A Regime-Splitting Model of Decentralized Policy Implementation

Abstract

This paper develops a regime-splitting process model of decentralized policy implementation in order to integrate two theoretical approaches rooted in the tension between local flexibility and national control. I estimate 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. My 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 counter-intuitive: 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.