Introduction
VARIETIES OF REBELLION
Lukumbi Village, Uganda, 1981
Word of the rebels came first in the form of rumors. "There are men who move at night," he was told. "They live deep in the forest." "They are strangers to this zone." But Samuel had never seen them with his own eyes.1
Government soldiers, however, were known to Samuel and his neighbors. They came in packs, demanding to know where the guerrillas were hiding. Out of fear, people would sometimes offer information. Samuel recalled one person who volunteered to take government soldiers to a rebel camp. They shot him from behind as he led them into the forest. The government troops claimed he was plotting to have them ambushed.
Soldiers maintained a regular presence in the village: knocking on doors, hurling threats, and exacting punishment on those who refused to cooperate. Most of the soldiers were of another ethnic group from another region of the country, and the enmity between locals and those in the military stretched back decades into Uganda's colonial and immediate postcolonial experience. Political sympathies in the village thus lay with the men hiding in the forest. But the soldiers had some local collaborators - representatives of the government's political party, chiefs who owed their authority and wealth to political elites in the capital, and groups of youths from minority ethnic groups in the area. They informed government soldiers about the presence of guerrilla units and identified community members who were offering support and comfort to the insurgents.
1 Interview, Semuto, November 19(B), 2000. The letter following the interview date is used to distinguish among multiple interviews conducted on a single day. "Samuel" is used as a pseudonym to protect the anonymity of the respondent.
1
Inside Rebellion
These collaborators often disappeared. At the time, Samuel knew nothing more than that the rebels had come to take them away. He later learned that informers underwent a process of political indoctrination in the forest. If they accepted responsibility for their actions and agreed to support the rebellion, the rebels welcomed them into the movement. If they refused, they were killed.
So, the first time Samuel saw the rebels, he was scared. It was 2:00 in the morning. He and his father were outside herding cattle that had escaped from their kraal. They were stunned to see a group of men moving in the dark, entering the forest with bags on their heads. His father shrieked, catching the attention of the men who rushed over quickly to quiet them down. Samuel recognized a local leader among the men in the group. He spoke to the father and encouraged him to offer his support to the rebels. But he also warned them: "in case you report us, we will come and kill you with your children."
Soon after this encounter, Samuel and his father began to supply food to the insurgents in the forest. Although they feared the rebels at first, the behavior of the government soldiers solidified their support for the insurgency. Government troops continued to wreak havoc in the village, killing people and raping women. Samuel recalled thinking that the rebels were different. While the government soldiers were intent on killing them, the insurgents played by different rules.
In the ensuing months, the rebels dispatched political cadres into the villages, tasked with organizing resistance councils in support of the movement. Formed following a public meeting in which cadres described the political goals of the insurgency, the councils were elected by community members to administer the areas with "justice and impartiality." They had primary responsibility for maintaining security in the zone; local militias were organized to track the movements of government troops and warn of impending attacks. The councils also ensured a steady supply of recruits to the insurgents, using their local knowledge to root out thieves and lazy types. Samuel moved quickly to join a resistance council for his village.
This brought him into closer contact with the rebels. He found that the rebels were "so disciplined because they hated the government soldiers for their misconduct among the civilians." There were acts of indiscipline because, as Samuel remembered, "it's human." But resistance councils were encouraged to report this misbehavior to insurgent commanders. And these violations of the "laws the soldiers had" were punished.
2
Introduction
Martngué Village, Mozambique, 1979
Luis was only seventeen when the rebels arrived without warning in September 1979.2 They first cornered a local leader and sliced off his ear, accusing him of supporting the government. The insurgents then gathered the population in the center of the village. They killed seven people in a public display of violence. Each victim was accused of having family members that supported the government. Young boys were also abducted that day Luis recalled - taken away from the village and not seen again during the course of the war.
The rebels remained in the village for three months following the first attack. Life under the insurgents was difficult. Luis's father was a wealthy cultivator who was quickly identified by the group as it sought to secure sources of food. The rebels arrived at their home one day on the outskirts of the village, and shot their guns in the air, demanding contributions of food. Luis' family complied by offering a contribution, but it was deemed insufficient. The soldiers then robbed them of all the food in their house. Neighbors also suffered at the hands of the rebels. Luis explained that husbands with beautiful wives were obligated to make trips to the rebels' base to deliver food; when they arrived, the wife would be asked to stay, while the husband was sent home. If they refused to leave their wives behind, they were beaten severely.
On occasion, the insurgents spoke of "political things." They organized one public meeting in which a rebel commander explained the purposes of the war: "they were fighting against the government and its system," Luis remembered. An antigovernment posture had resonance in central Mozambique because the government was viewed as biased in favor of the ethnic groups of the south. Moreover, its campaign of socialist transformation in the countryside, which was just getting underway, threatened to undermine local practices of governance, land rights, and cultivation that were highly valued by the population. But the message of the insurgents was clouded by their behavior. Theft, abduction, and rape were common practices even during this first visit of the rebels to Marínguě.
The rebels' stay in the village was short-lived. After three months, a government counterattack forced the insurgents to flee to the bush. To protect people and make the village easier to defend, dispersed patterns of land
2 Interview, Maringue, May 23(B), 2001. "Luis" is used as a pseudonym to protect the anonymity of the respondent.
3
Inside Rebellion
holding were quickly replaced by communal villages at the government's behest. On a daily basis, villagers went to the fields to work the land, but returned to the confines of the communal village at night, where they were watched closely by soldiers and locally organized militias.
The insurgents returned in 1982, again without warning. They launched a daylight raid on four communal villages near Maríngué, including the one where Luis and his family now stayed. When villagers heard the shooting, they fled to the bush and hid wherever they could. The rebels looted household belongings and then burned each and every home to the ground. Government soldiers were unable to repel the attack, and members of the local militia, like everyone else, fled quickly to protect their lives. Luis and his family, along with many of his neighbors, remained on the outskirts of the village. They constructed ramshackle houses near the land they worked and never returned to the center of town.
They elected instead to live in a zone of rebel influence. The village itself was too much of a target, and government forces had proven unable to offer them protection. But insurgent areas were not much better. Luis described life under the rebels as "very bad." People were forced to make weekly contributions of food to the insurgents; those who refused were punished severely. Civilians had no choice but to comply with the requests made by the rebels. While Luis understood the purposes of their military campaign, he could not make sense of their coercive tactics or brutality.
An old African saying likens the experience of civilians in wartime to that of the grass underneath the feet of dueling elephants. The grass is trampled by two outside forces over which it has little control. But it would be a mistake to imagine that civilians lack agency in all civil wars or that the abuse of noncombatants is simply a by-product of the battle between opposing armies. Civilian populations - their interests, their resources, and their support - figure centrally in the political and military struggles that plague many developing countries. And we see in these stories that the civilian experience of war differs across contexts. Understanding why requires a sustained look "inside rebellion" in an effort to determine why some insurgents who choose to challenge the state turn out to be thugs, and others, revolutionaries.
The Puzzle
One conservative estimate of the direct death toll from civil wars since 1945 exceeds 16 million, more than five times as many people as have died
4
Introduction
in interstate wars.3 In the 1990s, over 90 percent of deaths caused by war occurred in internal conflicts.4 The lingering effects of violence, including disease, famine, and the destruction of economic and social infrastructure, substantially (even exponentially) increase the numbers of those who perish as a result of fighting in developing countries.5
Journalists and scholars who write about civil war assume that violence against civilians is one of its fundamental characteristics. War correspondents report on untold human suffering in the Democratic Republic of the Congo and Colombia but never stop to ask why the war in Congo has claimed nearly 100,000 lives directly in battle, while Colombia's civil war, which has lasted more than four times as long, is responsible for only one-fifth the killing. Analysts explore the tactics and strategies of insurgents fighting in the bush in Sierra Leone and Nepal but fail to grapple with the reality that while rebels in both countries sought to capture state power and remove undemocratic regimes, those in Sierra Leone hacked, raped, and pillaged their way through the countryside in a war that cost more than 10,000 lives, whereas insurgents in Nepal transformed local structures of governance, mobilized large numbers of civilians, and killed fewer than 1,000 people in nearly 10 years of fighting. Scholars discuss the hundreds of thousands of innocent civilians who have perished in Chechnya and Mozambique but never think to inquire why the Russian government bears responsibility for most of the killing in Chechnya, but insurgent forces were largely responsible for the violence in southern Africa. From conflict to conflict, we are made keenly aware that the primary victims of violence in civil war are noncombatants caught in the midst of fighting. Yet we know surprisingly little about why some civil wars are so much more violent than others or why some groups commit horrendous atrocities and others do not.
Scholars who write about the violence that characterizes civil war tend to begin by exploring state behavior as a critical first step in making sense
3 James Fearon and David Laitin, "Ethnicity, Insurgency, and Civil War," American Political Science Review 97 (2003): 75-90.
4 A new dataset on battle deaths in civil war offers a strong empirical basis for comparing human suffering in warfare across conflicts and over time. See Bethany Lacina and Nils Petter Gleditsch, "Monitoring Trends in Global Combat: ANew Dataset of Battle Deaths," European Journal ofPopulation 21 (2005): 145-66.
5 The best recent study documenting the indirect consequences of war is Hazem Ghobarah, Paul Huth, and Bruce Russett, "Civil Wars Kill and Maim People - Long After the Shooting Stops," American Political Science Review 97 (2003): 189-202.
5
Inside Rebellion
of the consequences of warfare.6 One cannot fault that starting point: some of the most extreme cases of civilian brutalization have come at the hands of national governments and their militaries. State violence has included communist mass killings such as those in the Soviet Union, China, and Cambodia; ethnic genocides like those of Armenia, Nazi Germany, and Rwanda; and counterinsurgent massacres of the type perpetrated in Guatemala and Afghanistan. Yet rebel groups often share responsibility for the violence inflicted upon noncombatants, and the tactics, strategies, and patterns of violence exhibited by nonstate actors in civil war remain largely unexplored.
Some rebel groups abuse noncombatant populations, while others exhibit restraint, discipline, and control. Insurgent leaders in some countries transform local structures of governance, engaging civilians in the process of affecting political change; others build administrative machineries that do nothing more than extract resources. In some contexts, rebel groups kill their victims selectively, while in other environments violence appears indiscriminate, even random. Movements sometimes loot and destroy the property of civilian populations, while at other times they protect it from government attacks. In this book, I present a theory that accounts for the different strategies pursued by rebel groups in civil war, explaining why patterns of insurgent violence vary so much across conflicts. By "violence," I refer both to the character of insurgent attacks (the extent to which groups use force selectively to punish and prevent defection) and its aggregate level (the number of killings, abductions, rapes, and so on). Drawing on interviews with nearly two hundred combatants and civilians in three countries, I build my explanation by looking inside rebel organizations at their origins and structures. In focusing on origins, I highlight the factors that shape a rebel group's membership. In examining structures, I demonstrate how the profile of a group's membership constrains the organizational strategies its leaders can pursue, the structures of governance it can build in liberated
6 There is no shortage of sophisticated research on the causes of mass killing as perpetrated by states. For recent academic perspectives, see James Ron, Frontiers and Ghettos: State Violence in Serbia and Israel (Berkeley: University of California Press, 2003), and Benjamin Valentino, Final Solutions: Mass Killing and Genocide in the Twentieth Century (Ithaca, NY: Cornell University Press, 2004). For a more popular history of genocide in the twentieth century, see Samantha Power, A Problem from Hell: America and the Age of Genocide (New York: Harper Perennial, 2003).
6
Introduction
zones, and its capacity to use violence in a strategic, selective, and limited fashion.
The Argument
I argue that differences in how rebel groups employ violence are a consequence of variation in the initial conditions that leaders confront. Factors that raise or lower the barriers to organization by insurgent leaders - in particular whether material resources to finance warfare can be easily mobilized without civilian consent - shape the types of individuals who elect to participate, the sorts of organizations that emerge to fight civil wars, and the strategies of violence that develop in practice. My central finding is that rebel groups that emerge in environments rich in natural resources or with the external support of an outside patron tend to commit high levels of indiscriminate violence; movements that arise in resource-poor contexts perpetrate far fewer abuses and employ violence selectively and strategically.
The Mechanisms
Fighting an insurgency involves building an organization capable of challenging a government militarily. Many barriers to the organization of insurgency exist. Potential rebels must raise capital to finance the logistics of a military campaign, recruit foot soldiers willing to risk their lives in battle against a stronger government force, and generate support from civilians who can supply food, information about the location and strategies of government forces, and valuable labor in support of the movement. In confronting these challenges, rebel leaders may draw on two types of endowments: economic endowments, which come from diverse sources, including natural resource extraction, taxation, criminal activity, or external patronage; and social endowments, including shared beliefs, expectations, and norms that may exist in (or be mobilized from within) certain ethnic, religious, cultural, or ideological groups.
This book shows how the initial endowments to which rebel leaders have access shape the organizations that emerge and the ways in which different rebel groups ultimately use violence. First, resources shape the membership profile of a rebel group. That is, initial endowments constrain the set of recruitment tactics leaders can employ, altering the benefits and costs of joining in such a way as to affect the calculations individuals make
7
Inside Rebellion
about whether to participate in an insurgency. My argument begins, then, with the most fundamental - and perhaps most studied - aspect of rebellion: participation.7
Attracting recruits to participate in civil war is not an easy task. The work of rebellion is difficult and potentially dangerous. And when a rebel group sweeps to power and transforms the political regime in a country, it is difficult for it to exclude nonparticipants from the new freedoms that come with political change.8 So while the potential costs of participation make joining unattractive, the promised benefits may not tip the balance. The onus is on leaders of rebel groups to develop appeals that motivate participation in high-risk collective action.
Recognizing that potential recruits trade off the costs and benefits of participation, rebel leaders often offer selective incentives to motivate participation.9 As Samuel Popkin first argued, they find ways to distribute material benefits such that participants are rewarded for exerting effort and nonparticipants are excluded. At the same time, rebel leaders develop appeals around ethnic, religious, cultural, or ideological claims, reminding individuals of their membership in or affiliation with aggrieved groups, playing on their allegiance to a particular set of ideals, or activating norms of cooperation and reciprocity in order to motivate participation.
This book builds on the insight that recruitment strategies depend a great deal on the incentives that are likely to motivate individual participation, but it extends the discussion to reflect an additional consideration. I break with the common assumption that all potential recruits are of the same value to a rebel group, recognizing instead that rebel groups can attract both
7 James Scott's research on the sources of peasant protest provides the intellectual foundation for many subsequent studies of political violence and civil war, including this one. In exploring the causes of anticolonial movements and protests against the expansion of markets, Scott highlighted the ways in which market forces disrupted traditional peasant ways of life, threatening local institutions and, ultimately, peasants' ability to survive. His focus on the logic underlying an individual's calculus to resist social change is reflected in the major studies that have followed. While many have varied their assumptions about how peasants behave, key contributions have continued in Scott's tradition of highlighting the choices peasants make. See The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast Asia (New Haven: Yale University Press, 1976).
8 Olson's classic statement on barriers to collective action has influenced subsequent work on organization in contexts ranging from insurgency and war to political organization and community action. See The Logic of Collective Action: Public Goods and the Theory of Groups (Cambridge, MA: Harvard University Press, 1965).
9 Samuel Popkin, The Rational Peasant: The Political Economy of Rural Society in Vietnam (Berkeley: University of California Press, 1979).
8
Introduction
high-commitment and low-commitment individuals. High-commitment individuals are investors, dedicated to the cause of the organization and willing to make costly investments today in return for the promise of rewards in the future. Low-commitment individuals are consumers, seeking short-term gains from participation. The problem is that even though potential recruits are aware of their level of commitment, rebel leaders do not have access to this information. The recruitment process therefore involves both motivating participation and attempting to attract the right kind of recruit to the organization.
A group's endowments shape the potential strategies that its leaders can employ.10 Groups with access to economic resources are able to translate those endowments into selective incentives, or payoffs, in order to motivate individuals to join the rebellion. Resource-constrained groups must develop alternative strategies. They make promises about the material benefits that may accrue to individuals in the future and the collective benefits that the country will reap from a rebel victory, but these promises are only credible where leaders draw on social endowments that tie them to potential followers by means of ethnic, religious, or ideological ties. They can also mobilize within ethnic networks, religious organizations, formal and informal associations, and communities to activate norms and expectations that promote or reinforce cooperation.
Different initial endowments, then, create a situation in which there is variation in the opportunity that participation presents to potential rebels, and rebel groups attract different types of people depending on the costs and benefits of participation. Where participation is risky and short-term gains are unlikely, rebel groups tend to attract only the most committed investors. I call the movements in which they engage activist rebellions. Where
10 My focus on variation in the initial conditions rebel leaders confront has its roots in the literature on social movements. Resource mobilization theories focus attention on how the resources and organizational capabilities of groups help to explain their mobilization potential. See J. Craig Jenkins, "Resource Mobilization Theory and the Study of Social Movements," Annual Review of Sociology 9 (1983): 527-53. A second strain of argument highlights the importance of political opportunities in paving the way for collective action. Exogenous changes in the environment, such as openings in access to power, electoral realignment, and cleavages within elite groups, make it possible for resource-poor movements to emerge. A key work in the literature on political opportunity structures is Sidney Tarrow, Power in Movement (Cambridge: Cambridge University Press, 1992). More recent research links variations in endowments and opportunities to the strategies movement leaders employ, demonstrating how broad structural factors constrain the repertoires of action available to different groups. See Doug McAdam, Sidney Tarrow, and Charles Tilly, The Dynamics of Contention (New York: Cambridge University Press, 2001).
9
Inside Rebellion
participation involves fewer risks and individuals can expect to be rewarded immediately for their involvement, groups tend to attract consumers, who take part in what I call opportunistic rebellions.
The membership profile of a rebellion then affects its internal organization and the strategies it pursues in war. Rebel leaders confront a series of difficult choices as they design their organizations and engage civilians. Two merit particular attention here. They must decide how to ensure that orders are followed and they must extract the resources they need from civilians without destroying their base of support and sustenance. The nature of the strategic dilemma leaders face at each step in this process of organizational growth, and the options available to them as they respond, are themselves a function of the resource environment in which the group formed and its profile of recruits. Activist movements can maintain internal discipline by drawing on established norms and networks enabling them to decentralize power within their armies; opportunistic rebellions must permit indiscipline in order to maintain their membership, while holding on to the reins of military strategy. Activist insurgents can often obtain resources by striking cooperative bargains with noncombatant populations; opportunistic groups tend to employ coercive tactics because they cannot credibly commit to non-abusive behavior.
The outcome I ultimately seek to explain is how rebel groups use violence. Linking differences in the initial conditions leaders confront to variation in the membership and internal structure of groups helps to make sense of the character and level of violence committed by rebels against civilian populations. Structures of internal control and external governance shape the capacity of rebel groups to discipline the behavior of their members and influence the expectations of civilians about the types of behavior they will see when the rebels come to town. Where social and political ties can be employed to develop effective organizations, rebel leaders have a greater capacity to use violence strategically. Because they have clear guidelines about how combatants should behave and strong mechanisms for enforcing discipline, activist insurgencies are better able to selectively identify targets, implement attacks, and discipline the use of force. The short-term orientation of opportunistic insurgencies, on the other hand, tends to be detrimental to civilian populations. Without local ties, opportunistic groups have more difficulty identifying potential defectors and are prone to make mistakes. A constant demand for short-term rewards also drives combatants to loot, destroy property, and attack indiscriminately. A group's early missteps then initiate a cycle of civilian resistance and retribution by group members
10
Introduction
that spirals quickly out of control. The indiscriminate character of insurgent behavior results in higher aggregate levels of violence as civilian resistance makes it increasingly difficult for opportunistic insurgencies to operate.
These patterns of insurgent behavior, established early on in the conflict, tend to persist over time. This is true in particular for opportunistic movements that, as they attempt to maintain their membership in response to battlefield losses and changing government strategies, have little flexibility to adapt their internal organizational structures and a limited capacity to alter the expectations of civilians about how the rebels will behave. Activist rebellions also exhibit tremendous resilience, although the norms and expectations that tie members together and link groups to the communities in which they operate are vulnerable to fraying in the face of disruptive shocks that alter the membership of the group or undermine its internal structures that facilitate cooperation. When such changes do occur, activist movements tend toward opportunism, a change in strategy which is difficult to reverse. Figure 0.1 summarizes the relationship between resource endowments and rebel violence described in this study.
This book thus treats the question of why rebel groups abuse civilian populations in some contexts and not in others as the result of a process of organizational formation.11 In particular, it shows how the initial endowments accessible to rebel leaders shape and constrain their strategies as they respond to five distinct challenges of rebel organization. Other scholars interested in civil war violence have overlooked the different origins and internal characteristics of groups, focusing instead on dynamics of contestation and territorial control. Theories of contestation locate the sources of insurgent violence in the desire of losing rebel groups to signal their resolve; theories of control predict high levels of violence in places where insurgent groups are present but unable to control territory
11 This argument implies substantial path dependence in rebel behavior. Theories of path dependence were initially applied to explain the divergence in economic performance across countries. Douglass North showed how North and South America embarked on distinct trajectories of economic development even though after independence they shared constitutional forms, abundant resources, and similar international opportunities. The key difference is that North America benefited from its English colonial legacy of decentralized patrimony, whereas South America was stuck with the centralized authoritarianism and clientilism of its former Spanish rulers. See Institutions, Institutional Change, and Economic Performance (Cambridge: Cambridge University Press, 1990). More recendy, such arguments have been applied to explain divergence across forms of political organization. See Robert Putnam, Making Democracy Work: Civic Traditions in Modern Italy (Princeton: Princeton University Press, 1993).
11
Five challenges of rebel organization Initial endowments
Challenge 1
Recruitment: Motivating participation, obtaining high-quality recruits
Two organizational strategies Economic endowments Social endowments
1
Opportunistic strategy
Appealing to short-term, material interests —> attracts a membership pool of "consumers"
1
Activist strategy
Making promises, activating norms, appealing to nonmaterial interests —> attracts "investors"
Challenge 2
Challenge 3
Control:
Ensuring that combatants follov orders
Governance:
Managing civilians in areas of territorial control
Permit indiscipline to maintain membership but control military strategy from the top
Extract resources through coercion
Maintain discipline by establishing norms and sharing power within the army
Obtain resources by striking a bargain with civilians
Challenge 4
Challenge 5
Violence:
Punishing individuals or groups for acts of defection
Resilience:
Maintaining rebel membership over time
Employ violence indiscriminately and to a significant degree
Coercive tactics persist as groups find it difficult to change their membership and alter civilians' expectations
Use force selectively and in a limited fashion
Disciplined behavior often persists, but activist insurgencies are more vulnerable to disruptive shocks that change membership and alter expectations
Figure 0.1.  The Relationship between Resource Endowments and Rebel Violence.
Introduction
unilaterally. Both approaches link insurgent violence to the relative weakness of an insurgent group. In contrast, this book finds that insurgent violence follows from an organization's material strength. Perhaps surprisingly, it argues that resource wealth, which one might imagine would be a prerequisite for a group's possessing long time horizons and forward-looking behavior, is associated more often with high levels of indiscriminate violence, as leaders prove unable to use their wealth in support of their groups' social purposes.13
To demonstrate how a group's initial endowments condition its strategies, I draw on empirical materials related to four rebel groups that were engaged in three different conflicts: the National Resistance Army (NRA) in Uganda (1981-86), the Mozambican National Resistance (Resistencia Nacional Mocambicana, or Renamo) in Mozambique (1976-92), and two factions of Peru's Shining Path, or Sendero Luminoso (1980-92). The NRA, which successfully challenged and overthrew the regime of Milton Obote, was responsible for comparatively little of the violence Uganda experienced; it was government-sponsored massacres that resulted in the deaths of hundreds of thousands. The NRA was recognized instead for its discipline and restraint and for the cooperative relationships it built with an ever-growing network of civilian supporters. Renamo received international attention unprecedented at the time for the abuses it committed against Mozambique's civilians during a nearly twenty-year struggle against the government. Abduction, rape, and the hacking off of limbs and other body parts were all-too-common parts of the civilian experience of the war in Mozambique. In Peru, the national organization of the Shining Path adopted violence as a strategy for cleansing the countryside yet built an organization that implemented that violence strategically and
12 Lisa Hultman makes the argument about signaling resolve in a paper on civil war violence. See "Killing Civilians to Signal Resolve: Rebel Strategies in Intrastate Conflicts" (paper presented at the annual meeting of the American Political Science Association, Washington, DC, September 3, 2005). On territorial control, see Statins Kalyvas, The Logic of Violence in Civil War (New York: Cambridge University Press, 2006).
13 Resource wealth has been linked to short-term-oriented behavior in other contexts as well. The impact of sudden, unexpected wealth on small firms is covered in Olivier Blan-chard, Florencio L6pez-de-Silanes, and Andrei Shleifer, "What Do Firms Do with Cash Windfalls?" Journal of Financial Economics 36 (1994): 337-60. As with the behavior of states in the context of resource booms, the authors find that managers invest cash windfalls inside the firm, even when the investment opportunities are not attractive, as a way of ensuring the success of the firm with them at the helm. They call this the "agency model of managerial behavior."
13
Inside Rebellion
systematically, with errors and missteps punished severely Its regional committee in the Upper Huallaga Valley (UHV), on the other hand, made a name for itself for its indiscipline and mistreatment of civilians who lived in its zones of control.
The specific question that I address is why rebels in the National Resistance Army and the Shining Path showed such restraint in the use of force against civilians, while insurgents fighting under the banner of Renamo and the regional committee in the UHV committed serious abuses. My answer is that the strategies rebel groups pursue follow from their organizational structures. I show that groups organized largely on the basis of economic endowments, such as Renamo (supported by an external patron) and the regional committee of the Upper Huallaga Valley (supported by the drug trade), are populated by opportunists, lack mechanisms for disciplining behavior, and tend to commit widespread abuses against civilians. Rebellions organized around social endowments, such as the National Resistance Army (organized around ethnicity) and the Shining Path (organized around ideology), attract committed recruits, establish structures that facilitate cooperation and discipline, and employ violence selectively, controlling combatant behavior to a significant degree. Table 0.1 compares rebel violence across these four groups.
The interpretation I advance in this book is not intended to draw out all the pathways through which variation in the initial conditions rebel leaders confront shapes the strategies their groups ultimately pursue. The particular initial conditions I emphasize (economic and social endowments) are only one part of the larger landscape from which groups emerge. Other factors matter for the viability of insurgency, shaping both who decides to rebel and the nature of the organizations that evolve to wage civil war. The most important factor outside of those discussed here is the strength of the state's bureaucratic and military machinery. Where national governments are strong - able to implement their policies and police the countryside without challenge - the barriers to organization are often too high for insurgent groups to develop. Where no state exists or the government is so weak it does not rule outside of the capital, the barriers to organization may be so low that almost anyone can launch a rebellion. In this book, I consider civil wars in which the relative power of the state was similar and state power therefore cannot be the major explanation for variation in the structure and strategy of the insurgent groups. In all three countries, the state was weak enough that an insurgent organization could develop a rural base for insurgency; at the same time, it was strong enough that,
14
Introduction
Table 0.1. Variation in Rebel Violence
			Share of	
		Government-	Killings	Character
	Estimate of	Sponsored	Attributed	of Rebel
	Battle Deaths	Mass Killing?	to Rebels	Volence
Uganda	104,800	Yes		Selective
(1981-1986)				
Mozambique	145,436	No	Vast majority	Indiscriminate, brutal
(1976-1992)				
Peru (1980-1992)	69,280	No	Difference between rebel and state killings not statistically significant	Selective
Peru-Huallaga	ca. 14,000	No	Difference	Selective at first, later
Valley			between rebel	indiscriminate,
(1980-Present)			and state killings statistically significant, with rebels killing more	increasing brutality
Sources: For data on combat-related deaths in Uganda, Mozambique, and Peru, see Bethany Lacina and Nils Petter Gleditsch, "Monitoring Trends in Global Combat: ANew Dataset of Battle Deaths," European Journal of Population 21 (2005): 145-66. Coding of government-sponsored mass killing is from Benjamin Valentino, Paul Hum, and Dylan Balch-Lindsay, "Draining the Sea: Mass Killing and Guerilla Warfare," International Organization 58 (2004): 375-407. The share of rebel killings in Mozambique is taken from Africa Watch, Conspicuous Destruction (New York: Human Rights Watch, 1992). For the share of rebel killings in Peru and Peru-Huallaga Valley, see Patrick Ball, Jana Asher, David Sulmont, and Daniel Manrique, "HowMany Peruvians Have Died?" (AAAS report, American Association for the Advancement of Science, Washington, DC, August 2003). Assessments of the character of rebel violence in Uganda are made in Ondoga Ori Amaza, Museveni: From Guerrilla to Statesmen (Kampala: Fountain, 1998). For the character of rebel violence in Mozambique, see Alex Vnes, Renamo: Terrorism in Mozambique (London: James Currey, 1991). The characterization of rebel violence in Peru and Peru-Huallaga Valley is from the Final Report of the Comision de la Verdad y Reconciliation, available at http://www.cverdad.org.pe, vol. 1, Chapter 3, and vol. 4, Chapter 1.4.
from the perspective of the rebels, the conflict was asymmetric, with the government's military representing a real and credible threat.
My purpose here is to develop a simple, general model that links the resources groups have at their disposal in organizing violence to the strategies rebels pursue in their relations with noncombatant populations.
15
Inside Rebellion
The recognition that internal conflict represents a significant threat to human well-being in developing countries and that understanding the origins and nature of civil war is a pressing intellectual challenge is already reshaping the academic and policy agenda as it relates to conflict. Recent research linking natural resources to the onset and duration of conflicts raises as many questions as it answers about the motivations of insurgent groups and the factors that affect their viability. The emergence of powerful rebel groups whose behavior is as abhorrent and destructive as that of states that maim and massacre in fits of genocidal destruction has drawn critical attention to the task of opening the black box of insurgent organizations. The principal purpose of this book is to bring these strands of thinking together, first, by shifting the focus of a burgeoning body of research from the causes of civil war to the equally pressing question of the determinants of violence within civil war; second, by highlighting the essential role that nonstate actors play in the process of organizing violence; and, third, by developing theory and empirics that link factors we already know are critical to understanding the viability of insurgency to the question of why and under what conditions insurgent groups abuse innocent civilians.
Civil War, Violence, and Organization
To investigate issues of rebel strategy and behavior and to link patterns of violence to the characteristics of insurgent organizations is to adopt a focus on civil war, violence, and organization that not only builds on a growing tradition of research but also departs from conventional approaches in a number of important ways. It is common practice to treat civil wars as a form of political violence with three major characteristics: (1) they involve fighting between agents of a state and organized nonstate groups that seek to capture control of the government or over a region or to influence government policy by means of violence; (2) the fighting kills at least 1,000 people over its course and 100 on average in every year; and (3) at least 100 people die on both sides of the conflict.14 While a transparent and defensible definition has allowed for significant progress in understanding
14 Fearon and Laitin, "Ethnicity, Insurgency, and Civil War." A similar definition of civil war is employed in most quantitative studies of violence. See J. David Singer and Melvin Small, "Correlates of War Project: International and Civil War Data, 1816-1992," ICPSR study no. 9905 (Ann Arbor: Inter-University Consortium for Political and Social Research, 1994); and Michael Doyle and Nicholas Sambanis, "International Peacebuilding: A Theoretical and Quantitative Analysis," American Political Science Review 94 (2000): 778-801.
16
Introduction
the causes of civil war, recent research challenges the idea that the mechanisms leading to conflict can be usefully compared across a large set of countries and types of warfare.15
In shifting our analysis of civil war from the macro level to the micro level, it is important to distinguish among different types of conflicts, as the distinct characteristics of warfare shape the strategies of organization that groups ultimately pursue. Rather than draw difficult-to-defend distinctions between wars based on the expressed motivations of belligerents, I adopt a more transparent typology that distinguishes among wars in which groups seek to capture the center, conflicts fought over secession, and wars in which belligerents use violence but have no interest in achieving territorial control of any sort (notably terrorist groups). Trying to understand the organization of violence in a broad range of civil wars is a worthy goal; however, a more narrow focus on theory building in the context of the classical case of insurgency offers enormous leverage in identifying the distinct organizational challenges groups confront and the various factors that shape the choices of rebel leaders. In particular, the imperative of capturing a national territory creates a unique set of opportunities and constraints that may or may not hold in other types of warfare. The prospect of territorial control disciplines rebel behavior across geographic regions because it embeds insurgents in an interaction with civilians that, if they are successful, will be repeated over time. Rebel groups seeking control of the state constitute 56 percent of belligerent groups in civil wars fought since 1945, making the set of cases under consideration here part of a considerable population of armed groups.16 Of course, one implication of this approach is that, to the extent that the organization and strategies of violence differ in secessionist wars and conflicts not fought over territory, the account presented here is limited in its explanatory power. I return to the question of how my model applies to secessionist movements in the conclusion.
In addition to the way in which it defines civil war, this account differs from many treatments of conflict in the outcome it seeks to explain. Whereas most studies of civil war analyze where and when conflict takes place or the conditions under which wars can be brought to an end, this
15 Nicholas Sambanis, "Using Case Studies to Expand Economic Models of Civil War," Perspectives on Politics 2 (2004): 259-79.
16 Monica Toft has gathered as yet unpublished data on the objectives of belligerent groups since 1945.
17
Inside Rebellion
study considers why some civil wars are more violent, brutal, and destructive for civilian populations than others. This approach recognizes that patterns of violence and abuse vary in important and measurable ways across conflicts, over time and across space within conflicts, and across belligerent groups.
Following the lead of scholars of genocide and state-directed violence, I adopt an understanding of violence that recognizes its multifaceted nature and seeks to measure variation in its intensity.17 I assume that the violent strategies implemented by armed groups comprise a range of behaviors that include, but are not limited, to killing. Counting the number of deaths due to violence in war is not an easy task, but capturing the dynamics of more nuanced patterns of belligerent-civilian interaction in the processes of recruitment, resource extraction, and governance poses an even more difficult challenge. Making sense of patterns of abuse requires a combination of methods of analysis: the counting of combat deaths, the coding of human rights violations and atrocities, and qualitative analysis of in-depth interviews with noncombatants about their experiences of war. In adopting multiple methods, I conceptualize variation in the intensity of violence as being reflected in both the character and the level of violence committed against civilians. While variation in levels of violence is relatively easy to assess, differences in the character of violence are reflected in its selectivity and brutality. Violence is selective if it targets individuals or groups that threaten to undermine a rebel organization; selective violence has a tactical purpose for the group. When violence is used selectively, civilians can be relatively certain that cooperation can be exchanged for the right to survive. Indiscriminate violence makes no distinction among potential victims, neither protecting supporters nor punishing defectors. Brutality refers to behaviors, including amputation, rape, abduction, and pillaging, that often accompany rebel attacks but are above and beyond what is required to send a signal of the costs of defection.18 A major goal of this study is thus to
17 See, for example, Barbara Harff, "No Lessons Learned from the Holocaust: Assessing Risks of Genocide and Mass Political Murder Since 1945," American Political Science Review 97 (2003): 57-73; and Benjamin Valentino, Paul Huth, and Dylan Balch-Lindsay, "Draining the Sea: Mass Killing and Guerrilla Warfare," International Organization 58 (2004): 375-407.
18 This study departs from other recent explorations of violence in its exclusive focus on insurgent behavior. In contrast, Kalyvas looks at how the use of violence varies across groups within a conflict, testing theories of violence on geographic and temporal patterns of coercion by both insurgent and government actors. See Logic of Violence.
18
Introduction
introduce a way of thinking about rebellion that provides analytical leverage in making sense of different patterns of warfare perpetrated by nonstate actors, with obvious implications for both conflict prevention and conflict resolution.
A final way in which this book differs from current treatments of civil war is in the primacy it ascribes to the issue of how violence is organized. "Organization" here refers to the internal characteristics of a movement: its membership, policies, structures, and culture. Studies explaining why some countries experience civil wars and others do not or why some civil wars last so much longer than others have taken organization as a given. Such research typically links macro-level factors, such as a country's wealth, ethnic diversity, or regime type, to the onset or duration of violence without being specific about the micro-level processes through which war is actually carried out. Most analysis in this area begins with an economic theory of rebellion in which groups in some way trade off the costs and benefits of mounting resistance, but few studies attempt to document an empirical basis for linking the factors that affect the viability of insurgency to the recruitment of insurgents, the building of organizations, or the implementation of violence. With a careful study of the internal dynamics of rebellion, this book aims to fill the gap.
In doing so, it seeks to move the discussion of insurgent organization beyond a weak typology of "new" and "old" civil wars that has recently gained favor in academic and policy circles toward an understanding of different rebel behaviors that is rooted in a theory of how groups form.19 Many are right to point out variation in the characteristics of civil war and its perpetrators, but to explain this variation in terms of Cold War and post-Cold War dynamics is simplistic, theoretically unsatisfying, and empirically wrong.20 Following the lead of scholars such as Roger Petersen, Elisabeth Wood, and Stathis Kalyvas, this book instead grapples with variation in the conduct of warfare by looking inside insurgent organizations at how they
19 Descriptions of "new civil wars" first gained widespread prominence due to the work of Robert Kaplan. See "The Coming Anarchy: How Scarcity, Crime, Overpopulation, and Disease Are Rapidly Destroying the Social Fabric of Our Planet," Atlantic Monthly 44, February 1994. Academic scholars picked up on the distinction in subsequent work. See David Keen, The Economic Functions of Violence in Civil Wars, Adelphi Paper 320 (New York: Oxford University Press for the International Institute for Strategic Studies, 1998); and Mary Kaldor, New and Old Wars: Organized Violence in a Global Era (Stanford: Stanford University Press, 1999).
20 For a powerful critique of the distinction between new and old civil wars, see Stathis Kalyvas, "New and Old Civil Wars: A Valid Distinction?" World Politics 54 (2001): 99-118.
19
Inside Rebellion
form.21 It identifies a set of common organizational challenges that rebel leaders confront. Making sense of the incentives that motivate participation in insurgency, the strategies groups pursue in ensuring that combatants follow orders, and the challenges they face in establishing territorial control puts a premium on uncovering the choices and strategies of - and the constraints facing - those who participate in violence. It is at this level that theorists can begin to develop logically coherent and testable theories to account for the variation in behavior observed in conflict. This book thus draws inspiration and insight from studies of organizational design in literatures that explore social movements, criminal organizations, political parties, and competitive firms. At the same time, it builds on the tradition of scholarship on insurgency and peasant organization that, in studying the dynamics of communist movements in Southeast Asia, first drew attention to the internal characteristics of rebellions that account for their emergence, growth, and effectiveness.22
Structure and Agency
I account for variation in the internal structures of rebel organizations by examining the initial conditions leaders confront. My argument about the determinants of violence builds from the idea that different organizational types evolve depending on the viability of insurgency; factors that heighten or lessen the barriers to organization receive primary attention. This approach is well accepted in literature on the political economy of rebellion. Where I challenge conventional views about the determinants of strategy more directly is in my contention that patterns of violence are a direct consequence of the endowments leaders have at their disposal as they organize. I argue that groups commit high levels of abuse not because of ethnic hatred or because it benefits them strategically but instead because their membership renders group leaders unable to discipline and restrain the use of force - and membership is determined in important ways by the endowments leaders have at their disposal at the start of a rebellion.
21 Roger Petersen, Resistance and Rebellion: Lessons from Eastern Europe (New York: Cambridge University Press, 2001); Elisabeth Wood, Insurgent Collective Action in El Salvador (New York: Cambridge University Press, 2004); Kalyvas, Logic of Violence.
22 See, for example, Paul Berman, Revolutionary Organization: Institution-Building within the People's Liberation Armed Farces (Lexington, MA; Lexington Books, 1974). This followed on earlier work by Lucian Pye (Guerrilla Communism in Malaya [Princeton: Princeton University Press 1956]) and Phillip Selznick (77re Organizational Weapon: A Study of Bolshevik Strategy and Tactics [Santa Monica, CA; RAND Corporation, 1952]).
20
Introduction
Conceptualizing strategy in this way divorces it from agency. Decisions about recruitment, organization, and violence cease to be driven by the actions of individuals and become, instead, choices made under binding constraints imposed by the resources a group has at its disposal and the membership it has attracted to participate. Leadership, skill, and ideology all take a backseat to broader, macro-level factors that structure the universe of possibilities individual rebels confront. Rebel organizations are transformed, at least in theory, from groups denned by the personalities and ideologies of their leaders to teams of would-be rebels shaped by conditions that affect the viability of challenging the state. Violence becomes the natural outcome of a path of organizational evolution rather than a strategic choice made in response to changing conditions on the ground.
Some readers may question this approach. They will point to the powerful influence of ideologies in conditioning individual behavior, arguing that variation in the content and character of political appeals must be central to explaining violence. They will highlight the critical role played by leaders of various insurgent groups in obtaining outside support, crafting opposition coalitions, and militarily outwitting the enemy as evidence of the central-ity of leadership. They will point to the idiosyncratic character of warfare itself, with substantial change taking place over time in the strength of the enemy, the depth of the support base, and the role of external actors - all of which militate against an argument that implies substantial path dependence in behavior. While recognizing that the endowments leaders have at their disposal certainly shape their strategies, they will emphasize that such endowments are neither fixed nor exogenous and that leaders play a role in generating material resources and fostering social ties. These are important challenges to my argument, and I wish to offer an initial response.
When I highlight the importance of structure over agency, I am not denying that leadership, charisma, and skill affect the strategies rebel groups adopt in civil war. Nor am I suggesting that strong, coherent ideological platforms have no impact on the decisions leaders make about who to recruit and how to do it, what types of training to require of participants, and which strategies of mobilization to implement among civilian populations. Fighting wars (at least, fighting them successfully) requires leadership. Organizers must motivate and challenge untrained peasants to take up arms and engage enormous risks in fighting for a cause. The ability to inspire and lead must be a part of the story of group formation and survival. In the same vein, rebel organizations actively promote strong ideological positions, and from communist insurgencies to fundamentalist religious movements to groups
21
Inside Rebellion
organized around ethnic ties, these ideological foundations are evident in the strategies groups use to recruit and the structures they build to govern their movements. Nonetheless, my point in focusing attention primarily on the ways in which resource endowments shape the strategies leaders have at their disposal is simply that crucial explanatory variables such as leadership and strong ideologies are themselves endogenous to the process of group formation. A convincing explanation of rebel behavior must be able to account for the emergence of strong leadership in some contexts and not in others, and it must enable us to explain why some groups with strong ideological foundations form at some times in some places but not at other times in other places. This book's focus on factors that affect the viability of insurgency - in particular, on the economic and social endowments that can be activated by insurgent leaders - helps to make sense of the conditions under which leadership and group solidarity are observed. In developing the argument in subsequent chapters, I take seriously the idea that rebel leaders can sometimes shape the endowments accessible to them in organizing violence but show that differences in the viability of insurgency can account too for the leaders that come to the fore and the endowments they have at their disposal.
Looking Ahead
This book is organized into three parts. Part I, comprising Chapters 1 through 4, focuses on the structure of rebel organizations. It shows how the initial conditions rebel leaders confront shape the strategies they choose in recruiting participants and managing a growing insurgent organization.
Chapter 1 introduces in more detail a basic economic framework for understanding rebel organizations. It locates this book's focus on organizational structure in the context of two previous approaches to studying the dynamics of rebellion: those that view rebel groups as social movements and those that view them as states. Chapter 2 describes the four rebel groups that form the basis for comparative analysis. It provides the reader with a concise summary of the three civil wars, reviewing the conditions that contributed to the rise of the insurgency and introducing the key players involved in the process of organizational design.
Chapters 3 and 4 then address the factors that account for variation in the internal structures of rebel organizations. Chapter 3 explores the challenge posed by recruitment, as rebel leaders seek to overcome free rider problems
22
Introduction
that impede collective action while also motivating potential participants who are likely to be committed to the organization. The model shows how variation in the initial conditions rebel leaders confront gives rise to varying strategies of recruitment, shaping the membership (and trajectory) of each rebel organization. Chapter 4 focuses on the training and management strategies of rebel groups. It highlights the difficulty of maintaining organizational control as groups decentralize their operations during wartime, and it explores the strategies groups employ to ensure discipline among their forces. This chapter shows how investments in training and decisions about organizational hierarchy depend in important ways on the nature of the membership a group has recruited.
Once the factors that explain variation in the internal structures of rebel groups have been established, Part II tackles the central question of this book: Why are rebel groups abusive toward civilians in some contexts and not in others? Chapter 5 explores the nature of the political structures that groups build to mobilize civilian support and extract the resources necessary to maintain the organization. It identifies a central problem that groups face in gaining civilian compliance: how to credibly commit to extract only what is necessary for the sustenance of the movement. Drawing on evidence from interviews with combatants and civilians in rebel-held zones, the chapter shows how the characteristics of insurgent organizations constrain the range of governance structures they can employ in areas of civilian control.
Chapter 6 presents evidence on how rebel groups use coercion as a tool to maintain civilian support and build their militaries. Drawing on new quantitative data from Uganda, Mozambique, and Peru, it charts differences in the intensity and character of violence across the four rebel organizations. Taken together, Chapters 5 and 6 show how differences in the internal structures of groups, linked to the resources they have at their disposal, can account for variation in patterns of violence across civil wars.
The model linking resources and group structures to patterns of violence implies substantial path dependence. Chapter 7 explores the power of the argument by examining the evolution of rebel groups over time. It explores key instances of combat success and failure in each conflict, highlighting the ways in which groups sought to hold their organizations together by reinforcing rather than reforming internal structures and practices established in the earliest stage of the conflict.
In presenting the argument, this book proceeds in a fashion atypical of social science research. Rather than outlining a theory at the outset and walking the reader through four extended case studies, I instead move
23
Inside Rebellion
from one step in the process of organizational evolution to the next. In each chapter, narrative description mixes with analysis as the stories of the National Resistance Army, Renamo, and the two factions of the Shining Path motivate and reinforce the theoretical arguments I advance.
In Part III, Chapter 8 takes the argument beyond the initial set of cases to explore its explanatory power in other contexts. It first tests the insight linking resources to recruitment in the cases of Sierra Leone and Nepal. The chapter then turns to quantitative data on the level and character of violence across civil wars for cross-national evidence linking variation in the costs of organizing insurgency to patterns of civilian abuse. The book's final chapter relaxes the assumption that rebel leaders begin with access to a fixed set of endowments and presents evidence that groups built around social endowments get crowded out in countries where the barriers to organization are low. It then concludes with a discussion of the policy implications of differences in the structure of rebel groups for policy makers who seek to influence and restrain the behavior of nonstate actors in civil war.
24