This dissertation is an analysis of civil war duration. It presents a theoretical model that argues that conflicts with multiple parties will be harder to resolve through negotiation and therefore of longer duration because, in the absence of agreement, parties will continue to fight to try to achieve their goals. This model is tested using two methods: a large-n statistical study of the effect of the number of parties on the duration of civil war and a qualitative analysis of the negotiation strategies pursued by groups during the civil war negotiations in Rwanda and Burundi.
This project presents an advance on the existing literature on civil war because it moves beyond the narrow assumption that civil wars are fought between two parties and instead examines how multiple parties may make civil war settlement more difficult. A veto player framework, commonly used in the American and Comparative political science literature to explain governmental bargaining, is adapted to explain why negotiations are less likely to produce agreement in these types of conflicts. This veto player approach is subjected to rigorous empirical testing using new yearly data on civil war and a case-study analysis of the civil war negotiations in Rwanda and Burundi.
Broader Impact: This project makes a theoretical and empirical contribution to the existing literature on civil war. Theoretically, the project demonstrates that multi-party conflicts have different dynamics than the two-party conflicts assumed in the existing literature. In particular, it shows that negotiations are more difficult in multi-party conflicts because groups have incentives to block agreement. Empirically, this project will generate quantitative and qualitative data that will be of use to scholars, policymakers and practitioners interested in understanding why civil wars last as long as they do. There currently exists almost no systematic data on the number of parties involved in civil wars. As part of this project, the student has collected yearly data on all parties to all civil wars begun since 1945. This is an important contribution since most existing statistical analyses of civil war test theories using only country-level variables that do not directly test the theories. This project creates a dataset that includes important data at the group-level that will allow me to more accurately test the effect of group dynamics on the duration of civil war. These group-level data will also be useful to other scholars who can use them to test their own theories about civil wars. Additionally, although most existing theories of civil war duration and termination are based on assumptions about the way bargaining works, very little systematic analysis of negotiations in civil wars has been conducted. In this dissertation, the student collects data on the day-to-day negotiation process in the Rwandan and Burundian civil wars which allow for systematic testing of the mechanisms identified in this and other bargaining models.
