The long-run and intergenerational education impacts of intergovernmental transfers

This paper provides regression discontinuity evidence on long-run and intergenerational education impacts of a temporary increase in federal transfers to local governments in Brazil. Revenues and expenditures of the communities benefiting from extra transfers temporarily increased by about 20% during the 4 year period from 1982 to the end of 1985. Schooling and literacy gains for directly exposed cohorts established in previous work that used the 1991 census are attenuated but persist in the 2000 and 2010 censuses. Children and adolescents of the next generation --born after the extra funding had disappeared-- show gains of about 0.08 standard deviation across the entire score distribution of two nationwide exams at the end of the 2000s. While we find no evidence of persistent improvements in school resources, we document discontinuities in education levels, literacy rates and incomes of test takers' parents that are consistent with intergenerational human capital spillovers.

On making causal claims: A review and recommendations

Social scientists often estimate models from correlational data, where the independent variable has not been exogenously manipulated; they also make implicit or explicit causal claims based on these models. When can these claims be made? We answer this question by first discussing design and estimation conditions under which model estimates can be interpreted, using the randomized experiment as the gold standard. We show how endogeneity--which includes omitted variables, omitted selection, simultaneity, common methods bias, and measurement error--renders estimates causally uninterpretable. Second, we present methods that allow researchers to test causal claims in situations where randomization is not possible or when causal interpretation is confounded, including fixed-effects panel, sample selection, instrumental variable, regression discontinuity, and difference-in-differences models. Third, we take stock of the methodological rigor with which causal claims are being made in a social sciences discipline by reviewing a representative sample of 110 articles on leadership published in the previous 10 years in top-tier journals. Our key finding is that researchers fail to address at least 66 % and up to 90 % of design and estimation conditions that make causal claims invalid. We conclude by offering 10 suggestions on how to improve non-experimental research.

Saving lives: Evidence from a conditional food supplementation program

Many governments in developing countries implement programs that aim to address nutrionalfailures in early childhood, yet evidence on the effectiveness of these interventions is scant. Thispaper evaluates the impact of a conditional food supplementation program on child mortality inEcuador. The Programa de Alimentaci?n y Nutrici?n Nacional (PANN) 2000 was implementedby regular staff at local public health posts and consisted of offering a free micronutrient-fortifiedfood, Mi Papilla, for children aged 6 to 24 months in exchange for routine health check-ups forthe children. Our regression discontinuity design exploits the fact that at its inception, the PANN2000 was running for about 8 months only in the poorest communities (parroquias) of certainprovinces. Our main result is that the presence of the program reduced child mortality in cohortswith 8 months of differential exposure from a level of about 2.5 percent by 1 to 1.5 percentagepoints.

Partisan targeting of inter-governmental transfers & state interference in local elections: evidence from Spain

[spa] En este trabajo examinamos si, en la asignación de transferencias, los gobernantes regionales discriminan a favor de los gobiernos locales controlados por el mismo partido político, y si las perspectivas electorales de los gobiernos locales mejoran si están políticamente alineados con el gobierno regional. Con una nueva base de datos que considera 3.000 municipios españoles durante el período 2000-07 y un diseño de discontinuidad en la regresión, documentamos un efecto robusto de importante magnitud: en elecciones ajustadas, los municipios alineados con el gobierno regional reciben, en media, un 83% más de transferencias per cápita y su gobernante obtiene un 10% más de votos en las elecciones locales. También demostramos que el efecto de la alineación política es mayor: (i) si las elecciones regionales y locales se celebran el mismo día, (ii) en regiones donde las elecciones regionales son menos competidas, y (iii) en regiones con más recursos presupuestarios.

Partisan targeting of inter-governmental transfers & state interference in local elections: evidence from Spain

[spa] En este trabajo examinamos si, en la asignación de transferencias, los gobernantes regionales discriminan a favor de los gobiernos locales controlados por el mismo partido político, y si las perspectivas electorales de los gobiernos locales mejoran si están políticamente alineados con el gobierno regional. Con una nueva base de datos que considera 3.000 municipios españoles durante el período 2000-07 y un diseño de discontinuidad en la regresión, documentamos un efecto robusto de importante magnitud: en elecciones ajustadas, los municipios alineados con el gobierno regional reciben, en media, un 83% más de transferencias per cápita y su gobernante obtiene un 10% más de votos en las elecciones locales. También demostramos que el efecto de la alineación política es mayor: (i) si las elecciones regionales y locales se celebran el mismo día, (ii) en regiones donde las elecciones regionales son menos competidas, y (iii) en regiones con más recursos presupuestarios.

Essays in labor and public economics

Dans ma thèse, je me sers de modèles de recherche solides pour répondre à des questions importantes de politique publique. Mon premier chapitre évalue l’impact causal de l’allégeance partisane (républicain ou démocrate) des gouverneurs américains sur le marché du travail. Dans ce chapitre, je combine les élections des gouverneurs avec les données du March CPS pour les années fiscales 1977 à 2008. En utilisant un modèle de régression par discontinuité, je trouve que les gouverneurs démocrates sont associés à de plus faibles revenus individuels moyens. Je mets en évidence que cela est entrainée par un changement dans la composition de la main-d’oeuvre à la suite d’une augmentation de l’emploi des travailleurs à revenus faibles et moyens. Je trouve que les gouverneurs démocrates provoquent une augmentation de l’emploi des noirs et de leurs heures travaillées. Ces résultats conduisent à une réduction de l’écart salarial entre les travailleurs noir et blanc. Mon deuxième chapitre étudie l’impact causal des fusillades qui se produisent dans les écoles secondaires américaines sur les performances des éléves et les résultats des écoles tels que les effectifs et le nombre d’enseignants recruté, a l’aide d’une stratégie de différence-en-différence. Le chapitre est coécrit avec Dongwoo Kim. Nous constatons que les fusillades dans les écoles réduisent significativement l’effectif des élèves de 9e année, la proportion d’élèves ayant un niveau adéquat en anglais et en mathématiques. Nous examinons aussi l’effet hétérogene des tueries dans les écoles secondaires entre les crimes et les suicides. Nous trouvons que les fusillades de natures criminelles provoquent la diminution du nombre d’inscriptions et de la proportion d’élèves adéquats en anglais et mathématiques. En utilisant des données sur les élèves en Californie, nous confirmons qu’une partie de l’effet sur la performance des élèves provient des étudiants inscrits et ce n’est pas uniquement un effet de composition. Mon troisième chapitre étudie l’impact des cellulaires sur la performance scolaire des élèves. Le chapitre est coécrit avec Richard Murphy. Dans ce chapitre, nous combinons une base de données unique contenant les politiques de téléphonie mobile des écoles obtenues à partir d’une enquète auprès des écoles dans quatre villes en Angleterre avec des données administratives sur la performance scolaire des éleves. Nous étudions ainsi l’impact de l’introduction d’une interdiction de téléphonie mobile sur le rendement des éleves. Nos résultats indiquent qu’il y a une augmentation du rendement des éleves après l’instauration de l’interdiction des cellulaires à l’école, ce qui suggère que les téléphones mobiles sont sources de distraction pour l’apprentissage et l’introduction d’une interdiction à l’école limite ce problème.

Essays in political economy

Cette thése relie trois articles sur l'économie politique. Ces articles analysent à la fois théoriquement et empiriquement si, et dans quelle mesure, trois phénomènes politiques différents (les partis politiques, les guerres civiles et les menaces externes), et leur interaction, influent sur les résultats économiques. Le premier chapitre étudie l'impact de la présence au pouvoir des politiciens de nouveaux partis politiques sur la taille du gouvernement. Le chapitre se concentre sur les municipalités colombiennes, où les nouveaux partis politiques ont été nombreux et fructueux au cours des dernières années. Les estimations par régressions sur discontinuité montrent que les dépenses publiques et les recettes fiscales sont significativement plus élevées dans les municipalités gouvernées par un maire d'un nouveau parti politique. En utilisant des informations sur la politique locale et des caractéristiques des nouveaux partis, je soutiens que ce résultat peut être expliqué par le fait qu'il y a moins d'information sur les politiciens de nouveaux partis que les politiciens des partis traditionnels. Le deuxième chapitre développe une nouvelle explication de l'impact des guerres civiles et des conflits interétatiques sur le state-building qui repose sur l'idée que les protagonistes de ces deux types de conflits peuvent avoir un lien (ethnique ou idéologique). Un premier résultat montre que la force de ce lien détermine si les conflits contre des adversaires internes (i.e. guerres civiles) ou des ennemis externes (i.e. conflits interétatiques) sont complémentaires ou se substituent, conduisant à plus ou moins d'investissement en capacité fiscale. La théorie prédit également un rôle non trivial de la stabilité politique dans la relation entre les deux types de conflits et la capacité fiscale: un deuxième résultat montre que, bien que la stabilité politique se traduit par moins de capacité fiscale, plus de stabilité n'implique pas plus de state-building. Leur équivalence dépend du niveau de cohésion des institutions. Un nouveau mécanisme par lequel plus de stabilité politique peut impliquer moins de state-building est proposé. En outre, il est démontré que des corrélations dans les données cross-country sont compatibles avec la théorie. Le troisième chapitre examine la relation entre la probabilité d'occurrence d'un conflit intérieur violent et le risque qu'un tel conflit "s'externalise" (c'est à dire se propage dans un autre pays en devenant un conflit interétatique). Je considère une situation dans laquelle un conflit interne entre un gouvernement et un groupe rebelle peut s'externaliser. Je montre que le risque d'externalisation augmente la probabilité d'un accord de paix, mais seulement si le gouvernement est suffisamment puissant par rapport aux rebelles, et si le risque d'externalisation est suffisamment élevé. Je montre comment ce modèle aide à comprendre les récents pourparlers de paix entre le gouvernement colombien et le groupe le plus puissant des rebelles dans le pays, les FARC.

Cambios en el régimen de pensiones en Colombia y su efecto en el margen intensivo de la oferta laboral

En Colombia, como en muchos países de América Latina, en los años 80 y 90 se hicieron cambios importantes en los regímenes de pensiones. Este trabajo hace un análisis de uno de esos cambios en Colombia. El cambio consistió en aumentar el tiempo de cotización necesario para reclamar los beneficios pensionales y la inclusión del salario dentro de la fórmula del monto de pensiones. Para este propósito se estudia el impacto sobre la oferta laboral de un cambio exógeno en estas condiciones usando un diseño de regresión discontinua. Se encuentra un efecto positivo sobre las horas promedio trabajadas en la semana.

Parametric pension reform and the intensive margin of labor supply, evidence from Colombia

We analyze the effect of a parametric reform of the fully-funded pension regime in Colombia on the intensive margin of the labor supply. We take advantage of a threshold defined by law in order to identify the causal effect using a regression discontinuity design. We find that a pension system that increases retirement age and the minimum weeks during which workers must contribute to claim pension benefits causes an increase of around 2 hours on the number of weekly worked hours; this corresponds to 4% of the average number of weekly worked hours or around 14% of a standard deviation of weekly worked hours. The effect is robust to different specifications, polynomial orders and sample sizes.

Do more decentralized local governments do better? an evaluation of the 2001 decentralization reform in Colombia

In this paper I evaluate the impact of the 2001 decentralization reform in Colombia. I use data from Colombia's municipalities. I look at the effect of the 2001 reform on enrolment in pre-college schools. While all municipalities received earnmarked national transfers, withthe reform some of then now have more responsabilities to provide education (deeper decen-tralization) than others. Particulary important, the reform entitle the more decentralizedmunicipalities to sign subsidy contracts with private school. Departments (the regional gov-ernments) are entitle to sign this type of contracts for the less decentralized municipalities.Since the rule for municipalities to receive more responsabilities follows and exogenous popu-lation threshold, I can implement Regression Discontinuity Design. Enrolment is measuredthrough two variables: the number of students enroled in public schools and the number of subsidized students enroled in private schools. Results sugest that more decentralized mu-nicipalities subsidize more students in private schools. The difference is significant at all thelevels of pre-college school for the period 2004-2006. In 2005, the difference accounts for20% of enrolment in private schools and 3% of population of school age. Besides, there are not significant differences among municipalities regarding enrolment in public schools.

Congress composition and electoral advantage

In 2003, an electoral reform changed the mechanism to assign seats in the Colombian Congress. I simulate the 2006 Senate elections using the previous assignment mechanism to determine which senators benefited from the reform, i.e. would have not been elected had the reform not been made. With the results of the simulation, I use a regression discontinuity design to compare the senators that would have been barely elected anyways with those who would have lost, but were near to be elected. I check the differences in the amount of law drafts presented, the attendance to voting sessions, and a discipline index for each senator as proxy of their legislative behavior. I find that the senators benefiting from the reform present a different legislative behavior during the 4-year term with respect to the senators that would have been elected anyways. Since the differential legislative behavior cannot be interpreted as being better (worse) politician, I examine if the behavioral difference gives them an electoral advantage. I find no difference in the electoral result of 2010 Senate election in terms of the probability of being (re)elected in 2010, the share of votes, the share of votes within their party list, and the concentration of their votes. Additionally, I check the probability of being investigated for links with paramilitary groups and I find no differences. The results suggest that political reforms can change the composition of governing or legislative bodies in terms of performance, but it does not necessarily translate into an electoral advantage.

Exames curriculares e resultados educacionais: uma análise do exame nacional do ensino médio

Esta dissertação avalia os efeitos do Exame Nacional do Ensino Médio (ENEM) sobre o ensino médio brasileiro. Esta avaliação baseia-se na teoria de que exames curriculares com a qualidade de gerar conseqüências diretas sobre os alunos, como no caso do ENEM, são capazes de influenciar positivamente o aprendizado dos alunos por dois canais: por aumento de seu esforço e pela pressão que exercem sobre professores e diretores. O presente trabalho avalia especificamente o segundo canal, sobre o qual o ENEM deve atuar por meio da divulgação ao público das notas médias por escola em todo o país. Neste sentido, estima-se o efeito da divulgação das notas do ENEM sobre os seguintes resultados educacionais: proficiência em Matemática e Língua Portuguesa no Saeb, alocação de insumos escolares e comportamento dos professores. Para tanto, dois métodos são propostos: i) diferenças em diferenças, utilizando dados dos Saeb de 2005 – um ano antes do início da divulgação – e 2007 – um ano após a primeira publicação – e ii) regressão com descontinuidade, explorando o fato de haver uma regra estrita para entrada da escola na divulgação – ter ao menos 10 concluintes do ensino médio participando do ENEM. Os resultados mostram que o fato uma escola ter seu resultado médio no ENEM divulgado não parece influenciar o desempenho de seus alunos nem a quantidade de seus insumos escolares, por nenhum dos métodos propostos. O método de diferenças em diferenças aponta ainda que os professores de escolas submetidas ao tratamento dessa divulgação parecem ter maior enfoque em atividades voltadas a resolução de exercícios e fixação de regras gramaticais, reação que é interpretada por parte da literatura como adversa. Estes últimos resultados, no entanto, podem estar refletindo apenas tendências temporais preexistentes e não um efeito de tratamento legítimo. Conclui-se então que o ENEM não deve estar servindo como indutor de melhorias dos resultados educacionais via pressão dos alunos do ensino médio. Porém, é possível levantar a hipótese de que este tipo de efeito ainda pode ser captado em prazos mais longos ou para grupos de escolas diferentes das usadas na análise.

The effects of tax incentives for small firms on employment levels

This paper will examine the effects of tax incentives for small businesses on employment level evaluating a program with this purpose implemented in Brazil in the 1990s. We first develop a theoretical framework which guides both the de nition of the parameters of interest and their identi cation. Selection problems both into the treatment group and into the data sample are tackled by combining fixed effects methods and regression discontinuity design on alternative sub-samples of a longitudinal database of manufacturing firms. The results show that on the one hand the size composition of the treated fi rms may be changed due to the survival of some smaller firms that would have exited had it not been eligible to the program. On the other hand, the treated firms who do not depend on the program to survive do employ more workers.

O crime de drogas e a violência em São Paulo: uma análise a partir da Lei de Drogas

O objetivo desta pesquisa é estudar o efeito da Lei nº 11.343/06 (Lei de Drogas) sobre o crime de tráfico e porte de drogas e a relação entre crimes de drogas e outros crimes. Para isso, são exploradas as variações da Lei de Drogas, através de análises de regressões com descontinuidade e com variável instrumental, além de estimações com dados em painel, em busca de um efeito causal entre drogas e violência. Como resultados, a Lei de Drogas parece não ter efeito negativo significativo sobre crimes de drogas. Por outro lado, crimes de drogas apresentam uma associação negativa sobre crimes de furto e uma relação positiva com crimes de formação de quadrilha. Para cada redução de 100 crimes de drogas (por mil habitantes) associa-se um aumento de 3,6 crimes de furto (por mil habitantes) e uma diminuição de 27 crimes de formação de quadrilha (por mil habitantes). Não são encontrados efeitos robustos sobre roubos, homicídios, latrocínios, estupros, crimes de lesão corporal e porte de arma de fogo.

Test score disclosure and school performance

In this paper we test whether the disclosure of test scores has direct impacts on student performance, school composition and school inputs. We take advantage of the discontinuity on the disclosure rules of The National Secondary Education Examination (ENEM) run in Brazil by the Ministry of Education: In 2006 it was established that the 2005 mean score results would be disclosed for schools with ten or more students who took the exam in the previous year. We use a regression discontinuity design to estimate the e ects of test disclosure. Our results indicate that private schools that had their average scores released in 2005 outperformed those that did not by 0.2-0.6 in 2007. We did not nd same results for public schools. Moreover, we did not nd evidence that treated schools adjusted their inputs or that there was major changes in the students composition of treated schools. These ndings allow us to interpret that the main mechanism driving the di erences in performance was the increased levels of students', teachers' and principals' e ort exerted by those in schools that had scores publicized.