Three essays in Economics of crime, corruption and conflict

This thesis examines how violence—whether perpetrated by the state, produced by war, or embedded in governance—shapes institutional behaviour, public goods provision, and citizens’ perceptions of corruption. It integrates administrative records, geospatial data, and nationally representative survey microdata across three self-contained chapters. The first chapter studies how death-investigation laws affect the visibility of police killings in the United States. Using a county-year dataset (2013–2019) and a multi-outcome stratification framework, it shows that where law enforcement can certify causes of death—especially under sheriff-led systems—reported police killings are roughly 40% lower than expected relative to comparable adjacent counties. Underreporting co-moves with reclassification into “circumstances undetermined” homicides and with greater withholding of homicide statistics. The second chapter investigates post-conflict environmental governance in Kosovo by linking NATO bombing intensity, the siting/opening of post-war landfills, and infant mortality. Difference-in-Differences and Triple-Differences designs show that infants born within 6km of landfills after opening in heavily bombed municipalities face a 3.3–4.6 percentage point higher risk of dying before age one, representing a 118–164% increase relative to the baseline. These effects are concentrated in externally financed landfills built during the reconstruction period, and are accompanied by higher perceptions of local corruption. The third chapter examines how forms of wartime violence shape long-run perceptions of corruption in Kosovo. Combining municipality-level civilian and armed-group casualties with UNDP Public Pulse surveys (2010–2023), civilian-targeted violence is associated with higher perceived corruption—especially toward local institutions—whereas armed-group casualties are linked to lower perceived corruption. Analyses using individual-level exposure to human rights violations provide more consistent and robust evidence of these long-term effects, underscoring the importance of personal experience in shaping institutional trust. Taken together, the chapters show that institutional design conditions accountability, post-conflict governance is fragile, and the legacies of violence are heterogeneous across perpetrators and institutional tiers