Who Your Friends are Matters: Types of Supporters and Child Recruitment by Rebel Groups

2.1 Factors of Child Recruitment: General, Supply and Demand Factors

The factors associated with the child recruitment may be examined from three distinct perspectives: a general and systematic perspective, a supply perspective and a demand perspective.

Firstly, researchers have indicated that the proliferation of small arms increases the probability of rebels utilising child soldiers, as these weapons are light enough for children to manage (Bloom and Horgan 2019; Haer 2019; Singer 2006; Stohl 2002). The substantial population of children has also had an impact on child soldiering (Daillaire 2011; Haer 2019). For instance, existing studies have demonstrated that the considerable number of young people in Africa increases the probability of child recruitment (Haer 2019; Peters et al. 2003). Indeed, a number of African countries, including Angola, have been found to have a higher proportion of young people than adults (Achilihu 2010; D’Costa 2015).

Secondly, there is a possibility that some children decide to join armed groups of their own volition, a factor which is generally explained by push and pull factors (Gutiérrez-Sanón 2010; Haer 2019). ‘Push factors’ are the aspects which motivate (push) children to join the armed forces. The dearth of educational opportunities is one possible factor that can act as a catalyst for children to join armed conflicts (Goodwin-Gill and Cohn 1994; Haer 2019; Machel 1996; Wessells 2016). In the Democratic Republic of Congo, where educational opportunities are limited, children often enrol in military schools, as the paucity of alternative employment opportunities may lead children to perceive joining armed conflicts as a viable career option (Brett and MacCallin 1996; Haer 2019). Conversely, there are ‘pull factors’ which induce (pull) children to make the decision to join armed groups due to positive rewards such as the pursuit of basic necessities, including shelter, sustenance, and attire. (Brett and MacCallin 1996). Furthermore, it has been identified that norms are a significant contributing factor in children’s involvement in armed conflicts (Lasely and Thyne 2015; Machel 1996). The desire for prestige and power has also been identified as another motivation for children to engage in armed conflicts (Chen 2014; Francis 2007; Wessells 2005).

Finally, it is also common practice for finding reason of child recruitment from demand side perspective. For example, the Lord’s Resistance Army (LRA), the National Patriotic Front of Liberia (NPFL), and the Revolutionary Armed Forces of Colombia (FARC) have reportedly threatened, coerced, and abducted children to join their respective groups (Human Rights Watch 2008). The prevailing rationale for coercive recruitment is that children are more compliant and more susceptible to indoctrination than adults when exposed to drugs and alcohol, thereby increasing their risk-taking behaviours (Annan et al. 2006). In Sierra Leone, for instance, drugs were frequently used as a means of influencing children and inciting their involvement in violent activities (Human Rights Watch 2008). Furthermore, children are frequently enlisted to offset deficiencies in the adult military population, particularly when groups encounter challenges in recruiting enough adults (Brett and MacCallin 1996; Haer 2019; Singer 2006; Tynes and Early 2015; Woods 1993). This occurs because children are a cost-effective solution for armed groups seeking to enlist combatants, as they are unlikely to demand compensation and are relatively obedient (Beber and Blattman 2013; Vautravers 2009).

2.2 Impact of External Supporters on Rebel Groups

In the meantime, research on external supporters and their impact and relationship with rebels is being extensively examined. Firstly, the aims of supporting rebels are discussed. As scholars point out, external backers may support rebels to replace an existing regime with their preferred one by pressuring the beneficiary (Carothers 2006; Coyne 2008) or to establish a new government that is amenable to their interests (Salehyan 2010; Vanneman and James 1976). Additionally, some external supporters provide technical and financial assistance to rebel groups with the anticipation that these groups could evolve into political parties, which they may establish connections even after the conclusion of hostilities (Carothers 2006; Vanderhill 2013). Other studies indicate that, depending on resource allocation by external supporters, conflict diffusion (Karlén 2017), and the cohesion or fragmentation of rebel groups (Tamm 2016) can be affected.

Secondly, scholars also identified the divergent motivations and outcomes made by different external supporters. As the act of attacking another nation is inherently risky and costly, state supporters may support rebels to attack the government of rebel groups instead of using their military forces to save on expenditures (Byman 2006; Corbin 2011). Consequently, providing support to rebels may allow the principals (state supporters) to enhance efficiency and productivity through the deployment of a more cost-effective power (Corbin 2011). In a similar vein, state supporters are more likely to provide support to moderately strong rebels, who possess sufficient power to inflict costs on the targeted government, as these rebels are more likely to be structured or cohesive (Salehyan et al. 2011). Sharing a common identity and ideology also affects state supporters to decide backing rebels (Thaler 2022). Alongside the active research on state supporters, other studies focus on different supporters. For instance, the existing research suggests that rebel groups who are supported by non-ethnic supporters are likely to become fragmented due to the low trust between them, whereas ethnic supporters are likely to build a trustworthy relationship that enhances the cohesion within groups (Ives 2021). Moreover, recent studies suggest that diaspora support rebel organisations for the purpose of kinship and change in the situation of the homeland (Daub 2025). Similarly, Petrova (2019) suggests that rebels receiving support from the diaspora are more likely to adopt non-violent tactics by drawing parallels between state supporters and the diaspora. Furthermore, another study indicates that rebel groups supported by state actors are more likely to increase attacks on civilians because external support reduces the incentive for rebel groups to seek civilian support (Salehyan et al. 2014).

The motivations and outcomes of support from the third parties are also suggested to be multifaceted. In accordance with the findings of Petrova (2019), this research hypothesises that the outcomes of child recruitment are contingent upon who rebels receive external support from. By examining two different actors, I argue that rebel groups are more motivated to recruit child soldiers when they have rebel supporters than in the absence of such supporters.

3.1 Rebel Supporters and Child Recruitment by Rebel Groups

Rebel groups are likely to cooperate with each other during armed conflicts as, 50 % of rebel groups’ activities between 1946 and 2008 established alliances with one another (Akcinaroglu 2012; Gade et al. 2019). Forming a collaborative relationship brings benefits to rebel supporters, as this relationship allows them to build a more organised and resilient group that enables them to defeat their targeted opponents (Topal 2024). Given this background, this section suggests that receiving support from other rebel groups is likely to encourage rebels to recruit child soldiers with ease from the following potential scenarios.

Firstly, in case when a supporter already uses child soldiering as their strategy could be considered. It is crucial for armed groups to acquire novel tactics from disparate groups, as this may influence their development and success (Asal et al. 2022). The adoption of novel tactics by armed groups can be a costly decision; however, if a group acquires a tactic from a supporter, it can assess the benefits of utilising that tactic against the costs incurred. Indeed, the collaboration between these groups has been shown to facilitate the diffusion of strategies at reduced costs (Horowitz 2010). As discussed in existing literature, collaboration between groups has the potential to enhance their collective capabilities and facilitate the dissemination of specific tactics, such as the targeting of civilians, across the groups concerned (Asal et al. 2022). Such interactions are likely to occur in various settings, including training centres, during meetings, and umbrella organisations (Tynes 2018). In a similar vein, the utilisation of child soldiers can also be facilitated and transmitted when tactical intelligence is exchanged through group cooperation due to the nature of networks between armed groups^[3] (Heyman and Mickolus 1981; Tynes 2018). It can thus be surmised that this cooperative network, which serves as a forum for the discussion and refinement of tactics, may also facilitate the diffusion of the tactic of child soldiering with lower costs and risks (Tynes 2018).

This is supported by the case of the National Patriotic Front of Liberia (NPFL) and the Revolutionary United Front (RUF) in Sierra Leone. When Charles Taylor’s NPFL exercised control over 90 % of Liberia’s territory, the conflict in Liberia spilled over into Sierra Leone (Gershoni 1997). The RUF, which was the primary belligerent in Sierra Leone, received support from the NPFL. The initial objectives of the RUF were to establish a new Sierra Leone, proclaiming ‘liberation’ and ‘democracy’, however, the brutality of the conflict resulted in a decrease in civilian support for the group (Denov 2010). While the RUF received support from the NPFL, the RUF also adopted the NPFL’s ideology and strategies, including the use of violence against civilians and other forms of human rights abuses (Day 2015; Nuxoll 2014). Consequently, the RUF’s original ideological framework became effectively obsolete due to the influence of the NPFL (Restoy 2006). The RUF’s utilisation of a significant number of child soldiers was influenced by the NPFL, as the initial composition of the RUF’s military forces was largely comprised of Liberian troops (Restoy 2006; Utas and Jörgel 2008). Indeed, it is estimated that between 10,000 and 30,000 children were involved in the RUF movement, playing a pivotal role during the conflict (Abraham 2001; Denov 2010; Murphy 2003).

Second, the cooperation between rebels can increase their aggressiveness by enhancing the durability of rebels, as connections can lead to an escalation in violence and the adoption of new tactics (e.g. Asal et al. 2012; Horowitz and Potter 2014; Phillips 2014). This is because exchanging resources allows groups to acquire requirements such as personnel, weapons, or information and pooling resources (Phillips 2014). Consequently, these rebel groups are less inclined to seek support from civilians, thereby increasing the probability of their engagement in violent actions against civilians (Salehyan et al. 2014; Sawyer et al. 2017; Zhukov 2017). In essence, rebels who are accustomed to perpetrating violence against civilians are more likely to engage in forced recruitment (Cohen 2013; Eck 2014).^[4] Furthermore, existing study indicates that rebels with abundant resources are more prone to engage in the forced child recruitment as having more resources may decrease dependence on civilians (Haer et al. 2020). Consequently, this suggests that rebels may find it more straightforward to recruit child soldiers when having rebel supporters.

Thirdly, despite having rebel supporters allowing rebels to maintain their forces, given that rebel groups are less economically wealthy or organised than their opposing governments, or other potential types of supporters that rebels who receive support can consider taking cost-effective strategies to effectively maintain their power. The use of children could therefore be an attractive option for rebel groups, as children are less expensive labour, and they are more obedient and malleable than adults (e.g. Beber and Blattman 2013). Moreover, the recruitment of child soldiers is likely to prolong armed conflicts, thereby providing rebels with an opportunity to defeat state forces (Haer and Böhmelt 2017). Furthermore, child soldiers can contribute to the sustainability of rebel groups by instantly increasing the number of combatants, given the substantial size of the child soldier population (Haer and Böhmelt 2016).

Lastly, having rebel supporters can facilitate rebels’ use of child soldiers, given the weak supervisiory system. Despite arguments that some rebel groups may consider their reputation or adhere to international regulations (e.g. Jo 2015; Jo and Thomson 2013), it is difficult to expect rebels to respect human rights protection without strong monitoring systems or immediate punishment, as they are not directly bound by these regulations. Indeed, different from making an alliance with the state government, it is difficult to codify militant partnerships because of their clandestineness (Mattes 2012; Byman and Kreps 2010) and also, they are operating in anarchic environments which are less likely to have enforcement systems to each other (Vinci 2008). Consequently, providing support to rebel groups that utilise child recruitment as a tactic may not hinder rebel supporters from offering their backing. Though rebel supporters may be reluctant to continue supporting rebel groups that recruit child soldiers, monitoring these groups is challenging for supporters due to a lack of capability and supervisory systems.

In conclusion, having rebel supporters may allow rebels to recruit child soldiers more easily, regardless of whether they acquire knowledge from their supporters, due to the characteristics that rebel supporters possess and the imperfect partnership mechanism between rebels. Taken together, these points lead to the following argument:

3.2 State Supporters and Child Recruitment by Rebel Groups

Compared to rebel supporters, rebel groups that receive support from state sponsors are more likely to be constrained in their decision-making processes. As state supporters, which are foreign governments, find it difficult to intervene directly in the conflict as this could be a costly option, they instead invest material support in rebel groups to achieve their goals (Salehyan et al. 2014). Therefore, state supporters may prefer to support rebel groups who do not deviate from their values. In comparison to other categories of supporters, state supporters may possess a greater capacity to influence rebel groups to adhere to their directives when they establishing relationships with rebels, which in other words means that rebels’ decision-makings can be constrained by states supporters. State supporters prefer to provide support to rebel groups that are relatively weak, as such groups are less likely to take actions which are contrary to the interests of supporters (referred to as ‘agency slack’) (Hawkins et al. 2006; Lidow 2016; Salehyan et al. 2011; Sinno 2008; Tamm 2016).

Secondly, the research suggests that state supporters may exercise a greater influence over recipients compared to other supporters. Supporting rebel groups can be a risky investment for a state government (Marshall 2019; Salehyan et al. 2014). Even in cases where a supporting actor has entered into a contractual agreement with their recipient, information asymmetry between the supporter and recipient can still occur. This phenomenon is termed ‘adverse selection’ (Marshall 2019; Salehyan 2010). Should a rebel group engage in actions that contravene the principles of their supporter, this could result in a significant deterioration in the reputation of the supporter, both nationally and internationally (Marshall 2019; Salehyan 2010; Salehyan et al. 2014; Vanneman and James 1976). In particular, the use of violence against civilians results in a loss of reputation for state supporters among their domestic audiences (Berkowitz 2018). Consequently, state supporters may be disinclined to provide support to rebels who may not violate their principles and deteriorate their reputation. Therefore, even if insurgents gain support from state supporters and obtain a certain capability, they are less likely to recruit children, which may deteriorate the image of the sponsor.

Taking the aforementioned points into consideration, external state supporters are likely to impose constraints on the activities of their agents in order to mitigate their reputational risks. In particular, state supporters who are committed to the protection of human rights or the establishment of democratic governance are likely to oppose any violations of these principles. In such cases, state supporters are likely to eschew providing support to rebel groups that are considered to be high-risk and costly (Marshall 2019; Putnam 1998) and may opt to constrain or withdraw their support from these groups. Faulkner (2019) argues that external democratic state support acts as a deterrent to rebel groups recruiting children as soldiers. However, not only democratic societies, but also some authoritarian governments, are committed to the protection of human rights. Indeed, a significant number of authoritarian regimes states have ratified international humanitarian regulations on child recruitment. Despite the fact that their intentions to respect and ratify human rights treaties may differ from those of democratic states (Hollyer and Rosendorff 2011), violations of the regulations may also have a negative impact on their reputations, which may be exacerbated if they provide support to rebels violate human rights.^[5] This is potentially because if state supporters continue to provide support to such a group that engages in human rights abuses, it is probable that their reputation will deteriorate further, as it may suggest that these state supporters are encouraging rebel groups to engage in human rights abuses by providing resources to do so. Therefore, regardless of regime type,^[6] rebel groups with state supporters may be less motivated to actively recruit children than with other supporters in order to maintain the relationship with their supporters.

In conclusion, I hypothesise that state supporters may hinder rebels in making their own decisions, due to the strength of their restriction on rebel movements, which differs from the relative flexibility of a loose partnership between rebel supporters. Based on the aforementioned points, this study hypothesises as follows:

4.1 Dependent Variable: Use of Child Soldiers

A binary variable is employed to measure the usage of child soldiers by rebel groups. If a rebel group recruited child soldiers in a given year, the variable is coded as 1; otherwise, it is coded as 0. Table 1 presents the distribution of child recruitment by rebel groups, which have been clustered into rebel dyads. In total, 302 rebel groups were observed. Approximately 77 % of the rebel groups were found to have engaged in child soldiering. This research does not distinguish between forced and voluntary recruitment of children, given that children may be involved in armed conflicts by coercion, but also by their own decision.

4.2 Independent Variables: Types of Supporters

The UCDP External Support Dataset (Meier et al. 2023) is employed for the purpose of classifying the types of supporters. The dataset provides information regarding the recipient side, thereby enabling the researcher to ascertain the identity of the supporter in a given year. This study examines the influence of two distinct external actors – rebel supporters and state supporters – on the engagement of rebel groups in child soldiering. As previously stated, the term “rebel supporters” describes different rebel groups, whereas the term “state supporters” refers to foreign state governments.

In order to test the aforementioned hypotheses, a dichotomous variable is employed. If rebel groups received support from other rebel groups, the value is taken to be 1; otherwise, it is 0.^[10] The variable indicating the presence of state supporters is coded in a manner analogous to that employed for rebel supporters. As illustrated in Table 2,^[11] approximately 17 % of rebel groups received support from other rebel groups. Conversely, external state supporters provided assistance to approximately 35 % of rebel groups during the same period.

4.3 Control Variables

Several control variables which may affect rebel groups’ use of child soldiers are used in this study. Models include conflict duration because rebel groups may engage in child soldiering when a conflict lasts longer in order to replenish the number of soldiers and prevent a military shortage (Haer et al. 2020). This variable originates from the Non-State Actor (NSA) dataset (Cunningham et al. 2009, 2013).

Models also include battle-related death as a control variable, reflecting the intensity of a conflict. If a conflict causes significant fatalities, rebel groups are more likely to engage in active child soldiering (Haer and Böhmelt 2017; Singer 2006), as they do when a conflict lasts longer. This data derives from the UCDP/PRIO Battle-Related Deaths Dataset (Gleditsch et al. 2002).

The models furthermore control GDP per capita, sincepoverty can result in children becoming involved in armed conflict that the study statistically supports the increase of child recruitment under conditions of poverty (Achvarina et al. 2007). Concurrently, rebels in developing countries prefer to use children who are less likely to ask for incentives (Vautravers 2009) that those groups may save resources. Therefore, in order to control for the poverty of a state, the variable of GDP per capita is controlled, which is calculated using the natural logarithm (Gleditsch 2002).

The study controls for the youth population by taking the natural logarithm. Larger populations of children can be one of the possible reasons for increased child recruitment (Dallaire 2011; Haer 2019). It refers that if there are more children in the population, more children can be potential soldiers. The data is sourced from the World Bank.

Rebel groups having their own political wing may deter them from actively using child soldiers because they may seek to be perceived as legitimate actors both domestically and internationally by adhering to humanitarian regulations (Jo and Thomason 2013). To assess this variable, the research utilises the NSA dataset (Cunningham et al. 2009, 2013). If rebel groups have their own political wings, it takes 1, and otherwise 0.

I control rebel groups’ fighting capacity compared to the government’s force. As previously discussed, if rebels are sufficiently powerful, they are less inclined to rely on children as rebels may recruit children for the purpose of increasing their own strength (Haer and Böhmelt 2016). The NSA dataset (Cunningham et al. 2009, 2013) is utilised for measurement. I coded 0 when rebel groups’ fighting capacity is lower relative to government forces. It takes 1 if the fighting capacity is moderate, and 2 when the rebel groups’ fighting capacity outstands the government forces.

It is expected that rebel groups can more easily recruit people when they control a certain place compared to when they do not (Haer et al. 2020; Kubota 2013), and this assumption can be applicable to children as well. Territorial control takes an ordinal value: if rebel groups do not have territorial control, it takes 0 whereas takes 3 if rebel groups have strong control of the territory. I use the NSA dataset (Cunningham et al. 2009, 2013).

If governments recruit children coercively, rebel groups may also actively recruit children (Haer et al. 2020). This is because, if a state disrespects its citizens, the citizens (including children) may have an incentive to cooperate with rebels to improve their condition of human rights situation in their country (Høiskar 2001). Conversely, another possibility is that a low level of human rights protection in a certain country may facilitate a rebel that is active in that country and which may also be disrespecting the human rights of civilians, including children. Haer et al. (2020) code forced recruitment by the government as a dichotomous variable: 1 if there was forced recruitment and 0 if not.

Table 2 also includes the variance inflation factor, which detects multicollinearity between variables, for key independent variables and control variables. Across variables, no variables exceed a value of 5, thereby indicating that the variables examined in this study are not affected by multicollinearity.