Perverse decentralization is developed as a framework for understanding federal bargains and ethnic outbidding, with a focus on decentralization as a strategy of conflict management in multi-national African states. The argument advanced is that decentralization can displace conflict from the centre to the local arena when boundary design, fiscal authority, and security command are misaligned, encouraging elites to mobilize ethnicity within subnational institutions. Drawing on controlled comparison of three federal or quasi-federal systems; subnational quantitative analysis of conflict events (ACLED) before and after decentralization; process tracing of constitutional design episodes; and interviews with constitutional law practitioners, the study engages debates in federalism theory, ethnic outbidding, power-sharing, and political settlements. It develops the concept of “perverse decentralization,” in which devolution fragments political competition at the local level without resolving national distributional conflicts. Three core claims are advanced. First, devolution reduces conflict only where subnational units possess credible fiscal authority, clear boundaries, and safeguards against security sector capture by dominant local coalitions. Second, ethnic entrepreneurs exploit ambiguities in federal design by translating national distributional struggles into local contests over offices, districts, and customary authority. Third, comparative evidence from Ethiopia, Nigeria, and South Sudan shows that decentralization becomes perverse when it multiplies veto points without establishing impartial arbitration and stable revenue-sharing rules. The study concludes that constitutional design must be paired with clear intergovernmental fiscal compacts and dispute-resolution