Greed and Grievance in Civil War
Author(s): Paul Collier and Anke Hoeffler
Source: Oxford Economic Papers , Oct., 2004, Vol. 56, No. 4 (Oct., 2004), pp. 563-595
Published by: Oxford University Press
Stable URL: https://www.jstor.org/stable/3488799
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to
Oxford Economic Papers
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
( Oxford University Press 2004 Oxford Economic Papers 56 (2004), 563-595 563
All rights reserved doi:10.1093/oep/gpf064
Greed and grievance in civil
By Paul Collier* and Anke Hoefflert
*Centre for the Study of African Economies, Unive
tCentre for the Study of African Economies, Unive
Road, Oxford OX2 6NA; e-mail: anke.hoeffler@econ
We investigate the causes of civil war, using a new data se
Rebellion may be explained by atypically severe grievance
a lack of political rights, or ethnic and religious divisions
it might be explained by atypical opportunities for build
While it is difficult to find proxies for grievances and op
political and social variables that are most obviously relate
explanatory power. By contrast, economic variables, which
ances but are perhaps more obviously related to the viabi
considerably more explanatory power.
1. Introduction
Civil war is now far more common than international conflict: all of the 15 major
armed conflicts listed by the Stockholm International Peace Research Institute for
2001 were internal (SIPRI, 2002).
In this paper we develop an econometric model which predicts the outbreak
of civil conflict. Analogous to the classic principles of murder detection, rebellion
needs both motive and opportunity. The political science literature explains
conflict in terms of motive: the circumstances in which people want to rebel are
viewed as sufficiently rare to constitute the explanation. In Section 2 we contrast
this with economic accounts which explain rebellion in terms of opportunity:
it is the circumstances in which people are able to rebel that are rare. We discuss
measurable variables which might enable us to test between the two accounts
and present descriptive data on the 79 large civil conflicts that occurred
between 1960 and 1999. In Section 3 we use econometric tests to discriminate
between rival explanations and develop an integrated model which provides a
synthesis. Section 4 presents a range of robustness checks and Section 5 discusses
the results.
This analysis considerably extends and revises our earlier work (Collier and
Hoeffler, 1998). In our previous theory, we assumed that rebel movements incurred
net costs during conflict, so that post-conflict pay-offs would be decisive. The
core of the paper was the derivation and testing of the implication that high
post-conflict pay-offs would tend to justify long civil wars. We now recognize that
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
564 P. COLLIER AND A. HOEFFLER
this assumption is untenable: rebel groups often more than co
the conflict. Here we propose a more general theory which
tunities for rebellion against the constraints. Our previo
conflated the initiation and the duration of rebellion. We now
This paper focuses on the initiation of rebellion.' Our sampl
cross-section analysis of 98 countries during the period 196
sive coverage of 750 five-year episodes over the period 1960
analyse double the number of war starts. Further, we expand
variables to a more extensive coverage of potential determinan
ness to select a preferred specification.
2. Rebellion: approaches and measures
2.1 Preferences, perceptions, and opportunities
Political science offers an account of conflict in terms of motive: rebellion occurs
when grievances are sufficiently acute that people want to engage in violent protest.
In marked contrast, a small economic theory literature, typified by Grossman
(1991, 1999), models rebellion as an industry that generates profits from looting,
so that 'the insurgents are indistinguishable from bandits or pirates' (Grossman,
1999, p.269). Such rebellions are motivated by greed, which is presumably sufficiently
common that profitable opportunities for rebellion will not be passed up.2
Hence, the incidence of rebellion is not explained by motive, but by the atypical
circumstances that generate profitable opportunities. Thus, the political science and
economic approaches to rebellion have assumed both different rebel motivationgrievance
versus greed-and different explanations-atypical grievances versus
atypical opportunities.
Hirshleifer (1995, 2001) provides an important refinement on the motiveopportunity
dichotomy. He classifies the possible causes of conflict into preferences,
opportunities, and perceptions. The introduction of perceptions allows
for the possibility that both opportunities and grievances might be wrongly
perceived. If the perceived opportunity for rebellion is illusory-analogous to the
'winners' curse'-unprofitability will cause collapse, perhaps before reaching our
threshold for civil war. By contrast, when exaggerated grievances trigger rebellion,
fighting does not dispel the misperception and indeed may generate genuine
grievances.
Misperceptions of grievances may be very common: all societies may have groups
with exaggerated grievances. In this case, as with greed-rebellion, motive would
not explain the incidence of rebellion. Societies that experienced civil war would
be distinguished by the atypical viability of rebellion. In such societies rebellions
..........................................................................................................................................................................
1 On the analysis of th
2 By the 'Machiavelli T
to exploit someone els
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 565
would be conducted by viable not-for-profit organizations, pur
agendas by violent means.
Greed and misperceived grievance have important simil
of rebellion. They provide a common explanation-'opportu
describe the common conditions sufficient for profit-seekin
rebel organizations to exist. On our evidence they are observa
since we cannot observe motives. They can jointly be contraste
account of conflict in which the grievances that both motivate
are assumed to be well-grounded in objective circumstanc
high inequality, or unusually weak political rights. We now t
for opportunities and objective grievances.
2.2 Proxies for opportunity
The first step in an empirical investigation of conflict is a cl
definition of the phenomenon. We define civil war as an inter
least 1,000 combat-related deaths per year. In order to dis
massacres, both government forces and an identifiable rebel
suffer at least 5% of these fatalities. This definition has becom
the seminal data collection of Small and Singer (1982) and Sin
We use an expanded and updated version of their data set that
over the period 1960-99 and identifies 79 civil wars, listed in T
explain the initiation of civil war using these data.3
We now consider quantitative indicators of opportunity, start
ties for financing rebellion. We consider three common sources
resources, donations from diasporas, and subventions from h
Klare (2001) provides a good discussion of natural resource e
diamonds in West Africa, timber in Cambodia, and cocaine in C
we proxy natural resources by the ratio of primary commodity
each of the 161 countries. As with our other variables, we m
five years, starting in 1960 and ending in 1995. We then con
five years as an 'episode' and compare those in which a civil w
flict episodes') with those that were conflict-free ('peace episod
statistics give little support to the opportunity thesis: the conf
average slightly less dependent upon primary commodity exp
episodes. However, there is a substantial difference in the dis
episodes tended to have either markedly below-average or mar
dependence, while the conflict episodes were grouped around t
3 Later in the paper (Table 6) we examine whether our results are sensitive to th
Sambanis (2002) comes to a similar conclusion.
4 We list the data sources and definitions in the Appendix.
5 The standard deviation of primary commodity exports is 0.11 for the conflict
peace episodes.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
566 P. COLLIER AND A. HOEFFLER
Table 1 Outbreaks of war
Country Start of End of Previous war Gdp Secondary
the war the war sample schooling sample
Afghanistan 04/78 02/92
Afghanistan 05/92 Ongoing
Algeria 07/62 12/62
Algeria 05/91 Ongoing
Angola 02/61 11/75
Angola 11/75 05/91
Angola 09/92 Ongoing *
Azerbaijan 04/91 10/94
Bosnia 03/92 11/95
Burma/Myanmar 68 10/80
Burma/Myanmar 02/83 07/95
Burundi 04/72 12/73
Burundi 08/88 08/88 *
Burundi 11/91 ongoing
Cambodia 03/70 10/91
Chad 03/80 08/88
China 01/67 09/68
Columbia 04/84 ongoing
Congo 97 10/97
Cyprus 07/74 08/74
Dominican Rep. 04/65 09/65 *
El Salvador 10/79 01/92
Ethiopia 07/74 05/91
Georgia 06/91 12/93
Guatemala 07/66 07/72
Guatemala 03/78 03/84
Guinea-Bissau 12/62 12/74
India 08/65 08/65 *
India 84 94 *
Indonesia 06/75 09/82 *
Iran 03/74 03/75
Iran 09/78 12/79 *
Iran 06/81 05/82 *
Iraq 09/61 11/63
Iraq 07/74 03/75
Iraq 01/85 12/92
Jordan 09/70 09/70
Laos 07/60 02/73
Lebanon 05/75 09/92
Liberia 12/89 11/91
Liberia 10/92 11/96
Morocco 10/75 11/89
Mozambique 10/64 11/75
Mozambique 07/76 10/92
Nicaragua 10/78 07/79
Nicaragua 03/82 04/90
Nigeria 01/66 01/70
(continu
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 567
Table 1 Continued
Country Start of End of Previous war Gdp Secondary
the war the war sample schooling sample
Nigeria 12/80 08/84 * *
Pakistan 03/71 12/71 * *
Pakistan 01/73 07/77
Peru 03/82 12/96 * *
Philippines 09/72 12/96 * * *
Romania 12/89 12/89 * *
Russia 12/94 08/96
Russia 09/99 Ongoing *
Rwanda 11/63 02/64
Rwanda 10/90 07/94 * * *
Sierra Leone 03/91 11/96 * *
Sierra Leone 05/97 07/99 * *
Somalia 04/82 05/88 * *
Somalia 05/88 12/92 * *
Sri Lanka 04/71 05/71 * *
Sri Lanka 07/83 ongoing * * *
Sudan 10/63 02/72
Sudan 07/83 ongoing * * *
Tajikistan 04/92 12/94
Turkey 07/91 ongoing *
Uganda 05/66 06/66 * *
Uganda 10/80 04/88 * * *
Vietnam 01/60 04/75 *
Yemen 05/90 10/94
Yemen, Arab Rep. 11/62 09/69 *
Yemen, People's Rep. 01/86 01/86
Yugoslavia 04/90 01/92
Yugoslavia 10/98 04/99 *
Zaire/Dem. Rep. of Congo 07/60 09/65
Zaire/Dem. Rep. of Congo 09/91 12/96 * * *
Zaire/Dem. Rep. of Congo 09/97 09/99 * * *
Zimbabwe 12/72 12/79 * *
Note: Previous wars include war start
natural resources are sufficien
be so well-financed that rebel
make the net effect of natu
may also reflect differences
Section 3). Further, primary
that may cause civil war, suc
economic mismanagement (Sa
conflict risk may be due to r
financial opportunities.
6 Collier (2000) provides an illustrati
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
568 P. COLLIER AND A. HOEFFLER
Table 2 Descriptive statistics
Sample No civil war Civil war
(n = 1167) (n = 1089) (n 78)
War starts 0.067 0 1
Primary commodity exports/GDP 0.168 0.169 0.149
GDP per capita (const. US$) 4061 4219 1645
Diaspora (relative to population of 0.017 0.018 0.004
country of origin)
Male secondary schooling (% in school) 43.42 44.39 30.3
GDP per capita growth (average for 1.62 1.74 -0.23
previous 5 years)
Previous war (% with war since 1945) 20.8 18.5 53.8
Peace duration (months since last conflict) 327 334 221
Forest cover (%) 31.11 31.33 27.81
Mountainous terrain (%) 15.82 15.17 24.93
Geographic concentration of the 0.571 0.569 0.603
population (Gini)
Population density (inhabitants per 150 156 62
square km)
Population in urban areas (%) 45.11 46.00 32.7
Ethnic fractionalization (index, 0-100) 39.57 38.64 52.63
Religious fractionalization (index, 0-100) 36.09 35.98 37.70
Polarization a= 1.6 (index, 0-0.165) 0.077 0.077 0.076
Democracy (index, 0-10) 3.91 4.07 1.821
Ethnic dominance (% with main ethnic 0.465 0.465 0.452
group 45-90%)
Income inequality (Gini) 0.406 0.406 0.410
Land inequality (Gini) 0.641 0.641 0.631
Note: We examine 78-rather than the 79-war starts a
two outbreaks of war during 1970-74. We only include o
A second source of rebel finance is from
(1996) review the evidence, an example bein
Tamils in north America. We proxy the size o
living in the US, as given in US Census data.
in other countries, it ensures uniformity in
same legal, organizational, and economic env
population as a proportion of the population
analysis we decompose the diaspora into tha
which is exogenous to conflict, but here we
These do not support the opportunity thesis
the conflict episodes.
A third source of rebel finance is from
the government of Southern Rhodesia pu
Mozambique. Our proxy for the willingness
military opposition to the incumbent gover
Cold War each great power supported rebelli
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 569
power. There is some support for the opportunity thesis: only
broke out during the 1990s.
We next consider opportunities arising from atypically low c
be paid, and their cost may be related to the income forgone b
Rebellions may occur when foregone income is unusually low
mists regard this as fanciful we give the example of the Russia
Whites, both rebel armies, had four million desertions (the ob
ment problem). The desertion rate was ten times higher in sum
the recruits being peasants, income foregone were much high
(Figes, 1996). We try three proxies for foregone income: mea
male secondary schooling, and the growth rate of the economy.
the conflict episodes started from less than half the mean in
episodes. However, so many characteristics are correlated wit
that, depending upon what other variables are included, the pr
interpretations. Our second proxy, male secondary school
advantage of being focused on young males-the group fro
recruited. The conflict episodes indeed started from lower
but this is again open to alternative interpretation: education m
of conflict through changing attitudes. Our third measure, th
economy in the preceding period, is intended to proxy new in
Conflict episodes were preceded by lower growth rates. This
evidence that the lower is the rate of growth, the higher is
unconstitutional political change (Alesina et al., 1996).7 Althoug
are all consistent with atypically low forgone income as an opp
could also be interpreted as an objective economic grievance.
The opportunity for rebellion may be that conflict-spec
military equipment) is unusually cheap. We proxy the cost of
time since the most recent previous conflict: the legacy of weap
organizational capital will gradually depreciate. Empirically
preceded by far longer periods of peace than conflict episode
this supports the opportunity thesis, it could also be interpret
gradual decay of conflict-induced grievances.
Another dimension of opportunity is an atypically weak go
capability. An unambiguous indicator is if the terrain is favora
and mountains provide rebels with a safe haven. We measure
country's terrain that is forested, using FAO data. We could fin
7 The economic growth literature concentrates on the analysis of political insta
economic growth (see for example Barro 1991, 1997). Alesina et al. (1996)
equation system of economic growth and political instability. They present su
that political instability reduces growth. Lower growth does not seem to c
defined as the number of government changes. However, when they define p
narrowly as unconstitutional government changes they find that lower growth
political instability.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
570 P. COLLIER AND A. HOEFFLER
on mountainous terrain: proxies such as altitude tend to misc
and rugged uplands. We therefore commissioned a new index
John Gerrard. The descriptive statistics (Table 2) suggest that t
conflict episodes 25% of the terrain is mountainous, versu
episodes, although there is no difference in forest cover. Geog
the population may also inhibit government capability: Herbst
Zaire is prone to rebellion because its population lives arou
country. We measure dispersion by calculating a Gini coeffic
dispersion.8 In fact, the concentration of the population is slig
peace episodes (0.57) than prior to war episodes (0.6). Simil
density and low urbanization may inhibit government capabilit
to war episodes both population density and urbanization are
A final source of rebel military opportunity may be social c
religious diversity within organizations tends to reduce their
(Easterly and Levine, 1997, Alesina et al., 1999, Collier, 2001). A
army may be in particular need of social cohesion, constrain
single ethnic or religious group. A diverse society might in t
opportunity for rebellion by limiting the recruitment pool. T
measure of ethnic diversity is the index of ethno-linguistic fr
measures the probability that two randomly drawn people wil
ethnic groups. We could find no measure of religious fractio
constructed one equivalent to that of ethnic fractionalization
Barrett (1982). If ethnic and religious divisions are cross-cuttin
lization is multiplicative rather than additive. We could f
religious and ethnic divisions and so we construct a proxy tha
mum potential social fractionalization.9 The thesis that social
opportunity is not supported by the descriptive statistics: co
atypically high fractionalization. This seems more consiste
interpretation, to which we now turn.
2.3 Proxying objective grievances
We consider four objective measures of grievance: ethnic or re
tical repression, political exclusion, and economic inequality.
8 For the calculation of the Gini coefficient we used the population data per
the income Gini coefficient, the Gini coefficient of population dispersion will
concentrated in a relatively small area of the country.
9 If there were e equally sized ethnic groups and r equally sized religious gr
social fractionalization would be measured simply by the product er. Since both
ethnic and religious fractionalization range on the scale 0-100, their product
religious or ethnic homogeneity whereas there is social homogeneity only if
therefore measure social fractionalization as the product of the underlying ind
is the greater.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 571
Ethnic and religious hatreds are widely perceived as a cause
Although such hatreds cannot be quantified, they can evid
societies that are multi-ethnic or multi-religious and so ou
various dimensions of diversity. Our previously discussed meas
zation are pertinent: inter-group hatreds must be greater in
fractionalized than in those which are homogenous. Howe
source of inter-group tension is not diversity but polarization
allowable class of measures of polarization is quite limited.
measure due to Esteban and Ray (1994)
P= K 7l+crjd (1)
i=1 j=l
where 7i denotes the percentage of people that belong to group i in the t
population, i= 1,... , n. This measure of polarization depends on the parameters
K and a. K does not change the order, but is used for population normalization
Esteban and Ray show that ac is bounded between zero and 1.6. We calcul
the polarization measure for three different values of a, 0, 0.8 and 1.6, us
primary data on ethnic composition."1 In addition we investigate the variant o
the Esteban-Ray measure proposed by Reynal-Querol. These measures indeed d
tinguish polarization from fractionalization: their correlation coefficient rang
between 0.39 (a= 1.6) and 1.0 (a = 0).12 The descriptive data does not sugge
that polarization is important: conflict and peace episodes have very similar m
values (Table 2).
We measure political repression using the Polity III data set (see Jaggers and
Gurr, 1995). This measure of political rights ranges 0-10 on an ascending ordin
scale. Political rights differ considerably between conflict and peace episodes.
also investigate the Polity III measure of autocracy, and a measure of polit
openness published by Freedom House (the 'Gastil Index'). The quantitative poli
tical science literature has already applied these measures to conflict risk. Heg
et al. (2001) find that repression increases conflict except when it is severe.
Even in democracies a small group may fear permanent exclusion. A potential
important instance is if political allegiance is based on ethnicity and one ethn
group has a majority. The incentive to exploit the minority increases the larger
the minority, since there is more to extract (Collier, 2001). Hence, a minority
l?The link from polarization to conflict is proposed by Esteban and Ray (1999) and Reynal-Qu
(2000) and is common in the popular literature.
l Our data source was Atlas Narodov Mira, USSR (1964). The Esteban-Ray measure includes a coe
cient d that denotes the degree of antagonism between two different ethnic groups. Obviously, in la
samples such as we are using this is not observed. Following Reynal-Querol (2000) we assume that
distance between any two ethnic groups is unity whereas that within the group is zero, so that d ha
properties: d = 1 if i f j and d = 0 if i =j.
12 For a = 0 the polarization measure is equal to the Gini coefficient.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
572 P. COLLIER AND A. HOEFFLER
may be most vulnerable if the largest ethnic group constitutes
term this ethnic dominance. In Table 2 we define it as occurring
group constitutes 45-90% of the population. On this definitio
important: it is as common in peace episodes as in conflict ep
The opening page of Sen's On Economic Inequality (Sen, 197
relation between inequality and rebellion is indeed a close one'.
to induce redistribution, and rich regions may mount secession
empt redistribution.l3 We measure income inequality by the
by the ratio of the top-to-bottom quintiles of income. We me
by the Gini coefficient of land ownership. The data are from
(1996, 1998). Inequality is slightly higher prior to the conflict
2.4 Scale effects
Our measures of opportunity, such as primary commodity exports, income, and
school enrolment, are scaled by measures of country size. For given values of these
variables, opportunities should be approximately proportional to size. Grievance
might also increase with size: public choices diverge more from the preferences
of the average individual as heterogeneity increases.'4 We are, however, able to
control for three aspects of heterogeneity: ethnic, religious and income diversity.
Empirically, the conflict episodes had markedly larger populations than the peace
episodes.
3. Regression analysis
As set out above, the proxies for opportunity and objective grievances are largely
distinct and so can be compared as two non-nested econometric models. There is,
however, no reason for the accounts to be exclusive and the aim of our econometric
tests is to arrive at an integrated model which gives an account of conflict risk in
terms of all those opportunities and grievances that are significant.
We now attempt to predict the risk that a civil war will start during a five-year
episode, through a logit regression. Our dependent variable, civil war start, takes a
value of one if a civil war started during a five year episode (1965-69,..., 1994-99).
Episodes that were peaceful from the beginning until the end are coded zero.
Ongoing wars are coded as missing observations as to not conflate the analysis
of civil war initiation and duration.15 If a war ended and another one started in the
same period we coded these events as one. Some of our explanatory variables are
13 This is analogous to the theory of tax exit proposed by Buchanan and Faith (1987).
14 Mounting diversity is the offset to scale economies in the provision of public goods in the model of
optimal county size proposed by Alesina and Spolaore (1997).
15 This approach contrasts with our initial work (Collier and Hoeffler, 1998) in which we used a tobit
procedure to study the duration of civil war (on a much inferior data set) and argued that the same
factors that determined duration would determine the risk of initiation. Collier et al. (2004) establishes
that this is wrong: initiation and duration are radically different processes.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 573
time invariant. Those that are not are measured either for th
period (e.g. 1965) or during the preceding five years (e.g. gro
in order to avoid endogeneity problems. Our results rest on h
experienced an outbreak of war differed from those that sust
We start with the opportunity model (see Table 3). The first r
excludes per capita income and diasporas. Because per capita inc
Table 3 Opportunity model
1 2 3 4 5
Primary comm
exports/GDP
(Primary com
exports/GDP)2
Post-coldwar
Male secondary schooling
Ln GDP per capita
GDP growth
Peace duration
Previous war
Mountainous terrain
Geographic dispersion
Social fractionalization
Ln population
Diaspora/peace
18.149
(6.006)***
-27.445
(11.996)***
-0.326
(0.469)
-0.025
(0.010)**
-0.117
(0.044)***
-0.003
(0.002)
p=0.128
0.464
(0.547)
p= 0.396
0.013
(0.009)
p=0.164
-2.211
(1.038)**
-0.0002
(0.0001)
p= 0.109
0.669
(0.163)***
18.900
(5.948)***
-29.123
(11.905)***
-0.207
(0.450)
-0.024
(0.010)**
-0.118
(0.044)***
-0.004***
(0.001)
16.476
(5.207)***
-23.017
(9.972)**
-0.454
(0.416)
-0.837
(0.253)***
-0.105
(0.042)***
-0.004
(0.001)***
17.567
(6.744)***
-28.815
(15.351)*
17.404
(6.750)***
-28.456
(15.366)*
-1.237 -1.243
(0.283)*** (0.284)***
-0.002 -0.002
(0.001) (0.001)
0.014 0.008
(0.009) (0.008)
-2.129
(1.032)**
-0.0002
(0.0001)
p=0.122
0.686
(0.162)***
-0.865
(0.948)
-0.0002
(0.0001)**
0.493 0.295
(0.129)*** (0.141)**
700.931
(363.29)**
Diaspora corrected/peace
(Diaspora-diaspora
corrected)/peace
N
No of wars
Pseudo R2
Log likelihood
688
46
0.24
-128.49
688
46
0.24
-128.85
750
52
0.22
-146.86
595
29
0.25
-93.27
0.296
(0.141)**
741.155
(387.636)*
823.941
(556.024)
595
29
0.25
-93.23
Notes: All regressions include a constant. Standard errors in parentheses. ***, *, * indicate s
the 1, 5, and 10% level, respectively.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
574 P. COLLIER AND A. HOEFFLER
in secondary schooling are highly correlated, they canno
regression (p = 0.8). Our diaspora measure is available only
and so we explore it as an addendum. The variables included i
permit a sample of 688 episodes (from 123 countries), includ
Primary commodity exports are highly significant. Althoug
linear, the risk of conflict peaks when they constitute around
is a high level of dependence. The other proxy for finance, th
has the expected sign but is insignificant. The foregone ear
both significant with the expected sign: secondary schoolin
reduce conflict risk. Our proxy for the cost of conflictnumber
of months since any previous conflict (back to
between this interpretation and the danger that the proxy
fixed effects, we add a dummy variable that is unity if there
post-1945. Our proxy has the expected sign and is on the bord
while the dummy variable is completely insignificant. When t
dropped (column 2) the proxy becomes highly significant and
changed. The proxies for military advantage also have the e
marginally significant: mountainous terrain, population dispe
tionalization. Finally, the coefficient on population is positive
The third column replaces secondary schooling with per
permits a larger sample-750 episodes (from 125 countrie
Per capita income is highly significant with the expected n
the change of variable and the expansion of sample have littl
results-social fractionalization becomes significant and p
loses significance. There is little to chose between these
model-secondary schooling gives a slightly better fit, but p
mits a slightly larger sample.
In the last two columns of Table 3 we introduce our dia
many observations are missing, the number of war episodes w
radically reduced. In order to preserve sample size we therefo
parsimonious version of the model. We drop four sample-co
explanatory variables: social fractionalization, population dis
terrain, and the rate of growth in the previous episode. The
variables are thus per capita GDP, primary commodity export
number of months since the previous conflict. Even with t
deletions, the sample size is reduced to 29 war episodes (a
However, all the included explanatory variables remain signif
diaspora is not directly significant in the initiation of conflic
it is significant when interacted with the number of month
ous conflict. 'Diaspora/peace' divides the size of the diaspora
..........................................................................................................................................................................
16 We differentiate the proba
the risk is at its maximum
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 575
previous conflict. The variable is positive and significant: a la
erably increases the risk of repeat conflict.
While this result may indicate that diasporas increase the risk
their finance of rebel organizations, it is also open to a more a
tion. Diasporas are endogenous to the intensity of conflict: wh
people emigrate to the USA. Hence, the size of the diaspora mi
intensity of conflict. The result may therefore be spurious: in
have a higher risk of repetition. To test for this we decompos
into a component which is exogenous to the intensity of con
endogenous component. For this decomposition we estimated
reported in Appendix 1. The size of the diaspora in a census ye
a function of its size in the previous census, time, per capita in
of origin, and whether there was a war in the intervening
predicts the size of the diaspora with reasonable accuracy. Fo
to a conflict we replace the actual data on the size of the diasp
from this regression. Thus, all post-conflict observations of d
which are purged of any effect of the intensity of conflict. Th
actual and estimated figures is then used as an additional varia
part of the diaspora which is potentially endogenous to the i
Both of these measures are then introduced into the regressi
previous single measure of the diaspora. The results are r
column of Table 3. The purged measure of the diaspora remai
the size of the coefficient is only slightly altered (it is not signific
that on the endogenous diaspora measure). This suggests th
substantial causal effect of the diaspora on the risk of conflict
also guides our interpretation of why the risk of conflict repet
is maintained. Recall that in principle this could be either beca
fade, or because 'rebellion-specific capital' gradually depreciat
sporas slow these processes? Diasporas preserve their own hatr
finance rebellion. However, it is unlikely that the diaspora's
influence attitudes among the much larger population in the
By contrast, the finance provided by the diaspora can offset
rebellion-specific capital, thereby sustaining conflict risk.
In Table 4 we turn to objective grievance as the explanation
ping all the economic measures of opportunity.17 We retain th
since a previous conflict, since (subject to our discussion abov
preted as proxying fading hatreds. In the first column we also
measures due to considerations of sample size. This enables a
850 episodes and 59 civil wars.
1 We retain the two geographic measures, population dispersion and mount
their exclusion does not affect the results, non-economists often find the pro
opportunity affects conflict plausible and inoffensive, while contesting the role
We retain population size as a scale variable.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
576 P. COLLIER AND A. HOEFFLER
Table 4 Grievance model
1 2 3
Ethnic fractionalization 0.010 0.011 0.012
(0.006)* (0.007)* (0.008)
Religious fractionalization -0.003 -0.006 -0.004
(0.007) (0.008) (0.009)
Polarization a = 1.6 -3.067 -4.682 -6.536
(7.021) (8.267) (8.579)
Ethnic dominance (45-90%) 0.414 0.575 1.084
(0.496) (0.586) (0.629)*
Democracy -0.109 -0.083 -0.121
(0.044)*** (0.051)* (0.053)**
Peace duration -0.004 -0.003 -0.004
(0.001)*** (0.001)*** (0.001)***
Mountainous terrain 0.011 0.007 -0.0001
(0.007) (0.009) (0.009)
Geographic dispersion -0.509 -0.763 -1.293
(0.856) (1.053) (0.102)
Ln population 0.221 0.246 0.300
(0.096)** (0.119)** (1.133)**
Income inequality 0.015
(0.018)
Land inequality 0.461
(1.305)
N 850 604 603
No of wars 59 41 38
Pseudo R2 0.13 0.11 0.17
Log likelihood -185.57 -133.46 -117.12
Notes: All regressions include a constant. Standard er
the 1, 5 and 10% level, respectively.
Column 1: the two measures of fractionalization and
The four proxies for ethnic and religious
view of the attention that the phenomen
significant at 10% with the expected sign
tion are insignificant with the wrong sign
Nor are the three measures jointly signif
with the expected sign-repression incr
previous conflict is again highly signific
more likely to be proxying rebellion-s
second and third columns we introduce
respectively. Although the sample size is
18 At this stage we measure polarization with a = 1.6
largest ethnic group constitutes 45-90% of the popu
where we investigate robustness to alternative defini
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 577
episodes of which 41 (income) and 38 (land) are wars. Neither
significance.19 All three grievance models have very low expla
pseudo R2 of 0.17 or lower.
We now turn to the question of which model, opportunity or
a better explanation of the risk of civil war. Since the two mo
i.e. one model is not a special case of the other, we use the J
Davidson and MacKinnon (1981). As shown in the first two colu
find that we cannot reject one model in favor of the other.20 T
while the opportunity model is superior, some elements of the
likely to add to its explanatory power. We therefore investigat
the two models as presented in column 3 of Table 5.
Since this combined model includes income inequality a
our sample size is much reduced (479 observations). In column
ity (which is consistently insignificant). Omitting inequality
size to 665. In this combined model neither democracy, e
fractionalization nor the post-Cold War dummy are significa
are statistically significant or close to significance and th
sonable (pseudo R2 of 0.26). Since both the grievance and o
are nested in the combined model, we can use a likelihood rat
mine whether the combined model is superior. We can re
the restrictions proposed by the grievance model, but not by
model.21
Although the combined model is superior to the opportunity and grievance
models, several variables are completely insignificant and we drop them sequentially.
First we exclude the post-Cold War dummy, then religious fractionalization,
..........................................................................................................................................................................
19We also tried the r
insignificant.
20 The J-test is based on
in terms of the two di
(1) p = f(opportunity)
(2) p = f(grievance)
Based on these logit re
popportunity and fgriev
(1) p = f(opportunity, p
(2) p = f(grievance, p?PP
According to the J-tes
between the two diffe
opportunity model in f
we reject the grievance
Table 5, grievance is sig
model.
21Using the same sample as for the combined model (n=665) we obtain the following results:
Opportunity model versus combined model, 5 degrees of freedom, Likelihood Ratio Test (LRT) statistic
7.85 (p = 0.165); grievance model versus combined model, 6 degrees of freedom, LRT statistic 29.64
(p = 0.000).
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
Table 5 Combined opportunity and grievance model
1 2 3 4
Primary com
exports/GDP
(Primary commodity exports/GDP)2
Post-coldwar
Male secondary schooling
Ln GDP per capita
(GDP growth) t-1
Peace duration
Mountainous terrain
Geographic dispersion
Ln population
Social fractionalization
19.107
(5.996)***
-30.262
(12.008)***
-0.208
(0.457)
-0.021
(0.011)**
-0.108
(0.044)***
-0.0003
(0.002)
0.005
(0.010)
-1.976
(1.049)*
0.489
(0.193)**
-0.0002
(0.0001)***
Ethnic fractionalization
Religious fractionalization
Polarization
0.0005
(0.0014)
0.001
(0.008)
0.053
(1.101)
-0.022
(0.136)
0.008
(0.007)
-0.005
(0.008)
-9.338
37.072
(10.293)***
-69.270
(21.697)***
-0.873
(0.644)
-0.029
(0.013)**
-0.045
(0.062)
-0.0003
(0.0015)
0.005
(0.012)
-4.032
(1.490)***
0.927
(0.250)***
-0.0008
(0.0003)**
0.041
(0.019)**
0.015
(0.020)
-25.276
23.385
(6.692)**
-36.335
(12.998)
-0.281
(0.459)
-0.022
(0.011)
-0.108
(0.045)*
-0.003
(0.001)*
0.015
(0.009)
p=O.11
-1.962
(1.149)
0.697
(0.181)**
-0.0005
(0.0003)
p=0.11
0.023
(0.015)
0.014
(0.019)
-15.992
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
(8.734) (13.390)* (10.518)
Ethnic dominance (45-90%) 1.210 2.020 1.5
(0.648)* (0.915)** (0.746
Democracy -0.036 -0.018 -0.042
(0.054) (0.062) (0.054)
Income inequality 0.025
(0.024)
Grievance predicted value 0.765
(0.413)*
Opportunity predicted value 1.044
(0.21 1)***
Primary commodity exports
oil dummy (9.
(Primary commodity exports/
oil dummy (38
N 665 665 479 665 68
No of wars 46 46 32 4
Pseudo R2 0.24 0.25 0.24 0.26
Log likelihood -126.69 -125.29 -89.55 -124.60 Notes:
All regressions include a constant. Sta
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
580 P. COLLIER AND A. HOEFFLER
then democracy,22 then polarization, then ethnic fraction
mountainous terrain, yielding the baseline model of colum
with per capita income replacing secondary enrolment in colu
reduction in the model is accepted, and no additions of variab
previous models are accepted. The baseline model and its varian
results although the variant has less explanatory power an
significance (ethnic dominance and geographic dispersion).
Our baseline model allows us to calculate the change in the
starts for different values of the explanatory variables. We prese
in Appendix Table A2. At the mean of all variables the risk of
11.5%. Our model predicts that a hypothetical country with
teristics found in our sample would have a near-certain risk of
all the best characteristics would have a negligible risk. We no
variable affects the risk of civil war (keeping all other variab
values).
The effect of primary commodity exports on conflict risk is both highly significant
and considerable. At peak danger (primary commodity exports being 33% of
GDP), the risk of civil war is about 22%, while a country with no such exports has
a risk of only 1%. The effect is sufficiently important to warrant disaggregation
into different types of commodities. We categorized primary commodity exports
according to which type of product was dominant: food, non-food agriculture, oil,
other raw materials, and a residual category of 'mixed'.23 Of the many potential
disaggregations of primary commodity exports permitted by this data, only one
was significant when introduced into our baseline regression, namely oil versus
non-oil. The results are reported in column 7. We add variables that interact the
primary commodity export share and its square with a dummy variable that takes
the value of unity if the exports are predominantly oil. Both variables are significant:
oil exports have a distinct effect on the risk of conflict. However, the effect is
modest: at the average value of primary commodity exports oil has the same effect
as other commodities. Low levels of oil dependence are somewhat less risky than
other commodities and high levels of dependence are somewhat more risky. The
disaggregation slightly reduces the sample size, does not change the significance of
any of the other variables, and substantially improves the overall fit of the model.24
Recall that the other proxies for financial opportunities, the Cold War and
diasporas, are not included in this baseline. The end of the Cold War does not
22 We tried different specifications to test for the effect of political repression by investigating non-linear
effects, by including the autocracy score instead of the democracy score, and by using the difference
between the two variables as suggested by Londregan and Poole (1996). We also tried the Freedom
House measure of political freedom, but neither of these alternative political repression measures were
found to be significant.
23 We would like to thank Jan Dehn for providing us with the data that enabled this disaggregation.
24 Furthermore, using data from Dehn (2000) we investigated whether contemporaneous export price
changes altered the risk of conflict. We could not find any evidence to support this hypothesis.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 581
appear to have had a significant effect. Diasporas are excluded
purely for considerations of sample size. In the parsimonious v
are included, their effect on the risk of repeat conflict is substan
of peace, switching the size of the diaspora from the smallest to
post-conflict episodes increases the risk of conflict six-fold.
The proxies for earnings foregone have substantial effects. If
for secondary schooling is ten percentage points higher than th
war is reduced by about three percentage points (a decline in t
to 8.6%). An additional percentage point on the growth rate red
by about one percentage point (a decline from 11.5% to 10.4%
for the cost of rebellion is also highly significant and substant
civil war there is a high probability of a re-start, the risk being
declines over time at around one percentage point per year.
The only measures of rebel military advantage that survive i
population dispersion and social fractionalization. Consis
hypothesis, countries with a highly concentrated population h
of conflict, whereas those with a highly dispersed population h
(about 37%). Consistent with the hypothesis that cohesion is i
effectiveness, social fractionalization makes a society substan
mally fractionalized society has a conflict risk only one quart
genous society.
Only one of the proxies for grievance survives into the bas
namely ethnic dominance. If a country is characterized by et
risk of conflict is nearly doubled. Thus, the net effect of increase
the sum of its effect on social fractionalization and its effect on ethnic dominance.
Starting from homogeneity, as diversity increases the society is likely to become
characterized by ethnic dominance, although this will be reversed by further
increases in diversity. The risk of conflict would first rise and then fall. Note that
while these measures in combination are superficially similar to the hypothesized
effect of polarization, our measure of polarization itself is insignificant.
Finally, the coefficient on the scale variable, population, is highly significant and
close to unity: risk is approximately proportional to size. We have suggested that
proportionality is more likely if conflict is generated by opportunities than by
grievances.
4. Robustness checks
We now test these baseline results for robustness. We consider the sensitivity bot
to data and to method. With respect to data, we investigate the effect of outlying
observations, and of different definitions of the dependent and independent vari
ables. With respect to method, we investigate random effects, fixed effects and ra
events bias.
We investigate outlying observations using two different methods. First, we
inspect the characteristics of the 46 conflict episodes used in the baseline regressio
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
582 P. COLLIER AND A. HOEFFLER
and second, we use a systematic analysis of influential data po
is unbalanced as between events and non-events, the potential
arises predominantly among the 46 conflict episodes. Of these
were first-time conflicts and 22 were repeat conflicts.
First, the classification of events in Romania in 1989, and in
and 1981 as civil wars is in various respects questionable. They
highly atypical of conflict episodes. Both had secondary scho
higher than the other conflict episodes, and Iran also had
primary commodity export share. In Table 6 column 1 we dr
observations. No results are overturned, but the performanc
improves and all variables are now significant at the 1% or 5
There are four observations of highly negative growth: Ang
(now the Democratic Republic of the Congo) in 1990-95, Iran
in 1980-84. All of these growth collapses appear to be genuin
different countries. We now check whether the result that the
conflict risk is dependent upon these four observations, deleti
Iran and Romania (Table 6, column 2). Growth remains signific
cient is only slightly reduced. Hence, we can conclude that th
conflict due to slow growth is not confined to episodes of gro
more continuous relationship.
We next analyse whether our regression results are sensitive
influential data points. Based on the methods developed by P
examined which observations may be influential and investigat
these observations from our baseline model changed our resu
influential observations: Congo 1995-99, Iran 1970-74, an
However, when we omitted these three observations from our
all fit of the regressions improved (from previously R2 = 0.24
of the coefficients remain statistically significant. (Table 6, co
We now investigate the possibility that a few countries with
export ratio account for the non-monotonic relationship to c
might imply that the reduction in conflict risk only occurre
of commodity dependence. Four peaceful countries have part
of primary dependence: Saudi Arabia, Guyana, Oman, and Tri
In Table 6 column 4 we present our baseline model exclud
primary commodity exporters. The non-monotonic relationsh
commodity exports and the risk of conflict remains signific
results.
We next turn to questions of variable definition. The most contentious aspect of
the dependent variable is distinguishing between whether a country has a single
long war or multiple shorter wars interrupted by periods of peace. In the above
analysis we have been guided by the judgement of the political scientists who built
.........(1997) pp.98-101 provides a discussion of influence in limited dependent variable models.
25 Long (1997) pp.98-101 provides a discussion of influence in limited dependent variable models.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 583
the original data sets. Some peace periods are, however, quite s
better to conceptualize these as interludes in a single war. We f
those wars that were separated by peace periods of less than
tinuous wars (Table 6, column 5). The baseline results are
redefinition. We then reclassified those wars separated by less
tinuous wars (Table 6, column 6). The only result to be affected
rate becomes marginally insignificant (p = 0.12), although its
changed.26
We investigated how robust our results are to the definition of ethnic dominance
and social fractionalization. In the baseline we define ethnic dominance as the
largest ethnic group constituting 45-90% of the population. We investigate other
definitions that either vary the range of the population or use the share of the
largest group regardless of its size. As the range is changed from 45-90% th
significance level and the coefficient are both reduced, while if the definition is
changed more radically to being the population share of the largest group it is
completely insignificant. We also find that 'social fractionalization', our measure o
cross-cutting cleavages, dominates the other possible aggregation procedures for
ethnic and religious diversity. When this measure of fractionalization is included
with the ethnic and religious diversity indices either together or individually, it i
significant whereas the underlying indices are not significant.
In the baseline we use only the most extreme measure of polarization over the
range proposed by Esteban and Ray (1994). However, if this measure is replaced by
either the lower bound (a =0), or the central measure (a =0.8) the results are
unaffected: polarization remains insignificant and the other variables remain sig
nificant. We also experimented with the alternative measure proposed by ReynalQuerol
(2002), and with the number of ethnic groups, but with the same result.27
In Table 7 we investigate a number of different estimation issues. We concentrat
on the analysis of random effects, fixed effects, time effects, and a correction for
rare events. We re-estimated our models using random effects. For the baseline
model we find that the panel data estimator is not different from the pooled
estimator, i.e. we accept the hypothesis that we can pool across the observations.28
The estimation of fixed effects logits was only possible on a very small sub-sample
of the observations. The countries for which the dependent variable does not vary
over time (the majority of countries experienced only peace) cannot be included in
26 We also examined the effect of time since the previous conflict in more detail by including the natura
logarithm of the peace variable or its square, however, a linear decay term provides a better fit. Note tha
the measure of peace since the end of the civil war is somewhat imprecise since we only measure it from
the end of the war to the initial year of each sub-period. A duration model of post-war peace would allow
a more detailed analysis of this peace effect, however, the duration model results in Collier et al. (2004
support the results presented in this paper.
27All of these robustness checks are presented in Collier and Hoeffler (2002).
28 A LRT provides a x2 statistic of 0 (p = 0.998). Thus, we cannot reject the null-hypothesis that the pane
data and pooled estimator provide the same results.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
Table 6 Robustness checks
1 2 3
Excluding Excluding Excluding Exclud
Iran and Iran and Romania influential high pri
Romania and growth data points commo
collapses exporters
Primary commodity exports/GDP
(Prim. com. exports/GDP)2
Male secondary schooling
(GDP growth) t- 1
Peace duration
Geographic dispersion
19.696
(6.608)***
-34.090
(14.356)**
-0.035
(0.01 1)***
-0.140
(0.047)***
-0.004
(0.001)***
-2.114
(1.080)**
19.029
(6.671)***
-33.250
(14.609)**
-0.037
(0.011)***
-0.100
(0.052)**
-0.003
(0.001)***
-2.272
(1.090)**
28.745
(7.862)***
-59.818
(17.781)***
-0.041
(0.011)***
-0.137
(0.046)***
-0.004
(0.0011)
-2.890
(1.136)***
18
(6.0
-2
(12
-0
(0.
-0
(0.
-0
(0.0
-2
(1
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
Social fractionalization
Ethnic dominance
Ln population
N
No. of wars
Pseudo R2
Log likelihood
-0.0002
(0.0001)**
0.727
(0.368)**
0.747
(0.174)***
674
42
0.25
-118.40
-0.0002
(0.0001)**
0.732
(0.370)**
0.743
(0.175)***
671
39
0.22
-116.17
-0.0003
(0.0001)***
0.655
(0.372)*
0.899
(0.195)***
685
43
0.29
-114.04
-0.0
(0.0
0.6
(0.3
0.7
(0.1
6
4
0.
-12
Notes: All regressions include a constant. Standard errors in parentheses. ***, **, * ind
Column 2: We exclude the following three growth collapses: Angola 1970-74, Iraq 198
Column 3: We exclude the following three influential data points: Iran 1970-74, Rom
Column 4: We exclude the countries with the highest primary commodity export to G
average primary commodity export to GDP ratio is 0.504 (sample average 0.158).
Column 5: We exclude the following war starts: Angola 1975 and Somalia 1988.
Column 6: We exclude the following war starts: Angola 1975, Mozambique 1976, Sierr
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
586 P. COLLIER AND A. HOEFFLER
Table 7 Estimation issues
1 2 3 4
Random
effects time dummies
Primary commodity
exports/GDP
(Primary commodity
exportslGDP)2
Male secondary schooling
(GDP growth) t-1
Peace duration
Geographic dispersion
Social fractionalization
Ethnic dominance
(45-90%)
Ln population
T70-74
T75-79
T80-84
T85-89
190-94
195-99
N
No of wars
Pseudo R2
Log likelihood
18.937
(5.865)***
-29.443
(1 1.782)***
-0.032
(0.010)***
-0.115
(0.043)***
-0.004
(0.001)***
-2.487
(1.005)***
-0.0002
(0.0001)**
0.670
(0.354)*
0.768
(0.166)***
688
46
-128.21
35.850
(14.436)***
-65.967
(26.964)***
0.007
(0.033)
-0.045
(0.072)
0.011
(0.002)***
115.363
(74.562)
-0.007
(0.006)
0.010
(1.410)
145
44
-38.18
18.895
(5.988)***
-29.815
(12.098)***
-0.031
(0.010)***
-0.129
(0.047)***
-0.004
(0.001)***
-2.447
(1.018)**
-0.0002
(0.0001 )**
0.682
(0.359)*
0.762
(0.170)***
0.725
(0.602)
0.578
(0.608)
1.137
(0.602)*
-0.013
(0.757)
0.802
(0.677)
-0.492
(0.921)
688
46
0.26
-124.30
17.161
(6.535)***
-25.594
(14.355)*
-0.029
(0.010)***
-0.110
(0.040)***
-0.004
(0.001)***
-2.394
(1.085)**
-0.0002
(0.0001)**
0.644
(0.336)*
0.726
(0.151)***
688
46
Notes: All regressions include a constant. Standard errors in parentheses. , , * indicate significance at
the 1, 5, and 100 level, respectively
the analysis. Although the fixed effects test is very severe, the non-monotonic effect
of primary commodity exports remains significant. Were the effect of primary
commodity exports dependent only upon cross-section data, it might suggest
that the variable was proxying some other characteristic such as geography.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 587
However, the fixed effects regression uses only changes in prim
dependence, and so reduces the scope for alternative interpretati
We analysed whether time effects matter by including time d
model. Based on a log likelihood ratio test we cannot reject the
the time dummies are zero.30
Finally, in the last column of Table 7 we use a recently develo
method for rare events data (King and Zeng, 2001). The event w
occurs in only about 7% of our observations. King and Zeng sho
logit estimation tends to underestimate the probability of rare ev
used their correction procedure. The differences between the sta
and the rare events corrected results are negligible with all varia
the same levels. The mean of the predicted probabilities obtaine
events logit regression is 0.072. Thus, we find that the correcte
similar to the logit results.
We examined a number of different model specifications. We fo
the following geographic and demographic characteristics were s
coverage, population density and the proportion of young men ag
also investigated the potential endogeneity of income to civil war
we are measuring income prior to war the endogeneity only arise
more than one war. Since the first war will have reduced incom
wars the correlation between income and war could in principle re
causation. To control for this we re-estimated excluding repeat
variable remained highly significant.
5. Interpretation and conclusion
Using a comprehensive data set of civil wars over the period 1960
regressions to predict the risk of the outbreak of war in each fiv
find that a model that focuses on the opportunities for rebellio
29 We also investigated the effect of commodity prices. Since prices are exogenou
contemporaneous with the episode being predicted, whereas our value-based prox
experimented with both the level of export prices and with the change in pri
period. However, in either form when added to the baseline regression the variable
fact that lagged values of exports are significant even in the fixed effects regression
respond to changes in values, but the response is evidently not so rapid as to giv
price response. Potentially, the effect on conflict risk captures the 'voracity effect
Tornell (1999) whereby an increase in the price of a natural resource export would
increment in value to be devoted to conflict. Our results suggest that there may be su
it is lagged.
30 The LRT statistic is 7.83, 6 restrictions (p = 0.251).
31 The proportion of the population living in urban areas was statistically significant when we excluded
the geographic concentration of the population. However, when we included both proxies for the
concentration of the population, the geographic concentration measure remained statistically significant
while the proportion of the population living in urban areas was marginally insignificant (p = 0.11).
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
588 P. COLLIER AND A. HOEFFLER
whereas objective indicators of grievance add little explanatory
robust to a range of tests for outliers, redefinitions, and alter
One factor influencing the opportunity for rebellion is the av
We have shown that primary commodity exports substantial
risk. We have interpreted this as being due to the opportuniti
provide for extortion, making rebellion feasible and perhaps
alternative explanation would be that primary commodity de
governance and so generates stronger grievances. However, we
economic performance-the level, growth, and distribution
political rights (which appear not to affect the risk of conflict
wish to discount the possibility of an effect working through
we cannot control), there is plenty of case study evidence supp
interpretation. Another source of finance for which ther
evidence is diasporas. We have found that diasporas substa
risk of conflict renewal, and it is hard to find an alternative
result.
A second factor influencing opportunity is the cost of rebellion. Male secondary
education enrollment, per capita income, and the growth rate all have statistically
significant and substantial effects that reduce conflict risk. We have interpreted
them as proxying earnings foregone in rebellion: low foregone earnings facilitate
conflict. Even if this is correct, low earnings might matter because they are a source
of grievance rather than because they make rebellion cheap. However, if rebellion
were a protest against low income, we might expect inequality to have strong
effects, which we do not find.
A third aspect of opportunity is military advantage. We have found that a
dispersed population increases the risk of conflict, and there is weaker evidence
that mountainous terrain might also advantage rebels. It remains possible that these
are correlated with unmeasured grievances.
Most proxies for grievance were insignificant: inequality, political rights, ethnic
polarization, and religious fractionalization. Only 'ethnic dominance'-one ethnic
group being a majority-had adverse effects. Even this has to be considered in
combination with the benign effects of social fractionalization: societies characterized
by ethnic and religious diversity are safer than homogenous societies as long as
they avoid dominance. We have suggested that diversity makes rebellion harder
because it makes rebel cohesion more costly. It would be difficult to argue that
diversity reduced grievance.
Finally, the risk of conflict is proportional to a country's population. We have
suggested that both opportunities and grievances increase with population. Thus,
the result is compatible with both the opportunity and grievance accounts.
However, grievances increase with population due to rising heterogeneity. Yet
those aspects of heterogeneity that we are able to measure are not associated
with an increased risk of conflict. Hence, a grievance account of the effect of
population would need to explain why unobserved, but not observed, heterogeneity
increases conflict risk.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 589
One variable, the time since a previous conflict, has substa
heals. Potentially, this can be interpreted either as opportun
may reflect the gradual depreciation of rebellion-specific ca
increasing cost of rebellion, or the gradual erosion of hatred
found that a large diaspora slows the 'healing' process. The k
diasporas to finance rebel groups offsets the depreciation of
capital, and so would be predicted to delay 'healing'. The
lends support to the opportunity interpretation.
Opportunity as an explanation of conflict risk is consisten
interpretation of rebellion as greed-motivated. However, it is
grievance motivation as long as perceived grievances are suf
to be common across societies and time. Opportunity can acco
of either for-profit, or not-for-profit, rebel organizations.
not therefore imply that rebels are necessarily criminals. But
motivate rebels may be substantially disconnected from the
of inequality, political rights, and ethnic or religious identity
Acknowledgements
We would like to thank Todd Sandler and two anonymous ref
comments. Anke Hoeffler gratefully acknowledges funding from
Norway (NFR).
References
Alesina, A., Baqir, R., and Easterly, W. (1999). 'Public goods and ethnic divisions', Quarterly
Journal of Economics, 114, 1243-84.
Alesina, A., Oetzler, S., Roubini, N., and Swagel, P. (1996). 'Political instability and economic
growth', Journal of Economic Growth, 1, 189-211.
Alesina, A. and Spolaore, E. (1997). 'On the number and the size of nations', Quarterly
Journal of Economics, 112, 1027-56.
Angoustures, A. and Pascal, V. (1996). 'Diasporas et financement des conflits', in F. Jean and
J.-C. Rufin (eds), Economie des Guerres Civiles, Hachette, Paris.
Barro, R.J. (1991). 'Economic growth in a cross section of countries', The Quarterly Journal
of Economics, 106, 407-43.
Barro, R.J. (ed.) (1997). Determinants of Economic Growth, MIT Press, Cambridge, MA.
Barrett, D.B. (ed.) (1982). World Christian Encyclopedia, Oxford University Press, Oxford.
Buchanan, J.M. and Faith, R.L. (1987). 'Secession and the limits of taxation: towards a
theory of internal exit', American Economic Review, 77, 1023-31.
Collier, P. (2000). 'Rebellion as a quasi-criminal activity', Journal of Conflict Resolution, 44,
839-53.
Collier, P. (2001). 'Ethnic diversity: an economic analysis of its implications', Econ
Policy, 32, 129-66.
Collier, P. and Hoeffler, A. (1998). 'On the economic causes of civil war', Oxford Eco
Papers, 50, 563-73.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
590 P. COLLIER AND A. HOEFFLER
Collier, P. and Hoeffler, A. (2002). 'Greed and grievance in civil war',
WPS 2002-01, Oxford University.
http://www.economics.ox.ac.uk/CSAEadmin/workingpapers/pdfs/200
Collier, P., Hoeffler, A., and Soderbom, M. (2004). 'On the duration of
Peace Research, 41, 253-73.
Davidson R. and MacKinnon, J.G. (1981). 'Several tests for mode
presence of alternative hypotheses', Econometrica, 49, 781-93.
Dehn, J. (2000). 'Commodity price uncertainty, investment and sh
economic growth', D.Phil. thesis, University of Oxford.
Deininger, K. and Squire, L. (1996). 'A new data set measuring inco
Bank Economic Review, 10, 565-91.
Deininger, K. and Squire, L. (1998). 'New ways of looking at old
growth', Journal of Development Economics, 57, 249-87.
Easterly, W. and Levine, R. (1997). 'Africa's growth tragedy: policie
Quarterly Journal of Economics, 113, 203-49.
Esteban, J.-M. and Ray, D. (1994). 'On the measurement of polarizat
819-51.
Esteban, J.-M. and Ray, D. (1999). 'Conflict and distribution', Journal of Economic The
87, 379-415.
Figes, 0. (1996). A People's Tragedy: The Russian Revolution 1891-1924, Pimlico, Lon
Fearon, J. (2004) 'Why do some civil wars last so much longer than others?' Journal of
Research, 41, 275-301.
Gerrard, A.J.W. (2000). 'What is a mountain?', mimeo, DECRG, World Bank,
Washington, DC.
Grossman, H.I. (1991). 'A general equilibrium model of insurrections', American Economic
Review, 81, 912-21.
Grossman, H.I. (1999). 'Kleptocracy and revolutions', Oxford Economic Papers, 51, 267-83.
Hegre, H., Ellingsen, T., Gates, S., and Gleditsch, N.-P. (2001). 'Toward a democratic civil
peace? Democracy, political change, and civil war, 1816-1992', American Political Science
Review, 95, 33-48.
Herbst, J. (2000). States and Power in Africa, Princeton University Press, Princeton, NJ.
Hirshleifer, J. (1995). 'Theorizing about conflict', in K. Hartley and T. Sandler (eds),
Handbook of Defense Economics, Vol. 1, Elsevier Science, Amsterdam, 165-89.
Hirshleifer, J. (2001). The Dark Side of the Force: Economic Foundations of Conflict Theory,
Cambridge University Press, Cambridge.
Jaggers, K. and Gurr, T.R. (1995). 'Tracking democracy's third wave with the Polity III data',
Journal of Peace Research, 32, 469-82.
King, G. and Zeng, L. (2001). 'Logistic regression in rare events data', Political Analysis, 9,
137-63.
Klare, M.T. (2001). Natural Resource Wars: The New Landscape of Global Confli
Metropolitan Books, New York.
Lane A. and Tornell, P.R. (1999). 'The voracity effect', American Economic Review,
22-46.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 591
Londregan, J.B. and Poole, K.T. (1996) 'Does high income promot
Politics, 49, 1-30.
Long, J.S. (1997). Regression Models for Categorical and Limited
Sage Publications, London.
Mauro, P. (1995). 'Corruption and growth', The Quarterly Journ
681-712.
Pregibon, D. (1981). 'Logistic regression diagnostics', The Annals of Statistics, 9,
Reynal-Querol, M. (2000). 'Religious conflict and growth: theory and evidence'. Ph.D
London School of Economics and Political Science.
Reynal-Querol, M. (2002). 'Ethnicity, political systems and civil war', Journal of Confl
Resolution, 46, 29-54.
Sachs, J. and Warner, A.M. (2000). 'Natural resource abundance and economic growt
in G.M. Meier and J.E. Rauch (eds), Leading Issues in Economic Development, 7th ed., Oxfo
University Press, Oxford.
Sambanis, N. (2002). 'What is a civil war? Conceptual and empirical complexities of
operational definition', mimeo, Yale University.
Sen, A. (1973). On Economic Inequality, Clarendon Press, Oxford.
Singer, D.J. and Small, M. (1994). 'Correlates of war project: international and civil war da
1816-1992', data file, Inter-University Consortium for Political and Social Research, An
Arbor, MI.
Small, M. and Singer, J.D. (1982). Resort to Arms: International and Civil War, 1816-1980,
Sage, Beverly Hills, CA.
The Stockholm International Peace Research Institute (2002). Yearbook of World
Armaments and Disarmaments, Oxford University Press, Oxford.
Summers, R. and Heston, A. (1991). 'The Penn World Table (Mark 5): an expanded
set of international comparisons, 1950-1988', The Quarterly Journal of Economics, 99,
327-68.
USSR (1964). Atlas Narodov Mira, Department of Geodesy and Cartography of the
Geological Committee of the USSR. Moscow.
World Bank (2000). World Development Indicators, Washington DC, data file.
Appendix 1
1. A simple migration model
Our estimation of migration is based on the following model
diasit =1.163 diasit - 0.0002 lnGDPi,t_l + 0.003 wari,t_ + 0.003. T80 + 0.005 T90+ 0.013
(0.045)*** (0.001)** (0.03) (0.002) (0.002) (0.008)
Where dias denotes diaspora which is measured as the ratio of emigra
total population of the country of origin. The variable war is a wa
t - 1 it takes a value of one if the country experienced a civil war
The method of estimation is OLS. The data is measured at the beg
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
592 P. COLLIER AND A. HOEFFLER
i.e. 1960, 1970, 1980, and 1990. The regression includes time dumm
significant.
Based on this simple migration model we estimated the size of the diaspora at time t.
diasit = xit '*
For countries which experienced a previous civil war we used these estimated values to
correct for a possible endogeneity problem. We replaced a total of 64 observations. For
countries which did not experience a civil war we use the actual diaspora data. In order
to obtain values for 1965 we took the averages of this corrected diaspora data measured in
1960 and 1970, and analogously for the values for 1975 and 1985. For 1995 we use the
observations measured in 1990.
Appendix 2
1. Calculating the marginal probabilities
In our regressions we estimate the probability of a war breaking out during a five-year
period, and the model can be written in the following general form
Yit = a + bXit + cMi, t_ + dZi + uit (A2.1)
where t and i are time and country indicators. The dependent variable is a dumm
indicating whether a war broke out during the five-year period, so that Yit is the
war. The explanatory variables are either measured at the beginning of the p
example, income per capita, primary commodity exports/GDP, population), or
previous five-year period (for instance, per capita income growth, or are time in
changing slowly over time (for example, social fractionalization).
The expected probability Pit of a war breaking out can be calculated by using th
coefficients obtained from equation (Al.1):
a + bXit + cMi t- + dZi = it (A2.2)
ewit
pit = 100 (A2.3)
(l+ eit)
Appendix 3
1. Data sources
Democracy The degree of openness of democratic institutions is measured on a sc
zero (low) to ten (high). Source: http://www.cidcm.umd.edu/polity/index.html. The d
described in Jaggers and Gurr (1995).
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
Table A2 Marginal probabilities
Variable Coefficient Mean (1) (2) (3) (
of X At the Worst Best primary 10%
mean commodity/ men in gr
GDP = 0.
Primary commodity 18.937
exports/GDP
(Primary commodity -29.443
exports/GDP)2
Male secondary schooling -0.032
(GDP growth),_ -0.115
Peace duration -0.004
Geographic dispersion -2.487
Social fractionalization -0.0002
Ethnic dominance 0.670
(45-90%)
Ln population 0.768
Constant -13.073
p
0.158 2.992 6.060
-0.735 -3.015
0 6.060
0 -3.015
44.489 -1.406 -0.032 -4.645 -1.406
1.618 -0.186 1.508 -1.660 -0.186
347.5 -1.286 -0.004 -2.19 -1.286
0.602 -1.497 0.000 -2.415 -1.497
1790 -0.376 -0.004 -1.465 -0.376
0.439 0.294 0.670 0 0.294
30,500,000 13.230 16.049 9.136
-13.073 -13.073 -13.073
-2.043 8.160 -16.312
0.115 1.000 0.000
13.230
13.073
-
-
-
-
-
-1
-1.255 -2
0.222 0
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
594 P. COLLIER AND A. HOEFFLER
Diaspora We used the data on the foreign born population from
the Census and divided these numbers by the total population in
http://www.census.gov/population/
Ethnic dominance Using the ethno-linguistic data from the origin
1964) we calculated an indicator of ethnic dominance. This variable
if one single ethno-linguistic group makes up 45 to 90% of the tot
otherwise. We would like to thank Tomila Lankina for the translat
source.
Forest coverage We used the FAO measure of the proportion of a
is covered in woods and forest. Source: http://www.fao.org/forestr
GDP per capita We measure income as real PPP adjusted GDP p
data set is the Penn World Tables 5.6 (Summers and Heston, 1991
available from 1960-92 we used the growth rates of real PPP adj
from the World Bank's World Development Indicators 1998 in or
for the 1990s. These GDP per cpaita data were used to calculate th
rate over the previous five years.
Geographic dispersion of the population We constructed a
population on a country by country basis. Based on population d
generated a Gini coefficient of population dispersion for each count
that the population is evenly distributed across the country and a va
total population is concentrated in one area. Data is available for
prior to 1990 we used the 1990 data. We would like to thank Uwe
Bank's Geographic Information System Unit for generating this da
data sources: Center for International Earth Science Informa
Columbia University; International Food Policy Research Inst
Resources Institute (WRI). 2000. Gridded Population of the W
Palisades, NY: IESIN, Columbia University. Available at http://sedac
Inequality Inequality was either measured as income inquality
Squire, 1996) or as inequality in land ownership (source: Deini
lished). Both inequality measures are provided as a Gini coefficien
Male secondary school enrolment rates We measure male seco
rates as gross enrolment ratios, i.e. the ratio of total enrollment, r
population of the age group that officially corresponds to the l
Secondary education completes the provision of basic education t
level, and aims at laying the foundations for lifelong learning
by offering more subject- or skill-oriented instruction using m
Source: World Bank Development Indicators, 1998.
Mountainous terrain The proportion of a country's terrain w
measured by John Gerrard, a physical geographer specialized in
measure is based not just on altitude but takes into account plate
The data are presented in Gerrard (2000).
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms
GREED AND GRIEVANCE 595
Peace duration This variable measures the length of the peace peri
the end of the previous civil war. For countries which never expe
measure the peace period since the end of World War II.
Population Population measures the total population, the data sour
World Development Indicators 1998.
Primary commodity exports/GDP The ratio of primary comm
proxies the abundance of natural resources. The data on primary co
GDP were obtained from the World Bank. Export and GDP data ar
US dollars.
Social, ethnolinguistic, and religious fractionalization We proxy social fractionalization
in a combined measure of ethnic and religious fractionalization. Ethnic fractionalization is
measured by the ethno-linguistic fractionalization index. It measures the probability that two
randomly drawn individuals from a given country do not speak the same language. Data are
only available for 1960. In the economics literature this measure was first used by Mauro
(1995). Using data from Barrett (1982) on religious affiliations we constructed an analogous
religious fractionalization index. Following Barro (1997) we aggregated the various religious
affiliations into nine categories: Catholic, Protestant, Muslim, Jew, Hindu, Buddhist, Eastern
Religions (other than Buddhist), Indigenous Religions, and no religious affiliation.
The fractionalization indices range from zero to 100. A value of zero indicates that the
society is completely homogenous whereas a value of 100 would characterize a completely
heterogeneous society.
We calculated our social fractionalization index as the product of the ethno-linguistic
fractionalization and the religious fractionalization index plus the ethno-linguistic or the
religious fractionalization index, whichever is the greater. By adding either index we
avoid classifying a country as homogenous (a value of zero) if the country is ethnically
homogenous but religiously divers, or vice versa.
War data A civil war is defined as an internal conflict in which at least 1,000 battle related
deaths (civilian and military) occurred per year. We use mainly the data collected by Small
and Singer (1992) and according to their definitions (Singer and Small, 1984) we updated
their data set for 1992-99.
This content downloaded from
147.251.110.131 on Tue, 22 Mar 2022 12:08:37 UTC
All use subject to https://about.jstor.org/terms