Essays on the impact evaluation of social programs and public sector reforms

This thesis contains three essays on the impact evaluation of social programs and public sector reforms. Chapter 2 investigates whether the effects of a cash transfer program persist or wear off in the long-run. I study the first two phases of Bono de Desarrollo Humano (BDH) in Ecuador, each of which lasted about five years. My identification strategy uses a regression discontinuity design and relies on the fact that at the threshold of eligibility, the second assignment to treatment (in 2008/9) was independent of the first assignment (in 2003). This allows me to disentangle the impact of a short exposure to the program (treatment during one phase) from a long exposure (treatment during two phases). Most of the gains in enrollment and schooling were achieved in the short-run among children that started treatment when they were about to start elementary school, eleventh grade or Baccalaureate. However, an extended exposure to BDH was not enough to keep raising children’s education. Regarding labor market outcomes, BDH had a negative (not statistically significant) impact on the probability of working among young children but did not increase job opportunities among young adults in the long-run. Chapter 3 evaluates the impact on in-hospital mortality of a reform that made all health professionals working part-time switch to full-time contracts at public hospitals in Ecuador. I take advantage of the staggered adoption of the reform and hospital panel data to implement an event study to evaluate the impact of the reform. The results for the sample of admissions to the emergency department show that mortality in public hospitals decreased by 0.1% on the adoption year and by 0.2% one year later. Results were robust to several alternative specifications and to the inclusion of pre-reform characteristics that could have been used by policymakers to decide the order of implementation. More importantly, I show that the effects reported in this paper cannot be attributed to changes in other quality indicators at the hospital level like the length of stay or by changes in the patient mix. Chapter 4 studies the impact of increased liability risk facing physicians on the use of c-sections and on indicators of maternal and infant health. I take advantage of a legal reform that led to the hardening of sentences for cases of professional malpractice in Ecuador. I use a difference-in-difference strategy that compares the outcomes of two neighboring countries, Ecuador and Colombia, before and after the introduction of the reform and perform several parallel trend tests on the outcomes of interest and test for the stability of the demographic composition of both countries to support my identification strategy. During the five quarters following the reform, Ecuadorian doctors reduced the c-section rate by 1.1% among women aged 15 to 24 years, and by 0.9% among women aged 25 to 34 years. The c-section rate remained unaffected for women aged 35 to 44 years, possibly because doctors have less discretion over riskier births. Interestingly, the observed reduction in the c-section rate did not affect the health outcomes of mothers or newborns