What drives armed groups in peace processes? : a case study of collective demobilization in Colombia between 1982-2016

(1)

What drives armed groups in peace processes?

A case study of collective demobilization in Colombia between 1982-2016

Name: Paco Mens

Student ID: 10672052

Supervisor: Dr. Abbey Steele Second reader: Dr. Ursula Daxecker

Date: June 2018

Program: Master Thesis Political Science, International Relations Track University: University of Amsterdam

(2)

(3)

2 Abstract

Since 1964, Colombia has been experiencing a violent internal conflict with a number of armed groups. Unfortunately, the countless initiatives that have been started over the years in attempts to secure peace have seen limited success. These many failed negotiations together with the duration of the conflict reveal the difficulty of determining the precise conditions necessary for the leadership of an armed group to sign a peace agreement with the government and to collectively demobilize. Previous work on the Colombian Conflict has failed to structurally compare the peace processes of different armed groups in order to determine what the decisive factors are in their decision to lay down their arms. In this thesis, I have examined the role of a number of factors in these peace processes: the goals of an armed group, commitment problems, the security dilemma, conflict intensity, and prospects for reintegration. I argue that a close examination of these factors combined can adequately explain why an armed group decides to collectively demobilize.

Acknowledgements

This thesis would not have been possible without the help and support of a number of people, to whom I want to express my gratitude: Dr. Abbey Steele, who challenged me to improve the academic quality of this work and who provided me with the insights I needed to make this a good thesis. My parents, for supporting me throughout my whole education. They cheered for me when I learned how to count, and they motivated me to get the most out of my education. Jerome Mies, whose wise advice from the other side of the world made me choose this master’s programme in the first place. Wendeline van Seventer, who motivated me week after week to keep up with my schedule. Carolin Vahar-Matiar and Sarah van Veller, whose support throughout the thesis and intelligent comments were indispensable.

(4)

3

Chapter 1. Introduction

At the end of 2016, the government of Colombia and the Fuerzas Armadas Revolucionarias de

Colombia (FARC) finally signed a peace agreement in which the FARC agreed to lay down its

weapons and attempt reintegration into Colombian society. Although one armed group, the

Ejército de Liberación Nacional (ELN), remained active at that stage, the agreement was seen

as a huge step towards at last bringing the Colombian conflict that started in 1964 to an end. By the time of the 2016 agreement, the conflict had led to the displacement of about 7 million people and had caused over 220,000 casualties (Miroff 2016).

The Colombian conflict has been characterised by the multiplicity of armed actors and almost continuous attempts by the government to reach peace agreements with these actors. In the course of the conflict, some of these armed groups have negotiated peace deals in which they agreed to collectively demobilize. These collective demobilizations have taken place at various points in time and under different conditions. As such, the Colombian conflict allows for a comparison between the varying instances of collective demobilization.

For the purposes of this research, demobilisation will be defined as the disarmament and dissolution of force structures and the transition of combatants into civilian status (Tanner 1996 in Knight & Özerdem 2004). It is the second step in the Disarmament, Demobilization, and Reintegration (DDR) framework that has been applied extensively in modern peace-building efforts. As such, it forms the bridge between the associated processes of disarmament and reintegration. Collective demobilization then, refers to the demobilization of an entire armed group (Norman 2017).

The goal of this thesis is to establish which factors have influenced the likelihood of collective demobilization of armed groups in the Colombian civil conflict. In the early 1990s, four armed groups decided to accept the latest conditions offered by the government and demobilize collectively, while other groups refused. Amongst the latter was the FARC, which engaged in countless rounds of negotiation with the Colombian state prior to signing the 2016 peace agreement. Another group, the ELN, has to this day been unable to reach an agreement with the government despite extensive efforts on both sides. This leads one to wonder what accounts for the differences in the decision-making process of these armed groups. In searching for an adequate explanation of these differences, this thesis will address the following research

(6)

5 question: what factors have influenced the probability of collective demobilization in the

Colombian civil conflict between 1982 and 2016?

Existing research on collective demobilisation, or at least literature using this terminology, is limited. However, the concept is closely related to well-studied concepts such as peace agreements, conflict resolution, negotiation theory, and peace processes, since it is the desired outcome of peace processes with armed groups This thesis will build upon and adapt existing concepts from these strands of literature to shed light on the determinants of collective demobilization. Important concepts include theories about ‘ripeness’ and the ‘Mutually Hurting Stalemate’ as proposed by Zartman (2000), spoiler problems in peace processes as identified by Stedman (1997), the respective roles of greed and grievance (Collier & Hoeffler 2004), and the security dilemma (Sambanis 2002).

These concepts will be used to analyse the different peace processes that have developed between the Colombian government and the armed actors within the conflict. This topic has received much scholarly attention. Notable contributions include Nasi (2009), Chernick (1988, 1999 & 2009), Rampf & Navarro (2014), and Bouvier (2009). Despite the wealth of research on the Colombian peace process, few systematic analyses have been conducted attempting to explain why an armed group chooses to lay down its arms at a specific point in time. Some authors have analysed the peace process of one specific armed group, but to the best of my knowledge there exists only one article that systematically compares the varying factors that have caused armed groups to accept a peace agreement (Nasi 2009).

Nasi analysed the peace processes in Colombia between 1982 and 2002, while taking the different governments as his units of analysis. The present research will build on Nasi’s work, while taking the armed groups as units of analysis instead. By taking the perspective of the armed groups, I am hoping to gain a better understanding of their decisions within a peace process. Lustenberger (2012) also applies the perspective of the armed group in his analysis of the conditions under which armed groups are willing to start negotiations. I will extend his analysis to find out which factors cause armed groups to successfully conclude those negotiations through the process of collective demobilization.

This will also allow me to compare the collective demobilization process of a pro-state militia, the Autodefensas Unidas de Colombia (AUC), to the processes of the leftist guerrilla organizations to see whether substantial differences exist in the demobilization processes of

(7)

6 these armed actors. Furthermore, I can extend the analysis until 2016, the year in which the most important armed group within the Colombian conflict decided to collectively demobilize.

Understanding what determines collective demobilization, or phrased alternatively, what causes armed groups to accept a peace agreement, is of significant importance. Not only could it accelerate the current negotiations between the Colombian government and the ELN, it could also provide valuable information for peace processes elsewhere. While being aware of the uniqueness of the conflict in Colombia, this thesis will attempt to reach conclusions that may provide useful insights for other peace processes as well.

Five sets of factors existing in peace processes in Colombia will be analysed: the goals of an armed group, commitment problems, the security dilemma, conflict intensity, and prospects for reintegration. I expect that these five factors together accurately explain the decision-making process of an armed group in a peace process. Because of the theoretical underpinnings of these factors and their logical connection the likelihood of collective demobilization, I argue that these five factors represent a close approximation of the relevant mechanisms that have influenced the different peace processes.

In this thesis, I will start with an overview of the literature associated with this topic, to provide a clear insight into the relevant theory. Second, I will elaborate on the hypotheses that I aim to test. Third, I will provide some notes on the research strategy that I will use for my analysis, and explain why I have made certain choices. Fourth, I will present the necessary background information on the Colombian conflict and its different armed groups to provide context for the analysis. Fifth, I will commence with the main part of my thesis, the analysis. In this section I will systematically test the hypotheses outlined in the literature review by providing evidence in favour of and against them. Finally, I will finish with a conclusion in which I will also outline the limitations of this thesis and the possibilities for further research.

(8)

7

Chapter 2. Literature Review

In the following section I will present the existing literature on civil wars, negotiations, and peacebuilding. I will give an overview of the mechanisms that are expected to influence a peace process with an armed group in a civil conflict. With this overview, I will place my own research in the context of the existing literature on this topic.

2.1 Civil war literature

As the incidence of interstate wars declined and as civil wars became increasingly common, the academic interest into civil wars1_{rose (Kissane 2015). In the past two decades the literature}

on civil wars has witnessed an explosion of research into its causes and consequences. Such conflicts have a profound impact on any nation that experiences it, which makes it a crucial endeavour to gain a better understanding of this phenomenon (Blattman & Miguel 2010).

A civil war, as defined by Gersovitz and Kriger (2013), is “a politically organized, large-scale, sustained, physically violent conflict that occurs within a country principally among large/numerically important groups of its inhabitants or citizens over the monopoly of physical force within the country” (p. 160). They further add that a civil war is usually characterised by an incumbent government who faces an insurgency that tries to take over the state or to induce political reforms. This insurgency then forms an armed group, and their challenge to state power initiates the outbreak of a civil war (Gersovitz & Kriger 2013).

Causes of civil war

The mechanisms that cause the onset of civil war are also important in understanding its development and termination. Without knowing why an armed group has decided to mobilize in the first place, it is difficult to understand why they might decide to demobilize. As such, literature on the causes of civil war and its termination are closely connected. Similarly, an ongoing conflict can be understood as the repeated decision of armed groups to continue fighting, and understanding why they make this decision is key in understanding why they decide to stop fighting at a specific point in time (Cederman & Vogt 2017).

A recent study on civil war identified three explanatory logics that have been used most prominently in existing literature on the causes of civil war. These logics can be summarised

1_{The Colombian conflict has at times surpassed the arbitrary threshold of 1,000 casualties in one year to be}

(9)

8 as greed, grievance, and opportunities (Cederman & Vogt 2017). The explanatory logic focusing on greed argues that civil wars start because it allows individuals to maximise their utility, mainly in a materialist sense. It stands in contrast to the grievance-based hypothesis which puts more emphasis on political and socioeconomic injustice as the main causes for conflict. A third set of explanations minimises the importance of motives, and instead argues that opportunities for actors to engage in violent mobilization best explain the outbreak of civil conflict. These explanatory patterns are far from mutually exclusive. In practice, the vast majority of conflicts contain a mix of these three sets of explanations and many more (Cederman & Vogt 2017).

Collier and Hoeffler (2004), who demonstrate a greed-based logic, found that the duration of conflicts increases together with commodity prices in countries that are economically dependent on exports. This supports the theory that conflicts should be viewed as a lucrative business for rebel organizations and that conflicts can be fuelled by opportunities for rebels to secure financial resources. Similarly, Fearon (2004) highlights the influence of illegally traded goods such as coca on the duration of conflicts. Such business opportunities allow armed groups to sustain themselves financially and to expand their operations.

Some authors challenge the greed-grievance dichotomy in the analysis of civil war (Vinci 2006; Arnson & Zartman 2005). Vinci (2006) notes that “there has been a call to take into account a more holistic approach, including addressing more traditional features of conflict analysis such as the role of politics and strategic necessity” (p. 25). In his view, wars are not just about greed and grievance, but also about power and survival. He uses a group-level analysis and applies the perspective of the armed group. With this perspective, he assumes that it is “by understanding how groups are motivated that we can understand how they will act and thereby describe and explain their relations” (Vinci 2006, p. 28). He maintains that armed groups are, next to greed and grievance, motivated by the pursuit of power and the maintenance of security.

2.2 Negotiation theory

Armed groups may have multiple motivations to engage in and continue a conflict. Similarly, there are multiple factors that affect their decision to engage in a peace process and collectively demobilize. For collective demobilization to occur, a set of conditions must exist that all parties in the negotiation process are willing to accept. There must exist a bargaining range, or a Zone Of Possible Agreement (ZOPA). The ZOPA exists of all possible agreements that would generate more benefit for both sides than the alternatives they have (Spangler 2003).

(10)

9 Furthermore, the actors involved in the negotiation process must believe that it is possible to obtain a better outcome than the status quo through negotiation (Zartman & Faure 2005). Zartman and Faure assume here that parties entering into a negotiation process are genuinely committed to reach a peace agreement and are not just using negotiations as a military strategy. Sometimes, however, “the decision to negotiate can be a way for the rebel group or the government to win time or legitimacy, without any intention to peacefully settle the conflict” (Walch 2016, p. 76). Thus, peace negotiations only indicate a legitimate willingness to reach peace when the parties can show a credible commitment to the peace process (Zartman & Faure 2005). Such signs of commitment can include continuation with negotiations despite attempts to spoil the process, or costly concessions (Lustenberger 2012).

Ripeness

A conflict between two parties is considered ‘ripe’ for resolution, when both parties perceive the current situation as undesirable. More specifically, they need to be unable to unilaterally escalate the conflict at an acceptable cost. If this is the case, the two parties find themselves in a mutually hurting stalemate (MHS). Usually, the MHS is induced by an impending, past, or recently avoided catastrophe (Zartman 2000). The stalemate can be visualized as either a

plateau, where there exists a situation in which both parties are constantly incurring more costs

than benefits, or a precipice, when the situation predictably and suddenly deteriorates (Zartman 2000).

Another necessary condition for a conflict to be ripe for resolution is the mutual recognition by both parties that the situation can be improved through negotiation. “Parties do not have to be able to identify a specific solution, only a sense that a negotiated solution is possible for the searching and that the other part shares that sense and willingness too” (Zartman 2000, p. 228). Thus, for a conflict to be ripe for resolution the parties need to perceive the current situation as a MHS and they need to perceive negotiation as a way out of the current situation.

2.3 Mechanisms affecting the peace process

Ripeness is a useful concept to explain the timing of negotiations, but it is not a factor in itself. Rather, it is a state that the conflict is in, and it needs to be complemented by factors that explain how the ripeness comes about. Lustenberger (2012) aims to find out what factors induce armed groups to start negotiations. His theory suggests that the decision of an armed group to enter into negotiations can be explained by push and pull factors. Push factors are those factors that

(11)

10 make a continuation of the conflict costlier (MHS), whereas pull factors make the prospect of negotiations more appealing (way out).

The literature identifies many factors that affect the peace process. As it is beyond the scope of this research to examine every possible factor, those that are most important for peace agreement negotiations with armed groups in the context of civil conflict have been selected. These factors are all expected to influence both the costs of continuation of the conflict and the prospects of negotiations. The relevant concepts in the literature that are connected to these factors are spoilers, ideology, commitment problems, the security dilemma, and leadership. In the following section these concepts will be explained and discussed.

Spoilers

As put forward by Stedman, a significant barrier for peace in the context of internal conflict is the presence of spoilers. Spoilers are defined by Stedman as “leaders and parties who believe the emerging peace threatens their power, world view, and interests and who use violence to undermine attempts to achieve it” (Stedman 1997, p. 178). The existence of spoilers is the result of an almost inescapable characteristic of conflict: not everyone wants peace, not everyone wants it in the same way, and not everyone wants it at the same time. Those who are not happy with the peace at hand have an incentive to prevent it from being reached. Sometimes, those that are negotiating the peace are the spoilers of their own peace agreement (Stedman 1997).

Stedman’s typology of spoilers identifies three types: total, limited, and greedy. These types determine the behaviour of a spoiler and determine which spoiler management technique should be applied. A total spoiler can only be satisfied by achieving total power. Such spoilers cannot be incorporated into a peace process and should therefore be dealt with militarily. A limited spoiler is similar to the total spoiler in the sense that their demands are not subject to change, but different in the sense that it does not seek total power. For the Colombian peace process the most relevant category is the third: the greedy spoiler. This type of spoiler deals with the peace process in an opportunistic way. Its demands depend on the capabilities they have and the opportunities that are present within the context of peace negotiations.

Following Greenhill & Major (2007), this thesis will use a slightly altered version of the typology proposed by Stedman. They argue that structural factors remain more important than individual preferences of spoilers. “That is, the type of spoiler does not determine the kinds of outcomes that are possible; instead, the kinds of outcomes that are possible determine the type

(12)

11 of spoiler that may emerge at any given time” (Greenhill & Major, p. 8). According to this idea, all potential spoilers are essentially of the greedy type, even though they may attempt to come across as one of the other types as a negotiation strategy. Then, the most important factor determining spoiling is the opportunity structure.

Actors are presented with a number of available options that induce them to change their goals, without affecting their preferences. An actor that has the power to unilaterally affect the peace process in their favour will likely do so, whereas a less powerful actor will continue to pursue peace. As such, the concept of a spoiler remains crucial for the analysis of a peace process, but the opportunity structure is what eventually determines actions, through the influence it has on the goals of actors.

Ideology

In the typology described above, armed group behaviour is expected to directly follow from the existing opportunity structure. However, different levels exist within the greedy spoiler type, and these levels are related to the ideological commitments of an armed group. “ideology is best understood, we suggest, as a set of more or less systematic ideas that identify a constituency, the challenges the group confronts, the objectives to pursue on behalf of that group, and a (perhaps vague) program of action” (Sanín & Wood 2014, p. 214).

Sanín & Wood (2014) argue that ideology does not just have instrumental value. Sometimes, armed groups “act on normative commitments in ways not reducible to instrumental reasoning, and some groups constrain their strategic choices for ideological reasons, often normative concerns prescribed by their ideology” (p. 213). Consequently, those armed groups that have strong ideological commitments are more closely associated with the limited spoiler type. The behaviour of these armed groups does not directly follow from the existing opportunity structure, but is also influenced by their ideological convictions.

Commitment problems

Related to spoilers and equally troublesome for any peace process are commitment problems. Turning in your weapons and disintegrating command structures will drastically affect the bargaining power of an armed group and will therefore limit the incentive for the state to keep its promises. Therefore, achieving peace can be difficult even in an environment where all actors are in favour of collective demobilization (Humphreys & Weinstein 2007).

(13)

12 The demobilizing armed group can also renege on its commitments after concessions have been made by the government. Alternatively, an armed group may simply not be able to keep its promises. If, for example, an armed group decides to demobilize in exchange for political reforms, it is possible that individual combatants refuse to demobilize together with their leaders. Then, the government has made concessions without achieving the desired result (Sambanis 2002).

Such commitment problems can be overcome by an external actor who is able to guarantee the implementation of a peace agreement (Walter 1997). Walter (1997) found that armed actors in a civil war almost always fight until the end unless an outside power steps in to guarantee the peace agreement. However, examples do exist where commitment problems were overcome, in which the government managed to sign peace agreements with multiple armed groups at multiple moments without an external actor being present to guarantee the implementation of the agreements (Nilsson 2008; Nasi 2009).

A potential explanation for this finding is related to the concept of partial peace. Nilsson (2008) finds that the exclusion of one rebel group from a peace agreement does not affect its durability. The presence of armed actors excluded from a peace agreement does not have to aggravate commitment problems. It can alleviate commitment problems because it provides the state with an extra incentive to come across as a trustworthy negotiating partner. The security dilemma may become more severe, but armed groups have usually already calculated these risks into their decision to sign the peace agreement (Nilsson 2008).

Security dilemma

Closely linked to commitment problems is the security dilemma. This dilemma finds its origins in realist international relations theory. It is usually applied to explain the emergence of conflict (Norman 2017). “Regime collapse results in domestic anarchy comparable to the international system, and groups in “stateless” societies will build military capacity in response to uncertainty about others’ intentions, signaling aggression and a spiraling into armed conflict” (Norman 2017, p. 6).

In this thesis, the concept of the security dilemma will be used to analyse why armed groups refuse to collectively demobilize. Collectively demobilizing in an institutionally weak environment is a risky endeavour for an armed group. “Even if all parties favor the dissolution of their military factions, in an institutionally weak environment, mutual mistrust may result in

(14)

13 an unwillingness to take the first step toward demobilization” (Humphreys & Weinstein 2007, p. 535).

In a civil conflict, the state has difficulty to guarantee the security of the people, and people might feel safer under the protection of an armed group, or as a member of an armed group. If that group then decides to demobilize, they lose this protection and again become dependent on the state to guarantee their security. If an armed group is uncertain about the capability or the willingness of the state to provide their security after demobilization, they may refrain from demobilizing (Norman 2017).

Leadership

Leadership has the potential to affect both commitment problems and the security dilemma, as well as the prospects for reintegration. Lieberfeld (2016) proposes a slight variation of the ripeness theory as articulated by Zartman, arguing that leadership behaviour can often not be explained while employing the rationality assumption that Zartman utilizes. Instead, leaders are irrationally constrained by ideology, previous statements, and electoral considerations. A newly installed leader, especially from the government side, can be a more effective negotiator because of his or her ability to dissociate themselves from the strategies of previous administrations, and to apply a holistic perspective (Lieberfeld 2016).

“Once official negotiations begin, the government’s own political prospects become dependent on a successful outcome, which raises the government side’s commitment to negotiation” (Lieberfeld, p. 19). With this addition, Lieberfeld adds internal power dynamics of the negotiating parties into the concept of ripeness. For an agreement to be feasible, the parties must not only find themselves in a MHS with negotiation as a ‘way out’, they must also believe that a resolution of the conflict is in their own political interest. This holds for the government actors as well as for non-state actors.

(15)

14

Chapter 3. Hypotheses

It is beyond doubt that determining which factors regulate the likelihood of collective demobilization of an armed group is an incredibly difficult task. In the context of any civil war there are many factors that influence the decision-making process of the negotiating parties and the Colombian conflict is no different. However, gaining a better understanding of the factors that influence the probability of collective demobilization is a crucial component of research on peacebuilding. For this reason, following Nasi (2009), I will argue for a number of factors and causal mechanisms to be especially relevant in the decision-making process of armed groups. “Perhaps none of these factors individually explains a peace process’s success or failure, but understanding their relative impacts may help us to better assess what factors are likely to advance or hinder movements toward peace in the future” (Nasi 2009, p. 40).

H1: As the goals of an armed group become more limited, an armed group is more likely to collectively demobilize.

The demands of an armed group in a negotiation process are likely to be closely associated with their goals. All other things being equal, the government is more likely to accept a set of modest demands compared to more ambitious ones. The concept of the ZOPA accurately illustrates this relationship. The government wants to minimise its concessions. If the goals of the armed group become more limited, they will accept smaller concessions from the government, resulting in an expansion of the ZOPA. Thus, a peace agreement is more likely reached when the goals of an armed group are limited.

H2: As commitment problems are effectively mitigated, an armed group is more likely to collectively demobilize.

Commitment problems are especially significant in the context of a civil war. In the context of Colombia perhaps even more due to the limited interference of outside actors in the peace process. If the government can somehow credibly commit to the concessions from the peace agreement, an armed group is more likely to sign it. Likewise, the government will be more likely to sign a peace agreement if the ability of an armed group to break its promises is limited.

H3: As the security dilemma becomes increasingly limited, an armed group is more likely to collectively demobilize.

The security dilemma is especially relevant in a multi-party civil war context in which partial peace is established. The ongoing conflict and the remaining active armed groups can provide

(16)

15 extensive security risks for demobilizing combatants. If these security concerns are effectively minimised, an armed group will be more likely to sign a peace agreement.

H4: As conflict intensity increases, an armed group is more likely to collectively demobilize.

Increasing conflict intensity can contribute to the parties’ perception of a MHS, and thereby has the potential to make a conflict ‘ripe’ for resolution. As outlined before, the conflict is ripe for resolution when both parties perceive a MHS and a ‘way out’ of that stalemate through negotiation. If the conflict is going on at a high intensity, both parties are incurring high costs, and they are more likely to perceive the situation as a MHS. Therefore, high conflict intensity will increase the chances of collective demobilization.

H5: As the prospects for reintegration improve, an armed group is more likely to collectively demobilize.

The expected utility for the leadership of an armed group improves when they know that they and perhaps their constituents will be able to build up a new life after collective

demobilization. Especially relevant here are the limited prospect for persecution and the attitude of the leadership. If amnesty for war crimes is promised as a part of the peace agreement, the leadership of an armed group will be more likely to sign it. Likewise, if the leadership of an armed group perceives the current leader as having a relative favourable attitude towards the armed group, their expected prospects for reintegration will increase.

Outcomes of peace processes with armed groups are not exclusively connected to the

variables that have been proposed in this section. Rather, these variables are those that I think are the most strongly supported by the existing theory and research. Throughout the analysis, I will be aware of alternative causal mechanisms that have not been included in the present theoretical model. I will conclude my analysis with a short discussion of these alternative explanations.

(17)

16

Chapter 4. Research strategy

4.1 General Strategy

The proposed theory will be tested on the peace process between the government of Colombia and various armed groups from 1982 until 2016. The Colombian conflict has been chosen because it allows for a structural comparison between different peace processes within the same country context. This will enable me to distinguish more confidently what factors account for the observed variation. I will start by discussing the background of the conflict, in order to gain a better understanding of the relevant dynamics. Then, I will engage in a structural comparison of the peace processes of different armed groups to determine what factors ultimately lead to their decision to collectively demobilize.

Contrary to other literature on the Colombian peace process (Nasi 2009, Chernick 1999), I will use the armed group as my unit of analysis, rather than the government administrations. The downside of this is that I will have to repeat some of the relevant circumstances in my analyses, but it will allow me to gain a more in-depth understanding of the decision-making process of each armed group. Since my theory is mainly written from the perspective of the armed groups, this will add to the strength of my analysis.

I have chosen for a qualitative case study approach because it will allow me to delve deeper into the motives for every armed group to remain active or to demobilize (Cooper 2014). The Colombian conflict is tremendously complex, and only a focused qualitative study can take into account all the complexities that influenced the course of the conflict. “Case studies provide an opportunity to glean nuance that statistical approaches often overlook” (DeRouen et al. 2010, p. 336).

In this thesis I will apply George and Bennett’s method of structured, focused comparison. With this method, “the researcher writes general questions that reflect the research objective and that these questions are asked of each case under study to guide and standardize data collection, thereby making systematic comparison and cumulation of the findings of the cases possible. The method is ‘focused’ in that it deals only with certain aspects of the historical cases examined. The goal of such a study is to explain an observable variation in the dependent variable” (George & Bennett 2005 in Dukalskis 2015, p. 850).

(18)

17 Comparing the peace processes of different armed groups within one state allows me to more accurately distinguish which variables caused the collective demobilization to occur. At a specific moment, contextual factors such as the geography and the capabilities and attitude of the state are held constant for every armed group. With these variables held constant, they cannot account for the observed variation in the dependent variable (Dukalskis 2015).

In the formulation of my hypotheses, I have purposely used the term ‘armed group’, without specifying whether this armed group is a pro-state militia or a rebel group fighting against the state. Applying the same model to both of these types of armed groups will allow me to find out to what extent the causal mechanisms associated with collective demobilization are different from one another. A more focused comparison between the decision-making processes of these types of armed groups is beyond the scope of this study, but in this way this study will add to the literature on the differences between rebel groups and pro-state militias.

The last concept worth discussing here is path dependence. I purposely use the word concept here since path dependence can hardly be treated as a factor or a variable. “Most generally, path dependence means that where we go next depends not only on where we are now, but also upon where we have been” (Liebowitz & Margolis 2000 p. 981). In this thesis, path dependence will be considered when analysing the decision-making processes of armed groups and the Colombian government.

4.2 Operationalization

Many of the relevant variables in this study cannot be objectively determined, since they are by definition subjective. The security dilemma, for example, depends on the perception of the conflict parties and is therefore difficult to establish empirically ex ante. Similarly, even if a correlation can be objectively determined in the way that the causal mechanism proposes, it is impossible to be certain that the variation in the dependent variable was indeed caused by the proposed variables.

These problems are minimised in two ways. First, the proposed causal mechanism is backed up by a logical argument and by existing theory, as described in the hypothesis section. This increases the likelihood that observed correlation indeed implies a certain degree of causation. Second, data sources will be utilized to test the strength of the causal mechanism. These sources most importantly consist of scholarly articles on the Colombian peace process, and to a lesser extent also consist of available public discourse on the peace process by the Colombian

(19)

18 government, members of armed groups, and non-academic observers. This second group of source material will be examined more critically, as these actors may have incentives to mispresent information.

Most important for this study and perhaps easiest to measure is the dependent variable: collective demobilization. It is important to distinguish the concepts of individual, partial, and collective demobilization. The first refers to individual decisions to leave an armed group in order to become a civilian. Even when a large group of individuals decides to demobilize together, it still counts as individual demobilization. Collective demobilization, however, refers to the decision of the highest leadership of an armed group to demobilize. Even when a large number of individuals or a whole regiment does not follow the decision of their leaders to stop fighting, it still counts as collective demobilization in this thesis. It is also possible that only one part of an armed group decides to demobilize. For example, if the leadership of one regiment decides to demobilize against the will of its superiors, then that regiment still has demobilized, but the armed group as a whole has not. In this thesis, this will be described as partial demobilization.

In the Colombian conflict, instances of collective demobilization always came together with a peace agreement and vice versa. Because of this, the distinction between the two concepts within this context is subtle and of limited analytical importance and they will be used interchangeably.

This thesis, according to the definition described before, analyses 6 moments of collective demobilization in the Colombian peace process between 1982 and 20162. In the analysis, individual demobilizations, partial demobilizations, periods without negotiations, ceasefires, and failed negotiations will also be considered to find out which factors account for the observed variation in the dependent variable. Therefore, the dependent variable could also be described as ‘events in the peace process’.

The goals of an armed group will be estimated by looking at discourse of the members of armed groups, the content of peace agreements, and a close inspection of the negotiation process. Both across-time and across-unit variation in the goals of armed groups will be used for the analysis. Two concepts closely related to the goals of the armed group will also be used while testing the first hypothesis.

2_{In chronological order: M-19, EPL, MAQL, PRT, AUC, FARC. The peace process of ELN has not resulted in}

(20)

19 Firstly, the ideological strength of an armed group. A distinction is often made between greed- and grievance-based armed groups, where the latter represents those groups strongly motivated by ideology (Cederman & Vogt 2017), and who are likely to bring more ambitious demands to the negotiating table (Chinchilla 2010). Secondly, the capabilities and opportunities of an armed group will be considered. Rationally acting armed groups will alter their goals depending on the bargaining position that they have in the negotiation process. As such, the capabilities and opportunities serve as indicators for the goals of an armed group.

Both commitment problems and the security dilemma will be estimated by looking at existing analyses from scholars on the Colombian conflict. For both variables, the capacity of the state to guarantee the security of combatants after demobilization will be considered. For commitment problems, the content of promised concessions in peace negotiations, incentives to renege on made promises, internal cohesion, and the presence of spoilers will be considered when estimating the extent to which commitment problems affected the peace process. For the security dilemma, the presence or absence of hostile armed groups and their ability to target demobilizing combatants will receive special attention.

Conflict intensity will be estimated by looking at the total damage dealt and felt by a specific armed group. Conflict intensity in itself does not affect the chances of collective demobilization. It only does so when it contributes to the perception of a MHS and thereby ripens the conflict. Therefore, the analysis section will discuss whether conflict intensity caused the perception of a MHS among the warring parties.

For the armed groups, the total number of casualties suffered over a certain period will be analysed. For the government, I will investigate the number of casualties that the armed group has caused over a certain period. This distinction is necessary because the casualties inflicted by other actors in the conflict have a limited impact on the government’s perception of a MHS. Civilian casualties, however, are important in this respect. Violence against civilians can be applied as a strategy aimed at pressuring the government into concessions (Hultman 2007). For the armed group, by contrast, it does not matter who inflicts the casualties on them for their perception of a MHS. If a rebel group is being targeted by a pro-state militia, for example, they may be equally induced to reach a peace agreement. The number of casualties will be estimated by using the Uppsala Conflict Data Program (UCDP) (Allansson et al. 2017).

Since the data from the UCDP only starts in 1989, and to decrease measurement error in estimating the true conflict intensity, two other datasets will also be analysed; the dataset on

(21)

20 Selective Assasinations (1981-2012) and on Massacres (1980-2012). The fact that these datasets highlight a different aspect of violence perpetrated by the armed groups will allow me to get a better idea of their total repertoire of violence (Sanin & Wood 2017). Both these datasets are compiled by the Centro Nacional de Memoria Histórica (CNMH) and are derived from multiple sources such as news reports and from organizations such as Noche y Niebla (CNMH 2012). More information on how the data has been gathered and what characteristics are needed for an event to be included in the dataset can be found on their website3_{. Unlike the}

UCDP dataset, these datasets do not include any information on the number of casualties that an armed group has suffered. However, the data can be used to estimate how much an armed group contributed to the perception of a Hurting Stalemate for the government in a certain period. For a significant portion of the data, it is unknown who was the perpetrator of the violence. Thus, if one armed group is especially good at concealing their violent acts, the data could be biased.

Prospects for reintegration will be estimated by looking at the conditions that are offered to demobilizing combatants by the government. Since the expected prospects for reintegration also influence the peace process, the attitude of current leadership and the amount of individually demobilizing combatants will be used as indicators to estimate changes in this variable over time. A Colombian president with a favourable attitude towards an armed group could increase their expected prospects for reintegration. A more direct measure of this variable is the presence or absence of amnesty and anti-extradition laws. These will also be used as indicators to estimate this variable.

(22)

21

Chapter 5. Background

In order to gain a better understanding of the behaviour of armed groups, it is crucial to know the context in which these groups emerged. The Colombian conflict is incredibly complex, and it is closely connected to historical events that took place before the start of the current conflict. In the following section, the root causes of the conflict will be given a closer look, which will facilitate understanding of why the conflict has developed in the way it has.

5.1 Origins

The origins of the modern armed conflict in Colombia are commonly traced back to a violent period between 1948 and 1958, known as ‘La Violencia’. La Violencia can roughly be described as a partisan civil war between the two dominant political parties in Colombia; the conservative and the liberal party (Chernick 1999). The conflict ended when these two parties came up with a power sharing agreement, hoping to bring the partisan violence to an end for once and for all. This agreement marked the beginning of the National Front, which stipulated that each of the parties would alternate every four years in delivering the president. By that time, in 1958, La Violencia had caused the death of around 200,000 people (Nasi 2006).

The agreement between the two biggest parties was moderately effective at curbing violence between these two, but it excluded other parties to join the political system. There was no space for a third political force, that represented the interests of the workers, the farmers, or the least favoured social class in general. This limited possibility of political participation is viewed by some as one of the main reasons that so much violent opposition emerged against the government (Garibay 2003; Steele 2017). If those who wanted something different for Colombia than the National Front, they could not make any changes through a democratic process. As a result, they resorted to violence to get what they want.

Other than creating new problems, the agreement also failed to solve some of the underlying problems that served as the root causes for violence in Colombia. One of these root causes is summarised as the agrarian problem (el problema agrario). The Colombian agricultural sector is characterised by a large concentration of rural property; a large portion of the best lands are owned by a small percentage of large landowners, leaving the rest of the Colombian farmers dissatisfied (Posada 1987).

(23)

22 One of the political streams that was excluded from Colombian politics was communism. With the global rise of communism in the 1950s, the National Front’s worries about communist opposition also began to rise. It was considered by the National Front that certain rural communities served as breeding grounds for communism, and they attacked these communities to prevent the opposition from getting too strong. The agrarian problem, the limited democracy, and these attacks, ultimately led to the creation of Colombia’s first and most famous insurgent organization: the FARC.

Inspired by the Cuban revolution and the global rise of communist movements, the FARC was subsequently joined by the ELN and the EPL. These three armed groups were all leftist guerrilla organizations who fought for more equality in Colombia. The guerrillas were distinguished from one another because of ideological differences, military strategies, and geographical positions.

5.2 Development

During the first ten years of the existence of these guerrilla groups, their influence and power were limited (Messaoudi Rodriguez 2011). However, the state did not manage to completely eliminate the insurgencies. The power of these armed groups continued to increase during the 1970s, which was also the moment when the Movimiento 19 de Abril (M-19) came into existence.

As a response to the rising power of the guerrilla organizations and the state’s lacking ability of the state to fight them, opponents of the guerrillas formed paramilitary organizations. Some of these pro-state militias emerged as private armies owned by drug-lords and large landowners to defend them against the guerrillas, whereas others were created by civilians themselves to curb the increasing power of leftist organization. The paramilitaries arguably represent the most brutal force in the Colombian conflict (CNMH 2013).

Closely connected to the rise of the paramilitaries but also to the conflict in general was the Colombian drug trade. In 1975, 70% of the world’s marijuana supply was produced in Colombia. With the income generated from this business, drug cartels were able to become very powerful. This tendency only accelerated when Colombia became increasingly involved with the production of cocaine, a much more lucrative business. Colombian narcotraficantes bribed many government officials, thereby decreasing the legitimacy and capacity of the state.

(24)

23 They also benefit from instability in the country, providing them with an incentive to fund both guerrilla and paramilitary organizations (Pardo 2000).

The drug trade not only aggravated the conflict because of the violence between drug cartels, but it also became an important source of funding for some of the armed groups. Especially when the government managed to defeat the largest drug cartels in the 1990s, the gap they left behind was filled by the armed groups. Paramilitary as well as guerrilla organizations became increasingly involved with the drug trade (Pardo 2000). Income from the drug trade allows insurgency groups to perpetually finance their operations, thereby making them significantly harder to defeat by the government. As such, “there will be no peace without dealing with the drug plantations” (Pardo 2000, p. 71).

The peace process with these armed groups formally started under the Betancur administration in 1982. Ultimately, this peace process resulted in a ceasefire agreement with the FARC and the M-19, and the creation of a political party by the FARC: the Unión Patriótica (UP). The ceasefire collapsed soon after, and many members of the UP were assassinated by the paramilitaries (Theidon 2007). Supporters of the UP were also forced to displace on a large scale, causing public support for the UP to decrease rapidly (Steele 2011).

During Barco’s (1986-1990) and Gaviria’s (1990-1994) administration the peace process achieved its first big successes, with the collective demobilization of the M-19, EPL, and several other armed groups. As a part of this process, Colombia also obtained a reformed constitution, which finally marked the end of the National Front. The conflict was far from over, however, since the two largest guerrilla organizations (ELN and FARC) and the paramilitaries remained active (Chernick 1999).

These developments were to some extent influenced by the end of the Cold War and the existing momentum against communism in the international arena. In 1989, all communist regimes in Central and Eastern Europe were overthrown, and other civil conflicts such as those in Nicaragua (1990) and El Salvador (1991) were also lost by the communists (Hironaka 2009). The result of this was that many of the leftist organizations in Colombia felt that they would also be on the losing side. Instead of remaining their faith in a complete military victory, they would rather choose a political solution to the conflict (Rampf & Navarro 2014).

In the years after, Ernesto Samper (1994-1998) and Andres Pastrana (1998-2002) served as the presidents of Colombia. While Samper’s ability to negotiate peace agreements was limited because of alleged links with drug cartels, he did manage to establish a demobilization process

(25)

24 for combatants. “Law 418 of 1997 provided amnesty to members of politically motivated groups who turned themselves in, as long as they were not guilty of crimes against humanity” (Steele 2017, p. 103). As a result of this law, many combatants decided to demobilize individually. When Pastrana came to power, he promised to change the strategy from individual to collective demobilization. He began talks with the FARC and gave them a demilitarized zone as a sign of good will. The FARC used this zone to consolidate and expand its capacity (Steele 2017).

The peace process achieved another breakthrough under the Uribe administration (2002-2010). Uribe was elected with the promise to start a ruthless military campaign against the country’s guerrilla organizations and did so with success: “After more than a decade of sustained military pressure facilitated by more than $10 billion in US assistance through Plan Colombia4, however, both groups have been rooted out of their traditional strongholds and persuaded to negotiate settlements with the government” (Angelo 2017, p. 135).

While pursuing a military strategy with respect to the guerrilla organizations, Uribe negotiated with the paramilitaries. The paramilitaries at this point were loosely organized as the AUC5. Uribe, who has been accused of collaborating with the paramilitaries, negotiated a peace deal with them, causing them to collectively demobilize between 2003 and 2006 (Valencia Agudelo 2007).

The last chapter of the Colombian peace process starts with the election of Juan Manuel Santos in 2010. In 2012, negotiations with the FARC started, which ultimately led to the signing of a peace agreement in 2016. Currently, in 2018, negotiations with the remaining armed groups (ELN, EPL) persist and remobilized paramilitary groups and some dissidents of the FARC continue to plague the country. Big steps have been made towards total peace, but for now the conflict in Colombia is on-going (Keijzer 2017).

In the two tables below, one can find an overview of the Colombian presidents that served between 1982 and 2016, and the years in which the different armed groups in Colombia have been active.

4_{Plan Colombia was an aid programme funded by the US aimed at combating guerrilla organizations and drug}

cartels.

(26)

25

Table 1

President Tenure Party

Julio César Turbay 1978-1982 Liberal

Belisario Betancur 1982-1986 Conservative

Virgilio Barco 1986-1990 Liberal

César Gaviria 1990-1994 Liberal

Ernesto Samper 1994-1998 Liberal

Andrés Pastrana 1998-2002 Conservative

Álvaro Uribe 2002-2010 Colombia First

Juan Manuel Santos 2010-2018 National Unity

Table 2

Armed group Years in which active

FARC 1964-2016 ELN 1964-present EPL 1967-19916 M-19 1970-1990 AUC 1997-2006 MAQL 1984-1991 PRT 1982-1991

(27)

26

Chapter 6. Analysis

In the following section I will analyse the impact of five factors on peace processes with armed groups in Colombia between 1982 and 2016. A summary of my findings can be found in the table below. A brief analysis of the table already reveals that some of the variables are not related to events in the peace process as hypothesised. This will be further investigated in this chapter.

(28)

27

Table 37

Armed

group Year Result

Limited goals Limited commitment problems Limited security dilemma High Conflict Intensity High prospects for reintegration M-19 1984 Ceasefire X X EPL 1984 Ceasefire X FARC 1984 Ceasefire X X M-19 1990 Peace agreement X X X MAQL 1991 Peace agreement X X X X EPL 1991 Peace agreement (Partial) X X X X PRT 1991 Peace agreement X X X FARC 1991 Failed Negotiations X X X ELN 1991 Failed Negotiations X X FARC 1998 Failed Negotiations X X ELN 1998 Failed Negotiations X AUC 2003 Peace Agreement X X X X X FARC 2016 Peace Agreement X X X ELN 2016 Failed Negotiations X X X

7_{All inputs in bold represent instances of collective demobilization. For the events in the peace process where an}

agreement was eventually signed, I have chosen the year in which the agreement was signed. For the failed negotiations, I have chosen the year in which those negotiations started.

(29)

28

6.1 M-19

Background Origins

The M-19 (Movimiento 19 de Abril) was the first guerrilla organization that mainly operated from urban areas. Its name is derived from the presidential elections of the 19th of April 1970, which were supposedly fraudulent. The movement is part of the so-called second wave of armed groups that emerged in Colombia in the 1970s, and they initially emerged as a splinter faction of the FARC. Their ideology was a mixture of nationalism and revolutionary socialism, and their main aim was to open up democracy in Colombia (Leonard et al. 2013).

Development

M-19 was very successful at generating publicity for their cause. When they managed to steal around five thousand weapons from the main arsenal of the Colombian armed forces, a wave of repression started, which resulted in the imprisonment of many M-19 members. In an attempt to free their members from prison, they decided to capture the embassy of the Dominican Republic in 1980. With this action, they took several important people hostage, including the US ambassador (Leonard et al. 2013).

By the beginning of the 1980s, popular support for the armed groups and their increasing power forced the Colombian government to engage in negotiations with them. Eventually, these negotiations resulted in ceasefire agreements with M-19, EPL, and FARC. These agreements, signed in 1984, proved not to be enduring. Violence between the parties continued, and mutual distrust increased. In 1985, the agreement officially collapsed when the M-19 attacked the Palace of Justice. In this attack, twelve Supreme Court judges were killed, and M-19 suffered many casualties as well. Perhaps emboldened by the successful attack on the Dominican Republic embassy a few years prior, they had overestimated their capabilities and went too far (Messaoudi Rodriguez 2011).

Peace agreement

After the attack, public support for the movement declined rapidly, and repression from the government increased. By the end of Barco administration (1986 – 1990), they agreed to lay

(30)

29 down their weapons and convert into a political party. Two years later, “the Colombian Congress approved a referendum to set up a constituent assembly to reform the constitution as to allow for greater political participation by all forces, fulfilling one of the main demands of the guerrillas” (Leonard et al. 2013, p. 573).

This constituent assembly was a major accomplishment for the M-19 and arguably the most important event in Colombian politics of the 20th_{century (Durán et al. 2008). This assembly}

allowed the widening of democratic spaces, and the possibility to participate in it induced other armed groups to demobilize as well. As part of the peace agreement, M-19 was also allowed to legally start a political party (Durán et al. 2008).

This political party, the Alianza Democrática M19 (AD-M19), managed to get a significant vote share in the constitutional assembly (27,3%). This share allowed Antonio Navarro Wolff, a leader of the M-19, to participate in the Consitutional Assembly of Colombia as one of three co-presidents. These initial results gave new hope to those who had already given up on transitions from illegal armed groups to legitimate political parties. However, the successes proved not to last. Even though some of the members have successfully pursued a political career, the AD-M19 as a party on its own formally ceased to exist in 2000 (Durán et al. 2008).

Analysis

RQ: Why did M-19 collectively demobilize in 1990?

Goals of the armed group

M-19 can be characterized as an armed actor that is pragmatic and flexible in the way they define their goals and aim to achieve their goals. At first, they started as an urban guerrilla movement. When state repression increased, they partially moved their operations to the countryside. At first, they wanted to materialize their revolution by overthrowing the government. After the capture of the Dominican Republic embassy they proposed a peace deal. “This was an ideological break, because Latin American revolutionaries were born under the influence of phenomena such as the Cuban Revolution, which had the slogan of ‘win or die’, not conciliate or negotiate. Therefore, daring to propose a negotiated solution to the conflict was considered an absolute heresy” (Patiño 2001 in Durán et al. 2008, p. 21).

(31)

30 This proposed peace deal enjoyed broad popular support, as 77% of the population was in favour (Chernick 1988). The presidential elections at the time were dominated by the dilemmas surrounding peace and guerrilla violence. The president who won, Belisario Betancur, adopted parts of the proposed M-19 agreement, and thereby added a significant political dimension to the M-19’s endeavours. “More than any other actor, the M-19 had influenced the direction of the political debate in the final years of the Turbay Administration and during the first months of the presidency of Belisario Betancur. Its call for a ceasefire and national dialogue became the basic reference point for the other political sectors. During the congressional debate, they were directly consulted” (Chernick 1988, p. 65).

These developments made the M-19 realize that they had significant bargaining power over the Colombian government. During the early stages of the Colombian peace process, the M-19 was considered as the most important actor. “Something occurred which had never before been achieved in Colombia: the constitutionally-mandated head of state of Colombia publicly entered into talks with the leadership of an armed guerrilla movement. That means that we are a belligerent force in this country” (Interview with Alvaro Fayad8 1985 in Chernick 1988, p. 511).

The dominant position of the M-19 also meant that they received a lot of repressive attention from the Armed Forces of Colombia. Ultimately, this caused the gradual decline of the capabilities and the popular support of the M-19, which culminated in the attack on the Palace of Justice. After this attack, many people withdrew their support for the armed group. As a result, their bargaining position had weakened.

Another development within the M-19 also supported the peace process. M-19 had initially emerged as a radical group seeking to install socialism through force. But as time passed, they became increasingly moderate and realized that democracy was the best path to achieve their objectives. They formally established this new strategy at the Seventh Conference in 1979 (Durán et al. 2008). The leader of M-19 in 1980, Jaime Bateman, proposed a negotiated solution to the situation with the siege of the Dominican Republic Embassy: “That was the first attempt at peace in Colombia: Bateman’s proposal for a negotiated solution through dialogue, truce, and amnesty” (Patiño 2001 in Durán et al. 2008, p. 16).

(32)

31 In this period, they did not eschew violence against the state’s armed forces, even though they eventually wanted to achieve their goals through democracy. The increasing numbers of civilian casualties together with the weakening of the military capacity of the armed group led the group to a further reconsideration of their strategy. They decided that peace was the best way to reconnect with the people, and to recover their influence on what they referred to as ‘the democratic project’ (Durán et al. 2008). In conclusion, the decreased capabilities of M-19 together with the moderation of their ideology paved the way towards the collective demobilization of M-19.

Commitment problems

One aspect that contributed to the failure of the ceasefire agreement in 1985 was the limited participation of the military. The proposed agendas lacked proposals to reform the military, and the Colombian armed forces had developed a significant degree of autonomy (Chernick 1988). Without the support of the military, it was hard to create an environment in which the safety of demobilized combatants could be guaranteed.

Internal cohesion was also lacking on the side of the M-19. Up until his death in 1983 the undisputed leader of M-19 was Jaime Bateman. After his death, however, M-19 struggled for a long time to find a worthy successor. The power struggle resulting from that caused fractures within the leadership of the organization. Negotiating with a fractured party is extra difficult, as it increases the probability that the peace process will be spoiled, which may be a factor explaining the failure of the ceasefire agreement (Walch 2014). In 1986, Carlos Pizarro became the new leader of the organization. He managed to “re-establish a common policy and a political project aimed at peace. Pizarro suspected that recovering M-19’s political prominence was directly dependent on a clear drive for peace, and he took that direction resolutely” (Durán et al. 2008, p. 17).

When the final peace agreement was signed, commitment problems still had not been fully solved. On the one hand, the M-19 could not guarantee that all of their combatants would demobilize with them. On the other hand, the Colombian government could not guarantee that they would stick to the commitments they had agreed on in the peace agreement. However, there did exist a number of factors that mitigated the commitment problems experienced by the two parties.

Important for the government was that many other armed groups were still active at the time of the agreement with M-19. As such, they had a clear incentive to prove themselves as a

(33)

32 trustworthy negotiating partner. They even hoped to attract other armed groups into signing a peace deal as well, with the promise of participation in the constituent assembly. This promise of a constituent assembly would also have been a costly promise to break politically.

For the M-19 it played a role that their military capabilities had already declined significantly. Because of this, there was already less bargaining power they were giving up by signing the peace agreement. Moreover, the fact that they had already been promoting a democratic solution to the conflict since 1979 provided a clear sign to the Colombian government that they were actually committed to this solution.

Security dilemma

The failure of the ceasefire agreement in 1985 can partially be attributed to the lacking ability of the state to guarantee the security of M-19’s leaders. Prior to the establishment of the ceasefire agreement, one of the most prominent leaders of the M-19 was killed. About a year later, Antonio Navarro Wolf, who was leading the negotiation process with the Colombian government at the time, survived a grenade attack. Shortly after this attempt, the leadership of M-19 declared that the government had broken the ceasefire agreement and retreated into the mountains (Chernick 1988).

By the time the M-19 signed the final peace agreement, the security dilemma was far from fully solved. The failed experience of the FARC’s transition into political life was still freshly engraved in the memories of the demobilizing combatants. Between 1989 and 2005, 17.8% of the demobilised combatants of the M-19 were killed, a total of 160 combatants (Villarraga 2006, p. 80-81). Even though less than the majority of these incidents were politically motivated, it still shows that the security environment can hardly be described as safe for the demobilizing combatants. This shows that despite the existence of such an environment, an armed group may still decide to collectively demobilize.

Conflict intensity

The Mutually Hurting Stalemate only began to emerge after the siege of the Palace of Justice in 1985, and was the result of a number of developments that occurred around the same time. First, the M-19 had suffered significant military losses in this attack and in the subsequent period as a result of state repression. Second, M-19 was perceived by the government as the most dangerous of the armed groups because of their ability to convert their political vision and their actions into a way of pressuring the government (Messaoudi Rodriguez 2011). Third,

What drives armed groups in peace processes? : a case study of collective demobilization in Colombia between 1982-2016