Alliances in intrastate violent conflict
A systemic interpretation of identity and power in the
Sierra Leone civil war
Radboud Universtity – Nijmegen School of Management Master Thesis for Political Science - Conflict, Power and Politics
Hugo Triesscheijn - 1014007 Date: 04-08-2019
Supervisor: Dr. Romain Malejacq Wordcount: 18.839
Abstract
In this thesis I look into the effects of the distribution of power and ideology amongst the armed actors of a civil war on their alliance formation behaviour. By determining how many powerful armed groups with different ideologies are present within a conflict I consider an armed conflict to be bipolar (two such groups) or multipolar (more than two). In bipolar conflicts armed groups form alliances with groups with the same identity. In multipolar conflicts two powerful armed group with differing ideologies that perceive a third armed group to be threatening based on its ideology and power can also form an alliance. Furthermore, weak armed groups with an ideology different to those of the powerful armed groups can form alliances with these powerful armed groups because they do not threaten them due to the power difference. Moreover, I expect the weak armed groups that ally with a powerful armed group with a different ideology to cede some influence in their internal affairs, which can result in a change in ideology for the weak armed group. I do not expect this to happen if the ideological difference is ethnic in nature. I find that this framework is congruent with the developments in the Sierra Leone civil war, as long as ideology is the dominant motivation for armed groups.
Abbreviations
AFRC: Armed Forces Revolutionary Council APC: All People’s Congress
CDF: Civil Defence Force
ECOMOG: ECOWAS Cease-fire Monitoring Group ECOWAS: Economic Community of West-African States EO: Executive Outcomes
IMF: International Monetary Fund IR: International Relations
NPRC: National Provisional Ruling Council NPFL: National Patriotic Front of Liberia PGM: Pro-Government Militias
RUF: Revolutionary United Front SLA: Sierra Leone Army
SLPP: Sierra Leone People’s Party
UNAMSIL: United Nations Mission in Sierra Leone
UNOMSIL: United Nations Observation Mission in Sierra Leone WSB: West Side Boys
Tables
Index
Abstract 2 Abbreviations 3 Tables 3 1. Introduction 5 The Puzzle 6 Methodology 9 Relevance of research 12 Outline 122. Ideology and power from a systemic perspective 14
Threats and alliance formation 14
Defining the concepts 16
Systemic ideological polarity 19
Alliance shelter 22
Alternative theories 25
Patronage 25
Ethnicity 26
Territorial control 27
3. The Sierra Leone civil war 28
Sierra Leone before the civil war 28
The Sierra Leone civil war (1991-2002) 30
Actors in the war 33
The Revolutionary United Front 34
The Sierra Leone Army 36
The Civil Defence Forces 38
4. Explaining alliances in the Sierra Leone civil war 41
Systemic ideological polarity and alliance shelter 41
The relative power of armed groups 41
Power, ideology and alliances 43
Patronage alliances 45
Ethnic alliances 46
Territorial control alliances 47
Congruence and causality in the civil war of Sierra Leone 48
5. Conclusion 51
1. Introduction
Most of the armed conflicts since 1945 have been intrastate conflicts.1 Close to half of those
conflicts consist of more than two armed groups at a single point in time.2 Examples include
the conflicts in Sudan (1983–2005), Afghanistan (1992–1996 and 2001–present), and Syria (2011–present). Such conflicts tend to last longer and result in higher numbers of civilian casualties due to the presence of multiple armed groups, which complicates conflict resolution.3 Furthermore, these conflicts, civil wars, are often recurrent and contribute greatly
to the deconstruction of already weak or failing states.4
A civil war, or intrastate conflict, is an armed conflict between armed groups within the same state or country.5 These armed groups include the government of the territory in
which the conflict is taking place, the pro-government militias (PGMs) and rebels that operate within this territory. These armed groups function as crucial military and political actors, whose actions have major consequences for conflict dynamics and significantly contribute to the complexity of the conflict. Although scholars have studied these armed groups, their relation to each other has until recently received very little academic attention.6 This limits the
understanding of civil wars because alliances between armed groups in a conflict can be decisive for both the conduct and conclusion of the conflict.7 A better understanding of
alliance formation in civil wars contributes to the understanding of intrastate conflicts themselves and might help in mitigating their most harmful effects. Furthermore, knowledge of alliance formation behaviour can provide new insights into why armed conflicts flare up after they have seemingly ended.
Scholars have long treated intrastate conflicts as dyadic conflicts in which a government and a rebel group seek to defeat each other in a competition for political power in or control over the state.8 Yet in multiple faction civil wars, the government, rebel groups and
PGMs all compete for political power with each other.9 The dyadic model is not
representative of such complicated conflicts and must be adapted.10 In an attempt to further
1 Buhaug, Gates, Hegre and Strand, 2007
2 Buhaug, Gates, Hegre and Strand, 2007; Walter, 2019
3 Hultman, 2007; Wood and Kathman, 2015
4 Rotberg, 2002; Quinn, Mason and Gurses, 2007; Zeigler, 2013 5 Kalyvas and Kenny, 2010
6 Buhaug and Rød, 2006; Gent, 2011
7 Akcinaroglu, 2012
8 Coggins, 2015; Otto, 2017 9 Otto, 2017
academic understanding of alliance formation in civil wars, scholars have recently turned towards other (sub-)disciplines, most notably International Relations (IR).11
IR scholars concern themselves with the actions of states. These states, realists scholars argue, are positioned in a system of anarchy and are therefore free to adjust their behaviour. For states, alliances are instruments of national security policy.12 Alliances are
agreements between states to help each other through military means. Concepts such as ‘power’ and ‘threats’ have been used to theorise about the forming of alliances between actors. Scholars concerned with alliances in civil war argue that IR paradigms are applicable because rebel groups in a civil war operate in an anarchic system, just like states.
The application of these IR paradigms has generated various insights about alliances between armed groups in a civil war. The most important of these insights concern the use and risks of alliances. Akcinaroglu demonstrates that alliances can bring benefits such as increased security and the possibility to carry out joint military operations against an adversary.13 Alliances can prove essential to the survival of an armed group in a civil war.
However, Gade, Gabbay, Hafez and Kelly theorise that alliances can also function to limit decision-making autonomy and decrease options for an armed group to expand its power base or capture resources.14 Also, alliances come with the risk of betrayal by an allied faction.15 For
these reasons, armed groups must be careful when considering with which faction they ally themselves.
The Puzzle
What do armed groups in a civil war take into account when they form alliances? Christia argues that the power of a potential ally influences this decision. She argues that rebels looking for an alliance seek to form ‘minimal winning coalitions’. This is a collective of armed groups that expect to be big and powerful enough to win a conflict. At the same time, the members aim to keep the coalition as small as possible to maximise their political gain from being part of the victorious alliance.16
However, Pedersen and Walther, in a recent study of alliance formation in the Syrian civil war, state that armed groups sought to create a ‘maximum winning coalition’. This
11 Bond, 2010; Christia, 2012 12 Snyder, 2007
13 Akcinaroglu, 2012
14 Gade, Gabbay, Hafez and Kelly, 2019 15 Sztompka, 1999
coalition formed between armed groups with relatively similar ideological preferences.17
Therefore power, although important, cannot be the sole consideration in alliance formation decision-making. Christia’s framework of ‘minimal winning coalitions’ is thus an incomplete model for alliance formation and requires further theorising.
Another important consideration, as Bond demonstrates, is whether a potential alliance partner can be trusted. The ideology of an armed group is a major aspect of trust between armed groups.18 However, the current understanding of the effect of ideology as described by
Bond cannot explain changes in alliances between armed groups if the ideology of armed groups does not change. Yet, such changes in alliances between seemingly ideologically constant armed groups do occur. Bond calls for further study on system-level explanations for alignment, which focuses on power and ideology distribution. As such, recent scholarly explanations for alliance formation in a civil war are incomplete and further study of alliance formation in civil wars is required to understand it better.
In a civil war, rebels stand to gain from alliances. Alliances might even be necessary for survival against a threat. At the same time, rebels do not want to be taken advantage of through an alliance with a partner that seeks to exploit them or might drag them into new and unrelated conflicts.19 These two forces, one pro-alliance formation and one contra-alliance
formation, work as a balance. My argument is that that these forces are (partially) dependent on the system within which the armed groups find themselves. Both ideology and power attributes can function as pro- and contra-alliance formation forces as they influence the threat assessment of armed groups. However, their effect on the alliance formation decision-making of armed groups is determined by the system, the collective of armed groups with a certain power and an ideology.20 In other words, the ‘value’ of these characteristics must be
interpreted in relation to the other armed groups in the conflict. As such, to understand armed group alliance formation behaviour, it is essential to understand what kind of systems are theoretically possible, how these systems influence alliance formation behaviour, and which system is present in a specific conflict.
Ideology is found to have a significant effect on the likelihood of alliance onset. However, until now, this effect has been theorised as an isolated mechanism.21 Yet, some IR
scholars argue that, besides the perception of one ideology by another, the distribution of
17 Pedersen and Walther, 2018
18 Bond, 2010
19 Bapat and Bond, 2012
20 Haas, 2014
ideologies is highly influential in alliance formation processes.22 These distributions are
defined by the number of poles, distinct ideologies professed by powerful actors, in the system. This theory thus emphasises the role of systemic ideological polarity. Unipolar and bipolar systems create incentives for effective balancing against threats based on stable alliance systems, whereas ideologically multipolar systems can result in incentives for either effective or ‘under’ balancing, a situation in which actors need to increase their military capacity or form alliances to be able to deter or defeat an enemy, but fail to do so.23 As multi
factional civil wars in which ideology is the source of conflict always consist of a government and multiple armed groups, the effects of unipolar systems are of little relevance for research on these conflicts.
In the brief description of systemic polarity above, I have already mentioned the concept of power. I further define this concept in the next chapter. An actor can only count as a pole if it is a powerful actor and the abovementioned theory only explains the alliance behaviour of actors with an ideology that is represented in a pole. What can we then say about weak actors with an ideology that is not shared by any powerful actor? Apart from Gade, Gabbay, Hafez and Kelly’s recent effort, scholars have not yet attempted to formulate a theory of alliance formation behaviour in civil wars between actors of different power levels.24
Bailes, Thayer and Thorhallsson, argue that weak states can demonstrate different alliance formation behaviour than powerful states. These weak states require protection and do not pose a direct threat to a powerful state. As a result, the weaker state can form an alliance with a strong state and accept limitations from this alliance partner in order to protect itself from a threat. Therefore actors of different power levels should not be treated as if they were identical. If this mechanism applies to alliance formation in civil wars, the effect of relative power difference on alliance formation behaviour requires recognition and further theoretical development.
In this thesis, I attempt to uncover the influence of the system in which armed groups in a civil war find themselves and the effect of relative power differences on alliance formation behaviour. The main research question I seek to answer with my research is: ‘What
is the effect of the systemic characteristics on alliance formation in a civil war?’ This
question can be divided in two sub-questions: ‘What is the effect of the systemic ideological
polarity of the system on the alliance formation in civil war?’ and ‘What is the effect of the relative power difference between actors on the alliance formation in a civil wars?’
22 Haas, 2014; Gause, 2017; De Keersmaeker; 2017
23 Haas, 2014; Gause, 2017; De Keersmaeker; 2017 24 Gade et al., 2019
To answer these questions, I set out to test the effects of the system in which the armed groups find themselves and the distribution of specific attributes amongst the armed groups on the threat assessments of these factions. The central argument builds on the current understanding of the effects of ideology and power on alliance formation. I understand these effects in the context of the system of armed groups in a civil war. I expect the effect of ideology on alliance formation behaviour to differ between armed groups whose ideology is represented in a pole and armed groups whose ideology is not represented in a pole.
Methodology
To provide an answer to my research question, I construct a theoretical framework based on existing theories. This deductive approach allows me to formulate expectations and test these in an empirical case. The framework I develop is based in a constructivist position. As such, I start from the position that people act based on certain social constructs.25 However, these
social constructs, I argue, can only be interpreted through a systemic view. Therefore, both macro and micro variables are required to inform the framework I develop. Although social scientists have for a long time not combined the variables, structuration theory provides a means of doing so.26
Structuration theory, developed by Anthony Giddens, gives primacy to neither agency nor structure and posits that both should be taken into account. As such, it requires an evaluation of both micro and macro variables. Structuration theory is an ‘analytical’ rather than a ‘substantive’ theory. It is about the analysis rather than the substance of the social world. Structuration theory helps the researcher define what kinds of entities are present in the social world and how their relationships can be conceptualised. As such it provides a conceptual framework or meta-theory for thinking about real world social systems, but it does not tell us what particular kinds of agents or what particular kinds of structures to expect in any given concrete social system.27 In the framework I develop, this means that the armed
groups do not need to behave identically. Their relations with each other and the behaviours that result from these relations can be different and can change over time as the groups transform.
Within structuration theory, two schools of thought exist: economic and reflective.28
The economic school of thought sees actors as heavily constrained by their place within a
25 Craig, 2010
26 Giddens, 1984
27 Wendt, 1987
structure. The reflective school interprets these constraints more loosely by pointing out the ability of individuals to redefine their interests according to their experiences. Wendt, subscribing to the second school of thought, sees structure not as causal but as ‘possibilistic’, presenting conditions that allow, but do not dictate, events to occur.29 Furthermore, this
approach accepts that actors influence their structure and vice versa. By combining this structural research with historical research, both the possibilities and real outcome can be studied. I study both the actors and system of an actual, or historical, development, to analyse their relations through my framework.
Structuration theory is based on the ‘scientific realist’ approach, rather than the ‘empiricist’ approach to research.30 The scientific realist approach, although multiple
interpretations exist, states that ontological status can be ascribed to unobservable entities and that to infer causality it is necessary to identify an underlying, often unobserved, causal mechanism. As such, ‘the “scientific realist” explains how, often unobserved, causal mechanisms which make observable regularities possible works.’31
I use the reflective variant of structuration theory in my thesis as the foundation of my framework. I develop this framework through the deductive approach, starting with theoretical expectations and testing these on an empirical case. A case study is a ‘detailed examination of an aspect of a historical episode to develop or test historical explanations that may be generalisable to other events.’32 The study of multiple phenomena of interest, the
formation of alliances, makes this study a within-case-study. This allows for the gathering of useful data on a particular phenomenon, even with modest time and resources.
As a case I use the civil war in Sierra Leone (1991-2002). This violent conflict has ended and is extensively researched. As such, it allows for a body of secondary literature that can reasonably be expected to provide a representative history of the conflict, which ensures the internal validity of my research. It furthermore saw a change in alliances between armed groups and thus contains the dependent variable in which I am interested. I specifically study the conflict in the period March 1991 to October 1999. After this period the personal aims of leaders of armed groups become dominant over ideological consideration.33 This is
problematic for the application of my framework. While this conflict can provide insights into many conflict-related social and political phenomena, the scope of this research allows only
29 Wendt, 1987
30 Wendt, 1987
31 Wendt, 1987, p.354. Italics in original. 32 George and Bennet, 2005, p.5
for a specific focus on alliance formation between armed groups in this conflict. Three armed groups and their alliance formation behaviour are the subjects of the case study.
The case study serves to test the empirical applicability of the framework I develop. The use of a case study helps me to achieve high conceptual validity and allows me to compare alternative explanations with the explanation I propose. As such, it enables me to assess the ability of the framework to explain the outcome in a particular case. I thus employ the congruence method.34 Rather than establishing the causal process, the congruence method
tests the consistency between the predictions of a theory and the empirical outcome. This study thus explores the possibility of a causal relation rather than establish this relation. This function is especially useful for theory building.35 Congruence is present when the variation(s)
in characteristics that are theorised to cause a change, the independent variables, match the expected change of a result, the dependent variable. Ascertaining the value of the characteristics, their expected effects, and consecutively comparing this effect with the actual results can establish congruence. The establishment of congruence indicates the possible existence of a causal relation. The existence of this causal relation is not established. As such, the congruence method functions as a plausibility probe for a certain theory. If congruence is established, the next research step is to test the spuriousness, causal priority and causal depth of a theory. Although this falls outside the scope of this research, I attempt to provide some advice as to how these can be tested.
While the confirmation of theoretical expectations within this study does not automatically establish a new generalisable theory, it does signal a new insight, which upon further research and broader application could inform general theory. I thus perform theory building and exploratory research in an attempt to identify the potential of a causal mechanism. A single case study serves this purpose. This means that the external validity of my research is limited and generalisations require further study.
Relevance of research
My study of armed group behaviour in civil wars builds on the recent theoretical developments concerning the study of rebels, pro-government militias and governments, which are essentially capable of identical behaviour. On top of this, I differentiate between expected group behaviour based on the relative power of an armed group. This distinction is neutral in the sense that it applies to all armed groups in a civil war. My research thus
34 George and Bennet, 2005
contributes to a holistic understanding of the selection of alliance partners. Furthermore, I combine previous findings on the effect of micro-level information, the characteristics of armed groups, to form a systemic understanding of the effects of the distribution of these characteristics. The final theoretical contribution of my research is the understanding of these systemic distributions in a non-deterministic manner. In other words, I move away from direct and singular causality in favour of a possibilistic, or enabling, understanding of systemic effects on alliance formation.
Besides contributing to the academic discourse, my study also produces insights for policy formulation. Alliances and changes in alliances of armed groups significantly influence the options for both conflict resolution and conflict recurrence.36 The insights from this study
can thus guide effort for sustainable resolution of armed conflict. Another relevant policy aspect is the protection of civilians, which is the primary objective of United Nations peacekeeping missions and is high on the agenda of the international community in general.37
Changes in the balance of power in an intrastate conflict affect incentives for the use of violence against civilians.38 The formation and end of alliances greatly affect the balance of
power within an intrastate conflict and thus influences these incentives.39 On top of this,
changes in alliances have even been linked to the emergence of new organisations, preferences, and identities within a conflict.40
Outline
In the next chapter of this thesis, I discuss why armed groups form alliances, what I understand to be an armed group, and how I operationalise its characteristics, power and ideology. Furthermore I present the framework, which looks at these characteristics through a systemic lens and predicts alliance preferences for armed groups. The framework differentiates between armed groups with an ideology represented in a pole and those with a different ideology. Finally, I discuss several alternative theoretical explanations for alliance changes in civil wars. In the third chapter, I briefly discuss the socio-political situation in Sierra Leone before the civil war and then I cover the conflict itself. In this chapter I furthermore identify the characteristics of the armed groups in the civil war as discussed in the previous chapter. In chapter four, I apply these characteristics of the armed groups to my framework. I also analyse the explanatory value of the alternative theories I have presented
36 Cunningham, 2016
37 Hultman, Kathman, Shannon, 2013 38 Wood, Kathman and Gent, 2012 39 Christia, 2012
and discuss how these relate to my framework. I find that my framework is congruent with the developments in the Sierra Leone civil war. In the fifth and final chapter, I discuss the implications of my findings, the strengths and weaknesses of my framework and offer some concluding remarks regarding potential future research subjects.
2. Ideology and power from a systemic perspective
Threats and alliance formation
Waltz’s balance of power theory argues that states balance against the most powerful states in order to be able to defend themselves against these powers.41 He studied the actors
within a system that behave under conditions of anarchy. In 1985, Walt adjusted this theory when he argues that states instead seek to defend themselves from the biggest threats, not from the biggest powers. He defined threats as an aggregate of power, proximity, offensive capabilities, and offensive intentions.42
Thucydides was perhaps the first to write about threats and the need to react to them by allying with or against the threatening actor. Threats can be verbal and non-verbal, through words or actions. However, a threat does not unambiguously speak for itself, it must be interpreted.43 This means that the perception of the actors involved matters greatly. Scholars
have argued for various explanations about the source and problems of threat interpretation. To understand where the framework I develop fits in this academic debate, I briefly present the major explanations.
Rationalist theorists focus on incomplete information explanations for threat interpretation. They argue that rational actors perceive threats and act on these threats because they have incomplete information. If actors had complete information, they would be able to reach an agreement based on their relative power. Threats would thus not occur, as the actors would forgo conflict in favour of a non-violent redistribution of resources.44 Another
complexity regarding threat is the security dilemma. An actor seeking security from a second actor other might increase its military standing to ensure the other actor does not attack. The second actor, however, can interpret this as an increased threat of becoming the subject of attack itself. As a reaction, the second actor will also increase its military standing, leading to a rising spiral of military expenditure, perceived threats and counterbalancing, even though both actors seek security.45 Another scholarly focus looks towards the effects of ideas and
ideology and how these influence the threat assessments of actors.46 The final scholarly
41Waltz, 1979
42Walt, 1985 43 Stein, 2013 44 Stein, 2013
45 Jervis, 1978; Fearon, 2011
attempt to understand threat perceptions focuses on the structural attributes of the political system the actors find themselves in.47
The framework I develop in this thesis is designed to test the effects of the ideological polarity structure in which the armed groups find themselves on their threat assessments. As such, I combine the last two academic interpretations on threat perception discussed above. By applying the established threat perception approach of ideology influence with the, until now sparsely used, structural threat approach, I aim to contribute new knowledge to the intrastate alliance formation debate. As such, my framework consists of the domestic actors present within a conflict, but importantly also looks at the relative power distribution of actors within this system and their respective ideology. In Walt’s terms, the aggregate power, proximity, and offensive capabilities of the actor that is sending the threat must be of such a level that the threat-receiving actor perceives it as a threat.
Aggregate power refers to the resources an actor possesses to exert influence. As the armed groups in a civil war seek to influence the outcome of this conflict, the power they command to do so is the most important. Therefore, the military capacity of the armed groups is the most important aspect of their power. As this measurement of power refers to the credibility of threats from one group to another group, the military aspect of power is central to this effect. Therefore, I understand the power of an armed group to be equal to the military capacity of the group.
The proximity, in a geographical sense, tells us that the source of a threat must be close enough to execute it in order for it to be deemed credible. Within a civil war this characteristic is hard to clearly define, as often lines of separation are not as absolute as they are in international relations. Therefore, I assume that all parties within the conflict, if powerful, are geographically close enough to function as the source of a threat.
Offensive capabilities indicate the military capacity of an armed group to execute offensive military action. In the international sphere, the possession (or lack) of specific capabilities can significantly influence the possibility of one state to attack another one. While the distinction between offensive or defensive weapons is not absolute, the kind of weapons a state possesses and how it plans to use them matters. It might even affect the security dilemma discussed earlier as a state would not perceive a threat if the other state owned primarily defensive weapons.48 The credibility of a threat in the international sphere is thus greatly
affected by the possession or lack of offensive means. However, in an intrastate conflict, the
47 Allison & Zelikow, 1999; Haas, 2014 48 Biddle, 2001
means of warfare are often much simpler. While this does not make intrastate conflicts less deadly, on the contrary, it does mean that the difference between ‘offensive’ and ‘defensive’ is close to non-existent.49 Therefore the difference in “force posturing and military doctrine”,
as Walt describes it, between offensive and defensive battle plans is insignificant. I therefore do not include the indicator of offensive capabilities to interpret a threat in the framework on alliance formation behaviour of armed groups in a civil war.
The final aspect of threats as defined by Walt is the aspect of offensive intentions. How can an armed group, or rather the leadership of an armed group interpret the offensive intentions of other armed groups present in the civil war? Following findings by Bond, I argue that the ideology of an armed group functions to indicate this aspect of threat perception.50
The ideology of an armed group informs the goals it seeks to achieve and thus communicates to other armed groups which utility of cooperation or potential threat it can expect. The greater the ideological differences, the greater the perceived threat to core interests by the leadership. Leaders with the same or similar ideologies are more likely to trust each other and see their goals as compatible.
Defining the concepts
The actors that are the subject of study in this thesis are the armed groups in a civil war. The armed groups can include the government, rebel groups, and PGMs. PGMs and rebels have mostly been studied in isolation.51 This limits the kind of actors that are subject to
analysis and thus limits the explanatory power of any resulting findings. More important however, it assumes fixed factional alignment.52 As in this research I seek to further
understanding of alliance formation and changes in alliances, I reject this assumption.
I define armed groups by their political aspirations, the possession of a command and decision-making structure and their use of violence to achieve their goals. Furthermore, these armed groups originate from the country within which the civil war is fought. This definition excludes non-state actors that are economically motivated, such as trafficking organisations or criminal gangs and external intervening forces. This definition closely resembles the definition Bond uses in her research.53 However it allows for the inclusion of the government
and PGMs as actors within the system that is analysed. This ensures that the effects of the
49 Renner, 1998; Boutwell and Klare, 2000; Musah, 2002 b; Fearon and Laitin, 2003 50 Bond, 2010
51 Staniland, 2015 52 Otto, 2017 53 Bond, 2010
government is not overlooked within the framework and allows the analysis of the formation of an alliance between the government and a rebel group. As such, the government and rebels are treated as equal actors in this framework. In this framework, the government and rebels behave based on their own strategic calculations similar to sovereign states in the anarchic international system.54
I conceptualise these armed groups as rational and benefit maximising. For that reason, I understand alliance formation as the result of strategic interaction in which the groups seek to achieve security cooperation with other groups. In other words, armed groups seek alliances and cooperate in order to gain (mutual) protection and to achieve their political goals in the conflict. I treat armed groups as homogenous groups, for which the leadership functions as a proxy. This is a simplification of reality and thus limits the explanatory power of the theory. However, it ensures the feasibility of the research. Furthermore, since the interaction of an armed group with other armed groups is the subject of this research, it makes sense to focus on the relations between rather than within those armed groups.
Armed groups, in my framework, have two defining attributes: their level of power and their ideology. As argued above, these characteristics determine whether other armed groups see them as a threat, in a reaction to which they can seek alliance partners. In 1957, Dahl defined power as the ability of A to force B to do something it would otherwise not do.55
Later on, power became defined as the control over resources or events and outcomes.56
However, what is needed to achieve this control can vary, and as such, the measurement of power can be problematic.57 Another interpretation treats power as a relation between actors
and focuses on the context within which the actors seek to exercise their power.58 The
concepts ‘hard’ and ‘soft’ are an attempt to differentiate between military power and other forms of power, where ‘soft’ power works through co-optation, persuasion or attraction instead of the (threat of) use of force of ‘hard’ power.59 Economic and diplomatic actions are
often described as forms of soft power.60 Previous research on alliance formation behaviour in
intrastate conflict measured military or hard power through military capabilities and demographic resources.61 I define the power of an armed group as the ability of this group to
54 This attribute is in line with both Bond, 2010 and Christia, 2012. For further information see Posen, 1993; Fearon, 1995, 1998; and Kaufmann, 1996
55 Dahl, 1957
56 Hart, 1976
57Gubaidullina and Insebayeva, 2013
58Hart, 1976
59Nye, 1990
60Gelb, 2009
influence the outcome of the civil war and strive towards its political goals. Armed groups can be either ‘powerful armed groups’ or ‘weak armed groups’. Although the ability to threaten another armed group is not necessarily binary, this simplification allows for further theorising. The labels ‘powerful armed groups’ and ‘weak armed groups’ do not follow from absolute values of military power, but from a relative distribution. This means that military capacity is an important, perhaps the most important, measurement of power, but not the only one. Armed groups can also, for example, attempt to influence the outcome of the intrastate conflict through diplomatic or economic means. These ways of exerting power, however, are significantly harder to measure and, for that reason, I exclude these from my measurement of the concept of power. As such I determine the power of an armed group through its military potential, which I measure primarily through the membership size of the armed groups. Other factors that inform the military potential of an armed group are the quality of its arms and the discipline of its members. The quality of the arms of an armed group is determined by looking at the kinds of weapons this group possesses. Discipline is increased through training and is usually indicated through the use of a ranking system and drills.62 Such information can
substantiate the analysis of power but is not always readily available.
The ideology of an armed group refers to philosophy or doctrine that the leadership of an armed group expresses as motivating the groups’ behaviour. This package of ideas or beliefs points towards grievances of an armed group and frequently proposes a plan of action for the achievement of the goals that are linked to these grievances.63 Armed group ideologies
differ from highly structured doctrines to loosely related ideas vaguely advanced by leaders. Similarly, the coherence of the plan of action can vary. The ideology of a group functions to socialise combatants into a more or less coherent group, prioritises goals and coordinates actors.64 The first studies of motivations in intrastate conflict focussed on the ethnic
denomination of a group of people.65 In more recent literature, political ideology has become
central to the understanding of the motivations of armed groups.66 The ideology of an armed
group can be but is not limited to, conservative, reactionary, communist, liberal, separatist, or racist. Furthermore, armed groups can also adopt a hybrid ideology, combining various strands of ideology.67 Armed groups formed around an ideology can attempt to claim state
62 Vinci, 2006
63 Sanin and Wood, 2014; Hughes and Sasse, 2016 64 Sanin and Wood, 2014
65For example Fearon and Laitin, 1996; Hardin, 1997
66 Bond, 2010; Haas, 2014; Sanin and Wood, 2014 67 Bond, 2010
power in the name of the ‘ideological group’ they claim to represent.68 At the same time, this
means that the leaders of an armed group must generally act in accordance with the proclaimed ideology or risk disobedient or rebelling followers.69 Armed groups loyal to an
individual instead of an idea lack a mechanism to hold the leadership accountable and therefore fall outside the scope of this concept.70 Although the ideology of a group is usually
constant, it can change over time.71 The ideology of an armed group can be seen through
rituals and rules, through the use of symbols, and is often codified in manifests and texts.72
Systemic ideological polarity
Power is at the basis of a threat. If an actor is not powerful, it cannot credibly threaten other actors. Furthermore, powerful actors are themselves less vulnerable to threats and hostilities from other actors. An increase in power thus offers protection against annihilation and increases the likelihood of achieving political goals. Powerful armed groups are thus relatively free in the pursuit of their goals when compared to weak armed groups. But, I argue, when looking for alliances, powerful armed groups need information about the aspirations of other armed groups to decide who is a potential ally and who is not.
There is a relation between ideas and policy, especially foreign policy - which alliance formation is a part of.73 Ideas can inform and affect policymaking through discrediting
alternative interpretations of reality in the minds of leaders. The formulation of an ideology automatically creates a boundary between the ‘in-group’ and the ‘out-group’.74 Haas argues
that the line between ‘in-’ and ‘out-group’ is an essential determinant for the relations of amity and enmity, which concern the constellation of fears, threats and friendship of the actors, in international politics.75 I argue this is also the case for intrastate conflicts.
The ideology of an armed group informs the goals it seeks to achieve.76 Therefore, the
ideology of an armed group communicates certain information about the armed group to other armed groups in the conflict. Importantly ideology affects, or gives meaning to, what states and other political organisations see as a threat.77 This functions through an increased sense of
fear that the opposing ideology group will grow at the expense of one’s own group. In other
68 Kaldor, 1999; Hughes and Sasse, 2016 69 Gurr, 1970; Hafez, 2017
70 Schneckener, 2009 71 Bond, 2010
72 Arjona, Kasfir and Mamphilly, 2015 73 Goldstein and Keohane, 1993 74 Rousseau, 2006
75 Haas, 2014
76 Gertz, 1964; Bond, 2010
words, the ideology of group leaders affects the understanding of danger to their own interests, namely the preservation of their power and protection of their ideological group. The greater these ideological differences between leaders of two armed groups are, the greater the threat they perceive. However, if the ideologies of leaders closely resemble each other, they are likely to see their goals as similar and compatible. As a result, they are more likely to cooperate to preserve themselves and achieve those goals. The ideology of the leaders of an armed group thus affects the policies they pursue.78
Therefore these ideologies have a profound effect on the threat perceptions of these leaders, and by extension of their armed groups.79 Haas argues that, on top of the differences
between ideologies, the number of ideology poles within a specific system influences the threat perception of armed groups.80 This, in turn, informs alliance formation behaviour as
certain groups are seen as more threatening or as sharing a greater common interest. He expects that the specific content of these ideologies, whether they are national, clan, religious, linguistic, or any other sort of grievance, do not matter. The framework constitutes a structural theory of ideologies that examines their systemic effects. However, to assess the threat perception between leaders of armed groups, the perceived (in)compatibility of their goals does require analysis.81
In my framework, informed by Haas, I combine the micro variables of power and ideology attributes of an armed group into a systemic appreciation of the distribution of power and ideology. I combine this with the understanding that ideology and power both affect alliance behaviour. More importantly, the relative distribution of these attributes within a system informs the alliance formation behaviour of armed groups. This analysis focuses on “the number of prominent, distinct ideological groups that are present in a particular system.”82 This enables the differentiation between uni-, bi-, and multipolar systems. Each
pole of a system is made up of one or more powerful armed groups with a (shared) ideology. I assume that the incumbent government of a specific intrastate conflict a powerful armed group, facing at least one hostile ideological pole. If no opposing ideology pole existed, there would be no armed conflict. As such, in each civil war, there are at least two poles present within a conflict system. If more than two powerful armed groups with different ideologies are present, a civil war becomes multipolar. The polarity of a conflict influences the alliance
78 Goldsteina dn Keohane, 1993; Rousseau, 2006; Haas, 2014
79Haas, 2014
80 Bond, 2010; Lichbach, 1995; Gade et al., 2019 81 Haas, 2014; Hafez, 2017
formation behaviour of rebel groups according to ideology. In this framework, I thus assume that armed groups have an ideology. If an armed group lacks an ideology, this framework is not equipped to determine its threat perception and the resulting alliance preferences.
In a bipolar system, two distinct ideology-based groups are present. In a situation of conflict and ideological polarity, leaders will perceive their differences as fundamental and mutually exclusive and therefore as threatening.83 As a result, groups with the same ideology
will perceive low threat levels and significant common interests between them. Armed groups with different ideologies will perceive high threat levels and low levels of common interests. This, in turn, informs their alliance formation behaviour, thus resulting in alliances between armed groups with the same ideology. According to Haas, some actors will attempt to minimise the costs of forming alliances, yet the shared ideology will create strong binding forces that limit this tendency.84 As such, I expect armed groups to form an alliance if they
have the same or a similar ideology and a group with a different ideology threatens them. I expect this to be true for both powerful and weak armed groups.
In multipolar systems, with three or more powerful armed groups with a distinct ideology, an additional and more complicated mechanism manifests itself. To be clear, the primary incentives to form alliances with armed groups with the same or similar ideologies are still present. Thus, I expect that alliance along these lines will still form and I expect weak armed groups to behave according to this mechanism. The alternative mechanism is as follows: the existence of more than two antagonistic groups within the conflict system creates the option for armed groups with ideology A and B to band together against ideology C. This is likely to happen if armed groups with ideology A and ideology B both perceive armed groups of ideology C to be a great threat to themselves. This effect is most apparent when the respective groups perceive the most threatening ideology groups also as the most powerful. However, if the individual rebel groups perceived threats do not coincide, for example, when A perceives C to be threatening but B does not, I do not expect them to form an alliance.
The perceptions of rebel group leaders of certain ideologies need not be the same in different conflict systems. For example, in one conflict, leaders of armed groups with ideology X and Y might see each other as threats, while in another unrelated conflict, leaders with ideology X and Y do not. The perceptions between ideology groups, both with relation to differences in ideology and relative power levels must be assessed in each case. In other words, the existence of a multipolar system opens the possibility of cross ideology alliances.
83 Haas, 2014 84 Haas, 2014
But the formation of these alliances is dependent on the actual, or historical, perceptions between ideology groups.
Alliance shelter
The discussion above covers all possible categories of armed groups except one, weak armed groups with an ideology different from the powerful armed groups present in the conflict. I argue that weak armed groups with a dissimilar ideology can form alliances with powerful armed groups. They can do so because the powerful armed groups do not perceive them as a threat due to their lower level of power.
Bailes, Bradley, Thayer and Thorhallsson argue that small states are fundamentally different political, economic, and social units. They are more vulnerable and less threatening than great power states.85 In order to achieve military and diplomatic protection, these small
states can align themselves with a powerful state with a different ideology.86 It is important to
note that these small states are not completely without power and might excel in a particular mode of power, for example diplomatic grandstanding.87 Because of their vulnerability, small
states seek shelter with a larger state: they form an alliance. This shelter consists of diplomatic and military backing with regards to threats. Bailes et al. argue that ‘alliance shelter’ is a relation between two units that are not formally equal, apart from their capabilities. They see small states and their larger counterparts are fundamentally different units.88 This relation is
not one of pure hierarchy, nor a relation of formal equality and autonomy. Rather, the relation includes aspects of autonomy and hierarchy at the same time. This alliance shelter provides the small state with protection, at a potential cost. In exchange for protection, the small state can experience consequences as it might be required to yield effective control of its political decision-making. It could, for example, be forced to yield its say over political, economic or social developments in its internal sphere. In short, small states can form an alliance with a powerful state to acquire protection, yet at a potential cost.
Gade, Gabbay, Hafez and Kelly differentiate between alliances in their study of alliance networks in fragmented civil wars.89 They theorise a difference between symmetric
and asymmetric alliances. A symmetric alliance exists between groups of similar power to achieve capability aggregation. An asymmetric alliance exists between groups of dissimilar
85 Bailes et al., 2016
86 Bailes et al., 2016
87 Bailes et al., 2016
88 Bailes et al., 2016 89 Gade et al., 2019
power in an attempt of the powerful group to maximise decision-making autonomy.90
However, they find no empirical confirmation of this theorised relation in their study of the Syrian civil war.91 In this study, they look at the alliance formation process from the
perspective of the powerful armed group. Informed by Bailes et al., I theorise the alliance formation process from the side of the weak armed group. I expect that in their search for shelter from threatening armed groups, weak armed groups lacking a potential ally with a similar ideology will search for an alliance with a faction with a different ideology.
Christia argues that rebel groups can change their ideology narrative to solidify an alliance with a new armed group.92 According to her, a leader can select specific
characteristics from the ideology repertoire from his armed group that they share with their new allies and not with their enemies. With these characteristics, they can create an ideology narrative that serves as an important signal for the rank and file of the armed group that a new alliance has been formed. In exchange for protection, weak armed group might have to (gradually) change their ideology to match that of the more powerful armed group in order to preserve the alliance. Furthermore, similar to small states, these weak armed groups can be forced to yield some of its decision-making autonomy to the powerful armed group.
I do not expect all ideology changes to be possible in reality. For example, it seems highly unlikely that an armed group with expressed aims of improving the political situation of people of ethnicity A is willing to completely change their aims to improve the political situation for people of ethnicity B in exchange for ‘shelter’ because they understand their ethnicities as fixed and mutually exclusive.93 Plenty of examples of armed groups ‘adjusting’
their goals in exchange for security cooperation exist.94 This can result in a change of
professed ideology for the weak armed group. According to Kalyvas and Kocher, such weak armed groups can use a narrative to legitimate their violent behaviour.95 This does not mean
that they fully subscribe to the political goals of this narrative, yet they will often appear to do so to external observers and sometimes even in the eyes of participants.96
To sum up, I argue that the systemic ideological polarity and alliance shelter mechanisms, which I sometimes refer to as the systemic ideological polarity framework, can explain alliance formation and shifts during a civil war. The systemic ideological polarity
90 Gade et al., 2019
91Gade et al., 2019 92 Christia, 2012
93 Fearon and Laitin, 2000
94Christia, 2012
95 Kalyvas and Kocher, 2007
mechanisms explain the alliances powerful armed groups and weak armed groups with the same ideology form. The systemic ideological polarity mechanism differentiates between bipolar and multipolar conflicts. In bipolar conflicts armed groups with the same or a similar ideology will see each other’s goals as mutually compatible. As a result, they perceive little threat from each other and high potential to cooperate. This combination allows them to form an alliance. The opposite is true for armed groups with (widely) different ideologies. They perceive their goals as mutually and fundamentally incompatible and when powerful enough see each other as threatening. They will not seek to form an alliance, rather they will look for other alliance partners to protect themselves from the threat. In a multipolar conflict this mechanism still functions and a second mechanism becomes active. Multipolarity occurs when three (or more) different ideologies are present in a conflict. This forces the leader of a powerful armed group to assess the compatibility of goals with multiple other armed groups. If two leaders both perceive the goals of a third armed group as fundamentally incompatible and each other’s goals less so and if they both asses the power of the third group as threatening, they form an alliance. However, if such a shared interpretation of threats is not present, no alliance between powerful armed groups of differing ideologies are formed.
Weak armed groups that have no powerful armed group with a similar ideology with whom to ally themselves, still need to protect against threats. Because of this, they need to look for a powerful armed group with a different ideology to provide it with shelter. As the powerful armed group does not see the weak group as a threat due to their lack of power, the incompatibility of their ideology becomes less important for the formation of the alliance. However, the weak armed group might be forced to accept influence in its internal affairs to repay the powerful armed group for the protection in provides. This might result in a (gradual) change in the ideology of the weak armed group.
Alternative theories
Now that I have explained my framework, I discuss a few other suggestions from the literature with regards to alliance formation. These theories look at the effects of patronage, ethnicity, and territorial control on alliance formation. When talking about the actors in these theories, I use the term ‘armed actor’ instead of ‘armed group’ to signify that these theories do not necessarily confirm to my understanding of what motivates the behaviour of these actors.
Patronage
In an attempt to eliminate scope conditions such as ideological salience, the patronage alliance theory focuses on the condition of the state itself to determine the likelihood of alliance changes. The patronage alliance theory furthermore assumes that armed actors in a civil war seek political goals but are more fundamentally interested in their survival. The interaction between the government and armed actors determines what goals an armed actor can achieve. This theory is dependent on the rivalries between armed actors but does not explain what might cause them.97 A weak state, that is, a state with limited control over the
individuals and activities within its territory, is unable to formulate effective economic, institutional, and military policy.98 As a result, the weak state is unable to decisively win the
military competition over political concessions on its own.99 Weak states are thus more likely
to co-opt rebel groups than strong states to ensure the survival of their regime.100 Co-optation
occurs when a government invites an armed group into their patronage system by offering to fulfil their short term aims.101 This invites opportunistic behaviour from the armed actors in a
civil war, based on short term consequences. This theory thus focuses on the economic and political power of armed groups.
In its struggle to maintain its central position during a conflict, a government can attempt to manipulate the already tense relations between actors. A government may choose to lure rebel groups over to their side through co-optation. This can be done through, for example, offering military support in the form of supplies. For the rebels, this side switching would mean the achievement of limited (political) gains and the removal of the prospect of a risky and costly conflict.102 Furthermore, armed groups do not need to disarm, which ensures
their prolonged existence as a rebel group.103
PGMs also experience incentives to switch sides. Governments use PGMs for specific tasks and delegate responsibility to them. In exchange, PGMs receive political and material support. Weak states are less capable of promising future payoff to PGMs than strong states. Faced with a government that cannot continue to promise future benefits, PGMs are likely to switch sides. A weak state is less able than a strong state to control and punish such
97 Seymour, 2014
98 Fjelde and Soysa, 2009; Hendrix, 2010 99 DeRouen and Sobek, 2004
100 Otto, 2017
101 Day and Reno, 2014 102 Kreutz, 2010 103 Day and Reno, 2014
behaviour.104 As a result, both rebels and PGMs are more likely to switch sides when the civil
war is in a weak state than if it is in a strong state.105
Ethnicity
Various authors suggest that ethnicity is bound to determine alignments in civil wars. According to them, ethnic boundaries mark salient cleavages and ascriptive identities, which cause collective grievances.106 This understanding is based on the assumption that co-ethnic
individuals are concerned with the wellbeing of other members of their group. According to some, this is because of a primordial link between individuals with the same ethnicities. Violence further reinforces ethnic identities as ethnic groups gain means to limit defections and sustain collective action.107 Others suggest that information regarding loyalty and
trustworthiness is hard to collect and verify. As such, group leaders assume that people of the same or a similar ideology have similar political preferences.108 Armed groups can use ethnic
identity as a marker of underlying loyalties when facing uncertainty and apply strategies of ethnic preference to reinforce the salience of ethnic cleavages. As such, they reify the ethnic divide.109 Scholars studying alliance determination expect the formation of ethnic alliances to
occur especially in weak states.110
The alliances between groups in a civil war confirm to the ethnicity of these groups. Groups with the same or a similar ethnicity are expected to form alliances, and groups of opposing ethnicities not, due to the (dis)similarity of their political aspirations. Alliance shifts should be the result of identity shifts as groups acquire new ethnic identities or as boundaries between ethnic groups are redefined during the protracted use of violence.111
Territorial control
The territorial control alliance theory assumes, just like the patronage alliance theory that armed actors are primarily concerned with their survival. Several scholars argue that territorial control of armed actors informs alliance formation of weak armed actors.112 By controlling
territory, powerful armed groups can force collaboration despite underlying ethnic ties or
104 Clayton and Thomson, 2016 105 Otto, 2017
106 Horowitz, 1985; Walker, 1993; Kaufman, 2001; Petersen, 2002; Cederman, Gleditsch and Buhaug, 2012 107 Posen, 1993; Fearon and Laitin, 1996; Kaufmann, 1996
108 Davis and Moore, 1997
109 McLauchlin, 2010; Roessler, 2011 110 Kalyvas and Kocher, 2007; Kalyvas, 2008 111 Seymour, 2014
ideological commitments.113 The powerful armed group controlling territory can coerce other,
weaker, actors into collaboration by threatening their continued existence. The weak armed actors that are coerced thus choose collaboration over becoming the victim of atrocities committed by the new group controlling their home territory.114 While the coerced group
could also choose to fall back, ties to local property and communities bind them to their original location. This mechanism can only function if the progressing armed group seeks a local actor to impose control. The original theory focuses on rebels switching to the side of the government. However, the basic logic can also be applied in the opposite situation wherein government-affiliated armed actors join the side of rebel forces.
This theory predicts that when powerful armed groups take over a certain area, the weak armed groups located in that area will align with them. Changes in alliances should thus occur when changes in territorial control result in weak armed groups ending up on the other side of territorial delineation of powerful armed groups.115
113 Lyall, 2010 114 Kalyvas, 2006 115 Seymour, 2014
3. The Sierra Leone civil war
The Sierra Leone civil war began in 1991 and ended in 2002. Initially, the Revolutionary United Front (RUF) fought the Sierra Leone Army (SLA). However, in 1997, they formed an alliance in order to fight the Civil Defence Force (CDF), a cooperation of local defence militias that emerged during the conflict. The alliance between the RUF and SLA in this conflict was ‘unexpected’116 and the process of
emerging and unification of the CDF is a compelling case of the development of weak rebel groups. The SLA commits two coups during the conflict. These went by the names National Provisional Ruling Council (NPRC) and Armed Forces Revolutionary Council (AFRC). In the final stage of the war, from October 1999 onward, intensified international intervention and the fracture of the SLA into factions complicates the conflict dynamics significantly. In this period, the political landscape of Sierra Leone becomes defined by personal contact, which lies outside the scope of the framework developed in the previous chapter.117
Sierra Leone before the civil war
In 1967, the All People’s Congress (APC) was elected to lead Sierra Leone. A military coup prevented Siaka Stevens, the president of the APC, to take office. In 1968 a countercoup enforced the election results. Stevens quickly moved the authority to grant mining licences from the village chiefs to the Ministry of Mines, allowing him to award licenses to loyal supporters. This enabled Stevens to control the local strongmen without needing to use the unreliable army that prevented him from taking office in 1967. A failed coup in 1971 again demonstrated the risk of a powerful army. In response, Stevens limited the army to 1.500 soldiers, most of which came from the northern region of Sierra Leone and owed their position to their loyalty to the APC. Still, the relation between the SLA and the APC regime remained troubled due to mutual suspicion. For two decades the APC kept the SLA out of power by ‘co-opting senior officers, arresting officers suspected of political ambitions, executing accused coup plotters, and building a loyal paramilitary force.’118 During the period preceding the war,
the SLA had attempted various coups. This demonstrates that the SLA, although officially subservient to the civil government, was in fact, a separate organisation with its own political goals. This furthermore demonstrates the complicated relationship between the civil government and the SLA.119
In 1978, after having sidelined political opponents with use of the newly acquired mining licence authority, the APC turned Sierra Leone into a one-party state. Many observers interpreted this as a calculated move to eliminate the opposing Sierra Leone People’s Party (SLPP). The conversion 116 Chauveau and Richards, 2008 p.536
117 Fanthorpe, 2001, Utas and Jörgel, 2008
118 Bakarr Bah, 2011 p.204-205
into a one-party state allowed the privatisation of state resources by a small elite. This elite, with Siaka Stevens at its centre, created a patronage system. This meant that revenue from national resources ended up in personal bank accounts instead of state coffers. Patronage was distributed along party lines, which were mainly organised according to ethnicity.120 Beside causing grievances with the
portion of the population that was not part of this clientelist structure, it eroded state services. At the end of the Cold War, the centre of the clientelist structure shifted from local clients to foreign firms. As a result, unemployment increased sharply, and the younger population lacked any form of social mobility.121
The clientelist system also functioned on a smaller scale. The village chiefs were able to distribute the mining profits to political allies, marginalising rival politicians at the same time. The village chiefs furthermore enjoyed a dominant position in the customary law system. They were empowered to, for example, incorporate new members into their village structure, issue fines, arrange marriages, distribute land, and extract free labour for community projects. As such, many chiefs were in a position to (and many did) exploit their communities. Patronage became a requisite for education, employment and the acquisition of farmlands. As a consequence, the Sierra Leonean society contained many estranged youths and illicit diamond miners in the years before the conflict.122
The term ‘lumpen’ is frequently used to describe these social and economic outcasts. Unable or unwilling to re-attach themselves to local communities led by a chief, these outcasts fell into a political vacuum. They were thus forced to create their own communities, disconnected from mainstream political processes. Frustrated by their lack of economic and social perspective, they turned to illegal ways of providing for themselves, such as illicit mining. Because of their poverty and lack of prospects for improvement, these outcasts were susceptible to the promises of other institutions, for example, armed groups.123
By 1985, Stevens was at the end of his second presidential term and could not be elected again. Stevens picked General Joseph Momoh as his successor because, as the leader of the largely hollowed out army, he could not challenge Stevens’ clientelist economy. Stevens died in 1988. Momoh’s lack of control became apparent as diamond exports collapsed by 1989, worsening the countries already bad financial situation.124 This resulted in pressure from the International Monetary
Fund (IMF) to allow foreign firms into the mining industry.
The RUF formed somewhere during the early 1980s. Its first members were activists that sought to overthrow the APC. They received training in Libya, where they met the Liberian guerrilla leader Charles Taylor. From 1989 to 1991, the RUF supported the National Patriotic Front of Liberia
120 Ndumbe, 2001; Zack Williams, 1999
121 Richards, 2003; Reno, 2007; Wlodarczyk, 2009; Bakarr Bah, 2011 For an in-depth appreciation of the effects of the clientelist structure, see Richards, 2003 and Wlodarczyk, 2009.
122 Luke and Riley, 1989; Abraham and Sesay, 1993; Reno, 1995; Richards, 1996; Reno, 1999; Fanthorpe, 2001; Reno, 2007 and Richards, 2011
123 Richards, 1995; Abraham, 2001; Murphy, 2003 124 Reno, 2003
(NPFL) in its guerrilla war and spent time planning their incursion into Sierra Leone. During the Sierra Leone civil war, the RUF received financial and material support from the NPFL.125
The Sierra Leone civil war (1991-2002)
The civil war started in 1991 when the RUF entered Sierra Leone from Liberia. Momoh expanded his previously thinned out and underfunded army from 1.500 to 6.000 soldiers to fight the armed incursion.126 Frustrated by of a lack of supplies due to corrupt senior officers in Freetown, the capital
of Sierra Leone, Sergeant Valentine Strasser headed a coup in April 1992 which resulted in the NPRC. The NPRC was a military junta made up of SLA officers. Promising to defeat the RUF, the NPRC expanded the SLA to 14.000 soldiers, mostly by recruiting unemployed youths. Although the supplies to the SLA increased and the UN got involved in the negotiations, Strasser failed to achieve a ceasefire with the RUF and, by 1995, the RUF had advanced within 20 kilometres of Freetown.127
The NPRC, attempting to defeat the RUF, employed mercenaries, most notably Executive Outcomes (EO). In exchange for lucrative mining and security contracts, this transnational mercenary conglomerate supported offensive operations and trained both SLA units and local defence militias. These local defence militias cooperated with the SLA against the RUF. They had started as small groups of local young men that sought to protect their ethnic groups and villages from attacks by the RUF. As such, they mostly remained in their area of origin. The most famous of these were the
Kamajors, the name and initial members of which were Mende and came from a traditional hunter
society. Many Sierra Leoneans joined these groups during the war, often going through a ritual of initiation or selection process. In 1995 the local defence militias recaptured the significant Kono mining area from the RUF.128
In 1996 Captain Julius Madaa Bio, the leading intermediary between the NPRC and EO, replaced Sergeant Strasser in a palace coup. Bio, steered by public protests, organised elections won by the SLPP, which had been banned by the APC in 1978. The election of a civil government severely reduced the political influence of the army officers. Ahmed Tejan Kabbah, the SLPP president, continued peace negotiations that were initiated by Bio and Sankoh, the leader of the RUF. The RUF and the SLPP-led government signed a ceasefire in March followed by the Abidjan peace agreement in November 1996.129
At the same time, increasing pressure from international donors and the IMF forced Kabbah to implement austerity measures. He was forced to quit spending government funds on EO and the mercenaries left the country. Their departure was also enforced in the Abidjan peace agreements. Thus
125 Peters and Richards, 1998; Abdullah, 1998; Reno, 2003; Richards, 2003; Chauveau and Richards, 2008; Binningsbø and Dupuy, 2009; Obi, 2009; Wlodarczyk, 2009
126 Gershoni, 1997
127 Howe, 2001; Reno, 2003; Richards, 2003; Reno, 2004, Utas and Jörgel, 2008
128 Musah, 2002 a; Reno, 2002; Reno, 2003; Richards, 2003; Utas and Jörgel, 2008 129 Reno, 1997; Peters and Richards, 1998; Richards, 2003; Utas and Jörgel, 2008
