Presidential Use of Centralization and Politicization

Although the strategies of centralization—the creation of policy in the White House—and politicization—the use of political appointees to influence the bureaucracy—have long been considered foundational to the modern presidency, previous work had largely addressed these strategies independently and informally. This dissertation addresses both issues by presenting and then empirically testing formal models of the president’s centralization and politicization decision. The first paper of the dissertation formally models the strategic trade-offs the president faces in creating a policy when both strategies are available, with one model focusing on policy creation and the other incorporating policy implementation. Among other new, counterintuitive predictions, I find that politicization should decrease when there is extreme ideological distance between the president and agency, in contrast to the previous literature. I also find that centralization and politicization act as substitutes, even when no such assumption is made about their relationship. The following papers test predictions from the models using a variety of data and methods. The second paper employs original survey items from the Survey on the Future of Government Service, a large-scale survey of federal government executives across three presidential administrations, to create the first measures of presidential centralization by policy area. In accordance with the theory, I find the first direct evidence that presidents are more likely to centralize as ideological distance between president and agency increases. I also create the first dependent variable to directly contrast centralized and politicized policy influence. With this measure, I show that centralization and politicization do not serve as complements, but that politicization is increasingly replaced by centralization as agencies grow more ideologically distant from the president. The third paper supplements these analyses by evaluating factors across time that affect centralization. With an archival-based measure of centralized policy creation of randomly sampled policies from the Eisenhower through Clinton administrations, I test how presidential use of centralization is affected by contextual factors such as divided government and centralized capacity. As predicted, centralization increases as centralized capacity expands, as well as when congressional ideology diverges from the president's preferences. I conclude with suggestions for future research