In this dissertation I analyze three different models of delegation. First, I explore how delegation incentives change when the principal may lose power in the future in "Delegation and Political Turnover". One principal, called a legislator has policy making authority in the first stage. He can implement a policy location himself or he can develop to an expert. This expert can then implement both a policy location and location-specific quality. An election occurs, and the first legislator may be replaced by a second legislator with different policy preferences. The legislator in power then again decides whether to implement policy himself or delegate to an agent. I show that the first legislator can use delegation as insurance against possible future loss of power. The agent chooses a policy designed to be retained by both legislators in the second stage. Without delegation, the second legislator can always get his own ideal point, a terrible outcome for the first legislator. I then extend the model to show how political turnover influences the first legislator to invest in the agent's capacity. I also show that when the first legislator has a choice of agents, he prefers an agent opposed to the second legislator.,The second chapter, "Delegation to an Overconfident Expert"(with Scott Ashworth) is concerned with expert overconfidence. Policymakers often delegate partial decision-making authority to experts. Although monetary transfers can be useful in aligning the decision maker's and expert's policy choices, they are typically not observed in practice. We analyze a principal-agent model where the policymaker and agent have identical preferences over state-contingent policy, but disagree over the accuracy of the expert's information. The policymaker believes the expert to be overconfident in the precision of the signal he receives about the state of the world. In this case, the optimal mechanism is a delegation interval, without transfers.,Finally, in "Delegation, Information, and Federalism" I study a principal-agent model of federalism. One central government can delegate to two districts or make policy herself. The districts however have a monopoly on local information. There are coordination externalities between all levels of government that reward policy closeness regardless of local characteristics. The central government prefers to delegate when its specific conditions are moderate and externalities are low. When conditions are extreme and externalities are high, the central government prefers to implement a uniform policy across all levels government. This contrasts to local government decision making. While local governments also prefer a uniform policy when externalities are large, the uniform policy is most beneficial when local conditions are moderate.