Introduction: Armed Groups and Multi-layered Governance

Abstract

In this special issue we broaden the academic debate on rebel governance by examining additional armed actors – militias, police and foreign intervenors, and the ‘layers’ of governance they add. We develop the notion of ‘multi-layered’ governance to capture the complexity of these cases. We consider ‘mediated stateness’ as a special case of multi-layered governance. We discuss ‘polycentricity’ as an equivalent concept, but deem multi-layered governance more appropriate. The following articles discuss rebels’ legitimation strategies, armed opposition factions, auxiliary armed forces, mediated stateness and intervention by foreign powers to highlight the roles of different actors and the resulting impacts on governance.

Introduction

Who governs during civil war? We argue multiple actors with intersecting influences on governance of civilians. Not only the incumbent government, and not only those who rebel against it. The uncertainties unleashed by civil war lead to different types of armed groups intent on civilian protection or predation, military advantage or uneasy collusion with the state. Principal actors, either the state or rebels, may organise armed groups, while others form and act independently. Beyond these domestic agents, powerful external actors may also influence how these armed groups govern civilians. During civil war, control over territory becomes divided between different warring parties, with some areas under full control by either the government, the rebels or their auxiliaries, while in other areas control is contested, often shifting between a multiplicity of protagonists (Kalyvas 2006, pp. 210–220). Even in areas of full control, it would be wrong to assume a priori that armed agents created or financed by principal actors necessarily pursue the latters’ strategic objectives. In contested areas, deviation by junior players will probably be even greater. As Andreas Mehler correctly insists, ‘the precise interplay between violence actors/protectors regarding complementarity, active arrangements or contradiction/competition’, is an area of research that deserves more attention (2004, p. 545).

Throughout these territories instances of governance emerge, whether mainly following the regulations of the state, the practices of a variety of non-state armed groups or a combination of both (Zanker et al. 2015; Mampilly 2011, p. 26). We follow Rosenau’s observation that governance is a more encompassing phenomenon than government, because it embraces informal, non-governmental mechanisms of persons and organisations as well as formal institutions (1992, p. 4). Indeed, we extend it, noticing that formal authority is usually weak or non-existent during civil war and is sometimes replaced by fluid networking among distrustful actors. ¹ Rebel groups are involved in governance ‘whenever they engage residents in an area they significantly control to pursue a common objective’ (Kasfir 2015, p. 22).

While recent scholarship has studied rebel-civilian interactions through the concept of ‘rebel governance’ (Arjona 2016, Arjona et al. 2015, Duyvesteyn et al. 2015, Kasfir 2005, 2002, Mampilly 2011), we expand the academic debate on governance by non-state armed actors during civil war in two ways. First, we widen the notion of ‘rebels’ to ‘armed groups’ in order to pay special attention to governance by a diversity of other armed actors allied to the rebels, the state or neither. Second, we introduce the notion of ‘layering’ to situate these actors at different levels ranging from the local to the international. We articulate our notion of multi-layering by relating it to other concepts that could also account for multiple governance actors, specifically mediated stateness and polycentricity. Throughout we refer to the specific cases in the articles that follow. Our larger objective is to open new paths toward research into the implications for governance created by the complicated interconnections among all the armed actors generated by civil war.

Armed Groups and Governance

As soon as rebellion breaks out, the state faces competitors for civilian loyalty. At that point, a state-centric analysis is no longer sufficient. In our view, governance cannot be understood then as a monolithic or unified concept, that is, as the prerogative of a formal government in full control of a territory. We do not accept that in conditions of war government necessarily breaks down and produces ‘unruled spaces’ totally devoid of governance. It is in fact the multiple, dynamic, emergent and often contested nature of governance interactions and arrangements under conditions of civil war or uneasy collaboration that we put centre-stage in this special issue.

Since governance is often not confined either to the incumbent government, or to rebel groups during civil war, we supplement the notion of rebels opposing states with the wider notion of various armed groups opposing states, and as well cooperating with or opposing each other. As Jentzsch and colleagues argue: ‘research on internal armed conflict must depart from its tendency to equate the concepts of ‘armed group’ and ‘rebel group’ and abandon its canonical focus on state and insurgent actors as sole providers of wartime political order and violence’ (2015, p. 11). In order to analyse how they govern, it is essential to understand how armed groups relate to each other in particular conflicts.

Who rebels against whom? Some armed groups oppose the government, while others fight on its behalf, have loose coalitions with both or multiple sides, or with none. Some armed groups rebel against the political order imposed by other rebel groups. Other armed actors flee to third states where they maintain an edgy co-existence. In addition, armed groups often switch alliances (Christia 2012, pp. 32–54). The presence in some cases of ‘oligopolies of violence’, a concept that refers to a ‘fluctuating number of partly competing and partly cooperating actors of violence’, suggests the complexities that must be analytically disentangled (Mehler 2004, p. 539). In that sense, ‘rebellion’ should be conceived as a relational notion. Some of these armed groups have the willingness and the capacity to govern aspects of daily life in territories they hold. Some do not. Our research priorities should include an examination of all armed groups active in civil war, to see whether or not they engage in rebel governance. ² Each of these forces may take any side – state, rebel or its own – during the civil war, and may change sides as well. To limit militias to the government side, the conventional view (see Schneckener 2017, pp. 799–800 and passim), constricts analysis, not only of civil war, but also of governance.

In order to govern, or influence governance by other actors, these armed groups need the ability – or the credible threat – to use violence, possess some degree of territorial control and confront civilians living in that territory (Kasfir 2015, p. 25). Armed groups in civil war that meet these criteria may be rebel groups or other armed organisations such as militias, para-militaries, police, self-defence forces, vigilantes or combatant units formed by warlords, clan chiefs, or strongmen (Schneckener 2006, Staniland 2012, Mukhopadhyay 2014, Jentzsch et al. 2015, Dirkx and Terpstra 2016). As Reno shows in his account of the Kamajor auxiliaries in Sierra Leone for example, these militias may be exceptionally well-organised and may not display predatory behaviour (2007, pp. 330–332). They provided a level of political order and protection against rebels of the Revolutionary United Front (RUF). Strongmen in Afghanistan possess the means to govern through their coercive capacities, their ability to extract revenues and their social linkages to important civilian groups (Mukhopadhyay 2014, pp. 50–51). In other words, the ability to use force and the possession of territorial control with civilians residing in those areas are conditions met by a variety of armed groups active in civil wars. Several of those different types of groups involve themselves in governing civilians, as the various empirical case studies in this special issue show.

It is possible to adopt this broader definition of armed groups instead of rebels, without the loss of explanatory power already existing in the current rebel governance literature. For example, we may still inquire if armed groups, whether rebels, self-defence forces or militias, with a Maoist organisational structure are more likely to develop an effective governance system as Mampilly argues (2011, p. 78). Or, to extend another plausible proposition, we may ask whether the capacity of armed groups such as self-defence forces or militias to form a unified political command explains the likelihood of their participation in governance (Mampilly 2011, p. 80).

Two novel ideas (among others) emerge from our expansion of the types of armed actors that can influence the governance of civilians. First, the extent of state or rebel organisational control of armed auxiliaries may help determine how important the auxiliary is in shaping civilian governance. Second, some armed groups, particularly village militias created by the domestic state, a foreign actor, by rebels or independently by local civilians, are likely to know more about the people they are ordered to engage than their organisers. If engaged, they are in a strong position either to influence regulation of those civilians or be delegated to regulate them, particularly where the state or rebel group is more concerned with its larger military campaigns.

Organisational Control and Armed Auxiliaries

On the first point, while armed auxiliaries may be entirely autonomous or completely dependent, control will be a matter of degree and may amount only to supervision or coordination. What counts is how auxiliaries relate to their superiors. As James Q. Wilson, the leading scholar of bureaucratic behaviour reminds us, ‘the most important thing to know is how that coordination [or supervision or control] is accomplished’ (1989, p. 24). To which we hasten to add, ‘or not accomplished’. A senior armed actor that creates a militia will try to direct its activities to further its goals. How and whether it can control its subordinate amidst the uncertainties of civil war is a fascinating question about which we need to learn more. Principals may have tight, slack or no control over militias, even those they formed themselves. Control may be centralised, even exercised through a hierarchical command structure, or, given the confusion engendered by civil war, delegated to agents expected to act within formal or informal guidelines. Contested areas dictate different control strategies for militias or police from those securely held by either the state or rebels. The intrusion of external forces further complicates ties between senior partners and their armed auxiliaries. In any case, as the literature on rational choice reminds us, we need to inspect for shirking. Agents are likely to act in their own interests when they think they can get away with it. We return to this notion in the next section on multi-layered governance.

We also need to pay attention to individual military entrepreneurs who form relatively autonomous armed groups to protect their areas or to profit from civilians living in them. They may exert governance over civilians independently of the preferences of large rebel organisations or the national government. The outbreak of a civil war or its expansion to a new area may involve a single armed actor extending its influence or a confusing multitude of new actors attempting to control the resulting uncertainties. Nevertheless, civilians typically will have to sort out orders from all of them. Finally, we have to disentangle the relations between an intact rebel group that flees to a third state and that state’s government. To make matters more difficult, in assessing the degree of control in each of these cases, we need to remember that they are dynamic – what is tightly controlled and amicable one day may become far looser and more hostile the next.

As the succeeding case studies inform us, relations among multiple armed groups run the gamut posing complicated problems of governance. The expansion of the civil war in South Sudan to Equatoria included both the Sudan People’s Liberation Movement in Opposition (SPLM-IO) and several small newly established local forces that partly cooperated and partly competed (Kindersley and Rolandsen this issue). During the Mozambique civil war (1976–92), Renamo kept tight hierarchical control over the ‘Mujeeba’, the militias it created to keep order in villages in areas it controlled (Jentzsch this issue). Meanwhile, the Frelimo government only loosely coordinated its activities with Naparama, an independent militia, eventually siding with it after it decided it could trust its leadership.

In the Afghan civil war, the three-way relations among the local police forces in Kunduz Province, the Afghan state and the American army present a different pattern of interactions between principals and agents – one that changed over time. The tenuous control the Afghan state exerted over its rural police forces was reduced by the disproportionate investments of American army finance in these police to support US counterinsurgency objectives and then further undercut by the subsequent abrupt withdrawal of US material support (Dirkx this issue). Consequently, these local police forces often shirked by pursuing their own interests at the expense of goals intended for them by their principals.

The Palestine Liberation Organisation (PLO) demonstrates the changes in control over time (Stel this issue). When the PLO fled to Lebanon in 1970 after it was defeated and expelled from Jordan by the Jordanian army, it encountered a weak national government. In this situation, the PLO was able to act as an independent governing actor, despite its lack of sovereignty. In 1982, the PLO leadership was forced out of Lebanon and upon its return in the 1990s it had lost much of its autonomy and had to settle for a far more subordinate status.

Local Knowledge and Armed Auxiliaries

Regarding the second idea, the military and political benefits of deep knowledge of terrain and people is almost a truism of civil war studies. Guerrilla organisations operating in rural areas often have an initial advantage in local intelligence over conventional state armies. In areas that have experienced civil wars in the past, local people also can call on their past experience with insurgent techniques. For this reason, more numerous and powerful state armies are usually considered necessary to defeat smaller and poorly armed rebels. However, as successful rebels seize more territory, their leaders are likely to ‘lose’ the local knowledge that villagers always have at their fingertips. Armed auxiliaries created by the state’s army or by rebels can restore this resource. Local militias possess intimate knowledge of place and people, almost always superior to their senior partner. It is their loyalty that is in question. They provide their principals with several critical resources. Whether loyal or not, if not prevented, they are likely to facilitate or even conduct governance. Through local connections, they are usually more adept in maintaining order, mobilising popular support and providing critical intelligence by unmasking informers.

When the South Sudan insurgency spread to Equatoria, it was propelled by the accumulated skills in guerrilla warfare that local civilians had acquired over 50 years of past civil wars (Kindersley and Rolandsen this issue). In Mozambique, Renamo mostly limited governance by its local militias, the mujeebas, to security largely for its military purposes. It deployed them to impose order by preventing civilians from leaving their villages, interrogating them about supporters of the government and executing informers and defectors (Jentzsch this issue). On the other hand, Afghan local police forces, far more familiar with the residents in their districts than US Army or national Afghan officials, governed them by exploitation in stark contradiction to the expectations of their senior actors (Dirkx this issue). Even in its weakened state, when its leadership returned to Lebanon after 1990, the PLO continued to administer services in the Palestinian refugee camps and gatherings (Stel this issue). Elsewhere, Mampilly observed that local Mai Mai, in this case independent of the larger foreign backed Rassemblement Congolais pour la Démocratie-Goma (RCD-Goma), were far more successful than the latter in gaining the confidence of residents in the localities they controlled (2011, p. 207).

Multi-layered Governance

Once we recognise that several armed groups may operate in the same civil war, we need to ask how each might affect the governance of civilians in those areas where several armed groups are active simultaneously. Current analyses of rebel governance have generally adopted a single-level focus. Since the interactions between civilians and rebels or the state in a particular territory are the most prominent features of many civil wars, this emphasis is not surprising. However, where the governance environment contains multiple actors, friendly, hostile or both, we obviously require additional layers of analysis. ³ Extra layers emerge when political actors beyond any single armed group shape governance dynamics during civil war. Consequently, in our approach, we call for analysis that is sensitive to multiple often hostile actors with different agendas who have a voice in civilian governance.

A number of concepts could serve that purpose, but only if they are purged of often-unrecognised normative assumptions that are likely to undercut their application in civil war conditions. Leading candidates that could account for multiple streams of governance include our choice, ‘multi-layered governance’, as well as ‘polycentricity’, and in certain specialised cases, ‘mediated stateness’. Because articulation of all three concepts has focused on explaining governance within constitutional rules and under peaceful conditions, each carries at least an implicit commitment to certain norms that impede analysis of governance provided by armed actors in civil war. The specific difficulty in applying any of these approaches to civil war situations is to avoid any presumption that they necessarily entail a system in which multiple actors share a goal of cooperation or are expected to work together.

We choose multi-layered governance precisely because it lacks the highly specified analytic development of polycentricity. The former has been used mostly as a descriptive concept in rich and poor parts of the world, although only within administrative frameworks and under peaceful conditions. Still, it is the absence of close theoretical attention that makes it the more neutral and therefore, as discussed below, the more appropriate concept. We also consider mediated stateness, because it is especially useful in those cases where a rebel group and a state act together under conditions of both cooperation and hostility.

Multi-layered, or multi-level, governance is conventionally conceived (as is polycentric governance), first as a system in which the parts are supposed to engage together and second, as one in which cooperation is a shared norm, whether or not it exists. While most analyses of multiple governments, whether nested or overlapping, presume as much, often implicitly, an analysis of civil war cannot. A few instances illustrate this problem. Nauclér, for example, defines ‘multilevel governance’ as ‘the distribution of power and competences through the vertical layering of public authority from the supra-state to local government level’ (2005, p. 85). Plainly (and sensibly) Nauclér asserts that a consciously planned system of governance will more efficiently resolve administrative problems stemming from administrations responsible to overlapping constituencies than would be the case if each level of government were left to compete with the others.

In the same vein, the concept of multi-layered governance has generally been applied to relatively peaceful or post-conflict situations to study governance dynamics on various levels below the nation-state and on levels above. Beyond the nation-state, examples include the governance dynamics between supranational organisations like the European Union (EU) and its member-states (Deschouwer 2003, Clark and Rohrschneider 2009). Below the nation-state, it includes layers of subnational (self-)governance such as Scotland’s power and competencies within the United Kingdom or subnational governance arrangements in Belgium (Wolff 2009). Here, again the authors presume the problem the actors they discuss (political parties, voters and states, respectively) must solve is how to work together within a set of rules that has been constitutionally established. Leonard makes this norm explicit when it is usually just assumed: ‘Thus the effectiveness of all governments relies upon numerous actors at multiple levels within and beyond the state, connected to one another and the various functions of governance through complex and divergent networks’ (Leonard 2013, n.p.).

Multi-layered governance is also a dynamic concept. The influence of different layers can change over time and space. For example, federalism as practiced in the United States is also multi-layered government. Over time, the national government has supplied ever more of the money (over 80 per cent) spent by the states and therefore comes close to dominating state policy (Rose 1973, p. 1162 and passim). Nevertheless, as Rose notes, American states are still laboratories for innovation. But their variance either lagging or anticipating ‘national policy patterns’, has become increasingly limited (1973, pp. 1172–1173).

Mediated Stateness: A Special Case of Multi-layered Governance

We treat mediated stateness as a special case of multi-layered governance among armed groups. Mediated stateness, or a ‘mediated state’, introduces the analysis of instances of cooperation between rebels, other armed groups and the state with which they co-exist. Here, the adversarial interaction or use of violence is replaced by a pragmatic form of accommodation and compromise. While cooperation between them seems contradictory, we identify four types of mediated stateness involving states and armed actors. In many cases, insurgents operating within the country they rebel against make use of state resources that the state may find inconvenient to withdraw. Second, mediated stateness provides a method to analyse civil wars where the state forms alliances with independent indigenous militias. For example, the state may tolerate rebels that restrict themselves to particular areas (Mehler 2004, pp. 539–540). Third, mediated stateness is also a useful approach to interpret the uneasy collaboration among armed actors where rebels flee with civilian supporters to a second state while maintaining a sufficiently coherent organisational identity to act with some autonomy in particular administrative areas. If the rebels are powerful relative to the second state’s government, the geographic and functional areas in which the rebel organisation can act will be extensive and the two will have a tense relationship. If the rebels are relatively weak, the relationship will be the more commonplace case of sanctuary or refugee relations, usually with the voluntary consent of the state.

External intervention by powerful foreign actors poses a fourth form of mediated stateness. Where an external actor, often another country, allies with the state and invests significantly more resources in local armed actors, usually militias or police, than the state, it can heavily influence, even dominate, the policies of the state and the local armed actor. Since the state has formal sovereignty, it cannot be completely eliminated from the relationship, even when the foreign actor works directly with local forces. In this case, there is three-way mediated stateness. This cooperation may be hostile, even coerced, rather than voluntary. In each case, the nature of cooperation is likely to change over time as the balance of forces among these actors changes. All four situations, whether armed groups are operating in their own state or in another, suggest the rich explanatory power that mediated stateness can provide. We sketch the development of the concept and instances drawn from our case studies below.

Mediated stateness became a contemporary political concept when Ken Menkhaus introduced it as a pragmatic statebuilding solution for Somalia. He used it narrowly to cover situations ‘in which a central government with limited power and capacity relies on a diverse range of local authorities to execute core functions of government and mediate relations between local communities and the state’ (2006–07, p. 103). He took the idea from an account of the uses of maps by premodern rulers to overcome the decentralised authority structure characteristic of late feudalism (Menkhaus, 2006–07, 103 citing Voekel 1998, n.p.). As a medieval historian reminds us, ‘[p]owers of legitimate coercion and control were not concentrated in royal hands but were widely diffused throughout society’ (Given 1990, p. 248). The application of this characterisation of feudalism holds much promise not only for explaining situations of mediated stateness generally, but also for cases of rebel governance more commonly where civilians are caught between the state by day and rebels by night. Menkhaus made creative use of the parallel between statebuilding in Somalia and the distribution of authority among rival feudal institutions that made any royal policy initiative, including taxation, dependent on elaborate mediation.

Menkhaus envisioned the mediated state as ‘a partnership’ between a national government and ‘local intermediaries and rival sources of authority to provide core functions of public security, justice, and conflict management’ (2006–07, p. 78). Nora Stel offers a broader and more useful definition that stresses the degree of interdependence of state and non-state actors in governance (this issue). Thus, in applying mediated stateness to governance by armed groups, we greatly expand its use, while retaining its core idea of a partnership among organisations possessing independent sources of authority and military strength.

Consider how the four types of mediated stateness we note above apply to the cases our contributors present. The three elements we need to examine are: a sovereign state, an armed actor with authority that is not premised on the state and some cooperation between the two (or possibly more than two). The question is always to specify the degree and the limits of their interactions. For the first type noted above, the state and a rebel group sometimes collude on a limited basis in specific areas, each for its own reasons. While the appropriation of state resources by rebels is a well-understood tactic, tacit cooperation between rebels and states often remains unnoticed. In such cases, mediated stateness is a useful approach to explain the causes and dimensions of these implicit (and invariably denied) forms of cooperation. In Sri Lanka, for example, the Liberation Tigers of Tamil Eelam (LTTE) did not completely reject or replace state institutions in rebel-held territory. Instead, they took over the most strategic ones (governance sectors such as security and justice) while assuring the continuation of other services by state institutions (education and health care for example). The latter were implemented under LTTE regulations and control, leading to a type of hybrid governance in areas of LTTE territory (Mampilly 2011, pp. 112–113; Terpstra and Frerks forthcoming). For the Sri Lankan government this served its strategic goal of showing the outside world some capacity of sovereignty in the Northeast, while for the LTTE it was a way to keep the services provided without having to spend financial resources on them.

Second, states mediate security governance when they cooperate with local militias. In Mozambique, the Frelimo-controlled state overcame its suspicion of the Naparama, which formed independently and used spiritual ceremonies to mobilise followers, contrary to Frelimo’s strongly held secular and anti-witchcraft beliefs. They formed an informal alliance to provide a measure of security governance in opposition to Renamo (Jentzsch this issue). The mediated stateness approach is conducive to exploring the steps through which distrust and disagreements over policy and beliefs are overcome in order to forge an alliance during civil war. As Corinna Jentzsch observes, such alliances may be relatively loose, leaving the militia with significant operational autonomy (this issue). Particularly in areas where political authority is fragmented, localised ideas of ‘stateness’ may not necessarily be embodied in one particular office or centralised governance structure. As Nicki Kindersley and Øystein Rolandsen explain with regard to South Sudan, historical legacies of common knowledge, personal work experience and everyday practice have shaped the nature of the current wartime orders at different localities (this issue). Hence, armed groups, whether inside or outside the government, may need to negotiate with locally based organisers and supporters of different factions to mediate their rule.

Third, mediated stateness can account for the interdependence in civilian governance that develops when rebels defeated in one country flee with their followers to another country as relatively intact military organisations. The degree of influence these rebels exert in their new home will depend in part on how welcome they are, but even more importantly on the strength of the armed forces they encounter. The more powerful the rebels, the more autonomously they will be able to operate, and the more governance they have the capacity to deploy. At one end of this continuum, rebels may even extend some administrative services to citizens of their new state as well as providing them to their own civilian followers. Over time, the balance of strengths of the rebels and the new state may change. When that happens, we can expect the degree of governance to change as well. The nature of the collaboration between the displaced rebels and their host state may increase or reduce the delivery of services to civilians. Mediated stateness provides an excellent dynamic framework for assessing not only the degree to which each of the parties influences governance over civilians, but also the benefits that their cooperation actually provides.

As Stel explains, when the Palestine Liberation Organisation (PLO) fled to Lebanon in 1970, it more than maintained its own identity as a governance provider for Palestinian refugees and even for Lebanese citizens (this issue). Weak though the Lebanese government was throughout the first period of co-existence, it insisted on exercising its rights as the sovereign. In a second period, when the PLO leadership returned to Lebanon following its time in exile after the Israeli army expelled its military forces, the positions of relative power were reversed. Interdependence continued, but now more on the terms laid down by the Lebanese government.

As suggested above, the fourth type of mediated stateness is a three-way exchange involving an external actor that possesses independent authority whether or not it is a sovereign, the national state and local armed actors. The issue is the nature of their cooperation and what happens if it is withdrawn. To inquire into the Afghanistan case, as our example, we need to recognise that in addition to the Afghan state and the US military, there are also two types of ‘third’ actors, the arbaki, that form indigenous police forces and new local police forces, created as administrative units of the national government but funded (and supposedly supervised) by the US military (Dirkx this issue). As Toon Dirkx explains, some of these police units took advantage of American finance to strengthen their power as independent armed actors (this issue). But in doing so, they were not ‘mediating’ state authority. Instead, we interpret this outcome as an unanticipated consequence of the interdependence of the state and the external actor.

Polycentricity: An Equivalent Concept

We regard polycentricity as an equivalent concept to multi-layered governance. Polycentricity is defined as ‘a system of governance in which authorities from overlapping jurisdictions (or centres of authority) interact to determine the conditions under which these authorities, as well as the citizens subject to these jurisdictional units, are authorised to act as well as the constraints put on their activities for public purposes’ (McGinnis, 2011, p. 171). The seminal work of Vincent and Elinor Ostrom and others on common pool resources and public goods has sensitised policymakers and academics to the diversity and complexity of contemporary governance of complicated social systems, eventually also extending to governance aspects of the European Union (Hooghe and Marks 2003). In its reach, it seems roughly equivalent to multi-layered governance.

If it had no conceptual history, polycentricity might serve our purposes just as well. But, as Elinor Ostrom describes her ‘intellectual journey’ over the last 60 years in the published version of her Nobel lecture, it has been elaborated into a multifaceted coherent logical scheme that has taken older fundamentally dichotomous notions such as markets, states and collective action and reshaped them as continuous variables (2010, pp. 641, 644–645). Likewise, in recognising the omnipresent problem of incomplete information, it replaces the notion of the rational individual with the notion of ‘bounded rationality’ (Ostrom 2010, pp. 659–660). In addition, the approach attends to how learning can build the trust necessary to overcome a range of collective dilemmas (Ostrom 2010, p. 661). Our account only scratches the surface of the innovations accomplished in this research programme.

The attraction in this careful articulation of polycentricity lies in its capacity to find bases for cooperation to achieve long-range goals that appear obstructed by short-term rivalries. To achieve them depends on unearthing systemic elements that can solve collective action problems. Some armed actors in civil war cooperate (as exemplified above by the application of mediated stateness to aspects of our case studies), but we need an analysis that is also open to situations in which these actors are implacably opposed to the very existence of each other. In addition, to understand what armed actors are doing, we investigate more effectively with an approach that is neutral. We insist that while it is entirely appropriate to judge armed actors morally, we need first to understand them. The other problem with polycentricity for us is its explicit endorsement of normative goals as essential to achieve good administrative programs:

Extensive empirical research leads me to argue that instead, a core goal of public policy should be to facilitate the development of institutions that bring out the best in humans. We need to ask how diverse polycentric institutions help or hinder the innovativeness, learning, adapting, trustworthiness, levels of cooperation of participants, and the achievement of more effective, equitable, and sustainable outcomes at multiple scales. (Ostrom 2010, p. 665)

We know that the Ostroms and other practitioners of polycentricity would insist that their analyses must also be value-neutral. The larger issue is whether an approach designed for application to settled government can avoid hidden norms as well in conditions of civil war. We worry that it may not.

Multiple Layers of Governance

As stated above, we introduce the concept of ‘multi-layered governance’ to better understand and explain the influence of several armed groups on the governance of civilians during civil war. Recognising that some armed groups act independently while principals control others, we want to determine what influence each has in regulating civilians. Our strategy consists primarily in identifying the regulatory behaviour that each armed actor adds. For descriptive (rather than analytic) purposes, we array our cases by the increasing number of layers involved, moving from simple to complex cases of governance. In addition, we anticipate that for particular analytic issues certain layers will either be more important than others or will be the only one(s) necessary for explanation. We rely here on the case studies presented in the following articles to illustrate our conception. Our central concern is to identify how governance of civilians depends on the number and behaviour of the actors involved. To reiterate what we argued above, we regard armed actors’ cooperation or hostility an empirical question rather than a norm on which to evaluate them.

Armed Groups and Civilians

In many cases, the relationship between one armed group and civilians living in the area it controls consists predominantly of a single layer of governance. This is most likely when a rebel group has unambiguous domination over a specific territory, sometimes called a ‘liberation zone’, and the national government withdraws all support to local residents. Analytically, the best approach to certain research questions is to limit consideration to one layer. Explaining a rebel group’s efforts to create its legitimacy among local residents and their responses is such a question.

From a strategic perspective it is important for most, although not all, armed groups to induce compliance from the civilian population to achieve their political and military goals. Civilians provide invaluable opportunities for food distribution, intelligence and recruitment. In the interactions between armed groups and civilians, legitimacy can play an important role (Duyvesteyn 2017; Podder 2017; Schlichte and Schneckener 2015). In order to turn evasive power into durable political positions, armed groups need to develop legitimate claims to power that resonate among their constituents (Schlichte and Schneckener 2015, p. 411). International juridical legitimacy is almost impossible for rebels to obtain. Nevertheless, the civilian population frequently accords legitimacy on the bases of the Weberian traditional, charismatic and legal-rational sources. The legitimacy of armed groups resides in what could be called a tacit social contract between the ruler and the ruled. Armed groups can legitimately represent grievances of a marginalised segment of a population (Podder 2013, p. 19). James Worrall connects this point to the notion of political order, urging the exploration of ‘complex negotiations at multiple levels which create the ordered space in which rebels are able to govern’ (2017, p. 725 and passim).

Because the LTTE was known for its especially coercive policies, it poses a ‘hard’ case for testing legitimacy effects (Terpstra and Frerks this issue). Niels Terpstra and Georg Frerks show how the LTTE tried to induce compliance from the population through a mixture of coercion and legitimation strategies. They demonstrate how different strategies and acts by the LTTE succeeded in creating legitimacy and compliance among the Tamil population while others were less effective or caused resistance. The LTTE’s most effective legitimation strategies were discursively rooted in Tamil traditions and belief systems, particularly sacrifice and martyrdom, charismatic leadership and notions of a historical homeland. However, Terpstra and Frerks also observe that the LTTE propagated a strong reformative socio-political agenda attacking entrenched customs that did not increase its legitimacy among Tamil civilians.

While the single-layered perspective on rebel-civilian relations well serves some cases and research questions about governance (how rebels gain support, create allegiance or acquire legitimacy), for other situations we require an investigation of multiple layers of governance, as we discuss in the next section.

Multiple Armed Opposition Factions

In some protracted civil wars, there is a multiplicity of armed factions. This potentially creates new layers of governance. Some armed groups get mobilised by using their previous experience, knowledge and expertise (Kindersley and Rolandsen this issue). Rebel organisers may have gained work experience in government functions, or government training, while at the same time local government officials may temporarily align with insurgents. Therefore, it is difficult to delineate where local government structures begin and where the rebel organisations end. Moreover, different factions can align for a shared purpose, but may just as easily compete in a new stage of warfare.

Kindersley and Rolandsen in this issue analyse the multi-layered nature of armed factions in the Central Equatoria area within South Sudan as some armed groups began to develop governance structures in 2016–2017. They place the emergence of armed groups’ governance in its specific historical context showing how various forms of governance follow a historically embedded logic. Kindersley and Rolandsen move beyond binaries of civilians/rebels, and states/rebels, as these categories do not provide them with a better understanding of the governance dynamics. Instead, they look at the specific social histories of those involved in armed groups’ governance.

Senior and Auxiliary Armed Groups, and Civilians

The existence of multiple armed groups leads to more complex patterns of governance in civil war. The emergence of militias introduces an additional layer of governance, either with or without the consent of their senior fighting forces – if the militias regulate civilians at all. Both governments and insurgents may use auxiliary forces to pursue their political and military goals during civil war. Some auxiliary forces are formed by governments or insurgents, while others emerge as grassroots movements that are later co-opted by governments or rebels (Jentzsch 2014). The nature and degree of governance may depend on whether a senior armed actor forms a militia or an unaffiliated entrepreneur organises one. A characteristic of most auxiliary forces is their superior knowledge of their local geographical terrain and greater closeness to civilians to whom they are usually related. It seems likely that proximity leads auxiliaries to engage in some form of governance, the substance of which likely depends on how closely they are controlled by a senior force. Among auxiliary forces, independent militias are probably most likely to govern. Nevertheless, the rebels or the government may possess the capacity to accommodate or direct governance arrangements between the auxiliary forces and civilians.

Jentzsch analyses this additional layer of governance by discussing auxiliary armed forces that carried out specific functions in the civil war in Mozambique between the Frelimo government and the Renamo rebels (this issue). Renamo created its auxiliary armed forces, the mujeeba, controlled them closely and used them for purposes of supply, intelligence and attacks. Meanwhile Naparama, a local auxiliary force independently organised to defend villagers against Renamo, temporarily enjoyed military success, due in large part to its reliance on its leader’s ‘divine mission’ and use of protective local medicine that made its fighters fearless and frightening to its opponents. As Jentzsch observes, that had two consequences (this issue). First, it led to an arms-length alliance between Naparama and the government, despite the government’s strong opposition to reliance on any indigenous religious beliefs, including customary medicine. Second, it resulted in Renamo’s innovation by mimicking Naparama’s use of customary healers to initiate mujeeba, making them more effective opponents and thus developing a more successful strategy towards civilian security governance. These forces put the government under severe pressure and, due to Renamo’s tight control of its militias, resulted in harsh security governance for civilians. Thus, the success of the new auxiliary force fighting on the government side led to innovations on the rebel side, creating an even more powerful auxiliary force. In a similar change in the use of an auxiliary armed force, Tamil militias in Sri Lanka, initially set up to protect their own villages as ‘home guards’, were pushed to leave their villages and join other battlegrounds as military pressure on LTTE fronts increased. ⁴

The State, Its Armed Groups and Civilians

When the state mediates its relationship with an independent armed actor while opposing a rebel group, the situation for civilian governance becomes more complex. We suggest that it presents an additional layer of governance, possibly more. Central governments do not necessarily compete with other armed groups in all sectors of governance during civil war, rather they opt for a pragmatic form of engagement that allows them to govern with, or even through, those they consider rebel groups, even when they are rebels against other states. The critical distinction from the immediately preceding ‘layer’ involving armed auxiliaries is the lack of control the state possesses over its ally. Inevitably, this absence of control is better expressed as a continuum rather than a dichotomy. As we noted in discussing the concept of mediated stateness above, the Naparama’s relation to the Frelimo government is actually a case on this continuum.

Stel shows that while PLO governance in Lebanon is conventionally regarded as constituting a ‘state-within-the-state,’ it is better understood as a type of mediated stateness (this issue). In addition to studying the relations between the PLO and the Palestinian refugees living under its auspices, she demonstrates that it is necessary to investigate how the PLO negotiates with state officials, and how it also provides governance to Lebanese citizens in particular sectors. In addition, it is equally important to identify how the Lebanese state exerts its governmental influence over Palestinians living in refugee camps under PLO supervision. An important strength of Stel’s account is her analysis of the variation in the complex and wary relations between the Lebanese state and the PLO as they provided governance to Palestinian refugees and Lebanese citizens during two different periods. In the first period (1970–1982), the rebel group had the upper hand, while in the second the Lebanese state took the dominant role. Through this comparison, Stel is able both to show the dynamic character of mediated stateness and the range of possibilities for this combination of different layers of governance. Stel is careful to argue that dominance in either period did not amount to a one-way street. Instead, she demonstrates what she calls a ‘dialectical relationship’ in which both state and rebel organisation engaged in joint, if often tense, governance in which each of them preferred coexistence to a monopoly over decision-making (this issue).

Powerful External Actors, The State, Armed Auxiliary Actors and Civilians

The final layer of governance we discuss in this special section involves external armed actors, ostensibly invited by the state, who engage with local armed groups often with unanticipated consequences. This complicated scenario requires the analyst to disentangle relations first, between a foreign actor and the state, second, between the state and local armed groups and third, directly between that foreign actor and those local armed actors. Where an external actor is far more powerful than the state in terms of military strength and resources it is willing to invest, it will often take decisions that the state is reluctant to endorse but can do little about. These external actors apply their assets to undermine or empower local armed groups at the expense of other actors (Staniland 2012, p. 254). External support can be divided into different types: military, moral, political, or material support including training or technical assistance and the provision of a sanctuary (O’Neill 2005, pp. 142–148).

We focus our discussion here on the question of how an increase or decrease of support from powerful external actors confronting a somewhat reluctant national government may affect the governance dynamics between local armed groups and civilians. Increased external support for one armed group at the expense of another may affect the governance dynamics between those groups and civilians significantly. Among such cases, the influence of powerful external actors can be identified most clearly where governance by local actors ends up contradicting the policies of both the external actor and the state. Such cases are classic instances of ‘negative influence’, that is, intervention that has the opposite effect from that intended (Dahl 1976, p. 43). The final contribution by Dirkx traces instances of these reversals of intended governance in the Afghan civil war.

As part of its counterinsurgency, the American military lavishly funded new local police forces in Kunduz, Afghanistan over the opposition of the Afghan president (Dirkx this issue). But the Americans had poor intelligence about the individuals they selected to form these police forces and others who were recruited. Local police commanders and informal powerful figures manipulated, contradicted and rearranged their objectives to serve their own agendas. These armed actors’ governance towards the civilian population varied from protection and service provision to predation and violent attacks on local citizens. In one case, the national government ended a programme, forcing the US military to cut its support. The local police force became a militia and began to impose its own taxes on civilians (Dirkx, this issue). In another case, most of the commanders of these newly created US financed police units were informal clients of rival power figures. In short, police forces intended to protect their villages as part of US counterinsurgency strategy turned into relatively autonomous militias governing for their own interests, as Afghanistan’s president had warned (Dirkx this issue). Decreased support for some US-trained militias initially fighting on behalf of the Afghan government against the Taliban even led to the defection of these groups to the Taliban insurgency (Dirkx this issue). Hence, civilians living in that area shifted from a type of militia rule to Taliban governance.

Paradoxically, while these militias were an instrument to fight against the insurgency in order to establish the Afghan state’s monopoly on the legitimate use of force, their empowerment weakened efforts to centralise the power of the national government. Militia support yielded short-term counterinsurgency gains, but in the mid- to long-term, it unintentionally undermined the provision of security for local populations in Kunduz, the strength of the Afghan Government and as well US strategic interests.

Other cases also show how changes in external involvement create incentives for armed groups to become more prominent in governance. In Sri Lanka for example, the LTTE suddenly held control of much of the northern territories after the withdrawal of the Indian Peace Keeping Forces (IPKF) in 1990. The departure of the IPKF and decreased support from India pushed the LTTE to shift its strategy. The LTTE no longer had its sanctuary in India as it did in earlier days and it no longer received political support from India. From 1990 on, the LTTE became increasingly involved in governing the civilian population in the northeast of Sri Lanka (Terpstra and Frerks forthcoming). In such cases, the vacuum that external political actors leave behind helps explain how governance of armed groups emerges.

Conclusion

We propose two ways to push the academic debate on rebel governance forward. First, we demonstrate that widening the notion of ‘rebels’ to ‘armed groups’ helps to conceptualise governance by a multiplicity of non-state armed groups in conditions of civil war. Most obviously, governance may no longer be seen as the privilege of the sovereign government only. Instead, armed groups may create a tacit social contract with a particular segment of the population. The nature of interactions among armed oppositions factions, armed auxiliary forces, rebels, the state or external powerful actors may be conflictive, co-optive or cooperative. Relationships between armed groups and civilians may range from autonomous to full control by senior rebel forces, the formal state or powerful external actors.

Auxiliary forces typically have greater local knowledge than their senior partners and that may affect their governance roles. Inspecting auxiliary armed actors as junior partners liable to shirking by following their own interests instead of those of their principals is a new research direction worth attention. Like rebels, auxiliary forces, militias, police and other types of armed groups may participate in governing civilians. Through the possession of (contested) territorial control in populated areas and the ability to use force, armed groups of various ideologies, political goals and organisational structures become involved in civilian governance. Armed groups engage residents to pursue common objectives within territories under their control or create particular opportunities for their benefit. Institutionalising civilian compliance may provide armed groups with long-term strategic advantages, both in opposition to the government and vis-à-vis other armed groups.

Second, we introduce the notion of ‘layering’ to improve our understanding of how the multiplicity of armed actors, including states, external political actors and auxiliary forces created by higher level authorities, affect civilian governance. The creation of many actors creates the possibility that multiple streams of influence require examination in order to explain governance. We examine multi-layered governance, mediated stateness and polycentricity as approaches to discover the best framework for analysis. We find that while both the first and third can encompass multiple actors, the prior conceptual development of polycentricity contains hidden norms, namely an expectation of cooperation and willingness to work together. Because we are dealing with civil war, we need an empirical concept that does not necessarily presume cooperation. In this respect, we feel the field is better off with multi-layered governance because it lacks the sophisticated normative articulation that has marked the development of polycentricity. We find mediated stateness a valuable special case of multi-layered governance, one that is especially useful in interpreting the qualified, often reluctant, acceptance by states of auxiliary armed actors that have independent authority.

These layers may differ among governance sectors, as armed groups may be in conflict with the national government in one sector, while allowing that government’s service provision to continue on rebel-held territory in other sectors. At the same time, use of armed auxiliaries risks allowing their goals of governance to undermine those important to either the rebels or the state. Our contributors discuss variations located between coercion and compliance, cases of protection, predation and punishment, the design and application of different legitimation strategies, mediated stateness, the role of both external intervenors and the state in setting up auxiliary forces and militias and constraints involved in controlling them. We show that the resulting complexity of the governance interactions can best understood by a multi-layered analytic framework that extends beyond the binary of rebel group-civilian population.

Notes on Contributors

Nelson Kasfir is a professor of Government (Emeritus) at Dartmouth College.

Georg Frerks is a professor of Conflict Prevention and Conflict Management at the Centre for Conflict Studies, Utrecht University; and professor of International Security Studies at the Netherlands Defence Academy.

Niels Terpstra is a PhD candidate at the Centre for Conflict Studies, Utrecht University.

Acknowledgements

We are grateful to Jonathan Fisher and Paul Jackson as editors of Civil Wars, and two anonymous peer reviewers for their close reading of our work and their comments to help us improve the article. We also thank the contributors of this special issue for their suggestions to improve the article and the framework we introduce.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1. We find recent accounts of state construction and post-conflict management helpful in developing our account of armed actor multi-layered governance (see in general Leonard 2013, Hagmann and Péclard 2010).

2. In Mampilly’s work, militias occur mainly as challengers of rebel governance structures – referred to as ‘rival militias’ (2011, p. 3). We include militia as agents who may engage in governance without regard to whom they are allied (see for example: Dirkx this issue).

3. As Engel and Mehler (2005, pp. 90–92) note for example, local governance becomes increasingly complex when one takes into account the interfaces that emerge between local, subnational, national and external actors during violent conflict, resulting into different degrees and types of ‘stateness’.

4. Interview by Niels Terpstra, Trincomalee, Sri Lanka, 18 March 2016.