Optimal investment in human capital through effective K-12 schooling is critical for building a productive work force. This investment is particularly important for minority and low income students. My dissertation uses econometric techniques to analyze the effects of different educational programs on the academic achievement of elementary and middle school students in Arizona. The first essay evaluates the effect of Arizona's new English program, the 4-hour ELD block, on the achievement of students. In the 2008-2009 academic year, Arizona law required that English Language Learner (ELL) students to be separated from their native English-speaking peers and interact in the same classroom for 4-hour per day with other ELL students. In this study dynamic panel data methods and regression discontinuity design are employed to analyze the effect of the 4-hour ELD block program on the academic achievement of students. Using data from one school district during the school years 2006 to 2010, this study shows that this new program did not have a notable effect on the state-wide test scores of ELL students. The second essay assesses the long-run and short-run effects of Teacher Incentive Fund (TIF) program in Arizona. The TIF program is a nationwide performance-based compensation plan that provides incentives to teachers based on the student performance. The TIF program started in Arizona in the 2007-2008 school year and targets high need schools. Using a panel data set from 2006-2007 to 2010-11 school year, the effect of the TIF program on the achievement of students is estimated using the difference-in-difference method. Comparing the short-run and long run effect of this program indicates that the long-run effect is greater than that of the short-run. Finally, by utilizing a rich set of panel data from 2006-2007 to 2010-2011 school years, the third essay studies the raw and value-added achievement gap between first and second-generation students with native students. This study shows that native students outperform both groups of immigrant students in reading and math tests. Within immigrant students, second-generation students outperform first-generation students in reading but not in math, while the achievement growth of the second-generation students has a slower pace.

Thesis (M. Phil.)--University of Hong Kong, 1993.

Conforme aponta a literatura, alta rotatividade de professores pode dificultar o desenvolvimento do trabalho da equipe escolar, com possíveis consequências prejudiciais para a aprendizagem dos alunos, e sinalizar outros problemas na escola, como condições de trabalho relativamente ruins. Esta dissertação tem o objetivo de investigar os determinantes da rotatividade docente na rede estadual de São Paulo, usando os dados dos Censos Escolares de 2007-2011 e considerando a estrutura institucional da rede. A descrição da rotatividade coletiva no nível das escolas evidenciou o alto nível da rotatividade paulista em âmbito nacional, bem como que a maior parcela da rotatividade é explicada pela migração de professores entre escolas. A análise da estrutura institucional da rede revelou que existem políticas com potencial de afetar à mobilidade docente, como o adicional por local de exercício (ALE) e a bonificação por resultados, e que a Secretaria da Educação do Estado (SEE) tem buscado elevar os custos de migração interna da rede. Revelou também que parcela da rotatividade docente até 2010 pode ser vista como estrutural, na medida em que quase metade dos professores não tinha vínculo com a escola e, por isso, não tinha garantia de continuidade na mesma de um ano para outro. Há evidências de que o professor que mais migra tem menor qualificação e de que o aluno mais afetado por equipes escolares mais instáveis é o de menor nível socioeconômico, principalmente nas séries iniciais do fundamental. Dentre as políticas da SEE, a Bonificação por Resultados está associada à menor rotatividade em todas as etapas escolares (ensino fundamental e médio), porém seu efeito é pequeno, e o ALE parece ajudar a fixar o professor do fundamental I nas escolas. / According to the literature, high teacher turnover can be detrimental to school work, with possible damaging consequences to student learning, and indicate underlying problems, as relatively poor work conditions. The goal of this dissertation is to investigate the determinants of teacher turnover at schools managed by São Paulo State government using data from Censo Escolar 2007 to 2011 and considering the institutional features of the school administration. São Paulo has high collective turnover rates by schools compared to other Brazilian States and migration between schools accounts for most of the total teacher turnover. Between the relevant institutional features of the school administration are the existence of policies related to turnover, as the teacher bonus and the adicional por local de exercício (ALE), and the fact that the Education Department has increased the migration costs between schools. Also we can view the teacher turnover of São Paulo State as partly structural, because half of the teachers did not have a stable link with schools until 2010. There is evidence that the migrant teacher is less qualified and that schools serving economically disadvantaged and black students appear more vulnerable to turnover, especially elementary schools (first grades of ensino fundamental). The teacher bonus is associated negatively to turnover, but its marginal effect is very small, and ALE seems to help to retain teachers in elementary schools.

O retorno do ensino superior tem sido objeto de estudo na literatura econômica por décadas, mas pouca evidência foi documentada sobre o prêmio salarial de frequentar as universidades mais seletivas no mercado de trabalho. Especialmente no Brasil, país no qual o ensino superior é visto como o fator primordial para a ascensão social, torna-se cada vez mais relevante compreender o papel do diploma de elite nos rendimentos futuros. Neste trabalho, estima-se o prêmio salarial de frequentar a Universidade de São Paulo (USP), uma das universidades mais importantes e seletivas do país, aproveitando-se dos grupos dos aprovados e não aprovados ao redor da nota final de corte. Como estratégia empírica, utiliza-se a abordagem de regressões descontínuas (fuzzy RD) para solucionar possíveis problemas de seleção frequentemente discutidos na literatura. Na amostra de indivíduos empregados no setor formal, não se observa retorno salarial significativo de frequentar a USP. No entanto, o contrafactual é amplo e possivelmente inclui cursos superiores qualificados, como os oferecidos por outras instituições públicas. Isso, aliado à localidade da estimação, pode subestimar o efeito agregado. Ao investigar as maiores carreiras do vestibular, poucas apresentam retorno diferente de zero nas medidas salariais (rendimento médio anual e salário médio de dezembro). Contudo, não é possível excluir que diferenças salariais possam existir em regimes de trabalho mais flexíveis, como nas profissões liberais, ou que o mercado assalariado não distingua instituições de nível superior, uma vez que apenas salários formais são considerados na estimação. Por grupo socioeconômico, candidatos que fizeram ensino médio e fundamental estritamente em escola pública apresentam retornos positivos de frequentar a USP. Não foram encontradas diferenças de retorno salarial entre homens e mulheres nem entre PPI (pretos, pardos ou indígenas) e brancos. Por fim, também não são documentados efeitos sobre tempo de emprego formal e número de vínculos empregatícios. Os resultados encontrados incitam uma série de discussões sobre políticas educacionais, especialmente aquelas relacionadas à expansão de vagas no ensino superior. / The return to higher education has been studied in the literature for decades, but there is little evidence regarding the wage premium of attending the most selective universities in the labor market. Especially in Brazil, where higher education is a crucial element to upward mobility, a broader comprehension about the role of the elite degree in future earnings becomes even more relevant. Hence, I estimate the wage premium of attending the University of São Paulo (USP), one of the largest and most selective universities in the country, by taking advantage of two groups around the admission grade cutoff - those admitted and those not admitted at the university. This study utilizes the Regression Discontinuity Design (fuzzy RD) to solve possible selection problems frequently discussed by the literature. In the sample of those employed in formal sector, there is no significant wage return to attending USP with respect to the counterfactual. However, the counterfactual is broad and possibly includes qualified majors, such as those offered by other public universities. This fact, combined with local estimates, can underestimate the aggregate effect. When investigating the returns in the major careers of the college entrance exam, few of them exhibit returns different from zero in the wage measures (average annual income and the average salary in December). Yet, it is not possible to exclude that wage differences might exist in more flexible work relationships, as in the liberal professions, or that the formal labor market does not distinguish higher education institutions, since we only have earnings data from formal employment. When taking socioeconomic groups into account, only candidates who studied exclusively in public schools present positive returns to attending USP. There were no differences in the wage returns neither between men and women nor between non-white and white candidates. Finally, there are no effects on formal employment tenure and number of employment relationships. The results documented in this paper trigger a series of debates regarding educational policies, especially those related to the expansion of higher education.

This thesis is composed by three independent essays in economics of education, which aim to assess the impact of different public policies in Brazil. The first chapter studies the effects of a government scholarship program for low-income college students, the Prouni. Propensity score matching based on observable student characteristics and a proxy for previous student performance is used to deal with selection effects. The results are robust across different specifications, and suggest that students who received a scholarship perform better and take less time to reach their final year of college than comparable students. The estimated effects are higher among students with full scholarships than for students with partial scholarships, indicating that the amount of aid received matters. Results also indicate that full scholarship recipients also have a lower probability of working while in college. The second chapter provides evidence on a large-scale pay for performance program in the state of São Paulo, which awarded group bonuses to school teachers and staff conditional on improvements in student performance. Results using a difference-in-differences and triple-differences framework show that the program had overall positive effects on student performance, although improvements vary across grades and subjects. Although it could be expected that free-riding effects increase with the number of teachers in schools, thereby limiting the impact of the program, this effect seems to be modest. Results also show that initially low-performing schools improved much more than the average. The third chapter analyzes whether the provision of information on school quality affects students\' school choices. More specifically, it explores whether the publication of grades obtained at a standardized high school test (the Enem) resulted in changes in enrollments in high- and low-performing schools. A sharp regression discontinuity design is used, taking advantage of the fact that an exogenous rule determined that only schools with a minimum number of test-takers would have their results published. The results show that the disclosure of school grades did not significantly affect enrollment decisions, in neither private nor public schools. The findings remain unchanged when controlling for the degree of competition faced by schools or their socio-economic environment. / Esta tese é composta por três ensaios independentes em economia da educação, que têm como objetivo avaliar o impacto de diferentes políticas públicas no Brasil. O primeiro capítulo estuda os efeitos do Prouni, um programa que oferece bolsas para estudantes de baixa renda no ensino superior. A metodologia usada para lidar com efeitos de seleção é a de propensity score matching, com base em características observáveis dos alunos e uma proxy para desempenho inicial. Os resultados, que são robustos a diferentes especificações, mostram que os bolsistas têm melhor desempenho e levam menos tempo para chegar ao último ano de faculdade do que alunos comparáveis. Os efeitos estimados são maiores para bolsistas integrais do que para bolsistas parciais, sugerindo que o valor da bolsa é relevante. Os resultados também indicam que os bolsistas integrais têm probabilidade menor de trabalhar durante a faculdade. O segundo capítulo apresenta evidências empíricas relativas ao efeito do programa de incentivos para professores do Estado de São Paulo, no qual foram distribuídos bônus de grupo para professores e funcionários em função da melhora do desempenho dos alunos. Neste capítulo é usado o método de diferenças em diferenças e de diferenças triplas. Os resultados sugerem que o programa teve efeitos positivos sobre o desempenho dos alunos, no entanto, os efeitos encontrados variam entre séries e matérias. Embora poderia esperar-se que em escolas com um maior número de professores efeitos de free-riding sejam mais fortes e limitem o efeito da política, este efeito não parece ser importante. O estudo mostra também que escolas com desempenho inicialmente baixo apresentaram melhoras mais importantes do que a média. O terceiro capítulo analisa se a divulgação de informações sobre a qualidade das escolas afeta as escolhas dos alunos. Mais especificamente, procura-se saber se a divulgação das notas obtidas no Enem (Exame Nacional do Ensino Médio) afetou o volume de matrículas em escolas com alto e baixo desempenho. A metodologia usada é a de regressão descontínua, com base em uma regra exógena que determinou que somente as escolas com um número mínimo de alunos que fizeram a prova teriam seus resultados divulgados. Os resultados indicam que a divulgação de notas a nível de escolas não teve impacto significativo sobre as decisões de matrícula, tanto no caso de escolas privadas como públicas. Os resultados se mantêm quando é levado em consideração o grau de concorrência enfrentado pelas escolas, ou o seu ambiente socioeconômico.

O objetivo do presente estudo é contribuir para a identificação do efeito causal de reduções no tamanho da classe e ampliação da jornada escolar sobre o rendimento escolar em matemática dos estudantes brasileiros da 4ª série do ensino fundamental de escolas de área urbana da rede pública de ensino. Para tanto, foram utilizados a metodologia de Generalized Propensity Score Matching, no caso em que o tratamento de interesse (tamanho da classe) é uma variável contínua, e o estimador Nearest Neighbor Matching com correção de viés para tratamento binário (jornada escolar). A base de dados empregada foi o Sistema de Avaliação da Educação Básica (SAEB), realizada pelo Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira (INEP), referente ao ano de 2005. Os resultados sugerem a existência de um efeito positivo do aumento da jornada escolar sobre o desempenho escolar. A ampliação de quatro para cinco horas na jornada diária dos estudantes está associada a um aumento de 8,36 pontos na proficiência em matemática ou, equivalentemente, um movimento de 0,20 desvios-padrão na distribuição de notas. No caso do tamanho da classe, o efeito estimado de uma redução de 38 para 30 é de 10,67 pontos, um movimento de 0,26 desviospadrão na distribuição de proficiência. Num segundo momento, os custos associados a essas duas políticas foram avaliados, em contraposição aos benefícios, por meio de uma análise de custo-efetividade. Os resultados indicaram que a ampliação da jornada escolar de quatro para cinco horas diárias possui a maior razão benefício/custo, comparativamente às políticas de redução do tamanho da classe, quando as classes têm 33 alunos ou menos. Para classes maiores, a política de redução das classes é mais custo-efetiva. / This dissertation aims to contribute to the identification of the causal effect of class size reduction and school hours increase on student achievement. It focuses on Brazilian basic education students from public schools located at urban areas. The Generalized Propensity Score Matching methodology for continuous treatment and the Nearest Neighbor Matching with bias correction were applied to estimate the treatment effect of class size reduction and school hours increase, respectively. The dataset used in the analysis was the National Basic Evaluation System (SAEB) carried out by the National Institute for Educational Studies and Research Anísio Teixeira (INEP) in 2005. The results suggest that there is a positive school hours effect on student achievement. When the instruction time is increased from four to five hours, the corresponding increase in math test scores is 8.36 points or, equivalently, a 0.20 standard deviation movement on the grade distribution. In addition, the estimated class size effect of a reduction from 38 to 30 students is 10.67 points or, equivalently, a 0.26 standard deviation movement on the grade distribution. After estimations, a cost-effectiveness analysis was carried out in order to confront costs and benefits of these interventions. The results show the school hours increase is more cost-effective when class size is 33 or smaller. Otherwise, the class size reduction is more cost-effective.

This dissertation contains three essays in the economics of development. The first essay investigates the effects of the decentralization of governance over education to communities in terms of individual education outcomes. The next essay relates to the first by exploring the factors that drive communities to adopt decentralized governance, including forms of decentralized governance over education. The last essay returns to the topic of education by examining a policy aimed at decreasing the costs of post-secondary education for a minority group. Each essay probes these topics within the context of First Nations in Canada. The first essay examines the substantial impacts of education decentralization on high school attendance and completion through the analysis of First Nation education self-government agreements in Canada. These agreements are important institutional arrangements that transfer the authority over education from the federal government to First Nations. I exploit confidential microdata and exogenous variation in the implementation of education self-government agreements to perform the analysis. My results indicate that self-government agreements focused exclusively on education increase high school attendance by 5 to 9 percentage points and high school completion by 3 to 5 percentage points. However, the effects on high school completion rates under multi-sectoral self-government agreements implemented together with comprehensive land claim agreements and for self-government agreements that focus on education alone differ dramatically for women and men. High school completion improves by 8 to 11 percentage points for women, but drops by a staggering 17 to 25 percentage points for men. These results have important policy implications for education decentralization in general, along with implications for the particular case of First Nation education self-governance in Canada. The second essay identifies the determinants of decentralized governance by exploring the First Nation self-government agreement claim and implementation processes. I use a novel dataset on self-government agreements and confidential microdata to perform the analysis. My results support the notion that we can treat self-government treatment variables as exogenous, when controlling for reserve fixed effects. This is not an onerous condition to impose. Specifically, I do not find any factors of economic or statistical significance for claims for my richest and most-preferred specification, which includes controlling for reserve fixed effects. Contrary to the results for claims, I find that education and income are important factors for implementation, but only conditional on a reserve having previously made a claim. However, this significance disappears, once I relax this condition and compare the determinants of implementation against reserves that may or may not have made a claim. The third essay examines the substantial impacts of a targeted policy that provides postsecondary tuition and living expense subsidies for Aboriginal Canadians. To identify the effects of the policy, I exploit a reform of the policy's eligibility requirements in 1985 that lead to a large increase in the number of individuals with access to the subsidies. My results indicate that the reform lead to economically and statistically significant increases in the likelihood of attaining any post-secondary education for a group of women whose eligibility was particularly targeted by the reform and for women generally. These increases range from about 4 to 7 percentage points. The effects for men are positive, but much smaller and not significant.

<p>My three chapters are all related to the study of immigrants in how they impact the US</p> <p>economy. The first two chapters look at international students in particular and how they</p> <p>impact their domestic peers and the local college towns they reside in. The third chapter</p> <p>looks at immigrant workers and their effect on native workers’ propensity to consolidate to</p> <p>form labor unions.</p> <p>To be specific, the first chapter, titled How International students Affect Domestic Students’</p> <p>Achievement: evidence from the OPT STEM-extension, looks at the role of immigrants</p> <p>in shaping the educational outcome of domestic students pursuing STEM degrees</p> <p>in the United States. By utilizing the mass influx of international students after an immigration</p> <p>policy change (OPT-STEM-extension) in 2008, I investigate the peer effects that</p> <p>international students have on grades, attrition, and first-year salary of STEM graduates.</p> <p>I account for the common selection issues present in the peer-effects literature by looking</p> <p>at the yearly exogenous change in international student share in a specific course-instructor</p> <p>pair and controlling for rich individual ability and demographics. This was made possible</p> <p>by having access to administrative data of a land-grant university with one of the highest</p> <p>international student enrollments in the US. I find that international students tend to lower</p> <p>grades and persistence of domestic students in STEM. Still, this negative effect is more than</p> <p>compensated for in the increase in salary due to spill-over effects in learning for those who</p> <p>persist and graduate.</p> <p>My research aims to eventually aid policymakers in both the local educational institutions</p> <p>and the federal government. To this end, I have extended my analysis of international</p> <p>students by shifting my focus outside the classroom to the local economies of the college</p> <p>campuses. In my second chapter, titled International Students’ Effect on Local Businesses, I</p> <p>use the zip code-level Census data on small businesses to see how the influx of international</p> <p>students affected the regional college campuses. I find that international students have a</p> <p>significantly positive effect on job creation in the local economy. To my knowledge, this is</p> <p>the first data-driven-causal analysis of international students on local businesses in the US.</p> <p>12</p> <p>My third chapter is a co-authored work with Alex Nowrasteh and Artem Samiahulin</p> <p>titled Immigrants Reduce Unionization in the US. Here we attempt to relate immigrants to</p> <p>a more traditional labor economics topic: labor unions. Although there is a vast amount of</p> <p>literature on unions, we found that the literature that causally estimates immigrants’ effect</p> <p>on unions is severely lacking in the US setting. Using a combination of representative data</p> <p>such as the CPS, Census, and the ACS, we show that immigrants accounted for about onethird</p> <p>of the decline in unions since the 1980s. We based our paper on the theoretical model</p> <p>of Naylor and Cripps  1993  and borrowed George Borjas’s skill-cell method for our empirical</p> <p>method.(Borjas  2003 )</p>

Thesis advisor: Zhushan Li / Educational public private partnerships (PPP), referring to the shared delivery of education services by the government and private providers, have been increasing in recent decades, especially in low- and middle-income countries (LMIC). Yet to date, there has been limited research on their role in the education landscape, in part due to the difficulty of classifying PPP schools in large-scale datasets, which typically classify schools as either public or private. In addition, few studies have assessed PPPs and school quality assurance indicators typically associated with them.  The study had two purposes. First, to explore the possibility of classifying PPP schools in a large-scale dataset using a statistical method. And second, to use these classifications to examine the differences between PPP, public, and private schools on school quality assurance measures, including but not limited to achievement. These analyses were performed using data from the Program for International Student Assessment (PISA), with schools from six of the global emerging economy countries: Brazil, China, Indonesia, Mexico, Russia and Turkey.  Schools were classified using a two-step clustering method using funding and management variables. This revealed three good-quality clusters with a silhouette measure of cohesion and separation of 0.6 (IBM, 2015b; Wendler & Gröttrup, 2016). These were classified as public, private, and PPP based on the characteristics of each school type. With these classifications, the study assessed the relationship between school type and achievement in mathematics, science and reading, and 24 school quality assurance measures from PISA. The analyses controlled for school resources and socio-economic and cultural status. The study found that overall, PPP schools performed better than public schools on three indicators, and better than private schools on five indicators; public schools performed better than PPP schools on one outcome and better than private schools on three outcomes, although with mostly small effect sizes. Private schools did not outperform other school types on any outcome. A country wise analysis showed that these results differed by country. The study highlights the possibility of using two-step clustering to identify PPP schools, the effects of shared funding and management on school performance, and the importance of context in examining countries’ education policies. / Thesis (PhD) — Boston College, 2021. / Submitted to: Boston College. Lynch School of Education. / Discipline: Educational Research, Measurement and Evaluation.

The financial provisions of the HE Act (2004) were intended to introduce market forces into the relationship between higher education institutions in England and their full-time Home/EU undergraduates. The policies that underpinned that legislation were established by Parliamentarians during a period of intense public and political debate which accompanied the passage of the Act and now, as suppliers in a nascent quasi-market, universities are de facto responsible for their delivery. With that market beginning to stabilise, this research compares those political objectives with observed outcomes of the introduction of the Act. Primary data has been collected through semi-structured interviews with key decision-makers in six sample universities – chosen to reflect both the spread of institutions in the sector and the range of pricing policies in operation – and with those involved nationally in shaping the legislation. From this material, supported by secondary quantitative and qualitative evidence, university pricing strategies are considered, in the context of theories of marketing and of higher education management, to provide an understanding of how institutions have structured their financial offerings with the aim of targeting specific markets for applicants. Data from the sample institutions is then used to build a profile of the quasi-market that the suppliers within it are generating. Even though members of the sample have taken diverse approaches to price-setting, there are some clear consistencies that typify this emerging national market: prices are set through adjustments to bursaries, not fees; and complex financial offerings have created barriers to effective communication. Furthermore, as institutional managers analyse the effects of their own pricing strategies locally, they are observing these trends and thus, where any changes are being made, these tend towards simplification of bursary schemes and hence increasing homogeneity across the sector. However, with the (index-linked) £3000 cap which currently applies to fees, it is increasingly apparent that the current quasi-market for full-time Home/EU undergraduates at English universities has not reached its price-sensitivity point and hence, this research argues, the market is not operating fully. Therefore, while this study offers an understanding of motivations behind current institutional actions and the nature of the resulting quasi-market, it also explains why it is not feasible to extrapolate from this information to forecast how the market might work were regulatory parameters to be changed significantly.