Chasing the Key Player: a Network Approach to the Myanmar Civil War

Andrea Di Miceli (UCLA)

Abstract: I study the determinants of civil conflict in Myanmar. As governments in weak states often face several armed groups, they have to allocate resources to fight a subset of them strategically. I use a simple model to embed heterogeneity among rebel groups stemming from their network of alliances and enmities. The key insight is that, by attacking a group, the Myanmar army weakens its allies. Therefore, the model predicts that the Myanmar army strategically targets armed groups who are central in the network of alliances. To test the model's predictions, I collect a new data set on rebel groups' locations, alliances, and enmities for the period 1989-2015. Using geo-referenced information on armed groups attacked by the Myanmar army, the empirical evidence strongly supports the predictions of the model. A one standard deviation increase in a group's centrality increases the likelihood of conflict with the Myanmar's army by 1.2% (over a baseline yearly conflict probability of 6.4%), thus identifying a new determinant of conflict. This result is robust to variables measuring the opportunity cost of conflict such as rainfall and commodity price shocks. Since past (and expected) conflicts might affect alliances and enmities between armed groups, I pursue an instrumental variable strategy to provide evidence that the mechanism proposed is indeed causal.