Do more decentralized local governments do better? An evaluation of the 2001 decentralizing reform in Colombia

In this paper I evaluate the impact of the 2001 decentralizing reform in Colombia. I first develop a model of incomplete contracts a ̀ la Hart et. al. (1997) in which local governments choose what fraction of a divisible project is contracted with the private sector. I show that the fraction contracted with the private manager increases with decentralization if local govern-ments receive positive net gains from the (non contractible) innovations that makes the private manager. In the empirical part I use data of Colombia’s municipalities. I look at the effect of the 2001 reform on enrollment in precollege school. While all municipalities are subject to earmarked national transfers, with the reform some of them received more responsibilities in providing education (deeper decentralization) than the others. Particularly important, the reform entitles the more decentralized municipalities to sign subsidy contracts with private schools. Departments (the regional governments) are entitled to sign this type of contracts for the less decentralized municipalities. Since the rule for municipalities to receive more respon-sibilities follows an exogenous population threshold, I can implement Regression Discontinuity Design. Enrollment is measured through two variables: the number of students enrolled in pub-lic schools and the number of subsidized students enrolled in private schools. Results suggest that more decentralized municipalities have subsidized more students in private schools. The difference accounts for 20 % of enrolment in private schools and 3 % of population in schooling age in 2005. There are no significant differences among municipalities regarding enrolment in public schools.