Networks and Conflicts: Theory and Evidence from Fighting Groups in Africa

Civil wars often involve complicated webs of alliances and rivalries with multiple actors. Examples include recent conflicts in Somalia, Uganda, and the Democratic Republic of Congo. We investigate how the network of military alliances and rivalries affects the overall conflict intensity, destruction and death toll. To this aim, we construct and test a model that merges elements from network and conflict theory. More specifically, we embed a two-dimensional network of alliances and rivalries into a multi-agent contest success function. Fighting externalities working through the networks affect each actor's fighting strength. We derive a closed form characterization of the equilibrium, and characterize how the local and global network structure affects individual and total fighting efforts. The structural equation of the model is then tested for "The Great War of Africa", using a detailed, novel dataset covering the Democratic Republic of Congo from 1998 to 2011. This is an ideal setting for our analysis, with many fighting groups and a complex network of alliances and rivalries. After obtaining structural estimates of the fighting externalities, we perform a Key Player analysis (i.e. welfare analysis) to gauge which groups are the most destructive ones. The analysis reveals, among other things, that the most destructive groups are not necessarily the most active ones, but rather the most influential ones through their alliance and rivalry links.