Chapter 1: ‘Ordering the Security Arena’ provides the research framing of this book “The Security Arena in Africa: Local Order-Making in the Central African Republic, Somaliland, and South Sudan” by Tim Glawion



Chapter 1 – Ordering The Security Arena

Tim Glawion, German Institute of Global and Area Studies (GIGA)

The Security Arena In AfricaFrom the Book: The Security Arena in Africa: Local Order-Making in the Central African Republic, Somaliland, and South Sudan, pp. 16 – 57

Publisher: Cambridge University Press

Print publication: 30 January 2020

Published online: 16 January 2020


Chapter 1 – ‘Ordering the Security Arena’ – provides the research framing of this book. First, it discusses relevant academic works on arenas and combines them with the author’s own empirical insights. It defines ‘security’ and describes the two key dimensions of actors and their interactions. Second, descriptive commonalities and differences in ordering practices in the local security arena lead to analytical insights that call for more in-depth investigation – namely, the respective roles and interactions of state, non-state, and international actors; the dividing lines between inner and outer circles in the local arena; and different relations between peripheral local cases and their respective centers. In the methodological section, the chapter explains the selection of these three countries and the local arenas within them, how data was gathered through an explorative mix of methods, and its analysis through process tracing and the Comparative Area Studies approach.



Security, Security arena, Ordering, Order-making, Stable ordering, Fluid ordering, Fragility, Site-intensive methods, Comparative Area Studies

Chapter 1 – Ordering The Security Arena

In late November 2014, a South Sudanese military intelligence officer ‘went missing’ in the village of Garia, near Mundri town. Local security actors organized a joint visit to the village in early December. The situation was tense: what had the intelligence officer been doing in Garia? Was he keeping the peace by investigating the construction of a rebel training camp? Or was this yet another sign of the government’s paranoid repression of the people of Mundri? I was sitting in the acting commissioner’s office while state officials planned a trip to Garia to find out for themselves. They didn’t expect me to understand Arabic and spoke quite freely. Two important things were happening: first, most of the discussions were about finding enough cars and fuel to even make the journey to Garia. Second, a National Security Services (NSS) officer repeatedly walked in and out of the office, fully engaged in the planning. When, on our request, the acting commissioner invited my colleague and me to join them on their trip, the National Security officer intervened: ‘No, of course, they cannot go.’[1]

The local government, chiefs, youths, police, wildlife forces, and national security thus left without us. Most significantly, the army, whose clarifications other stakeholders sought, did not appear as promised. My trip to Mundri was one of the first I undertook during this study and indeed these impressions raised many questions that would guide me throughout my years of research for this book. Who are all these actors in the local arena that come together for the trip to Garia? Who are the actors that are left out and why? Why are there so many complex steps involved in visiting a place that should, in principle, be under constant state authority? Why is the national government’s role so contested in this local arena? And, most importantly, why is the situation so tense and security deteriorating? By traveling to eight further localities – thus a total of nine – I started developing arguments along the difference between an inner and outer circle and of center–periphery relations as well as the forms of ordering that hopefully can provide some answers to these questions.

Recent writings on public authority, areas of limited statehood, and oligopolies of violence have generated an awareness of the need to study security on a local level: they have put security at the heart of the analysis, taken crucial steps away from the country container bias and the state’s monopoly of violence, and have engaged in comparative analysis. However, no studies within these strands have combined the three dimensions of security, at the local level, in comparison at the same time. While other concepts show great merit in their respective fields, I propose the security arena as a concept that can more adequately combine investigating the issue of security, within non-monopolized local contexts, and through a comparative lens. The security arena concept thus grants an entry point to pinning down the issue of security and the actors engaging in this issue.[2] Rather than looking simply at individuals and their actions, the security arena concept calls for the categorization of groups of individuals that comprise actors – such as a rebel group or a local police department – and the ways they relate to one another through myriad interactions, rather than merely in one-off encounters. This shall facilitate the analysis across a multitude of very different localities of how actors interact around the issue of security and, by doing so, engage in creating a political order. But, first, I turn to the merits and limits of existing strands of literature to draw lessons to be heeded for the security arena concept.

Authors who study areas of limited statehood conceptually differentiate statehood from governance.[3] This approach makes areas, rather than countries, the locus of interest. Areas of ‘limited’ statehood can be found even within established states and vice versa: so-called failed states, such as Somalia, can contain areas with some degree of statehood, e.g. Somaliland. Consequently, I will compare different localities across countries, rather than painting whole countries with one brush. The distinction between statehood and governance enables social scientists to analyze the provision of goods even when statehood is limited. In Somalia alone, examples of governance in areas of limited statehood range from indigenous justice provision[4] all the way to health services provided by external actors.[5] However, a differentiation of ‘governance’ from the struggle over a force monopoly – that is, statehood – only seems expedient for issues such as health and education. Who provides security and who holds the force monopoly, however, seem to be intricately linked.

Rather, another strand of literature more accurately describes these political struggles as the production of public authority:[6] some authors observe instances of bottom-up competitions for legitimacy between different power poles,[7] while others conceptualize hybrid orders between the modern state and indigenous customary actors and narratives.[8] Therefore, different actors and symbolic ideas form new institutions that are a mix of what can be called state or non-state and customary or modern.[9] This public authority is contested and re-evaluated in daily struggles over popular legitimacy that can be gained through public services provision and by invoking narratives of state and tradition. The level of security is explicated by the struggle over authority: security declines during processes of exclusion[10] and improves when attributions of public authority become accepted.[11] The public authority literature sets the bar of local knowledge very high, which has led to credibly grounded case studies. This study, however, sets its focus on gaining comparative insights from multiple cases.

The oligopolies of violence literature enables such comparison and still convincingly conceptualizes non-monopolistic settings. An oligopolistic setting is the antithesis to Max Weber’s ideal-type state. First, it opens two of Weber’s three dimensions to analytic debate: the territory and people of concern depend on the research question at hand. Second, rather than the prime dimension – the use of physical coercion – being monopolized, it is split between and competed over by different poles; it is oligopolized: ‘Oligopolies of violence comprise a fluctuating number of partly competing, partly cooperating actors of violence of different quality.’[12] By assuming the rational choice both of beneficiaries and of providers, this model and comparable others then attempt to explain in which economic – fewer extractable resources means more security[13] – and market – the more monopolized the better[14] – contexts security improves or decreases.[15] Rational choice models have their limits, some of which the authors themselves admit. In the security market, paradoxically, more competition leads to less security[16] because the shares in the market diminish and become more competitive.[17] However, this is by far not always the case. More fragmented settings can also mean that no single group has the capability to engage in mass atrocities and the entrance of a strong peacekeeping force can at times halt violence quickly. While the market analogy thus grants a good entry point to analyze the inherently competitive nature of oligopolistic settings and the dangers of cartel formations,[18] the analysis would be hampered by seeing all actors as purely rational, security as a marketable good, and competition as having singular effects.

The underlying assumptions of these different strands only partly allow for the comparative analysis of local security in a non-monopolized political order. Theories of areas of limited statehood call for territorially differentiated intra-state comparison and a clear focus on the good that is provided. While this allows for the focused comparison of certain goods, such as health, I do not find it expedient for the differentiation of security ‘governance’ from the struggle over the force monopoly that is statehood. Writers developing the concept of public authority propose in-depth evaluations of the social conditions and contexts influencing local struggles over order and security but thereby also make small to mid-N comparison – that is, from two to around a dozen cases – difficult to achieve. The oligopolies of violence literature define parameters to investigate and compare different security markets but would reduce perceptions to surveys and comparative analysis to rational choice and market conceptions. Thus, while each strand of literature has particular merits, none of the four fulfills the specific three needs to conduct this analysis.

Drawing from the merits of each strand, the security arena provides a concept through which the researcher can (1) focus on perceived security (and not violence, conflict, or peace), (2) identify and analyze social and political aspects of local oligopolistic political orders, and (3) generalize gained insights on security provision through a comparative design.

Conceptualizing the Security Arena

I combine the issue of security with the actor-oriented concept of an arena[19] to facilitate a process analysis of the perceptions and agency of actors witnessed during my fieldwork. This concept enables each of the nine arenas in three different countries to be viewed in and of themselves, while at the same time granting common lines along which similarities can be unveiled.

Security as the core of the matter is an issue that naturally lends itself to the analysis of processes. A person can create security as well as undermine it. A person can feel secure or insecure as well as varying shades in between. Studying security is therefore inherently broader than studying conflict, as it opens avenues to people’s perceptions,[20] and narrower than studying peace, which is heavily based on normative theory.[21]

The arena is a concept that brings together actors and their interactions around the spatial dimension of the locality under investigation. My comparison across the nine arenas builds on establishing similar knowledge of actors’ characteristics and their interactions around the issue of security throughout the security arenas under scrutiny here. Based on the literature and insights from the field, I define security as the felt durability of physical integrity and insecurity as the felt threat of physical harm. I conclude that a security arena is composed of actors that interact on the issue of physical integrity around a predefined center of study.

In 1995, Emma Rothschild published the article ‘What Is Security?’.[22] In it, she traces the changing historical contexts of the contractual nature of security as the basis of politics and society and claims that, in the new liberal age, security concerns the conditions for personal liberty. In addition to her rich historical testimony, the article shows that security is a fluid concept that is hard to pin down. In fact, authors either define security vaguely[23] or narrow it down to the absence of physical threats to an individual or a societal group.[24] In both cases, many security studies investigate only insecurity and conflate this with the occurrence of conflict.[25] Defining one concept – security – by the absence of a second – conflict – begs the question as to why the author did not just analyze the latter concept from the start.

The ongoing debate on security in political science is driven by scholars who choose a deductive approach, including prominent contributions from the sub-discipline of international relations. Authors have analyzed the creation of security communities,[26] the global privatization of security,[27] and the downward hybridization of state security.[28] Scholars of critical security studies problematize the political manipulation of the concept of security[29] and the discussion of human security proposes a widening of the term ‘security’ to domains beyond the physical.[30] David Baldwin proposed a guideline on developing security research. Two key questions – security for which values and for whom? – and five optional questions should specify the concept of security used for each respective study.[31]

While these recent studies have indeed questioned the state-centric viewpoint in security studies, they do not go far enough: security in the local arena needs to be judged from the bottom up, in and of itself, rather than as a deviation from state-centric norms.[32] In line with a recently proposed vernacular understanding of security,[33] I find that security has an experienced dimension and thus would ask: ‘what makes you feel secure or insecure?’

I ask this question directly to interlocutors in the three countries to discover what matters in people’s lives. A number of commonalities span individuals’ responses across all three countries. First and foremost, statements on security and insecurity, in almost all instances, include the naming of actors: ‘The presence of the army creates a lot of insecurity’,[34] ‘il y a un problème de sécurité qui se pose … des exactions de la part des hommes en armes’,[35] and ‘the peace in here is very well, sometimes youth makes problem, sometimes clan problems’.[36]

While respondents specifically name categories of actors, they are less clear in explaining how these actors impact security. In fact, many replies suggest something more than an immediate threat to physical integrity. Take this definition of security provided by the sub-prefect of Obo (CAR): ‘La sécurité, c’est l’absence de troubles, d’actes qui font la confusion parmi la population; A fuir, à être inquiet de sa vie et sa famille. Vivre en paix, il n’y a rien de plus important que la paix.’[37] This absence of trouble is more than the absence of conflict proposed in the narrow definition of security from other studies. Aside from direct physical harm, trouble causes confusion, displacement, and worry. On the other side of the CAR, in Paoua, youth express themselves similarly: ‘Depuis 2003 il n’y a pas de paix, il y a beaucoup de désordre.’[38] In South Sudan’s Buseri, women describe their feeling of insecurity as ‘people do not feel really secure, not even in the house. You have the feeling that anything can happen any time.’[39] A few thousand miles east, in Hargeisa’s Daami quarter (Somaliland), youth frame their security worries almost identically, as: ‘It can change at any time, any time people can fight and it can deteriorate, one moment it is good, another it is bad.’[40]

If insecurity is trouble, disorder, a feeling that anything can happen at any time, then the opposite can be found in Zeila, Somaliland: ‘Security is very tight, it is different to other regions; if you go and your house is open and you go to Hargeisa or Borama in the summer and come back it will still be left alone.’[41] In Somaliland’s Baligubadle, people go one step further and call the security situation excellent, despite conflict being far from absent: ‘Last year there was fighting between Arab and Ogadeen [clans]; … we collected diya [blood-money] for those persons … submitted [it] and the problem was solved.’[42] The acting commissioner of South Sudan’s Mundri, in contrast, has the feeling that his security channels have broken down: ‘Insecurity is caused by cattle keepers … When they came they were supposed to see the authorities … But they just sit wherever they want and give a hard time to our people … So people fear.’[43] People thus strongly relate security to how they perceive their security arena is ordered and feel insecure when the order breaks down, leaving malignant forces to cause ‘trouble’.

Rather than proposing an objective standard, assessing security is based on subjective statements of respondents. Thus, security is the felt durability of physical integrity and insecurity is the felt threat of physical harm. In other words, subjective security is more than an immediate feeling; it involves an expectation of the future.[44] I positively define the term security (rather than conflating it with the absence of conflict), while not extending it beyond the realm of physical integrity, which would strain the argument. Physical integrity is not an easily quantifiable variable and thus the aspect of perception plays a crucial role. As indicated by the quotes above, perceptions of security are shaped by ascriptions to actors providing security or threatening harm and by the workings of the security arena between providing acceptable forms of ordering and promoting what people locally would perceive as disorder.[45] I will analyze this through the concept of the security arena.

Three theoretical and three empirical studies furnish the basis of the security arena concept: from the angle of theory, Nelson Kasfir’s political arena; Stephen Hilgartner and Charles L. Bosk’s public arena model; and, more recently, Tobias Hagmann and Didier Péclard’s negotiating statehood.[46] On the empirical side, Thomas Bierschenk and Jean-Pierre Olivier de Sardan analyzed the local political arena in the Central African Republic (CAR); Alice Hills, the Somali security arenas; and Manfred Öhm, local institutions and state-society relations in South Sudan.[47] All of these contain elements I want to draw on in this study.

While each study has a different emphasis, ranging from social problems[48] to activities,[49] all six studies centrally posit ‘actors’. In fact, actors are the central constituents of the arena and therefore merit close attention.[50] Bierschenk and Olivier de Sardan place their analytic emphasis on three main power poles: the village chiefs, the farmers’ organizations, and the churches.[51] Hagmann and Péclard prominently cited Bierschenk and Olivier de Sardan’s empirical study of the CAR[52] when they developed their own negotiating statehood framework, in which they emphasize actors’ resources and repertoires. Actors are varyingly described as ‘personae’, ‘operatives’, ‘power brokers’, ‘power poles’, and ‘local institutions’.[53] Rather than suggesting that these studies all research the same issue under a different name, I argue that understanding differences between actors is in itself key to understanding how they impact security in the arena.

The image of an arena invokes ideas of action, struggles, and competition. Surprisingly, Kasfir’s research on the political arena is the only study out of the six that grants an in-depth appreciation of actions by analyzing how people enter and leave the political arena and how this affects political development.[54] In contrast to other conceptual terms, such as ‘structure’, ‘setting’, or ‘order’, the terms ‘arena’ and ‘ordering’ suggest a central emphasis on actions and processes. It is precisely through emphasizing actions that the security arena concept aims to appreciate human error, misunderstandings, and contradictory interpretations of one event. Through interactions, actors can form patterns that impact security processes in the arena. Actors engage and shape the security arena through varying forms of actions. Examples of this can be found in Öhm’s case studies in Yambio and Thiet, in which he links war hierarchies and relations to the specific state-building trajectory of South Sudan, which ultimately led to what he calls the ‘SPLA state’, a political system in which the state and the Sudan People’s Liberation Army (SPLA) armed group became inextricably linked.[55]

In sum, a security arena is composed of actors that interact on the issue of physical integrity around a predefined center of study. The reduction of the security arena to two relational dimensions – actors and their interactions[56] – around a spatial dimension defining the research interest draws from existing works studying the arena while further specifying the concept. To situate comparative insights across arenas on how and why varying actors interact in different forms around the issue of security, I delve further into the characteristics and traits of actors and interactions. For the actors, these characteristics are their organization, attributions, and resource flows. Actors have interests that they pursue through varying strategies that can be observed as incidences of interaction in the arena.

Actors are the elementary units of the arena. In an area, geography takes precedence, while, in an arena, actors shape geography. Norman Long encouraged the actor-oriented viewpoint in order to move away from determinist interpretations of social change and to move towards identifying strategies and rationales.[57] What constitutes an actor in this study is based first and foremost on empirics: who are the people, groups, and institutions respondents talk about when discussing security? While respondents judge many actors as relevant to their security, some of these named actors do not officially concern themselves with matters of security but were described as relevant to respondents’ perceived security. In their answers, respondents would list a variety of actors that span along a spectrum of very structured institutions, e.g. the US Army, to rather loose categorical descriptions, e.g. the youth or the farmers. While I include all mentioned actors, I attempt to distil from respondents’ statements only those individuals and groups who are embedded with the power of agency.[58] Thus, while the ‘youth’ from the Daami quarter of Somaliland’s capital might generally be categorized as threatening by some respondents, in analyzing concrete incidences I can get closer to who is the driving force within this category – that is, where the agency lies. Actors endowed with agency can vary along characteristics of organizational structure, attributions, and resource flows (see Table 1.1).

Table 1.1 Characteristics and examples of more stable and more fluidly ordering actors

Institutionalized organization:

Members of the Somaliland army near Zeila are assigned there through national recruitment and deployment processes. No other person can become a member. Soldiers have designated roles, tasks, and rather regularized modes of interaction with other actors and individuals.

Clear attributions:

The state judge of Bangassou (CAR) can thoroughly describe his official role as a provider of justice through deep knowledge of the country’s penal code, all the while situating his official role within a deficient working context. Simultaneously, other respondents within the Bangassou arena would ascribe him a role similar to his self-depiction.

Regularized resource flows:

The UN troops in the CAR and South Sudan have highly complex resource attribution systems from their contributing countries and the UN. They receive their salaries regularly, as well as food and lodging to sustain themselves. They are also fitted with weapons, vehicles, and technological infrastructure.

Personalized organization:

The Bongo chief of Buseri (South Sudan) holds strong personal authority. His role ranges from moral guidance to the integration of newcomers to conflict resolution. He deals with issues and actors as they come and as he sees fit.


Diffuse attributions:

The vigilantes of the same town (Bangassou) are not  able  to clearly define their own role within the community – ranging from forceful action to engaging in intercommunity deliberation – and are highly contested within the arena: to some, they are protectors, to others thugs.

 Ad hoc resource flows:

Elders of the Gabooye minority in Daami (Somaliland) receive only a token salary from the state. They depend on other business activities for their survival and have to negotiate with the (mostly poor) members of their community to collect contributions for the payment of compensation deals.

What respondents label an actor can range on the organizational spectrum from institutionalized, e.g. the US military in CAR’s Obo, to personalized, e.g. quarter defense groups in CAR’s Bangassou. On the one hand lie actors that assign membership, responsibilities, and codes of conduct according to rules and regulations. On the other end lie actors that assign membership and responsibilities according to individual leaders and personal decisions drive behavior. Through the institutionalization of an actor, members ensure continuity even when composition fluctuates, while, within personalized actors, individual members have much more leeway for independent action.

When attributing characteristics to actors, different respondents at times used clear markers and recognized traits or would describe actors diffusely and in contradictory manners. I investigate how comprehensive self-attributions are and how such attributions are contested in the wider arena. Clear attributions from within and without are indicators of actor cohesiveness. Diffuse attributions, on the other hand, allow members to shape their actor as they see fit.

Actors seek both material and immaterial resources to pursue their agendas. Material resources in the security arena include arms, vehicles, and personnel. Immaterial resources include personal or institutional authority, laws, documents, and invoking the divine. Resource flows can range from regularized to ad hoc. Resources flow regularly depending on an actor’s role (e.g. because someone is an officer of a certain rank) or due to ad hoc gains, such as leveraging fees at checkpoints, and often a mix of the two with different levels of importance.

Besides actors, inhabitants populate each arena. They are more than simple subjects to the activities of security actors. Inhabitants can become part of the actors through recruitment or by forming new actors themselves, they can grant and withdraw support and adherence to actors, and they debate which claims they perceive as legitimate.

The ways in which actors can relate to one another in the security arena are limitless. Respondents would voice discontent with another actor, threaten retaliation to an attack, describe a recent clash, speak of the possibility of dialogue or even a successful recent discussion, or mention the material goods they received from someone else or the support they are providing to others. In focus group discussions, participants could draw the arena as they saw it, with the aid of different shapes and colors of cardboard. I simplified the vast array of possible interactions to three broad forms of ‘threaten’, ‘protect’, and ‘support’ – very similar to the forms of interaction proposed by Ana Arjona.[59] In retrospect, respondents had difficulties distinguishing between ‘protect’ and ‘support’ and likewise, from a theoretical angle, the former can be subsumed as a sub-form of the latter. Broadly defined, threats include the (alleged) intent to engage in violence as well as the act itself.[60] Supportive acts are those in which an actor uses its own resources to further the aims of a second actor.[61] Both supportive and threatening acts always emanate from actors and deliberately or unintentionally target certain other actors and individuals of the arena. Nevertheless, both support and threats can be very diffuse and become more so the less people are able to attribute them to clearly defined actors, target groups, or the ability and willingness to follow through on them.

Actors can use multiple interactions along the spectrum, from supportive to threatening. An actor might use support towards certain actors and threats towards others or segments of the population. These interactions are the most visible forms of processes in the arena. They are visible expressions of the actors’ larger strategies that they pursue to further underlying interests. However, the further one goes beyond observable actions, the less clearly a process can be pinned down. This study is empirical and therefore my emphasis lies on observable interactions based on (to me, mostly) unknown interests and strategies.

Take an auxiliary police officer who levies a fee by pointing a gun at a traveler. The threatening interaction can be observed but the strategy is less clear: by wearing a uniform, he might be signaling his continuing adherence to an actor and he might also want to retain this actor’s influence (e.g. over travelers) in the arena. His underlying interest could be ‘greed’ because he wants more than he deserves – whoever defines this – or it could be ‘grievance’, such as when his salary is not paid or is insufficient to put food on the table and send his children to school.[62] Thus, while I speak of ‘actors’, my empirical data grants insights only into a collection of individual acts. However, through triangulation of multiple sources I distinguish truly one-off individual acts from those that are attributable to a larger actor – in other words, if the police officer did this once on his own terms or if the way the police established itself as an actor is what makes such acts of extraction possible and recurrent.

In sum, the security arena focuses on how actors of varying types interact through threatening and supportive actions. In a next step, when individual interactions are put together and seen in a broader picture, they can be analyzed to discern the forms of ordering actors promote in the security arena on a spectrum ranging from fluid to stable ordering. When actors with different approaches to ordering meet in the arena, they can try to impose their form of ordering on the other or negotiate forms of cohabitation.

Stable to Fluid Ordering Spectrum

The varying settings of peace and conflict that countries experience are not just the outcomes of the decisions of actors but are part and parcel of the ways actors order their political system.[63] Too few scholarly contributions have searched for the order behind seemingly chaotic violence.[64] Those scholars who have investigated this have found varying but targeted aims behind even the most extreme forms of violence.[65] Kalyvas went as far as to pronounce ‘order is necessary for managing violence as much as the threat of violence is crucial in cementing order’.[66] The ways actors create and undermine security can be seen as the directly observable building blocks of how actors order their larger political system. Thus, rather than looking at an established order, ‘a set of predictable behaviors, structured by widely known and accepted rules which govern regular human interactions and behaviors’,[67] I investigate the processes through which actors in the security arena try to bring about their desired order. In other words, I look at processes of ordering. Randall Collins, among other sociologists, investigated at the micro level how individual and group interaction tend to escalate into violence and what factors most often inhibit situations from spiraling out of control.[68] The security arena concept remains micro-level enough to allow for the incorporation of individual and group-level spiraling processes,[69] while broad enough to compare varying cases.

Ordering The Security Arena – Chapter 1 From The Security Arena In Africa

Table 1.2 Characteristics of the stable–fluid ordering spectrum

Through the concept of the security arena, I propose a new process-driven concept to investigate how ordering impacts security (see Table 1.2). Stable ordering in the security arena becomes visible through hierarchical channels between actors and fixed distributions of responsibilities. Fluid ordering, on the other hand, is marked by horizontal actor relations with shifting distributions of responsibilities. To continuously enforce hierarchies and claim dominance over certain domains in the security arena, actors seeking to stably order the arena depend on regularized and, relative to other actors, strong resource flows. Actors that fluidly order the security arena can make do with occasional, lesser resources to navigate shifting horizontal relations without seeking dominance.

A key reason for choosing a type of ordering is the impact this has on security. While stable ordering lends more predictability, fluid ordering creates more modifiability. The debate on structure and agency is thus not only theoretical[70] but one that people live out on the ground. Some actors tend towards deliberately creating structures that will regulate methods to address issues of security in the arena: for instance, when the police are solely responsible for criminal cases, a person would have to challenge the order to seek justice for a murder with another influential actor he trusts. Other actors are more inclined to deliberately open flexible avenues for the pursuit of matters of security: in this case, a person could freely decide to report a thief to either the police, the gendarmerie, or an elder.

Most importantly, both forms of ordering coexist in an arena, either complementing one another or becoming the objects of violent struggles over the ‘right’ ordering forms.[71] Competition arises when an actor promoting stable ordering tries to extend its control into spheres of more fluidly ordering actors. Reciprocally, more fluidly ordering actors might feel threatened by the rising power of a dominant actor and seek to undermine its stabilizing claim. On the other hand, stable and fluid forms of ordering can cohabitate if the involved actors do not wish to take on the others’ domains. They can even benefit each other, for instance when fluid ordering actors need a ‘strong hand’ or when stably ordering actors want to avoid the resource drain of all-encompassing control.

My analysis finds that neither side of the ordering spectrum provides more security per se. While stable ordering can provide predictable security, institutionalized, highly resourced, control-seeking actors can also create exactly the opposite – insecurity. Fluid ordering avoids predictable insecurity by opening modifiable pathways to seeking security. However, under constant flux, often with fewer resources, and the aim to stay elusive, fluid ordering actors cannot claim guarantees and therefore might enable the recurrence of insecurity. Security deteriorates when actors turn to violent activities in pursuit of their contending preferred forms of ordering. Security improves when actors agree on the ways in which (different parts of) the arena should be ordered. It is thus not fluid or stable ordering per se that explains variation of security among the nine cases but the struggles over contending forms of ordering. I investigate this hypothesis from the angle of the opportunities provided by the shape of the arena and from the angle of how actors order the arena in order to pursue their aims.

Actors and their interactions add relational dimensions to the arena that deliberately avoid confining analysis to predefined geographic containers. However, there is also a spatial aspect to the arena that remains hard to pin down. I argue that the arena is a scalable concept and the ‘area’ of the arena depends on deciding on a center and a level of analysis. My analysis principally revolves around ‘local’ arenas:[72] here, the midpoints are the administrative centers of the nine towns I selected for further studies. Who is and is not part of the arena are questions of ordering relevance. When I ask people downtown who matters to their security, all actors these people mention are considered, regardless of whether they are 100 meters or 100 kilometers away. The security arena is made up of actors that interact around a predefined core or, as I call it, an ‘inner circle’. All areas beyond this inner circle are part of the outer circles.

The inner/outer distinction at the local level is mirrored by relations between center and periphery on the national and international levels. While investigating local cases, there are certain decisions and processes that cannot be explained solely on the local level, because the arena in some instances is only a periphery to a higher-ranking center of decision-making.[73] When traveling to the capital of each country, I analyze the national level arena and all actors and events become relevant that relate to politics of the nation – loosely, what happens within the countries’ borders. When looking at the international level of the security arena, the centers become multiple: depending on the issue at hand, powerful neighbors, headquarters of international institutions, or transnational armed groups might draw the attention as varying centers of the arena.

Inner–Outer Circle

Scholars of borderland studies have put a strong emphasis on analyzing local frontiers, linking local political trajectories to the ordering of international[74] and internal borders.[75] Igor Kopytoff, in particular, looks at the African frontier and depicts the distinct possibilities created by the relations between smaller centers and people moving to the frontier:[76] there, he argues, people can establish new political formations aside the pre-existing, organized societies.[77] Achille Mbembe shows how a Cameroonian independence movement sought the possibilities of the ‘maquis’ to continue its agenda even during government repression. In the forests and bush, the revolutionary committees could continue mobilizing the populace but, at the same time, disperse into the surroundings in the event of an attack by the much stronger colonial force.[78] The creation, handling, and erasure of dividing lines is thus a crucial object of study I wish to investigate through the concept of the security arena.

In the security arena, one possible delimitation can be drawn between an inner and an outer circle of the arena. This blurred line roughly relates to geographic indicators but goes beyond physical markers by focusing on the possibility of interaction. In Map 1.1,[79] we see an agglomeration of buildings at the bottom which is Baligubadle (the inner circle). To the south it borders Ethiopia (dashed line); towards the north the density of houses quickly thins. I have cropped the satellite image at the northern exit and show only a few hundred meters of the twenty-kilometer-wide plain of Gumburaha Banka (the outer circle) that one has to cross before reaching the top part of the image, Gumburaha village. Within the image I have marked the buildings of several key security institutions, the two market squares, and an exemplary temporary settlement in the grazing plains.

The inner circle is characterized by physical proximity, which creates the potential for daily interaction between inhabitants of different social groups, the interdependency of livelihoods, and the existence of numerous personal links between members of the different actors and social groups. Therefore, security in the inner circle is based on the necessity of continuous cohabitation.

The outer circle, on the other hand, is marked by irregular interaction or interaction based on particular social links (relatives) or activities (trade). Interviewees ascribe rumors, unclear intentions, and inconsistent labels to actors in the outer circle. Such rumors are facilitated by the scarcity of information in the countryside compared to the city and have played key roles since the French Revolution[80] to current spikes in violence.[81] Security in and from this surrounding sphere of the unknown is based on being able to face the irregular (rather than everyday security challenges).

While the dividing line between the two circles often has visible demarcations on the ground, such as rivers, forests, deserts, and population density, humans shape these natural boundaries to order the arena. I thus focus my analysis on the struggles revolving around the drawing, crossing, and at times erasing of the dividing line.

Ordering The Security Arena – Chapter 1 From The Security Arena In Africa
Map 1.1 Baligubadle town, Banka plains, and Gumburaha village, Somaliland


Local arenas are not isolated contexts but relate to varying centers. Without a view beyond the local, some aspects of security would be impossible to understand,[82] such as why states send forces to the respective locality (or refrain from this), who deploys and funds peacekeepers, or how armed groups move in and out of local arenas in the changing national context of peace and conflict. I argue that there are two key elements to the relations between center and periphery: the degree of a connection and the way of creating such a connection.

Jeffrey Herbst, in his landmark work, has shown that states do not necessarily seek to control their hinterlands, especially when population density is low and the security imperative for the regime is minimal.[83] It is thus more fruitful to investigate center–periphery relations that take into account the particulars of African state formation with a view both to the past and to the present. Kopytoff proposes looking at the extension of control and authority as frontiers. This grants a dual lens to the center, which extends along frontiers, and to the frontiersmen and women who leave the center for the particular opportunities of the frontier.[84] There are actors that deliberately seek to leave the range of the central state to avoid the potential violence and extraction emanating from it.[85]

Some actors engage multiple levels, in what Klaus Schlichte and Alex Veit called ‘coupled arenas’.[86] Intervention forces and international non-governmental organizations (NGOs) thus have to negotiate their actions where they have their headquarters, at the national level in the intervened country, and on a local level, where they carry out their programs. This often creates unintended side effects as the decisions made by stakeholders acting within the headquarter or national context are carried out in the local arena, which operates under different political logics. Some people use these different logics to their advantage. Actors can use the peripheries to position themselves beyond the control of central powers to expropriate the populace and prepare for an assault on the capital.[87] ‘The standard international legal practice almost always equates sovereign power with control of the capital city’,[88] which makes it such a lucrative target in the first place but also distinguishes it from large areas with low population densities that central powers have difficulties in reining in.[89] Centrality and peripherality have to do with access and the closeness of people and actors.[90] Both sides have agency to interact with one another[91] and use the particularities of their sphere in the bargain. The same applies to the international sphere, where nations find themselves in more central or peripheral positions of economic chains and political decision-making.[92] Far from being the secondary choice of actors, James C. Scott has shown that leaving the center to live in the peripheries can be a deliberate choice for people seeking to avoid the perils of control, such as taxes and conscription.[93] While the center and periphery possess different characteristics, they are always in relation to one another and with some degree of friction, as the people in the periphery evading central control ‘represent . . . a constant temptation’.[94]

Central actors can choose to relate to the peripheries on a spectrum ranging from imposing their will – be it through governance or extraction – to meeting peripheral people for negotiation.[95] Mamdani argues more towards the former when he discussed how central powerbrokers impose themselves on rural subjects.[96] David K. Leonard more towards the latter end of the spectrum, when describing how urban African elites mediate their relations to the periphery through community-based social contracts.[97] In analyzing the center–periphery dimensions relevant to the local security arenas, I focus on the historical evolution of the state and its security apparatus. The relevance of the state and the structure of security authorities vary greatly among the three countries and in large part this variation can be traced back to long-held economic structures and sociopolitical settings – as Fernand Braudel calls them, the ‘moyenne’ and ‘longue durée’.[98]

The View of State, Non-State, and International Actors

Stable ordering, fluid ordering, and the struggle over their cohabitation can be assessed from multiple angles by looking through the lenses of state, non-state, and international actors. I make a distinction between actor types also because this was a key differentiation made by respondents on the ground. While such descriptions are not always clear and entirely distinct, I try to differentiate them for analysis. State actors are those that primarily claim to draw their means to order the security arena from the official state’s laws and resources. Non-state actors are those that derive their means to order primarily directly from the people and from resources they extract without a detour over the state. International actors are those that primarily derive the resources and personnel to order the arena from outside the borders of the country they are involved in. While each can use aspects of other types – such as a state governor being a locally supported person or a traditional chief receiving a small state salary – assigning a type depends on what the actors themselves and other respondents perceive as the main source of making claims to ordering the security arena.

The African state has acquired a bad reputation. Jean-François Bayart and colleagues speak of its ‘criminalization’, in which state leaders mix political influence with economic stakes to gain a profit in illegal sectors.[99] Patrick Chabal and Jean-Pascal Daloz have famously concluded that ‘Africa works’ only for elites, who use ‘disorder as political instrument’.[100] Especially in regard to the three states under scrutiny here, the devastating impact state actors can have on security becomes obvious. In the CAR, rulers have either neglected the peripheries or met them with extreme brutality. In South Sudan, where violence is used by multiple actors, state security forces are committing some of the most horrific atrocities of the ongoing civil war. Somaliland’s declaration of independence in 1991 was a response to the Somali state’s indiscriminate use of violence against Somaliland’s people, including its bombing of Hargeisa in 1988. Whatever the alleged particularities of the African state, the state in its bureaucratized form can also create violent open spaces in which state employees commit horrifying atrocities.[101] Some authors nevertheless call for a strengthening of the state as the best way to stabilize African states and recreate security.[102]

I propose that the state is neither, in itself, problematic nor helpful in creating security. Rather it acts as a powerful narrative around which actors position themselves and play out their struggles over ordering the arena.[103] State actors are defined, on the one hand, through the bureaucratic narrative of their position based in documents, laws, appointments, and officialized hierarchies – processes that tend towards the stable ordering end of the spectrum. On the other hand, state actors use their assigned roles to create fluidity to avoid the disintegration of their forces and for the ability to create a personal profit from an assigned position.[104]

State actors can create security through stable enforcement when they have the means and will to provide for predictable security. However, state actors undermine security when, through their enforcement, they weaken other actors that local inhabitants prefer entrusting with their security. Especially under the observed contexts of resource scarcity, state actors can also create security by facilitating flexible resolutions outside the official law.[105] This fluidity can also undermine security, however, when state actors use unaccountable violence against civilians.[106] Third, state actors and their officially stable ordering narrative coexist with other actors and forms of ordering in all nine cases, either competing or complementing one another. I thus also take a close look at the ordering forms of non-state actors.

Non-state actors that attempt to order the arena always have to relate in some way to the presence of state actors and their claims to a force monopoly. Within the distinct ways non-state actors impact the security arena, they often mix fluid and stable forms of ordering. State and customary institutions interact in ‘processes of mutual diffusion’.[107] Hybrid governance scholars analyze such organizational arrangements[108]  and Christian Lund labels the outcome ‘Twilight Institutions’.[109] Bruce Baker investigated how state and non-state actors link up to the benefit of overall security provision.[110] Such linkages can mediate the fact that historical social contracts for most Africans are not between the state and individuals but between the state and communities[111] – non-state actors can thus act as intermediaries. But I heed Kate Meagher’s words of caution: ‘The condemnation of non-state order as institutionally destructive has been replaced by its celebration as a vehicle of embedded forms of order and authority.’[112] I, therefore, look at both sides of the coin of local non-state ordering: processes of improving and of reducing security.[113]

Different non-state actors should also be distinguished by whether or not they are armed. Most interview partners from non-state actors in this research were unarmed and situated in the inner circle. On the other hand, concerning armed actors, a lot of information was gained not from them directly but about them from other respondents in the arena – which is nevertheless very important as it says much about inhabitants’ perceptions of security in the arena, which is the aim of this study. Armament grants non-state actors a very direct impact on the security arena as a means to impose authority on others and to use as stakes in negotiations. Armament, however, also makes actors a more visible target. Unarmed non-state actors have difficulties imposing their authority, which is often based on tradition or religion when faced with armed actors. They thus have to rely on negotiation but also can have more freedom in doing so, as they can move about the arena and engage a wide variety of actors without being perceived as a threatening target. Non-state actors relating to state and international actors bring the need for cohabitation between fluid and stable modes of ordering to the fore. Actors can be at odds with one another when they attempt to force their way of ordering on the other but they can also complement each other when they use their different abilities of ordering to their mutual advantage.

The literature describes particularities of non-state actors ranging from personalized rule to traditional institutions. On the personalized end of the spectrum, Mats Utas finds that ‘Big Men are alternative governors of peopled infrastructures’, who hold personal power, the ability to create a following, and the informal ability to assist people privately.[114] Conflicts are fought by Big Men over socioeconomic opportunities rather than politicized identities.[115] Similarly, William Reno describes how elites deliberately undermine state institutions to keep rivals off balance and channel resources through personal patron–client relations.[116] While these rivalries at times give rise to clashes of arms, fluid ordering lends particular possibilities to unarmed non-state actors as well. The literature on Somaliland’s pathways of providing flexible security resolution apart from the state and its written laws is especially proliferate.[117] Roger MacGinty proposes traditional peacebuilding to be commonly based on moral authority, public elements, storytelling, relationships, and local resources[118] – in other words, aspects that are more on the fluid end of ordering the security arena.

Some non-state actors attempt to position themselves more stably in the arena. Paul Jackson describes how warlords present themselves as alternative governors when the state collapses[119] and other scholars scrutinize how rebels with initially personalized rule start erecting state-resembling institutions to generate sufficient legitimacy to stay in power.[120] However, ‘conflating rebel governance with state order forces analysts to awkwardly transpose the state-formation framework onto an actor that actively resists the state’s attempts to project order within its ascribed territory’.[121] Therefore, other scholars prefer analyzing rebel governance as a distinct form of finding ways to regulate social life.[122] The focus of the literature on (armed) rebel governance should not mask the fact that unarmed non-state actors can also create lasting, regularized institutions that can be a key avenue for inhabitants seeking predictable security aside the state.[123]

The discussion of the ‘responsibility to protect’ has given international actors strong leverage to intervene in states that are unwilling or unable to protect their own populace.[124] As all three countries are ranked at the bottom of fragility indexes and conflict is ongoing, interveners have entered each of them.[125] However, the impact on the local security arena can vary starkly from mission to mission and contingent to contingent. These intervening actors have to position themselves towards the actors already present on the ground, including stable and fluid ordering actors.

From a stable ordering perspective, security dilemmas between consolidated actors are key to understanding insecurity.[126] The logic of power-sharing draws from this international theory assumption by seeking the inclusion of all competing poles.[127] However, Denis Tull and Andreas Mehler point to the rather crude side effect that international enforcement of power-sharing agreements between warring parties creates incentives for would-be leaders to take up arms in order to obtain a position in government.[128] Severine Autessere warns that internationally enforced peace-building shows a disregard for local causes and dynamics of conflict.[129] International actors viewing struggles through stable ordering lenses can indeed miss out on other more fluid aspects of the security arena.

Roger MacGinty, therefore, calls for international actors to tackle fluid aspects of local security head-on.[130] Bottom-up peace-building cannot simply be enforced top-down but necessitates constant dialogue and the acceptance of alternative modes of ordering.[131] Jeni Whalan shows that peacekeepers need to seek local legitimacy to have a positive impact on the ground but this fluid embedding is counteracted by the institutionalized mode of peace operations.[132] The fluid embedding of peacekeepers in the local security arena remains under-researched and will thus form a main part of this book.

Unfortunately, international peacekeepers tend to undermine fluid ordering actors and contribute to a (mostly imagined) stable state. While internationals should engage the whole spectrum of actors and their ordering modes to engage security on the ground, they often revert to ‘bunkerization’ to protect their modes of ordering by separating themselves from the wider local arena.[133] Whether or not an intervention can have a positive effect on security on the ground depends on how they engage the entire spectrum of actors and ordering forms in the arena. Intervention forces can have an immediately positive impact on security when they forcefully engage a violent actor and push it from the arena. However, with limited local knowledge such enforcements can also have negative repercussions on social cohesion and lead to the creation of new violence. More fluid ordering by internationals can allow for adaptation to the local context. However, fluid intervention also provides foreign soldiers with opportunities to pursue ulterior motives, such as seeking personal profit or committing sexual abuse. Interventions are particular in that a new actor with distinct forms of ordering enters an arena with already-established actors and modes of ordering. Interventions can have a positive impact on security when these new interveners are willing to engage the entirety of power dynamics in the arena, while they risk creating insecurity when they are biased towards certain actors of the arena.


Although the social sciences, and especially political science, bear a long tradition of comparative studies, local security has seldom been the object of such research.[134] I attempt this through the analytic approach of intra-regional Comparative Area Studies (CAS).[135] The CAS approach is used by scholars who wish to combine profound area knowledge with comparison across cases. Theory-building based on contextual knowledge can be advanced by challenging European and North American inventions with comparative, empirical observations on the ground.[136] Qualitative field research on the security arena necessitates a degree of flexibility and CAS present a rather broad notion that can be filled with methods and analytical tools fitted to the research question at hand. This section will discuss the challenges of case selection, the gathering of qualitative data on the security arena, and its structured analysis.

Case Selection

This analysis has two levels of cases: three country contexts – the CAR, South Sudan, and Somaliland – as well as nine local case studies as seen in Map 1.2 from west to east: Paoua, Bangassou, and Obo; Raga, Buseri, and Mundri; Zeila, Baligubadle, and Daami. These cases are broadly under-researched. In Somaliland, Zeila had not featured in any renowned publication on the country, whereas Baligubadle and Daami are also seldom researched[137] compared to the more often studied areas in the east of Somaliland.[138] In South Sudan the cases of Mundri, Buseri, and Raja have been discussed in few publications in recent years.[139] Save for some rare publications, the CAR has not been the subject of any recent published academic analyses grounded in fieldwork.[140] Apart from reports on armed groups around Obo and Paoua,[141] the CAR localities examined in this book have not been the focal point of academic research.

While ideally case selection of small-to mid-N comparison – that is from two to around a dozen cases – should be non-random to enable representativeness and causal leverage,[142] studying security brings with it logistical and safety issues that set limits to theoretical considerations. The countries were selected according to alleged commonalities in the reduced role of the central state in the life of its people: all three countries have only minimal public services and an undiversified economy. Numerous indices rank all three cases as some of the most fragile in the world[143] and the Fragile States Index has even ranked them as the three most fragile states for four years in a row.[144] These rankings, however, are deceptive as the countries vary in socio-demographics, culture, resource potentials, and, most crucially, in political outlook.[145] The main comparison will not be of countries, however, but of local security arenas. I thereby heed the call for a move down from the national level and towards bottom-up local comparisons.[146]

Ordering The Security Arena – Chapter 1 From The Security Arena In Africa
Map 1.2 Field research sites (small star = case study; big star = capital)

Within each country, the cases were selected broadly along three theoretical considerations: first, variation on the socio-spatial distance to the capital or a large administrative center. A peripheral setting enables the investigation of local dynamics more clearly, as those holding power at the central national level do not view the locality as a key part of their rule.[147] When the willingness to provide state security (or other significant services) to the population is limited, alternative providers and constellations come forth more clearly. All selected cases can be considered peripheral: some more obviously through their enormous distances from the capitals – Obo and Raja lie more than 1,000 kilometers from their respective capitals. Others, such as Daami quarter in Somaliland’s capital, were selected due to prior knowledge of the cases and their specific political and economic marginalization.[148] A second selection criterion was some limited variation on population size. This enabled talking to most of the relevant security stakeholders, whereas in large cities finding a manageable pool of respondents could create a selection bias. Third, the cases varied according to the presence or absence of different forms of security actors, most notably government institutions, alternative security actors, and international peacekeepers. Therefore, the varying relevance of each actor, its specific impact on security, and relational setting within the arena could be compared. And, finally, the cases were varied along their levels of security. Somaliland was added to the two original countries to include cases with rather successful security provision at a time when both South Sudan and the CAR had fallen back into heavy conflict.

The deliberately sought-out absence of a monopoly of force often coincided with contesting claims by various groups, creating security constraints for the researcher. Logistical issues, often linked to security threats such as roadblocks, further reduced the amount of possible research localities. Case selection confined by security and logistics had repercussions on the research design: avoiding cases with high levels of violence or latent instability could result in bias towards a rather optimistic look on non-state alternatives. Overcoming logistical constraints on the ground by aerial routes was often facilitated by large NGOs or international organizations, bringing with them a significant outside influence that again impacted statehood on the ground. In actual practice, cases were thus often selected according to, first, (subjectively) interesting security dynamics gathered on the respective locality on arrival in the capital and from talking to numerous locals and internationals. These cases were then narrowed down by the possibility to get there and the feasibility of safely conducting research. This selection calls the representativeness of cases into question and therefore the generalizability of the findings, which I discuss further in the ‘Limitations’ section.

Data Generation

I investigate each of the nine local arenas as a ‘semi-autonomous’ field that is bound by the ordering processes around the issue of security but is nevertheless related to other arenas and larger society.[149] Viewing a case as an ‘arena’ incorporates the social space principle, which states that political struggles are about making an explicit consensus out of individual thoughts[150] – hence the focus on actors and their actions in the arena. Field theory as ‘an analytic approach, not a static formal system’ informs this analysis of arena as a quest to make ‘complex social phenomena intuitively accessible’.[151] A key lesson from field theory that also applies to gathering data in the security arena is to reflect on the positionality of the researcher himself.[152] Indeed, my position and mobility in the arena varied, which changed the type and breadth of data I received. With these lessons in mind, I collected information through interviews, focus groups, non-participatory observation, and monthly reports in each of the nine local arenas. This array of methods did justice to the explorative nature of this research endeavor due to access restraints, concerns for security, and the novelty of the research approach.[153]

Data was gathered during field research trips to each of the three countries and each of the nine localities: one pilot field trip of one month’s length each and another research trip of two months each one year later (plus a short follow-up visit to the capitals of the CAR and Somaliland in the third year). After the first research stay, further research trips to South Sudan had to be canceled due to an upsurge in conflict in all selected research sites and the capital. On average, I spent four months in each country and two to three weeks per locality. Considering this relatively brief period per case, I used site-intensive methods that aim at generating in-depth case knowledge, while avoiding the conceptual baggage and time constraints of ethnography.[154] Lotje de Vries and I conducted most of the field studies in the CAR and South Sudan together and Andreas Mehler briefly joined us for research in the CAR. Having a team of two during most field stays was an incommensurable benefit to data validation: combined interview questions could probe broader and deeper, complexities were immediately discussed thereafter, and the at times draining conflict environment was jointly navigated. Just as the benefits of co-conducting fieldwork surpassed my expectations, likewise did splitting each field study into multiple trips: interviewees were positively surprised by my kept promise to return. Discussions gained a whole new level of trust and I could refer back to issues that were too sensitive the previous year.

Interviews with members of security actors and the populace offered rapid, albeit subjective, insights into the issues at stake in the security arena and into broader security perceptions.[155] Interview partners were selected in a snowball fashion: official state authorities had to be visited first and they would recommend (and authorize) talking to further state forces, such as the police, army, and gendarmerie. Additional interview partners were selected according to the names and stories presented during interviews, group discussions, and as suggested by local research assistants. More than 300 interviews were conducted. Each interview lasted between one and two hours, and, if permitted, interviews were audio-recorded. The interviews were not transcribed but notes were typed on the spot in near-exact wording (depending on the rapidity of the speaker). While I was fluent in two languages – English and French – and attempted to deepen my knowledge of further research languages – Arabic and Somali – many interviews had to be translated by local research assistants. I will thus use the information provided in the interviews only within the specific context it was given and not to interpret the specific usage of language.

Focus group discussions were the second most important method.[156] During the first trips, discussions with broadly defined societal groups (such as youth, women, elders) gave a sense of the contested security issues in the arena and provided the selection of further individual interview partners. During the second trip, participants were selected along certain key population groups (varying from arena to arena by ethnicity, age group, gender, or geographic categories) to enable the creation of comparable actor mappings. To do so, an attempt was made to select two groups of participants from different sides of the most salient sociopolitical divide and a third group of fairly neutral or mixed inhabitants: in Paoua (CAR), for example, youth from the ethnic Tali, a group of women from the ethnic Kaba, and a group of radio journalists. Such selection was not always straightforward. Research assistants would at times interpret selection criteria differently and some participants did not show up or left early. The discussion would proceed for close to two hours, during which the participants created a map of their perceived security arena with the aid of questions and flash cards along the following steps (see Figure 1.1): (1) introduction; (2) societal groups on oval shapes; (3) actors on rectangular shapes; (4) closeness of actors and groups; (5) and (6) interactions of threat, protection, and support on red, green, and blue arrows, respectively; (7) security incidence examples; and (8) perceptions of security. Continuously asking for real-life examples helped guide the abstract discussion. It was particularly insightful to have numerous focus groups discuss the same recent example of a key event – e.g. the arrest of an armed group leader – to gauge different people’s interpretations. A total of around fifty group discussions were conducted, that is, on average three actor mappings and three general group discussions per locality.

Ordering The Security Arena – Chapter 1 From The Security Arena In Africa
Figure 1.1 Steps of actor mappings

Monthly reports written by local research assistants recorded changes within each locality in small (monthly) increments. In these two-page reports, the assistant first filled out all changes in actor hierarchies, changes of key elites, the departure or arrival of new actors and populations, and some proxy indicators, such as fuel and sugar prices. Second, the report included security-relevant events of the month to build a continuous security event log for each locality. Some research assistants kept track of every slightest change in the locality, while others only mentioned large events. Also, the reports were often irregular due to a lack of internet or mobile phone connectivity, and, in the cases of South Sudan, they were interrupted completely due to the resumption of war. At the very least, these reports aided in bridging the gap between the first and the second trips, facilitating the tracing of exact changes within the security arena. On top of that, they were often the sole source for triangulating perceived security with actual events in the locality, as international media seldom covered the cases studied here. In total, more than seventy-five monthly reports complemented information gathering during periods away from the field.

Participatory observation and primary documents of intercommunal meetings and security measures were used as a final measure to triangulate narratives with actions. Owing to the relatively short stay in each locality, the insights gained from observations are limited. Nevertheless, interviewees were always sought out in their office or home (area) to gain a glimpse of the context within which they act. Thus, I visited the prison in Baligubadle and can compare its well-constructed building with the padlocked shack at the commissariat in Obo. Among others, I also met mediation committees in the CAR and elders in Somaliland and was given written documents on their conflict resolution activities. In total, around fifty primary documents, of which the number per arena ranges widely, contribute to the analysis.

Nevertheless, some actors, such as armed contraband groups, refused an exchange and state forces denied access to certain areas and actors, making the gathered data biased towards actors and inhabitants on the safer side of the arena. I will therefore indicate every actor whose members did not have the opportunity to voice their viewpoint with an asterisk (*) in actor mappings as well as other tables and figures. This confinement overlaps with a theoretical distinction between the inner and outer circle, as outer-circle actors were often harder to reach.

Data Analysis

The three countries – CAR, Somalia, and South Sudan – are most similar in a key characteristic:[157] the state holds no legitimate monopoly on the use of force throughout its territory and over all its people. Comparable indicators on low state budgets, minimal public service provision, and dire economic outlooks further support seeing the three countries as most similar. At the same time, they vary strikingly along security outcomes, ranging from rather high security in some parts of Somaliland to very low security in large parts of South Sudan and the CAR.

The local arenas are often very dissimilar to one another,[158] varying in population size, distance to the capital, and variety of actors. Nevertheless, certain processes in the security arena share remarkable commonalities: state actors use similar references to claim a monopoly on authority, while non-state actors use similar processes to find pockets of self-government outside the state, and peacekeeping missions adapt in comparable manners to very different surroundings. Processes in the arena play out both in relation to the context they find themselves in[159] and to more generalizable trends seen across contexts.[160]

I analyze processes in the security arenas to find patterns that resemble each other across different security arenas.[161] Process tracing tests alternative explanations of the effect by tracing events in fine detail.[162] This method enables a refutation of symmetric causality – bidirectional relation between cause and effect – and an understanding of the cause–effect relationship. I deliberately abstain from formulating hypotheses and testing them because the data I gain through the very explorative nature of my research does not support such an endeavor. Rather than comparing evidence for and against predefined theoretic assertions, I thus follow my broad sets of subjective data sources[163] to build a triangulated story that fits the overall picture found in and across arenas, to answer the question: what are the effects of varying forms of ordering on perceptions of security in the local arena?

As a key resource in drawing the broader picture of settings and processes in the security arena, I engage in focus group discussions to gain an impression of the arena. During my first trip to each arena, these discussions were rather open-ended, while, during my second trip, I guided participants to create comparable mappings of the arena. I try following up on all the actors respondents mention in these and other debates for interviews. However, the story always begins with the insights gained from inhabitants and actors of the inner circle and their perceptions of relevant actors and issues of security. From there, I move as far as possible (which frequently is not that far) towards the views of the actors in the outer circle that are often seen as spoilers. Unfortunately, the more insecure the investigated arena, the more my answers are biased towards perceptions in the inner circle. This is particularly the case for the level of security in the arena. I approximate security perceptions by asking people for their personal assessments and triangulate their perceptions with the unfolding events described to me in monthly security event logs written by research assistants, and descriptions provided in discussions and interviews (see also Box 1.1).

The output of this analysis will mostly take the form of written text, in part supported by graphs and tables. Thus some notes on the presentation of the written analysis are due: first, I use the present tense in the majority of the text for describing more general processes and perceptions in the security arena as I saw them during my visits, although these will have naturally evolved by the time of publication. I use the past tense, on the other hand, only when recounting specifically dated events. In most instances, I confine myself to the events that happened during my fieldwork or to recent events that were recounted in detail by local respondents while I was there. I thus assess the wide array of data that I have gathered over time in the security arena to trace processes of ordering. I then compare these respective processes across arenas to find generalizable patterns of how and why actors order the arena and of the impact of ordering struggles on security.

Ordering The Security Arena – Chapter 1 From The Security Arena In AfricaLimitations

Nevertheless, there remain limitations to this research endeavor that could benefit from further investigations, such as the explorative nature of the research, the unbalanced data pool, the confinements of the actor lens, and the restricted generalization of my findings.

The first concern is the truly explorative nature of this research. While the empirical novelty of the data gathered provides an important contribution to the debate, it also means that few context conditions could be predefined. Volatile security circumstances at times demanded quick adaptations to case selection and research timing. Most constraining was my inability to return to the three South Sudanese cases due to the outbreak of open conflict in the locations I visited during my first and only trip in 2014. This brought much less breadth and depth to the data gathered here in the form of less-structured focus group discussions, no actor mappings, and fewer interviews. Owing to the scarcity of primary data gathered on South Sudan, I exceptionally extend my data through reports by research assistants and media sources on the events that followed shortly after my own trip.

Second, conflict was a key reason why data differs from case to case. In many local security arenas, I could not fully probe outer circles and their potential spoilers. My research in South Sudan was the most confined spatially, to the inner circle, and temporally, to one research trip. The CAR and Somaliland received the same time in the field – two main trips plus a short follow-up. However, the outer parts of local arenas were much more accessible in Somaliland than in the CAR, where armed actors were present. All in all, while my study does investigate the crucial importance of the outer parts of an arena to security, most of my findings relate to how people of arena centers perceive that outer circle, rather than first-hand observations of and interviews with people in the outer circle. Nevertheless, through the explorative mix, I also discovered a range of very fruitful methods to be employed. I found that dividing the research into two main trips each helped gain local respondents’ trust even when total time per locality was limited. Conducting most of the field research in a team of two was a huge benefit to navigating the security arena more quickly, broadly, and safely. Local research assistants, despite their lack of formal training, were of particular value in gaining access, avoiding dangers, and keeping me informed between the two trips to each local arena or after the sole trip to South Sudan. Finally, the guided actor mappings were a fruitful tool to allow respondents to create comparable depictions of their security arenas across their different contexts. Overall, I thus gathered an abundance of data that by far outweighs the security constraints.

The third limitation concerns the confinement of the actor-driven security arena concept. While this made comparing characteristics of actors and processes of ordering security across arenas more straightforward, it also obscured most of the individual aspects of security in the local arena. Emotional, spontaneous, and unorganized aspects have thus remained underappreciated, although I did try to incorporate some of these aspects under the umbrella of fluid forms of ordering. The actor focus also overemphasized the role and perceptions of male individuals, because most of the security actors are led by men. I tried to explicitly seek out women’s representatives and conduct group discussions with women to gain their perspectives but overall their representation in the respondent pool remains far below their proportion in the local population.

Finally, choosing three countries with varying historical backgrounds and, within these countries, three different local arenas granted me some degree of ability to decipher ordering patterns across variegated contexts. Because the primary level of comparison is the local level (except for the national comparison in Chapters 2 and 3) the results of this study are not directly transferrable to other parts of the respective countries. The tense relationship between the populace and the South Sudanese government found in the three arenas studied, for instance, would most likely be even worse in places such as Bentiu[164] but better in a town such as Aweil.[165] The aim of this study is thus not to make generalizations on the three countries per se but on processes of ordering security that can be related to one another in the local arenas under scrutiny. In a second step, these findings can be theorized to hold value as explanations for ordering processes – not outcomes – in other local security arenas.

Local security studies have thus far been conducted mostly through concepts of areas of limited statehood, public authority, and oligopolies of violence. While each has its merits for studying either perceived security or local non-monopolized contexts or to make a comparison, none combines all three aspects. I, therefore, build on the established literature and relate it to my empirical insights to specify the security arena concept.

Analyzing nine local case studies across three countries calls for a concept that can frame data in a comparable manner. The security arena concept fulfils these requirements. The arena is made up of actors and their interactions. Actors range from more loosely described agglomerations, such as ‘the youth’, to highly structured institutions, such as the US army. An actor can broadly threaten or support another actor and mix actions to order the security arena on a spectrum ranging from fluid to stable ordering. The outcome of stable and fluid ordering can be security or insecurity for different parts of the arena’s inhabitants. I measure security as a subjective variable, as the felt durability of physical integrity.

I analyze larger processes of ordering the arena and their impact on security on a spectrum from fluid to stable: on the fluid end, actors vary their interactions frequently, making flexible adaptations possible; on the stable end, actors fixate hierarchies, making a change of ordering costly. Security through stable ordering is continuous but so is insecurity. Fluid ordering can avoid the institutionalization of insecurity but is also less adept at creating continuous security. In sum, stable ordering lends more predictability, while fluid ordering creates more modifiability.

Stable and fluid ordering are not equally feasible in every part of the arena at any point in time. Actors divide the arena into different spheres both in speech and through specific actions such as the erection of barriers at town entrances. The open space of the outer circle lends itself more readily to fluid ordering, while the constant interaction in the inner circle demands greater degrees of stable ordering among actors. On a higher level, the difference between an inner and outer circle is recreated by the relations between center and periphery.

I select cases within three countries that experience low degrees of state authority monopolization and a multitude of actors involved in security. Within each country, I choose three local arenas with varying attributes of socio-spatial distance to the capital, population size, and the varying presence of security actor types. I gather data through an explorative mix of methods such as focus group discussions, interviews, monthly reports, and observations. With the help of the CAS approach, I then trace processes within each local arena through the triangulation of my data sources, and subsequently compare the cases to find generalizable patterns of how different ways of ordering the security arena impact security.

The nine local security arenas under scrutiny here cannot be seen in temporal and geographic isolation. The histories of how actors in those arenas relate to varying centers on the national and international level have vast repercussions on the way ordering plays out in the local arena. In the next chapter, I turn to this broader historical and geographic context to establish why such marked differences exist between the three countries and turn to the local arenas themselves in more detail.


[1] Notes on observations, Mundri, South Sudan, 6 December 2014.

[2] Cf. Kalyvas 2008: 397.

[3] Whereas governance describes the ‘institutionalized modes of social coordination to produce and implement collectively binding rules, or to provide collective goods’, statehood ‘is defined by the monopoly over the use of violence or the ability to make and enforce central political decisions’; Risse 2011: 4, 9. See also Risse & Lehmkuhl 2007.

[4] Simojoki 2011.

[5] Schäferhoff 2014.

[6] An overview of public authority literature is given in Hoffmann & Kirk 2013.

[7] Bierschenk & Olivier de Sardan 1997.

[8] Boege et al. 2009.

[9] See Lund 2006.

[10] Wimmer 2008, 2013.

[11] See Vandekerckhove 2011.

[12] Mehler et al. 2010: 10.

[13] Branovic´ & Chojnacki 2011: 562.

[14] Lambach 2007; Mehler 2004: 545.

[15] Konrad & Skaperdas 2012.

[16] Mehler et al. 2010: 10–11.

[17] See Branovic´ & Chojnacki 2011: 560; Chojnacki & Branovic´ 2011: 97.

[18] Lambach 2007: 7.

[19] Cf. Long 1990.

[20] See Lind & Luckham 2017: 92.

[21] Cf. Galtung 1969; Lederach 1997, 2005; van Tongeren 2013.

[22] Rothschild 1995.

[23] E.g. Abrahamsen & Williams 2008: 540; Baker 2010: 599; Hills 2014a: 166.

[24] E.g. Chojnacki & Branovic´ 2011: 89; Mehler 2012: 50.

[25] E.g. Healy 2011; Joseph 2014; Mehler 2012.

[26] Adler & Greve 2009; Daase & Friesendorf 2010; Nathan 2006.

[27] Abrahamsen & Williams 2008, 2009; Krahmann 2008.

[28] Bagayoko et al. 2016.

[29] See Van Munster 2007.

[30] See Bajpai 2003; Owen 2004.

[31] Baldwin 1997.

[32] See Randeria 1999; Hönke & Müller 2012.

[33] Lind & Luckham 2017.

[34] Focus group discussion with Sultans, Raja, South Sudan, 27 November 2014.

[35] English translation: ‘A question of security arises .. . exactions by men in arms’, Interview with Catholic Abbot, Paoua, CAR, 24 February 2016.

[36] Focus group discussion with Women of Ilays women’s group, Zeila, Somaliland, 16 May 2016.

[37] English translation: ‘Security, it is the absence of trouble, of acts that create confusion among the population, to flee, to be worried about your life and family. To live in peace, there is nothing more important than peace’, Interview with Sub-Prefect, Obo, CAR, 8 February 2016.

[38] English translation: ‘Since 2003 there is no peace, there is a lot of disorder’, Focus group discussion with Youth, Paoua, CAR, 28 February 2015.

[39] Focus group discussion with Women, Buseri, South Sudan, 25 November 2014.

[40] Focus group discussion with Women, Daami, Somaliland, 25 April 2016.

[41] Focus group discussion with Youths, Zeila, Somaliland, 18 May 2015.

[42] Interview with Arab Sultan, Baligubadle, Somaliland, 30 May 2015.

[43] Interview with Acting Commissioner, Mundri, South Sudan, 8 December 2014.

[44] I thank an anonymous reviewer for this hint.

[45] Disorder exists when people feel there are no understandable avenues to seek security. This should not be confused with fluidity (or as many call it ‘fragility’), which is a particular ordering form that leaves avenues of seeking security open to negotiation between involved actors and individuals.

[46] Kasfir 1976; Hilgartner & Bosk 1988; Hagmann & Péclard 2010.

[47] Bierschenk & Olivier de Sardan 1997; Hills 2014a; Öhm 2014: chap. 4. Öhm’s is the only study of the six that does not explicitly use the term ‘arena’.

[48] Hilgartner & Bosk 1988.

[49] Kasfir 1976.

[50] Long 1990.

[51] Bierschenk & Olivier de Sardan 1997.

[52] Hagmann & Péclard 2010: 542.

[53] ‘Personae’ in Kasfir 1976. ‘Operatives’ in Hilgartner & Bosk 1988: 68–70. ‘Power brokers’ in Hills 2014a: 166. ‘Power poles’ in Bierschenk & Olivier de Sardan 1997: 443. ‘Local institutions’ in Öhm 2014: chap. 4.

[54] See Kasfir 1976.

[55] Öhm 2014: 215. See also de Vries 2011.

[56] Individual actions of key security relevance are also investigated.

[57] Long 1990.

[58] Meyer & Jepperson 2000. See also Förster 2015: 209–11.

[59] Arjona 2017.

[60] See the use of threats in the CAR: Lombard 2012 unpublished; Lombard 2016b.

[61] Numerous studies focus on why actors might choose to protect certain social groups in their bid to gain authority: Konrad & Skaperdas 2012; Mehlum 2002; Skaperdas 2001.

[62] On greed and grievance: Berdal et al. 2000; Collier & Hoeffler 2004. And select critiques: Cramer 2006: 166–7; Nathan 2005; Wimmer et al. 2009: 317–20.

[63] Richards 2005.

[64] Worral 2017: 709.

[65] Mkandawire 2002; Kalyvas 2006; Valentino 2014.

[66] Kalyvas et al. 2008: 1.

[67] Worrall 2017.

[68] Collins 2008; Collins 2012.

[69] See also Fearon & Laitin 1996.

[70] Cf. Giddens 1984.

[71] Cf. Helmke & Levitsky 2004.

[72] A good example of a local arena study in the Ivory Coast is Heitz 2009.

[73] Cf. Mac Ginty & Richmond 2013: 773; Meagher 2012.

[74] De Vries 2012; Vaughan et al. 2013.

[75] Justin & de Vries 2019.

[76] Kopytoff 1987.

[77] Ibid.: 10f.

[78] Mbembe 1996.

[79] Map data: Google, Digital Globe (2016).

[80] See Lefebvre 2014 [1932]: 157ff.

[81] Bhavnani et al. 2009.

[82] Luckham 2017: 114.

[83] Herbst 2000.

[84] Kopytoff 1987.

[85] Scott 2009.

[86] Schlichte & Veit 2007.

[87] Mkandawire 2002.

[88] Herbst 2004: 302.

[89] Herbst 2000: 145–6.

[90] Langholm 1971: 276.

[91] Kalyvas 2003: 486.

[92] Gidengil 1978.

[93] Scott 2009.

[94] Ibid.: 6.

[95] Cf. Doornbos 2010; Hagmann & Péclard 2010; Kasfir 1976: 161ff; Schomerus et al. 2013.

[96] Mamdani 1996.

[97] Leonard 2013.

[98] Braudel 1980.

[99] Bayart et al. 1999.

[100] Chabal & Daloz 1999. On why their theory does not hold up in the CAR, see de Vries & Mehler 2019.

[101] See Baberowski 2015.

[102] E.g. Mengisteab & Daddieh 1999; Fukuyama 2004; Call & Wyeth 2008.

[103] Cf. Lund 2006: 686.

[104] For the example of Chad: Debos 2011; Debos 2016.

[105] I also heed the call of Niagalé Bagayoko et al. who lament a shortage of research on ‘official security institutions, including the informal networks around them, and the complex ways they interface with non-state actors’ (2016: 20).

[106] See, for example, Marielle Debos’ and Joël Glasman‘s work on the fluidity of state security forces: Debos & Glasman 2012: 21.

[107] Boege et al. 2009: 17.

[108] Meagher et al. 2014; Albrecht & Moe 2015.

[109] Lund 2006.

[110] Baker 2010.

[111] Leonard 2013: 1.

[112] Meagher 2012: 1074.

[113] Cf. the relation between hybrid governance and security: Bagayoko et al. 2016; Luckham & Kirk 2013.

[114] Utas 2012: 6. Emphasis in original.

[115] See Utas 2012.

[116] See Reno 2002; Reno 1998.

[117] See Hoehne 2011; Renders 2012; Schlee 2013; Simojoki 2011.

[118] Mac Ginty 2010a: 349f.

[119] Jackson 2003.

[120] Mampilly 2011; Péclard & Mechuloan 2015.

[121] Mampilly 2011: 36.

[122] Arjona et al. 2015.

[123] Hobsbawm & Ranger 1983; Hoehne 2011; Logan 2013; Molomo 2009.

[124] ICISS 2001.

[125] At least south-central Somalia, not Somaliland.

[126] Kühn 2011.

[127] See Lijphart 1969; Wolff 2012.

[128] Tull & Mehler 2005; Mehler 2013.

[129] Autesserre 2010: 22–3. See also: Veit 2010.

[130] Cf. Mac Ginty 2012.

[131] See Mac Ginty 2010b.

[132] Whalan 2017.

[133] Duffield 2010; Fisher 2017.

[134] I thank Dominik Balthasar for this hint. Parts of this section were published at an earlier stage as a working paper: Glawion 2017.

[135] See Ahram 2011; Mehler & Hoffmann 2011; Ahram et al. 2018.

[136] Basedau & Köllner 2007: 113.

[137] A sole exception on Daami is Vitturini 2017.

[138] See most prominently the writings by Markus Hoehne: Hoehne 2011; Hoehne 2015.

[139] See Blocq 2014; de Vries 2015.

[140] E.g. Carayannis & Lombard 2015; Lombard 2016a; Marchal 2016.

[141] Cakaj 2015; Spittaels & Hilgert 2009; Chauvin & Seignobos 2014.

[142] Gerring 2008: 645.

[143] See the Fragile States Index: Messner 2017; the indices of the Organization for Economic Co-operation and Development (2015) ‘States of Fragility 2015’:; And Monty G. Marshall (2017) ‘State Fragility and Warfare in the Global System 2016’:

[144] Messner 2017. See online data center for rankings on all measured years:

[145] See Glawion et al. 2019.

[146] E.g. Ansorg 2017; Mac Ginty & Richmond 2013; Mehler 2009a; Schroeder & Chappuis 2014.

[147] Heitz 2009: 112; Pouligny 2009: 5–9.

[148] As Jeremy Lind and Robin Luckham suggest, margins ‘might exist in distant borderlands as well as the neighborhoods of capital cities’, depending on social and political factors (2017: 93).

[149] See Moore 1973: 722.

[150] See Bourdieu 1985a: 729; Bourdieu 1985b.

[151] Martin 2003: 24, 36.

[152] See Martin 2003: 25.

[153] See Cramer et al. 2011; Malejacq & Mukhopadhyay 2016; Read et al. 2006.

[154] Kapiszewski et al. 2015.

[155] See Rathbun 2008.

[156] See Hennink 2007; Ryan et al. 2014.

[157] Cf. Brady 2008; Gerring 2008: 668–70; Mahoney 2000: 399–402.

[158] Cf. Basedau & Köllner 2007: 119; Gerring 2008: 650–67.

[159] See Ahram 2011.

[160] See Lund 2014.

[161] Cf. benefits of process tracing to covariation in comparative politics: Hall 2003.

[162] See Bennett 2008; Bennett & Checkel 2015.

[163] See Jacobs 2015.

[164] See atrocities committed by government army and government-allied militias: ICG 2015: 15–16.

[165] See the positive image of the SPLA in Aweil: Gorur et al. 2014.


Next Chapter

The book is worth buying: Available from De Gruyter

This site uses Akismet to reduce spam. Learn how your comment data is processed.