‘With or Without You’: The Governance of (Local) Security and the Koglweogo Movement in Burkina Faso

In late 2014 and after more than two decades of a ‘semi-authoritarian’ regime, a popular insurrection in Burkina Faso led to the fall of Blaise Compaoré, president and leader of the ruling party. Due to — or parallel to — the political transition, factors of insecurity developed or were amplified, leading to a reconfiguration of the provision of security at two levels. At the central state level began a reflection around the governance model of security and the improvement of the practices of state security forces. At the local level, non-state security initiatives have multiplied. Drawing on insights from the study of local security provision and providers in the town of Tenkodogo, located in the Boulgou province (Centre-East region), and on its wider integration into the national framework and response to insecurity in Burkina Faso, this article raises and investigates three major questions. First, how is the governance of security (co)produced by (state and non-state) actors in a specific local configuration in Burkina Faso? Second, in what ways does this local experience compare with the state’s response to insecurity and with the nationwide expansion of the Koglweogo movement? Finally, what new perspectives can such reflection at the local and national levels offer to overcome the limits of current approaches regarding local security?


Introduction
In late 2014, after more than two decades of a 'semi-authoritarian' regime, 1 a popular insurrection 2 in Burkina Faso led to the fall of Blaise Compaoré, president and leader of the ruling party Congrès pour la Démocratie et le Progrès (Congress for Democracy and Progress) (Hagberg et al. 2018). What has followed is a complex political transition, essentially marked by the installation of a transitional presidency and government in November 2014, 3 by a failed coup in September 2015, 4 and by the presidential and legislative elections held in November 2015, won by Roch Marc Christian Kaboré and his party, the Mouvement du peuple pour le progrès (People's Movement for Progress).
Due to -or parallel to -the political transition, three major factors of insecurity developed or were amplified: organized and petty crime, armed groups, and terrorism. The first one mostly takes the form of property theft and the coupeurs de route phenomenon, literally 'road blockers,' armed gangs attacking automobiles and stability motorcycles. Their actions range from theft to kidnapping, rape, or murder. This criminality has been facilitated in particular by the widespread presence of small arms coming from the 2011 mutinies (Chouli 2012), the porosity of the country's frontiers, and the weak capacity of state security forces (SSFs). The second factor is more closely linked to Compaoré's loss of power. Burkina Faso was previously a refuge to Malian and Ivorian rebel movements and the fall of its regime 'seemed to put an end to a kind of "tacit accord between Burkina Faso and different rebel movements" ' (Hagberg et al. 2018: 58). Finally, Burkina Faso has endured a series of terrorist attacks, with responsibility claimed by jihadist groups. Initially, they took place mainly in rural areas, such as the northern part of the country and in the heart of Ouagadougou -for instance in January 2016 (targeting the Splendid hotel and Cappuccino restaurant), in April 2017 (targeting the Bravia hotel and Aziz Istanbul restaurant), and in March 2018 (targeting the French Embassy and Burkina Faso's military headquarters). These attacks in the capital drew national and international media and public attention at the time, leaving the raising insecurity in the north more in the shadow. Yet, since mid-2018, jihadist and other armed groups have increased their presence and activities against civilians and authorities in larger parts of the territory, especially in the north and the east. The security situation has deteriorated rapidly. Insecurity now affects almost all border areas, including along the coastal neighbouring countries in the south.
The transitional and (in)security context initiated a reconfiguration of the provision of security at two levels: at the central state level, a reflection on the governance model of security (historically largely modelled on the French security architecture) and the improvement of the practices of SSFs (PARSIB 2018); at the local level, non-state security initiatives (LSIs) have developed from localized groups into a much more widespread movement -most notably the Koglweogo groups. These self-defence groups aim to fill the vacuum left by the state in protecting the people, animals, and natural resources of their local community (Quidelleur 2017). To do so, however, they patrol, arrest, investigate, fine, imprison, or even torture alleged criminals, giving rise to debates and controversies over the groups' practices and even their mere existence (Hagberg 2018).
Against this background, this article investigates three major questions. First, how is the governance of security (co)produced by (state and non-state) actors in a specific local configuration in Burkina Faso? Second, in what ways does this local experience compare with the state's response to insecurity and with the nationwide expansion of the Koglweogo movement? Finally, what new perspectives can such reflection at the local and national levels offer to overcome the limits of current approaches regarding local security? Underlying these questions are the scholarly debates around the categorization of non-state security actors (Hendriks 2018), their legitimacy as a provider of security, the tensions emerging from their practices as well as the role of negotiation processes between state and non-state security actors in the reconfiguration of the state (Péclard and Hagman 2010).
This article draws on field research data collected in February 2018 on local security provision and providers in the town of Tenkodogo, the capital of the Boulgou province (Centre-East region). 5 With about 50,000 inhabitants, Tenkodogo is a key regional market positioned on an economic axis between Ouagadougou and the border with Ghana and Togo. At the time of study, the insecurity linked to jihadist and other armed groups was still perceived as distant. Organized crime and petty crime were seen as the main challenges in a generally more secure local context than in other parts of the country. The emergence of a Koglweogo group in such understudied settings, where insecurity was then perceived as rather low, sheds an original light on LSIs dynamics and their anticipated effectiveness and legitimacy. Leclercq and Matagne: 'With or Without You' Art. 4,page 3 of 22 Our research methodology has been designed around semi-structured interviews in Tenkodogo (February 2018) on the perception of (in)security, the providers of security services, and the relations and interactions with and between these providers. Interviews were conducted with all SSFs present in Tenkodogo (National Police and Municipal Police, Gendarmerie, Customs, Rangers, Army) with non-state security actors (six members of the Koglweogo and Dozos 6 self-defence movements, even if the latter is not active in Tenkodogo itself) as well as with political authorities ( Integrating results emerging from the specific case of Tenkodogo into the national framework and response to insecurity in Burkina Faso, our research yields four findings. First, in Tenkodogo, the SSFs (mainly the National and Municipal Police and the Gendarmerie) are perceived as legitimate but largely ineffective actors of governance, while the local Koglweogo group is also perceived as legitimate but largely because it is seen as effective. Second, although important tensions exist between those actors, there still are ambiguous collaborations between them and in fine an 'imperfect coproduction' (Gautron 2010) of local security. Nonetheless, what may be true for Tenkodogo at a certain point in time should not be generalized across the country, especially in a complex and rapidly changing security context (section 'Governing Security in Tenkodogo'). Third, and as a consequence, a nationwide state-centred approach -often promoted by the international community or the national state security apparatus -appears neither achievable nor without political and security risks. Moreover, a nationwide expansion of the Koglweogo movement would produce significant tensions affecting both the country's social cohesion (e.g., inter-ethnic and inter-community relations, relations with pre-existing LSIs, especially Dozo, urbanrural fracture) and state-society relations (Koglweogo-SSFs tensions, parallel governance structures) (section 'From Local to National: Towards a (Non-)state Monopoly of Legitimate Violence?'). Fourth, we argue that turning the focus to the ' end-user' of security services (Denney 2014) to address both local security and SSR and thus engage, where relevant, with non-state actors, would not only overcome the limits of state-centrism and state-rejection approaches in providing greater security to citizens but also reinforce state-society relations and instil greater coherence into the local-national nexus. To that end, we provide a series of concrete suggestions from which authorities and external donors can build on (section 'Conclusion: Building on Sources of Legitimacy to Operationalize an End-user Approach to Local Security').

Governing Security in Tenkodogo
The inductive reasoning of our approach to local security governance requires a preliminary clarification on two fronts: what does governance mean and how do we approach it? These elements are central to our exploration of security service providers in Tenkodogo, relations between those actors, as well as an openness to a multi-actor, pluralist understanding of security provision.
For more than two decades, governance, which we define here as the 'institutionalized modes of social coordination to produce and implement collectively binding rules, or to provide collective good' (Risse 2011: 9), has been at the heart of the democratization and statebuilding agenda in developing countries (Chandler 2010;Wesley 2008). Effectiveness and legitimacy constitute essential components of, as well as ways to assess, the quality of that governance (Schmelzle 2011;Brinkerhoff 2005). For the purpose of this article, effectiveness broadly refers to the level of intentional (perceived) 7 achievement of the goals sought (Etzioni 1964), thus not its efficiency (i.e., its net performance) nor the beneficial or harmful nature of its (un)intended effects (Underdal 2004). In turn, legitimacy refers to the level of acceptance, by the governed, of the governor's right to rule (Gilley 2009). As we will see, however, the origin of this legitimacy may differ and produce input, output (Scharpf 1999), or throughput (Schmidt 2013) legitimacy. Although these two components -effectiveness and legitimacy -are key to the quality of governance, we do not presume a specific relationship between them (Schmelzle and Stollenwerk 2018;Schmelzle 2011;Levi and Sacks 2009). Rather, we assess these two dimensions separately as they are, and explore our way back to their relation, or lack thereof.
Traditionally, the analysis of governance has been associated with the state and its apparatus, especially for the provision of security, considered its (exclusive) prerogative (Baker and Scheye 2007). More recently, however, the focus has begun to shift to other (non-state) actors. It is increasingly recognized that what have previously been labelled as 'ungoverned spaces' in developing countries are indeed governed, if not by the state, then by other actors (Keister 2014) and that there is a heuristic and operational potential in approaching these actors with a similar governance lens. The literature on the network or nodal governance of security echoes these concerns. This approach acknowledges the (increasing) ' diversity of auspices and providers of security' (Shearing 2005: 58) and underlines the importance of taking the plurality and agency of these actors seriously (Johnston and Shearing 2003;Dupont et al. 2007;Price and Warren 2017;Biaumet 2017). As such, not only does it provide for a widening of the analysis to non-state actors, but it also allows for a cross-examination of the governance of the security network and the relations between all security actors.

State security forces: legitimate but ineffective
Territorial coverage of SSFs is a major issue in Burkina Faso, not only for its lack thereof but also for its incoherence (Hagberg et al. 2018;PARSIB 2018;CGD 2017). For instance, in the Centre-East region where Tenkodogo is located, half of the 30 departments have no state forces, and the other half often have both a Police station and Gendarmerie on their territory. 8 Despite its centrality in national strategies or donor program (European Commission 2018; Ministry of Economy and Finance 2016), the mapping of the territorial presence of the SSFs themselves (and public transparency around it) has thus remained a sensitive topic as it pertains to the different ministerial tutelages (there are seven different SSFs and they each answer to a separate minister), ongoing tensions between forces, and resistance to potential reforms such as the merging of the National Police and Gendarmerie (cf. infra). Tenkodogo, as a town, department, and capital of the Boulgou province, does not, however, fall short of SSFs presence. The following forces are represented in Tenkodogo: 9 National Police (approximately 250 officers), municipal police (14 officers), Gendarmerie (approximately 300 officers out of 500 for the whole Centre-East region), customs (42 mobile officers covering the Tenkendogo-Ouargaye area), rangers (none specifically assigned to Tenkodogo but available upon request, 10 for instance, on sanitation issues), and the army (undisclosed number of officers). The presence of a last 'security' force should be mentioned despite its lack of enforcement capacity, the Volontaire Adjoint de Sécurité (Voluntary Deputy Security Officer). These officers, around 300 for the Boulgou region, are mostly supporting police forces for traffic-related missions and, at times, for administrative tasks such as encoding data or filling out forms.
The area was not faced, at the time of the field work, with high levels of insecurity. Security threats did not include -or rarely included -terrorism or major organized crime, unlike other parts of the country (such as the north or border regions). Rather, petty crime, bandits, and coupeurs de route represented the major sources of perceived insecurity in the area. Yet, despite the concentration of SSFs and a perceived low level of insecurity in Tenkodogo, the interviews conducted, including with SSFs themselves, largely converge towards a perception of ineffectiveness in the conduct of their mission. Reasons provided for this ineffectiveness include insufficient human and financial resources, high levels of corruption, lack of motivation, complacency or complicity with criminals, and tensions between forces, especially between the Gendarmerie and the National Police. These factors lead to a slow or non-existent response to and poor governance of local security situations, as well as a reluctance of the population to cooperate with SSFs, fearing reprisals from bandits. Nevertheless, although SSFs are deemed ineffective, they are considered legitimate as governance actors of local security. Legitimacy is bestowed on SSFs by their nature and mandate as state actors. Thus, they benefit from what Scharpf (1999) refers to as an input legitimacy, focusing on the sources of legitimacy, and raising expectations they do not fulfil, which in turn leads to a low level of output legitimacy. While criticized (or even replaced, see infra), SSFs are, however, never dismissed as the key presumed provider of local security services. As such, the (in)effectiveness assessment is based upon the expectations generated by their legitimacy. This connection between legitimacy and ineffectiveness of SSFs is critical both to understanding the nature of local security governance by non-state actors as well as to envisioning a multi-actor provision of security services.

Koglweogo: legitimate because (presumed) effective
Already existing in several peri-urban and rural areas since 1996 (Government of Burkina Faso 2010), the Koglweogo phenomenon has greatly intensified since 2015 as a reaction to the rise of insecurity. Literally translated as 'protector of the bush land' in Mooré, the language of the majority ethnic group Mossi, the term Koglweogo has come to define 'the sum of measures and strategies to undertake to secure community life in rural areas' (Hagberg 2018: 27;Kouraogo and Kaboré 2016). In Burkina Faso's rural areas, 80 per cent of the population lives off livestock and agriculture, rendering the protection of goods and cattle essential to buy food, pay for the education of children, or access basic health care (La Libre 2018). Although regularly (and pejoratively) described as militia groups -or militia groups in the making -in the public debate (e.g., Sankara 2016; Carayol 2016), their self-claimed focus on the security of and justice for the population brings them closer to the concept of 'vigilantism,' as asserted by Quidelleur (2017). While the use and meaning of the term itself has evolved (see Favarel-Garrigues and Gayer 2016), the Koglweogo displays the core characteristics of a vigilante group, namely their non-state nature and their resort to collective coercive practices in order to enforce certain norms and/or exercise 'justice,' with the latter mostly referring to punishment but which can also evoke, among vigilante members and their audience, a societal ideal (Favarel-Garrigues and Gayer 2016: 17; Baker 2008: 81). While diverse, Koglweogo groups share the common objective of 'taking matters into their own hands' by patrolling, investigating, pursuing, and arresting alleged criminals as well as by fining, imprisoning, beating, or torturing them. Likewise, although their financing structure has or may evolve in certain areas, for instance through the 'privatization' of their services, Koglweogo groups mostly live upon the fines imposed on thieves to pay for the gas used to pursue them and ropes to tie them up (Carayol 2016). Nevertheless, as is often the case with non-state security actors, we should not elude the complexity of the phenomenon and lose sight of its multi-dimensional and dynamic nature, which defies simple categorization (Hendriks 2018). As such, the Koglweogo and their repertoire of actions fluctuate between different categories of security actors -community vs. commercial nature of services; local vs. localized (or franchised) initiative; crime fighter vs. criminal; accredited vs. illegal local group; support to SSFs vs. rebel or militia group, etc.
Created in November 2017, the Koglweogo group for Tenkodogo town counted 180 members in February 2018, eight of them women. 11 Unsurprisingly, the creation of the group is mostly justified by the failure of the state to provide security and justice for the members of the community. A member of the group also invoked a territorial coverage claim: 'if we do not create [a group], it becomes a bandit's haven'. 12 This claim is interesting as it implies the apparent necessity for the group to achieve a form of territorial coverage similar (and parallel) to the one (supposed to be) provided by SSFs. Also, the group's formation in Tenkodogo runs contrary to the usual expectation that their creation is linked to the absence of SSFs (Da Cunha Dupuy and Quidelleur 2018).
Governance of local security by the Koglweogo in Tenkodogo reveals a complex picture of intertwinement between (perceived) effectiveness and multiple levels of legitimacy. On the one hand, interviews, including with SSFs, all recognize the group's effectiveness in fighting petty crime and bandits, and attribute a reduction of the (already rather low as compared with other areas) crime rate to the group's presence, and through this diminution, a renewed economic dynamism. When a theft occurs, intervention is quick, the presumed thief is arrested and sanctioned, and the owner recovers the item (or its value). There is a sense of quick retributive justice. As such, the group's legitimacy seems to be based largely upon its members' performance as governance actors, thus concurring to Scharpf's output legitimacy. On the other hand, the relatively low level of initial insecurity factors in Tenkodogo and the recent creation of the group call for a closer scrutiny of this effectiveness-legitimacy connection. Indeed, other factors also appear to be shaping the relationship between the group's effectiveness and its legitimacy. First, while it is clear that effectiveness is a key driver of the group's legitimacy, it is not clear whether it should be attributed to the Tenkodogo-based group or the Koglweogo movement in general. In other words, its legitimacy also seems to be the by-product of the reputation and anticipated performance of the movement, what Cord Schmelzle and Eric Stollenwerk refer to as 'benefit-of-the-doubt cooperation' (2018: 458). Thus, an input form of legitimacy.
Second, its effectiveness-legitimacy nexus also seems to be influenced by throughput patterns of legitimacy. Mostly mobilized in the context of the European Union (EU), this concept of throughput legitimacy refers to the ' quality of governance or of decisionmaking processes' (Iusmen and Boswell 2017: 459-60), whereby quality is characterized by ' efficacy, accountability, transparency, inclusiveness and openness to interest intermediation' (Schmidt 2013: 6). This concept helps us understand the 'how' of local security governance by the Koglweogo in Tenkodogo. The group mobilizes a quadruple anchorage allowing for a transition from an appealing anticipation over the performance of a non-created group (input) to its locally based creation, functioning, and performance (output): • At the societal level, the group is made up of members of the local community, 'Koglweogo have succeeded because they know both the men and the field. They know who does what in each town.' 13 • At the customary level, relations with traditional authorities are key to the implantation of the group. To be created, they had to be 'sanctioned' by the customary king. 14 • At the religious level, the group has been endorsed by all religions (their practices are 'inspired by the Bible and Quran', 15 with a special proximity with the Muslim faith). • At the traditional level, some of the Koglweogo's practices and their staging have an important witchery dimension (Quidelleur 2017: 66-69).
This overall anchorage and closeness within the community greatly fosters shared social goals and values (Schmelzle and Stollenwerk 2018: 15), even for the sensitive and often contested areas of security and justice (Ciorciari and Krasner 2018). As a leader of a religious community summarized, 'the Koglweogo sleep with the people, they are our brothers. It is the very heart of the population. They know each other.… the Koglweogo are the people, they're not a separate body.' 16 Although it appears paradoxical, another aspect of the group's throughput legitimacy originates in the mimicry of codes -e.g., outfits, membership cards, a certain degree of bureaucratization, practices (e.g., handling of weapons, criss-crossing the territory, interrogating suspects), or vocabulary belonging to institutionalized security forces (Quidelleur 2017: 58-61). Located in what Bhabha (1994: 86) calls the ' almost the same, but not quite,' this mimicry allows the Koglweogo group to both identify itself as a provider of security and differentiate itself from SSFs. Interestingly, this 'hybridity' is also key to understanding the (national) debate around the Koglweogo's practices. In fact, despite a widespread consciousness of the unacceptability of some of the Koglweogo's practices in regard to human rights and the rule of law (lack of due process, torture, etc.), 17 these are often counter-balanced in the discourse of local actors by two overlapping arguments: a 'lesser evil' claim (as one interviewee from a CSO notes, 'we thus must choose the lesser evil but at least they are effective' 18 ); and a rights-based hierarchy claim, summed up in the newsletter of the major Burkinabe's human rights defence organization: ' every human rights defender should keep in mind that security is a human right,' (Mouvement Burkinabé des Droits de l'Homme et des Peuples 2016: 6). As is often the case with vigilante groups, certain laws are violated in order to protect others. Consequently, a specific hierarchy of rights is produced based on a differentiation between rights that have to be protected and those 'for which the infringement is made necessary by the legitimacy of the fight led' (Favarel-Garrigues and Gayer 2016: 23).

Between tension and cooperation
Our analysis of the governance of local security in Tenkodogo reveals an ambiguous yet critical relationship between SSFs and the Koglweogo. First and foremost, as hinted by Fourchard (2018), vigilante groups do not exist in a security vacuum. Rather, they have to be examined and understood in relation to the state and its forces. Even if ineffective, the 'idea of the state' (Abrams 1988) remains strongly anchored in social imaginary of both state and non-state actors. As Hellweg (2012) concluded with regard to the Dozo in Côte d'Ivoire, to examine Koglweogo groups is to examine the state. And, in fact, the justification provided by actors in Tenkodogo for the group's existence always gravitates around the (in)capacity of the state to provide security and justice for the local community. In other words, the local group is not created because there is insecurity and injustice, it is created because the state does not properly fulfil its role as provider of security and justice. The creation is therefore as much a way to respond to insecurity as it is a way to call out the state for not fulfilling expectations.
On the one hand, the existence and practices of the Koglweogo group generate predictable tension with the SSFs. While the local security initiative is based upon the SSFs' ineffectiveness, the latter perceives it as unlawful competition. SSFs interviewees rarely insisted on a normative assessment of the Koglweogo's practices as such but rather delegitimized the group using national or international rights-based arguments (e.g., 'they do not want to respect regulations' said a member of the Gendarmerie; 19 ' our country has signed international human rights conventions' stressed a National Police officer 20 ). Thus, the framings of the mutual delegitimization process are opposed, the Koglweogo focusing on the lack of output legitimacy (they do not perform), and SSFs focusing on the lack of input legitimacy (they have no right to do what they do).
On the other hand -and despite these tensions -the shared objectives and spatial and social territory as providers of security at the local level has led to different forms of cooperation between the local Koglweogo group and the SSFs based in Tenkodogo. For instance, the National Police forces often rely on the Koglweogo for information ('when we have a situation, we call them for information,' 21 said a National Police officer) and reporting of incidents ('they contribute to signalling [incidents],' 22 according to a member of the Gendarmerie). Likewise, when the Koglweogo have finished their procedures, the person they have arrested and punished is often handed over to the National Police or Gendarmerie so that 'they do the justice proceedings.' 23 Far from a univocal relationship and apparent polarity, the local security initiative and the SSFs thus also legitimize one another through these collaborations.
As a result of this dual relationship, we argue that SSFs and the Koglweogo contribute to what Gautron (2010) defines as an 'imperfect coproduction of security' at the local level in Tenkodogo. As explained by a staff member of a decentralized administrative authority, 'local security initiatives are needed to access the territory, they are relays at the local level… it is more of a complementarity, a conjugation'. 24 Indeed, there are real partnership modalities between actors, as informal as these may be, around the general common objective of reducing insecurity for the community. However, ' divergences in representations, values, standards of actions and interests' limit the cooperation to 'sporadic exchanges and punctual collaborations' (Gautron 2010: 13).

From Local to National: Towards a (Non-)state Monopoly of Legitimate Violence?
While acknowledging the specificities of the context of Tenkodogo in early 2018, 25 this state of play at the local level provides interesting grounds for a wider reflection, at the national level, on the governance of (local) security in Burkina Faso. In particular, how are local realities and security arrangements integrated into the national responses to insecurity by both state and non-state actors? In this section, we investigate these nationwide frameworks of insecurity response, first by the state and its security apparatus -for which we highlight its state-centrism -and second by Koglweogo groups -for which we highlight the effects of its expansion on social cohesion and state-society relations.

The unattainable monopoly of legitimate violence: an incidental state-centrism?
The rising insecurity, the ineffectiveness of SSFs, the political transition, and the dissolution of Compaoré's military presidential guard renewed a reform agenda on -and international support for -how security was governed and provided in Burkina Faso. Two major lines of reform can be distinguished. First, the need for structural reforms. Presented as an initial step towards a complete SSR, a National Forum on Security was organized in late 2017. Its report highlights seven major areas: the need for a national security policy and strategy, sectoral strategies and implementation plans; the need for an anti-corruption strategy; the need to consolidate the governance of the security sector (e.g., revitalization of community policing, unification of Police and Gendarmerie forces); the need for a strategy on fighting terrorism and radicalization; the need to intensify the security aspects in the programs for the Sahel region; the need to improve the operational capacity of SSFs; and the need to consolidate trust between SSFs and the population (PARSIB 2018: 5 Simultaneous to these reflections are the conventional trainand-equip programs sponsored by external donors, including the EU (the training component of the PARSIB project focusing on the fight against terrorism) or France (through its police and military cooperation). Common to all these reflections and programs is their state-centric gravitation. The responses to insecurity are largely thought by the state, for the state. As in many other SSR contexts (see Baker and Scheye 2007; Denney 2014), statecentric logics are fiercely anchored in the national and international reflections, studies, indicators, 26 or programs regarding the governance of security in Burkina Faso; however, this state-centric response to insecurity faces numerous challenges. The increasing level of insecurity -especially the terrorist threat (ICG 2019) -the limited human and financial means devoted to the security sector, the economy of corruption surrounding SSFs, the complexity and sensitivity of a wide SSR, the resistance to in-depth change in the governance of security as well as the tensions and low level of trust between SSFs and the population and between different components of the SSFs (Kibora and Traore 2017; CGD 2017; Harsch 2017: 232-33) all render the effective state monopoly of legitimate violence across the national territory highly unlikely in the short and medium term.
Despite this state-centrism, local security initiatives such as the Koglweogo are not absent from reflections on security governance. Nevertheless, if their existence and importance as security service providers at the local level are largely recognized, 27 LSIs are mainly framed and dealt with as threats to security in policy or programmatic discussions. A striking example of this paradox is apparent in the differentiated framing of LSIs in the two major sections of the National Forum on Security report published in April 2018 -the summary of the discussion and the analytical synthesis. In the summary of the discussion, LSIs, self-defence movements, or the specific phenomenon of the Koglweogo were mentioned either in relation to the 'need to involve LSIs in the fight against terrorism' (PARSIB 2018: 19), in relation to the 'role played by the Koglweogo in the production of security,' the necessity to 'have a candid collaboration between the Koglweogo and SSF in order to provide a better security to local populations' (ibid.: 21), or in relation to the strengths, weaknesses, and ways towards a better framing and monitoring of LSIs (ibid.). In contrast, the analytical synthesis of the discussion presents these actors as the second major threat on national security after international terrorism -'proliferation of self-defence groups and their frequent indelicate behaviours' (ibid.: 30). Likewise, out of the 52 recommendations drafted at the end of the general report, only one concerns LSIs -the need for a supervision plan of these initiatives (ibid.: 37). As such, the report reflects a 'restrict and control' nature of the approach adopted towards these non-state security actors. While LSIs are not always excluded from programs, they are apprehended as peripheral actors to be supervised, inducing an ad hoc and coercive approach to their treatment. 28 They are not formally perceived as legitimate security actors to be accompanied or trained towards a better respect for human rights and the rule of law.
This restrict-and-control approach is nonetheless softened by the political ambiguity surrounding LSIs, especially the Koglweogo. Their equally strong popularity and denigration in certain areas or among certain communities (cf. infra) make their existence and practices a polarizing issue in the public debate (Hagberg 2018: 27-28; ASHA 2018: 4; Kouraogo and Kaboré 2016: 4). Especially in the context of the 2016 municipal elections, political actors avoided taking a strong stance either against the Koglweogo -which may have alienated the groups and their supporters, especially in rural areasor in favour of them. Likewise, state authorities have often had a benevolent attitude towards the movement (Da Cunha Dupuy and Quidelleur 2018). Sometimes called the 'supreme leader,' 'president,' or 'father' of the Koglweogo, the former Minister for Internal Security, Simon Compaoré (January 2016-January 2018), was notably seen as an ally to the movement, providing it with implicit support (Dabiré 2016;Kaboré 2016). The consequence of this ambiguity is a lack of real commitment by the state -on top of its likely inability -in applying either the rule of law in general when it has been violated by the Koglweogo or the specific legislations regarding LSIs. 27 This ambiguity thus translates a certain complacency (Kibora and Traore 2017: 17) beyond the 'public transcript' (Scott 1990) of restriction and control.
As the title of this section suggests, the state-centric nature of the response to insecurity is, however, incidental. If the expansion of the Koglweogo movement is currently largely portrayed as a threat, their emergence was initially largely apprehended positively -and encouraged -by state authorities. A striking example is the 2010 National Internal Security Strategy, which only refers positively to the Koglweogo: '[Their] creation is based on the necessity [for the population] to survive… The main strengths of the "Koglweogo" are: popular legitimacy; solidarity; the self-supported nature of their actions…' (Government of Burkina Faso 2010: 18-19). An interview at the Directorate of Community Police corroborates this perception: 'What really motivated us in the beginning was their autonomy.' In other words, state authorities initially saw the phenomenon as an opportunity to ' discharge' 29 the provision of local security to these actors (Hibou 2004; see also Fourchard 2018). In December 2019, the president validated a draft legislation that would allow for the recruitment of civilian volunteers to reinforce the security in areas affected by terrorist attacks, perceived by the Koglweogo as an encouragement to further contribute to the fight against terrorism (Malagardis 2019). It was unanimously adopted by the National Assembly in January 2020 (Douce 2020). This ambiguous and oscillating relationship between state authorities and the Koglweogo should also be understood through the lens of the historicity of nonstate participation to local security provision. Koglweogo groups are indeed neither the only current form of local security agencye.g., Dozo, 6 which have been mostly present in the western part of the country (Hellweg 2012) -nor a recent phenomenon -e.g., 'vigilance committees' in Bobo-Dioulasso (1970, Committees for the Defence of the Revolution (1983Revolution ( -1987 30  Since the beginning of the complex political transition, Koglweogo groups have multiplied and the phenomenon greatly expanded across the national territory. Initially mostly present in rural areas of the -mostly Mossi -Central Plateau where Police or Gendarmerie forces were absent, the movement has expanded to almost the entire national territory, with the exception of the Great West and the Cascades regions (Kouraogo and Kaboré 2016;Kibora and Traoré 2017;Da Cunha Dupuy and Quidelleur 2018;Hagberg 2018). In a very short time, the model has spread by contagion beyond its initial configuration and Koglweogo groups have been established in (peri-)urban areas (including in Ouagadougou), in mixed ethnic zones (e.g., predominantly Gourmanché or Gourounsi areas) and in places where Police and/or Gendarmerie station do exist, as our case of Tenkodogo illustrates. In 2018, it was estimated that 4,400 groups had been created (La Libre 2018). This expansion, however, stumbles upon internal and external tensions undermining its sustainability as well as the social cohesion and the state-society relations in the country.

Internal challenges of the expansion and durability of the movement
The rapid growth of the movement has led to a -largely tentative -hierarchical structuring of the scattered groups around an organizing and coordinating leadership at different echelons up to the national level with the president of the Koglweogo association. While these structures ensure the diffusion of the model, relay with state authorities, and strive to resolve internal disputes, they have not successfully prevented or solved the multiple challenges to the expansion and durability of the movement that have emerged.
Three major factors of division or unsustainability are important to highlight. Firstly, internal dissentions have emerged both between Koglweogo groups and between the leadership and grassroots members over the resort to certain practices and the compliance with state demands regarding their control. Two Koglweogo spirits or schools are sometimes compared to illustrate the main fracture lines. The spirit of 'Mané' -named after the Koglweogo's original location in the 1990s -mostly promotes an auxiliary role of the group whereby it only apprehends bandits and hands them over, along with seized items, to the Police or Gendarmerie. These are often recognized by state administrative entities and granted a permit to practice. The spirit of 'Zorgho' -named after the area where the Koglweogo have spread from after 2015 and largely representative of the general movement at the moment of the field study in 2018 -represents groups that do not hesitate to bear arms, inquire, arrest, imprison, torture, impose fines, collect taxes, or carry out roadside checks, deputizing for the state security and judiciary apparatus in the process (Soulga 2016;Sawadogo 2017;Aicardi de Saint-Paul 2016;Carayol 2016). For the latter, weapons are necessary to apprehend bandits, fines necessary to pay for the gasoline consumed during their missions, and physical abuse necessary to prevent bandits from repeating their crimes (Dabiré 2016). The national leadership's attempts at pushing groups to adopt practices compatible with human rights and the rule of law or comply with the LSI legal framework have thus been met with great resistance, challenging their authority (Kouraogo and Kaboré 2016).
The second factor of internal division is the distinctive local characteristics that undermine the unification of a national Koglweogo movement (Da Cunha Dupuy and Quidelleur 2018: 9-10). An example is the change of name by Koglweogo groups in Fada-Ngourma, in the east of Burkina Faso. In order to highlight their identity and cultural distinctiveness from the Mossi, Koglweogo groups in predominantly Gourmanché communities decided, in late 2016, to rename their movement and unite themselves under the association 'Tin Kubi u gogu', meaning 'let's protect the city,' As Da Cunha Dupuy and Quidelleur (2018: 10) rightly point out, 'the change of name of the association reveals a strive for emancipation and autonomy from the "Koglweogo central power" in construction, and renders less probable a national self-defence union under the same Koglweogo "unit."' Finally, from the continuity and structuring of the movement emerges what we may call 'intermediary objectives' (Lascoumes and Le Galès 2004: 14) that can be pursued by the movement, its groups, or individual actors. For instance, joining the Koglweogo can be a way to consolidate one's social status and one's prestige, and contribute to one's political trajectory (Quidelleur 2017). Likewise, institutionalization and bureaucratization can lead to higher financial needs while the success at tackling criminality may hinder revenues (if there is less crime), pressing groups to resort to other financing modalities. These can include the creation of a common cash box (instead of a single-operation financing system), the collection of taxes, the resort to donations (especially from merchants), or the commercialization of their security services as well as more ambiguous modalities protecting the interests of individuals rather than their communities (e.g., corruption, interest taking) (Kouraogo and Kaboré 2016).
These factors thus greatly compromise the development of the movement as a homogeneous phenomenon that can be globally apprehended at the national level. Moreover, the perceived ineffectiveness of the movement in the context of a quickly degrading security environment and the failure to live up to expectations may negatively affect its effectiveness-legitimacy nexus and lead to a greater rejection of the movement and its practices.

The rise of societal tensions and the decay of social cohesion
Despite their overwhelming popularity in the rural areas of the regions where they are mostly concentrated, 32 the Koglweogo's proliferation has sparked major tensions in certain territories, hampering the legitimacy of the movement's presence or operation among different components of the population. Four key -often overlapping -factors feed these tensions: the close association with Mossi's ethnic identity, the acceptability of practices, the competition with pre-existing security forces and the (perceived) rurality of the movement. Since the 1970s, numerous Mossi herder-farmers from the Central Plateau have emigrated to the western part of the country in hope of a better climate. This demographic shift has led some towns to be populated by a majority of Mossi newcomers and has initiated multiple conflicts, especially land disputes (Da Cunha Dupuy and Quidelleur 2018). As a consequence, the installation of Koglweogo groups in predominantly Mossi-populated areas near pluri-ethnic 'native' villages or in areas where Mossi are in the minority further intensified ethnic tensions -the latter perceiving this installation as a sign of a deepening of the 'conquest' or 'imperialism' of the Mossi in the region (Le Cam 2017). This tension resulted in confrontations between the Koglweogo and the population as well as the refusal, by local community leaders and entities, to approve the installation of groups in their town, as was the case, for instance, in Solenzo (Demaison Nébié 2017). The hostility towards the Koglweogo also pushed the minister of internal security to adopt a decree in January 2017 prohibiting their implementation in the western part of the country. Nonetheless, and despite the decree, the Koglweogo resumed their activities in some areas or even created new ones in others. Major episodes of violence have taken place since then, including in Tialgho, where a disagreement over an unpaid fine and the exasperation of the population over the Koglweogo's practices led to the death of six people (Bationo 2017). However, the most significant episode of violence occurred in the very beginning of 2019 following the assassination of Yirgou's village chief, his son and five others by attackers suspected to be linked to Islamist terrorism. A punitive expedition, led by the local Koglweogo group, targeted the Peul community, which the group accused of being complicit with the terrorists, resulting in the death of more than 40 Peul and the displacement of more than 6,000 (Jeune Afrique 2019). On top of the ethnic or community dimension, the presence of pre-existing LSIs in the area also induces tensions at the local level, especially with the Dozo whose presence is well established in the west of Burkina Faso. The competition for and over the provision of local security has had three major consequences. Firstly, the tensions led to multiple instances of implementation refusals (e.g., Makognandougou 33 ), ultimatums to leave, and violent clashes (e.g., Karangasso Vigué 34 ). Secondly, the rivalry induces a reconfiguration of preexisting groups. The Dozo, for instance, reshaped the identity of their movement to emphasize the predominance of security provision (versus their identity as huntergatherer, herbalist, or diviner) while presenting themselves as a auxiliary security force respecting the rule of law and as such the 'best partner of the State' (Quidelleur 2017: 105), which has not always been evident in the past (Hagberg 2018: 36). Finally, the tension over the Koglweogo's presence articulates a wider division between rurality and urbanity and the social, economic, and educational differences it involves. A 2017 documentary directed by Luc Damiba and Ismaël Compaoré, entitled Koglweogo Land captured demonstrators in Ouagadougou protesting against the presence of the Koglweogo in the Zongho district and the destruction of their headquarters. A demonstrator interviewed for the film argues 'if they want to stay in the bush, help the population, the breeders, we agree with that, we will support them, but coming in the city, torturing its inhabitants to death, we're going to react.' Overall, the proliferation of the Koglweogo movement in areas where its local legitimacy is absent, partial, or significantly questioned -in opposition to its implantation in Tenkodogo, for instance -undermines the social cohesion both within and between communities.

Undermining state-society relations
Finally, the rapid growth of the Koglweogo movement coupled with the current statecentric response has had multiple effects on the relationships between the state authorities, its security actors, Koglweogo groups and members, and the wider population. In the same way as with previous LSIs, the existence and practices of Koglweogo groups have sparked important tensions with the SSFs, especially the Police and Gendarmerie. Often the result of the arrest of Koglweogo members over torture or unlawful conduct, these frictions not only involve active members of the Koglweogo movement but also their supporters (Lefaso.net 2016), frequently forcing state authorities to 'negotiate' with them on a way out of the conflict (Koaci 2016). Tensions also stimulate the polarization of actors, hinting to different (simultaneous) outcomes. First, an increased defiance of state authority by the Koglweogo, fuelling the existing tensions with SSFs and further lessening the state's (input) legitimacy. As Hagberg (2018: 29) stresses, a ' complete rejection of local security initiatives by the state may run the risk of further undermining state legitimacy, while acting within state legality.' From there, the movement may move towards attempting to 'become' the state and going beyond its substitution. Likewise, the Koglweogo can move away from the 'idea of the state' and circumvent it through the reinforcement of other -and sometime competing -societal structures. Closely linked to the Mossi community, the Koglweogo movement has largely turned to its highest chieftaincy for local and national legitimacy. An example is the support provided by the Moogho Naaba Baongho (reigning monarch of the Mossi) following the greetings of more than 2,000 Koglweogo for the 2018 New Year in Ouagadougou. Considering the tensions with other communities explored above, further reinforcing the Koglweogo's proximity and association with its Mossi identity may distance it from other communities. Likewise, having a conflictual relationship with state authorities and being perceived negatively in non-Mossi communities may trickle down to revived tensions and exclusion along identity lines, an increased politicization and/or an instrumentalization of Koglweogo groups, turning them into real militias (Hagberg et al. 2017: 60;Hagberg 2018). Such diversion from its original objective of the provision of local security is not uncommon among vigilante groups, as the examples of the politicization of PAGAD by Islamists in South Africa (Schärf 2003) or the Dozo in Côte d'Ivoire (Bassett 2003) suggest.

Conclusion: Building on Sources of Legitimacy to Operationalize an Enduser Approach to Local Security
The links between security provision at the local and national levels are complex, featuring strong and opposing trends. At the local level, our study of Tenkodogo suggested an imperfect but existing co-production of security between state and non-state actors. At the national level, we observed that, on the one hand, the current state-centric approach is unlikely to produce a state monopoly of legitimate violence in the short or medium term and, on the other hand, that if the Koglweogo movement is overwhelmingly popular in some regions, its nationwide expansion involves numerous tensions and risks. The rapidly degrading security context and the current terrorism developments in Burkina Faso are both a threat and an opportunity for the Koglweogo groups and a new challenge for the governance of security at the local level. As mentioned, a perceived ineffectiveness in providing security at the local level would jeopardize its support from populations. But it has also 'renewed' its relevance as complementary to state forces, as shown by the passing of the law on voluntary civil participation to security provision in January 2020. It creates a potential new space of collaboration between the state and non-state actors.
The first two sections of this article showed how the sources of security actors' legitimacy -input, output, and throughput -are intertwined, context-specific, and potentially fast-evolving. This pleads for approaches to security governance that neither start from a state-centric input legitimacy perspective nor from a security outputs perspective without considering social cohesion and state-society relations, but that instead incorporate local communities' perceptions and participation, in other words throughput legitimacy. In this concluding section, we build upon the concept of end-user approach to security (Denney 2014) and recent reflections on the 'second-generation' SSR 35 to overcome the current conceptual and practical obstacles and improve the engagement with the plurality of security providers in Burkina Faso.
In the last decade, academic and policy actors alike have promoted an inclusive and citizen-centred understanding of SSR. Already in 2008, the OECD manual on SSR defined the concept as 'the structures, institutions, and personnel responsible for managing, providing, and overseeing security, including informal or traditional security providers, can build the credibility, legitimacy, and effectiveness of a society' (OECD 2008: 163;emphasis added). Nevertheless, the concrete translation of this inclusivity has been marked with ambiguity, and a general struggle to move beyond a state-centric approach and actively engage with non-state actors. SSR promotion is still mostly guided by 'making their security sector look more like our security sector' (Jackson 2018: 6). Likewise, any pluralism in the security sector is still largely perceived as a loss of state sovereignty, as the Pathways for Peace report by the United Nations and the World Bank (2018: 161) clearly illustrates: 'The state cannot delegate security functions to nonstate actors without eventually sacrificing sovereignty.' The 'first generation' of SSR is 'people-centred in theory…yet statecentric in practice' (Donais 2018: 3). The case of Burkina Faso clearly reflects such statecentric focus and selective engagement.
In turn, the recent literature on end-user approaches within a 'second generation' SSR provides insights to better engage with non-state security in contexts where SSFs are not able to deliver the level of security expected by the population over the whole territory. For the purpose of this article, we summarize them into four key principles. Firstly, it is important to recognize that involving non-state security providers does not necessarily forgo state sovereignty. The resort to non-state intermediaries does not in itself determine whether the state is reinforced or fractured (Hibou 1999: 14). Distinguishing the concept of sovereignty from the question of state power allows us to envisage the state as a leading governance actor that plays a mediating and regulating role, conciliating national orientations with citizen-led initiatives (Gautron 2010;Hibou 2004). Concretely, it enables recognition of all security operators, both state and nonstate, and their interconnectedness (Denney 2014). Secondly, a demand-driven approach, focusing on 'how those who use security and justice services experience security and justice' (Denney 2014: 259), should be the starting point of any SSR effort. Adopting such an approach prevents -or at least points tothe risk of inequitable security arrangements (Luckham and Kirk 2012: 5). In essence, existing security arrangements and the ways they are perceived by the population, and not an (imported) ideal type, should be the primary elements of context analysis. Thirdly, and as a consequence, it is imperative to accept the non-homogeneity and local diversity across the national territory. As our case shows, the patterns of local security governance greatly vary according to the social context (see Baker 2008: 155) and socio-political realities (Donais 2017: 2). This principle also enables a much-needed focus on the local-national nexus in the governance of security. Lastly, adopting a pluralistic approach to SSR does not forfeit the respect for the rule of law and human rights as its key strategic and programmatic drivers. Although the legitimacy of non-state security actors such as the Koglweogo is often grounded in the extralegal foundation of their practices (Donais 2017), realigning the latter with the rule of law is precisely one key objective of such reform. Moreover, we should not presume to an automatic or 'natural' greater capacity of state actors to achieve accountability and human rights protection (Baker and Scheye 2007: 517). If the supporters of non-state security participation often advocate for a different role of the state in the security governance framework, they also ' acknowledge the imperative of enveloping security governance within an enforceable rulesbased framework' (Donais 2017: 8). As such, human rights, accountability, and accessibility should most particularly remain the cornerstones of any SSR, and, more generally, at the heart of the regulatory role of the state (Lawrence 2012: 10).
More concretely, the translation of these principles into steps towards a better engagement with LSIs in Burkina Faso may include the following three elements, which can be part of a community policing strategy. First, general reflections on SSR should be complemented by reflections on flexible submodels of SSR that would be better suited not only for the different local contexts but also their rapid evolution (Sedra 2018: 14). Second, a 'positive development' approach towards LSIs and, hence, a move away from the current 'restrict-and-control' attitude, should be adopted. LSIs should be accompanied towards the respect of the rule of law and human rights standards through training and consciousness raising. Joint training activities with SSFs would also facilitate socialization of security providers at the local level. In the same vein, the establishment or reinforcement of cooperation protocols and communication channels between LSIs and SSFs should be fostered. Third, it is essential to set up common accountability mechanisms at the local level (Price and Warren 2017: 8). Public oversight of all security providers and the strengthening of civic spaces to allow it are crucial for a pluralist approach to local security, which is most likely to effectively address the risks of rising social and state-society tensions and security implosion in the country.

Notes
1 The semi-authoritarian character of the regime was mostly characterized by a duality between, on the one hand, the prevention of any political alternation, while, on the other, the fulfilment of internationally recognized 'good governance' indicators (Hilgers and Mazzocchetti 2010; see also Frère and Englebert 2015 Hagberg's (2018: 30) argument that 'people's perceptions, appraisals and responses are as crucial as "the facts" of insecurity.' In this article, we therefore explore and question not only the real but also the perceived a chievements in our assessment of the actors' governance of local security. 8 Overall, 15 Police stations and 13 Gendarmerie brigades are present in the Centre-East region. 9 The numbers of officers were provided by each force in February 2018. 10 Like for the other forces, the regional direction is based in Tenkodogo as it is the region's capital. 11 Digital communication with a member of the Koglweogo group of Tenkodogo allowed us to follow up on the discussion of their practices and membership. According to him, the number had risen rapidly to 500 in June 2018. 12 Authors' interview with Koglweogo member, Tenkodogo, February 2018. All quotes from the interviews -conducted in French -are our translation. 13 Authors' interview with Customs, Tenkodogo, February 2018. 14 Authors' interview with a member of a civil society organization,  (2017: 44-45) highlighted the support of the population for LSIs due to their effectiveness in fighting insecurity. It is, however, important to note that none of the survey questions in the study -carried out as part of the PARSIB program -dealt specifically with or even mentioned LSIs. 28 Most notably, a 2016 decree on LSIs defines as illegal almost all of Koglweogo's core current characteristics (e.g., bearing arms, wearing outfits comparable to the SSFs, having an emblem) and practices (e.g., physical abuse, detention and sequestration of presumed criminals, fines) (Government of Burkina Faso 2016, art. 12), thus greatly limiting the incentives for the groups to ask for or receive their license to practice legally as LSIs. 29 The ' discharge' is an indirect governing modality, limiting the cost of an important bureaucratic apparatus (Hibou 2004: 18-20). 30 Between 1983 and 1987, part of the CDR's role of implementing and defending Thomas Sankara's revolution at the local level was to contribute to the security of the area through the mobilization, awareness, and organization of the people (Conseil National de la Révolution 1984). 31 Following the introduction of the concept of community policing in the internal security legislation (2003), CLS were created in the framework of the 2005-2010 five-year community police plan. These committees had the objective of serving as mobilization and coordination platforms for the participation of local communities in the prevention and fight against insecurity. Even though around 3,000 CLS were created, the limited representativeness of the committees among communities and lack of continuous funding ultimately led to their ineffectiveness in tackling insecurity at the local level (Bayala 2011, 50-51; Wisler n.d.). 32 A recent poll, conducted by the association Action pour la sécurité humaine en Afrique (ASHA) in the Eastern, Upper-Basin, Central-West and Central regions, reveals that between 79 per cent and 92 per cent of the population surveyed, depending on the region concerned, fully or partially support Koglweogo's actions while only between 8 per cent and 21 per cent fully reject them (ASHA 2018: 6). 33 In November 2016, Koglweogo leader Boukary 'Le Lion' Kaboré was banned from installing a Koglweogo in the village Makognandougou by its traditional chief following a debate with the Dozo. Most notably, the chief argued that 'This role is played by the Dozos that work in symbiosis with the loyal defence and security forces' (Hagberg 2018: 29). 34 Despite the ministerial decree banning their presence as well as an ultimatum to leave given by Dozo leaders in February 2018, the active presence and local outcry over the physical abuse practices of Koglweogo led to a deadly confrontation between the two self-defence groups (Dofini 2018

Funding Information
This article is published with the support of the University Foundation of Belgium. The data for this research was collected during a field mission that took place in February 2018 as part of a research program on governance networks funded by Belgian development cooperation through its university cooperation platforms ARES-CCD and Vlir-UOS.