Abstract
Programs to disarm and demobilize former combatants are standard features of peace agreements, but demobilized rebels face many challenges. I document a threat to demobilized rebels that emerges when rebel groups fragment, namely, selective violence by splinter groups. Why do splinter groups target demobilized rebels? While previous research on risks for this population has emphasized stigmatization or criminality, I argue that rebel splinter groups selectively target demobilized members of their parent group owing to their skills, experience, and potential for intelligence sharing. I analyze the pattern of violence against demobilized members of the FARC, a major rebel group in Colombia. Areas where FARC splinter groups emerged saw increased rates of violence against demobilized FARC rebels; this violence disproportionately affected demobilized rebels compared with other civilians, and activity by other armed groups did not have similar effects. These findings highlight an understudied channel through which rebel fragmentation threatens peace agreements.
Introduction
On 7 September 2022, a former member of the Revolutionary Armed Forces of Colombia (FARC), historically Colombia's largest rebel group, was shot and killed in his home in the country's northeastern Catatumbo region (Ramírez, 2022). The former rebel had laid down arms when the FARC signed a peace agreement with the government in 2016, and in the ensuing years had worked with a cooperative set up by former combatants called the Association for Reconciliation and Peace in Catatumbo (REPAZCAT). Several months after the shooting, Colombian authorities captured the alleged perpetrators, and found that they belonged to Frente 33, a splinter group of the FARC that had remobilized after the 2016 peace accord. According to officials, the victim was killed for resisting demands that he join the splinter group (Fiscalía, 2023).
This was not an isolated case. By early 2025, more than 450 out of the roughly 14,000 members of the FARC who demobilized under the 2016 peace agreement had been killed (RCN, 2025), with many others injured or forcibly displaced (UNVIC, 2023). And, while the majority of these incidents remain unsolved, a substantial number of solved cases involved deliberate attacks by splinter groups of the FARC (Nuevo Siglo, 2020).
Does rebel group fragmentation pose a threat to demobilized rebels? While a growing literature studies threats to demobilized rebel combatants with focus on criminal involvement or stigmatization (Blair et al., 2021; Blattman and Annan, 2016; Peña and Dorussen, 2021), there is relatively little evidence regarding selective violence against this population. Yet killings of demobilized rebels by rebel splinter groups have been reported not only after the FARC's peace agreement in 2016, but also after a 1991 peace agreement with another Colombian guerrilla group called the Popular Liberation Army (EPL) (Valencia Agudelo, 2021), as well as a 2002 peace agreement between the Democratic Republic of the Congo (DRC) and groups like the Rally for Congolese Democracy (RCD-Goma) (Richards, 2016).
This paper argues that when rebel groups fragment during a peace process, splinter groups that remain at war are threatened by the demobilization of members of the parent group. Their demobilization deprives splinter groups of potential recruits with valuable experience and brings sensitive intelligence about these groups within reach of authorities. Splinter groups respond to these threats with selective violence against demobilized rebels.
The empirical section links the emergence of FARC splinter groups after Colombia's 2016 peace agreement to the surge in violence against demobilized FARC rebels. I demonstrate, first, that areas where FARC splinter groups emerged saw a significantly higher rate of homicides against demobilized FARC rebels compared with other areas; second, demobilized FARC rebels were singled out as targets by splinter groups, and saw a significantly larger increase in fatality rates than the general population; and third, this violence was a unique feature of FARC splinter groups, and areas with other armed groups did not see a similar pattern. I also provide qualitative evidence illustrating the mechanisms driving this phenomenon.
Violence against demobilized rebels represents a serious challenge to peace, inhibiting rebel reintegration, and potentially deterring armed groups from demobilizing. The findings of this paper advance the literature on Disarmament, Demobilization, and Reintegration (DDR) programs by identifying a threat to these programs emanating from rebel fragmentation that has been largely overlooked, while also contributing to the set of explanations for why some DDR programs succeed at their aims while others fail (Banholzer, 2014). In practical terms, these findings suggest that rebel groups at risk of fragmentation may require a model of DDR that prioritizes participants’ physical security. More broadly, these findings are relevant to the literature on the “downstream effects” of rebel group fragmentation (Rudloff and Findley, 2016), establishing an additional mechanism by which rebel splinter groups threaten post-conflict stability.
Prior research on threats to demobilized rebels
Research on security threats to demobilized rebels focuses primarily on two key challenges: their potential involvement in crime; and stigmatization by the civilian population.
Research across a number of contexts has found high rates of criminal recidivism among demobilized rebels (Blattman and Annan, 2016; Daly et al., 2020). Explanations focus on their lack of opportunities in the formal economy, skills and experience in combat and in the illicit economy, ties to criminal networks, and propensity to settle in high-crime areas (Blattman and Annan, 2016; Daly et al., 2020; Peña and Dorussen, 2021; Vásquez-Cortés, 2025). If demobilized rebels experience high levels of violence, this literature might point to their own involvement in risky or criminal activities.
The literature also highlights challenges associated with societal stigma against demobilized rebels. Civilians may perceive former rebels as dangerous or untrustworthy (Blair et al., 2021). Moreover, the civilian populations victimized by rebel violence may desire harsh retribution against former rebels (Godefroidt and Langer, 2023; Salih et al., 2018), or even take part in revenge killings (Nussio and Howe, 2016).
In sum, the dominant explanations for violence against demobilized rebels focus on either criminal activity or stigmatization. These arguments neither predict nor explain selective violence against demobilized rebels perpetrated by rebel splinter groups, yet accounts of such violence have been documented in a number of cases. In 1991, after Colombia negotiated an agreement with the EPL, almost 15% of the 2200 EPL members who demobilized were killed, many of them allegedly by EPL splinter groups (Valencia Agudelo, 2021). After a 2002 peace agreement between the DRC and groups including RCD-Goma, members of RCD-Goma who participated in a DDR program reported being tracked down by a splinter group of RCD Goma called the National Congress for the Defense of the People and threatened with violence and forced recruitment (Richards, 2016).
In explaining these patterns, my theory builds on and extends insights from the research on demobilized combatants and criminality. In this literature, network connections to recidivist peers combined with relevant skills such as expertise with weapons create opportunities for demobilized rebels to participate in risky activities like crime. I argue that these skills and connections also increase risks to demobilized rebels who are not recidivists by making them targets of forced recruitment. I also build on Richards’ (2016) finding that demobilized rebels in the DRC were targets of violence and forced recruitment, which I argue is consistent with the idea that splinter groups employ violence as a coercive recruitment and retention strategy.
I also contribute to a broader literature on how armed group capacity and access to resources influences patterns of violence (Dube and Vargas, 2013; Weinstein, 2006). Recent work within this literature has focused on whether stronger and well-resourced rebel groups are more or less likely to provide services to civilians (Albert, 2023), engage in terrorism (Crisman-Cox and Park, 2024), or escalate violence (Eastin et al., 2024). My results add to an emergent body of evidence suggesting that rebel fragmentation represents a negative shock to rebel resources and capacity (Perkoski, 2019), which can lead to worse behavior toward civilians (Faulkner and Doctor, 2021).
A strategic rationale for splinter group violence against demobilized rebels
Research suggests that peace negotiations often catalyze rebel fragmentation, as some members of a rebel group who reject peace form splinter groups even as other members of the parent group disarm and demobilize (Duursma and Fliervoet, 2021). Nascent splinter groups enter a difficult strategic environment (Perkoski, 2019), weakened by the demobilization of their parent group, and often facing competition with the state, other armed groups, and in some cases rival splinters (Cunningham et al., 2012).
In this initial stage, rebel combatants with years of experience in the parent group represent a scarce and valuable resource, but the same experience that makes them valuable inside of the splinter group renders parent group members a serious threat outside of the splinter group. Rebels with experience in the parent group are likely to have developed relevant expertise such as skill with weapons and knowledge of tactics (Bloom, 2017)—expertise that is particularly valuable to groups at critical operational junctures such as rebuilding after losses (Perkoski and Worsnop, 2025). Such experience is valuable and scarce for nascent splinter groups, as groups seeking to expand rapidly often rely on inexperienced recruits or even minors to fill their ranks (Bloom, 2017; Faulkner and Doctor, 2021).
Outside of the splinter group, rebels with years of experience in the parent organization pose a serious intelligence threat. Many civil wars involve irregular warfare, a mode of conflict in which information is valuable and scarce (Kalyvas, 2006). Members of a rebel parent group might possess information about the group's sources of arms and financing, its structure, the location of training camps or weapons’ caches, and the identities of covert operatives or collaborators. The release of this sensitive information about the parent organization could threaten splinter groups, which may rely on many of the same resources, personnel, and operating procedures, particularly early in their lifecycle. Moreover, even members of the parent group who do not intend to share information with the state or rival groups cannot credibly commit to this course of action, particularly if the state or other groups employ coercion to extract the information.
Beyond strategic intelligence, rebels from the parent organization might also possess information implicating splinter group members in war crimes. Many peace processes involve truth and reconciliation provisions that require testimony from demobilized combatants as a condition for amnesty. Testimony implicating splinter group members or leaders could increase their risk of prosecution and severe punishment if they are captured.
I argue that violence against demobilized members of the rebel parent group is motivated by the reality that members of the rebel parent organization are valuable inside of rebel splinter groups but dangerous outside of them. Selective violence against demobilized rebels who possess sensitive information serves the immediate goal of preventing this intelligence from reaching authorities, while also deterring other members of the parent group from sharing such information. Selective violence also serves as a coercive recruitment and retention strategy. It forces members of the parent group to choose between remobilizing as part of the splinter group or refusing recruitment and becoming potential targets. It also might at least temporarily deter members of the splinter group whose morale or commitment is wavering from attempting to demobilize. Over time, relying on coercion to recruit or retain members could prove unsustainable, 1 but nascent splinter groups’ time horizons are short, leading them to pursue strategies that are suboptimal in the long term, but serve the immediate imperative of survival (Faulkner and Doctor, 2021).
Study setting
The Colombian government reached a peace agreement with the FARC in 2016. Within a year of ratification, an estimated 95% of the group's members had disarmed and demobilized. The agreement stipulated that these demobilized members would receive a stipend worth 90% of the minimum wage, amnesty or reduced sentences for crimes conditional on providing testimony to a truth and reconciliation commission, and political representation.
Even as most FARC members were assembling at cantonment sites and turning in their weapons, several mid-ranking FARC commanders announced their opposition to the peace agreement and formed splinter groups commonly referred to as “FARC dissident” groups. Composed initially of just a few hundred members, dozens of these FARC splinter groups emerged in the ensuing years, their expansion facilitated by access to illicit profits from the drug trade (Wyer, 2024).
Meanwhile, of the roughly 14,000 former FARC combatants who eventually demobilized, more than 450 had been killed in violent attacks by early 2025. Hundreds of others were threatened, injured, or forcibly displaced.
Violence against demobilized FARC rebels was a subject of political controversy in Colombia. Members of the FARC's political party suggested that paramilitaries were responsible for the violence (EFE Newswire, 2019), and in some cases blamed the government for failing to provide adequate protection. Meanwhile, statements by the FARC's political opponents insinuated that the violence was related to criminal activity by the victims (Semana, 2020). Early scholarly explanations blamed societal stigma and persecution of former combatants (Valencia Agudelo, 2021), or suggested that they were caught in the crossfire in disputes for territorial control (Charles et al., 2020).
Official statistics meanwhile, revealed a surprising pattern. Although the government was unable to identify the perpetrators for the majority of attacks, the cases that were solved revealed that the most common perpetrators were FARC splinter groups. Some 44% of solved attacks were perpetrated by FARC splinter groups compared with just 11% for the ELN guerrilla group, the next most common perpetrator (Nuevo Siglo, 2020). While these statistics represent only the subset of cases that have been solved, they provide a degree of plausibility to the argument that rebel splinter groups selectively target demobilized rebels.
Observable implications
My theory predicts that when rebel groups fragment during peace processes, splinter factions have a strong incentive to perpetrate violence against demobilized members of the parent group. An observable implication in this case is that the emergence and expansion of FARC splinter groups post-peace agreement should increase violence against demobilized FARC rebels. I hypothesize that: (H1) areas where FARC splinter groups emerge should see higher levels of violence against demobilized FARC rebels compared with other areas.
My theory suggests that demobilized rebels are not incidental victims caught in the crossfire of conflict, but are singled out as targets by splinter groups. I hypothesize that the rate of violent attacks should be higher for demobilized FARC rebels than for the broader population (H2A), and this increase should be driven by FARC splinter groups, not by other armed groups (H2B).
Data
The outcome of interest is homicides against demobilized FARC rebels. I focus on the municipality-level rate of homicides against demobilized FARC members for 2017–2018, the first two years after the peace agreement. This period represents the initial phase of FARC splinter group emergence and expansion. This data comes from the Special Jurisdiction for Peace (JEP), a transitional justice body established by the peace agreement that monitors its implementation, including the status of demobilized rebels. The JEP data include cases where authorities have not identified the perpetrator. They do not include former FARC members who remobilized into armed groups. I calculate the municipality-level rate of homicides, using data on demobilized FARC rebels released by Colombia's National Agency for Reincorporation in 2025. To ensure that this measure is not influenced by post-treatment violence, I attribute demobilized FARC rebels who moved between municipalities owing to violence to their origin municipality rather than their destination. 2 The main specifications exclude municipalities with fewer than 10 demobilized FARC rebels; however the results are robust to a range of inclusion criteria (Appendix A.5).
The independent variable is a binary variable for whether FARC splinter groups were present in a municipality during the period of interest. This measure relies on data from Indepaz, a Colombian think tank, which assesses armed group presence by triangulating from news reports, reports from official government agencies and independent national NGOs, communication with local and regional organizations, and field work (Indepaz, 2013; Posso et al., 2021).
Results
Model 1 in Table 1 compares the rate of homicides against demobilized FARC rebels in municipalities with and without FARC splinter group presence. This unadjusted difference-in-means provides support for H1, indicating that a municipality with a FARC splinter group experienced roughly nine additional homicides per 1000 demobilized FARC combatants compared with a municipality without splinter group presence.
Effects of FARC splinter groups on homicide rates.
Notes: The dependent variable in models 1–3 is the average municipality-year rate of homicides per 1000 against demobilized FARC members for 2017–2018. In model 4 the dependent variable is the average municipality-year rate of homicides per 1000 for the general population for 2017–2018. The independent variable is a binary variable for whether FARC splinter groups were present in a municipality between 2017–2018. In models 1, 2, and 4 the sample includes all municipalities with a population of at least 10 demobilized FARC members. Model 3 further restricts the sample to municipalities with FARC splinter group presence between 2017–2018 (treated) or 2019–2022 (not-yet-treated). In models 2–4 I generate weights on the covariates of interest with kernel balancing, and estimate the model using weighted least squares. Weights are trimmed to the 99th percentile. All models use HC2 standard errors. ∗∗∗p < 0.001; ∗∗p < 0.01; ∗p < 0.05.
One challenge for inference is that the same conditions that allowed FARC splinter groups to emerge in certain municipalities might be correlated with the level of violence.
Prior research suggests that FARC splinter groups emerged in territory valuable for the cocaine trade (Wyer, 2024) and that there might be a link between the illicit economy and risky activity by demobilized rebels (Vásquez-Cortés, 2025). I employ multiple strategies to address such concerns. Model 2 adjusts for a number of observed pre-treatment variables that might influence the rate of violence: the municipality's recent economic performance, population size, terrain ruggedness, distance from military bases, historical level of coca cultivation, average homicide rates over the previous 5 years, and population of demobilized FARC members. I use kernel balancing to generate weights that render the control group— i.e. municipalities without FARC splinter group presence—similar to the treatment group across these relevant covariates (Hazlett, 2020). 3 The identifying assumption for this strategy requires that there is not some other unobserved variable that influences both the emergence of FARC splinter groups and violence against demobilized FARC rebels.
To address potential unobserved confounding, in model 3 I limit the sample to compare municipalities where FARC splinter groups emerged in 2017–2018 to municipalities where FARC splinter groups eventually emerged in 2019–2022, using these “not-yet-treated” units as the control group. Any unobserved non-time-varying factors that cause both the emergence of these FARC splinter groups and violence against demobilized FARC rebels should be present in this group of “not-yet-treated” municipalities.
Model 4 assesses the effect of splinter group emergence on a different outcome, the rate of homicides for the general population. The result indicates that the presence of a FARC splinter group in a municipality increases the overall homicide rate in that municipality by 0.11 per 1000 residents. The estimated effect of FARC splinter groups on homicide rates for demobilized FARC members from model 2 is roughly 65 times larger. Moreover, the difference between these estimates is large and statistically significant (95% CI: [1.20, 13.07]). This result supports hypothesis H2A that the rate of violence should increase more for demobilized FARC rebels than for the general population.
The analyses in Table 2 evaluate H2B, which hypothesizes that this effect is driven by FARC splinter groups, rather than other guerrilla or paramilitary groups.
Effects of FARC splinter groups vs. other armed groups.
Notes: The dependent variable in models 1–2 is the average municipality-year rate of homicides against demobilized FARC members per 1000 for 2017–2018. The independent variables in model 1 are two binary variables that take a value of 1 if either FARC splinter groups or other armed groups were present in the municipality between 2017–2018. In model 2 the independent variables are counts of armed clashes involving FARC splinter groups or other armed groups but not FARC splinter groups in 2017–2018. Both models control for the municipality-level population of demobilized FARC members. All models use HC2 standard errors. ∗∗∗p < 0.001; ∗∗p < 0.01; ∗p < 0.05.
Model 1 employs data from Indepaz on the presence of other armed groups such as the ELN and paramilitary groups during the same 2017–2018 period. On the extensive margin, there is a strong and positive relationship between the presence of FARC splinter groups and homicide rates against demobilized FARC members, but this relationship does not hold for ELN and paramilitary group presence. Model 2 focuses on the intensive margin, using data on armed clashes compiled by Colombia's Center for Historical Memory. A larger number of armed clashes involving FARC splinter groups is associated with a higher rate of homicides against demobilized FARC members, but a larger number of clashes involving only the other armed groups does not have an effect of similar magnitude.
These results support my theory highlighting the incentives for splinter groups to target demobilized members of their parent group, while casting doubt on some alternative explanations. I argue that splinter groups have a unique incentive to target demobilized rebels from their parent group, first, because of their imperative to recruit and retain experienced members as they struggle to rebuild, and second, because former members of the parent group possess inside information that could be damaging. The same incentives do not obtain for paramilitary groups or the ELN. These groups had little reason to pursue a potentially suboptimal coercive recruitment and retention strategy, as unlike the FARC splinter groups, they were not facing the type of critical operational juncture that might necessitate such measures. Also unlike the FARC splinter groups, the ELN and paramilitaries had no need to fear the release of sensitive inside information about the FARC to state security forces.
One alternative explanation for the results in Tables 1 and 2 is reverse causality; that is, that violence against former FARC members by other armed groups might motivate the emergence of splinter groups. While I cannot fully rule out such a concern, I show in Appendix A.8 that neither the presence and activity of other armed groups nor the rate of homicides against demobilized FARC members in the first years after the peace agreement has a strong positive relationship with the emergence of FARC splinter groups in later years.
These results are also inconsistent with the alternative explanation that demobilized FARC members were simply caught in the crossfire between armed groups, as such an argument would predict that any increase in armed group clashes should result in violence against demobilized FARC members, regardless of the identity of the groups involved. They also run counter to a narrative that focused blame for the killings on paramilitary groups. In 2019 for example, a senator from the FARC's political party asserted that paramilitary groups, not dissident FARC splinter groups, were behind the violence against demobilized FARC members (EFE Newswire, 2019). While there are some well-documented cases in which paramilitary groups or the ELN killed demobilized FARC members (RCN Radio, 2018), the evidence presented here indicates that FARC splinter groups have been responsible for a significant share of the violence against demobilized FARC members.
Exploring mechanisms
Why does the emergence of rebel splinter groups increase violence against demobilized rebels? I argued that splinter groups target members of the parent group as part of strategy to recruit or retain individuals with valuable skills and experience. In Colombia, several cases suggest that the FARC splinter groups used violence as part of a strategy to remobilize former FARC members. In the case described in the Introduction, reports indicate that Frente 33 had repeatedly tried to recruit the victim, who was a member of a cooperative of demobilized FARC members called REPAZCAT, before killing him. Frente 33 also threatened his family and subsequently forced them to flee (Fiscalía, 2023). This behavior is consistent with a coercive recruitment and retention strategy. The existence of a thriving community of demobilized FARC members in Catatumbo would represent a serious obstacle to Frente 33's efforts to remobilize experienced FARC members in the region, while killing a member of the cooperative who refused to remobilize and displacing his family sends a strong signal that refusing recruitment is not a viable option. Several other documented cases follow a similar pattern, with demobilized rebels being killed by FARC splinter groups after refusing recruitment (El Espectador, 2020; Tubarco, 2020).
I also argued that rebel splinter groups target demobilized members of their parent group because they possess valuable and potentially damaging information. In some instances, evidence suggests that violence against former FARC members was linked to specific intelligence they possessed. One demobilized rebel who was killed in the region of Arauca in mid-2017 had been the head of finances for a local FARC unit, and had reportedly been talking to authorities about turning in his former unit's hidden assets (León et al., 2020). While no perpetrator was identified in that case, the killing is consistent with a logic of suppressing information sharing. Reporting suggests that FARC splinter groups appropriated a significant share of their parent group's assets (Yagoub, 2017), providing them with a strong incentive to prevent information about these assets from reaching Colombian authorities.
In other instances, evidence indicates that violence against demobilized FARC members was intended to root out potential informants. In one case, a FARC splinter group called the Frente Dagoberto Ramos, which was competing for territorial control in the region of Cauca against both the Colombian government and other armed groups, conducted multiple attacks on demobilized FARC members in a DDR camp located in the region. The group eventually delivered an ultimatum to the local town council demanding the expulsion of all remaining demobilized rebels, alleging that the DDR camp was “augmenting the network of informants” in the area, and calling the demobilization camp a “nest of toads”, an epithet for informants (Cardozo, 2022; Soto, 2020).
Conclusion
This paper set out to explain a pattern that has challenged multiple DDR programs: selective violence against demobilized rebels. Its theoretical argument highlighted the incentives for splinter groups to target demobilized members of their parent group to prevent the sharing of sensitive information and to coerce experienced combatants to remobilize. The pattern of violence against demobilized FARC rebels in Colombia supports this argument, as the rate of violence against this population increased significantly with the emergence of FARC splinter groups.
As an increasing number of armed conflicts conclude in negotiated settlements (Fazal, 2018), cases like Colombia and the DRC in which rebel groups fragment into splinter groups during a peace process may emerge more frequently (Duursma and Fliervoet, 2021). Because demobilizing former combatants is an integral component of most peace processes, my findings highlight a critical challenge to peace processes: DDR programs are unlikely to succeed at their aims if governments cannot credibly guarantee the survival of participants. This is of particular importance in multi-party conflicts, where members of other armed groups in negotiations with the government may hesitate to demobilize if they anticipate experiencing a similar fate. In Colombia, where the government has also pursued peace talks with guerrilla groups like the ELN and paramilitary groups like the AGC, conflict analysts argue that the killings of former FARC combatants have been an obstacle to negotiations and a “strong deterrent to disarmament” (Rueda, 2024).
My results also highlight policy issues that warrant further attention from scholars and practitioners. First, I demonstrated that a meaningful share of the violence against demobilized rebels emerged as a consequence of rebel fragmentation. While researchers have cataloged a number of risk factors for rebel group fragmentation, my findings further underscore the urgency of identifying causes of fragmentation and investigating how governments can predict, prevent, or respond to it during and after negotiations. Second, examples like the DRC and Colombia indicate that standard models of DDR can be vulnerable to violent disruption. As data on DDR programs become increasingly accessible (Palik et al., 2025; Sharif, 2026), one aim for future research might be to identify components of DDR programs that render them resilient to such threats.
Supplemental Material
sj-pdf-1-cmp-10.1177_07388942261420595 - Supplemental material for Either with us or against us: The threat of rebel group fragmentation to demobilized rebels
Supplemental material, sj-pdf-1-cmp-10.1177_07388942261420595 for Either with us or against us: The threat of rebel group fragmentation to demobilized rebels by Frank Wyer in Conflict Management and Peace Science
Supplemental Material
sj-zip-2-cmp-10.1177_07388942261420595 - Supplemental material for Either with us or against us: The threat of rebel group fragmentation to demobilized rebels
Supplemental material, sj-zip-2-cmp-10.1177_07388942261420595 for Either with us or against us: The threat of rebel group fragmentation to demobilized rebels by Frank Wyer in Conflict Management and Peace Science
Footnotes
Acknowledgements
I am grateful to the CMPS editors and two anonymous reviewers for their comments. I also thank Darin Christensen, Chad Hazlett, Barbara Geddes, Leslie Johns, and Timothy Jones for helpful feedback in the early stages of this project, and the University of California Institute on Global Conflict and Cooperation for financial support.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
Supplemental material
Supplemental material for this article is available online.
