Felipe Cesar Marques

The objective of this dissertation is to assess the effectiveness of some Brazilian education policy alternatives, for all educational levels, regarding their capacity to promote good academic performance to their contemplated students, with special attention paid to the existent heterogeneity of Brazilian education. For this, the study is structured with three essays, which have the microeconomic educational production function as their theoretical framework and apply the econometric tools of impact evaluation. The first essay seeks to determine the impacts of early childhood education (ECE) access on public elementary schools fifth graders students’ performance throughout the distribution of 2017 Sistema de Avaliação da Educação Básica (Saeb) test scores, taking into consideration the endogeneity of ECE enrollment decision. By applying an instrumental variable unconditional local quantile treatment effects model, the results of the essay show a scenario where children with good performance benefit, while children with bad performance suffer negative effects from attending ECE. Differentiated results based on mothers’ educational level are also found. The second essay deals with the academic achievement differentials found at military administered schools compared to typical Brazilian public schools. The performance differentials of elementary school ninth graders and high school third graders at 2017 Saeb tests are evaluated with the Oaxaca-Blinder detailed quantile decomposition method. The results of this essay show that the low performing students benefit more from military schools than the high performing ones. For those, most of performance differentials can be explained due to differences on educational inputs, especially due to the better infrastructure found on military schools. Such pattern is even more evident on mathematics tests and among elementary school students. Finally, the third essay analyses the effectiveness of Brazilian’s licentiate degree distant education courses, considering the performance of students enrolled in distance education at the 2017 Exame Nacional de Desempenho dos Estudantes test as the outcome variable. Several propensity score based techniques are used to create the counterfactual for these students. The results of the essay vary, essentially, between null and negatives impacts for distance education, depending on the course under consideration. A secondary analysis is carried, limited by the sample of students enrolled in the Sistema Universidade Aberta do Brasil, but its results also don’t show a clear indication of the direction of the impacts. Taking the results of all of the essays into consideration, the dissertation demonstrate the complexity involved in the design of educational policies, that often end up creating some groups of beneficiated students at the expense of others, instead of ensuring good academic results to all of the students served by those policies.