Decentralization and Corruption: A Model of Interjurisdictional Competition and Weakened Accountability

Abstract

Decentralization has the potential to lower corruption and alleviate poverty across the world. The true effects of this process are unclear since there are relatively few studies on decentralization and many of these studies, both theoretical and empirical, give conflicting results. One major problem in the literature for the effects of decentralization on corruption has been sample selection bias. The main cross-country dataset for decentralization, the IMF's Government Finance Statistics (GFS), has data for only about 40 countries, and most of these are developed. I attempt to mitigate this sample-selection problem by first estimating a Heckman model for decentralization in order to predict values for unobserved countries and then using these predicted values to estimate decentralization's impact on corruption. My results show that decentralization has an insignificant effect on corruption, suggesting that decentralization alone may not be a useful tool for mitigating corruption.