The viability of civil war peace agreements

Abstract

Civil war peace agreements are prone to collapse. While some research suggests that multiple layers of power-sharing provisions lead to more viable agreements, others have suggested that negotiated settlements are not only more likely to return to violence, but that those cases will be more deadly as a result. We suggest here that previous research has failed to address the various ways that peace agreements emerge and that this context is crucial in explaining peace agreement viability. In some situations, rebels are likely to earn those concessions through battlefield success. Governments may feel compelled to address underlying grievances in order to stop the bloodshed and ultimately maintain some political clout in the postwar regime. In other cases, however, governments may feel pressure to engage in discussions and to provide concessions by outside actors, whether in the form of diplomatic intervention, economic, or foreign military intervention. Hypotheses are tested on all civil war peace agreements identified by the UCDP Peace Agreement Data-set for 1975–2011. Findings from logit and hazard models suggest that agreements brought about in the aftermath of military intervention on behalf of rebels are more likely to endure while those earned on the battlefield (i.e., stronger rebels) do not. In addition, mediation enhances peace agreement viability, while interventions on behalf of governments tend to undermine it.

Notes

1. This is not to suggest that diplomatic intervention cannot be costly. A mediator’s reputation may suffer internationally or nationally in such situations (Svensson 2009). We consider the cost of failed diplomatic intervention to be more acceptable, however, compared to other forms of intervention, particularly foreign military intervention.

2. The data-sets used in this project have been used to test similar questions (see, for example, Hoddie and Hartzell 2003; Svensson 2007; Cunningham, Gleditsch and Salehyan 2009; DeRouen Jr. et al. 2009; Nilsson 2010; Escriba-Folch 2010; Peksen and Olson Lounsbery 2012; Clayton 2013; DeRouen Jr. and Moller 2013; Hultquist 2013; Olson Lounsbery 2016). As a result, they are widely accepted as reliable tools to test the types of hypotheses examined in this project.

3. The CWM data-set is built upon the framework of the ACD (see Gleditsch et al. 2002). Specifically, the ACD provides civil war episode start and end dates. The CWM data code as one episodes that were mediated between these dates and/or after the end date but before another start date (see DeRouen Jr., Bercovitch and Pospieszna 2011).