Universidade Autónoma de Lisboa
ISSN: 1647-7251
Vol. 7, Nº 1 (May-October 2016), pp. 73-95
Ricardo Real P. Sousa
Assistant Professor at Autonomous University of Lisbon (UAL, Portugal) and integrated researcher
at OBSERVARE. He has a PhD from the International Institute of Social Studies (ISS) Erasmus
University of Rotterdam (EUR) in the Netherlands. He was part of the Research School in Peace
and Conflict (PRIO / NTNU / UiO) in Norway and associated with the African Studies Center (CEA)
Lisbon University Institute (IUL) in Portugal as a researcher on conflict. He has a Master of
Science in Development Studies from the School of Oriental and African Studies (SOAS) of the
University of London, a post-graduation diploma of advanced studies in African Studies and a
Bachelor (Hons) degree in Business Administration, both from the Lisbon University Institute.
Understanding the initiation of conflict is fundamental for the success of efforts in conflict
prevention. The validity of the mechanisms of the Greed and Grievance” model, alongside
leadership and external interventions are tested in four periods of initiation and
intensification of the conflict in Angola. All mechanisms are present but their relative
relevance varies throughout the conflict. Among the mechanisms identified in each period
the most relevant in the Cold War period are the international and regional interventions in
1961 and 1975 and in the post-Cold War period, the “greed” factors in 1992 (oil and
diamonds, poverty and war capital) and the UNITA leadership of Jonas Savimbi in 1998. The
case study provides evidence that “greed” and “grievance” can be interlinked (such as in
1992) and confirms the relevance of leadership and external interventions mechanisms.
Africa, Angola, Conflict, Greed, Grievance, Leadership, External interventions
How to cite this article
Sousa, Ricardo Real P. (2016). "Greed, grievance, leadership and external interventions in
the initiation and intensification of the civil war in Angola". JANUS.NET e-journal of
International Relations, Vol. 7, 1, May-October 2016. Consulted [online] on the date of
last consultation, observare.ual.pt/janus.net/pt_vol7_n1_art
Article received on 16 February 2016 and accepted for publication on 12 April 2016
JANUS.NET, e-journal of International Relations
ISSN: 1647-7251
Vol. 7, Nº 1 (May-October 2016), pp. 73-95
Greed, grievance, leadership and external interventions in the initiation and
intensification of the civil war in Angola
Ricardo Real P. Sousa
Ricardo Real P. Sousa
Several approaches have been developed to explain the initiation of Civil War. The
“Greed and Grievance” model popularized by Paul Collier attracted intense scrutiny
from researchers.
The model is based on a rational choice approach and contrasts the economic
opportunities in which people are able to organize and finance a rebellion (“greed”), i.e.
rebellion as a criminal enterprise, with social and political motives making people want
to rebel (“grievances”), i.e. socio-economic injustices felt by a social group. The model
is operationalized through a series of proxy variables. The opportunities of would be
rebels are: a1) the funding possibilities available, which can be revenues from natural
resources, remittances from diasporas or support from hostile governments; a2) the
recruitment costs of rebels, determined by alternative income levels; 3) the
accumulated war capital; a4) the capacity of the government to control the territory
measured in terms of how the terrain is appropriate for rebels (forest and mountains)
or how disperse the populations are, and; a5) the social cohesion in society and how
ethnic and religious factors facilitate the establishment and maintenance of conflict
groups. The grievances of would be rebels are: b1) religious and ethnic hatred between
groups; b2) the level of political repression; b3) the political exclusion of groups, and;
b4) the income inequality in the country (Collier and Hoeffler, 2004).
The application of this model to the Civil Wars between 1960 and 1999 concludes that
the main mechanism in Civil War initiation is “greed”, in the desire to acquire economic
benefits and therefore the perceived capacity to organise and maintain a rebellion. The
main “greed” factors are the existence of natural resources (specifically oil),
remittances from the diaspora, the low recruitment costs of fighters, military advantage
in terms of dispersed populations, and the war capital existing in the country (in time
since the last conflict) (Collier and Hoeffler, 2004). The only significant grievance factor
is political exclusion through ethnic dominance, at the same time that ethnic and
religious diversity decreases chances of conflict if ethnic dominance is avoided. Finally,
the size of the population is positively associated with conflict onset2 (these results are
summarized in the column “results” in table 1). Fearon and Laitin (2003) reached
similar results regarding the relevance of “greed” factors in explaining the initiation of
Civil Wars. But they consider that the low income variable is a proxy of lower state
capacity to repress rebellion and consequently lower costs for the rebels to sustain a
rebellion rather than a proxy of lower recruitment costs of fighters, as considered by
Collier and Hoeffler (2004). For Fearon (2005), if oil predicts Civil Wars, it is not so
much by being an entrepreneurial mechanism (as a tempting “prize” for those who
control the state) but mostly because oil producers have low state capacities to repress
1 The review of the text was funded by national funds through FCT, Fundação para a Ciência e Tecnologia,
as part of OBSERVARE project, reference UID/CPO/04155/2013. Text reviewed by Carolina Peralta.
2 This is interpreted more as a greed factor by increasing the likelihood of having secession-seeking
population sub-groups.
JANUS.NET, e-journal of International Relations
ISSN: 1647-7251
Vol. 7, Nº 1 (May-October 2016), pp. 73-95
Greed, grievance, leadership and external interventions in the initiation and
intensification of the civil war in Angola
Ricardo Real P. Sousa
rebellion relative to their levels of income per capita. Oil rich countries have less
incentive to develop the state apparatus required for revenue collection.
Following similar rationalist-positivist approaches, the validity of the model was tested
with quantitative and qualitative assessment. Quantitative analysis by Hegre and
Sambanis (2006) confirmed several of the model’s results, including the
entrepreneurial (Collier and Hoeffler, 2004) and state repression claims (Fearon and
Laitin, 2003).
The model was later reviewed to consider that it is the financial and military feasibility
of conflict that increases the likelihood of Civil War initiation. Feasibility is mainly
measured in terms of: the country being a former French colony and therefore under
the security umbrella of France, making rebellion less likely to succeed; the proportion
of young male in the country who are potential fighters, and; a mountainous terrain
that makes rebel military action feasible (Collier, Hoeffler and Rohner, 2009).
Therefore, the issue is not so much if there is a “grievance” motive, or if there is a
“greed” opportunity, instead if insurrection is feasible.
We test in this paper the original “greed” and “grievance” model for three reasons. One
is that the feasibility factors are difficult to test within a single case study, as they don’t
vary significantly across time. A second reason is that the results of the feasibility
model reconfirm the results of the original model in the sense that “greed” factors are
still significant and “grievance” are not (Collier, Hoeffler and Rohner, 2009). A third
reason is that there is no agreement on the greed” and “grievance” debate and this
debate has not been replaced by a “feasibility” debate.
The current “greed” and “grievance” debate focuses on which of the mechanisms
explains the initiation of Civil Wars, the epistemological foundations of the studies and
the policy implications of the results.
The “grievance” argument can be traced back to the “relative deprivation” theory that
proposed that psychological mechanisms associated with a frustration of not meeting
material expectations are at the root of conflict initiation (Davies, 1962; Gurr, 1970).
Tilly (1978) contested this argument considering that grievance factors are widespread
in society and conflict is not present in all societies. Instead, it is the capacity to mount
a rebellion, determined by access to material and organisational resources, that
differentiates societies where Civil War is initiated or not. With Gurr’s work (1970,
2000) on ethnic conflict, group level grievances gain another capacity to explain the
initiation of conflict.
In the tradition of the “grievance” argument, it has been suggested that rebellion
occurs in cases of multidimensional horizontal inequalities (Stewart, 2002). Horizontal
inequalities occur when social exclusion and poverty and identity or regional boundaries
take place simultaneously. Buhaug, Cederman and Gleditsch (2014) used horizontal
inequalities as a proxy for inequality, instead of the GINI coefficient used by Collier and
Hoeffler (2004) and Fearon and Laitin (2003), which reflects vertical inequalities -
inequality among values of a frequency distribution of income, economic interpersonal
inequality. They found that horizontal inequalities are an important factor in rebellions
and a better predictor of rebellion than vertical inequalities.3
3 Similar results were found by Cederman, Weidmann and Gleditsch (2011) and Østby (2008).
JANUS.NET, e-journal of International Relations
ISSN: 1647-7251
Vol. 7, Nº 1 (May-October 2016), pp. 73-95
Greed, grievance, leadership and external interventions in the initiation and
intensification of the civil war in Angola
Ricardo Real P. Sousa
The qualitative analysis of the model through a series of case studies (Collier and
Sambanis, 2005) confirms its main results. Nevertheless, it also identifies a series of
limitations and reflections, some of which this case study on Angola is particularly
suited to investigate.
One limitation of the model is the absence of leadership as a factor. This is mainly due
to the fact that leadership is hard to quantify. There are two theories on the role of
leadership in mobilizing ethnic groups. One, from rationalist and constructivist
accounts, suggest that there is a social construction of identity by political elites in
order to mobilize and manipulate ethnic groups into fighting (Gurr 2000). These differ
from the primordial perspectives, which consider that there is an innate conflict
propensity in ethnic identity (Brubaker 1995).
Another limitation is the absence of an account of the role of external interventions.
The Collier and Hoeffler (2004) theory uses a Cold War dummy variable to proxy for
this effect, finding no statistically significant relationship4. But this variable does not
capture the nuanced effect of exogenous variables on Civil War. The Cold War had
different periods of intensity, between the post-Second World War and 1991, when it
ended with different levels of involvement of external actors. Also, it had different
expressions at international and regional level. At the regional level, there can be
diffusion and contagion effects. Diffusion occurs through demonstration, where political
events in one country inspire political action in another. Contagion occurs through:
common ethnic groups across borders; accumulation of war capital (for instances small
arms) in specific regions; refugee movements, or external interventions (Sambanis,
2005). The effect of external interventions on Civil War is one of the understudied
relationships in the literature (Sambanis, 2002). External interventions support the
fighting parties, affecting their propensity to fight. Military interventions increase
directly the military capacity to fight, and economic interventions decrease the
coordinating costs of sustaining a rebellion by increasing the likelihood of success
(Elbadawi and Sambanis, 2000). Diplomatic interventions are normally intended to find
a non-violent solution to the conflict and through information sharing they can increase
the chances of reaching a political solution. Generally speaking, evidence has been
found of the escalatory effect of military interventions and de-escalatory effect of
economic and diplomatic interventions in Civil Wars (Regan and Meachum 2014, Sousa,
These limitations can be contextualized in broader epistemological considerations. It
has been argued that the rational choice approach and methodological individualism of
these studies fail to take into account social, relational and historical aspects (Cramer,
2002). Furthermore, statistical inference is distinct from causality and positivism can
fall into tautological explanations of the phenomenon, based on datasets detached from
the meanings that events have on the ground (Korf, 2006).
A main reflection of Sambanis (2005) is that one should be looking at “greed” and
“grievance” as alternative shades of the same phenomena, and not as competing
explanations. A few mechanisms can be considered to illustrate this hypothesis. For
instance, functional political institutions may decrease political grievances but at the
same time good economic performance can promote the stability of institutions and, in
4 The other exogenous variable used is remittances from the diaspora, but, for data reasons, this is limited
to remittances from the USA.
JANUS.NET, e-journal of International Relations
ISSN: 1647-7251
Vol. 7, Nº 1 (May-October 2016), pp. 73-95
Greed, grievance, leadership and external interventions in the initiation and
intensification of the civil war in Angola
Ricardo Real P. Sousa
this way, affect grievances. Also state failure or government illegitimacy leads to
domestic anarchy, in which case “greed” can be considered the pursuit of survival by
groups in society.
Finally, the relevance of this debate can be apprehended on its policy implications on
how to prevent Civil War. The “greed” explanation puts a focus on: economic growth
and diversification; control and management of natural resources, and; state strength
and external interventions to improve its capacity. “Grievance” explanations highlight:
the indivisibility of some issues, such as identity, ethnicity or religion; the need for
ethnic inclusion and fairer distribution of wealth in the country; mediated solutions
among parties, and external interventions in order to secure the commitment to peace
Inspired on the qualitative work of Collier and Sambanis (2005), the main contribution
of this paper is two-folded. One is to test the “greed” and “grievance” hypotheses as
alternative but also complementary explanations of Civil War alongside the normally
omitted variables of leadership and exogenous effects at international and regional
level, in the form of external interventions (or processes of diffusion). Another is to
apply the model to an historical case study on Angola, which has not been done before.
The Angolan Civil War spanned from the war of independence through the Cold War
and up to the post-Cold War era. It had more than 500,000 deaths, tens of thousands
of persons mutilated by anti-personnel mines and the displacement of approximately
4.1 million people.
This study follows the definition of Civil War advanced by Gleditsch et al (2002),
whereby it consists of a contested incompatibility regarding a government and/or
territory with the use of force by parties, and where at least one of them is the state or
government, resulting in at least 25 deaths in battle5.
The conceptualization of variables, in this case the dependent variable of Civil War, is
one of the challenges in quantitative studies (Sambanis, 2004). Broadly speaking, in
the case of Angola there were two types of war: the war of colonial independence
initiated in 1961, also called extra-systemic war, and an internationalized Civil War
since independence, between 1975 and 2002. The war of colonial independence has
particularities that differentiate it from the subsequent war (the issues and parties
involved) which could merit a separate analysis. But because the original analysis of
Collier and Hoeffler (2004) includes these types of war, it will also be considered. The
Civil War initiated after independence is internationalized because it had military
involvement of external actors. The question here is to consider if relapses into conflict
after periods of peace following a peace agreement should be considered a new Civil
War or not. In Angola the relapses had the same conflict parties fighting over the same
issue and therefore it does not fit completely into a classification of a new Civil War.
5 All the classifications used in the paper and dates of initiation/intensification of conflict are from this
source, unless otherwise identified. The periods and sub-periods used in this paper significantly match
those of Sambanis (2004) and Collier and Hoeffler (2004) with small differences of one year due to the
level of violence considered: Sambanis (2004) and Collier and Hoeffler (2004) consider the second period
to end in May 1991, the year of the Bicesse peace accords, even if technically both 1991 and 1992 are
considered as conflict years in Gleditsch et al (2002); Sambanis (2004) considers the third period to end
in 1994, and 1995 to be a year with very low violence not reaching the threshold to be considered in
conflict, while for Collier and Hoeffler (2004) the conflict that started in 1992 only ends in 2002;
Sambanis (2004) considers the fourth period to start in 1997, due to the escalation of violence in that
JANUS.NET, e-journal of International Relations
ISSN: 1647-7251
Vol. 7, Nº 1 (May-October 2016), pp. 73-95
Greed, grievance, leadership and external interventions in the initiation and
intensification of the civil war in Angola
Ricardo Real P. Sousa
Also the model of the “initiation” of Civil Wars aims to identify the key mechanisms
associated with a qualitative change on the political processes in a country, where
actors decide to move from non-violent conflict to violent conflict. The mechanisms
present in these cases are not necessarily the same as in situations of relapse. In
relapses, the fighting group and war capital already exist and this can have a decisive
effect on what factors explain Civil War initiation. This path-dependence is difficult to
analyse and is also present in the transition from the war of independence to
internationalized Civil War, where one can find some of the same fighting parties even
if fighting against a different party (different dyads). For these reason the analysis will
consider two initiations, the war of independence and the internationalized Civil War,
and, in the latter war, two intensifications of the conflict
Four periods are identified in the Civil War in Angola between 1961 and 20026. The first
period begins in February 1961 with the initiation of the war of independence against
Portugal and extends up to July 1974, when a ceasefire is signed between Portugal and
the nationalist movements. The second period begins in November 1975, when the
internationalized Civil War initiates and ends with the Bicesse accords in May 1991. The
third period starts with the intensification of conflict after the September 1992 elections
and ends in 1995 when the conflict intensity decreases significantly. The final period
ranges from March 1998, when the conflict restarts, to April 2002, when it ends7.
Table 1 resumes the values of the proxy variables of the Collier and Hoeffler (2004)
model for the years closest to conflict initiation in the period between 1960 and 1995.
It compares the values for Angola with the averages for all countries, countries where a
Civil War did not start and countries where a Civil War started. In Angola, the main
“greed” indicators are propitious to conflict initiation: for funding, natural resources are
above the average and the recruitment costs are below the average8 of the values for
countries where a Civil War started; at the same time, the possibilities for state control
are diminished as both geographical dispersion and population are higher than the
average in countries where a Civil War started, and; the grievance indicators are less
favourable to conflict initiation as the social fractionalization is high throughout but
without ethnic dominance.
The historical analysis in this paper identifies both “greed” and “grievance” factors in
moments of the initiation or intensification of Civil War in Angola9. It suggests that
exogenous variables for international and regional dimensions and the endogenous
variable of leadership improve the model´s explanatory power. The article follows a
chronological order of the four moments of initiation or intensification of Civil War, with
a description and analysis of the “greed”, “grievance” and exogenous dynamics. It then
proceeds to analyse the leadership dynamics, which is better understood across
6 This article does not analyse the Cabinda conflict, which occurs at the same time with similar but in many
cases particular dynamics.
7 Collier and Hoeffler (2004) consider the Civil War in Angola to have started in 1961, 1975 and 1992 and
ongoing in 1999, the last year of the dataset. Because 1996 and 1997 are years not classified in conflict
by Gleditsch et al (2002), 1998 is added here as another intensification of the conflict.
8 Except for higher economic growth in 1965 (not shown in table) and 1998.
9 Because for Angola there is no data on GINI and the horizontal inequalities indicators are non-variant in
the period, the evidence of the grievance is based on case studies.