This book offers insights and lessons that help us understand when the answer is “Yes”, and when it is “No”. It shows us how decentralisation can be designed to drive development forward, and focuses attention on how institutional incentives can be created for governments to improve public sector performance and strengthen economies in ways that enhance citizen well-being. It also draws attention to the political motives behind decentralisation reforms and how these shape the institutions that result. The book's purpose is to marry policy makers’ detailed knowledge and insights about real reform processes with academics’ conceptual clarity and analytical rigor. This synthesis naturally shifts the analysis towards deeper questions of decentralization, stability and the strength of the State. The book explores these in Part 1, with deep studies of the effects of reform on state capacity, political and fiscal stability, and democratic inclusiveness in Bolivia, Pakistan, India, and Latin America more broadly. These complex questions—crucially important to policy makers but difficult to address with statistics—yield before a multipronged attack of quantitative and qualitative evidence combined with deep practitioner insight. How should reformers design decentralisation? Part 2 examines these issues with evidence from four decades of reform in developing and developed countries. What happens after reform is implemented? Decentralization and Local Service Provision turns to decentralization’s effects on health and education services, anti-poverty programs, etc. with original evidence from twelve countries across Africa, Asia, and Latin America.

We review recent evidence regarding decentralization and state strength and argue that decentralization can deepen democracy without compromising state strength if adequately designed. We examine how decentralization affects five key aspects of state strength: 1) Authority over territory and people, 2) Conflict prevention, 3) Policy autonomy and the ability to uphold the law, 4) Responsive, accountable service provision, and 5) Social learning. We provide specific reform paths that should lead to strengthening in each. Decentralizing below the level of social cleavages should drain secessionist pressure by peeling away moderate citizens from radical leaders. The regional specificity of elite interests is key. If regional elites have more to lose than gain from national schism, they will not invest in politicians and conflicts that promote secession. Strong accountability mechanisms and national safeguards of minority rights can align local leaders’ incentives with citizens’, so promoting power-sharing and discouraging local capture or oppression. “Fragmentation of authority” is a mistaken inference; what decentralization really does is transform politics from top-down to bottom-up, embracing many localities and their concerns. The state moves from a simpler, brittler command structure to one based on overlapping authority and complex complementarity, where government is more robust to failure in any of its parts. Well-designed reform, focusing on services with low economies of scale, with devolved taxation and bail-outs prohibited, should increase public accountability. Lastly, we advance a novel way that decentralization can strengthen democracy: by allowing citizens to become political actors in their own right, the small scale of local politics should promote social learning-by-doing, so strengthening political legitimacy, state-building, and ‘democratic suppleness’ from the grassroots upwards.

The Australian Labor Party went to the 2007 election promising a new era of cooperative federalism that would end the ‘blame game’ between federal and state governments and re-energise reform and productivity agendas. On the evidence of the Council of Australian Governments (COAG) meeting on 26 March 2008, these agendas are advancing rapidly. The communiqué foreshadowed a raft of new commonwealth-state agreements, streamlined arrangements for special purpose grants and, perhaps most significantly,