REBELS AGAINST MINES? EXPLAINING REBEL RESTRAINT ON LANDMINE USE

(1)

REBELS AGAINST MINES?

EXPLAINING REBEL RESTRAINT ON LANDMINE USE

HENRIQUE SINICIATO TERRA GARBINO Master’s Thesis

Spring 2019

Department of Peace and Conflict Research, Uppsala University

Supervisor: Dr. Chiara Ruffa

(2)

Abstract

Instead of state governments, rebel groups have become the most prolific landmine users. However, rebels display significant variation in the way they restrict the use of landmines. While some armed groups seek to limit their effects to government forces, some indiscriminately lay mines irrespective of collateral damage, and others directly target civilians with landmines. Furthermore, some rebels have renounced the use of anti-personnel mines and engage in mine action. In this thesis, I seek to explain this empirical variation. I argue that civilian victimisation caused by landmines creates significant legitimacy costs to rebels, thus increasing incentives to exercise restraint. It follows that the more legitimacy-seeking behaviour rebels display, the more likely they are to exercise restraint on landmine use. I test this hypothesis in a structured focused comparison of three rebel groups from the Philippines. The case studies support the hypothesis, as legitimacy-seeking behaviour had a positive effect on restraint on landmine use in the selected groups. This thesis contributes to the field mainly in two ways. First, I conceptualise and measure restraint on landmine use, which had been neglected in previous studies. Second, I develop a theoretical argument specifically explaining variation in restraint on landmine use.

Keywords: non-state armed actor, legitimacy, restraint, landmine use, Philippines

(3)

Acknowledgments

First of all, thank you, Masumi Honda, for your insights and suggesting the theme of this research during one of many informal yet always inspiring conversations.

My sincerest gratitude to Chiara Ruffa, for your constant support, invaluable advice, and genuine interest in this research. Thank you for sharing your knowledge and experience and for pushing me always further.

Special thanks to Alfredo Lubang, Miriam Coronel-Ferrer and Soliman Santos, for your support and inspiring engagement in promoting a landmine-free world, mostly through the Philippine Campaign to Ban Landmines. Your enthusiasm was a key component of my motivation throughout this thesis.

Thank you, former and current Geneva Call staff Anki Sjöberg, Christopher Rush and Pascal Bongard, for taking the time to discuss the framework of this research in its early stages.

Thanks to Carlos Bittencourt, João Maurício Valdetaro, Marthe Hiev Hamidi Abd Abad, and Yordanos Melake for reviewing and commenting this thesis’s drafts at different stages. Special thanks to Sofia Jarvis, who followed me through every step of this research and provided specific and detailed feedback as well as motivation to carry through the whole process.

I would also like to immensely thank Rotary International for making this experience possible through the promotion of the Rotary Peace Fellowship, without which I would not be able to attend this master’s programme.

Finally, I would like to refer to Armin Köhlt for inspiring the title of this master’s thesis through his article “Rebellen gegen Minen”, published in the Swiss newspaper Die Wochenzeitung, edition 42, on 16 October 2008 (Köhlt 2008).

(4)

List of abbreviations

AFP Armed Forces of the Philippines

AP Anti-personnel

APM Anti-personnel mine

AT Anti-tank

ATM Anti-tank mine

ASG Abu Sayyaf Group

AV Anti-vehicle

BIAF Bangsamoro Islamic Armed Forces

CISR Center for International Stabilization and Recovery CMC Cluster Munition Coalition

CPP Communist Party of the Philippines

GC Geneva Call

GRP Government of the Republic of the Philippines

HRW Human Rights Watch

ICBL International Campaign to Ban Landmines ICG International Crisis Group

ICG-SPPP International Contact Group for the Southern Philippines Peace Process ICRC International Committee of the Red Cross

IEA Islamic Emirate of Afghanistan (the Taliban) IED Improvised explosive device

IHL International Humanitarian Law IMT International Monitoring Team IRRC International Review of the Red Cross

IS Islamic State

JMU James Madison University MILF Moro Islamic Liberation Front MNLF Moro National Liberation Front

NDFP National Democratic Front of the Philippines NGO Non-governmental organisation

(7)

NPA New People’s Army NSA Non-state actor NSAD Non-State Actor Data NSAG Non-state armed group

NSAWG Non-State Actor Working Group RGIL Rebel Groups and International Law

RPM-P Rebolusyonaryong Partido ng Manggagawa-Pilipinas (Revolutionary Workers’ Party of the Philippines)

SCBL Swiss Campaign to Ban Landmines

SPLM/A Sudan People’s Liberation Movement / Army UCDP Uppsala Conflict Data Program

UN United Nations

USA United States of America USD United States dollar

(8)

List of figures, graphs, and tables

Figure 1. Contextual factors influencing reduction in landmine use. ... 25

Figure 2. Influencing factors for restraint on landmine use. ... 25

Figure 3. Causal chain. ... 29

Table 1. Operationalisation of the independent variable: legitimacy-seeking behaviour. ... 37

Table 2. Operationalisation of the dependent variable: restraint on landmine use. ... 38

Table 3. MILF’s legitimacy-seeking behaviour. ... 45

Table 4. MILF’s restraint on landmine use. ... 48

Table 5. ASG’s legitimacy-seeking behaviour. ... 51

Table 6. ASG’s restraint on landmine use. ... 52

Table 7. NPA’s legitimacy-seeking behaviour. ... 56

Table 8. NPA’s restraint on landmine use. ... 58

Table 9. Legitimacy-seeking behaviour ranked per non-state armed group. ... 66

Table 10. Restraint on landmine use per non-state armed actor. ... 67

Table 12. Summary of extended analysis. ... 68

Table 13. General data on the non-state armed groups. ... 90

Table 14. General data on dyadic conflicts. ... 91

Table 15. General data on rebel compliance to international norms. ... 91

Table 16. General data on landmine use. ... 91

Table 17. Conceptualisation of the variation of restraint on landmine use. ... 92

Table 18. Attributed scores to observable variation on variables of interest. ... 93

Table 19. Minimum and maximum variation of the dependent variable. ... 93

Table 20. Minimum and maximum variation of the independent variable. ... 93

Table 21. Scale of overall level of variables of interest. ... 94

Table 22. Summary of assigned numeric values observed for the variables of interest. ... 95

(9)

A landmine is a perfect soldier – ever courageous, never sleeps, never misses.

Commander Saloth Sar (better known as Pol Pot), Khmer Rouge (Oldale 2012)

We used landmines more than the Government of Sudan and are paying the price for this now. We are victims of our own mines.

Commander Edward Lino, SPLM/A (GC 2004, 55)

(10)

1 Introduction

Landmines threaten civilians both during and long after conflicts have ended. According to the Landmine Monitor, more than 122,000 people have fallen victim of landmines only in the last 20 years.

Apart from their direct physical and psychological effects, landmines are responsible for yet another wide range of harms. They restrict access to basic resources such as food and water, limit the use of key infrastructure, and both force and restrict migration. Fortunately, new landmine contamination is limited as a growing number of states have signed and ratified relevant international treaties restricting the use of landmines and other similar devices¹. Nonetheless, landmines remain a root cause of death and human suffering in more than 50 countries, such as Afghanistan, Iraq, Libya, Myanmar, Nigeria, Pakistan, Syria, Ukraine, and Yemen (ICBL-CMC 2018).

Instead of national governments, however, non-state armed groups (NSAGs)² have become the most frequent users of landmines and the main drivers of new contamination. An initial assessment by Geneva Call (GC) reported that, in 2005, at least 60 rebel groups had used mines in 24 countries across five geographic regions, namely sub-Saharan Africa, Asia, Europe, Latin America and the Middle East.

Often portrayed as the “perfect soldier” due to their low cost, easy availability and high lethality, landmines became the weapon of choice of many rebel groups (Sjöberg 2004, 2005, 2006b). In contrast with state governments, rebels have considerably less incentives to comply with existing humanitarian norms. Engaging them in restricting or renouncing the use of landmines remains one of the most pressing practical obstacles towards a mine-free world.

However, anecdotal evidence suggests that the use of landmines and other explosive devices vary significantly in the level of restraint exercised by non-state armed groups. On one side of the spectrum, some rebel groups indiscriminately lay mines irrespective of the potential collateral damage to civilians (Sjöberg 2005), some even directly target children with booby-trapped toys (e.g. Holmes and van Heerden 2016), while others use landmines but restrict themselves to command-detonated anti- vehicles explosive devices against government forces. On the other side of the spectrum, some non- state armed groups have also committed to the total ban on landmines, by unconditionally renouncing the use of anti-personnel mines and similar victim-activated explosive devices. Even further, many

1 In recent years, only the government of Myanmar and Syria have planted new anti-personnel mines (ICBL-CMC 2017, 2018).

(11)

NSAGs even engage in mine action activities in one way or another (Sjöberg 2006a). The wide difference in rebel behaviour towards landmine use raises the question: what explains variation in the level of restraint of non-state armed groups on the use of landmines?

Even though previous literature has developed relevant theories on rebel restraint and compliance, landmine use has so far been left mostly unaddressed. Building on the legitimacy-based theory of rebel compliance (Jo 2015), I argue that civilian victimisation caused by landmine use creates significant costs to rebel groups by compromising their legitimacy, thus increasing incentives to exercise restraint.

In short, the proposed theoretical argument leads to the hypothesis that the more legitimacy-seeking behaviour rebel groups display, the more likely they are to exercise restraint on landmine use. This hypothesis is then assessed in a structured focused comparison of three cases drawn from conflict contexts in the Philippines, namely the Moro Islamic Liberation Front (MILF), the Abu Sayyaf Group (ASG), and the New People’s Army (NPA). This thesis found support to the theoretical argument, that is, legitimacy-seeking behaviour had a positive effect on rebel restraint on landmine use. An additional finding also suggests that not only the overall level of restraint increased with the rise of legitimacy- seeking behaviour across the cases, but it also qualitatively varied depending on which audience the armed group was mostly reliant on. The cases of MILF and NPA also show support to the causal mechanism, further suggesting that rebel groups learn the benefits of limiting the use of landmines and appropriate restraint and compliance into their own norms and culture. Nevertheless, insights into the decision-making processes of the armed groups remain a challenge.

By exploring a specific and underresearched area, this thesis contributes to the field of peace and conflict studies in four ways. First, based on technical and strategic aspects of landmine use, it conceptualises the dependent variable restraint on landmine use, which had been neglected in earlier studies. Second, it expands previous theories of restraint and compliance relating them to landmine use. Third, by doing so, it maps different influencing factors of rebel restraint on landmine use. Finally, this thesis develops its own theoretical argument explaining variation in rebel restraint on landmine use.

This paper aims at enhancing the understanding of different incentives and dynamics at play on NSAGs’ behaviour. Insights on this field could inform government authorities, civil-society organisations as well as humanitarian and mine action advocacy organisations engaging with non-state armed groups. More broadly, this research aims at contributing to the efforts of the International

(12)

Campaign to Ban Landmines and like-minded organisations in their struggle to limit the indiscriminate and unnecessary suffering caused by explosive hazards.

This thesis has seven sections. After this introduction (Section 1), I will first present the logic of restraint on landmine use (Section 2). Next, I will argue for the centrality of legitimacy in influencing rebel restraint behaviour towards landmine use (Section 3), followed by a description of the methodology used in this thesis (Section 4). Thereafter, I present the empirical data on legitimacy- seeking behaviour and restraint on landmine use under three case studies (Section 5). The following section includes both case-specific and within-case analyses and seeks to find support or refute the suggested hypothesis (Section 6). Finally, I provide an overview of the main findings and their implications for practice, as well as possible pathways for future research (Section 7).

(13)

2 Understanding the logic of restraint on landmine use

The aim of this section is threefold. First, it situates this thesis within the literature on rebel restraint and compliance with international norms, thus identifying a significant research gap in explaining restraint on landmine use. Next, restraint on landmine use is conceptualised based on the effects of mines on civilians depending on different patterns of use. Finally, I present an original typology and compilation of possible roots of restraint on landmine use, based on previous literature as well as practitioners’ inputs. Those steps are necessary to understand the logic of restraint on landmine use and then build my own argument on the centrality of legitimacy.

2.1 Literature review: rebel restraint and compliance

In the following subsections I will briefly describe the main theories on rebel restraint and compliance, highlighting the need for specific propositions explaining restraint on landmine use.

2.1.1 Previous literature

Significant academic attention has been devoted to the study of violence against civilians, more precisely on the causes and dynamics of one-sided political violence against civilians. However, scholars have given much less emphasis on explaining the cases when violence against civilians does not happen, i.e. cases of restraint and compliance to humanitarian norms (Straus 2012).

Throughout this paper, restraint is defined as deliberate actions limiting the use of violence (ICRC 2018), while compliance takes the meaning of adherence to humanitarian norms, more specifically International Humanitarian Law (IHL) (Jo 2015). Because IHL explicitly calls for restraint in the use of violence, it is often hard to distinguish the two concepts both in practice and theoretically.

Therefore, literature on both concepts will be addressed in this section. Nonetheless, it is worth noting that restraint can be exercised without compliance to humanitarian norms, such as when it is addressed only to specific groups.

Following the typology suggested by Straus (2012), restraint may stem from different micro-, meso-, and macro-level sources. Perhaps the most intuitive source of restraint – and relevant for Straus’s three levels – is compliance with international human rights and humanitarian norms. In this context, three main factors influence restraint, namely knowledge of the law, training in the law, or threat of

(14)

punishment under the law (ICRC 2018). While efficiency of IHL in restraining violence remains debatable (e.g. Jenks and Acquaviva 2014), Hathaway finds that international norms are more effective in restraining the use of violence in the presence of political competition, independence judiciary and private media (2002). Similarly, Jo (2015) and Stanton (2016) argue that rebel groups are more likely to comply with international norms when they are accountable to their constituencies as well as domestic and international actors.

Particularly relevant for understanding leadership decision-making processes, micro- or individual- level theories of restraint focus on values and morals, as well as social psychological processes, that limit violence. As a baseline, Collins argue that violence runs against the “basic mechanisms of emotional entertainment and interactional solidarity” (2008, 25). In his study about Nazi supporters, Monroe adds that individual support to violent actions is a function of personal moral choices shaped by self-identity (2011). In general, violence can be very costly to individual interests by inciting revenge and disrupting mutual-benefit systems (Straus 2012). Several studies point that individual violence is most often committed in situations of high risk and fear (e.g. Collins 2008; Straus 2006), or when faced with peer or vertical pressure (e.g. Milgram 1974; Zimbardo 2007; Browning 1998; Costa and Kahn 2008).

Meso-level sources of restraint follow a more institutionalist approach and focus on group dynamics.

As such, a recurrent factor emphasised in the literature on restraint is the role of civilians in shaping armed actors’ behaviour. In their work on genocide, both Helen Fein and Timothy Longman argue that civil-society organisations, particularly religious institutions, can be instrumental in either facilitating or restricting violence, though they were not capable of supressing it neither in Nazi- occupied Europe (Fein 1984) nor in Rwanda (Longman 2011). Robinson (2009) shows that non- governmental organisations can offer a platform to de-escalate violence, mainly through information exchange between national and international actors. Dependence on material support from local communities, as argued by Weinstein (2007), is also a source of restraint. In comparison to what he calls “opportunistic rebellions”, “activist” rebel groups depend on cooperative interaction with civilians for material support, such as provisions, shelter, and recruitment, and therefore have more incentives for restraint. Similarly, Wilkinson (2004) found that dependence on political support (i.e.

votes) from civilians creates incentives to protect the group’s constituency. International actors, such as peacekeeping forces, have also shown to be effective at limiting violence against civilians mainly by

(15)

increasing the costs of violence (e.g. Fjelde et al. 2018; Fortna 2008; Hegre et al. 2015; Hultman et al.

2013).

On the macro-level, studies in the realm of political sciences emphasize the costs of repressive violence (e.g. Robinson and Acemoglu 2006). Indiscriminate violence may trigger community, domestic and international condemnation and influence the level of support to violent groups. Mass-scale violence against civilians also entails opportunity costs, as significant resources have to be diverted from other efforts such as offensive campaigns against a military opponent (Straus 2012). Less obvious, however, are indirect costs of large-scale violence, such as decreased mutually-beneficial exchanges between groups (Chirot and McCauley 2010), impact on international trade (Harff 2003) or general economic disruption of economic activities (Straus 2012). Democratic political institutions may be an additional macro-level source of restraint. Democratic institutions, the argument goes, should be able to restraint the executive power in its pursuit of violent strategies (Horowitz 1976; Rummel 1994).

2.1.2 Explaining restraint on landmine use: a research gap

Previous scholarship has addressed the logic of violence and restraint for different patterns of violence against civilians, such as indiscriminate violence, genocide and ethnic cleansing, and gender-based and sexual violence. Similarly, scholars have concentrated in explaining compliance to humanitarian norms, focusing on, among other topics, child soldiering and the protection of prisoners of war, aid workers and other categories of non-combatants. While some scholars have developed general theoretical frameworks that could be generalized to other contexts, to my knowledge, only a small number of authors have just recently developed theoretical explanations of rebel restraint on landmine use (Gleditsch et al. 2018; Fazal and Konaev 2019). Those authors, however meaningfully contributing to the scarce academic literature on the topic, take a rather narrow measure of restraint as commitment and compliance to a total ban on landmines.

Landmines are inherently indiscriminate weapons which remain active long after conflicts have ended.

By design, anti-personnel landmines aim not to kill but to maim the enemy, so that the wounded and agonizing soldier would further consume the enemy’s resources and decrease their morale. Coupled with its secretive and unpredictable nature, landmines are highly effective in creating a permanent condition of uncertainty and fear. Whether they are placed to directly target civilians or as the result of the fighting between the warring parties, landmines can cause significant harm to civilians and lead

(16)

to disruption in the social fabric of affected communities. In addition to the physical harm leading to death and permanent disabilities to survivors, psychological trauma, fear, and stress are widespread in mine-affected populations. Communities also suffer collectively by restricted access to livelihoods, key infrastructure, water sources, and either forced to move or impeded to return to their homes because of not only the actual presence of landmines, but also the fear of their presence (UN 2018b).

Given its nature, landmine use considerably differs from other forms of violence against civilians, such as indiscriminate or sexual violence. First, the use of landmines entails significant group capacity mainly in terms of logistic and coordination capabilities. It follows that landmine use should reflect some strategic, rather than opportunistic, reasoning. Second, due to their static nature, landmines mostly bring about geographically localized effects. In relation to violence against civilians, it means that landmines are more likely to affect specific groups of people depending on where they are placed.

Third, most landmines remain active long after conflicts have ended, when virtually all mine victims are non-combatants. Accounting for this longstanding and usually delayed effect on civilians requires long-term perspective and strategic thinking. Fourth, taking all the previous points into consideration, civilians may be either the direct target of mines or the collateral damage of actions against the other warring parties; however, even in the latter case, the armed groups should have accepted the risk of civilians falling victims of their mines.

Having said that, existing theories on restraint and compliance need to be adapted or reframed when examining landmine use. For this purpose, however, it is first necessary to conceptualise what restraint on landmine use is and how it can vary.

2.2 Conceptualising restraint on landmine use

As a concept, restraint on landmine use can be divided in two components. First, landmine use reflects how landmines are employed by NSAGs. Landmine use varies in different categories, namely in its strategy, type of device, trigger mechanism, location, frequency, and record-keeping, marking and information-sharing, as it will be further discussed below. Second, restraint is understood throughout this thesis as deliberate behaviour to restrict the use of violence. Civilians are thus victimised by unrestrained violence either by direct one-sided violence or by the collateral damage of the conflict. Therefore, in this study, the level of restraint on landmine use should vary depending on how landmine use reflects restraint behaviour.

(17)

2.2.1 Strategy

The reasons behind landmine use may vary widely depending on the conflict context. Factors such as the level of power asymmetry, the phase of the conflict, the extent of territorial control, the availability of landmines, and the knowledge in producing improvised explosive devices may all influence the strategy behind the use of landmines. Nonetheless, a 2005 Geneva Call report identified four main strategies guiding mine use, namely defensive, offensive, economic gain and nuisance mining (Sjöberg 2005).

In defensive strategies, landmines aim to deter an enemy attack and restrict access to particular areas or routes of military value. Among non-state armed groups, landmines are usually used for defensive purposes in the protection of camps and arm caches, or for slowing down the movement of enemy troops. Mines may also be laid following a defensive rationale for the protection of group’s constituency, family members or key individuals (Sjöberg 2005). Landmines can also serve an offensive strategy, when their end goal is to kill or maim the opposing force. They also serve an offensive function when laid as part of an offensive strategy, such as to block escape routes during ambushes and counterattacks, or in direct targeting of government forces or individuals (Sjöberg 2005). When employed under an economic gain strategy, mines do not serve any direct military purpose, but economic interests. Landmines are often laid to protect an important source of revenue such as coca crops in Latin America or diamond and gold mines in sub-Saharan Africa. However, in some cases, NSAGs may use landmines to directly extract revenues from the population, such as charging road tolls or extorsion fees (Sjöberg 2005). Other types of mine use that serve no direct military or economic purpose are sometimes labelled nuisance mining. This strategy has been used to disrupt access to and the rebuilding of key infrastructure. Landmine use that is aimed deliberately at civilians in order to empty territories, deny use of basic facilities, displace communities, isolate a region, or simply spread terror also falls under this category (Sjöberg 2005).

In the spectrum of level of restraint, it follows that the offensive use of landmines represents the least harmful strategy towards civilians, followed by defensive, economic gain and nuisance mining.

2.2.2 Type of device

There are several possible categories used to identify landmines. Following the Ottawa Convention’s effects-based definition (UN 1997), mines may serve different purposes depending on their main target. The most common type of landmine are anti-personnel (AP) and anti-vehicle (AV, also referred to

(18)

as “anti-tank” or “AT” mines) and are respectively designed to be exploded by the presence, proximity or contact of a person or a vehicle. The so-called booby traps, usually put in the same category of anti- personnel landmines, are explosive devices disguised as otherwise harmless objects.

Although other types of mine exist, in this study I will limit the classification of the type of device to anti-personnel, anti-vehicle mines and booby traps, which are the focus of the main treaties restricting landmine use and are arguably more disruptive to civilian life than other types. Whether the device is factory-made or improvised will not be taken into account when discussing restraint on landmine use, as it is more likely to be a reflex of logistic capabilities and access to landmines than a strategic decision to limit their effects on civilians.

In this regard, it is implied that higher levels of restraint are exercised when rebel groups use anti- vehicle landmines, followed by anti-personnel, and lastly booby traps.

2.2.3 Trigger mechanisms

Mines can be activated by a wide variety of trigger mechanisms, such as pressure, pull, tension release, pressure release, vibration, magnetic, frequency, audio frequency, delayed action, electrical, or a combination of those (USA 1959). In relation to civilian harm, however, what matters most is whether the explosive device is victim-activated or command-detonated.

Regardless of the exact type of trigger mechanism, civilians face significantly more risk if the device is victim-activated, that is, if the device is designed to detonate by the victim only, without any external action. Command-detonated explosive devices, on the other hand, are monitored and set off by an operator at a given time, thus avoiding unnecessary collateral damage (USA 1959). That is not to say that command-detonated explosive devices present no risk whatsoever to civilians. First, they can be used to directly target civilians, for example, in terrorist attacks. Second, especially in the case of improvised devices, if they fail to detonate, the devices are left as an explosive hazard threatening the population (GC 2011).

Restraint on landmine use is thus reflected on the explosive device’s trigger mechanism, where the use of command-detonated devices represents a higher level of restraint in comparison with victim- activated devices.

(19)

2.2.4 Location

The location of landmines depends mostly in their strategic use. However, in comparison to government forces, non-state armed groups often have less capacity to lay large quantities of mines, and, instead of large and coherent minefields, NSAGs tend to place mines in strategic places (GC 2007b). Rebel groups are also more likely to deploy landmines more indiscriminately near civilian- dense areas (Kurklick and Wall 2010). Conversely, landmines resulting from interstate conflicts are more likely placed along national borders in more restricted areas, thus reducing their humanitarian impact (GC 2007b). It follows that whether landmines are laid in areas with higher or lower risk to civilians can thus indicate different levels of restraint by rebel groups.

2.2.5 Frequency

The frequency in use of landmines varies significantly among non-state armed groups. While some employ mines and other explosive devices as their weapon of choice, others use them only sporadically, given a specific “need” or context in the conflict (Sjöberg 2005). Again, frequency may be subject not to specific strategy or policy, but to contextual factors, such as group capacity, access to landmines, conflict dynamics, among others. Nonetheless, restraint on landmine use can also be demonstrated by the frequency rebel groups lay mines or other similar explosive devices. It follows that the lower the frequency of use, the higher the level of restraint.

2.2.6 Record-keeping, marking and information-sharing

When compared to professional militaries, non-state armed groups are less likely to follow international marking standards³ for their minefields (Sjöberg 2006a). For civilians, this means increased risk in inadvertently walking through minefields and increased costs in future mine clearance.

Likewise, rebel minefields usually do not follow conventional patterns, as the mines are randomly laid (GC 2007b; Kurklick and Wall 2010). However, according to a Geneva Call report, at least 30 non- state armed groups had engaged in 2006 in some kind of information-sharing or mine risk education to affected communities (Sjöberg 2006a). Therefore, restraint on landmine use should be reflected

(20)

also on whether minefields and mined areas are marked, and on how information about them is recorded and shared.

2.2.7 Direct restraint

In addition to the restraint in relation to the use of landmines, direct forms of restraint can also be found on the non-use of landmines. Direct restraint can be exercised by rebel groups in codes of conduct, internal policies, trainings, and doctrines, as well as in unilateral declarations, ceasefire arrangements or peace agreements. As an example, the Colombian National Liberation Army’s code of conduct explicitly mention the duty to inform the civilian population of the location of mined areas (Rojas 1998; IRRC 2011); and the Libyan National Transition Council’s internal procedures protects detainees from being used in mine clearance activities (IRRC 2011). Conversely, some rebel groups might not formally restrict their use of landmines but do so in practice. Other groups, however, might use landmines and other explosive devices in a virtually unrestrained way, as in terrorist tactics.

2.3 Explaining variation in restraint on landmine use

Current theories on restraint, i.e. the behaviour that indicates deliberate action to limit the use of violence, range from a rationalist to a more sociological approach. On the one hand, restraint may be the product of a rational examination of different economic, political, and military interests. In this calculation, armed groups consider factors such as their self-image as perceived by their constituency and other stakeholders (e.g. the international community) as well as the military advantage that the use of certain weapons or tactics may provide them (Bangerter 2011). On the other hand, restraint may be influenced by organisational factors, such as military culture, and both formal and informal socialisation mechanisms (Bell 2016). Likewise, recent research presented by the International Committee of the Red Cross found that NSAGs’ behaviour towards violence or restraint is the product of their sources of authority, beliefs, traditions and the group members themselves (ICRC 2018). Finally, contextual factors could lead to a lesser exercise of violence, but not necessarily mean genuine restraint. Because this study is ultimately intended to understand the reasons of restraint, it is important to point what restraint is not.

Accordingly, this section will first discuss contextual factors that may lead to decreased landmine use and then, mirroring the two main approaches, present potential restraint factors clustered as strategic and

(21)

moral incentives. This section is organised under an original typology developed for this study, based on different theories on restraint and compliance adapted to landmine use, as well as insights and intuitions drawn from practitioners.

2.3.1 Contextual factors

Not all non-state armed groups are able to employ landmines and not all reduction in landmine use necessarily means genuine restraint. A common example of this is the seasonal use of landmines, which are only seldom laid during winter in case of frozen soil or heavy snowfall (e.g. HRW 1997).

Furthermore, as previously argued, the systematic use of landmines requires significant logistic capability and group cohesion; therefore, a non-state armed group with decreased group capacity could also display reduced use of landmines (ICRC 2018). Likewise, decreased access to landmines and other explosive components, as well as technical expertise in production of handmade mines or improvised explosive devices (IEDs), will limit landmine use. Finally, landmine use may be reduced due to evolving conflict dynamics, as non-state actors experiencing major victories are more likely to reduce landmine use (Sjöberg 2005). Unrestrained behaviour may also be subject to other dynamics, even if the rebel leadership is committed to limit the use of violence. Reasons for unrestrained behaviour and non-compliance include conflicting military training and doctrine (Carswell 2014), absence of political training (Green 2013, 2016, 2017; Oppenheim and Weintraub 2017), and principal-agent problems (Hooper 2004).

2.3.2 Strategic incentives

Although seemingly counterintuitive, landmine use may inflict significant military and economic costs on rebels. First, rebels are constantly victimized by their own mines, either during the production of improvised devices, when laying landmines, or unwillingly activating them after they have been laid.

Own casualties are arguably largely underreported due to the rebels’ interests in portraying the image of a professional and cohesive group. However, the National Democratic Front in Myanmar, for example, stated that up to 80% of its handmade mine manufacturers died during when assembling improvised landmines (Moser-Puangsuwan 2001). Likewise, the Sudan People’s Liberation Movement’s Army (SPLM/A) is believed to have suffered significant casualties from their own mines

(22)

(GC 2004). In those cases, exercising restraint on landmine use is a matter of safety and morale for the rebel group’s own ranks, continuity of operations, and the armed group’s very survival.

Second, particularly in the cases where non-state armed groups hold control of territory, using landmines in the land rebels are fighting for entails an inherent contradiction, as mine contamination and future mine clearance might be excessively costly (McGrath 2001). Furthermore, non-state armed groups may be economically dependent on the revenues of the land, in which case denying access to it with landmines would decrease their revenues. Thus, it is expected that rebels would avoid contaminating productive land in their own territory.

A second category of strategic incentives to restraint are political and reputational costs. Landmine use can influence how the non-state armed group is perceived by its constituency, as well as other domestic and international audiences. Civilian support has long been considered a central determinant of civil war outcomes (Duyvesteyn 2018), as civilians provide recruits, food, information, and safe hiding places. It follows that dependence on civilian support creates restraint on the level of one-sided violence (Weinstein 2007) as well as incentives to protect the armed group’s constituency (Wilkinson 2004). Under a similar logic, de la Calle (2017) argues that civilian victimisation is driven by rebel strength, in the sense that weaker rebels would seek civilian support.

Furthermore, local communities are particularly vulnerable to landmines and other explosive devices.

Especially in rebel-controlled areas, civilians have less access to medical facilities, thus increasing the lethality of mine accidents (Walker 2001). A NSAG in Myanmar, for instance, has allegedly changed their mine use policy after realizing that up to 30 percent of the mine victims came from the rebels’

own ethnic group (Moser-Puangsuwan 2001). As a direct consequence of victimisation, individual and collective traumas, lack of access to livelihoods, and loss of property, local communities could decrease support to the armed group (McGrath 2001). It is possible that the decreased support is particularly targeted to the NSAG’s landmine policy and not the group as a whole, demanding the group to restrict their use of landmines and demine certain areas, as for example observed in Colombia (GC et al. 2009) and Senegal (GC and APRAN-SDP 2010).

Civilian victimisation also entails significant reputational costs to other domestic and international audiences, particularly towards “human rights-conscious audiences” (Jo 2015). Abiding to a shared legal framework – such as international law – plays in favour of rebel groups’ perceived political legitimacy. Exercising restraint on landmine use, therefore, entails compliance to international (and

(23)

sometimes domestic) law (Coronel-Ferrer 2001; Kurklick and Wall 2010) and could increase the likelihood of external support, political participation, and leverage in negotiations. As an example, Herr found that SPLM/A adhered to the landmine ban due to transnational pressure and fear of legitimacy loss (2010).

Likewise, compliance with a coherent legal framework may ensure practical and legal securities to the rebel group, in particular to its leadership, such as granting the legal status of combatants and reciprocity in treatment by government forces (Jo 2015). Accordingly, a recent study led by Kristian Gleditsch shows that decisions to commit to a landmine ban by both governments and NSAGs are mutually dependent. The authors propose a game-theoretic model to understand why both actors would first commit to the ban and, then, adhere to it in practice (Gleditsch et al. 2018). Likewise, Tanisha Fazal and Margarita Kovaev have demonstrated that groups that seek international recognition and have a strong military capacity are more likely to commit to a landmine ban (Fazal and Konaev 2019; Konaev and Fazal 2018). Similar to the previous points raised in this section, their argument focus on reputation costs and benefits of compliance to international norms in comparison to the military utility of landmines and other methods of war.

In the same vein, rebel leaders may fear criminal indictment for ordering the use of anti-personnel landmines. Accordingly, a report has found that rebel groups are more likely to exert restraint on landmine use in countries where the use of landmines and other similar victim-activated explosive devices has been criminalised by domestic law (Sjöberg 2005). The effectiveness of criminal justice in fostering compliance, however, is still debated (e.g. Jenks and Acquaviva 2014).

It is worth noting that the abovementioned strategic incentives are interconnected. Political and reputational costs may have direct impact on material support from both the rebel group’s constituency and domestic and international audiences, thus compounding to military and economic costs. Similarly, decreased military efficiency could lead to decreased political support.

2.3.3 Moral incentives

As suggested by Straus (2012), restraint may also derive from genuine commitment to humanitarian principles, whether they are based on humanitarian norms or the group’s own values, beliefs, and traditions.

(24)

As a baseline, indiscriminate violence is condemned in virtually all cultures, so it is expected that armed groups should avoid unnecessary civilian casualties, unless otherwise justified. Likewise, moral obligation has been found to influence decision-making even over material costs and strategic interests (Kreps and Maxey 2018).

Sanín and Wood (2014) explore the role of ideology in shaping rebel violence and restraint. Specific ideological and religious motivations can further influence increased restraint and compliance to humanitarian norms (Al-Dawoody 2017; ICRC 2018). The Islamic Emirate of Afghanistan (IEA, commonly referred to as the Taliban), for instance, officially considered the use of landmines “an un- Islamic and anti-human act”, which “would be punished in accordance with Islamic Law” (IEA 1998, 1–2). Similarly, the Revolutionary Proletarian Army – Alex Buncayao Brigade and the Revolutionary Workers’ Party of the Philippines issued a joint statement renouncing the use of landmines on ideological grounds:

We believe, therefore, that the destruction of lives and properties, as a consequence of armed conflict, is an antithesis to our desire for a better world. We believe that while we are fighting to achieve full human development and social progress, we must respect the lives of the people and of nature — uphold and promote human rights and protect the environment. So, based on our politico-military orientation, we always plan that there will be no single civilian that will be damaged in our military operations against the state, the oppressor. The use of anti-personnel mines, as a weapon for destruction, has been proven to have been extremely prejudicial to lives and safety (SCBL et al. 2001, 87–88).

Restraint may also be driven by interaction with potential victims. In Colombia, for instance, rebel groups oftentimes voluntarily marked mine areas or engaged in mine clearance to preserve the communities where they operated. Arguably, genuine interest in protecting civilians from the effects of landmines comes from ethnic and family ties, as well as continued interaction with the communities (GC et al. 2009).

Knowledge and acceptance of International Human Rights Law and International Humanitarian Law may also genuinely lead to restraint on landmine use. Practitioners have found that sustained engagement with non-state armed actors, in particular through education and awareness raising on humanitarian norms, constitutes an important step towards restraint on landmine use (Moser- Puangsuwan 2004; Reusse-Decrey 2005). State and non-state armed actors alike often lack long-term perspective on the humanitarian consequences of their own actions, hence the change in behaviour after engaging with human rights and humanitarian norms advocates (Sjöberg 2007). In Myanmar, for instance, the Karen National Union, a rebel group and landmine user, agreed to cooperate with mine

(25)

action organisations after a series of meetings conveying the relevance of international humanitarian law (Moser-Puangsuwan 2004).

2.3.4 Summary of incentives for restraint on landmine use

In short, reduced landmine use does not necessarily mean restraint, as it may be the result of other contextual factors, such as variation in logistics, technical expertise, weather, group cohesion and conflict dynamics (Figure 1).

Figure 1. Contextual factors influencing reduction in landmine use.

Genuine restraint may have different roots. In practice, it is unlikely that restraint is the product of only one influencing factor or that those factors act in isolation from each other. However, for the sake of simplicity, I have clustered the possible influencing factors for restraint on landmine use in strategic and moral incentives in the factors tree below (Figure 2).

Figure 2. Influencing factors for restraint on landmine use.

(26)

In this study, however, I will focus on one specific incentive influencing restraint which, I argue, has the most impact in non-state armed groups, i.e. legitimacy. In the next section I will present and develop the claim of why legitimacy plays a central role in influencing restraint over the other factors.

(27)

3 The centrality of legitimacy-seeking in rebel restraint

This section develops the main theoretical argument of this thesis and then focuses on one part of a broader causal chain, namely the relationship between legitimacy-seeking behaviour and rebel restraint on landmine use. It starts by arguing for the centrality of legitimacy-seeking as well as focusing on the main causal mechanism and its resulting hypothesis. Next, it conceptualises the independent variable influencing restraint on landmine use and explicates the scope conditions and contributions to the literature of the argument.

3.1 Theoretical argument

Among the different sources of restraint presented in the previous section, building on previous theories of rebel restraint and compliance (Fazal and Konaev 2019; Gleditsch et al. 2018; Jo 2015), I argue that the costs resulting from legitimacy loss are particularly significant to rebel decision-making and therefore should play a central role in influencing restraint on landmine use. As argued before, the systematic use of landmines depends on significant group capacity. It follows that landmine use – or restraint on it – is first a strategic choice stemming from the rebel leadership. Hence, to exert restraint, non-state armed groups should strategically decide that the costs of landmine use outweigh its benefits. In other words, the costs resulting from legitimacy loss should be perceived as higher than the military utility of landmines.

Legitimacy can be defined as “a particular quality that is conferred upon a social or political entity by those who are subject to it or part of it, thus granting it authority” (Bellina et al. 2009, 3). Thus, the most relevant “subjects” in understanding the sources of legitimacy for rebel groups are, as argued by Jo (2015), their constituency, domestic audiences, and international audiences. First, the importance of popular support to insurgencies has been widely studied. Rebel groups most often depend on their constituency for both material and non-material support, such as manpower, funding, materiel, intelligence, sanctuary and general compliance and obedience (Arjona 2017; Davis et al. 2012; Kitzen 2017). Second, building on Jo’s argument that “civil wars are more than armed conflicts” (2015, 16), more than purely military means, rebel groups need support from their domestic audiences in order to achieve their political goal, whether it is over territory or government issues. Relevant domestic stakeholders include the state government, political parties, civil society, private sector, among others.

In practice, rebels’ perceived legitimacy allows them to engage in political negotiations, establish

(28)

political alliances, and enhance their public image (Schlichte and Schneckener 2015). Finally, rebel groups engage with international audiences for several reasons. Foreign sponsors can be not only a valuable source of material support, such as funding and weapons, but also political, ideological, and moral support. This is particularly relevant for non-state armed groups aligned with global movements and seeking recognition by and accession to the international community.

Echoing humanitarian and mine action practitioners (e.g. SCBL et al. 2001; ICBL/NSAWG and GC 2003), I argue that civilian victimisation caused by landmines leads to political and reputational costs for rebels, in particular on their perceived legitimacy. The relationship between civilian victimisation and decreased legitimacy of armed groups has been well recorded and studied in different contexts, such as Latin America (e.g. Wood 2003; Arjona and Kalyvas 2011) and Eastern Europe (e.g. Petersen 2006), although most often than not in explaining rebel recruitment due to violence perpetrated by government forces. Legitimacy has long been considered a key determinant of civil war outcomes, being central to most rebel and government strategies (Duyvesteyn 2018). In addition, legitimacy- seeking behaviour has shown significant influence on rebel compliance to international norms (Jo 2015).

In short, the effects of decreased legitimacy should spill over onto different other areas, including losses in both material and non-material supportand decreased leverage in political negotiations. Thus, non-state armed groups will more likely perceive a decrease in the utility of landmines and exert restraint on landmine use. According to this argument, restraint is exercised in a direct relation to the armed group’s perceived political and moral legitimacy by its stakeholders. It follows that NSAGs should acknowledge the benefits of exerting restraint on landmine use to avoid related costs and maximize support to their cause.

Although landmines have been traditionally considered an indispensable weapon of war, especially given the power asymmetry between rebels and governments, the effectiveness to achieve their military purpose is largely overestimated. Evidence indicates that, even when used on a massive scale, landmines have usually yielded little or no effect on the outcome of hostilities (ICRC 1997).

Considering that, in practice, landmines are not an essential weapon of war, legitimacy costs stemming from landmine use should decrease the perception of their effectiveness, finally leading armed groups to exercise restraint.

(29)

Implicit in this argument is the assumption that non-state armed groups have an initial perception of landmine effectiveness. This might come from the symbolic status of landmines, but it also follows from logical deduction. Even when disregarding the costs stemming from the effects of landmine use, the process of acquiring, producing, and laying landmines can be significantly costly to rebels and they would not undertake those efforts unless perceived to be in their favour.

It follows that after non-state armed actors do employ landmines, I assume some degree of civilian victimisation as a consequence of landmine use. The effects of landmines on civilians can of course be minimal, but I argue that even the most limited use of landmines – such as the offensive use of command-detonated anti-vehicle mines in unpopulated areas – still presents a threat to civilian life, due to the device’s intrinsic risk of failure and the permanence of an explosive hazard.

It is also expected that decreased legitimacy due to civilian victimisation could have a delayed effect on restraint on landmine use, given the conflict dynamics. Armed groups do not necessarily predict the full extent of the costs of landmine use and only realize it after civilians are victimised by their landmines. Restraint is, the argument follows, a delayed effect of landmine use, meaning that it does not necessarily require a change in legitimacy-seeking behaviour to occur. Figure 3 summarises the causal chain implied in this argument.

Figure 3. Causal chain.

In short, the rest of this thesis will focus on the relationship between the variables of interest, rather than on following the causal mechanism leading to restraint on landmine use. This focus is justified given the exploratory nature of this study and lack of previous research, in addition to time and resource constraints. The main argument of the causal mechanism is that rebel groups experience decreased legitimacy once civilians are victimised by landmine; in turn, the perception of landmines effectiveness is decreased. It follows that legitimacy-seeking groups would attempt to minimise legitimacy costs and thus exercise restraint on landmine use. Deriving from the proposed causal mechanism, I will assess the hypothesis that the more legitimacy-seeking behaviour rebel groups display, the more likely they are to exercise restraint on landmine use.

(30)

The definition of legitimacy-seeking behaviour follows Jo’s conceptualisation (2015). Rebel groups seek legitimacy as a strategic goal to obtain governing authority in local, national, or international politics.

Rebels are also dependent on legitimacy to safeguard whatever material and non-material support is required to pursue their cause. In sum, legitimacy is both the goal and the means of a rebel group and is intrinsically connected to the group’s political goals, relationship with its constituencies, and sources of support. As argued by Jo, it follows that “the link between legitimacy-seeking behaviours and compliance behaviours lies in a rebel group’s governing history and their relationship with core constituencies both at home and abroad” (Jo 2015, 92).

3.2 Scope conditions

This argument should hold valid under specific scope conditions. Central to this thesis is the puzzle of why rebels exercise restraint on landmine use when there are incentives not to. Landmine use is often justified as a necessary weapon, given the military power asymmetry between rebel groups and government forces. Therefore, the suggested mechanism of rebel restraint on landmine use is more likely to work in intrastate conflicts, in comparison with interstate and non-state conflicts.

It has been previously argued that using landmines requires some degree of logistic capabilities, technical expertise, and group cohesion. Following ICRC’s typology for non-state armed groups (2018), mostly centralised – and not decentralised or community-embedded – NSAGs meet those conditions.

Finally, international law has experienced major developments in the previous decades, in particular regarding landmine use. Major benchmarks in this context are the adoption of the 1996 Mines Protocol (UN 1996) and the Ottawa Convention (UN 1997) by most countries. Because unrestrained use of landmines has become highly stigmatised, rebel legitimacy is more likely to be influenced by norm-compliant behaviour in recent conflicts. In short, I argue that the proposed argument is more likely to hold for post-Cold War conflicts, than for earlier conflicts.

3.3 Contribution to the literature

This theoretical argument expands previous analyses of rebel compliance and restraint to the often- overlooked phenomenon of landmine use. Even though landmine use shares many similarities with

(31)

other forms of violence against civilians, it deserves its own focus and theoretical propositions because of its unique nature. Landmine use requires significant group capacity, its collateral damage is assumed, and its effects on civilians are geographically localized, long-term, and often delayed. Moreover, this study further adds nuance to compliance to international norms by looking into other forms of restraint on landmine use. In this regard, the conceptualisation of restraint on landmine use offers a novel baseline for future studies.

(32)

4 Research design

Section 4 describes the research design used for this research. It starts by presenting and justifying the selected methodology and case selection strategy. Next, it operationalises the dependent and independent variables. The section then continues with a presentation of the data sources and structure of analysis to be used in this study.

4.1 Method and case selection strategy

In order to empirically test the suggested hypothesis, this thesis uses the method of structured focused comparison, where both the dependent and independent variables are measured according to a pre- defined set of questions across different cases. This method is appropriate for the current hypothesis given that it restricts the analysis to a particular aspect of interest in a particular time frame. It also allows for comparable data collection and analysis (George and Bennett 2005).

Furthermore, the comparative aspect of structured focused comparison is necessary to measure and analyse variation of restraint on landmine use among different cases. The case selection strategy for this thesis follows the exploratory most-similar design (or method of difference), i.e. cases with similar background conditions, but with different outcomes and unknown causal factors (Gerring 2006;

Gerring and Cojocaru 2016). The choice of a qualitative design over a large-N study is based on the relatively scarce data on landmine use and the exploratory nature of this research. Because this thesis deals with a rather underexplored topic and aims to contribute to theory building, a qualitative design is used to allow for a more detailed analysis and opportunities to refine its theoretical propositions.

4.1.1 Unit of analysis

The unit of analysis of this study is the non-state armed group. In practice, this means that data on the dependent and independent variables will be measured and analysed aggregated up to the group level, and, for comparison purposes, viewed as existing at one single point in time. There are several disadvantages and advantages that arise from this.

To start with, rebel groups are often fluid institutions. Splintering factions, shifting leadership and changing organizational structures are inherent to most non-state armed groups (Jo 2015). While group-level aggregated data does not capture the nuance of different fronts or evolving dynamics, it

(33)

does provide a clearer baseline for comparison for several reasons (Kenny 1996). First, the main interest of this inquire is in group behaviour; second, I do not expect significant within-case variation in legitimacy-seeking behaviour, but rather a lagged effect of civilian victimisation in rebels’ perceived legitimacy; third, as an exploratory study, this research opts for simpler measurement and analysis methods. Nonetheless, case-specific analyses will be done to account for some of this nuance, though it does not allow for cross-case comparison.

It is not in the scope of this study to dive into individual-level explanations of restraint or violence, but rather focus on group-level decisions and collective action. The choice of the group as the unit of analysis thus implies that behaviour is attributed to a collective entity, rather than to its members. This is justified by the fact that decisions on restraint on landmine use are most likely taken by the rebel group leadership and therefore it can be assumed to influence group action.

4.1.2 Conditions

Following the scope conditions of my theoretical argument outlined in the previous section, the population of cases was selected from dyadic conflicts from the post-Cold War period involving one non-state armed group and one government – that is, intrastate conflicts – in which landmines were systematically employed as a method of war.

The post-Cold War period indicates the consolidation, proliferation, and expansion of international humanitarian norms, particularly in relation to violence in internal conflicts. The period experienced a growing number of rebel groups engaging with humanitarian and human rights organisations, as well as the criminal indictment of rebel leaders under new accountability norms (Jo 2015). Moreover, it is during this period that the public awareness rose against anti-personnel landmines and the most relevant legal instruments in limiting or banning their use were established (UN 1996, 1997, 2008).

In accordance with the most-similar case study design, the selected cases should present maximum variation of the dependent variable and minimum variation of background factors (Gerring 2006;

Gerring and Cojocaru 2016). Ideally, the sample of cases should include rebel groups with both low and high levels of restraint on landmine use, which are similar in both contextual factors and strategic and moral incentives other than the main explanatory variable for this research, i.e. legitimacy-seeking behaviour.

(34)

4.1.3 Background factors

In order to maximize the effects of legitimacy-seeking behaviour on restraint on landmine use, other potential restraint sources should be controlled for in the case selection. The first of those are the contextual factors that might lead to the decrease of landmine use, but not necessarily genuine restraint. Those factors are: weather, logistics, technical expertise, group cohesion, and conflict dynamics. While holding the unit of analysis in the group level reduces the effects of weather conditions and evolving conflict dynamics, the selected cases should show sufficient levels of logistic capacity, relevant technical expertise, and group cohesion. Some aspects of conflict dynamics, such as government responses and military action, are also controlled by holding constant the government against the rebel groups fight against.

After realizing that their landmines are causing civilian victimisation, rebel groups could exercise restraint due to moral incentives or in pursuing legal securities. Therefore, the selected cases should present significant levels of one-sided violence or other cases of non-compliance with humanitarian norms, regardless the level of restraint on landmine use, meaning that the groups do not exercise restraint solely based on moral grounds nor for fear of criminal indictment.

The effects of other strategic incentives in restraint on landmine use, such as expected land use and rebels’

own casualties, are harder to isolate due to data collection challenges. First, rebel might have incentives for not laying landmine in areas they currently use or have interest in using in the future. Information on land use and how it may influence the decision of laying mines is barely available; and, although actual and expected land use could be measured and estimated by detailed conflict analyses, it deserves a study of its own. Second, rebels’ own casualties due to landmines are hardly reported by rebels themselves, who have incentives to omit their casualties, in order not to signal weakness to other warring parties. As such, those factors remain as potential alternative explanations for variation in restraint on landmine use.

4.1.4 Population and selected cases

The population of cases, i.e. landmine-user non-state armed groups, was drawn from the compilation of different databases on landmine use, non-state actors, and general conflict data. As a baseline, according to an initial assessment, at least 72 NSAGs were alleged or confirmed landmine users (Sjöberg 2005). Additional relevant data on group and conflict characteristics, rebel compliance and

(35)

landmine use for those groups were taken mainly from the Uppsala Conflict Data Program (UCDP) (Harbom et al. 2008; Pettersson and Eck 2018), the Non-State Actor Data (Cunningham et al. 2009), and the Rebel Groups and International Law database (Jo 2015).

Following a most-similar case selection strategy, I have selected three non-state armed groups active in the Philippines, namely, the Moro Islamic Liberation Front, the Abu Sayyaf Group, and the New People’s Army (see Appendix I). While this design is usually used only with two cases, the groups display a wide range of commonalities in terms of background factors, and an initial assessment shows significant variation on restraint on landmine use. Most significantly, all three cases are centralised non-state armed groups engaged in minor or intermediate conflicts against the Government of the Republic of the Philippines (GRP); they display moderate levels of command and control and low arms procurement and fighting capabilities (Cunningham et al. 2009; Harbom et al. 2008; Pettersson and Eck 2018). They all occur in the territory of the Philippines, which is signatory to the Ottawa Convention and the Mines Protocol (ICBL-CMC 2019; ICRC 2019) and has thus criminalised the indiscriminate use of landmines in its territory.

The cases differ, however, in type of incompatibility, ideology, group size, and relative strength in relation to the GRP. Likewise, all the three cases have a record of one-sided violence or non- compliance to international norms, but to different degrees (Jo 2015). Those differences, though certainly able to influence patterns of landmine use, should be reflected in the groups’ legitimacy- seeking behaviour, and thus indirectly accounted for. Whereas ideology and group size should reflect the relationship with its constituency; the type of incompatibility mirrors its relationship with the government and other domestic actors; and the relative strength of rebel groups is linked to the support received from foreign actors.

Regarding restraint on landmine use, according to an initial assessment (Sjöberg 2005; JMU/CISR 2004), MILF has shown a high level of restraint by committing to a total ban on landmines through Geneva Call’s Deed of Commitment (MILF 2002); NPA displays a moderate level restraint, since it has agreed to respect human rights and humanitarian law (GRP and NDFP 1998), but still justifies the use of command-detonated explosive devices against government (Sjöberg 2005); finally, ASG displays a low level of restraint, since it has not shown any significant effort to minimise civilian casualties, and rather has directly targeted civilians using explosive devices (Sjöberg 2005).

REBELS AGAINST MINES? EXPLAINING REBEL RESTRAINT ON LANDMINE USE