Designing Effective Peace Accords to Tackle Civil War Recurrence

Civil war is the most frequent and destructive form of armed conflict today. Nearly one third of societies that have experienced one civil war also experience a second or a third one and ninety percent of civil wars in the 21st century are recurrences of previous ones. We investigated peace processes that experienced repeated relapses into conflict before ultimately reaching an agreement that ended widespread conflict-related violence to explore: How can we design effective peace accords that prevent conflict recurrence? We built a dataset of Conditions of Recurrence (CoR), coding all the agreements in 14 peace processes which experienced repeated failures before achieving a stable peace accord from 1976 to 2015. We utilised machine learning in the form of decision trees to identify which provisions in peace agreements and contextual factors predict civil war recurrence. We subsequently test the strength of these relationships through regression analysis of the PAIC dataset (Fontana et al, 2020). Our empirical results suggest that provisions in peace accords matter above contextual factors when trying to prevent civil war recurrence. Specifically, we find that the prevention of recurrence is associated with UN mediation and provisions for the inclusion of women in post-conflict societies. These findings rebalance the emphasis of the existing literature on contextual and conflict-related factors as determining conflict recurrence. Finally, we explore two case studies of recurrent civil war to hypothesise some causal pathways linking UN mediation and provisions for the inclusion of women in post-conflict societies with the end of violent conflict: Sierra Leone and Liberia.