'There is nothing more important than corruption': The rise and implementation of a new development idea

The 1990s saw the emergence of a key new trend in development discourse, the so-called 'anti-corruption eruption'. Fighting corruption moved from being a fringe area of concern to forming the cornerstone of the new good governance agenda. As a policy it was embraced at both the national and international levels, by official aid agencies and NGOS, by international organizations and domestic governments. Initial research on this phenomenon focused on its origins, and the question of why it emerged when it did. Explanations centered on the changing role of aid in the aftermath of the Cold War, and on the emergence of new knowledge on the effects of corruption. These explanations drew from both state specific and policy diffusion approaches. Such explanations did not however explain how anti-corruption policy was actually implemented in practice, or how its popularity has persisted in the absence of significant success.;In this dissertation I answer these questions by combining the theoretical insights of constructivist models of policy diffusion with a more traditional donor-specific approach to offer a full and comprehensive picture of the main influences on the development and implementation of anti-corruption policy. For the former, I focus on the role of epistemic communities in the policy development process, while with regard to the latter I focus on independent state factors such as economic and strategic needs.;Through an in-depth qualitative analysis of the academic literature on corruption, I confirm the existence of a neo-liberal, economics-centered, anti-corruption epistemic community. I map the typology derived from this analysis to the actual anti-corruption projects pursued by OECD donors as well as to their policy discourse on the subject, further confirming the strong influence of the epistemic community on anti-corruption policy. In addition I devise a unique measure of anti-corruption aid. I use this in a probit model of the determination of anti-corruption aid so as to test the relative significance of donor specific factors vis-a-vis epistemic community influences. Overall both donor specific and epistemic community influences are found to be strong, with the latter being dominant in the absence of pressing donor specific concerns.