On the Beaten Path: Violence Against Civilians and Simulated Conflict Along Road Networks

Why do some conflict zones exhibit more violence against civilians than others? In answering this question, the literature has emphasized ethnic fractionalization, territorial control and strategic incentives, while overlooking the consequences of armed conflict itself. This oversight is partly due to the methodological hurdles of finding an appropriate counterfactual for observed battle events. In this chapter, we aim to test empirically the effect of instances of armed clashes between rebels and the government in civil wars on violence against civilians. Battles between belligerents may create conditions that lead to surges in civilian killings as combatants seek to consolidate civilian control or inflict punishment against populations residing near areas of contestation. Since there is no relevant counterfactual for these battles, we utilize road networks to help build a synthetic risk-set of plausible locations for conflict. Road networks are crucial for the logistical operations of a civil war and are thus the main conduit for conflict diffusion. As such, the majority of battles should take place in the proximity of road networks; by simulating events in the same geographic area, we are able to better approximate locations where battles hypothetically could have occurred but did not. We test this simulation approach using a case study of the Democratic Republic of the Congo (1998–2000) and model the causal effect of battles using a spatially disaggregated framework. This work contributes both substantively and methodologically to the literature on micro-foundations of civil war and reactive violence in two main ways: (1) It offers a tentative framework for crafting synthetic counterfactuals with event data. (2) It proposes an empirical test for explaining the variation of violence against civilians as a result of battle events.