Whereas no prior contribution has focused on the case of a medium-sized city in a developing country, such as Mexico, to explore how urban structure and its expansion has affected the spatial distribution of employment, three distinct, but related papers were developed, which combine urban economics literature and spatial sciences techniques to fill this gap and provide new evidence. The first paper, entitled "Spatial Distribution of Employment in Hermosillo, 1999 and 2004" identifies where employment subcenters are. Testing the presence of spatial effects, it concludes that an incipient process of employment suburbanization has taken place; however, the city still exhibits a monocentric structure. As a complement, a second paper, "Employment Density in Hermosillo, 1999-2004: A Spatial Econometric Approach of Local Parameters" tests if the Central Business District (CBD), despite suburbanization, maintains the traditional attributes of attracting activities and influencing the organization of employment around it. The CBD is still attractive, but its influence varies across space and economic sector, conclusions that were masked by global estimations. Thirdly, a study was essential to uncover how important is the urban structure and the suburbanization of jobs in explaining the dispersion resulting of households and workplaces (commuting). The paper entitled "Commuting in a Developing City: The Case of Ciudad Obregon, Mexico" examines this issue. To take advantage of the commuting information available, the study area was switched. In general, the results are consistent with those suggested by urban economics; moreover, the inclusion of workplace characteristics was a novelty to model commuting behavior and proves that space matters. Additionally, new evidence was provided to the field of spatial science through the applications of techniques able to expose the spatial effects associated with the distribution of employment, more specifically, the Exploratory Spatial Data Analysis(ESDA), Geographically Weighted Regression (GWR) with spatial effects, as well as the generalized multilevel hierarchical linear model (GMHL) were used. The new findings produced for this dissertation provide a more comprehensive understanding of urban dynamics and could help to improve the planning process. It is hoped that this dissertation will contribute to that development as well as stimulate further research.

Dans cette thèse, nous avons développé un algorithme de détermination d’une nouvelle forme de la matrice de poids spatiale qui prend en compte aussi bien les effets de proximité spatiale que les critères idiosyndratiques des unités régionales. Ainsi, nous avons proposé une approche originale de détermination d’une forme stationnaire du diagramme de Moran à travers laquelle nous pouvons déterminer les formes des associations spatiales de long terme. En ce sens, nous avons montré la conformité qui peut exister entre les résultats trouvés suite à l’application de cette approche et ceux découlant à partir de l’estimation du modèle de convergence absolue dans un cadre spatial. Dans une autre partie du travail, nous avons proposé deux nouvelles méthodes de détermination des régimes spatiaux sous forme de polarisation et de stratification. Une nouvelle approche de spécification des modèles spatiaux qui se base sur la robustesse du test de Moran dans la détection de l’autocorrélation spatiale globale et le bon choix de la matrice de poids spatiale, a été encore proposée. Nous avons utilisé l’approche d’Analyse Exploratoire des Données Spatiales, l’Approche d’Econométrie Spatiale et les méthodes que nos avons développées pour expliquer les différentes formes de disparités, des interactions spatiales et de convergence régionale en Inde en termes de taux d’alphabétisation au niveau national, pour l’ensemble de la population (Homme et Femme) et au niveau rural et urbain pour l’ensemble des districts sur la période de 1991-2001. Dans la dernière partie du travail, nous avons développé un modèle théorique des facteurs déterminants de l’alphabétisation en Inde. / In this thesis, we have developed an algorithm for determining a new form of the spatial weight matrix that takes into account both the effects of spatial proximity and the idiosyncratic criteria of regional units. Thus, we propose a new approach for determining a stationary form of the Moran scatter plot in which we highlight the forms of steady state spatial associations and the conformity that can exist between the results arising out of this method and those obtained from the model of absolute convergence in a spatial framework. In another part of the thesis, we propose two new methods of determining the spatial regimes in the form of polarization and stratification. A new approach to the specification of spatial models which is based on the robustness of the Moran test in the detection of global spatial autocorrelation and the proper choice of the spatial weight matrix has also been proposed. We used the approach of Exploratory Spatial Data Analysis, Spatial Econometrics approach and methods we developed to explain the different forms of disparities, spatial interaction and regional convergence in India in terms of literacy rate at the national level for the entire population (Male and Female) and rural and urban level for all districts over the period 1991-2001. In the last part of the work, we develop a theoretical model of the determinants of literacy in India.

O setor sucroenergético do Brasil encarou uma década de expansão da produção, seguida por um período de retração das atividades. Neste período, 80 usinas e destilarias deixaram de operar por motivo de falência das empresas ou decisões operacionais dos grupos controladores. A cana de açúcar possui características biológicas que determinam a necessidade de processamento industrial em poucas horas após a atividade de colheita, sob pena de grande perda de teor de sacarose. Essa questão tecnológica determina que a organização espacial da produção de açúcar e etanol seja preferencialmente realizada dentro de um raio geográfico reduzido em relação aos canaviais. O objetivo desta tese é analisar se as recentes paralisações e falências no setor sucroenergético do Brasil foram influenciadas pelas condições operacionais dos mercados regionais de cana de açúcar, assim como identificar os principais determinantes dos eventos de paralisação das atividades das usinas e destilarias entre 2001 e 2016. Indicadores de estrutura de mercado e concentração espacial são desenvolvidos e discutidos com base no conceito de mercado antitruste usado pelas principais agências de defesa da concorrência. A descrição da estrutura de mercado e a intensidade dos eventos de desligamento e falência em cada região são avaliados considerando a existência de múltiplas unidades de processamento que são de propriedade de uma mesma empresa em cada região. Os resultados indicam que a concentração e a disputa pela cana de açúcar como insumo por usinas localizadas próximas umas das outras estão espacialmente associadas aos eventos de falência e desligamento. No entanto, a análise de sobrevivência indica que o aumento do número de concorrentes demandando este insumo e a expansão do processamento de cana de açúcar no raio geográfico de colheita viável não foram fatores que contribuíram isoladamente para as falências e interrupções. Estes eventos foram menos frequentes entre as unidades industriais cuja empresa controladora conta com investidores estrangeiros na estrutura de capital e nos casos em que uma parcela mais alta dos insumos é obtida por meio de uma estrutura verticalmente integrada. / Brazil\'s sugar and ethanol industry faced a decade of expansion of production, followed by a period of retraction of activities. During this period, 80 mills and distilleries ceased to operate due to corporate bankruptcy or operational decisions of the controlling groups. Sugarcane has biological characteristics that determine the need for industrial processing in a few hours after harvesting, under penalty of great loss of sucrose content. This technological issue determines that the spatial organization of sugar and ethanol production is preferably carried out within a reduced geographic radius in relation to sugar cane fields. The objective of this thesis is to analyze if the recent shutdowns and bankruptcies in the Brazilian sugarcane industry were influenced by the operational conditions of the regional sugarcane markets, as well as to identify the main determinants of stoppage events of the maills and distilleries between 2001 and 2016. Indicators of market structure and spatial concentration are developed and discussed based on the concept of the antitrust market used by the major antitrust agencies. The description of the market structure and the intensity of the shutdown and bankruptcy events in each region are evaluated considering the existence of multiple processing units that are owned by the same company in each region. The results indicate that the concentration and the dispute for sugarcane as an input by plants located close to each other are spatially associated with the events of bankruptcy and shutdown. However, the survival analysis indicates that the increase in the number of competitors demanding this input and the expansion of sugarcane processing in the viable crop geographical radius were not factors that contributed in isolation to bankruptcies and interruptions. These events were less frequent among industrial units whose parent company relies on foreign investors in the capital structure and where a higher share of inputs are obtained through a vertically integrated structure.

Ce travail analyse sous différentes approches la pauvreté au Sénégal en s’appuyant notamment sur les données fournies par les deux dernières enquêtes auprès des ménages (ESAM 2-2002 et ESPS-2006) réalisées par l’Agence Nationale de la Statistique et de la Démographie en partenariat avec la Banque mondiale.Dans l’analyse de la pauvreté monétaire, nous faisons apparaître des différences importantes en termes de seuils de pauvreté dans les régions avec aux extrêmes la région Dakar 923,55 F CFA (1,40 €) et Tambacounda 515,70 F CFA (0,78€), ce qui suggère le peu de pertinence quant à l’utilisation d’un seuil établi au seul niveau national. Sur la base de ces seuils, les indices de pauvreté issus de la formule générique de Foster, Greer et Thorbecke (FGT) dévoilent une baisse du taux de pauvreté entre 2002 et 2006 de 57,1% à 50,8%, soit de 6,9 point dans l’ensemble du pays et un écart à la ligne de pauvreté passant de 18% à 16,4%. Cette baisse est particulièrement observée dans les régions de Dakar, Diourbel, Kaolack, Saint-Louis et Thiès. Au niveau départemental, les taux de pauvreté montrent une concentration importante dans les zones rurales et l’existence de poches de pauvreté enclavées dans les zones urbaines. L’estimation d’un modèle économétrique spatial met en évidence les facteurs socioéconomiques susceptibles d’expliquer les différences interdépartementales de taux de pauvreté constatées en 2006, notamment le degré de développement économique des territoires (urbanisation, emploi) ainsi que les comportements des ménages liés au niveau d’infrastructures (d’éducation, de santé et de fécondité).Par ailleurs, nous proposons un modèle dichotomique à partir duquel il est possible de mettre en évidence les déterminants de la pauvreté monétaire des chefs de ménage. Les résultats montrent que les femmes chefs de ménage ne sont pas la couche la plus pauvre. De manière générale, les disparités de pauvreté manifestes entre milieux urbain et rural sont largement corrélées à des handicapes en matière de d’éducation et à l’inégal accès aux moyens d’information et de communication.Nous abordons une analyse multidimensionnelle de la pauvreté au Sénégal, à travers une estimation des degrés de privation de certains besoins essentiels des ménages. L’approche par la théorie des ensembles flous utilisée à cet effet suggère que la pauvreté a faiblement diminué : 1 % contre 7 % pour la pauvreté monétaire. Contrairement à l’approche monétaire, la baisse de la pauvreté non monétaire observée concerne d’autres régions comme Kolda et Ziguinchor et les régions de Diourbel et Kaolack connaissent une hausse. L’estimation des indices flous unidimensionnels a permis d’identifier les domaines dans lesquels les ménages affichent le degré de privation le plus important : la qualité du logement, le niveau d'instruction et les moyens d’information et de communication, au-delà du revenu.Les profils de pauvreté monétaire aussi bien que multidimensionnelle sont d’excellents outils pour cibler les groupes les plus nécessiteux de la population. En revanche, ces outils restent muets sur la perception de ces pauvres quant à leur propre situation socioéconomique. En ce sens, une analyse économétrique des facteurs déterminants de la pauvreté ressentie au Sénégal en 2006 fait apparaître l’importance de certaines dimensions non économiques (exclusion sociale, culturelle et manque de concertation des intéressés sur les politiques de développement et de lutte cotre la pauvreté). / This paper analyzes different approaches in poverty in Senegal, relying in particular on data provided by the last two household surveys (ESPS-2-2002 and ESAM 2006) conducted by the National Agency of Statistics and Demography in partnership with the World Bank.In the analysis of monetary poverty, we reveal important differences in terms of poverty lines in regions with at extremes, Dakar 923,55 F CFA (1,40 €) and Tambacounda 515,70 FCFA (0,78€), suggesting little relevance to the use of a threshold at national level alone. On basis of these thresholds, the indices of poverty stemming from the Foster's generic formula, to Greer and Thorbecke (FGT) reveal a reduction in the rate of poverty between 2002 and 2006 from the 57.1 % to 50.8 %, that is 6.9 in the whole of country and a gap in the poverty's line passing from 18 % to 16,4 %. This decline is particularly observed in the regions of Dakar, Diourbel, Kaolack, Saint-Louis and Thies. At the departmental level, poverty rates show a significant concentration in rural areas and the existence of pockets of poverty enclaved in urban areas. The estimation of a spatial econometric model highlights the socioeconomic factors that may explain the interdepartmental differences in poverty rates observed in 2006, including the level of regional economic development (urbanization, employment) and household behavior related at infrastructure (education, health and fertility). Furthermore, we propose a dichotomous model from which it is possible to identify the determinants of income poverty of household heads. The results show that female-headed households are not the poorest layer. In general way, differences of poverty apparent between urban and rural areas are largely correlated with disabilities in terms of education and unequal access to information and communication resources.We are entering a multidimensional analysis of poverty in Senegal, through an estimate of the degree of deprivation of some basic household needs. The approach by the theory of fuzzy sets used for this purpose shows that poverty declined slightly: 1% against 7% for monetary poverty. Unlike the monetary approach, the observed decrease from non-monetary poverty affects other regions as Kolda and Ziguinchor and Kaolack and Diourbel saw an increase. The estimation of one-dimensional fuzzy indexes allowed identifying the domains in which the households post the degree of largest deprivation: the quality of housing, education and information and communication technologies, beyond income.The profiles of monetary poverty as well as multidimensional are excellent tools to target the most destitute groups of the population. However, these tools remain dumb on the perception of these poor people as for their own socioeconomic situation. In this sense, an econometric analysis of the determinants of poverty felt in Senegal in 2006 brings up the importance of certain non-economic dimensions (social exclusion, cultural and lack of consultation of stakeholders on policy development and cutter fight poverty).

Au regard des dégâts environnementaux au cours des dernières années, il est important de mettre en œuvre des politiques environnementales efficaces afin de lutter contre le changement climatique, de préserver la biodiversité, et de réduire les pollutions de l'eau et de l'air. Le secteur énergétique est l'un des principaux contributeurs à la détérioration de l'environnement. Ainsi, afin d'atteindre les objectifs environnementaux, l'établissement de politiques énergétiques efficaces est primordial. A cet effet, il semble indispensable de mener une démarche inclusive et transversale en impliquant l'ensemble des acteurs des différents échelons (locaux, régionaux, nationaux, internationaux,...). Dès lors, l'identification de l’échelon optimal est cruciale dans une optique d'élaboration des politiques énergétiques efficaces. La thèse s'attache précisément à éclairer les enjeux liés à la fiscalité énergétique. Ainsi, l'objectif principal de cette thèse est d'analyser la nature des interactions environnementales, qu'elles soient entre régions ou Etats, en termes de fiscalité ou de dépenses environnementales. Les contributions de cette thèse sont à la fois théorique et empirique. D'un point de vue empirique, cette thèse s'est d'abord focalisée sur l’étude de la fiscalité énergétique sous l'angle des interactions spatiales. En particulier, le chapitre 1 teste la présence des interactions spatiales des départements français via la vignette automobile. Ce chapitre fournit une analyse empirique de la réaction de la politique fiscale d'un département français suite à un changement de la politique fiscale de ces voisins. Tout d'abord, en utilisant une approche économétrique spatiale en panel, les résultats montrent l'existence d'une dépendance spatiale positive, suggérant l'existence d'un comportement mimétique des départements français lors de la détermination des taux de la vignette automobile. Ensuite, ce chapitre met également en évidence une relation entre les impôts directs locaux et ceux indirects locaux. En particulier, les résultats des estimations montrent que le taux de la taxe professionnelle (resp. la taxe sur le foncier non bâti) et la vignette fiscale sont des substituts alors que le taux de la taxe d'habitation (resp. la taxe sur le foncier non bâti) sont des compléments à la vignette automobile. Enfin, les résultats montrent que les départements avec des populations plus grandes, plus jeunes et plus âgées fixent des montants plus élevés de la vignette automobile. D'un point de vue théorique, le but du chapitre 2 est d'analyser les interactions fiscales entre plusieurs niveaux de gouvernements (en particulier deux niveaux de décisions), dans lesquels il existe plusieurs juridictions de niveaux inférieurs identiques, telles que les États ou encore les collectivités locales, et une juridiction unique de premier plan, telle que l'Etat fédéral, un pays ou une union communautaire, toutes ces juridictions pouvant taxer deux bases fiscales à savoir, le capital et la ressource énergétique. Nous déterminons grâce à la formalisation théorique les implications en termes de distorsions fiscales de l'architecture fiscale dans le cadre d'une politique énergétique. En particulier, en recourant au jeu de Nash, nos résultats montrent d'une part une substitution entre les taxes fédérales et locales sur l'énergie impliquant que les gouvernements locaux augmentent leurs taux de taxation de l'énergie en réponse à une baisse du taux de la taxe fédérale sur l'énergie, toutes choses égales par ailleurs. D'autre part, nous constatons que les taxes locales sur l’énergie ont une incidence positive sur la qualité de l’environnement, ce qui suggère que la décentralisation de la politique énergétique peut jouer un rôle crucial pour atténuer les dommages environnementaux. Enfin, à travers le chapitre 3, nous testons d'un point de vue empirique l'existence d'une interdépendance spatiale entre les pays de l'OCDE via les dépenses environnementales. / In the light of environmental damage in recent years, it is important to implement effective environmental policies to combat climate change, preserve biodiversity, and reduce water and air pollution. The energy sector is one of the main contributors to the deterioration of the environment. Thus, in order to achieve environmental objectives, establishing effective energy policies is paramount. To this end, it seems essential to carry out an inclusive and transversal approach by involving all actors at different levels (local, regional, national, international, etc.). Therefore, identifying the optimal step is crucial for effective energy policy development. The thesis focuses on clarifying the issues related to energy taxation. Thus, the main objective of this thesis is to analyze the nature of environmental interactions, be they between regions or states, in terms of taxation or environmental expenses. The contributions of this are both theoretical and empirical.From an empirical point of view, this thesis first focuses on the study of energy taxation from the perspective of spatial interactions. In particular, Chapter 1 tests the presence of spatial interactions of French departments via the car sticker. This chapter provides an empirical analysis of the reaction of the tax policy of a French department following a change in the tax policy of these neighbors. First, by using a spatial econometric panel approach, the results show evidence of spatial dependence, suggesting the existence of a mimetic behavior of French departments when determining the rates of the car sticker. Next, this chapter also highlights a relationship between local direct taxes and indirect local taxes. In particular, the results of the estimates show that the rate of the business tax (or the tax on undeveloped land) and the road tax sticker are substitutes, whereas the residential tax rate (or the tax rate on developed land) are complements to the road tax sticker. Finally, the results show that departments with larger, younger and older populations are setting higher amounts of the car sticker.From a theoretical point of view, the purpose of Chapter 2 is to analyze the tax interactions between several levels of government (in particular two levels of decision-making), in which there are several jurisdictions of similar lower levels, such as local authorities, and a single leading jurisdiction, such as the federal state, a country or a community union, all of which can tax two tax bases, namely, capital and energy resources. Through theoretical formalization, we determine the implications in terms of tax distortions of tax architecture in the context of an energy policy. In particular, using the Nash game, our results show, on the one hand, a substitution between federal and local taxes on energy, implying that local governments increase their energy tax rates in response to a fall in the rate of energy. of the federal energy tax, all other things being equal. On the other hand, we find that local taxes on energy have a positive impact on the quality of the environment, suggesting that decentralization of energy policy can play a crucial role in mitigating environmental damage.Finally, in Chapter 3, we empirically test the existence of spatial interdependence among OECD countries through environmental spending. To this end, we use data from 30 OECD countries for the period 1994-2014 and a wide range of economic and political control variables.

Neste trabalho, utilizamos uma equação hedônica para estimar o preço implícito de amenidades urbanas a partir de dados do mercado de imóveis no Município de São Paulo. São testadas duas especificações: uma com as variáveis explicativas originais e outra com a construção de fatores para correção de multicolinearidade. Além de considerar a configuração monocêntrica tradicional, propomos a inclusão de um vetor de acessibilidade para o caso de uma cidade duocêntrica. A investigação sobre as variáveis ambientais relevantes é fundamentada na interpolação espacial dos resíduos. Concluímos que a proximidade das estações de trem, a presença de áreas verdes e o zoneamento estritamente residencial valorizam o imóvel para fins residenciais, enquanto a criminalidade reduz o seu valor. / In this work, I use a hedonic equation to estimate the implicit prices of urban amenities based on housing market data for the city of São Paulo. I test two different specifications, one with the original independent variables, and another with factors to correct for multicollinearity. In addition to the traditional monocentric model, I include an accessibility vector for a duocentric city. A criterion to select the appropriate environmental variables based on residual spatial interpolation is introduced. I conclude that proximity to train stations, green areas and strict residential zones raise rents, while criminality reduces them.

A presente dissertação objetivou analisar o impacto dos royalties do petróleo no PIB per capita dos municípios da região Sudeste brasileira, calculando os efeitos diretos, indiretos e totais das receitas nas localidades beneficiadas. Com efeito, foram empregados modelos econométricos espaciais para controlar as interferências de componentes espaciais nas estimativas, e também para calcular o efeito spillover. Através da análise dos dados e dos testes estatísticos decidiu-se estimar o modelo espacial de Durbin, ou SDM, com a matriz de distância de Haversine inversa com raio de até 200 km. O efeito direto das receitas confirma os indícios de presença da maldição dos recursos naturais nos municípios da região Sudeste. O efeito transbordamento indica que os municípios do Sudeste afetam os seus vizinhos de forma não linear, mas pela média de dependência aos royalties, há indícios de que os efeitos negativos são mais presentes do que os positivos. Restringindo a amostra para os municípios do Rio de Janeiro, onde a dependência é maior, foi encontrado que os royalties não possuem efeitos nem positivos e nem negativos nos municípios beneficiados, porém, o efeito transbordamento é negativo e significativo, indicando, portanto, a presença de uma \"maldição regional fluminense dos royalties do petróleo\". / This paper aimed analyze the oil royalties impact upon Southeastern municipalities per capita GDP, estimating both the direct and the spillover effect of such revenues. The spatial econometric models were estimated to control the spatial interferences in the econometric estimations and estimate the spillover effect as well. The data analysis and the statistical tests indicated the Durbin spatial model as the best, or SDM, likewise the inverse distance Haversine spatial matrix considering a 200-km radio. The direct effect confirmed the evidences of \"natural resource curse\" presence in the Southeastern cities. The spillover effect showed the oil royalties no-linearly impact the neighbors of contemplated cities, however, considering the average oil dependency, there is an evidence of the negative impacts overcome the positive ones. Narrowing the sample to the Rio de Janeiro, where the dependency is worst, it was founded that oil revenues had no effect upon per capita GDP of these municipalities. Nevertheless, the spillover effect was negative and statistically significant, indicating, thus, the presence of a \"oil royalties fluminense curse\".

Na literatura econômica, há um grande interesse no estudo sobre a dinâmica do crescimento das cidades e os fatores que a influenciam. A motivação principal dos pesquisadores é verificar por que algumas cidades crescem enquanto outras permanecem estagnadas, e quais são os fatores que contribuem para o fluxo de migração de fatores de produção para os grandes centros urbanos. Neste sentido, o presente trabalho procura avaliar o impacto da infraestrutura pública sobre o crescimento econômico das cidades brasileiras - mensurado por meio do crescimento populacional e salarial. O acesso à infraestrutura é um dos principais fatores que determinam o desenvolvimento econômico e é considerado um dos maiores entraves para a o crescimento econômico do Brasil. Boa infraestrutura pode garantir serviços básicos, bem como externalidades positivas sobre a saúde, meio ambiente, produtividade e possibilidades de investimento. Os dados dos Censos Demográficos, utilizados nesta dissertação, mostraram que o país experimentou um forte crescimento no acesso à infraestrutura no período de 1970 a 2010. O acesso ao bastecimento de água, a coleta de esgoto, energia elétrica e a telefonia mais do que dobrou no período analisado. Tendo em vista que os municípios brasileiros são suscetíveis a diversas formas de interações entre si, é de se esperar que existam efeitos espaciais entre estes. Portanto, para analisar os impactos da infraestrutura sobre o crescimento econômico regional, foi utilizada a metodologia econométrica de dados em painel com dependência espacial. Como esperado, os resultados mostram que as variáveis de infraestrutura afetam positivamente o crescimento econômico regional, sendo o acesso à energia elétrica e a telefonia as variáveis com maior impacto. / In the economic literature there is a great interest in the study of the dynamics of the growth of cities and the factors that influence it. The main attraction of the researchers is to check why some cities grow while others remain stagnant, and what are the factors that contribute to the migration flow of production factors to the major urban centers. In this sense, this paper seeks to assess the impact of public infrastructure on economic growth of Brazilian cities - measured by population and wage growth. Since the access to infrastructure is one of the main factors determining economic development and is considered one of the greatest barriers to the economic growth in Brazil. Good infrastructure can guarantee basic services, as well as positive externalities in health, environment, productivity and possibilities for investment. Data from Demographic Censuses, used in this paper, showed that the country experienced strong growth in the access to infrastructure in the period of 1970-2010. The access to water supply, the sewage collection, electricity and telecommunications more than doubled during the period analyzed. In order to measure the effects of infrastructure on regional economic growth, we used the econometric analysis of panel data with spatial dependence, since the Brazilians\' municipalities are susceptible to diverse forms of interactions among them, is to be expected that exist spatial effects between them. As expected, the results show that the infrastructure variables positively affect the variation of income, being the access to electricity and telecommunications the variables with the greatest impact.

Esta dissertação investiga o impacto da aglomeração sobre o desempenho econômico dos Municípios Brasileiros entre 1970 e 2000. Para tanto, foram estimadas equações de crescimento combinadas com modelos econométricos espaciais. Os resultados indicam que a aglomeração teve efeito positivo sobre o crescimento econômico, sem indícios claros de congestionamento. Além disso, também foram encontrados sinais de externalidades positivas e spillovers de educação vindos de áreas vizinhas. Os resultados reforçam a necessidade de se considerar os efeitos espaciais quando se procura entender o desenvolvimento econômico no Brasil. / This paper investigates the impact of agglomeration economies on economic performance of Brazilian municipalities between 1970 and 2000. In order to achieve that, we estimate growth equations adopting standard spatial econometric models. The results indicate that agglomeration has positive impacts on economic development without clear indication of congestion effects. Moreover, we also found evidence of positive spatial externalities and educational spillovers coming from neighboring areas. The results reinforce the need for considering spatial effects when aiming to understand the economic development in Brazil.

Submitted by PPG Economia do desenvolvimento (economia-pg@pucrs.br) on 2018-05-14T20:42:09Z No. of bitstreams: 1 ANNE_JULISSA_ADUBER_PENALOZA_TES.pdf: 3031751 bytes, checksum: 27dc405ec8dc4ef1271f4c34d565e5c6 (MD5) / Approved for entry into archive by Sheila Dias (sheila.dias@pucrs.br) on 2018-05-18T15:00:23Z (GMT) No. of bitstreams: 1 ANNE_JULISSA_ADUBER_PENALOZA_TES.pdf: 3031751 bytes, checksum: 27dc405ec8dc4ef1271f4c34d565e5c6 (MD5) / Made available in DSpace on 2018-05-18T15:11:47Z (GMT). No. of bitstreams: 1 ANNE_JULISSA_ADUBER_PENALOZA_TES.pdf: 3031751 bytes, checksum: 27dc405ec8dc4ef1271f4c34d565e5c6 (MD5) Previous issue date: 2017-10-03 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior - CAPES / This thesis is composed of three articles on Health Economics. The first document seeks to analyze the possible spatial integration of the network of public hospitals in Colombia, with the aim of analyzing whether there is efficiency when remitting patients according to the type of care demanded in the hospital. Based on the estimation of a spatial panel model, we analyzed the total of hospitals that make up the public offer system, concluding that although there may be a certain degree of transfer in patient care, it is very low for the case of general medicine, given that the spatial coefficient of the SDM model, rho is less than one. Once the analysis for the static model was estimated, the dynamic SAR and SDM model was estimated, a dynamic model was estimated that decomposed the effects in direct and indirect, where it was evidenced that there is a long-term spatial distribution effect for the attention of appointments of general medicine determined by the number of health care personnel of the 4 closest hospitals, however the presence of transshipments was not found in the short term in general medical care among the hospitals of the public network, since these are the ones that behave as institutions providing basic services, without having a trained staff that allows better care from the development of better medical practices. The same article calculates the SDM, SAR, SAC models for highly complex hospitals, having as dependent variable the number of surgeries performed with a normalized distance matrix of 8 neighbors, concluding that the number of emergencies attended by the hospital affects negative in the total of surgeries practiced, that is to say, the hospitals of high level of complexity carry out in their majority programmed surgeries, nevertheless a change in the total of emergencies. In order to control the existence of random spatial effects in the error, the model is estimated. GSPRE shows that there is a spillover effect that varies little in the time given by the coefficient lambda = 0,10, which is statistically significant but small , and that if there are shocks over time that significantly modify the behavior between hospitals. This type of variation is related as it is expressed cite baltagi2013generalized with technological shocks, changes in the facilities of hospitals that allow to significantly change the capacity to perform surgeries. The second article analyzes decentralization in the health system, studying one of the main objectives in public policy: the prevention of diseases, analyzing the effectiveness of the program in performing preventive cytopathological exams in women, given the availability of resources in the basic care teams located in all the municipalities of Brazil. Based on the analysis of the National Melhoria do Acceso e da Qualidade de Aten??o B?sica -AMPprogram, a factorial analysis is performed that calculates the covariances between the personnel, medical and physical supplies available in the equipment, as well as features such as race, income, literacy level, among others, the decision of women to undergo the cytopathological examination, a model is realized that includes a system of structural equations. It concludes that decentralization is due more than to a system formed in a correct way according to national and international parameters, to a management of the bureaucrats, as it was shown, in the states where there is a deficit or an adequate number of health posts , the estimate had positive coefficients, but this beyond the flow of resources, depends on the political will of each one of the government spheres to increase the number of health establishments or health programs. In this way, and in family health programs such as the performance of cytopathological tests, it is taking place in the basic care network in Brazil, in compliance with the Health Pact established for Brazil. Finally, the analysis of the demand for private health plans is made from different variables of individual, family, social, economic and geographical characterization of individuals and units of family consumption or Brazilian households, based on a logit model. It allows determining that individuals of the amarela race have the highest probability of choosing to demand a private health plan in Brazil, specifically in the North, South and Southeast regions of the country. Thus, in the Southeastern region of Brazil, at a level of significance of 1% the probability of demanding private health insurance is greater for people of that skin color or race, an additional control is created: division by range of monthly per capita income level; In this way, it is observed how the possibility of demanding private health plans varies according to the race declared by the individuals associated with the level of income received in per capita terms by the Brazilian family consumption units, which reflects that with a per capita income monthly of 200 reais, regardless of the declared race, all individuals have the same probability of demanding a private health plan in Brazil. / Esta tesis esta compuesta por tres art?culos sobre Econom?a de la Salud. El primero documento busca analizar la posible integraci?n espacial de la red de hospitales p?blicos en Colombia, con el objetivo de analizar si existe eficiencia la momento de remitir los pacientes de acuerdo a el tipo de asistencia demandada en el hospital. A partir de la estimaci?n de un modelo de panel espacial, se analiz? la el total de hospitales que componen el sistema de oferta p?blica, concluyendo que a pesar de que puede existir cierto grado de transferencia en la atenci?n de pacientes es muy baja para el caso de medicina general, dado que el coeficiente espacial del modelo SDM, ? es inferior a uno. a vez realizado el an?lisis para el modelo est?tico se estimo el modelo din?mico SAR y SDM, se estimo un modelo din?mico que descompusiera los efectos en directos e indirectos, donde se evidencio que existe un efecto de distribuci?n espacial a largo plazo para la atenci?n de citas de medicina general determinado por el n?mero de personal asistencial de los 4 hospitales m?s cercanos sin embargo no se encontr? la presencia de transbordamientos en el corto plazo en la atenci?n de medicina general entre los hospitales de la red p?blica, dado que estas que estos se comportan como instituciones de prestaci?n de servicios b?sicos, sin disponer de un personal capacitado que permita una mejor atenci?n a partir del desarrollo de mejores practicas medicas. El mismo articulo calcula los modelos SDM,SAR,SAC para los hospitales de alta complejidad, teniendo como variable dependiente el n?mero de cirug?as practicadas con una matriz de distancia normalizada de 8 vecinos, concluyendo que el n?mero de urgencias atendidas por el hospital incide de manera negativa en el total de cirug?as practicadas, es decir, los hospitales de nivel alto de complejidad realizan en su mayor?a cirug?as programada, no obstante un cambio en el total de emergencias. con el prop?sito de controlar la existencia de efectos espaciales aleatorios en el errores se estimo el modelo GSPRE muestra que existe un efecto spillover que varia poco en el tiempo dado por el coeficiente ? = 0, 10 , que es estad?sticamente significativo pero peque?o, y que si existen shocks en el tiempo que modifican de manera significativa el comportamiento entre los hospitales. Este tipo de variaciones esta relacionado tal como lo manifiesta [21] con shocks tecnol?gicos, cambios en las instalaciones de los hospitales que permitan var?en de manera significativa la capacidad de realizar cirug?as. El segundo articulo analiza la descentralizaci?n en el sistema de salud, estudiando uno de los principales objetivos en la pol?tica p?blica: la prevenci?n de enfermedades, analizando la efectividad del programa en la realizaci?n de examenes citopatologicos preventivos en las mujeres, dada la disponibilidad de recursos en los equipos de atenci?n b?sica situados en todos los municipios de Brasil. A partir del an?lisis del programa Nacional de Melhoria do Acceso e da Qualidade de Aten??o B?sica ?PMAQ-, se hace un an?lisis factorial que calcula las covarianzas entre los insumos de personal, m?dicos y fisicos disponibles en los equipos, as? junto con caracter?sticas como la raza, ingresos, nivel de alfabetizaci?n, entre otras, la decisi?n de las mujeres de realizarse el examen citopatol?gico, se realiza un modelo que comprenda un sistema de ecuaciones estructurales. Se concluye as? que la descentralizaci?n se debe m?s all? que a un sistema conformado de manera correcta seg?n par?metros nacionales e internacionales, a una gesti?n de los bur?cratas, tal como se mostr?, en los estados donde hay d?ficit o un adecuado numero de puestos de salud, la estimaci?n tuvo coeficientes positivos, pero esto m?s all? del flujo de recursos, depende de la voluntad pol?tica de cada una de las esferas del gobierno para aumentar el numero de establecimientos de salud o de programas de salud. De est? manera, y en programas de salud familiar como la realizaci?n de examenes citopatologicos, se esta dando en la red de atenci?n b?sica en Brasil, dando cumplimiento al Pacto por la salud establecido para Brasil. Por ultimo se realiza el an?lisis de la demanda de planes de salud privados a partir de diferentes variables de caracterizaci?n individual, familiar, social, econ?mica y geogr?fica de los individuos y unidades de consumo familiar u hogares brasileros, a partir de un modelo logit. La permite determinar que los individuos de raza amarela, poseen la probabilidad m?s alta de elegir demandar un plan de salud privado en Brasil, espec?ficamente en las regiones Norte, Sur y Sureste del pa?s. De modo tal que en la regi?n sureste de Brasil, a un nivel de significancia del 1% la probabilidad de demandar un aseguramiento privado en salud es mayor para las personas de dicho color de piel o raza, se crea un control adicional: divisi?n por rango de nivel de renta per c?pita mensual; de esta manera se observa c?mo var?a la posibilidad de demandar planes de salud privados de acuerdo a la raza declarada por los individuos asociada al nivel de ingresos percibidos en t?rminos per c?pita por las unidades de consumo familiar brasileras lo cual refleja que con una renta per c?pita mensual de 200 reales, independientemente de la raza declarada, todos los individuos tienen la misma probabilidad de demandar un plan de salud privado en Brasil.