With strong conceptual arguments in its favor, decentralization is a popular and growing policy trend across the world. And yet dozens of empirical studies have failed to find convincing evidence that past reforms have worked. This begs two questions: 1)Why does decentralization produce indifferent results? and 2) Why is there so much centralization in the first place? The paper develops a simple model of a legislature in which municipal representatives bargain with central government agents over the allocation of public resources. By locating central government in a particular geographic space ¿ the ¿capital¿ ¿ and invoking self-interest on the part of its residents, I can answer both questions. I introduce the concept of residual power, which underpins the model and determines the flow of resources to districts. There is so much centralization because residual power is located in the capital, whose residents directly benefit from weak local governments.