249.23 (29)*** 190.47 (28)***
R
2
0.9094 0.9094
Observations 160 160
Departments 32 32
Note: Prais-Winsten regression with common AR-1 autocorrelation estimate (rho), fixed
effects (not shown), and panel corrected standard errors (in parentheses). Wald x
2
and R
2
reported by Stata 9, using xtpcse command with common AR(1) option.
Key:*p<0.10, ** p<0.05, *** p<.001, two-tailed tests.
414 Michelle L. Dion and Catherine Russler
model results in Table 2. These findings are consistent with expectations that
displacement disrupts agricultural activity and labour markets and thus coca
cultivation, and that a strong state presence precludes extensive illegal
activities, including coca cultivation.
To better understand the complex relationships among aerial eradication,
state presence, displacement, and coca cultivation, we also estimated a lim-
ited model of displacement. The category ‘displaced ’ was regressed on aerial
eradication and state presence to estimate the role of these two variables on
the rate of displacement. Figure 1 illustrates the path relationships among
eradication, state presence, displaced population and coca cultivation. The
path coefficients are the standardised betas from Model 2 and the model
of displacement described above.
This exercise confirms that though aerial eradication has only a mar-
ginal direct effect on coca cultivation (x0.001, p>0.100), it has a significant
effect on displacement (0.296, p<0.001) and thus does contribute indirectly
to local reductions in coca cultivation. While the direct effect of aerial
eradication on coca cultivation is modest (x0.001), the indirect effect via
displacement is larger (x0.080, the product of 0.296 and x0.270). The total
effect of aerial eradication on coca cultivation (x0.081, or x0.001 plus
x0.080) is still modest, compared to some of the other variables in Model 2.
The model suggests that much of the effect of aerial eradication on coca
cultivation occurs through displacement of the population, which implies
significant and unintended human and economic costs of the eradication
policy.
66
At a minimum, the result suggests that the effectiveness and ex-
ternalities of aerial eradication should be more closely examined.
Coca
cultivation
Aerial eradication
State
p
resence
Displaced
-0.001
0.296***
-0.270***
0.035
-0.172***
Fig. 1. Relationships among aerial eradication, state presence, displacement and coca cultivation. Note :
Standardised betas from Table 2, Model 2 and regression of displaced on aerial eradication and state
presence. *** p<0.001, two-tailed test.
66
Studies of the effect of displacement on welfare suggest that it reduces by a third the
consumption of the displaced. Ana Maria Iba´n
˜ez and Carlos Eduardo Velez, ‘ Civil Conflict
and Forced Migration: The Micro Determinates and the Welfare Losses of Displacement
in Colombia’.
Coca Cultivation in Colombia 415
Figure 1 also provides some insight into the relationship between
state presence and coca cultivation. A strong state presence clearly directly
reduces the extent of coca cultivation (x0.172, p<0.001).
67
Meanwhile,
state presence does not have a significant direct effect on displacement
(0.035, p>0.100). This finding suggests that weak state presence does not
necessarily correspond to increases in displacement due to violence between
illegal armed groups. Instead, a strong state presence is important because it
directly reduces the intensity of coca cultivation in a department.
Overall, the simple path model results illustrated by Figure 1 suggest
that future studies should more closely examine the complex relationships
between eradication, state presence, displacement, and coca cultivation. In
particular, the results suggest that aerial eradication affects coca cultivation
mainly through displacement, which entails significant human and economic
side effects, and that a strong state presence directly reduces the intensity of
coca cultivation, though state presence has little impact on displacement.
Most of the remaining indicators of state or market context are consistent
with existing theoretical expectations, although not all of the variables have
statistically significant effects on coca cultivation. For example, although
areas that receive displaced populations do have slightly higher coca culti-
vation rates, even when controlling for departmental level of economic
development, this effect is not statistically significant.
68
Likewise, coca culti-
vation is greater in departments with higher levels of corruption and lower
levels of economic development, though these effects are not statistically
or substantively very significant. Coca cultivation is significantly higher,
however, in agricultural regions, even after controlling for poverty and
level of economic development. Not surprisingly, extent of legal agricultural
production is one of the better predictors of coca cultivation because many
of the resources necessary to produce agricultural goods for legal markets are
the same as those necessary to produce for illegal markets: arable land, ap-
propriate climate and sufficient labour. At the same time, not all agricultural
67
Though we do not model the presence of illegal groups directly for reasons discussed
above, the direct effect of state presence on cultivation is consistent with studies arguing
that illegal armed group presence is associated with coca cultivation.
68
We also estimated the models using the CODHES data for departments receiving the
displaced. The results for all of the variables, except incoming displaced and aerial eradi-
cation, were nearly identical in terms of substantive and statistical significance to the
models reported in Table 2. In the case of incoming displaced, the effect on coca culti-
vation remained positive but was smaller and even less statistically significant, which
suggests that the displaced do not engage in significant coca cultivation. The coefficient for
aerial eradication became larger but remained insignificant. That substituting the
CODHES data for that of SUR only affected the coefficients of these two variables (and
even then they remained statistically insignificant), suggests that the SUR figures do indeed
under count those that are displaced by aerial eradication. Therefore, our results are
conservative.
416 Michelle L. Dion and Catherine Russler
areas have sufficient infrastructure to integrate producers into legal markets.
Given the hardiness of coca as a cash crop, it is ideal to overcome barriers to
market entry, including long distances or poor public infrastructure. The
model supports this argument because coca cultivation is higher in areas
with fewer paved roads, or poor access to legal markets for producers.
Even controlling for level of economic development and state presence, the
percentage of roads that are paved in a department has a substantively and
statistically significant effect on reducing coca cultivation. This suggests that
improving public infrastructure to provide agricultural producers with
greater access to legal markets may be a useful strategy for combating coca
cultivation in agricultural regions.
Finally, the models suggest that the relationship between poverty and coca
cultivation may be more complex than is often assumed. The statistical
analysis suggests that poverty has a statistically and substantively significant
inverted-U relationship with coca cultivation rates, where very low and ex-
tremely high levels of poverty are associated with little coca cultivation.
69
Figure 2 presents the predicted cultivation rates according to Model 2, at
0
2
4
6
20 25 30 35 40 45 50 55 60 65 70 75 80 85
Poverty
Predicted cultivation
(km2 per capita)
Fig. 2. Predicted effect of poverty on coca cultivation with 95% confidence interval. Note : Simulated results
estimated using CLARIFY in Stata with regress command with fixed effects. The model results were nearly
identical to those reported for Model 2 in Table 2.
69
Though the linear effect of poverty in Model 1 is statistically significant, the coefficient in
Model 1 is much smaller than its counterpart in the quadratic model. The quadratic term in
Model 2 is statistically significant. The results are consistent with an interpretation that the
coefficient of the linear parameter in Model 1 is biased downward because of omitted
variable bias. The jack-knife procedure confirms that the curvilinear model of poverty
better predicts actual coca cultivation than poverty as a simple, linear predictor. See the
Appendix.
Coca Cultivation in Colombia 417
levels of poverty found in the sample. The figure is based on simulations
using CLARIFY, to account for sampling variability.
70
Indeed, when poverty
exceeds 80 on the poverty index, as it does in some Colombian departments,
the model predicts no coca cultivation. Coca cultivation is highest in regions
of moderate poverty, peaking in those with UBN scores between 50 and
60. This suggests that regions with low rates of poverty will be less likely
to engage in significant coca cultivation and also that regions with extremely
high rates of poverty will also be too resource poor to have sufficient
resources to engage in coca cultivation. From a policy standpoint, this re-
lationship creates a conundrum because efforts to reduce poverty may have
an untended consequence of facilitating the expansion of coca cultivation.
This is why it is important to understand the myriad factors that contribute
to regional patterns of coca cultivation so that coherent policies can be
implemented not only to reduce cultivation but also reduce poverty and
promote economic development.
Conclusions
The model developed in this article elucidates and reveals some of the
complex factors explaining patterns of coca cultivation in Colombia since the
implementation of Plan Colombia in 1999. The preceding analysis presents
several findings, each with important policy implications for Plan Colombia
and long-term strategies for reducing coca cultivation in Colombia.
First, the analysis illustrates the complex relationships existing between
aerial eradication, displacement and coca cultivation. Aerial eradication, the
centrepiece of Plan Colombia’s efforts to stymie Colombian coca exports,
has a small overall impact on coca cultivation, compared to other variables.
Furthermore, the effects of aerial eradication on coca appear to derive not
directly from the fumigation of the plant itself, but rather from the associated
side effects of fumigation, namely, violence and indiscriminate disruption of
agriculture in coca growing regions. That aerial eradication principally affects
coca by generating significant levels of displacement suggests that the un-
intended human and economic costs of aerial eradication should be explicitly
considered and addressed by policy makers. The human and economic costs
of displacement due to aerial eradication may only perpetuate the poverty
and underdevelopment common to agricultural regions already most likely
to grow coca, creating an on-going cycle and pattern of transient coca cul-
tivation. At a minimum, aerial eradication policies should be complemented
70
Michael Tomz, Jason Wittenberg and Gary King, ‘ CLARIFY : Software for Interpreting
and Presenting Statistical Results’, Version 2.0, Cambridge, MA : Harvard University,
1 June 2001.
418 Michelle L. Dion and Catherine Russler
with sufficient funding to mitigate the human and economic costs and
reduce the likelihood that eradication will continue to generate a sub-national
balloon-effect.
Second, the results create a dilemma for policy makers hoping to reduce
coca cultivation by reducing localised poverty and marginalisation. Because
coca cultivation is concentrated in areas of moderate poverty and is seldom
found in areas of extremely low or high poverty, efforts to address extreme
poverty might have the unintended consequence of facilitating coca culti-
vation. The answer is not to abandon areas in extreme poverty, but to ensure
that poverty alleviation programs simultaneously address the other factors
that make coca cultivation attractive. For example, poverty alleviation pro-
grams should simultaneously address local public infrastructure and access
to agricultural markets to reduce the likelihood that extreme poverty is
replaced by coca cultivation.
Finally, the results suggest that coca cultivation thrives in economically
underdeveloped, agricultural regions where residents have the means to
partake in agricultural endeavours but lack access to legal markets due to
poor public infrastructure and a weak state presence. The policy implication
is that establishing strong local state capacity, reducing corruption through
improved accountability and transparency, and investing in public infra-
structure to support local access to legal agricultural markets should all help
reduce coca cultivation. Unfortunately, these are all policies that have been
underdeveloped or under-funded as part of Plan Colombia. Securing long-
term reductions in coca cultivation will require the strategic yet localised
funding, development and oversight of public and market infrastructures
that emphasise both accessibility and sustained viability. Given Colombia’s
large expanse of relatively ungoverned territory, institutionalised solutions
that foster government and communal accountability are clearly needed,
and Colombia’s nascent efforts in this terrain warrant internationalised
analytical, financial and logistical attention. Policy formulators should ac-
knowledge that the surest investments are not necessarily those of immediate
returns.
This is the principle argument of alternative development proponents
who seek increased funding for localised projects (UNODC, 2005). During
1999–2007, US$350 million was allocated to national level alternative
development activities, and the annual alternative development activities
implemented at the municipal and departmental levels increased from
US$3 million in 2000 to US$78 million in 2004.
71
These funding levels are
71
See UNODC, Colombia Coca Survey (2005). UNODC was contacted for this study, and
annual department-level allocation or receipt of alternative development funds have yet to
be compiled.
Coca Cultivation in Colombia 419
small relative to funding for military and eradication efforts. Increasing
emphasis on alternative development programmes to combat coca culti-
vation acknowledges the shortcomings of forced eradication as a long-term
policy. However, the inconsistent development, implementation, oversight
and dispersal of funds for alternative development projects since Plan
Colombia’s commencement has severely impeded both their potential
effectiveness and the availability of data on funding dispersal and project
results.
72
Though the effect of alternative development projects on local and
national coca cultivation levels may be difficult to monitor because their
implementation and results take longer to mature, our findings suggest that
such programmes, by addressing the structural sources of coca cultivation,
may have a longer lasting effect on cultivation in Colombia than short-term
strategies to fumigate current cultivation.
72
Ramı
´rez Lemus, et al., ‘Colombia : A Vicious Circle of Drugs and War ’ ; Isacson, ‘The US
Military in the War on Drugs.’
420 Michelle L. Dion and Catherine Russler
Appendix: Jack-knife Test Results
Department
Actual avg.
cultivation*
Model 1 Model 2
Predicted
cultivation Difference
Predicted
cultivation Difference
Atla´ntico 0.000 x13.386 13.386 x11.983 11.983
Casanare 0.000 x4.697 4.697 x6.729 6.729
Cesar 0.000 x2.423 2.423 x1.009 1.009
Huila 0.000 3.098 x3.098 8.634 x8.634
Quindio 0.000 16.092 x16.092 9.402 x9.402
Risaralda 0.000 x1.254 1.254 2.716 x2.716
San Andre´ s 0.000 x16.757 16.757 x15.473 15.473
Sucre 0.000 2.608 x2.608 x0.292 0.292
Tolima 0.000 x4.502 4.502 x2.728 2.728
Cundinamarca 0.007 x0.034 0.041 3.018 x3.011
Valle 0.017 x9.157 9.173 x8.919 8.936
Boyaca´ 0.018 x2.967 2.985 2.495 x2.478
Caldas 0.037 x7.734 7.771 x6.736 6.773
Santander 0.145 x4.205 4.351 x0.034 0.180
Magdalena 0.159 5.592 x5.433 10.192 x10.033
Choco´ 0.207 9.668 x9.461 6.964 x6.756
Cordoba 0.232 x2.708 2.940 x4.891 5.123
Guajira 0.274 x2.741 3.016 x3.689 3.963
Cauca 0.344 7.210 x6.865 13.318 x12.973
Antioquia 0.363 x7.804 8.167 x5.857 6.220
Bolı
´var 0.845 x8.622 9.467 x8.268 9.114
N. Santander 1.673 x3.954 5.627 0.128 1.545
Narin
˜o 1.876 x0.107 1.983 2.110 x0.234
Arauca 3.579 x8.013 11.592 x4.192 7.771
Amazonas 3.866 x1.128 4.995 3.425 0.441
Caqueta´ 4.952 x1.367 6.319 1.534 3.418
Meta 6.882 10.703 x3.821 16.917 x10.036
Guainı
´a 9.087 56.494 x47.407 43.046 x33.959
Vaupe´s 9.210 22.996 x13.786 8.316 0.894
Putumayo 12.237 x2.003 14.240 24.804 x12.567
Vichada 22.054 27.515 x5.460 20.349 1.705
Guaviare 47.926 22.903 25.023 27.489 20.437
Average 1.414 0.059
Note: *Cultivation in km
2
per capita.
Coca Cultivation in Colombia 421
