The changing political context in Colombia and its influence on armed group violence

(1)

Master Thesis Political Science: International Relations

Violence, Resistance and Restraint in War and Genocide

Philipp Wolf

11757663

The Changing Political Context

in Colombia and its Influence on

Armed Group Violence

Word count: 23.900

Submission date: 22

nd

of June 2018

Under the supervision of Dr. Jana Krause

(2)

Acknowledgements

I am grateful to my supervisor Dr. Jana Krause for guiding, helping and encouraging me throughout the thesis writing process. I express my gratitude to Dr. Abbey Steele who has kindly agreed to be the second reader of my thesis.

I thank my family and friends who have always been there for me, especially during the process of writing this thesis: my mom, for everything she has done for me; Nicole, Nadine and Jenny for emotional support; Athanas for technical wisdom; and last but not least Chantal and Laura for comments on earlier versions of the manuscript.

(3)

Abstract

How does the political context influence armed groups’ patterns of violence? In studies of strategic violence against civilians, one factor influencing armed groups’ patterns of violence that has not been given much attention to this day is the political context. In this thesis I argue that changes in the political context lead armed groups to strategically adjust their patterns of violence with regard to the reachability of the groups’ goals. The analysis shows that during the Colombian civil war the FARC and the ELN adjusted their patterns of violence to the political context to improve their chances of reaching their goals. Reacting to changes in the government’s position and or to resistance from civil society the armed groups strategically altered the intensity of violence and the repertoires in which they engaged. The results of my research support the literature dealing with armed group violence and bargaining and contribute to the civil war literature by studying armed group violence against civilians, showing that not only violence but also restraint is exercised strategically by armed groups that seek political concessions.

Keywords: strategic violence against civilians; strategic restraint; pattern of violence;

(4)

1. Introduction

On November 24th_{2016 the Colombian government and the Fuerzas Armadas} Revolucionarias de Colombia (FARC) signed a definitive peace agreement in Bogotá and ended a confrontation that had lasted over half a century. The path to the peace agreement was long and there were many obstacles to overcome such as criticism evoked by political opposition and the scepticism of the Colombian civil society. The official negotiations for the peace agreement started four years earlier in 2012; ten years after the last formal talks between the Colombian government and the FARC failed dismally. Indeed, Colombia has a long history of failed attempts to remove particular non-state armed actors from the battlefield through negotiations; let alone bring its civil war to a political end (Gomez-Suarez and Newman, 2013). As of the time of writing this thesis the Santos administration is still negotiating a peace agreement with the Ejército de Liberación Nacional (ELN), the last non-state armed actor that officially remains active and conducts attacks against civilians.

Striking violence, overblown demands and shattered visions of revolution have marked the path to a plausible end of the Colombian civil war. Scholars have dealt with numerous phenomena that have been present in the Colombian context. Gutiérrez-Sanín (2004) first rejected the idea that greed is the guerrillas’ main motivation. As the focus of civil war studies shifted to motivations and the functioning of armed groups, Gutiérrez-Sanín (2008) examined the structures and mechanisms behind the FARC and the Autodefensas Unidas de Colombia (AUC). Furthermore, the processes of disarmament, demobilization and reintegration accompanying the AUC’s disbanding (Porch and Rasmussen, 2008) were examined.

In the meanwhile, the broader civil war literature dealt with the reasons behind violence against civilians; a process initiated by Kalyvas’ (2006) groundbreaking work on violence and territorial control. Elisabeth J. Wood (2009) identified group structures and hierarchies as important features that can determine whether or not armed organisations restrain from certain repertoires of violence against civilians, and if these organisations are able to contain opportunistic violence by individual group members. Reed M. Wood (2010) added further reasons for strategic violence against civilians arguing that rebel violence against civilians depends on the rebels’ capabilities. Steele (2011) and later Balcells and Steele (2016) showed – with evidence from Colombia – that certain repertoires of violence fulfil strategic purposes and produce forced displacement.

(7)

Gutiérrez-Sanín and Elisabeth J. Wood (2014) thoroughly discussed the role ideology plays in structuring armed groups, and how it furnishes the base to socialize the individual members, brings them into line with the leaderships ambitions, and teaches them norms and values with regard to violence against civilians. They concluded that the explanation why certain armed groups restrain from engaging in specific repertoires of violence lies in the groups’ ideology. Already two years earlier, Thaler (2012) pointed to ideology as factor that influences restraint from violence against civilians. Hoover Green (2016) further contributed to the discussion of elements promoting group cohesion, stressing that political education allows commanders to bring rank and file into line with the organisation’s goals.

Hultquist (2013) pushed forward the bargaining literature in the field of civil wars, arguing that armed groups have to show a certain amount of force to be considered as being worthy negotiation partners by opponent governments. Stanton (2016) discussed armed groups’ engagement in violence or restraint from a strategic point of view. She argued that non-state armed groups engage in terrorism against democratic government opponents to coerce the government into making political concessions, such as for instance entering negotiations. However, Stanton did not account for variations in such terrorist practices over time, let alone armed groups’ strategic restraint from certain terrorist repertoires. In the meanwhile, scholars studying the Colombian civil war shifted the focus more and more towards the study of civilian agency and non-violent resistance (Kaplan, 2017) or linked ideology to changing terrorist practices, and discussed changes in military realities (Feldmann, 2017). The influence of the political realities on the guerrilla groups’ violent tactics, however, was not discussed.

Finally, Gutiérrez-Sanín and Elisabeth J.Wood (2017) made a significant contribution to civil war research by introducing a concept of pattern of violence. They pointed to misleading judgments of researchers due to measurement inaccuracies, and partially applied their new concept in the context of the Colombian civil war by examining the FARC’s and the AUC’s use of violence against civilians. The authors did so by almost exclusively dealing with changes in patterns of violence that had happened once agreements with the Colombian government were reached. Questions regarding the influence of possible negotiations or concessions on patterns of violence remain to be answered. The patterns of violence of the FARC and the ELN both varied considerably depending on whether informal talks were held or whether negotiations were within reach. For instance, the FARC declared a unilateral ceasefire several times during the talks with the Santos administration. Sometimes an armed group cut back its use of violence and exercised restraint in such situations, whereas

(8)

sometimes the group escalated its use of force. How can these variations in armed groups patterns of violence be explained when talks or agreements with the government are within reach? This leads to the question I aim to answer in this thesis that is:

How does the political context influence armed groups’ patterns of violence?

In the course of the Colombian civil war all parties involved, government and guerrilla groups, underwent many changes of positions and objectives (Gomez-Suarez and Newman, 2013). So far, little is known about how these shifts influenced the FARC’s and the ELN’s patterns of violence. I argue that changes in political context led armed organisations to strategically adjust their use of armed violence in a way that improved the organisations’ chances of getting political concessions. Studying such variations may provide insights into how armed groups react to changing prospects of getting political concessions. Do armed groups exercise restraint to improve their position towards their government opponent, or towards the public opinion that might pressure the government into changing its positions? Does civil society directly impact armed groups’ decisions regarding the targeting of civilians, or is it the government – the negotiating partner – that influences armed groups’ strategic calculations?

By analysing how the political context influences armed groups’ strategic decisions regarding violence against civilians, I aim at contributing to the current state of civil war research in several ways. First, this research will contribute to Stanton’s (2016) argument linking non-state armed groups and their use of violence when seeking political concessions. While Stanton mainly points at engagement in violence I will also discuss the implications of restraining from violence in connection with the seeking of political concessions. Second, by applying Gutiérrez-Sanín and Wood’s (2017) nuanced concept of patterns of violence when examining the relation between political context and guerrilla violence, I will draw a more detailed picture of the violence used to achieve political concessions. Third, this research will provide more information about the way the FARC and the ELN used strategic violence to reach their goals. Especially in the case of the ELN such information is rather scarce.

In order to test my argument, I will analyse qualitative data, i.e. reports, secondary literature, and to a great extent newspaper articles, providing information about shifts in the political context in Colombia between 1998 and 2016. I will link these shifts with the use of violence by analysing descriptive data from the Global Terrorism Database as well as further data from reports and secondary literature.

(9)

This thesis is organised in eight sections. First, previous research related to the topic of my research will be discussed in the literature review. Furthermore, I will point out how my own research contributes to the civil war literature. Second, I will lay out the theoretical approaches and considerations that build the foundation of my argument in the theoretical framework. This presentation will be followed by my own argument and its conceptualisation. Third, I will point out the research design which my thesis will follow. I will discuss the methodology of my research, justify my case selection, present the data on which I will rely, and finally I will operationalize the concepts laying at the core of my argument. Fourth, a brief overview of the Colombian civil war, the FARC and the ELN will be given. Then, I turn to the analysis where I will link changes in the political context to the two armed groups’ use of violence against civilians. Following this analysis, I will present my findings and discuss alternative explanations that could account for armed groups’ adjustments of their patterns of violence as well as limitations inherent in my analysis. Finally, this thesis will be ended by a conclusion.

(10)

2. Literature Review

This chapter discusses contributions and limitations of literature dealing with strategic violence against civilians, variations in violence against civilians and subjects specific to the Colombian civil war. First, I will explore how strategic violence has been approached in the civil war literature to get a sense of how different strategic purposes lead armed groups to the engagement in specific repertoires of violence. Second, I will point to contributions regarding the measurement of variation of violence and to explanations given for variation of violence. Third, literature dealing with the Colombian government and the government’s positions towards the non-state armed groups will be briefly discussed. Finally, I will specify my argument and lay out how it contributes to the current state of the literature.

2.1 Rational Approach Regarding Violence Against Civilians

The civil war literature so far has discussed various strategic reasons for armed groups to engage in violence against civilians. Kalyvas (2006) discussed the strategic use of violence linking the form and scale of violence against civilians with territorial control. However, his approach is tailored to lethal violence and does not account for other repertoires of violence against civilians. When violence is used strategically, armed groups engage in certain repertoires for specific reasons, and group leaders mostly seek to contain violence originating from individual motivations. Elisabeth J. Wood (2009) argued that for an armed group to minimize individually motivated violence the structures of the armed organisation are essential. Once a commander decides to restrain from a certain repertoire of violence, whether or not this restraint is maintained depends to a large extent on the structures and hierarchies present in the armed group. Furthermore, Wood made the point that especially leftist insurgent groups invest resources in educating their members politically to insure that the strategic decisions are carried out as intended. She thus added a normative element to considerations regarding the use of certain repertoires of violence. However, Wood’s analysis is restricted to sexual violence, and does not discuss how armed group structures influence the restraint from other repertoires. Thaler (2012) then showed that the strategic engagement in a specific pattern of violence does not only imply the use of certain repertoires of violence; it also means restraining from certain repertoires. The author, who looked at two Leftist Marxist liberation movements in Angola and Mozambique, found that Leftist groups rather engage in non-lethal attacks. Thaler argued that especially Marxist organisations heavily seek for civilian support, and the best way to gain civilian support, is by not using violence against

(11)

them. Restraint from violence against civilians is therefore not only explained strategically, but also normatively by linking it to an armed group’s ideology. The normative aspects of restraint in violence against civilians, then, were refined and strengthened by Gutiérrez-Sanín and Elisabeth J. Wood (2014). Nevertheless, their argument is restricted to forms of mass violence.

In the meanwhile, Hultquist (2013) linked strategic violence to relative power of armed groups and their government opponent. By doing a quantitative dyad-year analysis he showed that power parity between rebel group and government most likely enables the achievement of a negotiated settlement between the two parties. The author further stated, that weaker armed groups might resort to violence to pressure the government into negotiations with them. Hultquist, however, does not discuss the forms of violence armed groups engage to get negotiations

Finally, Stanton (2016) made a significant contribution to the civil war literature on strategic violence against civilians and strategic restraint. She got to the heart of it by writing that“[i]n many cases, leaders of an armed group instruct their members to deliberately attack civilian targets to achieve particular strategic objectives in the conflict.” (Stanton, 2016: 28) While linking strategic purposes to either “low-casualty” or “high-casualty”, she also described a repertoire that non-state armed groups are prone to use when fighting democratic governments in order to get political concessions: terrorism. However, Stanton fails to explain when, how and why armed groups might change their use of terrorism against a government opponent over time and what specific forms of terrorism they use. Furthermore, she leaves aside strategic restraint from violence to get political concessions in the context of an armed group facing a democratic government opponent.

2.2 Variation of Strategic Violence Against Civilians in the Civil War Literature and the Colombian Context

Kalyvas (2006) argued that armed organisations engage in violence against civilians that support the enemy. Scope and scale of that violence depend on the information the armed group has about enemy supporters. The more information an organisation possesses the more selectively it can target civilians. The accessibility of information in a certain area can vary across time and space, consequently leading to a variation in strategic violence against civilians. Reed M Wood (2010) made the point that non-state armed group violence against civilians varies with the armed groups’ relative capabilities such as varying civilian support and varying access to lootable resources. Another factor influencing the scope and scale of

(12)

violence against civilians is the one of changing military realities on the ground. González Muñoz (2015) as well as Feldmann (2017) discussed the impact of changing military realities on civilian targeting in the context of the Colombian civil war. González Muñoz (2015), by examining the reasons that led to the peace talks between the government and the FARC in 2012, found that the Colombian military regained strength forced the guerrilla groups to change tactics. As the state forces succeeded in pushing the armed groups out of the urban centres, these groups were not able to target civilians at the same scale and scope as before. The scholar concluded that the impossibility to end the conflict militarily led the government to negotiate with the FARC. González Muñoz’ discussion of varying violence against civilians remains, however, superficial and does not give detailed information about the change of balance of power he refers to. Feldmann (2017), analysing the military realities more precisely, identified variations in the frequency of terrorist practices by the FARC and the ELN. He demonstrated that the intensity of violence changed with altering war dynamics. Limitations of his contribution lay in the way he conceptualised guerrilla violence against civilians. The scholar defined acts of terrorism in such a way that basically every act of deliberately targeting civilians constituted an act of terrorism. Feldmann therefore looks at violence against civilians as a whole, instead of examining specific repertoires of violence, and leaves the motives behind specific repertoires aside for the most part.

Gutiérrez-Sanin and Wood (2017) pointed to the limitations of civil war research such as the works of Kalyvas (2006) or Stanton (2016) that solely refered to violence against civilians in broad and imprecise terms. Arguing that especially in the Colombian case neglecting the differences between repertoires of violence can result in misleading conclusions when measuring and comparing armed organisations’ patterns of violence, Gutiérrez-Sanín and Wood (2017) introduced a multidimensional concept of pattern of violence that constitutes a significant contribution to the field of civil war studies. The scholars define that

“the pattern of violence (A, X, T, L) is the matrix summarizing for organization A, subunit X, at time period T, and at location L the forms of violence, and for each combination of form, target and technique, the frequency in which X regularly engages.” (25)

Every pattern of violence therefore consists of four elements: repertoire, targeting, technique and frequency. Gutiérrez-Sanín and Wood’s concept thus allows to assess an armed organisation’s use of violence in a highly nuanced and differentiated way and hence to answer

(13)

a variety of questions: Has the frequency with which a certain target population was attacked changed over time? Has the repertoire they faced changed? Has the repertoire of violence stayed the same but the target population changed? Furthermore, the concept of pattern of violence allows for comparisons between different armed groups active in the same conflict. As an example of the power of their concept Gutiérrez-Sanin and Wood introduced a dataset about small and large massacres against civilians that occurred in Colombia, and compared the patterns of violence of the FARC and the AUC. Gutiérrez-Sanín and Wood finally argued that the patterns were influenced by the political context. However, the authors only do so in a limited manner, arguing that paramilitary groups changed their pattern once an agreement with the government was reached. Furthermore, Gutiérrez-Sanín and Wood briefly mentioned that the FARC was accused of substituting kidnapping with extortion while talking to the government (Gutiérrez-Sanín and Wood, 2017: 34). A more in depth analysis of how the political context can impact patterns of violence, especially before an agreement with the opponent government is reached, is still lacking. Additionally, the purpose of violence, whether it is employed strategically or opportunistically, is an aspect that is not treated by Gutiérrez-Sanín and Wood’s concept.

2.3 Variation in Political Context and Strategic Choices in the Case of Colombia

The Colombian civil war has been going on for half a century. Positions and objectives of governments and guerrilla groups have changed, and so have the conclusions that scholars drew from the civil war over the years. Research has focused on the motives and capacities of non-state armed groups and on the attitude with which different government administrations have tried to deal with them. On the side of the non-state armed groups the focus lied mainly on the study of the two biggest and most violent ones, the FARC and/or the AUC (Gutiérrez-Sanín 2008; Porch and Rasmussen 2008). One of the smaller organisations that did not get as much attention from researchers is, for instance, the ELN (Feldmann 2017; Gruber and Pospisil 2015). Numerous scholars have studied the period around the turn of the century when the Pastrana administration tried to negotiate with the guerrilla organisations. And a few researchers also made timid attempts to link the organisations’ pattern of violence with the political context (Rochlin, 2003). Several political scientists have studied the period of time between the failure of the FARC-Pastrana talks and the start of the FARC-Santos talks (González Muñoz, 2015).

Understandably, so far very few scholars have analysed the recent peace talks between the government and the FARC and compared them with failed ones from the past. Scholars

(14)

that did so vastly agree on lessons that both the FARC and the Colombian government have learnt over time. Nasi and Rettberg (Forthcoming) pointed at various lessons. First, under Pastrana the negotiations were entered with an “open agenda” (Nasi and Rettberg, Forthcoming). According to the two scholars this led the parties to approach the negotiating table with contrasting expectations. Then, while Uribe held office, the government wanted negotiations on “minimalist terms” (Nasi and Rettberg, Forthcoming), a demand that the FARC was not ready to meet. For the peace talks under Santos the parties agreed on an agenda containing concrete points of negotiations. Furthermore, the two parties agreed on key principles surrounding the talks such as “nothing is agreed until everything is agreed”, “this will be a negotiation by Colombians and for Colombians”, and “negotiations will take place amidst conflict” (Nasi and Rettberg, Forthcoming). In the meanwhile Gomez-Suarez and Newman (2013) also stressed the importance of the limited agenda during the recent talks, but concluded that the most important lessons learnt do not stem from the Pastrana era. The authors gave priority to how the negotiating point “political participation” that safeguards political guarantees and avoids the proliferation of violence was influenced by the failed attempt of transforming the FARC into a political actor in the 1980s. While Gomez-Suarez and Newman discuss the prospects of the peace talks in detail they leave aside the on-going conflict between the government, the FARC, and the ELN and do not discuss the influence that violence against civilians or the exercise of restraint might have on the negotiations.

González Muñoz (2015) identified several challenges for the Santos-FARC talks. Despite all the advantages of clear denominations of who negotiates with whom, the author saw the drawbacks of it. The solely indirect participation of civil society results, according to González Muñoz, in an unequal participation of the different actors. Furthermore, the information ban surrounding the peace talks also faces downsides. Not only does it hinder spaces of direct participation but it also critically impacts the formation of the public opinion. Finally, González Muñoz saw a major factor of instability in the principle „nothing is agreed until everything is agree”. This way of functioning would lead to the whole process’ legitimacy being questioned with every unilateral action of either side. Consequently, every action can cause tensions to rise and the public opinion to question the entire peace process (González Muñoz, 2015: 257). The author accurately points at the flaws of the Santos-FARC negotiations. However, if and how unilateral actions did influence the peace talks has not been analysed so far. Scholars have not yet discussed thoroughly the opportunities and challenges of the negotiations between the Santos administration and the ELN.

(15)

2.4 The Contribution of this Thesis

So far civil war scholars have linked strategic violence to the accessibility of information (Kalyvas, 2006), the goals that armed organisations pursue (Stanton, 2016), or to seeking a place at the negotiating table (Hultquist, 2013). When studying the repertoires of violence in which armed groups engage research has lacked precision until recently: often only lethal violence has been examined or only selective or indiscriminate forms of violence have been distinguished (Kalyvas, 2006). Variations in the use of violence against civilians have mostly been described in broad terms to this day and were linked to changes in an armed group’s access to information (Kalyvas, 2006), the group’s capabilities (Wood, 2010), its ideology (Thaler, 2012) or the realities on the battlefield (Feldmann, 2017).

In the context of the Colombian civil war most of the research has focused on the FARC’s and the AUC’s use of violence and has explained differences with ideology (Gutiérrez-Sanín and Wood, 2017) or military facts (Feldmann, 2017). Variations in the FARC’s and the AUC’s use of violence have been linked to peace negotiations and the pressure originating from Colombian civil society and the international community. These variations were mainly argued to consist of the substitution of one repertoire with another (Gutiérrez-Sanín and Wood, 2017). How violence varies in the time leading up to negotiations or agreements has not been studied in detail so far. Although Feldmann (2017) looked at variations of violence over time in the Colombian civil war, his findings have to be treated with caution as he left political realities aside and used flawed conceptualisations of repertoires.

This thesis, drawing on Gutiérrez-Sanin and Wood’s concept of pattern of violence and Stanton’s argument linking violence with concession-seeking, will examine the cases of the FARC and the ELN and argues that the political context leads to strategic adjustments in the armed groups’ patterns of violence. In doing so this thesis will contribute to the current state of the civil war literature in several ways. First, it will provide refined information regarding Stanton’s argument on how armed groups target civilians or exercise restraint when attempting to achieve political concessions. Second, the analysis will further provide detailed evidence on which repertoires were prominently used by the FARC and the ELN in what contexts. Third, this research will examine whether government positions or rather pressure originating from the pubic opinion lead armed groups to exercise restraint regarding certain repertoires. Finally, this thesis will generate more evidence on the ELN, an actor in the Colombian civil war that continues to be understudied.

(16)

3. Theoretical Framework

In the following chapter I will discuss the theoretical approaches and considerations that lay the foundations for my argument. First, I am going to clarify what distinguishes strategic from opportunistic violence. Also, relying on Stanton’s (2016) considerations, I will argue that armed groups base their decisions regarding violence against civilians on strategic calculations. In a next step, I will point out how an armed group can ensure that its members follow the leaders’ orders and only engage in violence that serves the group’s goals. Then, I am going to discuss different forms and repertoires of violence distinguishing crime from political violence. Finally, I will bring the essence of the three paragraphs together and formulate my argument that is followed by a thorough conceptualisation.

3.1 Strategic Violence

Parties of a conflict can engage in violence against civilians for a variety of reasons which might be strategic (Kalyvas, 2006; Stanton, 2016; Steele, 2011; Wood, 2010) or opportunistic (Manekin, 2013). Manekin (2013), while studying opportunistic violence by individual members of the Israeli army, distinguishes strategic and opportunistic violence as follows. Strategic violence serves to generate compliance, to gain information or to signal willingness to continue fighting. In a broader sense, it serves a certain strategic purpose. Opportunistic violence, on the other hand, refers to violence that originates from individual motivations instead of leadership decisions. Strategic violence against civilians during civil war, unlike opportunistic violence, is a form of violence that originates from rational decisions (Kalyvas, 2006). Kalyvas (2006) gives a simple feature that distinguishes opportunistic from strategic violence: the scale of the violence observed. The scholar states “individual motivations alone are unlikely to result in large-scale violence over a long period of time.” (Kalyvas, 2006: 26) Violence that occurs with a certain frequency can thus be said to originate from decisions of an armed group’s leadership.

Non-state armed groups decide upon the use of violence, its frequency and its form, or upon exercising restraint, based on strategic calculations about which behaviour will help them to reach their political goals (Stanton, 2016). The armed groups “thus also carefully weigh the costs of engaging in violence against civilians and the benefits of restraint.” (Stanton, 2016: 31) Armed groups aim at achieving or fostering political power. Reaching that goal requires a certain degree of civilian support. Consequently, armed groups weigh costs and benefits of the use of violence to not endanger their standing among civilians. As the public opinion changes over time, so may the calculations regarding targeting civilians.

(17)

Stanton brings into play several repertoires of violence that aim at achieving political objectives and that are decided upon through above-mentioned strategic calculations. Three different purposes of violence can be distinguished: control of civilians and territory, cleanse territory of a specific group of people, terrorize the civilian constituents of the opponent “in an effort to coerce the opponent into making concessions.” (Stanton, 2016: 10) Stanton subsequently differentiates between the “audience of violence” and the “target of violence” (Stanton, 2016: 43). Engagement in strategic violence of an armed group is often aimed at altering the behaviour of a specific group of people. This group constitutes the audience of violence. The target of violence, on the other hand, refers to who is physically targeted. Targets of violence again can be determined selectively or indiscriminately, consequently resulting in a low or high number of civilian casualties.

An armed organisation engaging in violence for the purpose of terror can do so in various forms and for several reasons. When an armed group is fighting a government opponent the targets of violence are the government’s constituents, i.e. the population entitled to vote. In that case the target audience of terrorism is usually the government (Stanton, 2016). Terrorism then can be divided into acts that cause a high or low number of casualties. Terrorist acts causing many deaths are, for instance, bombings of public places. Acts causing few to none deaths are, for example, bombings of civilian infrastructure such as power stations or bridges (Stanton, 2016). Especially in democracies, where the government can be held accountable by the targets of violence, armed groups use terrorism to coerce the government into making concessions. An armed group targets civilians to alter their position towards the government because this targeting may lead the population to hold their government accountable for the experienced suffering. Armed groups are thus more likely to make use of terrorism when fighting opponents that can be held accountable by civilian constituents (Stanton, 2016). However, it remains unclear what might prompt armed groups to step up or halt their engagement in violence against civilians, when the group aims at coercing the opponent government into making concession.

3.2 Strategic Versus Normative Considerations Regarding Violence Against Civilians

In order for an armed group to use violence against civilians in a way that helps the group achieving its political goals, the leadership of the armed group has to ensure that its rank and file follow suit (Stanton, 2016). Individual members’ compliance with the leadership’s decisions and directives is therefore essential for the functioning of an armed

(18)

group. A feature that fuels the creation of individual compliance and commitment to the cause and reliable structures is ideology. Gutiérrez-Sanín and Wood (2014) define ideology as

“a more or less systematic set of ideas that includes the identification of a referent group [. . .], an enunciation of the grievances or challenges that the group confronts, the identification of objectives on behalf of that group [. . .], and a (perhaps vaguely defined) program of action.” (215)

The ideology of a group therefore has an influence on the group’s hierarchy, its structure and the rules and norms to which the respective group adheres. Within the group “political education” helps to ensure that fighters of a group incorporate the ideology of the organisation as well as the norms and rules that are directly linked to the ideology (Hoover Green, 2016: 623-625). Political education, therefore, directly “connects conflict purposes to specific behavioural norms.” (Hoover Green, 2016: 624) An armed group that is fuelled by ideology thus engages in a pattern of violence that is guided by normative beliefs. Leaders can hardly opt for the engagement in a repertoire that contradicts the group’s core values. As long as the heads of an armed group make strategic decisions, for instance in favour of terrorist repertoires, that are in line with the group’s norms and beliefs, individual group members can be expected to follow suit. Consequently, variations in an armed group’s pattern of violence reflect strategic decisions of the group’s leadership.

3.3 Crime or Terrorism

Terrorism, as Stanton herself acknowledges, is a broad and contested term. The definition she gives is that “terrorism aims to convey a message to an audience other than those targeted with violence” (Stanton, 2016: 47). Such a definition leaves room for many interpretations and even more repertoires of violence. To briefly re-mention Gutiérrez-Sanín and Wood’s (2017) approach on measuring violence: the repertoire is one of four elements constituting an armed group’s pattern of violence. The remaining elements are targeting, technique and frequency. While the aspect of targeting is essential for Stanton, she does not go into detail regarding repertoires that are used in a terrorist manner.

Whether or not strategic violence against civilians is considered to be aiming at coercing a government opponent into making political concessions, respectively whether or not violence consists terrorism, often depends on how the violence as such is perceived. The circumstances under which violence is studied can change the perception of the exact same violence. Violence against civilians is mostly tackled from two angles: political violence or

(19)

criminology. Especially civil war scholars tend to take rather the former than the latter approach (Thyne and Schroeder, 2012). Violence that is considered to be political inflicts damage upon human beings or infrastructure with the intention of reaching a wider audience and thus is often referred to as terrorism. Henceforth the terms political violence and terrorism will be used interchangeably. Violence against civilians approached from a criminology point of view, does not differ much from terrorism or civil war scholars’ conceptions. The difference between violence as a crime and as a political statement does not lie in the act itself. Rather, it is defined by the motivations leading to the act, the perceived opportunities, the control mechanisms and the constraints connected to the act of violence (Thyne and Schroeder, 2012).

Therefore, various repertoires of violence can be seen as both acts of political violence or as crimes. Kidnapping, for instance, is such a repertoire. One speaks of kidnapping when “persons are forcibly seized and transported to a destination where they are held against their will in unlawful confinement” (Turner, 1998: 146). Kidnappings can be distinguished according to the motivations of the kidnappers. Four possible combinations of motivations are possible: money but no politics, money and politics, politics but no money, no money and no politics (Turner, 1998: 146-152). When one parent abducts his own child the act of kidnapping constitutes a crime. However, if a public figure is kidnapped and her release is tied to political concessions or serves to deliver a message to a wider public, then kidnapping can be an act of political violence or terrorism (Turner, 1998). In general, when kidnaping is conducted by an organisation that follows a political agenda every act of kidnapping can be subject to politicization; even a kidnapping that was conducted out of financial motives. Distinguishing between economic and political motives behind kidnappings and the difficulties related to doing so are reflected in the academic debate. Scholars looking at kidnapping as an act originating from social or political motives consider it an act of terrorism (Brandt and Sandler, 2009). Researchers focusing on economic reasons behind kidnapping regard it as a crime, as a violent act without political connotations (Pires, Guerette and Stubbert, 2014). The case of how differently kidnapping can be perceived is typical for how the perception of repertoires depends on the motives behind the act itself, and on the researcher’s approach to the subject (Thyne and Schroeder, 2012). Sexual violence constitutes another repertoire of violence that can be studied from criminology as well as from a civil war angle (Wood, 2009). This applies to assassinations as well; they can be committed by criminals or politically motivated organisations.

(20)

Other repertoires of violence are more unambiguously considered to be political and therefore studied as regular features of civil war rather than as isolated crimes. Feldmann (2017), for example, does not only consider kidnapping as a terrorist act, but also bombings, massacres and summary executions. Bombings of civilian or government infrastructure is seen as an act of political violence, and often occurs in civil wars, for instance, when an armed group is fighting an opponent government (Stanton, 2016). Massacres, the killing of “defenceless persons, such as civilians and soldiers who are not participating in military activities or in the conflict” (Gutiérrez-Sanín et al., 2015) are a prominent feature of civil conflict, too. Who is and who is not targeted depends, for instance, on regional, ethnic or social and political characteristics. Armed groups committing massacres do so deliberately and with “specific strategic objectives.” (Gutiérrez-Sanín and Wood, 2017: 31) Such strategic objectives are, for example, the punishing of suspected enemy collaborators, deterring other civilians (Kalyvas, 2006) or producing forced displacement (Steele, 2011).

3.4 The Argument of this Thesis

Stanton’s (2016) approach brings a political dimension into play by relating violence to political purposes. She does so, however, only in a rather basic way. Gutiérrez-Sanín and Wood (2017) add precision to how variations in the use of violence can be measured, but neglect the purposes behind specific repertoires. While they account for the strategic substitution of one repertoire by another, the strategic restraint from certain repertoires is not considered. Gutiérrez-Sanín and Wood’s concept of pattern of violence allows examining which repertoires exactly are related to the pursuit of political concessions while elements of Stanton’s approach enable to link the variation in patterns of violence to the political reality. The scholar links terrorism with the attempt to reach political concessions. She does so, however, imprecisely by solely differentiating forms of terrorism in terms of the casualty toll. Furthermore Stanton makes the point that democratic governments react to terrorism because they fear being held accountable by their constituents. However, Stanton does not explain when and how violence that aims at the opponent government as audience of violence varies and she does not take into consideration how restraint from the armed group’s side might change the government-constituents relation.

In democracies, governments are elected on the basis of the agenda they are advocating. When governments do not fulfil their agendas and do not satisfy the voters’ wishes, they are likely to be held accountable by the voting population in the next election. The agenda a government pursues varies with different presidents in office. One

(21)

administration may pursue a favourable position towards negotiating with the armed groups it is fighting while another administration might opt in favour of taking a hard stance against armed groups. Violence that is aimed at civilians can lead them to pressure the government into taking a more offensive position against the armed groups or to enter into negotiations with the armed groups (Stanton, 2016).

For an armed group to make sure that the violence its members engage in serve the group’s goals the organisation needs strong group cohesion, stable organisation structures, and clear hierarchies (Gutiérrez-Sanín and Wood, 2017; Hoover Green, 2016). These aspects ensure that decisions by the armed group’s leadership are carried out as intended. As I link armed organisations’ decisions with their use of violence, I intend to solely focus on violence that is “intentional on the part of the group” (Stanton, 2016: 28) and therefore ordered by the group’s leadership. Variations in an armed group’s pattern of violence therefore reflect changes in strategy originating from the group’s leadership.

As the political context changes, so do the strategic calculations of an armed group regarding violence against civilians. In contrast to Stanton (2016), I argue that when calculating the costs of violence armed groups seeking political concession do not only have the choice between engaging in low or high casualty terrorism, but they also can opt to exercise restraint in violence against civilians. Which strategy an armed group opts for – violence or restraint – is decided by taking into consideration the government’s position and the public opinion. These aspects form the political context in which armed groups decide which strategy is the most appropriate in pursuing their goals. As the political context changes, so does the calculation regarding the most effective pattern of violence, i.e. the engagement in violence or the exercise of restraint. The political context thus leads armed groups to strategically adjust their patterns of violence against civilians. Such adjustments consist in shifts from one repertoire to another, in a change of frequency with which one or more repertoires are being used, or in an absolute or partial restraint from violence.

3.5 Conceptualisation and Observable Implications

Political Context

Armed groups use terrorism to coerce the opponent government into making concessions (Stanton, 2016). Concession, here, has numerous meanings. It can consist of an act that precedes negotiations: the recognition of the armed group as a political player, the government’s agreement or willingness to talk to a specific armed group; the government’s agreement to formalize talks with the concerned armed group; the government’s fulfilment of preconditions to talks, for instance freeing prisoners of that group or exchanging prisoners,

(22)

cease military actions against that group or cease a specific territory to that group. Furthermore, concessions can concern the process of negotiation itself. Finally concessions can consist of the results of negotiations. Members of the concerned armed group can, for example, be granted immunity from criminal prosecution and/or the armed group can be granted political influence, for instance by receiving parliamentary mandates.

The prospects of an armed group in achieving political concessions are influenced by the political context. For the purpose of this research the political context is understood to be given and influenced by actions and interactions of the government, the armed non-state groups and the public opinion. The government can completely reject any form of concessions to a specific armed group. It can refuse talking to or negotiating with armed organisations. Furthermore a state can be unconditionally open to talks, or it can link possible talks to the fulfilment of preconditions. A change in government can influence positions and demands, as mentioned above. How an armed group stands towards talks with the opponent government depends on similar characteristics. Whether or not it agrees to negotiate with the government depends on the group’s objectives. The armed organisation can pose demands that have to be met before the armed organisation is willing to talk and/or it can reject preconditions demanded by the government. The public opinion, i.e. the population entitled to vote, can pressure the government and/or the respective armed group into taking or giving up certain positions.

Violence and Restraint

This thesis acknowledges the difficulties of unambiguously separating criminally motivated violence from political violence and takes the following approach concerning political violence, i.e. terrorism. For the sake of clarity this thesis treats the terms terrorism and political violence as synonyms and relies on the criteria given by the Global Terrorism Database (Global Terrorism Database [GTD], 2017) to distinguish between crimes and acts of political violence. Acts of political violence

“must be aimed at attaining a political, economic, religious, or social goal. [. . .] There must be evidence of an intention to coerce, intimidate, or convey some other message to a larger audience (or audiences) than the immediate victims. […] The action must be outside the context of legitimate warfare activities.” (GTD, 2017: 10)

(23)

As Stanton (2016) states that terrorism is violence that aims at an opponent government as the audience of violence, in this thesis all acts that fulfil the criteria of the GTD are considered to have the government as audience of violence.

Furthermore, this thesis codes violence in civil wars as follows: all forms of violence listed by the GTD are considered to be possible means to coerce the government into making political concessions and therefore are considered as terrorism. Kidnapping, one of the repertoires listed by the GTD, is seen as targeting the government if it fulfils the GTD criteria. If a kidnapping incident does not correspond with the requirements it is considered to be a crime and not an act of terrorism. Repertoires that do not fully correspond with GTD criteria and that can mainly be observed when non-state armed groups fight each other, such as massacres, for instance (Gutiérrez-Sanín and Wood, 2017), are not considered to have the government as the audience of violence, and therefore are not considered as terrorism.

Armed groups’ restraint from political violence is conceptualised as follows: Heavily inspired by Stanton’s framing of the term “restraint”, this thesis regards restraint from violence as behaviour indicating “a deliberate attempt to limit violence.” (Stanton, 2016: 30).

Observable Implications

The argument of this thesis is that the political context leads armed groups to strategically adjust their patterns of violence against civilians. This adjustment can, for instance, consist of a shift from one repertoire to another or in the change of frequency with which one or more repertoires are being used. As terrorism is linked to the pursuit of concessions, if this argument is valid, it is expected that changes in political context lead to changes in an armed group’s pattern of violence, more precisely in their engagement in terrorist repertoires. Variation of the political context is expected to alter an armed group’s calculation on the adequate amount of terrorism that enables the achievement of the group’s objectives. If the variation of terrorist repertoires corresponds with variations in criminal repertoires or classical civil war repertoires, such as for instance economic kidnapping, or massacring, then the argument of this thesis is weakened considerably. If variations in terrorist repertoires do not correspond with changes in the political context, then the argument of this thesis is invalid.

(24)

4. Research Design

This chapter will present the approach taken to test the argument of this thesis. First, I will outline the methodology that my research will follow. Second, I will introduce and justify the cases I chose to study. Third, I am going to briefly discuss existing data on the Colombian civil war before I will turn to present the data that will be used in this thesis in detail. Finally, I will operationalize the two concepts fundamental to my argument, the political context, and variation in violence against civilians.

4.1 Methodology

In order to test the argument of this thesis, a comparative analysis using a case-oriented approach will be conducted. Such a method allows for the study of a small number of cases in considerable depth and allows taking the broader context surrounding the cases into consideration (Della Porta, 2008). Within the frame of this thesis the cases consist of armed organisations. These selected cases will be approached through a within and cross-case analysis. This enables a detailed comparison of the selected armed groups and the influence of the political context on their patterns of violence.

The examination of the variation of patterns of violence of armed groups and of changes in the political context will be approached differently. On the one hand, variations in armed group’s patterns of violence will be traced by using descriptive data. On the other hand, changes in the political context will be derived from the analysis of qualitative data that relies on aspects of process tracing and discourse analysis. Content analysis allows approaching the language used in communication while it also takes into consideration the contexts in which they take place. Such an approach of content analysis is the most accurate when the intentions behind the communications can be grasped, as words create social contexts and shape views on the “social world” (Berg, 2006). Discourse analysis focuses on how “a political problem relates to a particular narrative in which it is discussed.” (Hajer, 2006: 66). It will be used in this thesis to trace the evolving political context by examining if one and the same phenomenon were referred to with different terms over time. Furthermore, when armed groups and an opponent government exchange demands or conditions for political concessions, what is being said, and how it is being said allows to make sense of the broader context beyond the simple words that are exchanged (Hajer, 2006).

(25)

4.2 Case Selection

To test my argument that the political context leads armed groups to strategically adjust their patterns of violence against civilians I will examine and compare the patterns of violence of the FARC and the ELN in Colombia from 1998-2016. The case selection is adequate for several reasons: First, the Colombian civil war has been going on for several decades and has been studied by many scholars and from different perspectives. The great amount of high standing research provides richness in quantitative and qualitative data. This allows me to answer my research question without doing fieldwork myself. Second, I choose the FARC and the ELN because at the time they both were active they were similar but not identical Leftist guerrilla groups with comparable objectives, and they were both opposing the Colombian government. If the political context has an influence on armed groups’ patterns of violence I expect the patterns of the two organisations to show similar variations.

The time span used to test my argument reaches from 1998 to 2016. In order to test my argument, this time span is divided into three periods: 1998-2002, 2002-2010, and 2010-2016. The division along these years is made because the time periods cover different Colombian governments and therefore are expected to represent variations in the political context. Starting the research earlier than 1998 would certainly make sense but would also demand more time and resources than available within the context of this thesis. Without a doubt, the division of the time span is debatable, as the political context varies steadily and continuously. However, tackling my argument along these time periods allows for contrasting the FARC’s and the ELN’s patterns of violence under different political circumstances, respectively different Colombian governments in office. Finally, the division of the entire time span of 1998-2016 into three phases also helps structuring the analysis more reasonably.

4.3 Data

The Data Available on Violence Against Civilians in Colombia

To test my argument, reliable data on the FARC’s and the ELN’s use of violence is essential. A prominent dataset in the study of civil wars is the Uppsala Conflict Data Program (UCDP). It gives detailed information on lethal violence committed by state as well as non-state actors (Eck, 2012). However, data solely portraying violence that resulted in casualties would hinder the accurate testing of my argument. If one aims at analysing an armed group’s behaviour in detail, studying homicides is not sufficient (Gutiérrez-Sanín and Wood, 2017). The UCDP, therefore, would only be able to provide an incomplete depiction of the FARC’s and the ELN’s patterns of violence. When looking at the period of time in question,

(26)

1998-2016, the UCDP draws a remarkably different picture than the GTD (discussed in detail below) used for my analysis. While both datasets show a peak in incidents around the turn of the century, only the GTD displays a rise of incidents, for both the FARC and the ELN, around 2010 when the groups once again intensified their violent activities. The sole conclusion that can be drawn from the UCDP when comparing the FARC and the ELN is that the ELN engaged less frequently in lethal violence against civilians (UCDP, 2017). The GTD attributes a total of 749 incidents of violence against civilians where no one was killed, and 262 such incidents are attributed to the ELN. It is therefore obvious that the UCDP and the GTD do not portray the FARC’s and the ELN’s use of violence against civilians in the same way. I argue that the UCDP does not represent an adequate source to test my argument, as violence aimed at coercing the government into making concessions does not necessarily have to be lethal.

In his research of the FARC’s and the ELN’s terror practices, Feldmann (2017) refers to the GTD but mainly relies on a dataset compiled by Justicia y Paz and the Centro de Investigación y Educación Popular (Centro de Investigación y Educación Popular, 2017). The data provided by the latter database on violence against civilians is extremely rich in detail and covers numerous repertoires of violence. One can, for instance, search for violent incidents that occurred in the context of peace negotiations or dialogues. The low number of incidents (16 in case of the FARC and 2 in case of the ELN) explicitly linked to peace negotiations for the time period of 2001 to 2017, however, raises questions. Furthermore, how and why the specific cases got attributed to peace talks does not become clear from reading accounts of the incidents. Moreover, much of the information provided by the database covers aspects, which I do not consider important for the testing of my argument. Therefore, instead of relying on the UCDP or the database of Justicia y Paz and the Centro de Investigación y Educación Popular, I will rely on the GTD to test my argument. The reason why I chose the GTD will be discussed in the following paragraph.

The Data Used in this Thesis

To observe the variation in patterns of violence of the FARC and the ELN from the 1990s until recently, I rely on various datasets and reports. First, the most important dataset is the GTD (2017). I chose to rely on the GTD for a variety of reasons: First, the GTD provides information on events that resulted in casualties as well as events where no one was killed. Second, it allows straightforward searching for violence that is linked to political motives, as GTD’s criteria regarding terrorism (discussed in the conceptualisation above) ensure that the

(27)

violent events had a predominant political motivation. Third, the GTD provides information on a variety of repertoires and targets. One can choose from nine different attack types, and repertoires: “armed assault”, “assassination”, “bombing/explosion”, “facility/infrastructure attack”, “hijacking”, “hostage taking (baricade incident)”, “hostage taking (kidnapping)”, “unarmed assault”, “unknown”. Fourth, for each recorded event, the perpetrator organisation, if known, is given, and every entry in the database comes with a short account of what happened and mentions the amount of civilians that were killed or injured. Fifth, as mentioned above, the GTD uses several criteria to assess the political nature of violent acts. Furthermore, when using the database, the researcher can choose whether to search for all incidents “regardless of doubt” (GTD, 2017) or to limit the search to unequivocal cases. I will only focus on unambiguous cases. Finally, the GTD covers the entire time span relevant for testing my argument.

Additionally, I draw on Gutiérrez-Sanín et al.’s (2015) dataset on small and large massacres in Colombia. The dataset gives a minimum and maximum number of victims for each massacre, mentions the perpetrator organisation, if it is known, and gives short accounts of the incidents. The Dataset covers the time span of 1980-2011. Finally, I consult various reports and newspaper articles that deal with the evolution of violence in the Colombian civil war. I base my research on “Basta Ya!” a report of the Centro Nacional de Memoria Histórica (Centro Nacional de Memoria Histórica [CNMH], 2016); “Auge y declive del Ejército de Liberación Nacional” a report by Echandía (2013) for the Fundación Ideas Para la Paz; a report on kidnapping by the CNMH (2013); articles of InSight Crime; and an article of the Crisis Group. Most newspaper articles used in the analysis come from El Tiempo, and fewer from El País, Semana, Verdad Abierta, and Colombia reports. I chose to rely on these reports and newspapers because several scholars refer to them in their research (Echandía 2013; González Muñoz, 2015; Gomez-Suarez and Newman, 2013; Gutiérrez-Sanín and Wood, 2017) and I therefore consider them to be credible and valuable sources for my analysis.

In order to trace the changing political context I rely on two kinds of sources. First, secondary literature constitutes the basic element supporting my examination of the evolving political context in Colombia. Especially the work of researchers that had direct access to members or internal documents of the FARC and the ELN play a role in my description of the political context. Second, I rely on articles of Colombian news outlets and reports of the organisations mentioned above, to examine shifts in government positions or changes in armed groups’ demands. The analysed articles and reports are about certain terrorist attacks, portray the government’s position towards the armed groups or vice versa, quote officials of

(28)

the FARC or the ENL or portray evolutions of the armed groups over time. To trace the FARC-government relations I analysed 28 different news articles, 2 reports and 4 pieces of secondary literature. For the analysis ELN-government relations I relied on 22 news articles, 3 reports and 2 pieces of secondary literature. I chose to mainly rely on El Tiempo for news articles because it has a vast accessible online archive and because numerous scholars also rely on articles of El Tiempo. I therefore assume that the newspaper constitutes a reliable source of qualitative data. As no academic research tool provided access to El Tiempo’s archive, I conducted the search of the archive manually. I sifted through the archive for the time periods in question using the keywords “FARC” and “peace”, respectively “ELN” and “peace” and selected the articles according to their relevance. I used the same search terms when looking at the other news outlets mentioned above.

Limitations

The descriptive as well as the qualitative data pose certain limitations to this research. First, in a variety of cases it is difficult to ascertain that a specific act of violence really aimed at the government. Some incidents might have been directed to rival armed groups as the target audience of violence but might have affected a wider audience because of the scope or brutality of the incident. Second, as discussed above, different databases and/or scholars adopt different measurements. The way of defining an act of political violence directly influences the amount of events observed. Scholars applying alternative definitions might therefore very well reach different conclusions. Third, other qualitative sources might lead to a slightly different interpretation of the political context. Penultimate, for the last time period from 2010 to 2016 data is considerably less abundant than in the preceding two time periods. Most of the data presented in the datasets and reports mentioned above are only recorded until 2010. For that last time period, the one lasting from 2010 to 2016, this thesis therefore mostly relies on the data of the GTD (2017) as its data ranges until 2016. Finally, data is not available for all forms of violence that were present in the Colombian civil war. This thesis restricts the description of the FARC’s and the ELN’s patterns of violence on repertoires that are backed by significant data: kidnapping, massacring, and attacks aiming at the destruction of private or public infrastructure. One of the repertoires that is left aside in this research but that counts as a classical form of violence in civil wars is, for instance, sexual violence. During the research it soon became apparent that data on this specific is scarce repertoire in the case of Colombia. Furthermore, Gutiérrez-Sanín and Wood (2017) already state that at least in the case of the FARC the engagement in sexual violence was extremely rare.

(29)

4.4 Operationalization

To test my argument that the political context leads armed groups to strategically adjust their patterns of violence against civilians I have to make sense of changes in political context, on the one hand, and to observe the evolution of the FARC’s and the ELN’s patterns of violence over time, on the other hand. To evaluate the political context for each time period I identify the objectives of the government and the guerrilla groups and assess the positions the government and the respective armed organisation had towards each other. I therefore rely on newspaper articles and chapters of reports that mention the terms “FARC” and “peace”, respectively “ELN” and “peace”, and cover the time periods in question. I also draw conclusions from secondary literature and reports.

As this thesis argues that patterns of violence vary with the political context, the patterns of violence of the FARC and the ELN will be analysed. A variation in a group’s patterns of violence is present when the frequency with which a specific repertoire is used changes over time. In order to examine variation in the use of terrorism, one has to clarify which repertoires are seen as terrorism and thus as violence aimed at coercing the government into making political concessions. I will examine terrorist repertoires following the GTD’s guidelines. As mentioned above the GTD lists nine repertoires as forms of terrorism. I leave aside the repertoires “Unknown” and “Unarmed Assault”, as the former lacks clarity and the latter was extremely rare for both organisations analysed here.

“Hijacking”, “hostage taking (barricade incident)”, and “hostage taking (kidnapping)” refer to actions depriving civilians of their liberty. In the analysis I will summarize these three attack types in the repertoire “kidnapping”. “Assassinations” are distinct from Gutiérrez-Sanín et al.’s concept of massacres, as assassinations are seen as the specific killing of public and/or political figures. “Armed Assaults” aim at causing physical harm through the use of firearms. “Bombing/explosion” and “facility/infrastructure attack” aim at destructing private property or public infrastructure; the former requires that explosives were used for the attack. The target “government general” refers to attacks on government buildings as well as current or former members of government. “Utilities” refers to attacks on facilities that serve to generate or transmit energy, such as oil pipelines or power lines. “Private citizens/property” refers to attacks on individual civilians or the public in general. The target “business” refers to companies “engaged in commercial or mercantile activity” (GTD, 2017: 31). This includes, for instance, attacks on restaurants or oil corporations.

(30)

In order to examine, whether these terrorist repertoires were varying distinctively compared to classical civil war repertoires, I will also trace the evolution of the frequency of massacres and economic kidnappings by the FARC and the ELN, as far as data is available.