Organizations matter in local governance: evidence from health sector decentralization in Honduras

Health systems strengthening is at the forefront of the global health agenda. Many health systems in low-resource settings face profound challenges, and robust causal evidence on the effects of health systems reforms is lacking. Decentralization has been one of the most prominent reforms, and after more than 50 years of implementation and hundreds of studies, we still know little about whether these policies improve, harm or are inconsequential for the performance of health systems in less-developed countries. A persistent problem in existing studies is the inability to isolate the effect of decentralization on health outcomes, struggling with heterogeneous meanings of decentralization and missing counterfactuals. We address these shortcomings with a quasi-experimental, longitudinal research design that takes advantage of a unique staggered reform process in Honduras. Using three waves of household survey data over 10 years for a matched sample of 65 municipalities in Honduras, we estimated difference-in-difference models comparing changes in outcomes over time between local health systems that were decentralized using one of three types of organizations [municipal governments, associations of mayors or non-governmental organization (NGOs)] and those that remained centrally administered. We find evidence of overall improvements between 2005 and 2016 in several service delivery-related outcomes, and additional improvements in decentralized municipalities governed by NGOs. NGO-led municipalities saw a 15% decrease in home delivery relative to centralized municipalities in 2016, a 12.5% increase in MCH facility delivery and a 7% increase in the use of a skilled birth attendant. There were no detectable positive treatment effects for vaccination, and a slight decline in the weight-for-length z-scores in NGO municipalities, but we find no systematic evidence of decentralization negatively impacting any maternal and child health outcomes. These findings highlight the importance of considering implementation context, namely organization type, when assessing the effects of decentralization reform.