Decentralized policy implementation

This article develops a regime-splitting process model of decentralized policy implementation to integrate two theoretical approaches rooted in the tension between local flexibility and national control. The author estimates a model that simultaneously assesses the ability of each approach to explain the outcome it is meant to map onto (case-level discretion for local flexibility and aggregate responsiveness for national control) as well as each approach's extensibility to the other approach's domain. The data for the study come from the implementation of eight primary statutes by the U.S. Environmental Protection Agency's regional offices. The results are counterintuitive: the national control approach largely explains case-level discretion, the domain of local flexibility, but retains some power for explaining aggregate responsiveness. The local flexibility perspective contributes to both case-level discretion and aggregate responsiveness. Both models work outside their traditional domains, but neither is a sufficient explanation for decentralized policy implementation.