This thesis starts by arguing that the civil conflicts that erupted in the African Great Lakes are rooted in a continuous pursuit of power, in which ethnic, regional and political identifiers are used by the contenders for power to rally community support. In an introductory chapter, I go back to the colonial era, drawing attention to Burundi and Rwanda, and then describe in more details Burundi's refugee crisis, ex-combatants' demobilization and the 2010 elections, all of which will be addressed in the subsequent chapters. <p><p>In the second chapter, entitled "On the Instrumental Power of Refugees: Household Composition and Civil War in Burundi", I study changes in household composition following household's exposure to civil war in Burundi. The analyses rely on a panel dataset collected in rural Burundi in 2005 and 2010. To address concerns over the endogenous distribution violence, I use an instrumental variables strategy using the distance to refugee camps, in which the Hutu rebellion was organized from the mid-1990s onwards. The analysis focuses on the impact of violence on demographic changes within households.<p><p>The third chapter, entitled "Who Benefited from Burundi's Demobilization Program?" and co-authored with Olivier Sterck (University of Oxford) and Philip Verwimp (ULB), assesses the impact of the demobilization cash transfers program, which took place from 2004 onwards in post-war Burundi. In the short run, we find that the cash payments had a positive impact on beneficiaries' consumption, non-food spending and investments. Importantly, it also generated positive spillovers on civilians in their home villages. However, both the direct impact and the spillovers seem to vanish in the long run. Ex-combatants' investments in assets were not productive enough to sustain their consumption pattern in the long run, as they ultimately ran out of demobilization money. <p><p>In the fourth chapter, entitled "From Rebellion to Electoral Violence. Evidence from Burundi" and co-authored with Andrea Colombo (ULB) and Olivier Sterck (University of Oxford), we aim at understanding the triggers of electoral violence in 2010, only a few months after the end of the war. We find that an acute polarization between ex-rebel groups -capturing the presence of groups with equal support - and political competition are both highly conducive to electoral violence. Disaggregating electoral violence by type, we show that these drivers explain different types of violence. Perhaps surprisingly, we find that ethnic diversity is not associated with electoral violence in post-conflict Burundi. <p><p>In the last chapter, entitled "Who Benefits from Customary Justice? Rent-seeking, Bribery and Criminality in sub-Saharan Africa" and co-authored with Olivier Sterck (University of Oxford), we have a closer look at the judicial system of Uganda, an important institution in a post-conflict economy. In many African countries, customary and statutory judicial systems co-exist. Customary justice is exercised by local courts and based on restorative principles, while statutory justice is mostly retributive and administered by magistrates' courts. As their jurisdiction often overlaps, victims can choose which judicial system to refer to, which may lead to contradictions between rules and inconsistencies in judgments. In this essay, we construct a model representing a dual judicial system and we show that this overlap encourages rent-seeking and bribery, and yields to high rates of petty crimes and civil disputes. <p><p>In Burundi, history has shown that instability in one country of the Great Lake region may destabilize the whole area, with dramatic effect on civilian population. Understanding the dynamics laying at the origin of violence, during and after civil conflict, is crucial to prevent violence relapse in any form, from petty criminality to larger scale combats. <p> / Doctorat en Sciences économiques et de gestion / info:eu-repo/semantics/nonPublished