This report, rooted in the conflict over the control of natural resource wealth, departs from the widely-accepted findings of two disparate literatures. First, while recent analyses correctly conclude that natural resources rents play a contingent role in development, this study deviates from the conventional wisdom attributing the variation of the resource curse to formal institutions. Secondly, as opposed to the recent wave of “political insurance” arguments that ascribe the creation of reforms to weak incumbents attempting to tie the hands of their successors, I argue that actors pursue similar institutional reforms for economic and political reasons. I build on these literatures by examining the commitment to a specific government institution—stabilization funds, which manage the fluctuations of natural resource rents and stop natural resource wealth from being a curse—across three natural resource-rich Latin American countries: Chile, Mexico and Venezuela. Paradoxically, because successful stabilization funds provide greater political benefits when rents are saved, I argue that these institutions only tie the hands of political successors from using rents for political purposes when they are created for economic purposes. / text

This thesis addresses three major issues in data mining regarding feature subset selection in large dimensionality domains, plausible reconstruction of incomplete data in cross-sectional applications, and forecasting univariate time series. For the automated selection of an optimal subset of features in real time, we present an improved hybrid algorithm: SAGA. SAGA combines the ability to avoid being trapped in local minima of Simulated Annealing with the very high convergence rate of the crossover operator of Genetic Algorithms, the strong local search ability of greedy algorithms and the high computational efficiency of generalized regression neural networks (GRNN). For imputing missing values and forecasting univariate time series, we propose a homogeneous neural network ensemble. The proposed ensemble consists of a committee of Generalized Regression Neural Networks (GRNNs) trained on different subsets of features generated by SAGA and the predictions of base classifiers are combined by a fusion rule. This approach makes it possible to discover all important interrelations between the values of the target variable and the input features. The proposed ensemble scheme has two innovative features which make it stand out amongst ensemble learning algorithms: (1) the ensemble makeup is optimized automatically by SAGA; and (2) GRNN is used for both base classifiers and the top level combiner classifier. Because of GRNN, the proposed ensemble is a dynamic weighting scheme. This is in contrast to the existing ensemble approaches which belong to the simple voting and static weighting strategy. The basic idea of the dynamic weighting procedure is to give a higher reliability weight to those scenarios that are similar to the new ones. The simulation results demonstrate the validity of the proposed ensemble model.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

This thesis develops flexible and principled nonparametric learning algorithms to explore, understand, and predict high dimensional and complex datasets. Such data appear frequently in modern scientific domains and lead to numerous important applications. For example, exploring high dimensional functional magnetic resonance imaging data helps us to better understand brain functionalities; inferring large-scale gene regulatory network is crucial for new drug design and development; detecting anomalies in high dimensional transaction databases is vital for corporate and government security. Our main results include a rigorous theoretical framework and efficient nonparametric learning algorithms that exploit hidden structures to overcome the curse of dimensionality when analyzing massive high dimensional datasets. These algorithms have strong theoretical guarantees and provide high dimensional nonparametric recipes for many important learning tasks, ranging from unsupervised exploratory data analysis to supervised predictive modeling. In this thesis, we address three aspects: 1 Understanding the statistical theories of high dimensional nonparametric inference, including risk, estimation, and model selection consistency; 2 Designing new methods for different data-analysis tasks, including regression, classification, density estimation, graphical model learning, multi-task learning, spatial-temporal adaptive learning; 3 Demonstrating the usefulness of these methods in scientific applications, including functional genomics, cognitive neuroscience, and meteorology. In the last part of this thesis, we also present the future vision of high dimensional and large-scale nonparametric inference.

We develop statistical methods for tackling two important problems in genetic association studies. First, we propose a Bayesian approach to overcome the winner's curse in genetic studies. Second, we consider a Bayesian latent variable model for analyzing longitudinal family data with pleiotropic phenotypes. Winner's curse in genetic association studies refers to the estimation bias of the reported odds ratios (OR) for an associated genetic variant from the initial discovery samples. It is a consequence of the sequential procedure in which the estimated effect of an associated genetic marker must first pass a stringent significance threshold. We propose a hierarchical Bayes method in which a spike-and-slab prior is used to account for the possibility that the significant test result may be due to chance. We examine the robustness of the method using different priors corresponding to different degrees of confidence in the testing results and propose a Bayesian model averaging procedure to combine estimates produced by different models. The Bayesian estimators yield smaller variance compared to the conditional likelihood estimator and outperform the latter in the low power studies. We investigate the performance of the method with simulations and applications to four real data examples. Pleiotropy occurs when a single genetic factor influences multiple quantitative or qualitative phenotypes, and it is present in many genetic studies of complex human traits. The longitudinal family studies combine the features of longitudinal studies in individuals and cross-sectional studies in families. Therefore, they provide more information about the genetic and environmental factors associated with the trait of interest. We propose a Bayesian latent variable modeling approach to model multiple phenotypes simultaneously in order to detect the pleiotropic effect and allow for longitudinal and/or family data. An efficient MCMC algorithm is developed to obtain the posterior samples by using hierarchical centering and parameter expansion techniques. We apply spike and slab prior methods to test whether the phenotypes are significantly associated with the latent disease status. We compute Bayes factors using path sampling and discuss their application in testing the significance of factor loadings and the indirect fixed effects. We examine the performance of our methods via extensive simulations and apply them to the blood pressure data from a genetic study of type 1 diabetes (T1D) complications.

We develop statistical methods for tackling two important problems in genetic association studies. First, we propose a Bayesian approach to overcome the winner's curse in genetic studies. Second, we consider a Bayesian latent variable model for analyzing longitudinal family data with pleiotropic phenotypes. Winner's curse in genetic association studies refers to the estimation bias of the reported odds ratios (OR) for an associated genetic variant from the initial discovery samples. It is a consequence of the sequential procedure in which the estimated effect of an associated genetic marker must first pass a stringent significance threshold. We propose a hierarchical Bayes method in which a spike-and-slab prior is used to account for the possibility that the significant test result may be due to chance. We examine the robustness of the method using different priors corresponding to different degrees of confidence in the testing results and propose a Bayesian model averaging procedure to combine estimates produced by different models. The Bayesian estimators yield smaller variance compared to the conditional likelihood estimator and outperform the latter in the low power studies. We investigate the performance of the method with simulations and applications to four real data examples. Pleiotropy occurs when a single genetic factor influences multiple quantitative or qualitative phenotypes, and it is present in many genetic studies of complex human traits. The longitudinal family studies combine the features of longitudinal studies in individuals and cross-sectional studies in families. Therefore, they provide more information about the genetic and environmental factors associated with the trait of interest. We propose a Bayesian latent variable modeling approach to model multiple phenotypes simultaneously in order to detect the pleiotropic effect and allow for longitudinal and/or family data. An efficient MCMC algorithm is developed to obtain the posterior samples by using hierarchical centering and parameter expansion techniques. We apply spike and slab prior methods to test whether the phenotypes are significantly associated with the latent disease status. We compute Bayes factors using path sampling and discuss their application in testing the significance of factor loadings and the indirect fixed effects. We examine the performance of our methods via extensive simulations and apply them to the blood pressure data from a genetic study of type 1 diabetes (T1D) complications.

No description available.

Submitted by Evelyn Edith Contreras Medrano (contrerasmedrano@gmail.com) on 2017-03-10T21:05:13Z No. of bitstreams: 1 EContreras 2016 - Local Institutions and Natural Resources vf.pdf: 2289932 bytes, checksum: 3e134974c30ddcb38184183d421e24e7 (MD5) / Approved for entry into archive by ÁUREA CORRÊA DA FONSECA CORRÊA DA FONSECA (aurea.fonseca@fgv.br) on 2017-03-20T13:10:47Z (GMT) No. of bitstreams: 1 EContreras 2016 - Local Institutions and Natural Resources vf.pdf: 2289932 bytes, checksum: 3e134974c30ddcb38184183d421e24e7 (MD5) / Made available in DSpace on 2017-03-22T18:33:30Z (GMT). No. of bitstreams: 1 EContreras 2016 - Local Institutions and Natural Resources vf.pdf: 2289932 bytes, checksum: 3e134974c30ddcb38184183d421e24e7 (MD5) Previous issue date: 2016-12-08 / After decades of research, there is still no consensus in the literature regarding the effects of natural resource abundance on the economic and political performance of a territory. This research aims to contribute to this discussion, by analyzing the role of institutions on explaining the relationship between natural resource-based revenues and the provision of public goods at the local level. In order to do so, I test the mechanisms previously proposed in the literature for explaining the natural resource curse effect at the national level (mediation and moderation effects of institutional quality), using cross-sectional data of Peruvian municipalities located in the Andean highlands, for the 2011-2014 period. The identification strategy proposed uses as source of exogenous variation for revenues, the location of natural resources and its value among the territory, and a set of rules established by law for the redistribution of natural resource-based revenues to the local governments. In order to deal with the endogeneity of institutional quality, I include 2SLS estimations, using the presence of 'Peasant Communities' (Comunidades Campesinas) as an instrumental variable. The results show some evidence of a positive effect of natural resource-based revenues on the provision of local public infrastructure (water, public lightning and rural roads), a null effect on education results, and a negative effect on health campaigns. However, regarding the role of institutional quality on explaining these effects, I find no significant effects for all of the outcomes and samples analyzed.

La chute récente du prix des matières premières a montré la vulnérabilité que pouvait impliquer une trop forte dépendance des économies aux revenus provenant des ressources naturelles. Dans le premier chapitre, les résultats que nous obtenons indiquent que les prix des matières premières sont un indicateur important du risque pays des exportateurs, ce qui n’est pas le cas des pays qui sont importateurs. Bien que les pays exportateurs soient aujourd’hui ceux dont le défaut externe est le plus probable, il n’apparaît pas de prime de risque supplémentaire liée à la détention de leurs obligations. Dans le deuxième chapitre, nous étudions le lien entre les termes de l’échange et le taux de change effectif réel entre et de exportateurs de matières premières . Il existe une relation de long terme entre les termes de l’échange des pays exportateurs et leur taux de change réel. Les termes de l’échange n’ont pas d’impact à court-terme sur les taux de change effectifs réels. Néanmoins, nous montrons grâce à l’estimation d’un modèle à seuil à transition lisse en panel PSTR , qu’ une forte baisse du prix du pétrole (entre 25% et 36%) donnent un pouvoir explicatif aux termes de l’échange. Même si la question est encore largement débattue dans la littérature, l’idée qu’il existe une "malédiction des ressources naturelles" a fait sa place. Les pays fortement dotés en ressources naturelles auraient en moyenne une croissance économique plus faible que ceux qui n’en sont pas pourvus. Dans le dernier chapitre, nous proposons une méta-analyse basée sur un échantillon de 67 études empiriques qui étudient le lien entre ressources naturelles et croissance économique. Sur la base d’un ensemble de coefficients estimés, nos résultats montrent qu’il existe une "faible" malédiction des ressources. / The recent drop in commodity prices showed the high vulnerability implied by being too much dependent on revenues stemming from natural resources. In the first chapter, we look into the way financial markets assess the market risk of twenty-two emerging economies. More precisely, the purpose of this chapter is to investigate how natural resources are incorporated in the way international investors perceived the ability to service external debt obligationsduring the 2003-2014 period. The results indicate that commodity prices are an important driver of sovereign spreads in the case of exporters while it is not the case for importing countries. In the second chapter, we investigate the link between energy prices and the real effective exchange rate of commodity-exporting countries. Estimating a panel cointegration relation between the real effective exchange rate and its fundamentals, we provide evidence for the existence of both energy and commodity currencies and we show that when the oil market is highly volatile (downwards), currencies follow an "oil currency regime", terms-of trade becoming an important driver of the real exchange rate A conventional wisdom has spread in the literature stating that a high endowment in natural resources is detrimental for growth, yet the debate is still ongoing In this chapter, we aim at providing quantitative results on the magnitude of the link between natural resources and growth found in the literature, as well as discussing, on quantitative bases, whether the sources of heterogeneity are significant. To this end, we implement a meta-analysis based on 67 empirical studies that investigate the link between natural resources and growth, totaling 1405 estimates. The results show a "soft" curse that may be reverted together with the importance of institutions in mitigating the curse.

Syftet med den här uppsatsen är att, utifrån olika exempel på ungdomsspråk, undersöka gymnasieelevers och svensklärares attityder till samt användning av ungdomsspråk i och utanför gymnasieskolans verksamhet med ett fokus på muntlig språkproduktion i svenskämnet. Vidare utreder uppsatsen om det finns likheter och skillnader mellan gymnasieelevers och svensklärares användning av ungdomsspråk samt vad för attityder de har till det. Uppsatsen visar också på om och hur gymnasieelevers prestationer i svenskämnet påverkas av deras användning av ungdomsspråk. Dessutom undersöks gymnasieelevers och svensklärares uppfattning om det ovanstående och om motpartens användning av ungdomsspråk. Två underliggande faktorer, det vill säga kön och ålder, undersöks också i koppling till de ovannämnda målgrupperna. Två enkäter distribuerades till gymnasielever och svensklärare vid en gymnasieskola där de fick besvara kvalitativa och kvantitativa frågor angående attityder till och användning av ungdomsspråk. Resultatet visade att gymnasieelever använder ungdomsspråk mer än svensklärare, både i och utanför gymnasieskolans verksamhet, exempelvis genom slang, svordomar och diskurspartiklar. Båda målgrupperna har olika attityder till ungdomsspråk där svensklärare kan se det som begränsande medan gymnasielever anser att det skapar gemenskap. Svensklärare och gymnasieelever har också varierande åsikter om huruvida elevers användning av ungdomsspråk påverkar deras prestationer i svenskämnet. Dock anser majoriteten av dem att det har en negativ inverkan. Likaså finns det både likheter och skillnader hos båda målgruppernas uppfattning av motpartens användning av ungdomsspråk. Det visas till exempel genom att svensklärarna anser att ungdomsspråk kan vara innovativt medan gymnasielever uttrycker att det är pinsamt om deras lärare försöker använda ungdomsspråk i klassrummet. / The purpose with this essay is to, based on different examples of youth language, examine upper secondary students’ and Swedish teachers’ attitudes to and usage of youth language in and outside upper secondary school with a focus on oral language production in the school subject of Swedish. This essay will also focus on if and how upper secondary students’ accomplishments in the school subject of Swedish are affected by their usage of youth language as well as what kind of view that upper secondary students and Swedish teachers have about this and about the opposite group’s usage of youth language. Two underlying factors, as in gender and age, will also be examined in relation to the previously mentioned target groups. Two surveys were handed out to upper secondary students and Swedish teachers. They got to answer qualitative and quantitative questions about attitudes to and usage of youth language. The result showed that upper secondary students use youth language more than Swedish teachers, both in and outside upper secondary school, for example through slang, curse words and discourse markers. Both of the target groups have different attitudes to youth language where Swedish teachers consider that it is limited while upper secondary students believe that it creates solidarity. Swedish teachers and upper secondary students also have different opinions about the fact that students’ usage of youth language might affect their accomplishments in the school subject of Swedish. However, the majority of both target groups believe that it has a negative influence. There are also similarities and differences between both target groups’ views on the counterpart’s usage of youth language. For example, the Swedish teachers believe that youth language can be innovative while upper secondary students express that it is embarrassing if their teachers try to use youth language in the classroom.