Dois ensaios em avaliação de políticas públicas

Abstract

The present dissertation consists of two independent essays in public policy evaluation, which estimates the causal effects of two major Brazilian policies, the Mais Médicos Program (PMM) and the Bolsa Família Program (PBF). The first essay aims to evaluate the PMM in terms of the provision of physicians, presenting estimates of its impact on hospitalization for ambulatory care sensitive conditions (ACSH). The differences-in-differences method was used with propensity score matching (double difference matching), using three specifications, a falsification test and also a dynamic endogeneity test to confirm the robustness of the results. For the application of this methodology, a panel of municipal data was constructed covering several variables related to socioeconomic, demographic and public health infrastructure characteristics in the cities for the period from 2010 to 2016. First, it is shown that the program had an effect on increasing the density of physicians. Then, the main results show a significant reduction in hospital admissions in treated municipalities with an increasing and perceptible effect in the second year of the program. The second essay aims to analyze the impact of PBF in Brazilian intern migration. The Average Treatment Effect on Treated (ATT) was estimated through Propensity Score Matching (PSM) using microdata from Brazilian Census 2010. The results show that the program stimulates recipients’ remigration to their home states and retain those already established in a locality. The estimates are robust in the possible presence of omitted variables and in different matching specifications. By exploring regional heterogeneity, we have shown that the impacts on return and retention are greater for the Northeast, the concentrating region of return migrations and Program beneficiaries. Through Dose Response Function it is also shown that the results are positively correlated with the monetary values transfer by the Bolsa Família, which may indicate a channel for the empirical results.