Si de nombreux efforts ont été déployés pour améliorer l’accès à l’éducation en Afrique subsaharienne, la proportion d’enfants qui interrompent leur scolarité avant la fin du cycle primaire reste encore très élevée. Pour tenter de répondre à ce défi majeur, cette thèse se propose d’examiner les déterminants de la demande d’éducation dans un environnement rural exposé à de nombreux risques. Le premier chapitre s’intéresse à la place essentielle, et pourtant peu étudiée, que tiennent les coûts d’opportunité du temps des enfants dans les choix éducatifs. Il s’avère dans ce cas nécessaire d’estimer la productivité du travail des enfants afin d’identifier les coûts susceptibles de compromettre leur scolarisation. Afin de mieux appréhender les choix en matière d’éducation, le deuxième chapitre dresse quant à lui une estimation des bénéfices de l’éducation dans un contexte agricole qui se distingue à la fois par un accès limité aux nouvelles technologies et par la prédominance d’exploitations familiales de petites tailles. Enfin, le troisième chapitre interroge les effets des chocs de productivité sur les décisions de scolarisation et sur les performances scolaires des enfants. Deux critères retiennent notre attention, l'âge auquel les enfants sont confrontés à ces chocs et leur intensité. Ce sujet nous semble d’autant plus pertinent que la fréquence de ces chocs ne cesse aujourd’hui d’augmenter. À travers ces trois chapitres centrés sur la Tanzanie rurale, cette thèse offre ainsi un aperçu du rôle des politiques publiques dans la protection et le développement de l’éducation. / Despite numerous investments that have been made to increase access to education in sub-Saharan Africa, a noteworthy share of children drop out of school prior to completing primary education. To address this issue, this thesis examines the factors that drive education decisions in a rural risky environment. The first chapter focuses on one of the core determinant of education investment that has been under-explored, the opportunity costs of education. To identify these costs that can significantly hinder education, we determine children's productivity on the farm and provide an estimate range of the value of one day of child labor. In order to better understand education decisions in rural sub-Saharan Africa, the second chapter assesses the different benefits of education in rural Tanzania, where family farm is the dominant structure in agriculture and where the technology level is low. Finally, the third chapter investigates whether productivity shocks are detrimental to educational achievement and children’s cognitive skills by considering two particular aspects, the age at which shocks occur, and the length of shocks. This subject is all the more relevant today when the number of productivity shocks is growing. Throughout these three chapters which focus on rural Tanzania, this thesis provides some insight into the role of public policies in protecting and promoting education.

Using a three-essay approach, I focus on two issues related to the labor market: the effect of changes in regulatory costs on informal sector employment, and the role of endogeneity in the relationship between education and earnings. In the first essay, I analyze the implications of regulatory costs on skill-based wage differences and informal sector employment. I use a two sector matching model with exogenous skill types for workers where firms have sector-specific costs and workers have sector-specific bargaining power. In general, there are multiple equilibria possible for this model. I focus on the equilibrium that best resembles the situation in the developing countries of sub-Saharan Africa and southern Asia. My results show that government policies which reduce regulatory costs decrease unemployment, earnings inequality, and the fraction of skilled workers in the informal sector. The different types of regulatory costs affect the skill premium differently and non-monotonically. In the second essay, I test the hypothesis of linearity in returns to education in the Mincer regression with endogenous schooling and earnings. I estimate the marginal rate of return to education using a polynomial model and a semiparametric partial linear model based on the standard Mincer regression. To perform the analysis, I use a control function approach for IV estimation with spousal and parental education as instruments. Results suggest that estimates not accounting for endogeneity understate returns at the tails of the education spectrum and overstate returns for education levels between middle-school and college. In the third essay, I empirically test the claim of Mookherjee and Ray (2010), based on a theoretical model of skill complexity, that "the return to human capital is endogenously nonconcave." I estimate the functional form of returns to education for India using a semiparametric partial linear model based on the standard Mincer regression. Marginal returns are estimated to test the nonconcavity of the functional form under both exogenous and endogenous schooling assumptions. My results show that the marginal rate of return declines during primary education and increases until high school, followed by stable returns for college and higher studies. However, the test of robustness of the functional form based on uniform confidence bands fails to reject the presence of nonconcavity in returns to education for India. This lends support to the claim of Mookherjee and Ray (2010).

Der Einfluss von Bildung gewinnt gesellschaftlich und politisch an Bedeutung. Auch im wissenschaftlichen Bereich zeigt sich dies über eine vielseitige Diskussion zum Einfluss von Bildung auf das Einkommen. In dieser Arbeit werden nationale und regionale Disparitäten in der monetären Wertschätzung von allgemeinem Humankapital aufgedeckt und diskutiert. Dafür werden verschiedene Verfahren diskutiert und basierend darauf Intervalle für die mittleren Bildungsrenditen bestimmt. Im ersten Abschnitt wird die Thematik theoretisch über zwei verschiedene Modellansätze fundiert und kritisch diskutiert. Anschließend folgt die Darstellung des aktuellen empirischen Forschungsbestands. Der Hauptteil der Arbeit beginnt mit der Darstellung des verwendeten Datensatzes und seiner kritischen Repräsentativitätsprüfung. Eine nähere Variablenbeschreibung mit deskriptiver Analyse dient zur Erklärung der verwendeten Größen. Darauffolgend werden bestehende Verfahren zur Schätzung von Bildungsrenditen diskutiert. Unter ausschließlicher Berücksichtigung der Erwerbstätigen zeigt das 3SLS-Verfahren die besten Eigenschaften. Bezieht man jedoch alle Erwerbspersonen in die Analyse mit ein, so erweist sich das Heckman-Verfahren als sehr geeignet. Die Analyse - zunächst auf nationaler Ebene - bestätigt weitestgehend die bestehenden Erkenntnisse der Literatur. Eine Separierung des Datensatzes auf verschiedene Alterscluster, Voll- und Teilerwerbstätige sowie Erwerbstätige in der Privatwirtschaft und im öffentlichen Dienst zeigen keine signifikanten Unterschiede in der Höhe der gezahlten durchschnittlichen Bildungsrenditen. Anders verhält es sich bei der regionalen Analyse. Zunächst werden Ost- und Westdeutschland separat betrachtet. Für diese erste Analyse lassen sich über 95 %-Konfidenzintervalle deutliche Unterschiede in der Höhe der Bildungsrenditen ermitteln. Aufbauend auf diese Ergebnisse wird die Analyse vertieft. Eine Separierung auf Bundesländerebene und ein weiterer Vergleich der Konfidenzintervalle folgen. Zur besseren statistischen Vergleichbarkeit der Ergebnisse wird neben dem 3SLS-Verfahren, angewendet auf die separierten Datensätze, auch ein Modell ohne die Notwendigkeit der Separierung gewählt. Hierbei ist die Variation der Regionen über Interaktionsterme berücksichtigt. Dieses Regressionsmodell wird auf das OLS- und das Heckman-Verfahren angewendet. Der Vorteil hierbei ist, dass die Koeffizienten auf Gleichheit getestet werden können. Dabei kristallisieren sich deutlich unterschiedliche Bildungsrenditen für Mecklenburg-Vorpommern, aber auch für Sachsen-Anhalt und Thüringen im Vergleich zu den restlichen Bundesländern Deutschlands heraus. Diese Länder zeichnen sich durch eine besonders hohe jährliche Verzinsung von allgemeinem Humankapital aus. Es folgt eine Diskussion über mögliche Ursachen für die regional verschiedenen Bildungsrenditen. Dabei zeigt sich, dass in den Bundesländern mit hoher Rendite das mittlere Einkommensniveau und auch das durchschnittliche Preisniveau tendenziell geringer sind. Weiterhin wird deutlich, dass bei höheren relativen Abweichungen der durchschnittlichen Einkommen höhere Renditen zu verzeichnen sind. Auch die Wanderungsbewegungen je nach Qualifikation unterscheiden sich. Unter zusätzlicher Berücksichtigung der Arbeitslosenquoten zeigt sich in den Ländern mit hoher Rendite eine tendenziell höhere Arbeitslosigkeit. Im zusammenfassenden Fazit der Arbeit werden abschließend die Erkenntnisse gewürdigt. Dabei ist zu bemerken, dass der Beitrag einen Start in die bundesländerweite Analyse liefert, die eine Fortführung auf beispielsweise eine mehrperiodische Betrachtung anregt. / The impact of education becomes politically and for the society more important. In research area the higher relevance is indicated by a distinctive discussion about, for example, the influence from education to earnings. This thesis shows national and regional disparities with respect to the monetary appreciation to general human capital. Therefore different econometric methods are discussed and based on that, intervals for the returns to education are estimated. The first chapter presents the theoretical fundament using two different models, which are critically discussed in the text. Afterwards the existing empirical literature is shown. The main part starts with a description of the dataset and a discussion about its representativeness. A closer examination to the variables follows using descriptive statistics and verbal explanations. The estimation part starts with a discussion about common methods to measure returns to education. Looking at employees exclusively, 3SLS gives best performance. However, by extension the analysis to the overall manpower Heckman-method is best. The national analysis confirms the results from the existing literature. Within the next step the dataset is separated to have a closer look at different cohorts, full- and part-time worker, as well as public- and private-sector worker. Comparing these groups with each other the estimated returns to education are not statistically different. However, results change by estimating regional returns to education for Germany. First the dataset is splitted into two regions, the eastern- and the western-region. Comparing the estimated 95 %-confidence intervals, differences are significant between both regions in Germany. Hence, it is worth to intensify the regional analysis. The next step is a further separation of the dataset to a federal-state-analysis. Again, the comparisons of the resulting confidence intervals show partially no overlapping confidence intervals. Separating the dataset gives no possibility to compare the coefficients with statistic tests. Because of that a new regression model without the necessity of separation is introduced. For this purpose the variation of the regions are included by interaction terms. This model can be estimated using OLS- and Heckman-method. The advantage of this procedure is that the schooling-coefficients can be tested on statistical equality. Irrespective which method is used, differences in the return to education can be estimated for Mecklenburg-West Pomerania especially and for Saxony-Anhalt and Thuringia. These three states tend to a higher monetary return to general human capital compared to the other federal states. Thereafter a discussion follows about possible causes for the estimated regional heterogeneous returns to education. It is shown that federal states with a high return to education lean towards a lower middle level of income and a lower price level on average. In addition to that, there is a relation between a higher relative deviation from average income and higher returns to education. Looking at migration movements between federal states, it is shown that there are differences in the qualification levels. Moreover, regions with a higher unemployment rate tend to result in higher returns as well. The summarizing conclusion appreciates the scientific findings. For that it has to be said, that this work gives a start for an analysis which should have a closer look to different parts of Germany. It gives an incentive to pursue with, for example, an analysis for more than one year.

Studies on returns to education disagree as to where these returns are highest. It is suggested that these disagreements are the result of inconsistencies in the data and methodologies used for their estimation. These disagreements specifically refer to where in education governments should invest in order to obtain the highest returns, based on the specific characteristics of their countries (i.e. level of economic development). The purpose of this dissertation is to estimate and determine whether returns to investments in education vary for groups of countries with different levels of economic development. Rates of return to investments in education were estimated through improvements in methodology and data comparability. Subsequently, the differences among these returns and their significance were observed. This study provides evidence to suggest where in education countries should invest based on their specific level of economic development in order to obtain the highest returns to these investments.

We assess empirically the changes in returns to education at the subnational level in Uganda using the Uganda National Household Surveys for 2002/2003 and 2005/2006. Our results indicate that average returns to schooling tended to converge across regions in the last decade. The overall trend in convergence of returns to schooling took place at all levels of educational attainment and this behaviour in returns to education is mostly driven by the dynamics of returns to schooling in urban areas. We analyse subnational convergence in returns to education and unveil deviant dynamics in Northern Uganda. We discuss the potential challenges to inclusive economic growth in Uganda which are implied by our results. (authors' abstract) / Series: Department of Economics Working Paper Series

As one of the fastest growing regions in the world, crop production and education remain two of the most important topics for the development of sub-Saharan African (SSA) countries. This dissertation is composed of three chapters that investigate the economic returns to education (Chapter 1 and 2) and assess the policy influence on fertilizer usage (Chapter 3) in two SSA countries, Zimbabwe and Ethiopia. Chapter 1 investigates the casual impact of improved educational attainment on household well-being as reflected by consumption level in Zimbabwe. We use the age-specific exposure to the 1980 education reform as the instrument for the household head educational attainment to identify the economic returns to education. We find that an extra year of household head schooling leads to an 8% increase in per capita household consumption on average when using the multiple rounds of the Income, Consumption and Expenditure Survey (ICES). The impact of enhanced education on household consumption is larger for rural and female-headed households and we also find some evidence that head educational attainment could affect consumption patterns, where additional schooling leads to slightly lower consumption share in food and higher share in non-durable goods. Chapter 2 extends this topic by utilizing a pseudo panel data constructed with multiple waves of repeated cross-sectional data, which allows us to use fixed-effect and other panel data methods to address the problem of unobserved "ability" bias. For pseudo panel, we use age, gender and some other time-persistent criterions to define the cohorts and replace the individual observations with the intra-cohort means. Individual time-invariant factors that influencing both education and consumption are transformed into cohort time-invariant factors, within transformation on the pseudo panel would eliminate such factors leads to achieve unbiased and consistent estimates on the returns to education. We find on average there is a 14% increase in monthly household per capita consumption for each one more year of education for the household head. By further disaggregating our population, we find female-headed households exhibit a return to education of around 15.3%, much higher than its corresponding OLS/IV estimates. On the other hand, we fail to detect such large discrepancy for the male-headed households, suggesting that the overall downward bias of OLS/IV estimates mostly come from female-headed households. Facing significant higher opportunity cost, Zimbabwean females are much less likely to furthering their education when compared to males with similar unobserved ability level which can be one of the major underlying reasons. Chapter 3 investigates the potential effect of fertilizer promotion polices on crop acreage and input intensities in Ethiopia. We use a fully calibrated multi-input and -output model based on the principle of positive mathematical programming (PMP) to assess the policy impact in four major agricultural states in the country. I analyze two policies designed to promote fertilizer use, namely fertilizer import expansion and a universal subsidy program. The results from the simulation model suggest that local farmers actively respond to these promotion policies by adjusting crop acreage and investing more in fertilizer input. However, when the availability of fertilizer in one region is fixed and local farmers face a binding constraint, the behavior responses to the subsidy program alone would be limited. / Doctor of Philosophy / Education and food production are two of the most important issues when we study the development in Sub-Saharan African countries, which are among the fastest developing regions in the world. The dissertation is composed of three manuscripts, aiming to evaluate the economic returns to education and the impact of fertilizer promotion policies in two of the SSA countries, Zimbabwe and Ethiopia. Chapter 1 investigates the returns to education as reflected by household consumption and finds significant positive effect of enhanced education on household well-being. We also find such effect is larger for rural and female-headed households which shed light on the policy of more public investment targeting female and rural education in developing countries. Chapter 2 further extends this topic by combining multiple rounds of survey data and finds larger educational effects on household consumption compared to the results in Chapter 1. Female household heads, facing more barriers in attaining higher education, are an important cause of the higher estimates of returns found in this chapter. Chapter 3 investigates how potential fertilizer promotion polices would affect the regional level of choices on crop acreage and fertilizer input intensities in the major agricultural states in Ethiopia. It finds local farmers will actively adjust their land and fertilizer inputs when facing a fertilizer import expansion in combined with a universal fertilizer subsidy program.

El primer capítulo de la tesis sobre "Estudios Empíricos sobre Capital Humano e Instituciones" presenta estimaciones de retornos a la educación en Alemania y analiza los determinantes de las preferencias educativas. El segundo capítulo utiliza estimaciones de retornos a la experiencia en el país de origen de inmigrantes en Estados Unidos para proporcionar evidencia empírica que el comercio aumenta la acumulación de capital humano en el trabajo, incluso en los países menos desarrollados, resolviendo la ambigüedad teórica si el comercio aumenta o disminuye "learning-by-doing". La acumulación de capital humano en el trabajo es también positivamente asociada con el PIB per capita, un alto nivel de educación y una mayor calidad de políticas e instituciones. El último capítulo analiza el efecto de recursos naturales y asistencia externa sobre la calidad de instituciones, proporcionando evidencia empírica que -contrario a la asistencia externa, la abundancia de mineral y combustible tiende a ser asociada negativamente con la calidad de instituciones si la fragmentación étnica es grande. / The first chapter of the thesis on "Empirical Studies on Human Capital and Institutions" presents estimates of returns to education in Germany and analyses the determinants of educational choices. The second chapter uses estimated returns to home country experience of US immigrants to provide empirical evidence that trade increases on-the-job human capital accumulation even in less developed countries, thereby resolving the theoretical ambiguity whether trade increases or decreases learning-by-doing. Similar to trade, GDP per capita, a high average level of educational attainment and stronger quality of policy and institutions are found to be positively associated with on-the-job human capital accumulation. The last chapter analyses the effect of natural resources and aid on institutions, providing empirical evidence that contrary to aid, mineral and fuel abundance tends to be associated with significantly lower quality of institutions if ethnic fractionalization is large.

En économie du développement, la culture d’évaluation s’est renforcée pour mieux orienter l’aide vers les actions qui ont démontré la preuve de leur efficacité. Dans ce contexte aux pratiques renouvelées, un essor des méthodes expérimentales s’est fait jour dans le domaine de l’évaluation, notamment à travers la méthode des évaluations aléatoires qui est fortement promue par le laboratoire J-PAL, et notamment par Esther Duflo. Contribuant empiriquement à la constitution d’un « fonds scientifique capitalisable » grâce aux connaissances mises à jour, mais comportant des failles persistantes, nous procédons dans cette thèse à une analyse critique de la méthode, avec l’appui d’une récente littérature économique française et anglophone.De plus, nous avons mené une enquête par questionnaire auprès de parents d’élèves du primaire à Madagascar, notre terrain d’étude. Pour la réaliser, nous nous sommes basés sur une évaluation aléatoire qui avait été menée dans ce pays par une doctorante affiliée au J-PAL, et qui concluait à un impact positif d’un traitement à base d’information statistique sur les rendements de l’éducation, délivrée aux parents d’élèves du primaire. En reproduisant ce traitement dans notre enquête, nous avons voulu savoir si l’on pouvait aboutir à des résultats similaires bien que la méthode employée varie. Nous retrouvons globalement les conclusions de l’auteure de l’évaluation aléatoire, bien que nous utilisions une méthode moins purement quantitative. / In development economics, the culture of evaluation is being strengthened to specify which development actions could be effective, and thus allows a better targeting of aid on the actions that have proven their effectiveness. Within this context of renewed practices, an expansion of experimental methods in the evaluation field has emerged, in particular through the method of randomized evaluations strongly promoted by the J-PAL laboratory, particularly by the emblematic figure of Esther Duflo. Contributing empirically to the composition of a « scientific capitalizable fund » thanks to the updated knowledge, but still including persistent limits, we set into this thesis about a critical analysis of this method, with the mobilization of some recent French and English economic literature.Moreover, we conducted a questionnaire survey to primary school pupils’ parents in Madagascar, our field study. To carry out it, we used a randomized evaluation that had been conducted in this country by a PhD student affiliated with the J-PAL and whose finding was the positive effect of some statistical information treatment on returns to education towards primary school pupils’ parents. By reproducing this treatment in our study, we wanted to know whether it was possible to achieve similar results even though the applied method varies. It’s globally matching up with the conclusions of the author of the randomized evaluation, even if we use a less purely quantitative method.

Cette thèse de doctorat porte sur la croissance «exclusive» et la vulnérabilité qui caractérisent le développement du Vietnam et plus largement de l'Asie aujourd'hui. Les chapitres traitent trois aspects importants de la vulnérabilité et du développement inclusif, à savoir : l’informalité (chapitre 1), le dilemme de l'éducation (chapitre 2) et l'emploi atypique (chapitre 3). Les contributions de ce travail reposent sur différents facteurs : la nouveauté et la pertinence des sujets de recherche ; le large éventail de données utilisées, quantitatives et qualitatives, y compris les Enquêtes sur les entreprises individuelles et sur le Secteur Informel au Vietnam, et les Enquêtes Nationales sur l’Emploi de divers pays d'Asie ; et enfin l’originalité de la méthodologie. Le chapitre 1 étudie l'hétérogénéité du secteur informel au Vietnam, en se basant sur une approche quali-quantitative unique. Le chapitre 2 se concentre sur la variation des rendements de l'enseignement supérieur dans la population vietnamienne en recourant à différents modèles d'estimation. Le chapitre 3 est la première étude qui étudie systématiquement les écarts de salaire induits par le statut de travail temporaire dans les pays en développement d'Asie. Dans l'ensemble, toute la thèse implique que le capital humain, l'emploi et le revenu sont des facettes interdépendantes du bien-être individuel et que certains phénomènes de développement doivent être analysés dans leur hétérogénéité. / This PhD dissertation is dedicated to the issue of ‘exclusivist’ growth and vulnerability characterizing Vietnam as well as developing Asia today. The chapters address three important aspects of vulnerability and inclusive development, namely: Informality (chapter 1), Education dilemma (chapter 2) and Non-standard employment (chapter 3). The contribution of this work lies in the novelty and relevance of research topics; the wide range of data used, both quantitative and qualitative, including Household Business and Informal Sector Surveys in Vietnam, and national Labor Force Surveys of various countries in Asia; as well as the originality of methodology. Chapter 1 investigates the heterogeneity of the informal sector in Vietnam, based on a unique quali-quanti approach. Chapter 2 focuses on the variation of the returns to higher education across the Vietnamese population with different estimation models. Chapter 3 is the first study that systematically examines the wage differentials induced by temporary job status in Asian developing countries. Overall, the whole thesis implies that human capital, employment, and income are interrelated facets of individual well-being, and that some development phenomena should be analyzed in their heterogeneity.

No description available.