If the effects of institutional reforms are long-term and unpredictable, why do leaders undertake them? We analyze reforms where the incentives of agents pursuing a change, and the effects of that change, are highly asymmetric in time and policy dimensions. We examine detailed evidence from decentralization in Bolivia and Pakistan, and explain very different outcomes with a simple game-theoretic model of the incentives and interactions amongst ruling and opposition parties, and civil society, over multiple periods. Politicians decentralize to solve a discrete, often short-term political problem. But decentralization is neither short-term nor discrete, but rather a deep change in the structure of public finance and authority with long-term effects on government, politics and society. Understanding the original problem is key to understanding the characteristics of the reform implemented, as well as its ultimate success or failure. Our analysis likely extends to a broader class of deep reforms with long-term consequences.