Abstract
Theories of civil war usually theorize the choices of civilians and combatants without considering the institutional context in which they interact. Despite common depictions of war as chaotic and anarchic, order often emerges locally. Institutions vary greatly over time and space and, as in peacetime, shape behavior. In this article, I propose a research agenda on local wartime institutions. To this end, I present original evidence on conflict areas in Colombia to illustrate the scope of variation, propose the concept of wartime social order and a typology, and discuss several ways in which research on wartime institutions can contribute to our study of civil war both at the micro and macro levels.
Understanding the choices of civilians and combatants is crucial to our research on civil war and postconflict dynamics. We want to know, for example, why people join rebels and militias, why families decide to flee, why combatants kill, why locals support or boycott counterinsurgency operations, and why former fighters successfully reintegrate into their communities or fail to do so. Even when we ask questions about macro-level outcomes such as the duration of war, the stability of peace agreements, or the effects of peacekeeping operations, our theories and interpretation of empirical results rely on assumptions about how actors make decisions on the ground.
Despite the general agreement that institutions—understood as rules that structure human interaction (North 1990)—shape behavior, the study of how civilians and combatants make choices in war zones has, for the most part, neglected the role of wartime institutions. Overlooking the effect of institutions in the analysis of individual and collective behavior would be astounding in any field in political science; however, it has endured in civil war studies perhaps because war is assumed to be chaotic and anarchic, as the widespread use of concepts such as failed states and collapsed governance suggests (Justino 2013).
Yet, the emergence of local institutions—and, with them, order—in the midst of war makes sense theoretically. To start with, war often weakens, and sometimes destroys, state institutions. Different literature has shown that in contexts where access to effective institutions is lacking, new informal institutions are likely to emerge. For example, rural communities that depend on limited, public natural resources often develop norms that facilitate collective action (Ostrom 1990). Illegal markets where property rights and contracts cannot be enforced by law also tend to develop their own parallel institutions (e.g., Gambetta 1996; Volkov 2000 Varese 2001; Skarbek 2011). The emergence of the state itself has been explained as a process whereby one actor offers institutions and protection in exchange for taxation, thereby transforming a situation of anarchy into one where clear rules allow for higher predictability, productive activities, and capital accumulation (e.g., Tilly 1985; Olson 1993). Even within contexts where institutions do exist, actors often attempt to provide private orderings to “realign incentives and embed transactions in more protective governance structures” (Williamson 2002; see also Dixit 2007, 438). These insights suggest that when prewar institutions are weakened in war zones, some sort of new institutions that establish order are likely to emerge.
The existence of wartime institutions should not be surprising for another simple reason: armed groups have incentives to create them. First, as Tilly (1978) suggests, in order to overcome their competitors, warring sides try to monopolize the means of violence, extract resources from local inhabitants and, at the same time, promote capital accumulation. Even though Tilly was referring to a long historical process, armed actors fighting civil wars are likely to learn that in order to advance their cause, they need to create a sustainable system of resource extraction to fund their operations. Such a system, in turn, requires some security and limited taxation for civilians to engage in productive activities (Olson 1993)—in other words, it requires institutions. 1
A second reason why armed groups are likely to create institutions has to do with the kind of warfare they engage in. Most contemporary civil wars are characterized by irregular warfare—that is, a contest “entailing an asymmetric rebel challenge launched from the country’s rural periphery” (Kalyvas and Balcells 2010, 415). In such wars, rebels’ modus operandi entails seizing control over pockets of territory, imposing a growing challenge to the state. Although violence is a key means to achieve and maintain such control (Kalyvas 2006), creating institutions to rule local populations is essential as well (Arjona 2010). The mere creation of order facilitates population monitoring and increases the odds of voluntary cooperation, which, as many have noted (Guevara and Davies 1985; Kalyvas 2006; Mao 1978) is crucial for maintaining control. In addition, specific institutions allow rebels to shape economic, political, and social affairs in ways that benefit their organization. Such institutions may, for example, facilitate recruitment, provide access to political networks, allow for the accumulation of material resources, and even put their ideology in practice by implementing promised reforms (Arjona, 2010; Mampilly 2011; Arjona, Kasfir, and Mampilly, in press).
Empirical evidence on civil wars across the globe supports these theoretical priors: some form of order often emerges in war zones, where clear rules are enforced. Furthermore, since these rules vary greatly across and within civil wars, the institutional arrangements that operate in war zones can be quite diverse.
In this article, I argue that such arrangements need to be incorporated in our study of civil war and propose a research agenda on local wartime institutions. I focus on the locality because war often segments territory, making localities the key locus of choice. To advance this research agenda, I first show that there is, indeed, great variation in wartime local institutions by relying on systematic, quantitative, and qualitative original data on Colombia. Second, I propose a way to conceptualize the emergence (and breakdown) of order as well as of the set of rules and arrangements that structure political, economic, and social interactions in war zones; for this purpose, I introduce the concept of wartime social order, present a typology, and assess its quality both theoretically and empirically. Theoretically, I argue that the typology identifies a variation that is relevant: we have reasons to inquire about the causes of this variation, and we can expect it to influence other important phenomena both during wartime and in its aftermath. Empirically, I use cluster analysis to show that the typology identifies distinct types—that is, they are internally homogenous but differ drastically from each other. I also argue that while the typology is parsimonious in that it identifies only three types, it still has high descriptive and explanatory potential. Finally, I discuss how this typology could advance our understanding of different wartime and postwar phenomena and make a plea for incorporating institutions to our study of micro-, meso-, and macro-level outcomes. My goal is not to provide a theory of wartime institutions; 2 rather, this article seeks to show that different institutional arrangements emerge in war zones giving place to new forms of order, argue that they warrant attention, and offer a way to conceptualize them.
I proceed as follows: In the first section, I discuss why we need a research agenda on wartime local institutions. In the second section, I present data on wartime local institutions in Colombia to give the reader a sense of the phenomenon we are to conceptualize and the scope of its variation. In the third section, I introduce the concept of wartime social order and the typology, and assess its quality. In the fourth section, I conclude by discussing specific ways in which this approach can contribute to our study of civil war.
Why Study Local Wartime Institutions?
Studies of civil war focus on research questions at different levels. At the macro level, they seek to identify the conditions under which civil wars start, end, and resume; why some produce a greater number of deaths than others; or how particular ways of ending a conflict shape postconflict paths. 3 At the micro level, recent research has focused mostly on the causes of killings (Kalyvas 2006; Balcells 2010; Metelits 2010), sexual violence (Wood 2009), and displacement (Steele 2010; Ibañez and Vélez 2008); the determinants of participation, mobilization, and recruitment (Petersen 2001; Wood 2003; Humphreys and Weinstein 2008; Arjona and Kalyvas 2011; Parkinson 2013; Petersen 2001; Wood 2003); and the individual-level effects of interventions to foster reintegration, reconciliation, and development (Humphreys and Weinstein 2007; Gilligan, Mvukiyehe, and Samii 2013).
Either explicitly or implicitly, studies at both levels rely on assumptions about the ways in which civilians and combatants make choices on the ground. This is obvious when we think of questions like recruitment or collaboration: we are inquiring directly about a decision of an individual. But even when we ask about macro-level outcomes, our theories tend to rely on some assumption about why people behave in the way they do. For example, theories of civil war onset rely on assumptions about why people launch rebel movements and why others decide to join them. To be sure, a theory might start with the wrong assumptions and still get the general causal link right; when it comes to deriving implications, however, micro foundations and mechanisms can make a stark difference as false assumptions may lead to wrong theoretical deductions and policy recommendations. Here is where the locality becomes crucial: if we want to model decision making—either because our question is about a choice or because we need to make assumptions about it—we need to rely on a realistic understanding of the context in which that choice is being made. In most civil wars, such context is the locality.
Civil war has a tremendous capacity to segment space (McColl 1969; Kalyvas 2006, 88). While a town lives under full control of the national army, the town up the hill lives under rebel control, and the one down in the valley is under dispute. In the blooming literature on the micro-level dynamics of war, scholars have started to take into account the role that local-level factors play in shaping different outcomes–for example how prewar local elections shape wartime violence (Balcells 2010) and displacement (Steele 2010; Balcells and Steele 2012), how state repression shapes civilian support for the rebels (Petersen 2001; Wood 2003; Lyall 2009) and how territorial control shapes violence and collaboration (Kalyvas 2006). However, few authors have attempted to conceptualize, systematically describe, and theorize those different local realities that emerge amid war, in which actors live and interact.
Recent studies give clues about how those local realities might look. The work of anthropologists on wartime governance (e.g., Vlassenroot and Raeymaekers 2004; Lubkemann 2008) depicts daily life in areas where rebels, incumbents, and international actors interact with civilians. Wood’s (2008, 539) analysis of the transformation of social processes in war makes a key contribution by highlighting different ways in which “social actors, structures, norms, and practices” change during war. Arjona (2009) discusses the coexistence of different local orders in war zones, showing evidence of variation in who rules, in what domains, with what enforcement mechanisms, and establishing what kind of relation with the local population. Staniland (2012) focuses on the different relations between states and insurgents—which he conceptualizes as wartime political orders—bringing to the fore the different ways in which power can be allocated between incumbents and insurgents. An emerging field on rebel governance identifies and theorizes variation in rebels’ ruling strategies (Kasfir 2005; Weinstein 2007; Metelits 2010; Mampilly 2011).
These studies have certainly improved our understanding of conflict areas. However, we need ways to conceptualize the overall institutional contexts in which actors live. In as much as these local realities are the locus of key choices, we need to conceptualize them and theorize their origins, how they function, and how they might shape decision making. It is useful to think of such realities as “regimes”: if there is something like local regimes in war zones, we need to incorporate them in our analyses.
Before proposing a concept and a typology that can move us forward in this direction, I present evidence of the existence of wartime local institutions and their variation.
Wartime Order and Institutions: Evidence on Colombia
Localities in conflict areas often become microcosms with their own political, social, and economic institutions. These can be either formal or informal, but in both cases people tend to know them well. In this section, I present evidence from what, to my knowledge, is the first systematic data set on wartime local institutions. I led a research team to collect the evidence in 2010 and 2012 on two random samples of Colombian localities where guerrilla or paramilitary groups had been present for at least six months since 1970. The goal of presenting these data is twofold: I aim to show, first, that institutions do exist in war zones and, second, that they vary greatly across time and space, across and within armed actors, and across and within localities. I start with a brief overview of the Colombian conflict; I then describe the method for gathering the data; and then I present evidence on economic, political, and social institutions operating in conflict zones.
A Brief Overview of the Colombian Conflict
The ongoing Colombian conflict started in the 1960s, right after a previous bloody war had ended. Several leftist guerrilla groups were formed, including the Revolutionary Armed Forces of Colombia (FARC), the National Liberation Army (ELN), People’s Liberation Army (EPL), the People’s Revolutionary Army (ERP), and the April 19 Movement (M-19). These groups described themselves as popular liberation movements seeking to bring about social justice and socialism.
The conflict had a low intensity for several years; in the 1970s, however, the guerrillas began to expand into new areas of the country. They moved from poor and isolated places to areas that were closer to the center and had higher incomes and resources (Echandía Castilla 1999; Vélez 1999). They engaged in extortion, kidnapping, taxation, and drug cultivation and trafficking, which provided abundant resources. During this decade, the FARC became the largest and most powerful of the guerrilla groups, followed by the ELN. According to available estimates, by the late 1990s, about three-fourths of all Colombian municipalities had some form of presence of guerrilla organizations (Echandía Castilla 1999). This growth, in terms of both geographical expansion and scope of activities, affected the interests of local elites in several regions of the country, particularly in the north.
In part reacting to the threats that the guerillas posed to them, and in part responding to national-level changes such as decentralization, local elites began to form paramilitary forces. Although a few were self-defense groups organized by peasants, most were set up by landowners, cattle raisers, emerald traders, and drug traffickers (Romero 2003). At first, these paramilitary groups operated separately in different areas of the country. They financed their operations with a combination of taxes on economic activities in areas under their control, voluntary and forced regular payments by locals, and drug trafficking. Even though the state did not create these groups directly, there is substantial evidence of collusion as well as of tacit approval, including negligence in stopping instances of massive victimization of civilians. 4 In addition, these groups managed to create very strong ties with local and regional political figures, which are well documented by journalists and academics. 5 In 1997, most paramilitary groups united under an umbrella organization called the United Self-Defense Forces of Colombia (AUC). Although some of the operations of the blocs were planned at the level of the AUC, each bloc preserved a high degree of independence. Overall, paramilitary groups were less disciplined than their guerrilla counterparts (Gutiérrez 2008).
Due to the growth of the guerrillas and the paramilitaries, the armed conflict escalated throughout the 1980s and reached its peak in the late 1990s. According to most sources, the amount of violence decreased in the mid-2000s (e.g., Centro Nacional de Memoria Histórica 2013).
Different peace negotiations and demobilization processes have taken place in the last two decades. In the late 1980s and early 1990s, several leftist guerrilla groups demobilized collectively and were given amnesty (like the M-19, part of the EPL, and the Quintin Lame). Thousands of individual members of the FARC and the ELN deserted under the Uribe government (2002–2010), but both groups remain active. Most paramilitary groups demobilized their members through negotiations with the government, although new groups known as BACRIM (the Spanish acronym for criminal bands) quickly emerged and are now active in many regions of the country. Although the guerrilla groups that remain active have been weakened, they have intensified their operations as a new peace process with the government is currently underway.
The Colombian conflict differs from many others in its duration: it is one of the longest ongoing internal armed conflicts. This exceptional length could raise doubts about whether the patterns found in Colombia are generalizable to other conflicts. However, while some regions have coexisted with armed groups for forty years, others became war zones only recently. In addition, the country exhibits internal variation in almost every dimension that one might expect to matter in an investigation of wartime institutions: some regions have valuable legal natural resources like gold, while others have coca leaves, and others lack any such goods; ethnicity varies across and within regions; both left-wing and right-wing groups operate; and state capacity varies greatly across time and space. Hence, despite its uniqueness as a long conflict, the Colombian case is well suited for investigating many aspects of the conduct of war.
Measuring Wartime Local Institutions
Although the importance of informal institutions has been largely acknowledged (e.g., Dasgupta and Sergaldin 2000; Helmke and Levistky 2004), methods to measure them are surprisingly scarce. Measuring informal institutions during conflict makes the task even harder. Using a novel approach, I led a research team to gather detailed data on wartime institutions in seventy-four communities throughout Colombia in 2010 and 2012. In what follows, I briefly describe the sampling strategy and data collection method.
I randomly selected a list of municipalities where at least one guerrilla or paramilitary group had been present for at least six continuous months in the past four decades, stratified by region to ensure geographical variation. I excluded five of the thirty three departments of the country (the equivalent to US states), because they are located in the Amazonian region or in the Caribbean sea, are underpopulated, and have only recently experienced the presence of armed actors. 6 In total, the two samples include thirty-one municipalities located in nineteen departments. The sample is quite diverse along several dimensions, including location, ethnicity, economic activities, abundance of legal and illegal natural resources, the structure of land tenure, and historical patron–client relations. By virtue of the geographical variation, the sample also includes very different conflict dynamics: some municipalities were strongholds of the FARC since the 1970s, while others have only experienced rebel presence since the 2000s. Likewise, some municipalities were bastions of paramilitary groups in the 1990s, while others encountered these organizations much later. Patterns of violence also vary greatly across municipalities, as do patterns of counterinsurgency and anti-narcotics policy. Table 1 shows descriptive statistics of proxies of ethnic composition, state presence, social conditions, infrastructure, natural resources, and violence.
Descriptive Statistics of Sampled Municipalities.
aBased on the multidimensional poverty index.
Once the municipalities had been selected, I relied on a short survey with a heterogeneous group of key informants to map out variation in armed groups’ involvement in local institutions. I then stratified localities according to this measure, and randomly selected between two and four communities in each municipality. Figure 1 shows the final sample of localities. 7
Sample of Colombian localities with presence of non-state armed groups.
In each selected community, my research team relied on focus groups to create timelines and identify key events in order to help respondents recall past events. Participants also collectively identified how a set of institutions changed over time. We then conducted a survey with each participant to gather more detailed evidence on the history of several local institutions, state presence, community organization, and the interaction between communities and armed actors over time. The survey used a structured questionnaire with both close- and open-ended questions. Participants were selected in order to have a heterogeneous group in terms of political views and interests. In most cases, the group included a teacher, a local leader, a merchant, and an elderly person. Women were present in all workshops. Using both the quantitative and qualitative data collected in the field, I created a data set with a rich, detailed description of how armed groups penetrated local communities, how civilians responded, and what kind of institutions operated over time. 8
In what follows, I present some of the results to illustrate the range of variation in wartime institutions. Given that in some localities several armed groups were present at the same time—sometimes establishing different institutions—the unit of analysis is not the locality-year, but the locality-armed group-year. Structuring the data in this way allows the assessment of what different armed groups did when operating at the same time and location. 9
Wartime Institutions Under Guerrillas and Paramilitaries
The sample includes 1,271 observations on 124 dyads made up of seventy-four communities and thirty-eight fronts or blocs of guerrilla and paramilitary groups, which interacted for at least six continuous months between 1970 and 2012. 10 Since this is a study of wartime institutions, the panel only includes observations for locality-years when at least one nonstate armed group was present. Given that some communities interacted with armed groups for thirty years while others did so for a few years, the panel is unbalanced.
Institutions vary greatly across and within armed groups, across and within localities, and over time. To simplify, I present the data, aggregating all locality-group dyads over time. In most cases, I show separate descriptive statistics for guerrillas and paramilitaries.
I start with a general description of these conflict zones. About 67 percent of all communities interacted with at least one guerrilla or paramilitary group for more than ten years between 1968 and 2012; in around 24 percent of the cases, armed groups were present between five and ten years; and in about 9 percent of the cases, presence lasted less than five years. This means that the sample includes communities that have interacted with armed actors for many years, as well as communities where such actors operated only for a few years. In almost 40 percent of the communities, only one group was ever present, while two or more groups operated in the remaining 60 percent of cases, sometimes simultaneously. For the most part, however, most communities interacted with one group at a time (91 percent of all locality-years), while periods with two or more groups were far less common (9 percent). When two or more groups were present, in about half the cases, they were fighting each other while in the remaining half they coexisted peacefully under some agreement.
Guerrillas were present in the places where they operated every day or at least once every week in about 60 percent of the cases. They showed up only once per month or a few times per year in almost 40 percent of the cases. The paramilitaries had a daily or weekly presence, which was more often than the guerrillas (in about 74 percent of the places where they operated). In general, however, both interacted quite frequently with local residents (Figure 2).
Intensity of group presence by side.
Turning to order, civilians describe most of the time they lived under the presence of an armed actor as one where clear rules—either formal or informal—regulated conduct: in about 90 percent of all observations, civilians knew what rules they had to follow and stated that they had clear expectations about how combatants would behave. Where do these rules come from?
The first way to investigate wartime institutions is by looking at who rules. We asked respondents about who, in general, run their communities at specific points in time, allowing for multiple choices in their answers. 11 Armed groups ruled in about 65 percent of all cases; state authorities such as the police or the major ruled in almost 25 percent of the cases; civilian leaders such as a priest or a civic leader ruled in close to 12 percent of the cases, including the indigenous cabildo or the Afro- Colombian communitary councils. Even though the guerrillas and the paramilitaries often ruled in the localities where they were present, there is substantial variation over time and space. In about 70 percent of all municipalities included in the sample, there are community-armed group dyads where combatants ruled in some years but not others.
This complex distribution of power brings to the fore an important aspect of wartime local governance: the interplay of state, societal, and rebel forces as creators of institutions. Social actors shape rebel strategy in various ways, as combatants have to react to authority structures, local norms, and social cohesion in their attempt to control civilian behavior (Arjona, 2010; Barter, in press). The state can also shape wartime institutions in those places where, despite the presence of nonstate armed actors, state agencies remain in place. There is great variation in the intensity of state presence in conflict zones: while weak states may be unable to provide any services in peripheral areas under rebel control, in countries with stronger states, public agencies may operate even when rebels own the monopoly over the use of violence, as the cases of India and Colombia indicate. Often, rebels directly seize resources from the coffers of public agencies; sometimes they redirect services and resources to their support base—a practice that Colombian scholars have called armed clientelism (Peñate 1999). As these various forms of interaction indicate, the relation between nonstate armed groups and the state is quite complex, as the former may influence, co-opt, coerce, or ally with public servants (Arjona 2009; Staniland 2012).
The complexity of these relationships can be illustrated by looking at national elections, which in Colombia have been held for years despite the ongoing war. The guerrillas forced people to abstain from voting at least once in about 17% percent of the places where they operated. They forced locals to vote for a particular candidate at least in one election in about half of the localities. The paramilitaries, on the other hand, did not forbid turnout but commanded locals to vote for a particular candidate at least once in about two-thirds of the places where they were present. In local elections, both groups often vetoed who could run for office, chose a candidate to support, and mobilized or coerced people to vote for that candidate. At the same time, in some territories, none of the groups intervened in elections despite having a presence there.
How do these social, state, and nonstate actors rule? Preserving public order is a key concern, and all three actors often established rules and enforcement mechanisms to regulate certain conducts. Those in charge usually forbade stealing, killing, and raping. Again, there is great variation in who becomes the authority figure in charge of these issues: in about two-thirds of the cases, civilians turned to combatants to solve problems related to public order; they only turned to the local government (the mayor or the police) in about 24 percent of the cases, and in about 12 percent locals relied on civic leaders, indigenous leaders, or community assemblies. Very few sought the local courts.
Civilians also relied on different kinds of institutions to solve interpersonal conflicts. When confronting a problem over a land border, in about two-thirds of the cases civilians turned to the armed actor. They also sought the mayor or the police (20 percent) and indigenous authorities, assemblies, or civic leaders (15 percent). A low percentage would resolve the problem in some other way, including turning to the courts.
Economic activities are also regulated in different ways in war zones. Mandatory contributions to the armed actors were common but not universal: guerrillas imposed such taxes in about two thirds of the cases, while paramilitaries did so in almost 85% of them.
Armed groups regulated legal and illegal economic activities in some localities but not all. Both guerrillas and paramilitaries regulated fishing, hunting, or wood extraction in almost half of the cases. The paramilitaries were more likely to intervene in illegal mining. Both armed actors regulated the cultivation of coca leaves. In a small percentage of communities, the guerrillas had a say on who would receive state subsidies.
In many places, guerrillas and paramilitaries established rules to regulate private conduct. Both regulated mobility in more than half of the localities where they were present and free speech (that is, what people could talk about) in close to 90 percent of the cases. They regulated personal appearance (like the use of skirts by women or earrings by men) or sexual behavior (like homosexual relations and prostitution) in around two-thirds of these places (Figure 3). There is variation within a single community over time as well, as some communities lived under these rules in some periods but not others.
Armed groups’ regulation of private conduct.
Institutions established by the armed groups were more or less formal depending on the armed group and the territory. In about half the communities where a guerrilla group was present, the commander discussed at meetings the specific rules that everyone had to follow; paramilitaries did so in about one-third of all the localities where they were present. In others, those rules were made clear in more subtle ways, such as through interventions by militiamen.
Conceptualizing Variation in Local Wartime Institutions
As the data show, institutions vary greatly: across different spheres of local life, within and across armed groups, and over time and space. As with any phenomenon, there are many ways in which we could conceptualize this variation and several typologies to capture it. I propose to approach the distinct institutional realities that emerge in war zones as forms of social order. Although fear and violence are common in conflict areas, chaos is seldom the norm because clear rules often exist. In many places, there is a sense of normality—even if different from that of peacetime—and people have expectations about what might happen. I refer to the existence of this predictability as order. I define wartime social order as the set of rules that structure human interaction in a given community during wartime, allowing for that predictability to exist.
Social order in a war zone can vary across multiple dimensions. I propose a typology on the basis of two (Table 2): first, whether a social contract has been established between the armed group and local residents, by which I mean that both sides follow certain rules of conduct. While this contract is seldom spelled out, every social order relies on an implicit notion of the duties or commitments of both the ruler and the ruled. I refer to the existence of such social contract as a situation of order. This dimension can be operationalized as the existence and enforcement of rules of conduct, which allow for predictability. 12
A Typology of Wartime Social Orders.
When combatants, the local population, or both fail to abide by a set of defined rules, there is no social contract between the two and unpredictability is high. I refer to this situation as disorder. When combatants do not abide by clear rules, local residents have few solid beliefs about the likely outcomes of alternative choices. Even if an armed group is in full control of civilian behavior, the absence of limits for those in power leads to high levels of unpredictability. A parallel with an impulsive dictator or the state of exception in a democracy serves to illustrate this situation: the government in power can have tight control over the population, but at the same time display unpredictable behavior. To be sure, this does not imply that combatants behave randomly or that internal rules that constrain their behavior toward civilians fully disappear; indeed, there is great variation across cases of disorder in armed groups’ tactics. Yet, civilians do not expect rebels to follow a set of rules and, therefore, the level of predictability is low.
Disorder may also emerge when local actors—common civilians, authorities, or other agents present in the locality—fail to abide by known rules, even if combatants themselves do so. Consider a village where local residents pick and choose which rules to follow or decide to disobey them entirely: there would be no social contract, and unpredictability would be high. 13
It is important to note that these social contracts do not necessarily entail legitimacy; in this sense, they are closer to a Hobbesian social contract than to a Rousseaunian one. Civilians may dislike the way in which rebels rule and even the very existence of a rebel government. But the mere existence of clear rules that both sides—the ruler and the ruled—follow indicates that there is a shared notion of how everyone behaves. There is a stark difference between this situation and one in which a social contract is absent and uncertainty reigns.
When a social contract between the local population and the armed group does exist, the form of order varies depending on the scope of the group’s intervention in local affairs, which can be narrow or broad.
Armed groups intervene broadly when they regulate activities beyond security and taxation such as politics, the economy, social relations, and private conduct. I use the term rebelocracy, or the rule of rebels, to denote situations where an armed group becomes the de facto ruler in this broad sense. The specific domains in which the armed group rules can vary, but for rebelocracy to exist, its intervention has to go beyond the maintenance of public order and the collection of material contributions. The channels through which the group rules can vary as well: in some places, it relies on combatants, who are permanently deployed in the locality and rule directly; in others, it relies on militiamen, who are part-time members of the organization and live within the community (and are often members of it) and report directly to a commander; in other cases, the group rules through a preexisting political party that becomes allied with the armed group, or through organizations that freely support it or that have been widely infiltrated, co-opted, or even created by the armed actor, like unions, boards, cooperatives, or even the formal local government.
Rebels intervene minimally when they interfere only to maintain their monopoly over the use of violence. They may also demand a material contribution from locals such as food or payments. However, other matters are in the hands of other local actors—be it state officials, traditional authorities, civic leaders, or others. I refer to this form of social order as aliocracy or the rule of others. 14
It is important to note that rebelocracy does not imply that sources of authority other than armed groups are absent. As the empirical section of this article illustrates, war zones can exhibit a complicated structure of authority where state officials, religious figures, and ethnic leaders play important roles even when combatants are the de facto rulers. While this typology focuses on the influence that nonstate armed groups gain, it does not rule out the presence and sway of other actors. The next sections I discuss why such a focus on armed groups can be productive for my study of both wartime and postwar phenomena.
Assessing the Quality of the Typology
How robust is this typology? Although there is no consensus on the criteria to evaluate concepts and typologies (Doty and Glick 1994; Gerring 1999), a good typology has to meet at least three conditions, beyond internal consistency: (1) it should identify variation that matters either because we have reasons to inquire about its causes or because we can expect it to shape relevant phenomena; (2) it should identify types where within-group variation is minimized and between-group variation is maximized; and (3) it should be parsimonious: it should identify as few types as possible while having the greatest descriptive and explanatory potential. In the remaining of this section, I show that this typology is parsimonious in the sense of being simple and, yet, having great descriptive potential. I also show that it yields distinct types that are internally homogeneous. In the next section, I discuss the relevance of the concept as both a dependent variable and an explanatory factor in our study of other phenomena.
The first way to assess whether the typology captures types that exist on the ground is by looking at actual cases (i.e., war zones or conflict areas) in very different contexts and seeing whether we can identify disorder, aliocracy, and rebelocracy. Although systematic data on wartime local institutions are hard to find, there is plenty of detailed, qualitative evidence on armed groups and conflict zones around the world. A survey of this literature suggests that the typology indeed captures three ideal types that are often found on the ground. 15
Accounts of armed groups bringing about disorder when occupying territories abound. The groups fighting in Sierra Leone and Liberia, for example, are best known for their predatory strategies and limited observance of rules (e.g., Ellis 1998; Weinstein 2007; Johnston 2004). However, even the groups that are known for ruling civilians often display this type of unconstrained behavior, especially when trying to take over a territory for the first time or when defending it from their enemies.
Situations of order—that is, where clear rules regulate conduct—are quite common, despite the widespread association of anarchy with war. Some authors actually describe the change brought by war as the emergence of a new order (e.g., Weber 1981; Vlassenroot and Raeymaekers 2004; Lubkemann 2008). What are the attributes of this new order? Case studies from very different civil wars provide evidence of both aliocracy and rebelocracy.
Several descriptions fit the definition of aliocracy. In these cases, rebel groups control a territory and behave like a minimal state, only regulating public order and taxation. Other spheres of life are regulated by rules that come from various sources, including the state, traditional authorities, and local leaders. Cases in which a group establishes an indirect form of rule fall into this category: an underlying agreement between the armed group and the community—or its ruler—leads the group not to interfere in civilians’ affairs, as far as locals meet a set of obligations. Renamo in most occupied territories in Mozambique established this form of presence. Traditional chiefs, known as regulos, ruled civilian affairs, but they had to ensure food provision to Renamo as well as civilians’ abidance by a set of rules (Geffray 1990; Young 1997; Weinstein 2007). Similarly, accounts of the interaction between the Sudan People’s Liberation Movement (SPLM) in Sudan and local communities in Tei Town portray the relation between the SPLM and civilians as a tense agreement, in which local chiefs assured some minimal compliance in exchange for greater safety of the community (e.g., Johnson 1998).
The existence of a social order of rebelocracy in war zones, in which armed groups rule more broadly, has received limited attention—often, even experts on civil wars doubt they exist at all. The emerging literature on rebel governance mentioned in the introduction has helped to counter this omission, especially by showing that armed groups often provide public goods. In an effort to show that rebelocracies are indeed common and warrant attention, I mention a few cases across the globe.
In Africa, descriptions of rebels providing public goods and creating new institutions abound. The Eritrean People’s Liberation Front, for example, provided health care, education, and dispute resolution schemes. They also implemented land reform and created a formal system of taxation and political councils (Cliffe 1984; Barnabas and Zwi 1997; Connell 2001; Pool 2001). The Tigray People’s Liberation Front (TPLF) in Ethiopia is also known for its provision of services and implementation of land reform (J. Young 1997). The National Resistance Army and the Rwenzururu Kingdom Government in Uganda have also been described as insurgencies engaged with ruling civilians comprehensively (Kasfir 2005; Weinstein 2007).
Insurgencies in Latin America have also established rebelocracies. Wickham-Crowley (1987, 1991) provides a comprehensive list of insurgent groups from the 1950s to the 1970s that acted as rulers in areas where they operated—including both well-known cases, like Cuba and Nicaragua, and more obscure ones, like Venezuela. In Cuba, the creation of administrative councils to deal with public health, the collection of taxes, and the enactment of new laws have been described by Guevara and Davies (1985) as well as by others (McColl 1969).
Asia and Europe are not an exception. The Liberation Tamil Tigers of Tamil Eelam (LTTE) in Sri Lanka ruled civilian affairs in a comprehensive fashion. Mampilly (2011) and Stokke (2006) offer a detailed account of their effective civil administration, which included education and health systems, a legal code with its corresponding judiciary, a police force, and even a bank. The Maoist Rebels of Nepal also created institutions to distribute land and food and set up courts to solve disputes (Kattel 2003). The resistance groups that fought against Soviet occupation in Afghanistan developed a bureaucracy that, while not sophisticated, was in charge of several regulatory tasks in local territories (Rubin 2002; Sinno 2008, 126-27). The Taliban also engaged early on in state-like activities in areas where they were present in Afghanistan (Sinno 2008). Mao’s Chinese People Liberation Army was also known for engaging in civilian rule. Provision of public goods and institutions by rebels in Europe was also documented in the case of Greece (McColl 1969; Kalyvas, in press).
Evidence on militias and counterinsurgent irregular groups is more difficult to find, as these groups are understudied. However, some case studies suggest that they also often establish rebelocracies. For example, in the different armed conflicts that Afghanistan has endured over the last decades, several nonstate armed groups often became the de facto guarantors of local order, provided public goods, and co-opted or eliminated other sources of authority in their areas of influence. Warlords’ organizations like Massoud’s and Wahdat—two of the many that were competing for power after the Najib regime collapsed in 1992—created and developed civilian institutions in some of the territories under their influence (Sinno 2008, 193, 217).
This evidence suggests that, indeed, the typology I propose captures variation that we see in war zones. Yet, it is a very simple typology that classifies rebels’ influence on institutions—when institutions do exist—into two discrete categories. Are rebelocracy and aliocracy capturing two distinct realities?
This typology was developed on theoretical grounds in 2009—prior to data collection—and was therefore not informed by the evidence presented in the previous section. A good test of the depth and parsimony (Gerring 1999, 380) of the typology is therefore to inspect the data in order to assess whether the typology captures “natural” groups. Using k-means cluster analysis, we can calculate the Euclidean distance between observations on the basis of measures of the dimension the typology is trying to capture—to wit, armed groups’ influence on local institutions. Based on this distance, we can identify two groups or clusters that are homogeneous—that is, each cluster contains elements that are as close as possible to the other elements in the cluster.
I use five indicators of armed groups’ influence on local institutions. Each is an index ranging from 0 to 1, measuring armed groups’ influence on a domain of local life based on a series of variables. “Public goods” measures whether the group provided education, health, or infrastructure either directly or by putting pressure on the local authorities; “economy” measures whether the group regulated different legal or illegal economic activities in the locality; “justice” measures whether the group became the de facto court—that is, if people turned to it to solve conflicts; “politics” measures whether the group intervened in locals’ decisions to vote and for whom; “rules over private conduct” measures whether the group established rules to regulate sexual practices, personal appearance, mobility, free speech, or domestic violence.
Plotting the means of the different proxies of armed groups’ influence on local institutions by cluster shows that there is a strong positive correlation between all the different indexes within each cluster, and a strong, negative correlation between all indexes across both clusters (Figure 4). This means that armed groups’ intervention in local institutions tends to be either broad or narrow, regardless of which sphere of local life we examine. This result suggests that the simple typology presented in the previous section does a good job of capturing two very distinct types whose elements share many attributes. The results of the cluster analysis are not sensitive to adding or dropping variables or to changing the seed.
Cluster analysis of wartime institutions.
If we look at other variables that capture additional kinds of intervention by armed groups on local life, the results remain unchanged. Figure 5 shows the scores of the locality dyads in each of the two clusters identified previously for two additional variables: “Social influence,” which refers to whether combatants participated for community social events by playing soccer, drinking beer, or attending parties with locals and “community meetings,” which captures whether combatants organized meetings to discuss the problems of the community. As before, cluster 1 shows low levels of interventions and cluster 2 shows high levels.
Cluster and others forms of influence.
In sum, armed groups’ intervention in different spheres of life tends to covary; that is, in most cases, when a guerrilla or paramilitary group intervenes in politics, it also intervenes in economic activities and social relations. Likewise, when a group abstains from intervening in one sphere, it tends to also neglect other domains of life. The typology captures well these two distant situations: under rebelocracy, armed groups intervene broadly and tend to influence greatly a myriad of activities and under aliocracy, on the contrary, they intervene minimally and have a much more limited capacity to transform local life.
Discussion: The Explanatory Power of Wartime Institutions
The evidence I have provided shows that institutions do operate in many areas of conflict, and that the level of influence that armed groups have on those institutions varies greatly. If we know—from political science, sociology, and economics—that institutions shape behavior, why have we overlooked their relevance when studying decision making in the midst of war?
My theoretical prior is that in wartime—as in peacetime—institutions can shape available alternatives and payoffs, beliefs, and preferences via different mechanisms. They can also transform the nature of civilian–combatant relations, which can in turn have different effects on both actors’ choices. In this section, I illustrate the explanatory power of wartime institutions by giving a few examples of how different types of wartime social order may shape phenomena at the micro and macro level, as well as our strategies to investigate them.
The first way in which taking into account wartime social orders can illuminate our study of civil war is by questioning the validity of common assumptions in theories of different phenomena. Macro-level theories usually assume that civilians have little agency—they are either supportive of rebels only for their ideological ties or coerced by them—and even micro-level theories that take agency into account tend to theorize choices within institutional vacuums. Similarly, armed groups are thought to rely solely on violence to achieve their ends, leaving many strategies—like transforming local institutions—out of the analysis. Inasfar as armed actors strategically shape institutions, and such institutions influence actors’ choices, the premises on which many theories are built should be reconsidered.
Incorporating institutions in our analysis of the conduct of war can improve theory building in at least three ways. First, armed groups’ strategies might be explained at least partially by their desire to establish certain institutions; second, once in place, those institutions can shape combatants’ choices; and third, those institutions can also shape civilians’ decision making. I illustrate the potential insights that can come from investigating these causal links with a few examples.
Violence could be better understood if we consider the institutional context in which it is used. Kalyvas (2006) has convincingly argued that selective violence against noncombatants at the local level is shaped by the distribution of territorial control between the warring sides. Following this theory, selective violence should be higher in areas where two or more armed groups fight for control (unless control is evenly shared), because in that situation, combatants have a high demand for information on defectors, and civilians have great incentives to provide it. Indiscriminate violence, on the other hand, is expected to be high in situations where combatants demand information but civilians fail to provide it. If we factor institutions in, however, we could expect variation not only across but also within territories under dispute. Local institutions can shape civilians’ decision to share information with armed actors; for example, communities living in rebelocracy might be more likely to share information than communities that have preserved their institutions and have, therefore, been less influenced by combatants. The implication would be that communities living under aliocracy are less likely to be targeted with selective violence but are more likely to endure indiscriminate violence than communities living under rebelocracy.
Furthermore, the existence of social orders in war zones suggests that violence serves other purposes beyond punishing and preventing collaboration with the enemy: it may be used to bring about a particular form of order and also to preserve it. Under this view, violence would thus be not only a means to deter defection to the enemy (Kalyvas 2006) or a by-product of poor recruitment (Weinstein 2007) but also a tool to enforce new institutions.
Considering institutions can also be crucial for our understanding of armed groups’ capacity to expand. It might be that controlling a territory is a function of the group’s capacity to bring about rebelocracy: rules on economic, political, and social affairs can shape local dynamics in ways that allow armed groups to acquire goods, information, and support. Understanding the conditions under which armed groups are able to set up the institutions they want can give us clues about where they expand, how they do it, and why they succeed or fail.
Investigating wartime social orders can also illuminate our study of civilian choice in war zones. A parallel between the existence of distinct social orders and regime types is useful to think about the effects that such variation may have on civilian behavior. As with any regime—like democracy or dictatorship—the specific characteristics of these social orders have far-reaching consequences on those living in them. They determine the set of forbidden behaviors and individual rights, the actor or organization that individuals seek for solving their conflicts, the persons and institutions they have to obey, the existence of channels to communicate with those who command them, and the availability of procedures to defend themselves when accused of misconduct. Even their private life—their attire and sexual conduct—can be subjected to strict regulation.
One of the shortcomings of the literature on civilian choices in civil war has been abstracting the institutional contexts in which such choices are made. This neglect is consequential for our study of key phenomena such as civilian collaboration, recruitment, and displacement. Whether a social contract exists between a community and an armed group, and what specific behaviors the group adopts, should be taken into account when trying to understand why civilians behave in the way they do in war zones. Furthermore, understanding civilian choice requires carefully theorizing how armed groups’ influence in so many aspects of local life can transform shared beliefs, create new sets of available alternatives, awaken emotions that change preferences, and create new ways of reading the local (and national) status quo.
Institutions can also be a crucial mediating variable of the effects of conflict. Research on the consequences of war on health, education, and economic well-being need to take into account the ways in which wartime institutions may catalyze or ameliorate the effects of war (Justino 2013). Similarly, studies on the political and social legacies of conflict need to consider the role of local institutions. Recent studies have found that violence increases collective action (e.g., Bellows and Miguel 2009; Blattman and Annan 2009; Gilligan, Pasquale, and Samii 2013); a thorough understanding of the ways in which social order is transformed by war is essential to identify causal paths.
Finally, the existence of wartime social orders also has implications for the validity of measures that are commonly used in studies of the micro dynamics of civil war. Scholars use different proxies of armed groups’ presence to investigate rebel behavior and war outcomes; those proxies often rely on simplistic assumptions about local order in conflict areas. For example, inferring that violence is a good proxy of presence (e.g., Acemoglu, Robinson, and Santos 2013) may lead to excluding precisely those places where armed groups have permeated institutions and local life to such extent that little violence is needed and, if used, is unlikely to be denounced, reported, and recorded.
Turning to the macro level, there are many ways in which understanding the role of local institutions can advance our study of the conduct of war at an aggregate level (Justino 2013). To start with, advances in our study of individuals’ choices illuminate questions on macro-level outcomes. As I mentioned before, all claims about the onset, conduct, and termination of war rely to some extent on assumptions about how actors react on the ground. The better our theories on individual choices are, the stronger the foundations of our theories on macro-level outcomes can be.
More directly, inquiring about wartime institutions can give us insights about how war evolves over time. If transforming local institutions and establishing rebelocracy are essential for holding territorial control over the long run, only certain rebel groups may be able to fight long wars. Clearly, rebelocracy is not equally likely everywhere; the more fertile the ground for rebel rule, the more likely it is that rebels keep their strongholds, as combating a group that has managed to rule populations tightly across the country is more difficult than confronting one that holds little sway over civilians. If correct, this line of reasoning could shed light on the conditions for the onset and duration of civil war. Furthermore, understanding wartime institutional change can shed light on questions about the termination of war, the post-conflict transition from political to criminal violence, and the durability of peace (Justino 2013).
Another implication has to do with democracy in contexts of civil war. If nonstate armed groups are likely to co-opt or capture existing authorities and combatants can manipulate elections, significant questions arise regarding the workings of democracy in contexts of civil war. On one hand, the capture of democracy raises questions about the effects of specific wartime dynamics on the quality of democracy (Arjona and Chacón 2013). It also leads to all sorts of normative concerns about the policy of building democracy in contexts of civil war. On the other hand, it raises questions related to the strategic use that armed groups can make of democracy as a means to acquire both power and legitimacy. What is the effect of democracy on rebels’ strength or bargaining power? How does democracy alter the odds of success of alternative means to end conflict?
Research on local institutions can also make a tremendous contribution to the study of counterinsurgency. The debate about how “to drain the water in which the fish swim” has taken armies around the world to try indiscriminate violence, selective violence, and civic–military operations. The United States has recently embraced an approach that emphasizes the provision of infrastructure and both private and public goods. If institutions shape civilian collaboration and, therefore, armed groups’ capacity to preserve territorial control, counterinsurgency should pay greater attention to institutions. Furthermore, recipes that work well in some contexts may fail in others precisely due to institutional changes brought by war. Assuming that wartime interventions operate in an institutional vacuum hinders our capacity to identify the effects of alternative policies.
Turning to postconflict studies, by identifying variation in civilians’ experience of war, this article calls for a more disaggregated approach to postconflict outcomes. The presence of armed groups brings about profound changes to local communities, shaping not only how war affects them (as victims) but also how they react to it (as agents). Variation in wartime social order is, therefore, likely to transcend the war, creating challenges and opportunities for reconciliation, reconstruction, and development (Arjona 2009). Furthermore, as Justino (2013) argues, in as far as institutional outcomes tend to be persistent, the wartime transformation of institutions may impact the very possibility of durable peace.
More generally, different kinds of institutions, including those coming from state agencies and traditional authorities, can be deeply transformed by the rules that operate during war. If fostering trust on the state, recovering the authority of traditional institutions, or promoting community cohesion are among the challenges that postwar societies face, understanding the ways in which war transforms social order is a necessary step. In addition, by virtue of modifying institutions, war may transform state-building trajectories. Future research needs to theorize the specific ways in which civil war in general, and wartime institutional change in particular, impacts the quality, the strength, and the evolution of the state.
Finally, wartime institutions are important not merely for their potential explanatory power of other wartime or postwar phenomena. Variation in local institutions during war is a phenomenon in and of itself that warrants explanation. Understanding why order emerges in war zones and what form it takes is an important question, as it relates to civilians’ experiences of war, armed groups’ strategies, and wartime transformation of key aspects of society. Even more, the question of why rebels—or counter-rebels—manage to rule communities or fail to do so is essentially a more general question about how order is created, preserved, and destroyed.
Footnotes
Acknowledgment
I am grateful to Laia Balcells, Kate Baldwin, Fernán González, Francesca Grandi, Francisco Gutiérrez, Yvan Guichaoua, Michael Kalin, Kosuke Imai, Patricia Justino, Stathis Kalyvas, Matthew Kocker, Jason Lyall, Zachariah Mampilly, Roger Petersen, Jonah Schulhofer-Wohl, Scott Straus, Andres Vargas, Elisabeth Wood, and two anonymous reviewers for their valuable comments. I want to thank the outstanding research assistance of Eduardo Alvarez, Andrés Felipe Aponte, Victor Barrera, Camilo Corredor, Laura Otálora, Silvia Otero, Natalia Perez, Javier Revelo, Camila Reyes, Alejandra Rodríguez, and Sara Zamora. I am also indebted to the many persons who agreed to share their views and experiences with us.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by funding from the U.S. Social Science Research Council (SSRC), the Harry Frank Guggenheim Foundation, Yale University, and the U.K. Economic and Social Research Council (ESRC).
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
