Decentralization of health care systems and health outcomes: Evidence from a natural experiment

While many countries worldwide are shifting responsibilities for their health systems to local levels of government, there is to date insufficient evidence about the potential impact of these policy reforms. We estimate the impact of decentralization of the health services on infant and neonatal mortality using a natural experiment: the devolution of health care decision making powers to Spanish regions. The devolution was implemented gradually and asymmetrically over a twenty-year period (1981-2002). The order in which the regions were decentralized was driven by political factors and hence can be considered exogenous to health outcomes. In addition, we exploit the dynamic effect of decentralization of health services and allow for heterogeneous effects by the two main types of decentralization implemented across regions: full decentralization (political and fiscal powers) versus political decentralization only. Our difference in differences results based on a panel dataset for the 50 Spanish provinces over the period 1980 to 2010 show that the lasting benefit of decentralization accrues only to regions which enjoy almost full fiscal and political powers and which are also among the richest regions.