Rebel institutions and negotiated peace

Abstract

Civil wars are difficult to resolve through negotiated settlements. Rebel institutions are thought to make negotiations more successful. I show, however, that this positive association does not hold. Rather, rebel service provision is negatively correlated with successful negotiated settlements. The well-established literature on commitment problems suggests that negotiated settlements are not reached because governments end negotiations amidst fears of rebel growth from civilian support derived from service provision. I offer an alternative explanation—strategic stalling—based on rebel incentives to realize the full long-term benefits of service provision. Qualitative evidence shows that observable implications of strategic stalling are observed in the cases surveyed.

Keywords

Commitment problems peace negotiations rebel governance

Introduction

Unlike interstate wars, civil wars typically endure for decades, and rarely end with a negotiated settlement. Between 1945 and 2009, the average rebel group existed for 15 years, and only 26% of them signed a negotiated settlement.¹ Many scholars have argued that a key impediment to ending civil war is the government's inability to credibly commit to uphold a settlement after the rebels have disbanded (Fearon 2004; Fearon and Laitin 2008; Walter 2002, 2009; Driscoll 2012; Glassmyer and Sambanis 2008). Civil war termination scholars focus heavily on how to mitigate the government's commitment problem, such as the presence of third-party guarantees and security sector reforms (Walter 2002; Toft 2010; Uzonyi and Hanania 2017; Licklider 1995). Rebel institutions, however, should also mitigate government commitment problems. Rebel governance mobilizes civilians who in turn hold the government accountable after civil war (Huang 2016), thereby resolving the government's commitment problem. A smaller strand of the literature examines the rebel side of the equation more closely (Mattes and Savun 2009: 739). Rebel institutions also abate difficulties rebels face when committing to governments. These institutions reinforce rebel centralization, giving them greater control over their fighters. In turn, rebel groups can commit to the government that their leaders can prevent factions from using violence to derail the peace process, or “spoil” the peace (Heger and Jung 2017; Zartman 1989). This leads to a strong a priori expectation that rebel service provision, a common aspect of rebel governance (Fukuyama 2013), should facilitate successful negotiated settlements.

Although rebel service provision and other institutions might not directly translate into battlefield superiority (Stewart 2020), they can create other forms of strength that are just as important to civil war, incentivizing rebels to divert resources from building their fighting capacity to providing services to civilians. Rebel service provision wins civilian support and facilitates state-building (Kalyvas 2006; Heger and Jung 2017; Stewart 2018), and rebel taxation gives rebels a steady base of resources for their fighters (Huang 2016). Rebel governance can solicit support internationally through diplomatic missions (Coggins 2015) and organize civilians domestically (Arjona 2014, 2016). These institutions also give rebels legitimacy, both domestically and internationally, while eroding legitimacy from the government (Ledwidge 2017; Mampilly 2015). Rebel governance and service provision translate into rebel capacity,² which improves rebels’ ability to bargain and increases the likelihood of a negotiated agreement (Walter 2009; Clayton 2013).

Existing theories, however, fail to consider an important characteristic of rebel service institutions: they are a long-term investment. Unlike guns and bullets, which do not increase in value over time, service institutions bring rebels increasing benefits over time (Kilcullen 2011; Arjona 2016). This important factor has not been seriously considered in theory building about the impact of rebel service institutions on peace. However, it does have important implications, and suggests two possible countervailing mechanisms at work. Both mechanisms predict a negative effect on successful negotiation, not positive as has been suggested in the literature. The first explanation is a rebel commitment problem driven by civilian mobilization, rather than from spoilers within a rebel group. Rebel service provision mobilizes civilian support for the group (Heger and Jung 2017; Kalyvas 2006), and pools of potential recruits increase a rebel group's ability to sustain conflict (Walter 2004). Governments do not trust rebels to refrain from regrouping through their increased pools of civilian collaborators. Therefore, they are unwilling to negotiate with the rebels. The second explanation is that, because rebel service provision is a powerful tool to win hearts and minds and gain civilian support over time (Kalyvas 2006: 128), if a rebel group can survive long enough they will continue to receive the dividends of their investment in service provision.

I use the Rebel Quasi-State Institutions dataset (Albert 2022) to show that, contrary to expectations in the literature, the correlation between rebel service provision and negotiated settlements is negative. Since there are two possible explanations for a negative correlation, I use additional quantitative and qualitative evidence, partly based on archival research, to differentiate between them. I show that rebel commitments are unlikely to explain the negative effect of service provision on negotiated settlements. Even though commitment problems are a common theoretical explanation as to why negotiated settlements are difficult to reach, they fail to explain why rebel service provision has a negative effect on reaching successful negotiated settlements. Furthermore, while it is difficult to prove directly that service provision incentivizes rebel stalling, the evidence more strongly supports the stalling explanation over a service-provision created rebel commitment problem.

This has direct implications for policy-makers. One of the biggest mistakes negotiators make is misunderstanding the nuances of rebel strength: rebel groups do not just invest in military strength. They invest in institutions which give them longevity. Mediators and policy-makers must take into account rebel institutional capacity when negotiating with rebels. Since rebel institutional capacity, particularly through service provision, is a different type of capacity than military strength, mediators need to understand how institutions impact rebel strategy. Misjudging rebel institutional strength will lead to mediation failure for two reasons. First, mediators need to understand that rebels with greater institutional capacity have stronger incentives to stall final agreements. Engaging in negotiations gives rebels the opportunity to expect a reprieve from fighting. This is often considered from the military perspective where governments’ concerns over rebel military mobilization are at times addressed. However, there is an institutional implication derived from stalling as well. Rebel service institutions often persist during periods of low violence, and because there is less need for offensive military action, breaks in fighting mean resources can be used for institutional maintenance. Time strengthens service provision institutions and increases the benefits they give to rebels. This leads to the second reason why civil war mediation will fail to produce a negotiated settlement when rebel institutional strength is misunderstood: successful peace deals will have to address the future expectations of strength derived from service provision. When rebels expect to survive long enough to realize the benefits of civilian service provision, accepting a peace agreement today based on their present capacity is unlikely to convince rebels to disarm.

Strategic stalling and rebel commitment problems: two competing theories

Previous scholarship has argued that service providing terrorist groups are more likely to have stable talks with the government, resulting in cessation of violence (Heger and Jung 2017). Providing services gives rebels a wider base of civilian support and they are more likely to have a centralized organizational structure, which is further solidified by service provision. Governments find rebel groups with a strong centralized command and control more attractive as negotiating partners, since this organizational trait decreases the probability of spoilers—those who derail peace talks through violence without the approval of the rebel leadership. Service-providing rebels are also expected to be stronger and more capable than other rebels since providing services corresponds to territorial control and greater centralization, and is costly (Kasfir 2015; Mampilly 2011; Stewart 2018; Heger and Jung 2017; Weinstein 2007; Kalyvas 2006, 2015). Stronger rebels are, in turn, more likely to reach a negotiated settlement (Walter 2009).

Even though the literature suggests a strong a priori expectation of a positive relationship between rebel service provision and conflict ending through negotiated settlement, I argue there are at least two possible theories that call this positive relationship into question. The first is rebel strategic stalling, which I argue is necessary to fully glean the benefits of investing in service provision. The second is the rebel side of the commitment problem. That is, rebels’ difficulty in committing to not remobilize during negotiations and the implementation of an agreement. However, both quantitative and qualitative evidence suggests that rebel strategic stalling is the more likely explanation as to why there is a negative relationship between rebel service provision and successful negotiated settlements. An overview of these theories is presented in Table 1, including observable implications for each explanation.

Table 1.

Overview of theories.

Explanation	Correlation between rebel service provision and negotiated settlements	Observable Implications
Rebel strategic stalling	Negative	Rebels will engage in talks and agree to ceasefires but delay signing any peace agreement
Rebel commitment problem	Negative	Governments will be wary of signing ceasefires and end the talks due to fears of rebel remobilization
Mitigates rebel commitment problem	Positive (Heger and Jung 2017)	Conflicts ending through a negotiated settlement

Rebel strategic stalling

The logic of strategic stalling is built on two assumptions. First, rebels invest their resources in things they believe will help them achieve their goals (Popkin 1979; Kasfir 2015). Second, rebel service provision, while perhaps not as cost prohibitive as some may argue (Arjona 2016: 60), is still costly (Stewart 2018). Together, these two assumptions imply that when rebels choose to invest in service provision they do so because they believe it will benefit them, in some way and at some point in time. Since rebels, like many organizations, face a budget constraint, investing in service provision naturally implies that there is less to invest in military capability (i.e. guns, ammunition, etc.).

Service provision, however, has the potential to benefit rebels immensely. As part of a “winning hearts and minds” strategy ultimately generating loyalty, service provision leads to civilian collaboration (Kalyvas 2006: 124, 128). Providing services also legitimizes rebels and their control, particularly for civilians (Ledwidge 2017; Stewart 2016). Rebels, much like the state, can use public education to socialize civilians into their ideology (Weber 1976). Parents in ISIS-controlled Mosul recognized this and consequently, despite the risk of severe punishment, kept their children home instead of sending them to the ISIS-run schools (Lafta et al. 2018). Rebel groups also recognize this. During an interview with the Kurdish Ministry of Education in Erbil, Iraq, controlled exclusively by the Kurdish Democratic Party until 2006, I was told that “teaching children to love their country [Kurdistan]” was one of the main goals of Kurdish education.³ In the development of the French state, education was used “to teach children national and patriotic sentiments, explain what the state did for them and why it exacted taxes and military service, and show them their true interest in the fatherland” (Weber 1976: 332). The “imagined community” of the nation is solidified through state education (Anderson 2016). Just as state education teaches civilians what legitimizes the government's authority, rebel education communicates and reinforces rebels’ legitimacy.

However, this is not limited to education. Other services also garner civilian support. They provide an avenue for regularized non-violent interaction with civilians, which is central to winning hearts and minds (Kilcullen 2011). Healthcare, for example, not only invests in a healthier population from which rebels can recruit, but it also communicates that they care about the health and welfare of civilians. Law and Order provision builds and reinforces rebel legitimacy, all while undermining the legitimacy of the government (Ledwidge 2017). Rebel legitimacy and evidence of rebel concern about civilian welfare both lead to increased civilian support for rebels. There are multiple paths—ideology, legitimacy, increased civilian welfare and regularized non-violent contact with civilians—through which rebel service provision leads to increased civilian support.

Although service provision provides many benefits, fully realizing these benefits can take time (Weinstein 2007: 171). On the one hand, guns, ammunition and military equipment immediately benefit rebels and can alter the balance of power for bargaining today. Service provision, on the other hand, requires time before the full magnitude of its benefits can be realized. Winning hearts and minds requires consistency to prove to civilians that the rebels will be sticking around. In one sense, the benefits are not rooted in the good itself, but in the civilian interaction over time from the service provision (Kilcullen 2011: 71). Service provision gives rebels with long-time horizons the ability to “[reshape] local life in accordance to its interests, needs, and principles” (Arjona 2016: 11). Rebel territorial control makes this easier (Kalyvas 2006: 128), but it is not always necessary (Ledwidge 2017: 33).

Since rebel service provision is advantageous but requires a longer time horizon, rebels need to survive long enough to realize the full extent of those benefits. If rebels are unable to survive, then service provision is an unwise investment. This incentivizes rebels to ensure their long-term survival, even at the cost of immediate military advantages. Strategically stalling gives rebels such an opportunity. Peace talks and negotiations are expected to be long, drawn out processes, and in some cases are even accompanied by lulls in fighting. Ceasefires, in particular, signal goodwill on behalf of the rebels and their willingness to negotiate.

However, ceasefires and negotiations also buy rebels time where they can focus on providing services and worry less about government offensive military action against them. And because rebels stall during negotiations to buy time to realize additional benefits from their service provision, negotiations will break down before a final peace agreement. The government, then, faces a conundrum. A government cannot offer rebels their future expected value today. Even if a government wants to end the civil war, they also want to remain in power, which means keeping their selectorate, regardless of its size, happy (De Mesquita et al. 2004). Giving rebels more than they are worth today would be viewed as kowtowing to the rebels and would put the government under scrutiny. This means that the government cannot offer enough concessions to get rebels to accept a deal today, when they expect more future concessions because of future increased capacity from service provision. Rebel strategic stalling, then, should lead to a negative correlation between conflicts ending through a negotiated settlement and rebel service provision. There should also be evidence of rebel willingness to sign ceasefires and negotiate with the government but without signing final settlements.

Rebel commitment problem

Commitment problems arise when parties cannot credibly commit to upholding the deal they strike today in the face of a power shift tomorrow (Powell 2006). In the civil war context, when rebels disarm, which is sometimes demanded prior to entering peace talks, it increases the government's power. Unless governments can credibly commit to the rebels that they will honor the agreement, rebels will refuse to negotiate or sign a peace agreement. This is why getting rebels to disarm is considered the most important obstacle to overcome to peacefully resolve civil war (Driscoll 2012). This is the government commitment problem. The other side is the rebel commitment problem. Theoretically, a rebel commitment problem occurs when the rebels decide to remobilize during negotiations or implementation of a settlement (Walter 2009). Because the power dynamic is generally in favor of the government in civil wars, little has been done to explore the rebel commitment problem.

Rebel service provision affects commitment problems in two ways. First, rebel service provision mobilizes civilians to keep the government accountable, even after they disarm (Huang 2016). This means that rebel service provision should help mitigate the government commitment problem. The second way rebel service provision affects commitment problems is through its ability to generate civilian loyalty and collaboration, and through redefining identities (Kalyvas 2006; Ledwidge 2017). As civilians move to genuinely support the rebel cause, rebels can recruit more, and better, fighters. Pools of potential recruits mean rebels retain the ability to remobilize, even if they begin to disarm (Walter 2004). This is the rebel commitment problem. Since service provision, especially over time, solidifies civilian loyalties, rebels face difficulties committing to not remobilizing their loyal civilians. If rebel service provision invokes rebel commitment problems then there should, like rebel strategic stalling, be a negative relationship between service-providing rebels and successful negotiated settlements. However, rebel commitment problems can be differentiated from strategic stalling because a rebel commitment problem would lead to a negative correlation between service provision and ceasefires since rebels can use ceasefire periods to remobilize. Qualitative evidence in support of a rebel commitment problem includes government statements expressing their concerns about rebels remoblizing or rearming during ceasefires/peace talks.

Data

There are two main datasets I used to conduct the analyses here. First, for data on rebel service institutions I used the Rebel Quasi-State Institutions dataset (Albert 2022). This dataset provided the most extensive time-series coverage of service institution variables for the Uppsala Conflict Data Program (UCDP) rebel groups.⁴ I chose to use the Rebel Quasi-State Institutions dataset for two reasons. First, it includes additional institutions and services to test the strategic stalling theory. Second, it is built from the UCDP armed actors list. This means that the Rebel Quasi-State Institutions dataset includes more groups that match with data on peace agreements, mediation, battle deaths and rebel group characteristics, all of which will be important control variables. The second dataset I used is the Conflict Termination dataset (Kreutz 2010), which codes for conflict termination type and, if applicable, the date a peace agreement was signed.

Main explanatory variables: rebel service institutions

There are a number of possible institutions that rebels could choose to create. Previous studies using cross-national data on service aspects of rebel governance have focused heavily on the major service sectors: education, health care and justice (Heger and Jung 2017; Huang 2016; Jung 2017; Stewart 2018, 2016). The data I use here cover these major service sectors, as well as other, less common, service institutions. To evaluate the theory here I used the Rebel Quasi-State Institutions dataset (Albert 2022) to create new variables of rebel service provision. The strategic stalling theory suggests that service provision over time will decrease the likelihood of seeing successful negotiated settlements that lead to the end of conflict.

To capture the effect over time I use count variables of the five most common rebel service institutions: Education, Health care, Law, Policing and Justice. These data come from the Rebel Quasi-State Institutions dataset (Albert 2022) and code for the presence or absence of these services for each year the rebel group existed. I took this binary data and created count variables over the years for each service. These variables count, over time, how many years the rebel group provided the service. Breaks in service provision over two years restart the count.

Outcome variable: negotiated settlements

The outcome of interest here is whether or not governments and rebels are able to reach a peace deal. There are several different types of negotiated settlements. Ceasefire agreements are between two parties that they will halt active fighting. A ceasefire may or may not include provisions for monitoring or withdrawing forces, but they usually allow rebels to remain armed. Peace agreements, however, usually require rebels to disarm and reintegrate (Walter 2009). Unlike ceasefires, peace agreements offer some sort of political solution to the conflict issue, which makes them an important step toward achieving peace. Because ceasefires do not require rebels to disarm, they are often easier to reach. Rebels do not have to worry about governments attacking them when they are vulnerable after disarming. Peace agreements are more difficult to reach because both sides must agree to a political solution. Sometimes governments and rebels will sign agreements establishing a framework for peace talks. These agreements, however, are not included here as negotiated settlements because they lack a political solution to the conflict. Since I am interested in negotiated settlements that end conflict, ceasefires are not considered a negotiated settlement here, and I focus instead on peace agreements.

The peace agreement variable I use here codes for a verbal or signed agreement by the major parties of the conflict at any time. The Conflict Termination dataset identifies the year a peace agreement was signed, and this data is used to create a binary variable for whether or not the parties are under a peace agreement, that is, if a peace agreement was signed in that year or a previous year. The panel for each rebel group ends when the group ends by (1) disbanding, (2) ceasing to engage in any activity or (3) renouncing violence and becoming a political party.⁵ Because each rebel group is coded over the course of their existence, not just the years where armed conflict is present, I need to establish rules for how to code the years after the peace agreement is signed. Rebel services provided under a peace agreement need to be identified because I consider a negotiated settlement to have failed, and code it as 0, when armed conflict occurs after two years without armed conflict. In cases where a peace agreement is signed but armed conflict continues, a negotiated settlement fails after the fifth year of armed conflict.⁶ Successful negotiated settlements are signed peace agreements that lead to a rebel group no longer existing.

I have included a failure mechanism because I am specifically looking at peace agreements that end conflict. Ultimately, understanding how conflict ends is the goal. There is little value in understanding when a peace agreement is signed if that peace agreement does not at least partially attempt to end the conflict itself. However, I have included in Section A.3 of the Online Appendix model results that only consider if a peace agreement is signed. In one version, the panel data is used and in another the data are collapsed to a rebel group unit of analysis. The results are substantively similar to those presented here.

Control variables

Since this data is observational, I control for three categories of potential omitted variables: (1) state-level factors that account for government decision to remain in peace talks and their ability to credibly commit; (2) conflict characteristics that make peace agreements more difficult to reach; and (3) rebel characteristics that might be correlated with the decision to create institutions and with the prospects of peace. A summary of the variables in each category is presented in Table 2.

Table 2.

Summary of control variables at the rebel group, state and conflict levels.

Rebel-level controls	State-level controls	Conflict-level controls
Rebel group size (ln)	Participatory democracy	Yearly battle deaths (ln)
Rebel group strength	GDP per capita	Military victory
Rebel group centralization	Primary enrollment ratio	Mediation attempt
Rebel territorial control	Infant mortality	Territorial conflict
Communist ideology
Ethnic based
Foreign state support for rebels

The first category captures state-level variables that might impact the government's decision to walk away from negotiations for reasons other than the rebels’ inability to credibly commit. The first is the level of participatory democracy.⁷ Participatory democracy captures the level of participation in all political processes. Higher participatory democracy indicates norms of engagement and discourse, which should make the government more willing to engage rebels in negotiations and increase their ability to credibly commit to a peace deal. The other state-level controls are gross domestic product per capita,⁸ primary education enrollment ratio⁹ and infant mortality rates.¹⁰ These controls account for the government provision of social services, and the rebels' decision to provide services in the first place.

The second category of controls are conflict characteristics that impact how difficult a conflict is to resolve. Mediation,¹¹ which involves the presence of a third party facilitating discussions between governments and rebels, makes reaching peace agreements easier, while higher numbers of battle deaths¹² make peace agreements more difficult to reach. Military victories¹³ capture cases where one side has a clear military superiority, which disincentivizes the winning party from agreeing to a peace deal. Finally, I control for separatist conflicts,¹⁴ which are positively correlated with mediation and with the presence of rebel service institutions.¹⁵ Separatist conflicts are wars where the rebels are fighting for independence or autonomy of a specific territory within the state. They differ from rebel territorial control, which captures when rebels physically control territory. Rebel territorial control can occur even when rebels are not fighting a separatist conflict, and rebels fighting for territory will not necessarily control territory.

Finally, I need to consider potential rebel characteristics that make peace agreements more likely. These variables, which come from the Non-State Actor dataset (Cunningham et al. 2013) and the FORGE dataset (Braithwaite and Cunningham 2019), are discussed in the following section. I include a longer discussion on these variables because to isolate the correlation between negotiated settlements and rebel service provision, I need to control for rebel characteristics that influence their ability to reach a negotiated settlement with the government.

Some rebel groups may possess characteristics that make it easier to achieve a successful negotiated settlement. Stronger rebels who can continue the armed struggle cause greater damage to the government or mobilize the population might give the government greater incentives to negotiate and settle (Walter 2009; Clayton 2013). Rebels with stronger command and control over their soldiers to deter defectors and spoilers are also more likely to reach a settlement (Heger and Jung 2017; Cunningham 2011). Others have argued that separatist conflicts are more difficult to resolve (Fearon 2004; Fuhrmann and Tir 2009), and other rebel characteristics, like ethnicity-based rebel groups or communist rebels, might impact the capability of reaching a settlement (Balcells and Kalyvas 2015).

The literature has found that as rebels approach parity with the government, mediation and negotiated settlements are much more likely (Mason and Fett 1996; Regan and Michael Greig 2008; Zartman 1989), and I find similar results using a logistic regression model with these variables to see the likelihood of a negotiated settlement. The results are presented in Table 3. Centralization increases the likelihood of a peace agreement. Rebel strength relative to the government, a categorical variable, is set with the base category as “much weaker”. Therefore, increasing in rebel strength from “much weaker” relative to the government, to “parity”, “stronger” and “much stronger”, are all positive and significant. Only 11 rebel groups, out of the 235 rebel groups in the Rebel Quasi-State Institutions dataset, are categorized by the Non-State Actor data as being stronger than the government at some point in time.¹⁶ There are 31 groups that are categorized as having parity with the government, which means that the overwhelming majority of groups are weaker than the government. The positive and statistically significant coefficients mean that stronger rebels are more likely to reach a successful negotiated settlement.

Table 3.

Likelihood of a negotiated settlement.

	Dependent variable: Peace Agreement
	(1)	(2)	(3)
Centralization	0.475 $^{* *}$
	(0.225)
Rebel size (ln)		0.114
		(0.107)
Rebel strength: weaker			0.615 $^{* *}$
			(0.301)
Rebel strength: parity			2.023 $^{* * *}$
			(0.369)
Rebel strength: stronger			1.617 $^{* *}$
			(0.708)
Rebel strength: much stronger			3.282 $^{* *}$
			(1.503)
Ethnic rebel group	0.097	0.386	0.049
	(0.296)	(0.281)	(0.264)
Communist ideology	$-$ 0.870 $^{* *}$	$-$ 0.819 $^{* *}$	$-$ 0.906 $^{* * *}$
	(0.359)	(0.342)	(0.329)
Separatist conflict	$-$ 1.479 $^{* * *}$	$-$ 1.265 $^{* * *}$	$-$ 0.924 $^{* * *}$
	(0.391)	(0.340)	(0.312)
Observations	1750	1624	1826

Note: Cubic time polynomial included.

$^{*}$ p $<$ 0.1; $^{* *}$ p $<$ 0.05; $^{* * *}$ p $<$ 0.01.

Standard errors clustered by rebel group. Logistic regression.

Base category for rebel strength is “much weaker”.

Rebel groups with a communist ideology or fighting for territory (autonomy or independence) are both less likely to reach a negotiated settlement, which is consistent with previous studies (Fearon 2004; Fuhrmann and Tir 2009; Balcells and Kalyvas 2015). The natural logs of rebel group size and ethnic rebel groups, however, are both insignificant.

Rebel services and successful negotiated settlements

To test the argument that service provision creates an incentive for rebels to stall, there should be a negative correlation between rebel services and a conflict ending with a negotiated settlement. I used panel data with a count variable for each rebel service that counts the number of years rebels have provided that service. There should be an institutional effect of these services where their impact increases the longer they have been provided. Rebel education, for example, should have a stronger effect on civilian support if it has been provided for a decade than in the first year it is provided. Any break longer than two years restarts the count.

Table 4 presents the results the from linear probability models.¹⁷ Owing to the high correlation between the different services I run individual models for each service and include a time count variable to account for how long the rebel group has existed. Models 1–5 use the service count variables, and the count over time variable for education, health, law, policing and justice rebel services. The coefficients are all negative; however, education, law and justice are also statistically significant. Although the coefficients are small, they represent the change for each additional year of rebel service provision. When rebels provide these services the average number of years the service exists is over 10 years for each of the top five services. Given this, the results here mean that over the life span of the top five service institutions, the likelihood of a negotiated settlement decreased by 1–2% compared with a rebel group that never provided these services.

Table 4.

Linear regression results for service count variables.

	Dependent variable: Negotiated settlement
	(1)	(2)	(3)	(4)	(5)
Education count	$-$ 0.001 $^{*}$
	(0.001)
Health count		$-$ 0.001
		(0.001)
Law count			$-$ 0.002 $^{* *}$
			(0.001)
Police count				$-$ 0.0004
				(0.0004)
Justice count					$-$ 0.001 $^{*}$
					(0.0004)
Battle deaths (ln)	$-$ 0.004 $^{*}$	$-$ 0.004 $^{*}$	$-$ 0.004 $^{* *}$	$-$ 0.003 $^{*}$	$-$ 0.004 $^{*}$
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002))
Mediation	0.052 $^{* *}$	0.055 $^{* *}$	0.060 $^{* * *}$	0.054 $^{* *}$	0.053 $^{* *}$
	(0.022)	(0.022)	(0.023)	(0.024)	(0.023)
Separatist conflict	$-$ 0.014	$-$ 0.016 $^{*}$	$-$ 0.014	$-$ 0.016 $^{*}$	$-$ 0.018 $^{*}$
	(0.009)	(0.009)	(0.009)	(0.009)	(0.009)
GDP per capita (ln)	$-$ 0.017 $^{* *}$	$-$ 0.017 $^{* *}$	$-$ 0.014	$-$ 0.015 $^{*}$	$-$ 0.016 $^{*}$
	(0.008)	(0.008)	(0.009)	(0.009)	(0.008)
Rebel strength: weaker	0.020 $^{*}$	0.018 $^{*}$	0.018 $^{*}$	0.021 $^{* *}$	0.024 $^{* *}$
	(0.010)	(0.010)	(0.010)	(0.010)	(0.010)
Control territory	$-$ 0.001	$-$ 0.002	$-$ 0.002	0.001	$-$ 0.001
	(0.012)	(0.012)	(0.011)	(0.012)	(0.012)
Communist	$-$ 0.014	$-$ 0.011	$-$ 0.018	$-$ 0.012	$-$ 0.011
	(0.011)	(0.011)	(0.011)	(0.011)	(0.010)
Number of foreign	0.004	0.004	0.002	0.003	0.003
supporters	(0.004)	(0.004)	(0.003)	(0.003)	(0.003)
Observations	1018	1031	1031	1025	1010
Year fixed effects	✓	✓	✓	✓	✓

Note: Time count variable of rebel existence included for yearly service count variables. Standard errors clustered by rebel group. Controls included but not reported: participatory democracy, primary education enrollment ratio, infant mortality, government, military victory, rebel military victory, centralization, and rebel size (ln). Yearly analysis including years under a negotiated settlement until failure or group disband.

$^{*}$ p $<$ 0.1; $^{* *}$ p $<$ 0.05; $^{* * *}$ p $<$ 0.01.

The results show a negative correlation between rebel services and a conflict ending with a negotiated settlement even when considering different services. Robustness checks in Section A.3 of the Online Appendix use event analysis where the event is signing an agreement, rather than ending the conflict, and the substantive results are similar to those in Table 4. Appendix A.4 also includes models using the other institutions. Section A.5 of the Online Appendix discusses why these results are opposite to those found with terrorist group service provision (Heger and Jung 2017).

While these results suggest that previous expectations regarding the relationship of rebel service provision and negotiated settlements are incorrect, they do not allow us to differentiate between a rebel commitment problem and a stalling explanation. However, these two explanations have different expectations of how episodes of armed conflict end. On the one hand, a rebel commitment problem arising from rebel service provision would lead to government unwillingness to come to any agreement, including a ceasefire, with the rebels. This is because in a commitment problem the rebels cannot commit to not using the ceasefire as an opportunity to rearm and strengthen their forces. Any observed breaks in fighting, then, would be the result of low activity by both sides. Stalling, on the other hand, would lead rebels to value ceasefires and the government, which is not concerned with the rebel commitment problem, would agree to ceasefires in hopes of ending the conflict. Since a rebel group can remain armed and active while continuing to provide services amidst lightened military expectations during ceasefires, they are advantageous for rebels. Strategic stalling, then, would lead to an increased likelihood of signing a ceasefire as it does not concern the government and is advantageous to rebels.

I collapsed the Rebel Quasi-State Institutions data into a rebel group-episode unit of analysis, based on the episodic years provided by UCDP. Yearly control variables were averaged for each episode. I ran a multinomial logit regression using the following episode outcome types: Peace Agreement, Ceasefire, Government Military Victory, Rebel Military Victory, Low Activity, and Other. The peace agreement outcome requires that the agreement provides a political solution to the conflict and must be signed within two years of when the episode ends. Ceasefires include those agreed to by both parties and may or may not include conflict regulation protocols. Low activity captures the cases where there is no formal end to the episode, rather there is at least one year where there were less than 25 battle-related deaths (Kreutz 2010).

The results for the multinomial logit models are presented in Table 5. Each row represents a separate model run with the same set of control variables. Ceasefire is the baseline category. This means that for Education count, which is number of years the rebels provided education during that episode, each outcome listed is compared with the chances of ending in a ceasefire. The comparison of interest is Low Activity, presented in column 2. The negative sign on the coefficient means that as the number of years rebels provide services increases, the chances of the episode ending with a ceasefire are higher compared with ending with low activity. However, only the education model produces a statistically significant coefficient. As another robustness check, the comparison between ending with a ceasefire compared with a peace agreement is also negative. This means that as the number of years rebels provide services increases, ending in a peace agreement is less likely relative to ending with a ceasefire. Although none of these coefficients are statistically significant, they are consistent substantively with the results in Table 4.

Table 5.

Multinomial logit results comparing outcomes with ceasefires.

	Dependent variable:
	Peace agreement	Low activity	Government victory	Rebel victory	Other
	(1)	(2)	(3)	(4)	(5)
Education count	$-$ 0.147	$-$ 0.119 $^{*}$	$-$ 0.524	$-$ 0.078	$-$ 0.038
	(0.093)	(0.071)	(0.399)	(0.170)	(0.091)
Health count	$-$ 0.142	$-$ 0.082	$-$ 0.245	$-$ 0.226	$-$ 0.031
	(0.105)	(0.072)	(0.252)	(0.212)	(0.093)
Law count	$-$ 0.021	$-$ 0.006	0.222	0.434	$-$ 0.010
	(0.087)	(0.075)	(0.165)	(0.403)	(0.133)
Policing count	$-$ 0.005	$-$ 0.043	0.304	$-$ 0.354	$-$ 0.043
	(0.083)	(0.077)	(0.197)	(0.276)	(0.113)
Justice count	$-$ 0.082	$-$ 0.082	$-$ 0.028	$-$ 0.007	$-$ 0.077
	(0.079)	(0.070)	(0.149)	(0.191)	(0.086)
Akaike information criterion	550.700	550.700	550.700	550.700	550.700

Note: Base category is ceasefire. Each service was run separately with same controls. Controls included but not reported: episode duration, number of foreign supporters, average democracy score, average primary education enrollment ratio, average infant mortality rate, average GDP per capita, separatist conflicts, average number of mediation attempts, average battle deaths, communist ideology, ethnic rebel group, average level of rebel centralization, rebel territorial control, rebel strength and average rebel size.

$^{*}$ p $<$ 0.1; $^{* *}$ p $<$ 0.05; $^{* * *}$ p $<$ 0.01.

Peace processes in Colombia and Northern Ireland

The negative correlation between rebel service provision and successful negotiated settlements means that there is no support for the argument that service provision reinforces centralization and prevents spoilers, thereby increasing the likelihood of a negotiated settlements. While the positive and statistically significant correlation between ceasefires and rebel service provision, relative to low activity, lends support in favor of the strategic stalling mechanism over a rebel commitment problem, it is still a correlation. Therefore, I supplement the results above with qualitative evidence on the negotiations that occurred between the Colombian government and the FARC between 1998 and 2002, and the British government and the Provisional Irish Republican Army (PIRA) between 1972 and 1976.

To test between the rebel commitment problem and strategic stalling theories, I explore three areas where these two theories offer distinct observable implications. The first is who walks away from the negotiations first. The second is if the government expresses concern over continued rebel service provision during the negotiations. Finally, did the government allow the rebels to create a political party? Table 6 summarizes the expectations of each theory.

Table 6.

Observable implications of theories for FARC and PIRA cases.

	Theory 1:	Theory 2:
	Rebel commitment	Theory 2:
Observation	Problem	Strategic stalling
Who walked away?	Government	Rebels (temporarily)
Concerned over rebel service provision?	Government concerned over service provision	Continued provision with no government concern
Rebel political partiesallowed?	Government actively prevent	Government allows them and rebels create them

These cases were selected for two reasons. First, they should be difficult cases for the theory proposed because they possessed several qualities that the literature argues should make negotiated settlements more likely. Both groups were centralized and growing in military strength during the time periods in question (Hennessey 2015: 8), and although the FARC eventually reached a peace agreement with the Colombian government, it was after 2012 and therefore not a positive case of negotiated settlement in the Rebel Quasi-State Institutions data. Additionally, 14 years separate the FARC case studied here, which did not end with a peace agreement, and when they did eventually sign a peace agreement in 2016. This sufficiently separates the two observations for a qualitative case study. The PIRA did, however, reach a peace agreement in 1998, and formally dissolved in 2005, making it a positive case of negotiated settlement in the data. The FARC and the PIRA also differed in their goals. While the FARC was looking to control the central government in Colombia, the PIRA wanted to separate from the UK and join Ireland.

Second, negotiations are complex with many different factors at play; however, the results in the preceding sections suggest that, on average, rebel service provision is negatively correlated with successful negotiated settlements. Commitment problems can arise from the government or the rebel side through different mechanisms, but rebel service provision should only have a negative impact on negotiated settlements through a rebel commitment problem. Therefore, I need to control, as best as possible, for the government commitment problem, which could still be present during negotiations. PIRA and the FARC are both ideal cases to do this because during the negotiation periods I examine, the government did not require either group to disarm. In fact, in the case of the FARC, the Colombian government even granted them full control¹⁸ of a sizable piece of territory. In civil war, the main source of a government commitment problem is that rebels are expected to disarm, which the British and Colombian governments did not demand during the negotiations. Since I am attempting to isolate the rebel side of the commitment problem, using cases with limited government commitment problems provides a clearer examination of the rebel side.

Who walked away from the negotiations?

To help distinguish between the two proposed mechanisms I first consider who walked away from the negotiations. If a rebel commitment problem is driving the negative correlation then there should be evidence of the government ending negotiations amidst fears of rebel regrouping, specifically related to the rebels’ service provision. Strategic stalling, however, would reveal a pattern of rebels temporarily walking away from negotiations. The PIRA declared a number of unilateral ceasefires between 1972 and 1976. The British government stated that the Army would be rolled back in response to genuine evidence of a cessation of violence, and that other practices, like photographing, Interim Custody orders and searches, would also be discontinued (TNA 1975b). This was not simply propaganda on the part of the government; British intentions of reducing the Army presence in Northern Ireland in response to a cessation of PIRA violence are confirmed through secret, internal documents (TNA 1974). The PIRA, in exchange for the ceasefire, made it clear that they expected to see concessions from the British government, like an end to internment, the release of political detainees or a public statement regarding British intentions of eventually leaving Northern Ireland. These ceasefires, however, were unilateral and the PIRA ended them,¹⁹ stating that the process was not moving quickly enough, citing violations by the British Army or complaining that too few prisoners were released. The British government, however, continued talks with the PIRA, sometimes through intermediaries and other times directly with PIRA representatives (TNA 1975a; Hennessey 2015).

Promises of successful negotiations with the FARC won Andrés Pastrana the Colombian presidency (Isacson and Rodriguez 2009: 23). In an effort to get the FARC to the negotiation table, in 1998 Pastrana offered a demilitarized zone roughly the size of Switzerland for the FARC to control in exchange for sitting down at the negotiation table (Braun 2007: 30). The military, which was withdrawn from the demilitarized zone, did not support Pastrana's decision to grant the FARC territorial control and believed the FARC would use the territory to expand and launch military operations (Kline 2007: 54-5). These concerns were valid and during the time period where the demilitarized zone was in place, the FARC grew to its peak membership, launched military operations and established extensive governance and service provision (Kline 2007; Dudley 2004; Nasi 2009; Beittel 2015). However, despite evidence that the FARC was strengthening and mobilizing during the negotiations, Pastrana remained at the negotiating table. It was the FARC that “froze” the peace talks four times between 1998 and 2002. Additionally, when Pastrana did eventually end negotiations it was attributed not to the FARC growth from their service provision, but to their hijacking a plane, which was landed in the demilitarized zone, and kidnapping a senator. After the FARC refused to hand over the hijacker, Pastrana ordered the military to retake the demilitarized zone (Kline 2007). Similar to the PIRA, peace talks with the FARC lend support for strategic stalling over a rebel commitment problem. While the Colombian military was clearly concerned with FARC growth and a power shift during these talks, ultimately, Pastrana controlled whether or not the government would negotiate with the FARC. Despite evidence that FARC growth did occur, it was a high-profile act of violence, not concerns over increasing strength from FARC service provision in the demilitarized zone, that led to the demise of the peace talks.

Concerns over service provision

Another way to distinguish between strategic stalling and rebel commitment problems is by looking at how rebels and governments treated service provision. If rebel service provision was generating fears of rebel growth and regrouping then the government should be expressing concern regarding rebel service provision. If rebels are stalling then rebel statements and actions should reveal continued service provision during negotiations, and evidence that they valued such provision. In the case of the PIRA, British comments regarding PIRA service provision, which was mostly justice and policing, are limited. However, in both internal memos (Hennessey 2015: 178–179) and parliamentary statements, the British government states that sustained cessation of violence includes ending “bombings, murders and kneecappings … [and] kangaroo courts” (TNA 1975b). However, the British government also assessed that PIRA activities were costing them Catholic support while support for the British had been increasing (TNA 1975c).²⁰

Even though it appears the British recognized that the PIRA needed the 1974–1975 ceasefires to regroup and regain civilian support, and recognized the possibility that the PIRA could come back stronger (Wilkin 1975; Hennessey 2015), the British still remained intent on trying to make the ceasefire last (TNA 1974). The PIRA's actions during the ceasefire, including reintroducing auxillary units believed to be for community policing (Wilkin 1975), suggest that they were using the ceasefire to repair the damage done to their Catholic support. The PIRA also stated that the British government was preventing them from protecting their people (Hennessey 2015: 197). This supports PIRA stalling. The British government,²¹ however, shows no concern that the PIRA's activities during the ceasefire would dangerously shift power in favor of the PIRA. Classified British Army internal reports outline PIRA activity during the ceasefire, which included training and acquiring weapons (Wilkin 1975). The British government also recorded PIRA violations of their ceasefire, and in a classified internal document, discuss using these publicly but note that “[i]t is easy enough to see the reason for not using [the PIRA violations] at all, notably the difficulty of reconciling them with the publicly stated reasons for continuing to release detainees during the past few months” (TNA 1975a: 2). Despite PIRA activity during the ceasefire, the British government continued releasing PIRA political prisoners. Therefore, PIRA negotiations did not break down owing to British government fears that PIRA service provision was increasing the capability and support of the group.

FARC's control of the demilitarized zone gave them free rein to provide services, which they did. Much like the British government, the Colombian government appears to have viewed the FARC's service provision only in the context of the military advantage that it could bring. When Pastrana declared the end of negotiations with the FARC, he added that, in addition to the increase in “terrorist acts” there were other activities like the FARC's infrastructure development (roads, bridges and an airstrip) which were for “illicit purposes” (Kline 2007: 120).²² The FARC issued a communique in response stating that they built infrastructure because the Colombian government failed to do so when it was in control of the territory and restated their commitment to the struggle of the people (Kline 2007: 121–122). In this case, the Colombian government did not appear concerned over FARC service provision, suggesting there was not a rebel commitment problem. The fact that the FARC provided services during the negotiations does, however, support the theory of FARC strategic stalling.

Rebel political parties

The final area where the rebel commitment problem and strategic stalling theories diverge is in the government's openness to the rebels participating in elections while providing services and remaining armed. If a rebel commitment problem is driving the negative correlation between service provision and successful negotiated settlements, then governments should be actively preventing rebel inclusion in the political process while rebels are providing services. Rebel service provision should increase their electoral prospects, which would in turn increase their bargaining position during negotiations. Strategic stalling, however, should produce the opposite: rebels, if allowed by the government, should engage in the political system and run in elections.

The PIRA's political party, Sinn Féin, was proscribed in 1956, but British authorities legalized it in May 1974, during the period of talks with the PIRA. The British government convened a Constitutional Convention, with elected members, in 1975 to discuss the political options for Northern Ireland. Sinn Féin was allowed to run in this election but chose not to (TNA 1975a: 6). Additionally, Sinn Féin had a policy of boycotting their seats and it was a number of years later, in 1983, when they discontinued their policy of abstention. This suggests that a rebel commitment problem was not why the talks with the PIRA broke down; however, Sinn Féin's reluctance to engage in the electoral process in Northern Ireland is not affirmation of PIRA stalling.

The FARC, however, provides better support for strategic stalling in regards to their political party. Their first political party, the Unión Patriótica (UP), was formed in 1984, following the Uribe Agreement. This agreement established a framework through which the FARC and the Colombian government would proceed to negotiate, and it had two important implications. First, the FARC was not expected to disarm. Second, the agreement paved the way for the FARC to enter mainstream politics (Dudley 2004: 46). Unfortunately, the UP became the target for the FARC's enemies. Alberto Rojas Poyu, a UP member, told Steven Dudley “Maybe the army wasn't able to finish off the guerrillas, but it was capable of finishing off the membership of the [UP] … They [the FARC and Communist Party] … made the UP an easy target for assassins” (Dudley 2004: 95). While the UP was targeted by paramilitaries and drug cartels, the Colombian government did not block UP electoral participation, and the UP won seats in 1986 and 1988 (Beittel 2015: 3).

The UP was also targeted by some within the Colombian security forces. Although this seems contradictory, it is important to remember that governments are not unitary actors. In the case of Colombia during the Pastrana presidency, there was contention between the Colombian presidency, who had the power to negotiate and sign an agreement, and the security forces. This mitigates concerns that the Colombian government allowed the UP to run in elections but then ordered the security forces to eliminate them. The power struggle between the Colombian presidency and the security forces was public and continual. In 1984, mere days before the Uribe Agreement was signed, General Landazábal gave an interview stating that there would be no ceasefire with the FARC. After the Uribe Agreement, General Landazábal began a media campaign to turn public support against the FARC (Dudley 2004: 39–43).²³ The Colombian government, however, still allowed the FARC's political party, the UP, to run in elections, which the UP did. Even though the UP was targeted for assassinations, the fact that they ran in elections supports strategic stalling.

Overall, evidence from the PIRA and the FARC cases on who ended negotiations, how the government reacted to rebel service provision and rebel involvement in elections, lends support for a stalling mechanism over a rebel commitment problem. Both the PIRA and FARC stalled negotiations by temporarily walking away, but remained open to dialogue with the government, which supports the stalling explanation. A rebel commitment problem would lead the government, not rebels, to end negotiations, which was not observed in the PIRA case. And, although Colombian President Pastrana did eventually end negotiations with the FARC, it occurred after a major incident of violence: hijacking a plane and kidnapping a senator. The FARC had spent three years providing services in the demilitarized zone but Pastrana attributed the end of the negotiations to the FARC's “terrorist” activity, which includes the military use of FARC infrastructure, rather than their mobilization of civilians through service provision. Additionally, throughout the peace talks, the FARC froze the process four separate times, but continued to state their willingness to continue negotiations with the government (Arenas 1990; Kline 2007: 121–122), which supports strategic stalling.

The lack of concern the British and Colombian governments had in regards to the PIRA and FARC service provision also supports stalling over a rebel commitment problem. In fact, the British acknowledged that the PIRA was using the ceasefires to regain support, but continued its policy of attempting to keep the ceasefire from breaking down by, for example, releasing detainees. The FARC and the PIRA, however, stated that they provided services to assist civilians. And, while the PIRA did not directly state this, the British government assessed that the PIRA's decision to declare a ceasefire was to regain Catholic support they had lost. Additionally, both groups were given the opportunity to run candidates in elections, which we would not expect to see if there was a rebel commitment problem. While PIRA's electoral behavior is not positive support for the stalling mechanism, it does not support the rebel commitment problem mechanism. The FARC's participation in elections does, however, support strategic stalling over a rebel commitment problem.

Conclusion

Rebels invest in service provision, diverting resources which could be used to improve their military capability, because they believe that service provision will benefit them. A burgeoning literature on rebel governance confirms this. Service provision wins civilian hearts and minds, which over time translates into collaboration (Kalyvas 2006), and can enhance rebel groups’ legitimacy and ability to organize civilians (Coggins 2015; Arjona 2014, 2016). Rebel institutional capacity is a source of strength, albeit different than military capacity. However, in civil war, which is an armed struggle over the right to govern, institutional capacity is also important. It may not directly enhance rebels’ fighting capacity, but it is directly beneficial in legitimizing their claims of governing, which can in turn increase support and lead to greater rebel capacity. Then, as the literature has suggested, rebels who provide services over time should be more likely to reach and successfully end a conflict with a negotiated settlement (Heger and Jung 2017). However, despite a strong a priori expectation of a positive relationship, the results here suggest that, on average, the relationship is actually negative.

I have argued here that there are two possible theoretical explanations why rebel service provision does not have a positive effect on ending civil conflict through negotiated settlements. The first is that the benefits of service provision are not immediate and consequently, rebels stall to increase the dividends of their investment. This incentivizes rebel stalling, particularly through negotiations. The second possible explanation is that service provision, which can increase civilian support, creates a rebel commitment problem, instilling government fears that rebels are using ceasefires and peace talks to regroup through their increased civilian support. Commitment problems are well established in the conflict literature (Powell 1999, 2006; Walter 2002); however, a survey of negotiations between the Colombian government and the FARC, and the British government and the PIRA, suggests that rebel commitment problems are unlikely to be causing the negative correlation revealed here. Rather, the evidence suggests that rebels will stall through negotiations to reap the benefits of their service provision.

The implications of this paper, however, reach beyond theory. Ending a conflict with a negotiated settlement is incredibly difficult, but the human cost of continued conflict is much steeper. Policy-makers and negotiators need to understand how different forms of rebel capacity, particularly institutional forms of capacity, create different incentives. Expectations of future benefits from an investment in service provision will lead to rebel stalling in negotiations. This, however, does not necessarily mean that rebels will never be willing to sign an agreement. It simply means that rebels may expect to get a better deal in the future, so they will delay signing a peace agreement. Mediators and policy-makers, therefore, need to consider how to make a settlement more attractive today to offset the expected future gain from rebel service provision, and seriously consider the state of government service provision. State investment in services can undermine rebel support, thereby decreasing rebel expectations about the gains from their service provision.

Supplemental Material

sj-pdf-1-cmp-10.1177_07388942221147508 - Supplemental material for Rebel institutions and negotiated peace

Supplemental material, sj-pdf-1-cmp-10.1177_07388942221147508 for Rebel institutions and negotiated peace by Karen Albert in Conflict Management and Peace Science

Supplemental Material

sj-r-2-cmp-10.1177_07388942221147508 - Supplemental material for Rebel institutions and negotiated peace

Supplemental material, sj-r-2-cmp-10.1177_07388942221147508 for Rebel institutions and negotiated peace by Karen Albert in Conflict Management and Peace Science

Supplemental Material

sj-csv-3-cmp-10.1177_07388942221147508 - Supplemental material for Rebel institutions and negotiated peace

Supplemental material, sj-csv-3-cmp-10.1177_07388942221147508 for Rebel institutions and negotiated peace by Karen Albert in Conflict Management and Peace Science

Supplemental Material

sj-csv-4-cmp-10.1177_07388942221147508 - Supplemental material for Rebel institutions and negotiated peace

Supplemental material, sj-csv-4-cmp-10.1177_07388942221147508 for Rebel institutions and negotiated peace by Karen Albert in Conflict Management and Peace Science

Supplemental Material

sj-csv-5-cmp-10.1177_07388942221147508 - Supplemental material for Rebel institutions and negotiated peace

Supplemental material, sj-csv-5-cmp-10.1177_07388942221147508 for Rebel institutions and negotiated peace by Karen Albert in Conflict Management and Peace Science

Supplemental Material

sj-csv-6-cmp-10.1177_07388942221147508 - Supplemental material for Rebel institutions and negotiated peace

Supplemental material, sj-csv-6-cmp-10.1177_07388942221147508 for Rebel institutions and negotiated peace by Karen Albert in Conflict Management and Peace Science

Footnotes

Acknowledgements

The author would like to extend special thanks to Hein Goemans, Jack Paine, Bethany Lacina, Jason Lyall and Gretchen Helmke. Additional thanks are due to Emily Vanmeter, Olga Gasparyan, Xiaoyan “Christy” Qiu, Cyanne Loyle, Danielle Jung, Megan Stewart, Doug Lemke and the anonymous reviewers for all their thoughtful and helpful comments.

Funding

This research was supported by grants from the Peter D Watson Center for Conflict and Cooperation at the University of Rochester.

ORCID iD

Karen Albert

Supplemental material

Supplemental material for this article is available online.

Notes

References

Albert

(2022) What is rebel governance? introducing a new dataset on rebel institutions, 1945–2012. Journal of Peace Research 54(4): 622–630.

Anderson

(2016) Imagined Communities. Revised edition. London: Verso.

Arenas

(1990) Paz, Amigos y Enemigos. Colombia: La Abeja Negra.

Arjona

(2014) Wartime institutions: A research agenda. Journal of Conflict Resolution 58(8): 1360–1389.

Arjona

(2016) Rebelocracy: Social Order in the Colombian Civil War. New York: Cambridge University Press.

Balcells

Kalyvas

(2015) Revolutionary Rebels and the Marxist Paradox. Working paper. cpd.berkeley.edu/wp-content/uploads/2015/04/Marxlns_4_15.pdf.

Beittel

(2015) Peace Talks in Colombia. Technical report, Washington, DC: Congressional Research Service.

Bolt

Inklaar

de Jong

, et al. (2018) Rebasing “Maddison”: New income comparisons and the shape of long-run economic development. GGDC Research Memorandum, 174: 1–41.

Braithwaite

Cunningham

(2019) When Organizations Rebel: Introducing the Foundations of Rebel Group Emergence Dataset. International Studies Quarterly 64(1): 183–193.

10.

Braun

(2007) “Que haiga paz!” The cultural contexts of conflict in Colombia. In: Welna

Gallon

(eds) Peace, Democracy and Human Rights in Colombia. Notre Dame, IN: University of Notre Dame, pp.23–59.

11.

Clayton

(2013) Relative rebel strength and the onset and outcome of civil war mediation. Journal of Peace Research 50(5): 609–622.

12.

Coggins

(2015) Rebel diplomacy: Theorizing violent non-state actors’ strategic use of talk. In: Arjona

Kasfir

Mampilly

(eds) Rebel Governance in Civil War. New York: Cambridge University Press, pp.98–118.

13.

Cunningham

(2011) Barriers to Peace in Civil Wars. Cambridge: Cambridge University Press.

14.

Cunningham

Gleditsch

Salehyan

(2013) Non-state actors in civil wars: A new dataset. Conflict Management and Peace Science 30(5): 516–531. http://cmp.sagepub.com/cgi/doi/10.1177/0738894213499673.

15.

De Mesquita

Morrow

Siverson

, et al. (2004) Testing novel implications from the selectorate theory of war. World Politics 56(3): 363–388.

16.

Driscoll

(2012) Commitment problems or bidding wars? Rebel fragmentation as peace building. Journal of Conflict Resolution 56(1): 118–149.

17.

Dudley

(2004) Walking Ghosts: Murder and Guerrilla Politics in Colombia. New York: Routledge.

18.

Fearon

(2004) Why do some civil wars last so much longer than others? Journal of Peace Research 41(3): 275–301. URL http://jpr.sagepub.com/cgi/doi/10.1177/0022343304043770.

19.

Fearon

Laitin

(2008) Civil war termination. Working paper. web.stanford.edu/group/fearon-research/cgi-bin/wordpress/wp-content/uploads/2013/10/civil-war-termination.pdf.

20.

Fuhrmann

Tir

(2009) Territorial dimensions of enduring internal rivalries. Conflict Management and Peace Science 26(4): 307–329. http://cmp.sagepub.com/cgi/doi/10.1177/0738894209106478.

21.

Fukuyama

(2013) What is governance? Governance 26(3): 347–368.

22.

Glassmyer

Sambanis

(2008) Rebel–military integration and civil war termination. Journal of Peace Research 45(3): 365–384.

23.

Heger

Jung

(2017) Negotiating with rebels: The effect of rebel service provision on conflict negotiations. Journal of Conflict Resolution 61(6): 1203–1229. https://journals-sagepub-com.web.bisu.edu.cn/doi/10.1177/0022002715603451.

24.

Hennessey

(2015) The First Northern Ireland Peace Process. New York: Palgrave Macmillan.

25.

Huang

(2016) The Wartime Origins of Postwar Democratization: Civil War, Rebel Governance, and Political Regimes. New York: Cambridge University Press.

26.

Isacson

Rodriguez

(2009) Origins, evolution, and lessons of the Colombian peace movement. In: Bouvier

(ed.) Colombia: Building Peace in a Time of War2. Washington, DC: United States Institute of Peace, pp.19–37.

27.

Jung

(2017) Determinants of rebel and terrorist organizational structure and the effects of counter-insurgency: Population dynamics and organizational ecology in rebellion. Working paper.

28.

Kalyvas

(2006) The Logic of Violence in Civil War. New York: Cambridge University Press.

29.

Kalyvas

(2015) Rebel governance during the Greek civil war, 1942–1949. In: Arjona

Kasfir

Mampilly

(eds) Rebel Governance in Civil War. New York: Cambridge University Press, pp.119–137.

30.

Kasfir

(2015) Rebel governance—constructing a field of inquiry: Definitions, scope, patterns, order, causes. In: Arjona

Kasfir

Mampilly

(eds) Rebel Governance in Civil War. New York: Cambridge University Press, pp.21–46.

31.

Kilcullen

(2011) The Accidental Guerrilla. Oxford: Oxford University Press.

32.

Kline

(2007) Chronicle of a Failure Foretold. Tuscaloosa, AL: University of Alabama Press.

33.

Kreutz

(2010) How and when armed conflicts end: Introducing the UCDP conflict termination dataset. Journal of Peace Research 47(2): 243–250.

34.

Lacina

Gleditsch

(2005) Monitoring trends in global combat: A new dataset of battle deaths. European Journal of Population 21(2-3): 145–166.

35.

Lafta

Cetorelli

Burnham

(2018) Living in Mosul during the time of ISIS and the military liberation: Results from a 40-cluster household survey. Conflict and Health 12(1): 4–11.

36.

Ledwidge

(2017) Rebel Law: Insurgents, Courts and Justice in Modern Conflict. London: Hurst.

37.

Licklider

(1995) The consequences of negotiated settlements in civil wars, 1945–1993. The American Political Science Review 89(3): 681–690.

38.

Mampilly

(2011) Rebel Rulers: Insurgent Governance and Civilian Life During War. Ithaca, NY: Cornell University Press.

39.

Mampilly

(2015) Performing the nation-state: Rebel governance and symbolic processes. In: Arjona

Kasfir

Mampilly

(eds) Rebel Governance in Civil War. New York: Cambridge University Press, pp.74–97.

40.

Mason

Fett

(1996) How civil wars End: A rational choice approach. American Journal of Political Science 40(4): 546–568.

41.

Mattes

Savun

(2009) Fostering peace after civil war: Commitment problems and agreement design. International Studies Quarterly 53(3): 737–759.

42.

Nasi

(2009) Colombia’s peace processes, 1982–2002. In: Bouvier

(ed) Colombia: Building Peace in a Time of War. Washington, DC: United States Institute of Peace, pp. 39–64.

43.

Popkin

(1979) The Rational Peasant. Berkley, CA: University of California Press.

44.

Powell

(1999) In the Shadow of Power. Princeton, NJ: Princeton University Press.

45.

Powell

(2006) War as a commitment problem. International Organization 60(1): 169–203.

46.

Regan

Michael Greig

(2008) When do they say yes? An analysis of the willingness to offer and accept mediation in civil wars. International Studies Quarterly 52(4): 759–781.

47.

Stewart

(2016) Civil war as state building: The determinants of insurgent public goods provision. Dissertation, Georgetown University.

48.

Stewart

(2018) Civil war as state building: Strategic governance in civil war. International Organization 72(1): 205–226.

49.

Stewart

(2020) Rebel governance: Military boon or military bust? Conflict Management and Peace Science 37(1): 16–38.

50.

TNA (1974) CJ 4/1225 A Plan for a Cease Fire JB Bourh. London: The National Archives of the UK.

51.

TNA (1975a) FCO 87/460 Letter from Northern Ireland Office to G W Harding. London: The National Archives of the UK.

52.

TNA (1975b) FCO 87/460 Parliamentary Statement: 11 February 1975. London: The National Archives of the UK.

53.

TNA (1975c) FCO 87/460 Visit of Frank Cooper by Arthur Galsworthy. London: The National Archives of the UK.

54.

Toft

(2010) Securing the Peace: The Durable Settlement of Civil Wars. Princeton, NJ: Princeton University Press.

55.

Uzonyi

Hanania

(2017) Government-sponsored mass killing and civil war reoccurrence. International Studies Quarterly 61(3): 677–689.

56.

Wagstaff

Jung

(2017) Competing for constituents: Trends in terrorist service provision. Terrorism and Political Violence 6553: 1–32.

57.

Walter

(2002) Committing to Peace: The Successful Settlement of Civil Wars. Princeton, NJ: Princeton University Press.

58.

Walter

(2004) Does conflict beget conflict? Explaining recurring civil war. Journal of Peace Research 41(3): 371–388.

59.

Walter

(2009) Bargaining failures and civil war. Annual Review of Political Science 12(1): 243–261.

60.

Weber

(1976) Peasants into Frenchmen: The Modernization of Rural France, 1870–1914. Stanford, CA: Stanford University Press.

61.

Weinstein

(2007) Inside Rebellion: The Politics of Insurgent Violence. New York: Cambridge University Press.

62.

Wilkin

CRC

(1975) Provisional IRA Reorganisation during the Ceasefire. Technical report. Lisburn: British Forces Headquarters.

63.

Zartman

(1989) Ripe for Resolution: Conflict and Intervention in Africa. Oxford, England: Oxford University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB

0.11 MB

0.10 MB

14.01 MB

0.24 MB

0.43 MB