Abstract
Unable to attract enough voluntary recruits, many rebel groups rely on force to fill their ranks. Given that the group used force to compel individuals to join, a coerced conscript would be presumed unlikely to be loyal and would be expected to desert at the first opportunity. Yet, groups that have relied on coerced recruitment retain their members just as well as, if not better than, rebel armies that rely on voluntary methods of recruitment. This is a puzzle. How do rebel groups maintain allegiance and prevent desertion, especially if they rely on abduction to staff their ranks? A recruit can be forced to join a rebel group, but continuing to rely on coercion to enforce retention is too costly and not sustainable. These groups must find a way to reduce the costs of retention. The solution to this puzzle rests in the mechanisms of socialization that shape the allegiance of forcibly recruited soldiers. Socialization mechanisms are traced through three outcomes: compliance (or Type 0 socialization), role learning (Type I socialization), and norm internalization (Type II socialization). Integrating socialization theory and a rational choice analysis demonstrates that mechanisms that alter preferences through Type II socialization are effective in retaining recruits; the highest level of retention occurs when several mechanisms work in concert. Illustrative case studies of the Lord’s Resistance Army from Uganda, the Revolutionary United Front in Sierra Leone, the Maoists in Nepal, and the Liberians United for Reconciliation and Democracy (LURD) show that a reliance on child soldiers, group assets (pecuniary and non-pecuniary), organizational structure, and the nature of military contestation shape when different mechanisms are effective or not.
Membership matters. To be viable, a group must have members, whether a political party, labor union, sorority, sports club, or a rebel army. An organization must convince people to join and persuade them to continue to participate. For a military organization engaged in armed conflict, failing to address these two tasks could be fatal – threatening the survival of the group and the lives of its members. For all military organizations, every desertion increases the risk of capture or death for the members that remain loyal. Recruitment and retention are imperative.
Facing an inability to attract enough recruits voluntarily, many rebel groups rely on force. Given that the group used force to compel individuals to join, they presumably have little to induce these individuals to remain of their own free will. Forced conscripts would presumably not be loyal to the group and would be expected to desert at the first opportunity. Yet, groups that have relied on forced recruitment could retain their members just as well, if not better than, rebel armies that rely on voluntary methods of recruitment. 1 Groups such as the Revolutionary United Front (RUF) in Sierra Leone, the Lord’s Resistant Army (LRA), Renamo in Mozambique, and others have been able to retain forcibly recruited soldiers from four to seven years on average. In contrast, groups that rely on voluntary recruitment typically retain their soldiers for two years. Many factors shape how long an individual will remain in an organization, but voluntary as opposed to forced recruitment is not a significant factor (Haer et al., 2013; Weierstall et al., 2013; Vermeij, 2013). One would think that soldiers forced to join would leave the group as soon as they could, but they do not. What then accounts for these long periods of time spent with these organizations? This leads to the central research question of this article: how do organizations that rely on forced recruitment retain the loyalty of their troops?
To understand the processes that foster retention, this article features the organization and the members that compose it. Exploring these mechanisms and how they differ provides new insights into understanding group allegiance. Between-group variation will be investigated, featuring the mechanisms that serve to retain recruits. The analytical focus of this article is on the mechanisms of socialization that shape the allegiance of forcibly recruited soldiers. Three types of socialization are examined: (a) compliance (Type 0), whereby punishment and fear motivate retention; (b) role learning (Type I), whereby a member of the group alters his or her behavior to fit in; and (c) internalization (Type II) that induces preference changes leading to group allegiance (see Checkel, 2017).
By solving the puzzle of how rebel groups that rely on forced recruitment retain their forces, we also gain more general insights as to why military groups employ different recruitment strategies under varying conditions and how they prevent desertion and foster allegiance among their troops. The article proceeds as follows. First, the phenomenon of forced recruitment is analyzed. I then examine the choice of desertion versus remaining within the organization. Next, the puzzle of why forcibly recruited soldiers remain loyal to the organization is explicated. The analytical heart of the article examines the role different paths of socialization play as a mechanism of retention.
Forced recruitment
Abduction, conscription, fear, and intimidation serve as the principal means of recruitment for many armies. Indeed, one-third of all African rebel groups have relied on coerced recruitment (Beber & Blattman, 2013; Eck, 2014). The manner of forcible recruitment varies. It may come in the form of armed soldiers entering a refugee camp (Achvarina & Reich, 2006), a school (Becker, 2010), or villages and forcing individuals to join their group at gunpoint. Conscription is another form of forcible recruitment. Many nation-states rely on mass conscription, whereby every young man is obligated to serve in the military. 2 Non-state actors also may rely on forms of conscription as a means of recruitment. The Liberation Tigers of Tamil Eelam (LTTE) in Sri Lanka for a period required all households under their control to provide one soldier; the household thereby decided who would serve or not, but the choice was forced (Becker, 2010).
Some groups rely almost completely on abduction as a means of recruitment. In the Democratic Republic of Congo (DRC), the Alliance of Democratic Forces for the Liberation of Congo-Kinshasa (AFDL) and the Congolese Democratic Rally (RCD) also relied heavily on forced recruitment (Haer et al., 2013; Eck, 2014; Haer & Banholzer, 2015). In Northern Uganda, the Lord’s Resistance Army (LRA) had been relying on forced recruitment at an astonishing rate of 90% for over two decades (Annan et al., 2006; Beber & Blattman, 2013; Obayashi, forthcoming). The estimated rate of forced recruitment in Renamo in Mozambique was also for a period late in the war as high as 90% (Weinstein, 2007). In Sierra Leone, the Revolutionary United Front (RUF) is estimated to have abducted 87% of its recruits (Humphreys and Weinstein, 2008); others report an abduction rate of 72% (Mitton, 2012). The point, however, is not the precise percentage. 3 What is important is that these organizations depended on forced recruitment.
Why force recruitment? This question can be answered by drawing on standard labor economics, by examining the underlying manpower needs of a military organization. Demand and supply for military labor are both important. 4 Just like the standard supply and demand graphs of economics, labor demand can be depicted in a two-by-two graph with price (i.e. the participation and compatibility constraints of the potential recruits) on the vertical axis and quantity (i.e. the number of soldiers) on the horizontal. The quantity of military labor demanded by armed groups can be taken to be an array of different quantities desired at various prices. The demand function thus arises from the marginal product of labor. The labor demand curve slopes downwards, as more labor is demanded at lower prices. The supply curve tends to be upward sloping. The standard labor supply curve is a function of the relation between the responsiveness (i.e. the elasticity) of recruits to the ‘wages’ (i.e. the pecuniary and material incentives) offered by a rebel group and a potential recruit’s reservation wage. The insights from this model do not necessarily rely on an assumption of market efficiency and are therefore more generalizable. 5
Most analyses of rebel group recruitment focus exclusively on supply factors, such as the degree and extent of poverty and education, or the ethnic and religious composition of a country. When government actions provoke grievances and a desire for retribution, this may also affect supply (Andvig & Gates, 2010: 78). Demand will be shaped by the goals of the group and the nature of the contest with the government, which will determine the technologies of conflict and the tactical nature of warfare.
In situations in which a group is unable to recruit enough manpower, it may resort to forced recruitment. In other words, under conditions of excess demand, groups may resort to forced recruitment. Warfare is a bloody contest that can alter the supply and demand for manpower. Such changes often lead to shifts from voluntary to forced recruitment. Most groups that depend on forced recruitment began by relying on volunteers (Eck, 2014). For example, the Communist Party Nepal-Marxist (CPN-M) altered its recruitment strategy when it began employing conventional battlefield tactics in 2004. ‘Virtually all resources were redirected towards the fighting effort, and the strategic offensive phase demanded such a large number of troops that the Maoists increasingly resorted to coercion. During this period, abductions became extremely common, and the number of abductees increased more than twenty-fold’ (Eck, 2014: 387). In the early phases of its conflict with the Sri Lankan government (1970s to mid-1990s), the LTTE relied on voluntary recruits. After 1995 they started relying on abduction. After 2002 the LTTE adopted the ‘one-family-one-recruit’ strategy (Obayashi, forthcoming). The change in recruitment tactics was notable: in 1994, one in 19 child recruits was abducted. In 2004, the pattern reversed and only one in 19 was a volunteer (Human Rights Watch, 2004: 16). From the late 1980s until 1994, the LRA abandoned its voluntary recruitment strategy after being unable to secure enough volunteers (Titeca, 2010). After 1994 until the mid-2000s, an LRA operation would abduct 100 to 200 recruits, sometimes as many as 800. In 2003 the LRA recruited over 6,000 soldiers (Obayashi, forthcoming). Changes in a group’s resource base can also affect its ability to meet the equilibrium reservation benefit stream for many recruits. No longer able to entice volunteers, rebel groups often resort to forced recruitment. When Renamo was receiving substantial economic support from South Africa, it made extensive use of economic incentives to induce primarily adults to join the organization. When that financial support dried up, Renamo began to rely on forced recruitment and to recruit children, some very young. In fact, almost 9% of the recorded, demobilized Renamo soldiers were ten years old or younger (Weinstein, 2007).
Some groups rely on a mix of recruiting strategies, involving lower rates of forcible recruitment without much change over time. The FARC had abducted approximately 20% of its recruits (Gutiérrez-Sanín, 2010: 122), while the pro-government paramilitaries (United Self-Defense Forces of Colombia) relied on voluntary or forced recruitment varying by region (Gutiérrez-Sanín, 2010: 132). In Liberia during its civil war, LURD forcibly recruited approximately 19% of the adults and 27% of the children, while MODEL forced about 13% of its adult forces and 9% of the children to join (Pugel, 2010: 171).
Many groups rely almost exclusively on volunteers. For example, the Tigrayan People’s Liberation Front (TPLF) never forcibly recruited anyone (Eck, 2014; Young, 2006). The Ejército de Liberación Nacional (ELN) in Colombia has recruited mostly volunteers (Gutiérrez-Sanín, 2010: 132). Of course, the distinction between volunteering and being forced can often be blurred, especially at the individual level of analysis.
Forced recruitment is particularly prevalent among rebel groups that abduct children (Beber & Blattman, 2013; Gates & Reich, 2010). In such cases, the degree to which an armed group enlists children also will be influenced by ‘the characteristics of the accessibility of recruits: the number of usable children vs. adults in the area; the ease of capturing a child compared to an adult; the existence of exceptionally good “fishing grounds” such as refugee camps or secondary schools; and so on’ (Andvig & Gates, 2010: 90). Under such conditions the labor supply of children is flatter than that of adults, raising the demand for child soldiers (Andvig & Gates, 2010; Blattman, 2012).
The calculation of a forced recruit is different from that of a volunteer. What does forced recruitment mean for the participation constraint (i.e. the price of joining)? For the agent, it means a choice of joining the rebel group or being killed. Death would bring the discounted utility to zero. Thus, for the forced recruit, the participation constraint only need be greater than zero for him to join. This implies a minimum reservation benefit level for the agent that is approaching but greater than zero. In other words, given a choice between death and joining the group, an individual will choose to join. Coerced participation thus allows the rebel group to lower the reservation utilities of potential recruits. This means that the group need not provide much in terms of positive rewards for joining. The implication is that the supply of labor shifts outward thereby increasing the quantity (number of soldiers) ‘employed’ by the group.
In contrast for a volunteer, the group can offer a benefit stream (meaning a set of rewards over time) that induces the person to join and remain with the organization. Somewhat like employment in the job market, if the compatibility constraint is not met, some groups allow for negotiated settlement determining the conditions of exit. Of course, this involves more than giving your employer three months’ notice. The principle, nevertheless, is the same. The contracting agent is free to leave the organization. The ELN in Colombia, for example, allows people to leave and return, ‘a grave offense that the FARC would never permit’ (Gutiérrez-Sanín, 2010: 132, 134). Groups that force the recruits to join, coerce them to stay.
Force alone is sufficient to induce participation, but inducing a forced recruit to remain in the organization by means of coercion is costly. A group that relies on forced participation would be able to get more people to join the group than they would if they had relied exclusively on voluntary participation, but this recruitment strategy produces several problems. The first problem for the group is that by forcing a recruit to participate, the group will not be able to sort out undesirable recruits (violent psychotics, lazy good-for-nothings, or government spies) from effective rebel soldiers. Such a hidden information problem regarding the type of recruit is referred to as adverse selection. 6 Forced conscription may cast the net of recruitment so wide that little filtering or selection is evident. This will affect a group’s military effectiveness generally, but the costs are truly significant in battle. The second problem created by a forced recruitment strategy is that the incentive compatibility problem (i.e. the price or wage it takes to convince a recruit to remain with the group) will be amplified. The implication is that soldiers recruited forcibly will not work hard and are likely to desert. Given that a recruit has joined the group having been forced to join, to retain the allegiance of the soldier the group must either constantly impose a credible punishment that serves to deter desertion or introduce a stream of benefits that induces the recruit to remain. If fear of punishment is the only factor keeping an individual in a group, he will choose to escape at the first good opportunity.
The third problem is related to the first two problems. Punishment and material reward schemes are costly and only go so far in mitigating the problem of retaining soldiers who have been forcibly recruited. Constant policing to prevent desertion is labor intensive and hence costly (Eck, 2014). Moreover, the problem of capturing deserters, in order to punish them, can sometimes be extremely difficult. In addition, coercive techniques may deter desertion, but if employed too liberally can increase the number of casualties and more generally decrease combat effectiveness (Lyall, 2015; King, 2014).
A fourth problem stems from the very act of forcible recruitment, alienating the community from which a group recruits (Eck, 2014). The act of forcing people to participate especially by means of abduction reduces the likelihood that others will volunteer to join the group. Despite these costs many groups nevertheless recruit by force.
Desertion or allegiance
Desertion is about leaving an armed group. The deserter may join another rebel group, the government’s army or a paramilitary organization, or return to civilian life. Since the focus of this article is on retention and allegiance, the act of desertion will be conceptually limited to an individual leaving the group and no more. Given forcible recruitment, one would expect an individual to desert at the first opportunity. A recruit might desert in the heat of battle, or escape at night, or run away when foraging. Ultimately, however, the puzzle I am trying to solve is not the decision to leave, but the decision to stay in a group that one presumably did not want to join in the first place. 7
Desertion as operationalized in this article is not about group factionalization, whereby a commander leaves an organization and takes his men with him. In such cases, the commander is defecting, while his men remain loyal to him rather than to the organization. For example, take the case of Colonel Karuna, who, after fighting for the Liberation Tigers of Tamil Eelam (LTTE) for over 20 years, formed a breakaway faction (the Tamil Makkal Viduthalai Pulikal (TMVP)) that allied with the governmental forces of Sri Lanka (Obayashi, 2014). The problem of mass desertion is not considered here. The focus is on individuals’ decision to desert or to remain.
To understand why an individual would stay in a group he was forced to join, we still need to understand the implications of desertion for the armed group. Desertion poses one of the most significant threats to the leaders of any type of military organization, whether a governmental army, a paramilitary, or rebel organization. A deserter may be able to reveal valuable information to the enemy regarding the location of troops, headquarters, leadership of the group, the group’s commitment levels, weaknesses or particulars relating to strategy or tactics that would put the group at a serious disadvantage to governmental forces (Obayashi, 2014, forthcoming). Desertions could spell doom for a small guerrilla army, terrorist band, or social movement.
Throughout the ages, military organizations have attempted to address the threat of desertion, especially in battle. Even the most modern and powerful armies in the world have reserved their harshest punishment for desertion under fire. ‘The German army executed at least 15,000 servicemen in the second world war. The Soviet army may have executed that number at Stalingrad alone’ (Strachan, 2006: 215). Such punishments, however, have a mixed record at best for deterring desertion. The Italian army ‘executed more soldiers than any other army in the first world war, but at the same time its disintegration in the field in October 1917 was probably more extreme than that of any other army in the war’ (Strachan, 2006: 215). In contrast, the British executed 315 men in World War I and none in World War II; yet they faced comparatively fewer desertions than other countries, and fewer of their own men deserted in World War II than in the first (French, 2000: 138, 242–243). Harsh punishment alone does not guarantee that desertion is deterred.
The puzzle of loyal abducted fighters
Groups that have relied on forced recruitment would be expected to suffer from high levels of desertion, but many do not. Many armed rebel groups depend on recruitment strategies that rely on force and are yet able to retain them for long periods of time. For example, the average length of service in the Revolutionary United Front (RUF) in Sierra Leone was 5.4 years; for the Lord’s Resistant Army (LRA) in Uganda it was 4 years; and for Renamo in Mozambique it was 7.4 years (Vermeij, 2013). At their peak period of recruitment, each of these groups was composed of over 85% forced recruits. ‘The case of the RUF illustrates aptly that violent recruitment practices can be maintained’ (Eck, 2014: 395). The LRA has managed to survive for 30 years and for much of this time has relied almost exclusively on forced recruitment. If the LRA were not so effective in retaining a significant share of those it abducted, the organization would no longer exist (Titeca, 2010). In stark contrast to these groups that could retain their troops for long periods of time, Liberians United for Reconciliation and Development (LURD) did not rely on forced recruits (only 20% to 25% were forced), but tenure with LURD was only about two years (Pugel, 2010; Vermeij, 2013). In a study of ex-combatants from seven different groups in the DRC, Haer et al. (2013) find that on average those abducted spent more time with their group than those who joined voluntarily. Indeed, for groups that rely on voluntary recruitment, the average term of duty is two years. To solve this puzzle, we need to answer the question: how are groups able to retain troops forced to join?
Another aspect of this puzzle relates to promotion within the organizations that recruit by means of force. Officers constitute some of the most loyal and committed soldiers in a military organization. This is especially true for armed non-state actors, whereby all officers must work their way up the ranks. In the case of the LRA, this pattern is clear. In the late 2000s, 77% of all officers in the LRA had been abducted forcibly into the organization (Obayashi, forthcoming). Similar patterns were evident with Renamo in Mozambique and the RUF in Sierra Leone. The point is that a sizable majority of officers in the LRA, Renamo, and the RUF had at one time been forced to join the group. How did this organization not only manage to retain these forced recruits in the organization, but to promote them and put them in positions of leadership?
Let’s summarize the argument so far. A rebel group will rely on forced recruitment when it does not have the resources to recruit voluntarily and needs more soldiers than it can recruit – in other words, when there is excess demand. By relying on forced recruitment, a group can meet individuals’ reservation participation constraints. By threatening to kill them, it needs not provide many rewards to get them to join. This presents four problems: first, adverse selection (recruiting non-ideal soldiers); second, incentive incompatibility (the inability to retain the allegiance of recruits); third, because of these two problems, costly policing; and fourth, forced conscription can alienate the community from which a group recruits, increasing the need to further rely on coerced recruitment. These four problems and the lack of resources should result in desertion, but we do not see it. Forced recruits stay with the group.
Solving the puzzle
By featuring the dynamic processes affecting the interactions within a rebel group, we can gain some analytical insights to help solve this puzzle. Processes of socialization help groups retain the allegiance of their members. Bridging the uneasy relationship between socialization and strategic choice offers insights for understanding these processes. Strictly speaking, conventional rational choice models assume that preferences are fixed. In contrast, socialization theory features changing preferences and the internalization of new norms. The theories would appear to be mutually exclusive, but there is no inherent contradiction in the two theories. Indeed, significant insights can be drawn by bringing together strategic choice and socialization theory (see Checkel, 2017).
Checkel (2017) defines three types of socialization: Type I, Type II, and Type 0. For Type I the mechanism of role playing is active, but does not involve the internalization of norms. The mechanism for Type II socialization is more complete. An individual undergoes a socialization process internalizing norms that work to promote allegiance to the group, even for those forced to join the group. Type 0 is essentially a pure rational choice explanation response to incentives without socialization.
Two central parameters are featured in the analysis: punishment (its cost to the organization and its deterrent effect) and the reservation wage of a potential recruit, which affects the participation constraint (cost of joining) and the compatibility constraint (cost of retention). For a group to retain its troops, it must find ways to reduce these costs. Socialization mechanisms work by reducing the need for punishment and increasing the value of being a member of the group. So, what can a rebel group do? First, the rebel group can continue to rely on coercive measures, but to reduce the costs of punishment and policing. This is a Type 0 socialization. Second, alter the value of a forcibly recruited soldier’s outside option, thereby making staying with the group relatively more attractive than leaving. This could also be considered Type 0, but I will argue that it is better classified as Type I socialization. Third, rely on drills and training as a means of socializing the abducted recruit. This is also Type I socialization. Fourth, socialize recruits through internalization of the group’s norms, via indoctrination, psychologically transformational experiences, or altering beliefs. These are Type II socialization. All these mechanisms are employed by groups that forcibly recruit to retain their forces. An alternative hypothesis is that recruits do not know their preferences before they are abducted and learn that they like being a member of the rebel group even though they were recruited by force. While plausible, this is not as compelling as the other explanations. Type II psychological transformational experiences offer the most effective and cost effective means of ensuring recruit allegiance to the group. 8
My analytical focus is on the variation between groups that tend to rely on forced recruitment and those that generally rely on volunteers. The scope conditions of this analysis are primarily limited to groups that recruit by force. Volunteer armies may exhibit similar socialization mechanisms, but the conditions in which they work differ considerably. For a volunteer, the group offers a ‘wage’ (a combination of pecuniary and non-pecuniary benefits) that satisfies the participatory constraint. Because of material enticements (money, ‘loot’, or drugs) or ideological, ethnic, or religious identity rewards, a volunteer will also stay in the organization as long as his compatibility constraint is met. For ideology, religion, or ethnicity to matter to an individual, socialization has begun before they join the group. Pre-joining socialization also provides a common set of preferences binding a group to civil society (or at least a certain population). Moreover, given that a person has voluntarily joined, coercion plays a smaller role in deterring desertion. Indeed, voluntary armies tend not to exhibit the elaborate coercion mechanisms evident in armies that depend on forced recruits (King, 2014; Eck, 2014; Käihkö, 2015). Moreover, punishments are also generally not as harsh in military organizations that tend to rely on volunteers (Cagoco-Guiam, 2002; Gutiérrez-Sanín, 2010). Indeed, coercive discipline would not help with recruitment and retention of voluntary forces (Strachan, 2006: 226).
Type 0 socialization
When socialization is not evident and decisions are based on rational calculation of incentives (both rewards and punishment, pecuniary and non-pecuniary), following Checkel (2017) I refer to this as Type 0 socialization. From the individual perspective, this calculation will be whether the group can meet the recruit’s compatibility constraint – the wage necessary to deter desertion. When soldiers have been forcibly recruited, coercion brought them into the group and it is what in the short run will keep them in the group. Punishment and a climate of fear undoubtedly keep conscripts from trying to escape. This can occur in an environment of no socialization (Type 0). Punishment can take on a wide variety of forms, ranging from a temporary minor reduction in utility to an extremely harsh punishment with permanent ramifications. Death sentences are common.
Punishment schemes that are purely based on rational choice assumptions (Type 0) can be effective in the short run. Nevertheless, such mechanisms of retention rely on harsh penalties and a sufficiently high probability of being detected to be effective. Retention depends on the nature of the punishment and the chances of being caught. For an organization that has many forcibly recruited soldiers, keeping costs down is important, otherwise, it should expect widespread desertion.
From the organizational perspective (or at least from the leaders’), retaining a forcibly recruited soldier is costly. So, what can the rebel group do to lower the costs of coercion? For groups that use conscription rather than abduction as a means of forcible recruitment, punishment may not necessarily be limited to punishment of the agent himself. The penalty can also be extended to family members. This may be easier than trying to track down, capture, and punish the deserter himself. The LTTE’s method of ‘one household one recruit’ relied on a penalty whereby the family of a deserter was punished. Given the nature of their recruitment policy, information about which family the deserter belonged to could be ascertained (Obayashi, forthcoming). Nevertheless, the ability to match deserters and their families requires considerable organizational capabilities. Indeed, the problem with this mechanism as a means of deterring desertion is that it is extremely expensive for an organization to engage in this kind of policing. The information requirements and logistics of carrying out indirect punishment are extremely costly.
Another option is to enforce ghastly and highly symbolic punishment early on in an abducted recruit’s involvement with the group. The grisly spectacle can serve as a deterrent and thereby decrease the need for costly punishment in the future. Rumors of horrific punishments that await those who do try to leave can keep retention rates high. What matters is both the severity of the punishment that can befall a defector and the perceived probability that it will be carried out. The choice of individual for punishment might also be randomized to increase the terror within the organization and thereby decrease risk of desertion. Executions, however, do not seem to be effective in deterring desertion from national militaries (Strachan, 2006). In weak military organizations, which would define any rebel group relying on forced recruitment, such harsh punishments threaten to ‘rive the already frail organization’ (Käihkö, 2015: 3).
Another option is to spread the costs of policing within the organization. For example, in some units of the RUF in Sierra Leone, all soldiers were paired with a ‘buddy’ (typically not a friend). If you failed to sound the alarm when your buddy deserted, the penalty was execution (Gates, 2011). In such cases, the punishment extended beyond the individual, to other members of the organization. Punishment thereby functions as a deterrent to a deserter and compels the buddy to inform his superiors. The weakness in the system is that it creates an incentive to collude and defect together.
Spreading the costs of policing need not be limited to a ‘buddy’ system. Punishment matters not just to the one who tried to desert but to those who served as the executioners – those who punish. Punishment is not carried out by the leadership, nor by a punishment specialist (such as an executioner that arose as an institution in ancient kingdoms), but by fellow recruits, who were also recruited with force. This mechanism serves to move monitoring away from the typical vertical hierarchical pattern of supervisor–subordinate, officer–soldier towards a horizontal control mechanism whereby peers (or individuals at a similar rank within the organization) serve to monitor for desertions. The horizontal pattern of monitoring multiplies the possibilities of detecting defection or desertion as compared to the vertical. Such mechanisms, however, are difficult or impossible to implement top-down. They arise through socialization processes. ‘Without shared norms, coercion remains illegitimate’ (Käihkö, 2015: 25).
Punishment alone, even if made relatively cheap, proves to be inadequate for retaining forced recruits. The reservation wage of a potential recruit, which affects the participation constraint (cost of joining) and the compatibility constraint (cost of retention) also needs to be reduced. Type 0 socialization, with a total reliance on reducing the costs of coercion, does not provide a compelling solution to understanding why forced recruits remain in rebel groups. Coercion is no doubt important, but not as the sole mechanism of retention.
Type I socialization
Type I socialization involves learning to play a role in the organization. The individual need not internalize the new norms, but his compliance is manifest in his behavior. A new recruit will learn norms and rules from others, through processes of emulation, imitation, and experience. The main point though is that this type of socialization does not involve the internalization of new norms or the reordering of preferences.
Re-identification is a powerful socialization mechanism, whereby a new identity is constructed. Rituals, uniforms, and (even) mirrored sunglasses can create a sense of separate identity. Tattoos, scarring, head-shaving, and other ways of designating themselves to be part of a group are also employed – just as they are employed by urban gang members (see also Rodgers, 2017). These socialization mechanisms need not lead to an internalization of new norms, but they do work to create a separate identity distinguishing the recruit as a member of the group. This sentiment of identity is epitomized in the words of a RUF combatant who during an attack screamed: ‘you bastard civilians; you don’t like us and we don’t like you’ (Mitton, 2012: 115).
‘Training and socialization to the armed group take place both formally, through the immersion experience of “boot camp”, and informally, through initiation rituals and hazing. The powerful experience of endless drilling, dehumanization through abuse at the hands of the drill sergeant, and degradation followed by “rebirth” as group members meld recruits into a cohesive unit, whereby loyalties to one another are felt more strongly than previous loyalties, such as those to family’ (Wood, 2008: 546). Training plays a major role in providing cohesion in the modern volunteer national army (King, 2014). Training facilitates the quality of ties between soldiers and in turn shapes the achievement of unit goals and individual performance as well as an individual’s commitment to the group (Haer & Banholzer, 2015). Training facilitates the cohesion of the small military unit, the platoon (King, 2014).
Religious or spiritual groups often tap into rituals as a means of transformative socialization. Through a process of cleansing or rebirth a new recruit is inducted into the group. Xygalatas et al. (2013) find that extreme rituals enhance pro-sociality, thereby promoting group cohesiveness and cooperation. Their research features the role of physical pain. Military training, especially boot camp, can be seen to exhibit some of these characteristics, especially the pain part.
Another option to encourage Type I socialization is to alter the value of the outside option for a forcibly recruited soldier. The implication is that by making it costlier to leave the organization, the reservation wage to meet the compatibility constraint is lowered. This makes it cheaper to retain the recruit. An example of this kind of action would be to force the new recruit to commit an atrocity in his/her home village, severing a child’s connections to his family and community, thereby making it more difficult to return (Gates, 2011). Both the RUF and LRA made extensive use of this practice. 9 Cohen (2017) examines a similar mechanism affecting female soldiers who had participated in gang rapes in Sierra Leone. Having nowhere in particular to go, a recruit may be induced to stay with the group. But not because of positive incentives: abducted RUF recruits did not feel safer inside the organization (Humphreys & Weinstein, 2008). In such a situation, the recruit identifies as a member of the group, but has not necessarily internalized the norms of the group.
These Type I mechanisms serve to foster cohesion in a military organization. But do they deter desertion? Without internalization of group norms and adaptation, commitment levels will remain low. At the right time, the recruit will leave. Evidence from DDR programs show that such deserters adapt easily to life outside the rebel group. Beber & Blattman (2013) show that those who never accepted the values and norms of the LRA were the first to escape. Those who did accept them remained in the organization.
Type II socialization
Type II socialization involves changes in identity and preferences. These changes exhibit a certain degree of stickiness, meaning that people do not ‘shift loyalties from group to group depending on which group maximizes status markers’ (Johnston, 2005: 1031). Rather, stickiness is rooted in identification with the values of a group. Internalization of values and norms, in turn, alter beliefs and preferences, and thereby affects behavior.
Ideological indoctrination is a classic form of socialization and it is typically top-down. Indoctrination plays a big role in ideological groups. The Maoists in Nepal and the LTTE in Sri Lanka and the CNP-M in Nepal relied heavily on indoctrination programs, and relied on forced recruitment. The process of indoctrination begins before an individual joins the group. Children are targeted, especially through the school systems, to take part in indoctrination programs (Becker, 2010). Through indoctrination processes, the group instills a sense of fighting for a cause, and thereby creates value for functional preferences that might not have been there before. Religious indoctrination can work similarly. The will of God, as well as moral outrage at violence committed against family and neighbors by state forces, constitute important aspects of doing the right thing, playing to functional rewards or pleasures of agency (Gates, 2002; Wood, 2003). Indoctrination, however, is not cheap. It is time-consuming and labor intensive (Eck, 2014).
Religious or spiritual groups often tap into rituals as a means of transformative socialization. The LRA, for example, draws on the local Achioli culture where Christianity and animist spiritualism are mixed. Joseph Kony, the leader, is deemed to possess great spiritual power. Interviews with former LRA rebels indicate an very strong belief in Kony’s spirits (Bevan, 2007; Titeka, 2010; Mergelsberg, 2010; Vermeij, 2013). The all-seeing spirits are believed to be ever present and monitoring what you do. Belief in Kony’s control of spirits allows the LRA to outsource policing – spirits do it. Of course, the belief in the ever-watchful spirits is what reduces the costs of coercion. If one truly believes, desertion is futile as it will be inevitably detected. Punishment is certain. The internalization of these beliefs thereby alters behavior fundamentally.
The all-encompassing spiritual order of the LRA also serves as marker of ‘pure’ insiders and ‘impure’ outsiders (Titeca, 2010). Strict ‘holy rules’ dictate the way an LRA member interacts with the outside world (Mergelsberg, 2010). Belief, behavior, and identity work together to reinforce allegiance to the group.
Violence and the fear it generates can play a critical transformative role in Type II socialization in rebel groups that abduct their members. Cultivating a culture of violence is a very powerful means of socialization. Fear and violence begin with the act of abduction. Shortly afterwards, the new recruit is often forced to kill, typically someone from his or her own home village. Preparations for such killings and the spread of fear can be achieved through three specific steps. The first step is authorization – through approval, encouragement, or explicit orders by the leadership. The second step is routinization, which assures the continuation of violence. Once violence becomes routine, it becomes easier for it to be employed on a regular basis by members of the organization. The third step is internalization: at a psychological level, a recruit learns to adapt; he learns to take command of the violence, thereby harnessing fear. Instead of fear of punishment motivating behavior, the socialized soldier relishes the violence, controlling it. Fear is conquered. A culture of violence defines the organization and serves as the basis of identity with the group. Similar processes have been evident in the LRA (Bevan, 2007; Mergelsberg, 2010; Titeka, 2010; Haer, Banholzer & Ertl, 2011), RUF (Mitton, 2012, 2015), and in various groups in the DRC (Weierstall et al., 2013; Haer et al., 2013).
Transmission of the culture of violence is mostly maintained horizontally rather than vertically. At the time of initial abduction and early phases of induction into the group, individuals are controlled by their emotions, especially fear. As abducted fighters serve as the punishers – the ones meting out violence – some eventually are transformed into soldiers who control fear. Being tasked with engaging in brutal violence, within the group and outside the group, indelibly alters beliefs, norms, and fundamentally, behavior. The extreme emotional aspects of fear relate to the processes of ritualization of violence and the role of pain that Xygalatas et al. (2013) identify. ‘The particularly brutal forms of violence exacted by child soldiers were shaped by their own agency and were by nature often beyond the control of a fragmented and inconsistent RUF leadership. The impression of those witness or victim to such violence that it was in some sense “wicked” or “mindless” accurately describes the manner in which, far from betraying a rational consideration of conflict aims or a particular grievance, young combatants acted according to their brutal education in violence which even led them to enjoy the cruelty of their atrocities’ (Mitton, 2012: 119). 10 The process is not so much top-down as bottom-up.
Socialization Type II may be more effective for children than adults. The LRA, RUF, and Renamo, which relied extensively on the abduction and socialization of children, found it much easier to socialize children. It was much easier to make the children become good, integrated rebels (Beber & Blattman, 2013). As noted by Wessells (2006: 34–35): ‘through violence or threat of violence, young children can be trained to obey commands that many adults would contest or find ways around’. The key point here is that the children as opposed to adults internalized the norms of the organization and developed a deep allegiance to the organization (Blattman & Annan, 2010; Beber & Blattman, 2013). The proportion of children recruited is likely to constitute one of the principal scope conditions of these mechanisms. Belief in the spirits of the LRA is highly associated with age (Beber & Blattman, 2013). Norms relating to extreme violence against civilians are also likely to be limited to groups with a large proportion of children (Mitton, 2012, 2015).
An alternative mechanism
An alternative explanation that does not involve any of the mechanisms identified above would be ex-post screening (Obayashi, forthcoming). The abducted recruits may learn that they like being soldiers. Before abduction, they simply did not know the rewards available. Only by being forced to join the group was this information revealed. The process is not due to socialization of any type. Given the wretched lives most recruits had before joining a group, the appeal of soldiering should not be underestimated. Until experienced, the adrenaline rush of warfare and the comradery of the group cannot be known.
Nevertheless, the degree to which this mechanism alone solves the puzzle of forced recruits’ allegiance is limited. For this alternative mechanism to explain higher rates of retention in groups that rely on forced recruitment, it would have to occur at higher rates than for other groups. Yet, the portion of newly recruited volunteers who find that they like being soldiers more than they anticipated should be similar. The portion is probably similar to the number who re-enlist. A more likely explanation is that this alternative mechanism works hand in hand with socialization processes to deter desertion.
Illustrative cases
The Lord’s Resistance Army (LRA) has been a remarkably resilient organization, founded in 1987 and still operating as of July 2017. Since 1994 the LRA has depended on abduction as a means of recruitment (Doom & Vlassenroot, 1999). The group has almost no material resources; no one in the organization is paid; yet it persists, sustained by abducting new recruits and persuading a good share of them to remain with the group. Cleansing or rebirth are emphasized as fundamental to the recruit being inducted into the group. Violence, often committed in a new recruit’s home village, raises the cost of deserting and returning home. These violent acts work to create an insider vs. outsider identity. Violence permeates the culture of the organization.
The LRA is organized in small group operations of about 10 to 15 soldiers (the size of a squad in the US army). This structure allows for a high degree of mobility, ideal for guerrilla warfare. The small unit structure, moreover, reflects the organizational advantages King (2014) sees in modern militaries (though his basis for analysis is the platoon and not the squad), serving to create an effective bond between soldiers. Continuous training also helps form and develop these bonds. Nonetheless, the strongest mechanism to retain recruits to the LRA is a belief in spiritual monitoring and punishment, which effectively deters desertion. Socialization Type I raises costs of an outside option, while training and an organization structure based on small groups work together with the Type II socialization associated with the LRA belief system to create an intense in-group/out-group identity marker. The LRA does have deserters, make no mistake. The point is that they can retain a significant portion of those forced to join the group.
The Revolutionary United Front (RUF) was created in 1991 and ended as a military organization in 2002 after waging war in Sierra Leone for 11 years. Defeat came at the hands of intervening British troops in late 2001. Unlike most rebel groups that have relied on forced recruitment, the RUF depended on abduction from the very beginning (Eck, 2014).
The RUF also relied on a combination of mechanisms to retain abducted recruits. The RUF had no strong ideology with which to indoctrinate and no core religious beliefs. The buddy system, whereby two recruits were paired, lowered costs of coercion. Unlike the LRA, material rewards were also distributed. Individuals offered money or diamonds were much more likely to continue to participate in the RUF (Humphreys & Weinstein, 2008: 448–449). Several Type I socialization mechanisms were employed: forcing a new recruit to commit an atrocity in his home village raised the costs of the outside option, creating a sense of the in-group and out-group; training and small group organization enhanced the sense of group identity. Type II socialization occurred in the form of violence transformation, which developed bottom-up in the organization. ‘Through the mass abductions of young and impressionable children, the RUF leaders exposed youngsters to a desensitizing environment in which violence and atrocities were not only trivialized but were actively rewarded’ (Mitton, 2012: 119). These mechanisms focus very much on the role of violence and fear. Only through the psychologically transformative aspects of overcoming fear do these mechanisms work together. The RUF was also able to sustain its army of coerced recruits for about 11 years, but it did suffer problems of fragmentation during the war, and disintegrated as a fighting force after the British intervened.
The military wing of the Communist Party Nepal-Maoist (CPN-M), the People’s Liberation Army (PLA), was founded in 1996 and fought a bloody civil war until 2006. The CPN-M is a deeply ideological group that for the most part depended on volunteers for fighting a guerrilla war. By the end of 2001, the Maoists controlled significant territory and felt confident that it was time to directly engage the Royal Nepal Army (RNA) in combat. Casualties mounted. Manpower needs grew. Abductions became more frequent. More people were abducted in ‘2002 alone than in the entire 1996–2001 period’ (Eck, 2014: 387). Time devoted to ideological indoctrination and military training decreased significantly (Cowan, 2010; Eck, 2014). The result was substantially reduced levels of retention. The PLA to some extent got around this problem by drawing on local populations tasked with bearing equipment and picking up the dead and wounded and discarded arms from the battlefield. Their service ended with the conclusion of the battle. For such short periods, the Maoists could recruit and retain using coercion alone, but not for extended periods. This case underscores the limits to relying on coercion and the inability to retain the allegiance of troops who are forced to participate.
The fourth case is Liberians United for Reconciliation and Democracy (LURD), which fought a civil war against the government of Charles Taylor and involved another rebel group, the Movement for Democracy in Liberia (MODEL) from 1999–2003. LURD and MODEL relied only partly on forced recruitment, though LURD abducted more and preyed on displaced people (Pugel, 2010). LURD allocated material rewards (money) and non-material rewards (family protection) to attract their troops (Pugel, 2010). These rewards also worked to keep the recruits with the group. Coercion and punishment played a very limited role in retaining troops (Käihkö, 2015, 2016). Weak organization and a lack of shared norms ‘contributed to the limited use of executions’ and when executions were conducted intragroup, conflict erupted (Käihkö, 2015: 14). Discipline tended not to be top-down, but horizontal. Peers viewed ‘in-group cohesion as a way to increase their chances of survival in a violent environment’ (Käihkö, 2015: 24). As in the LRA and RUF, small group dynamics and bottom-up socialization and norm enforcement served to bring organizational cohesion to LURD. This case too underscores the limits of coercion alone in retaining troops.
These illustrative cases demonstrate that coercion alone did not induce long-term retention. While coercion was present in all groups, it was most successfully employed when administered horizontally and not vertically. Small group dynamics also played a critical role in socialization processes. Often two socialization mechanisms operated in conjunction with one another. Type I processes, whereby a new recruit committed an atrocity in his home village, made desertion costlier. Type II mechanisms made staying in the group a positive experience. Both worked by affecting the reservation wage of the recruit.
Conclusion
How do organizations that rely on forced recruitment retain the loyalty of their troops? Armed rebel groups employ a variety of retention mechanisms that serve to bond individuals to the group and maintain their allegiance. Coercion alone may effectively enhance recruitment, but it is not effective in maintaining group cohesion and fostering troop retention. Much of the existing civil war literature assumes retention to be a mere continuation of recruitment, but the mechanisms of retention cannot be explained by extending the logic of the initial decision to join. Recruitment and retention are separate processes. Thus, a person will join a violent group for a set of reasons, but will remain for others. In the case of rebel groups that rely on forced recruitment, continuing to rely on coercion to enforce retention is too costly and not sustainable. Four problems result from forced recruitment: adverse selection, incentive incompatibility, costly policing, and alienation of the population from which the group recruits. To address these problems the costs of coercion and the reservation wage of the recruit must be reduced. Reducing the costs of coercion only works in the short run. The long-term solution rests in the mechanisms of socialization that reshape the allegiance of forcibly recruited soldiers. Some of these mechanisms work better for children, especially adolescents, than for adults.
This research can be extended in several ways. The theory presented here could be cross-sectionally tested using the rebel group as the unit of analysis. New data would need to be collected on group recruitment and retention figures, as well as indicators of the various socialization mechanisms. This study could also be extended by examining general patterns of retention and recruitment across all organizations (see also Hoover Green, 2017). Another extension would be to use the central aspects of the analysis employed here to analyze military effectiveness. One could also compare nation-state militaries in terms of conscription and voluntary armies and their relative effectiveness.
Socialization mechanisms, especially those involving an internalization of the norms of the organization, are especially effective at promoting group allegiance. Top-down mechanisms, such as ideological indoctrination, are costly. Norms that emerge from the bottom-up or are at least maintained among peers are effective, but they risk factionalizing the organization. Discipline and cohesion work best at maximizing retention. This is true for rebel groups that depend on forced recruits, those that rely on volunteers, and the armed forces of nation-states.
Footnotes
Supplementary material
Supplemantary material for this article is available at https://www.prio.org/JPR/Datasets/ and 
.
Acknowledgments
I thank Chris Butler, Jeff Checkel, Dara Cohen, Marianne Dahl, Lee Ann Fujii, Amelia Hoover Green, Francisco Gutiérrez, Helge Holtermann, Bertrand Lescher-Nuland, Jason Lyall, Kazuhiro Obayashi, Dennis Rodgers, Håvard Strand, Scott Strauss, Henrik Urdal, Libby Wood, and especially the anonymous reviewers at JPR for helping me improve this article; and special thanks to Jeff Checkel for organizing this special issue of JPR.
Funding
This research was supported by grant QZA 13/0365 provided by the Norwegian Ministry of Foreign Affairs.