Abstract
Post-conflict justice is an integral component in maintaining stability and building peace in the aftermath of civil conflict. Despite its instrumental function, scholars routinely find that policymakers’ choice of justice is shaped by the structural conditions of the post-conflict environment, with outright victories leading to retributive forms of justice and negotiated outcomes yielding restorative forms of justice. However, existing literature conflates ceasefires and peace agreements into a single phenomenon, thereby overlooking the independent effects of each outcome. Leveraging the dual sovereignty framework, this article argues the conflation of negotiated outcomes is problematic because peace agreements and ceasefires generate different post-conflict environments. Relative to ceasefires, peace agreements lead to a reduction in the degree of dual sovereignty because they resolve a conflict’s incompatibility, thereby encouraging efforts to move society beyond war through restorative forms of justice. Due to the persistent threat of recurrent war generated by high levels of dual sovereignty, policymakers following ceasefires will be inclined to pursue retributive forms of justice that may target political opponents or potential defectors to bolster organizational strength. Statistical analyses confirm the underlying expectation that ceasefires and peace agreements yield different post-conflict justice outcomes. Peace agreements, relative to ceasefires, are more likely to be followed by the implementation of amnesties and reparations, whereas ceasefires exhibit a greater probability of yielding purges in the post-conflict environment.
Keywords
On 4 October 1992, the warring factions of Renamo and Frelimo signed the Rome General Peace Accords, bringing an end to a decades-long war in Mozambique that resulted in the deaths of approximately 145,000 combatants and nearly one million civilians (Domingues, 2011). Rather than seek accountability for individuals responsible for wartime abuses, Mozambique’s leaders sought to implement restorative justice measures, offering property reparations to refugees and internally displaced persons while concurrently offering amnesties to the war’s participants (Cobban, 2007; Binningsbø et al., 2012). A reversal of this dynamic played out in Azerbaijan in 1994 when the state paused hostilities with the Republic of Nagorno-Karabakh through a ceasefire agreement (Wagner & Druckman, 2017). Although the conflict ended through a negotiated settlement, neither side implemented programs for reparations or amnesties; rather, Azerbaijan’s government pursued charges of treason against a former defense minister and military commander for the surrender of territory during the conflict (Cornell, 2011; De Waal, 2013). Similarly, the post-conflict periods following the civil wars in Nepal and Moldova, conflicts that ended with ceasefires in 1964 and 1992, respectively, were succeeded by trials rather than efforts at restorative justice (Binningsbø et al., 2012).
In these cases, the post-conflict governments faced a crucial question concerning the means by which the state and society would deal with wartime atrocities. Although each conflict ended in a negotiated outcome, the post-conflict justice outcomes diverge considerably across conflicts. These cases raise important questions concerning our understanding of post-conflict justice implementation. If the conflicts ended through negotiations between the warring actors, why did the resultant governments implement different post-conflict justice strategies? Why do some decisionmakers in post-conflict states pursue accountability through retributive forms of justice, while other policymakers seek programs centered around restorative justice mechanisms?
A recurring answer in the literature stems from the recognition that the end of a conflict shapes what is possible in the post-conflict environment (Flores & Nooruddin, 2009; DeTommaso, Schulz & Lem, 2017). Recent research on the effect of conflict termination and post-conflict justice shows how military victories and negotiated outcomes may impact the implementation of different forms of justice (Binningsbø et al., 2012; Reiter, Olsen & Payne, 2013; DeTommaso, Schulz & Lem, 2017; Kim & Hong, 2019). Within this literature, multiple studies identify negotiated outcomes as mechanisms leading to the implementation of restorative forms of justice via truth commissions, amnesties, and reparations over trials, exiles, or purges (Binningsbø et al., 2012; DeTommaso, Schulz & Lem, 2017; Kim & Hong, 2019). However, in identifying this relationship, all of these studies treat ceasefires and peace agreements as a single phenomenon, thereby overlooking the disparate effects of different types of negotiated instruments.
Departing from this conflation of concepts, I argue that ceasefires and peace agreements create two distinct types of post-conflict environments that impose different constraints and offer divergent incentives to policymakers who are deciding whether to implement post-conflict justice. Using the dual sovereignty framework as a guide, I expect that post-conflict environments marked with low levels of dual sovereignty provide policymakers with greater discretion and incentives to pursue restorative justice mechanisms as compared to periods of high dual sovereignty, which encourage the implementation of retributive justice measures in an effort to consolidate power in the event of renewed conflict. By resolving the critical incompatibilities, I argue peace agreements are more likely than ceasefires to reduce levels of dual sovereignty, ameliorating the willingness and capacity of rebel groups to engage in further conflict. Thus, I expect to observe restorative forms of justice being adopted following peace agreements. Conversely, in the wake of ceasefires, policymakers intent on maximizing power given the persistent threat of future conflict will be more inclined to implement retributive justice mechanisms. Leveraging data from the Post-Conflict Justice dataset (Binningsbø et al., 2012), I find modest support for ceasefires being more likely to lead to the implementation of purges, indicating an aversion of restoration in the face of potential recurrent conflict. Conversely, the empirical analysis provides robust evidence revealing that peace agreements are far more likely than ceasefires to give way to the implementation of amnesties and reparations in the post-conflict period.
This article contributes to the literature’s knowledge of post-conflict justice and peacebuilding in three ways. First, it calls into question the treatment of all negotiated outcomes as being equivalent in the context of understanding post-conflict justice implementation. Second, my theory shows how different post-conflict environments can incentivize the use of certain justice mechanisms to consolidate power when the risk of conflict recurrence is high. Lastly, post-conflict justice has been shown to exert positive effects on a number of different outcomes, including peacebuilding (Cobban, 2007; Druckman & Albin, 2011; Albin & Druckman, 2012; Wagner & Druckman, 2017; Druckman & Wagner, 2019), economic development (Appel & Loyle, 2012), democratization (Wiebelhaus-Brahm, 2010; Olsen, Payne & Reiter, 2010), and human rights (Sikkink, 2011; Stewart & Wiebelhaus-Brahm, 2017). Given the importance of post-conflict justice in these different areas, a circumspect understanding of the forms these justice regimes are likely to take, given the way in which a conflict ends, can aid scholars and policymakers alike in achieving a more sustainable peace.
Seeking justice in the post-conflict environment
Post-conflict justice typology
Although post-conflict justice can stem from normative or functional considerations, policymakers’ preferences are likely constrained by other forms of volatility and the degree to which recurrent war still poses a threat (Flores & Nooruddin, 2012; Sikkink & Kim, 2013; Walter, 2015; Hegre & Nygard, 2015; Loyle & Davenport, 2016; Girelli, 2017). Previous literature identifies a conflict’s end as a salient factor in shaping the constraints imposed on a state in the post-conflict environment (Binningsbø et al., 2012; DeTommaso, Schulz & Lem, 2017; Kim & Hong, 2019). Because post-conflict justice implementation is necessarily a decision relegated to the aftermath of war, policymakers are likely to be constrained when deciding whether to implement post-conflict justice.
Recent scholarship has examined how conflict termination impacts the implementation of post-conflict justice. Binningsbø et al. (2012) find that concessions in the form of restorative justice mechanisms are more likely to be offered by policymakers following a negotiated outcome. 1 DeTommaso, Schulz & Lem (2017) contend that post-conflict justice mechanisms should not be considered in isolation; rather, many forms of post-conflict justice can occur simultaneously, necessitating consideration of post-conflict justice as a regime rather than as individual mechanisms. Their regression analysis finds a positive and statistically significant correlation between negotiated outcomes and restorative justice regimes. Most recently, Kim & Hong (2019) describe how post-conflict justice implementation stems from political motivations of policymakers, who locate opportunities for the pursuit of certain types of post-conflict justice and encounter constraints toward others depending on the process by which a conflict ends. In particular, their empirical analysis finds domestic trials to be more likely following outright victories whereas restorative forms of justice follow in the aftermath of a negotiated outcome.
While each of these studies improves our understanding of post-conflict justice by highlighting the critical link between conflict and justice, the analysis in each study is limited in that they all collapse ceasefires and peace agreements into a single concept, thereby treating all negotiated outcomes as the same. In contrast to this conflation of negotiated outcomes, the existing literature on civil war termination identifies numerous aspects in which ceasefires and peace agreements differ, showing how variation across agreements can lead to different outcomes in the post-conflict environment (Hartzell & Hoddie, 2003; Gates et al., 2016; Wagner & Druckman, 2017). I follow this tradition in arguing that peace agreements and ceasefires differ because the former incorporates a resolution to a conflict’s incompatibility, which increases the likelihood of demobilization, demilitarization, and integration provisions being attached, whereas ceasefires only place a hold on hostilities. These differences necessitate severing ceasefires from peace agreements to evaluate how they lead to potentially different post-conflict justice outcomes.
Negotiated outcomes, dual sovereignty, and post-conflict justice
The literature has long recognized a difference between ceasefires and peace agreements, both at the interstate and intrastate levels (Walter, 2002; Hartzell & Hoddie, 2003; Fortna, 2004; Werner & Yuen, 2005; Fazal, 2013; Wagner & Druckman, 2017). This separation often turns on two key characteristics of negotiated outcomes: (1) whether the agreement addresses a conflict’s incompatibility and (2) the degree of variation in the specific terms of the agreement. The incompatibility is at the heart of what drives a rebel group in its campaign against the government. It stokes the willingness of dissidents to continue to take up arms in pursuit of a specific goal (Gurr, 1970; Mason et al., 2011). Absent a resolution to the underlying incompatibility by way of a peace agreement, rebels likely maintain some motivation to consider armed conflict until victory is achieved, which can also be leveraged by rebel entrepreneurs to re-mobilize support for the movement and renew the conflict. Conversely, a focus on agreement provisions can take many forms, with studies evaluating the variation in power-sharing provisions concerning political power, military affairs, territorial considerations, economic development, and security sector reforms (Hartzell & Hoddie, 2003; Toft, 2010). Within these broad categories, scholars have also analyzed the influences of disarmament, demobilization, and integration programs (Harbom, Hogbladh & Wallensteen, 2006). However, for a few reasons, the salient factor distinguishing peace agreements from ceasefires is the resolution of the incompatibility separating the warring parties.
First, both ceasefires and peace agreements have been observed to contain different forms of conflict regulatory provisions, including disarmament and demobilization requirements (Hartzell & Hoddie, 2003; Kreutz, 2010). For example, 37 of the 57 ceasefires recorded in the UCDP Conflict Termination dataset included provisions to manage the underlying conflict (Kreutz, 2010). Yet, it is also possible that neither type of agreement incorporates these provisions (Joshi, 2010). Ultimately, variation across negotiated outcomes rooted in conflict management is not a sufficient basis to capture adequately the differences between ceasefires and peace agreements. 2
Conversely, the resolution of the incompatibility is key to understanding how peace agreements and ceasefires differ. Common definitional conceptions separate agreements based upon the instrument’s treatment of the conflict’s incompatibility (Kreutz, 2010). Moreover, the definitional components of these agreements are reified in both the literature and policymakers’ discussion of the functional aspect of ceasefires, that is, ceasefires are viewed as a means to ease tensions and clear a space for future talks aimed at resolving the underlying conflict (Fortna, 2004; Werner & Yuen, 2005; Kolås, 2011; Akebo, 2016). For example, both the 1994 and 2002 ceasefires involving the Liberation Tigers of Tamil Eelam (LTTE) in Sri Lanka and the 1994 ceasefire agreed to by the Provisional Irish Republican Army were reached in the hopes that the cessation of hostilities would lead to future peace talks (Peiris, 2009; Akebo, 2016; Dixon, 2018). These examples reflect how ceasefires can represent a decision to postpone future military ventures in order to facilitate subsequent negotiations toward a comprehensive peace agreement in which a conflict’s incompatibility can be addressed. Critically, though, an attempt to resolve the issues giving rise to armed rebellion in the first instance is absent from ceasefires, thereby leaving unresolved issues separating the belligerent parties.
Even if certain regulatory provisions in ceasefires mirror those in peace agreements, the mobilizing force of the incompatibility remains following a ceasefire (Kreutz, 2010). 3 Due to the failure to resolve the issues that mobilized armies to begin hostilities, the likelihood of renewed conflict continues to pervade the conflict-state’s atmosphere. This fraught context leads to vastly different post-conflict environments that, in turn, bear on the decisionmaking of post-conflict policymakers.
Linking negotiated outcomes and dual sovereignty
Considerable literature links the threat and occurrence of conflict recurrence to structural conditions in the post-conflict period (Flores & Nooruddin, 2012; Hegre & Nygård, 2015; Walter, 2015). Given this background, we might expect the structural conditions surrounding the implementation of post-conflict justice to be linked to the propensity of states to devolve into renewed conflict, which is critical given these divergent expectations for the resolution of conflict incompatibilities across peace agreements and ceasefires. Quinn, Mason & Gurses (2007) leverage Tilly’s (1978) dual sovereignty framework to offer a theoretical lens through which we can understand conditions likely to produce recurrent civil war, specifically related to the persistent challenge to the state’s authority and monopoly on the use of force.
Although considerable work suggests that negotiated outcomes maintain high levels of dual sovereignty, rendering conflict recurrence more likely, other studies suggest that certain agreement provisions can reduce the likelihood of recurrent conflict (Licklider, 1995; Walter, 1997; Hartzell & Hoddie, 2003; Toft, 2010). For example, agreements with third-party guarantees of enforcement (Walter, 1997, 2004), power-sharing provisions centered on military, political, economic, territorial, and security issues (Hartzell & Hoddie, 2003; Toft, 2010; Wagner & Druckman, 2017; Druckman & Wagner, 2019), and provisions imposing constraints on state power (Gates et al., 2016) reduce the likelihood of recurrent conflict and foster peace. However, these features of negotiated outcomes address only one aspect of dual sovereignty and cannot distinguish between ceasefires and peace agreements given that conflict management can be incorporated into either type of agreement.
Reducing the organizational capacity of previous conflict belligerents is important, yet it does little to impact the resolve with which parties entered into conflict. An integral source of continued support for dissidents in the postwar period is the persistence (and perhaps exacerbation) of conditions linked to the initial incompatibility motivating conflict (Gurr, 1970; Mason et al., 2011). Opposition groups will be more likely to derive support from aggrieved individuals if the incompatibility remains an open question; therefore, in contrast to ceasefires, peace agreements that address and resolve the incompatibility will diminish the ability of former combatants to generate support for their respective causes. Thus, agreements with resolutions to the underlying incompatibility will reduce the degree of dual sovereignty in the post-conflict environment. Coupled with mechanisms of conflict management, peace agreements should do more to reduce dual sovereignty than ceasefires. 4
Implementing post-conflict justice under conditions of dual sovereignty
The degree to which dual sovereignty exists in the aftermath of war imposes different constraints on policymakers while simultaneously shifting incentive structures that encourage the implementation of different post-conflict strategies. Importantly, the confluence of constraints and incentives for renewed war explains the differential implementation of post-conflict justice mechanisms in the wake of different types of negotiated outcomes.
As previously noted, ceasefires tend to merely return the gunpowder to the keg without removing the threat of a spark. In terms of constraints, the postwar government will be less inclined toward implementing restorative justice mechanisms because they either signal a weakness in the resolve of the state to continue fighting or reflect a reduction in the state’s capacity to defeat the rebels (Dancy, 2018). Alongside this constraint, both rebels and government actors will be incentivized to turn to retributive forms of justice as a way to identify potential defectors and maximize military capacity. That is, each side in the conflict must maintain military readiness and organizational cohesion in order to be ready in the event war is reinitiated. Thus, government and rebel elites will implement measures to inspire or coerce loyalty while removing dissent. These dynamics can be observed throughout the course of the Sri Lankan civil war. During the conflict, the warring parties reached ceasefires in 1994 and 2002. Although the 1994 agreement initiated a series of talks between adversaries, multiple violations of the ceasefire were lodged by one side against the other, including allegations of continued efforts to restore and bolster the military capacity of the warring forces (Akebo, 2016). The 2002 ceasefire proved equally unsuccessful, with numerous violations alleged that included the killing of political opponents of the LTTE (Peiris, 2009; Akebo, 2016).
The context following a ceasefire is considerably different from what tends to follow from a peace agreement. Since negotiators pursue peace agreements in the hope of resolving salient issues to enact a lasting peace, the post-conflict environment encourages (or may even require) elites to compromise on the underlying incompatibility, which will also incline elites toward strategies aimed at reducing the military capacity of belligerent parties. The resolution of the incompatibility coupled with the demobilization of the warring parties diminishes the need to prepare for future war, which would include the use of retributive justice to deter defections. Furthermore, rebel leaders are only likely to support these agreements if Dual sovereignty, conflict termination, and post-conflict justice
Combining the incentives and constraints imposed by the post-conflict period, as shaped by different conflict outcomes, Figure 1 displays the expectations to be derived from the reasoning previously set forth. 6
Distribution of post-conflict justice mechanisms and conflict termination
Based upon this theoretical framework, I propose two hypotheses:
Hypothesis 1: Civil wars ending with ceasefires are more likely to be followed by the implementation of retributive forms of post-conflict justice, especially as compared to those ending with a peace agreement. Hypothesis 2: Civil wars ending with peace agreements are more likely to be followed by restorative forms of post-conflict justice, especially as compared to those ending with ceasefires.
Research design
Post-conflict states have several available options in crafting a post-conflict justice regime (Olsen, Payne & Reiter, 2010; Binningsbø et al., 2012). The theoretical expectations outlined herein suggest a differential impact of political factors on a state’s selection of each form of post-conflict justice, stemming from different power dynamics in post-conflict environments linked to negotiated outcomes. I test the aforementioned hypotheses using data derived from the Post-Conflict Justice (PCJ) dataset generated by Binningsbø et al. (2012) on four types of post-conflict justice for civil wars occurring between 1946 and 2005: trials, purges, amnesties, and reparations. 7 Although linked with a civil conflict, each unit of analysis is the five-year, post-conflict period, thereby providing data on the implementation of post-conflict justice that occurred after a conflict’s termination. Each indicator is binary, that is, the data take the form of a 1 for the implementation of the particular mechanism in the post-conflict period and a 0 otherwise.
The main variables of interest turn on different forms of conflict termination. I leverage the UCDP Conflict Termination (CT) dataset to operationalize conflict outcomes, which provides data for civil wars ending with peace agreements, ceasefires, one-sided victories, and instances of low activity (Kreutz, 2010). 8 The empirical analysis in the next section includes separate measures for ceasefires, outright victories, and low activity outcomes with peace agreements set as the baseline category. Table II provides descriptive data concerning the distribution of different forms of implemented post-conflict justice mechanisms across the four types of conflict outcomes.
Although ceasefires and peace agreements are followed, in part, by the implementation of each of the four forms of justice in the sample, it is clear from the data that post-conflict periods succeeding peace agreements are more likely to observe the implementation of restorative justice mechanisms, although the descriptive data provide little insight on the propensity of either outcome to create an environment for retributive forms of justice. Notably, outright victories appear far more likely to yield all forms of post-conflict justice, providing preliminary evidence to suggest that outright victors, operating under conditions of low dual sovereignty, have greater discretion to implement justice mechanisms to achieve their particular goals.
Despite preliminary evidence for the second hypothesis, a robust analysis of the relationship between negotiated outcomes and post-conflict justice necessitates a consideration of other conditions that could be related to termination and post-conflict justice. Given the binary nature of each dependent variable, I use logistic regression and include robust standard errors clustered on each conflict-state as a means to address heteroskedasticity. To account for possible confounding factors, the resultant regression analysis includes a number of control variables identified in the literature.
One group of potential confounders can be derived from the previous conflict’s dynamics. Reiter, Olsen & Payne (2013) identify both the duration and severity of a previous conflict as determinants of post-conflict justice implementation; therefore, I include measures for each characteristic. In terms of a conflict’s duration, I calculate the number of days each conflict in the PCJ dataset lasts and take the natural log of the resulting calculation for each observation. In terms of a conflict’s severity, I leverage two separate measures: one accounting for the degree of hostilities between combatants and a second measure intended to capture the violence perpetrated against civilian populations in the midst of war. The first measure is obtained by calculating the number of battle-related deaths occurring during each civil war and taking the natural log of each calculation per observation. Similar to the data for conflict duration, the numbers of battle-deaths per conflict are obtained from the PCJ dataset. The data for the second measure of conflict severity are taken from the Political Instability Task Force dataset that codes whether a government or non-state actor engaged in a campaign of genocide or politicide during a conflict (Marshall, Gurr & Harff, 2018). I dichotomize this variable to indicate whether a civil war involved any instance of either genocide or politicide.
Alongside conflict dynamics, I also include several control variables capturing characteristics of conflict actors. First, I account for the material capabilities of each side in the conflict using data on the strength of rebel groups found in the UCDP’s Non-State Actor data (Cunningham, Gleditsch & Salehyan, 2009). I account for whether international actors became involved in the conflict using data from Rustad & Binningsbø (2012) regarding the operation of United Nations peacekeeping missions in post-conflict states, with a 1 indicating the presence of a peacekeeping force and a 0 otherwise. Lastly, I include a count of the number of mediation attempts undertaken throughout the duration of each conflict using data from the Civil Wars Mediation dataset (DeRouen, Bercovitch & Pospieszna, 2011; DeRouen & Bercovitch, 2012). Looking to the post-conflict period, previous literature describes multiple instances where post-conflict elites implement multiple forms of justice (Cobban, 2007; Binningsbø et al., 2012; DeTommaso, Schultz & Lem, 2017); therefore, I include a control variable for the total number of justice mechanisms implemented following each conflict in the dataset, and the variable ranges from an absence of justice to four total mechanisms.
The last group of control variables accounts for state-level characteristics that may influence the willingness and ability of officials to implement post-conflict justice. Democracies are often viewed as more willing to resolve political disputes and controversies through less violent and less retributive means while also imposing constraints on the executive that are often absent in non-democracies (Mitchell, 2002; Bayer, 2010). To capture facets of democracy, I account for each conflict-state’s regime type using the Polity index, which is a 21-point scale ranging from –10 (stable autocracy) to 10 (stable democracy) (Marshall, Gurr & Jaggers, 2018). In the empirical analyses, I include the index variable to account for the full range of potential regime characteristics attendant to the governing structures of different states in the sample. 9 Alongside a state’s institutional structure, previous research has highlighted the monetary cost associated with implementing post-conflict justice; therefore, I include a measure for each conflict-state’s GDP per capita, the data for which are taken from the Maddison Project Database (Bolt et al., 2018). The final two control variables account for sociodemographic characteristics of each conflict state, the first measuring the overall population of the state, the data for which are also derived from the Maddison Project Database (Bolt et al., 2018) and the second accounting for ethnic diversity, using data from the Ethnic Power Relations dataset (Wimmer, Cederman & Min, 2009).
Analysis
As noted above, the analysis tests the effects of different negotiated outcomes on four types of post-conflict justice using logistic regression analysis. Pursuant to the two hypotheses, I anticipate the coefficients for ceasefires to be positive for trials and purges while expecting the ceasefire coefficients to be negative in relation to amnesties and reparations. Table III presents the results for the four types of justice.
Conflict termination and post-conflict justice mechanisms
Clustered standard errors are in parentheses. Peace agreements are the reference category.
**p < 0.01, *p < 0.05, † p < 0.1.
Change in effects shifting between negotiated outcomes
The effects in Table IV were calculated for anocratic states while holding discrete variables at their modal values and continuous variables at their means.

Distribution of post-conflict justice regimes
Alongside the analysis concerning the main independent variables of interest, a number of important associations between post-conflict justice and some control variables are uncovered. First, most of the control variables across the four models fail to achieve statistical significance. However, conflict dynamics, in terms of both severity and duration, are related to the implementation of post-conflict justice, albeit at odds with one another across the different models. These results further indicate how influential the dynamics of the conflict can be in shaping a post-conflict environment that impresses certain incentives and imposes various constraints on policymakers.
The second interesting finding pertains to the effect of concurrent justice processes implemented in the post-conflict justice period. In every model reflected in Table III, the Total post-conflict justice mechanisms variable is positive and statistically significant. Thus, the implementation of other forms of justice in the post-conflict environment increases the likelihood that each of the four individual types of justice will be implemented as well. Both descriptively and anecdotally, the existence of multi-justice regimes is not an anomaly. Of the 150 post-conflict environments in which a form of post-conflict justice was adopted, approximately 25% experienced two justice processes and roughly 13% observed three processes. These empirics argue for a closer examination of the effect of negotiated outcomes and post-conflict justice regimes. 10
Conflict termination and post-conflict justice regimes
Clustered standard errors are in parentheses. Peace agreements are the reference category.
**p < 0.01, *p < 0.05, † p < 0.1.
In the analysis that follows, the absence of post-conflict justice is used as the baseline category. Given the operationalization of the dependent variable as a measure divided into four categories that do not follow a logical ordering, multinomial logistic regression is the most apt approach to test the underlying hypotheses.
The support offered for the second hypothesis in Table III finds further confirmation in the results presented in Table V. In terms of restorative justice regimes, the coefficient for ceasefires is negative and statistically significant, indicating the comparatively positive influence peace agreements have on subsequent implementation of restorative forms of justice. Interestingly, the positive effect attained in analyzing conflict termination on the isolated implementation of purges in the post-conflict period disappears in the analysis in Table V, although the coefficient for ceasefires on retributive forms of justice is positive.
Marginal effects for conflict outcomes on restorative justice regimes
The marginal effects in Table VI were calculated for anocratic states holding all discrete variables at their modal values and continuous variables at their means.
One of the notable implications of these findings is that the differences between negotiated outcomes have stronger effects in terms of restorative justice implementation as compared to the pursuit of retributive justice. In fact, the only statistically significant finding supporting Hypothesis 1 stemmed from the effect of ceasefires on the implementation of purges. How can this disparate impact be explained? First, the ability of states and rebel groups to identify and sanction individuals within their ranks may require a degree of institutionalization within their bureaucracies and the funding necessary to carry out investigations, judicial processes, and removal procedures. Although states and rebels may have structures of governance, their institutional capacity and economic health will likely be dampened and degraded by the hostilities, making it more difficult to implement retributive forms of justice. This relates to the second factor linked with international intervention. External states and organizations may assist domestic actors by rebuilding key infrastructure or providing aid to facilitate the implementation of post-conflict justice; however, external actors may be reluctant to assist in the pursuit of justice if such processes aim at punishing political opponents or stem from illiberal purposes. Furthermore, the use of retributive justice in this manner may incur international condemnation, reducing the legitimacy of incumbent governments or rebel movements. Leveraging the data on UN peacekeeping operations in the sample, only 6% of trials, 0% of purges, and 14% of exiles coincided with peacekeeping operations. These data provide preliminary evidence that these factors may help explain why the restorative justice results are more robust than those related to forms of retributive justice.
Additional tests
The primary analyses reflect the importance of disaggregating negotiated outcomes for understanding the implementation of individual types of justice as well as different types of justice regimes. However, the findings herein may be artifacts of other theoretical or statistical properties not explicitly accounted for by the statistical methods utilized in the analyses in Tables III and V, and I employ additional tests to check the robustness of the results.
First, the sizes of the samples used in the analyses in Tables III and V are relatively small. One limitation potentially attributable to a small sample size is the greater amount of noise introduced into the analysis, inflating standard errors and rendering it more difficult to identify true effects between variables. This challenge should confer greater confidence in the results in Tables III and V, which detect effects despite the size of the samples; however, given the number of observations relative to the overall population of civil wars, it could be the case that the effects arise from a particular subset of cases that are not representative of the population, thereby revealing false positives. 11 To mitigate this concern, I ran a series of robustness checks that began by omitting certain control variables that decreased the sample size in the original analysis and incrementally adding these variables into separate models. Across each of the models estimated using this process, the correlations reflected in Table III generally remained. Although small sample sizes may implicate certain limitations on statistical analysis, the additional measures undertaken mitigate these concerns and heighten the impact of the statistically significant findings presented in Tables III and V.
Another concern may arise because the decision to enter into a negotiated outcome is correlated with the choice of post-conflict justice implementation. 12 In order to test this potential, I ran a series of bivariate probit models to estimate, in the first stage, the effect of conflict dynamics on the likelihood of a conflict ending in a negotiated outcome; in the second stage, I turned to an estimation of the potential for the different types of agreements to effect the implementation of justice mechanisms. The results of the bivariate probit regression analysis yield substantially similar results to those in Tables III and V. Once again, ceasefires, relative to peace agreements, exhibit a positive and statistically significant effect on the likelihood of policymakers pursuing purges after civil conflict. Furthermore, the coefficients for ceasefires are negative and statistically significant in relation to the implementation of amnesties and reparations in the post-conflict environment. Taken together, these results provide additional evidence for both hypotheses. 13
Conclusion
Previous work has examined the relationship between post-conflict justice and conflict outcomes by treating peace agreements and ceasefires as being the same. However, the empirical analysis herein demonstrates the unique effects of these negotiated outcomes. With respect to ceasefires, officials will be constrained in their ability to implement restorative justice while also being incentivized to leverage retributive forms of justice as a means of control and power. The opposite dynamics exist following peace agreements, after which policymakers will be encouraged to implement restorative justice mechanisms to assist the state and society in moving past its history of conflict.
The differentiation in outcomes stemming from salient features of ceasefires and peace agreements opens up avenues of further research concerning the relationship between conflict outcomes and post-conflict justice. Outside the dichotomy of peace agreements and ceasefires, contexts on each end of the dual sovereignty spectrum exhibit considerable variation in the relationships, de jure or de facto, between states and rebel groups. For example, Staniland (2012) outlines a six-part typology of wartime political orders that can define contours of cooperation or conflict between warring parties without resolving underlying issues or demobilizing one or both of the competing sides. While the empirical results herein reveal that ceasefires and instances of low activity are quite similar with respect to their effects on post-conflict justice relative to peace agreements, future research should investigate how variation within these contexts of dual sovereignty may influence decisions to pursue post-conflict justice. Ultimately, answering these questions concerning the determinants of post-conflict justice will help scholars and policymakers translate the goals of post-conflict justice from the aspirational to the actualized.
This article offers several practical implications for stability, human rights, and peacebuilding. While present exigencies may compel negotiators to pursue any agreement to quell hostilities in the interim, it is important to recognize the potential downstream effects of the different types of negotiated outcomes. If negotiators intend for ceasefires to clear a path toward a comprehensive peace agreement, an understanding of the post-conflict consequences is imperative for the success of this process. If the insecurity and fear instilled among former combatants in the wake of ceasefires incentivize leaders to consolidate power and loyalty through trials or purges, the retributive methods employed may constitute further abuses and atrocities, compounding existing questions of wartime behavior that will require redress by elites either in negotiations over a peace agreement or in the overall aftermath of civil war. Rather than clearing the path of peace, ceasefires may lead adversaries to erect additional roadblocks along the way.
Footnotes
Replication data
Acknowledgements
I wish to thank Jessica Maves Braithwaite and Faten Ghosn for their crucial feedback and support throughout the process of working on this article. My thanks also go to Alex Braithwaite, Chad Westerland, Kathy Powers, the editor, and the anonymous reviewers for their helpful comments and recommendations. Finally, I am grateful for the encouragement and support of my family and friends in these scholarly endeavors.
Notes
References
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.
