Inequality in Educational Opportunity and Achievement: Evidence from Brazil

不平等无处不在。在某些情况下，它被认为是社会分裂和经济停滞的根源。普遍的共识是，教育在促进社会经济流动性和消除收入不平等方面发挥着重要作用。从这个意义上说，必须让不平等现象普遍更为严重的发展中国家的经济政策了解教育选择的动态。尽管如此，现有的大多数关于高等教育的经验文献都集中在发达经济体，部分原因是数据的可用性。基于巴西相对未经探索的数据，我提出的研究旨在揭示教育不平等的联系。这些是我将在工作中解决的关键问题：是什么情况促使大学生在完成学位之前辍学？如果大学提供的学历具有高劳动力市场回报，那么这些是整个收入分配中的所有人都享有的，还是只有一部分人享有？在可能抵消劳动力市场力量的情况下，国家资助的财政援助是否有效？巴西为这一分析提供了理想的环境，并在尚未开发的微观数据方面提供了巨大的机会。我计划使用的主要数据来源之一是高等教育年度普查，它使我能够跟踪每个巴西学生在整个大学生涯中的进展，将个人的社会经济特征与他们入学和毕业或辍学的确切日期联系起来。为了解决上述问题，我计划将我的工作分为三个方面，这在巴西是前所未有的。经济援助的重要性-在这第一阶段，我将调查的因果关系影响的经济援助对大学的表现（入学，辍学和完成）使用回归不连续设计。此外，探索其他人口统计学特征-如性别，种族，年龄，研究领域-可以用来了解是否有不同群体的学生异质性的反应。这本身就很有意思，并提供了发展B部分和C. B部分的背景。当地劳动力市场条件的作用-B部分的目的是评估当地劳动力市场是否以及在多大程度上有利于形成受过高等教育的工人，或者是否反而吸引未完成大学教育的年轻人。在这里，我的目标是确定当地市场的影响程度的比例转移到行业的份额使用巴蒂类型的工具变量或“多仪器”的方法。大学生涯的动态模型-最后，我将利用离散选择动态规划。这种模式应该由一系列连续的二元选择组成：首先选择是否进入大学，然后选择在随后的每一个阶段继续学习，直到完成学位，或者退出学习计划。这个模型及其估计将使我能够解决几个与政策相关的反事实问题，例如：大学入学率和完成率如何受到学费变化的影响？如果当地劳动力市场繁荣，辍学率会如何变化？最后，我将特别关注其他社会科学，如社会学，政治学和公共政策的相关工作，以便更好地评估政策影响，建立可能的结果互补性，并确定下一步研究可能需要的方法改进。

Inequality is ubiquitous. In some cases, it is believed to be the root cause of social division and economic stagnation. The general consensus is that education plays an important role in promoting socioeconomic mobility and levelling income inequality. In this sense, it is essential to inform economic policy in developing countries, in which inequality is generally more acute, about the dynamics of educational choice. Still, most of the existing empirical literature on higher education has focused on developed economies, partially due to data availability. Relying on relatively unexplored data from Brazil, my proposed research aims to shed new light to the education-inequality link.These are the key questions I will address in my work: what circumstances drive university students to drop out before completing their degrees? If universities provide qualifications with high labour market returns, are these enjoyed by all individuals across the entire income distribution or only by some? Does state-sponsored financial aid work in the presence of possibly offsetting labour market forces?Brazil offers an ideal environment for this analysis, as well as great opportunity in terms of still untapped microdata. One of the main data sources I plan to use is the Annual Census of Higher Education, which allows me to track the progression of each Brazilian student throughout their entire university career, linking individuals' socioeconomic characteristics to the exact date when they enrol and graduate or drop out of the university system.To address the above issues, I plan to divide my work in three strands, which have never been done before for Brazil.A. The importance of financial aid - In this first stage I will investigate the causal impact of financial aid on university performance (enrolment, drop-out and completion) using regression discontinuity design. Additionally, exploring other demographic characteristics - such as gender, race, age, field of study- can be used to understand if there are heterogeneous responses across different groups of students. This is interesting in itself and provides the background to develop part B and C.B. The role of local labour market conditions - The objective of part B is to assess if and to what extent local labour markets favour the formation of highly educated workers, or if instead, they attract young people before they complete their university education. Here, I aim to identify the degree to which local markets are affected by proportional shifts to industry shares using a Bartik-type of instrumental variable or "multiple instrumentation" approaches.C. A dynamic model of university careers - Finally, I will make use of discrete choice dynamic programming. The model should consist of a series of sequential binary choices: first choosing to enrol or not in college, and then choosing to continue in every subsequent period until completion of the degree, or withdraw, from the programme of study. This model and its estimates will allow me to address several counterfactual questions relevant to policy such as: how are university enrolment and completion rates affected by a change tuition fees? How do drop-out rates change if the local labour market is in a boom? Finally, I will pay special attention to the related work that from other social sciences, such as sociology, political science, and public policy, in order to better assess policy implications, establish possible complementarities in results and identify methodological improvements that might be needed for next research.