A absorção é a etapa determinante da recuperação de acetona no Filter Tow, por reduzir a emissão de acetona e trazer melhorias à economia. Por isso, ela constitui o objeto deste estudo, que inclui desde a revisão dos conceitos fenomenológicos, considerando a escolha do melhor modelo de equilíbrio líquido-vapor, passando pela execução dos balanços coerentes de massa e o estabelecimento da Fotografia da situação original, relatando um caso complexo de transposição de pratos para recheios. Contudo, a operação das colunas de absorção é influenciada por outras variáveis de caráter desconhecido, ruídos em relação à fenomenologia, o que se pretende avaliar, justificando-se desenvolver estudo para avaliar seus efeitos. Com o objetivo geral de otimizar absorção, uma análise estatística foi executada a partir do levantamento de dados operacionais, utilizando todas suas variáveis, sejam as fenomenológicas, como os ruídos, com o objetivo específico de obter modelos empíricos que complementem as simulações fenomenológicas, aumentando sua abrangência. Para a execução da análise estatística, os conjuntos de dados históricos foram levantados e validados pelos balanços coerentes de massa e pela Fotografia, o que tornou possível sua evolução, desde a seleção das variáveis, até estabelecer os modelos de regressão, com os quais pode-se obter um novo modo de controle, que estabiliza a operação, possibilitando a otimização. Em termos ambientais, a utilização destes modelos resulta em redução de até 15% das perdas de acetona para o ambiente, como também de consumo energético, com uma economia da ordem de 1 milhão de reais por ano, sem quaisquer custos adicionais. / Absorption is the key step of the acetone recovery at Filter Tow production, for reducing the acetone emission and improving economics. For this reason it is the subject of this study, including the revision of phenomenological concepts, considering the choice of the best vapor liquid equilibrium model, passing by the improvement of coherent mass balances and establishing the Photography of the original situation, detailing a complex case of transposition of sieve trays to structured packing. However, the operation of the absorption columns is influenced by other variables, with unknown impacts, noises in relation to the phenomenology, justifying the development of this study, for evaluating them. With the general objective of optimizing the absorption, a statistical analysis is performed from collecting operating data, considering all variables, phenomenological and noises, with the specific objective of getting empirical models complementing the phenomenological simulations, increasing their comprising. For performing the statistical analysis sets of historical data have been collected and validated by coherent mass balances and the Photography, making possible its evolution, from the selection of the variables till establishing the regression models, and with them getting a new control way, which stabilizes the operation, allowing the optimization. In environmental terms, the use of these models results in up to 15% decrease in acetone losses to the environment, as well as power consumption with a saving of approximately 1 million reais per year, without any additional costs.

In der vorliegenden Arbeit werden verschiedene Fragestellungen zur Integration und Transformation auf den europäischen Agrar- und Ernährungsmärkten untersucht. Eine Weiterentwicklung der Gemeinsamen Agrarpolitik, die anstehende Osterweiterung der Europäischen Union (EU) sowie Auswirkungen von Ausländischen Direktinvestitionen (ADI) im Ernährungssektor mittel- und osteuropäischer Transformationsländer werden mit Hilfe von partiellen und allgemeinen Gleichgewichtsmodellen simuliert. Das Ziel der Arbeit ist die separate, quantitative Analyse ausgewählter Aspekte der Integration und Transformation sowie das Aufzeigen von Wechselwirkungen zwischen ihnen. Die Ergebnisse sollen dazu dienen, die politischen Rahmenbedingungen in der europäischen Agrar- und Ernährungswirtschaft zu verbessern. Als Hauptinstrumente einer neuen EU-Agrarpolitik werden einheitliche Boden- und Arbeitssubventionen verbunden mit einem weiteren Abbau des Außenschutzes untersucht. Diese würden zu wesentlich geringeren Verzerrungen auf den Produktmärkten und zu deutlichen Budgeteinsparungen gegenüber der derzeitigen Situation führen. Eine einheitliche Bodensubvention mit einer weitergehenden Liberalisierung wird auch als Politikoption für die EU-Osterweiterung analysiert. Der Handel mit Agrar- und Ernährungsgütern innerhalb einer erweiterten EU würde sich in einigen Szenarien verdoppeln. Durch Handelsgewinne und Transferzahlungen aus dem EU-Budget käme es in den Beitrittsländern zu einem Wohlfahrtsgewinn in Höhe von etwa zwei Prozent des Bruttosozialprodukts. Die Bedeutung von ADI im Transformationsprozeß sollte nicht überschätzt werden. Die Modellrechnungen ergeben, daß der Zustrom von ADI seit 1992 zu einem zusätzlichen jährlichen Wachstum des Bruttosozialprodukts von unter einem Prozent in den mitteleuropäischen Ländern und in den Nachfolgestaaten der Sowjetunion geführt hat. Durch zusätzlichen Technologietransfer profitiert die Agrarwirtschaft nicht immer von ADI in der Nahrungsmittelverarbeitung. Grund hierfür sind zum Teil Einsparungen bei der Verwendung landwirtschaftlicher Rohprodukte. Schließlich zeigt die Analyse von ADI in der polnischen Zuckerindustrie, daß handelspolitische Eingriffe, wie z.B. Produktionsquoten, die lokalen Auswirkungen von ADI stark beeinflussen können. ADI sind nur dann deutlich wohlfahrtssteigernd, wenn sie auf relativ unverzerrten Märkten getätigt werden. Die verbesserte Wettbewerbssituation in der Zuckerindustrie würde zu teilweise deutlichen Gewinnen für die Zuckerrübenproduzenten führen. / This dissertation consists of four essays covering various aspects of integration and transition on European agricultural and food markets. Further reform of the European Union's (EU) Common Agricultural Policy, a prospective Eastern enlargement, and the effects of foreign direct investment (FDI) in food industries of the transition countries are analyzed using partial as well as general equilibrium modeling approaches. The overall objective of the study is to quantify these processes separately as well as to demonstrate various interactions between them. The results should be useful for improving the political and economic environment in the European agricultural and food sector. In the discussion about further reform of the Common Agricultural Policy, uniform payments on agricultural land and labor have been proposed in connection with further reductions of border protection. The analysis shows that these policy options would be much less distortionary on product markets and they would lead to significant reductions in budget expenditures compared to the present situation. Furthermore, a uniform payment on land together with further liberalization is also analyzed as a policy option for the EU Eastern enlargement. Trade in agricultural and food products in an enlarged EU would double in some scenarios. Gains from trade and transfer payments from the EU budget would add up to a welfare gain of about two percent of total gross domestic product in the new member countries. The impact of FDI in the transition process should not be over-estimated. The model calculations show that total FDI inflows since 1992 induced additional economic growth of less than one percent per year in the Central European countries and the former Soviet Union. Additional transfers of new technologies into the food industry do not necessarily improve the situation in primary agriculture. The reason is partly input saving technical change which reduces the demand for agricultural raw products. Finally, the analysis of FDI in the Polish sugar industry reveals that trade policy interventions, like production quotas, have a strong influence on the local impact of FDI. Foreign investment is welfare improving to the local economy only if it occurs on more or less undistorted markets. Improved competition in the sugar industry would lead to considerable gains for local sugar beet producers.

Vietnam und Thailand sind zwei der führenden Exporteure für Reis. Zurzeit hat die staatliche Preispolitik den stärksten und meisten Einfluss auf die vietnamesische Branche für Reis. Es beschränkt die Exportmenge, um die Menge für den nationalen Nahrungsmittelbedarf sicherzustellen. Diese Regulierung hält die vietnamesische Produktion unter ihrem vollständigen potenzial, im Vergleich zu der Reisproduktion in Thailand, zurück. Viele Experten unterstellen, dass Vietnam mehr Aufmerksamkeit der nationalen Strategie zur Ernährungssicherung schenkt als es notwendig wäre. Das verursacht Marktverzerrung und schwächt den landeseigenen Reisexport im Vergleich zu Thailand. Um die Auswirkungen der staatlichen Preispolitik mit einer quantitativen Methode zu untersuchen, wird über ein Gleichgewichtsmodell drei verschiedenen Szenarien eruiert: (1)Die Regulierungsrichtlinie für Reispreise wird wöchentlich aktualisiert; (2)Die Regulierungsrichtlinie für Reispreise wird monatlich angepasst; (3)Die Regulierungsrichtlinie für Reispreise wird quartalsweise überabeitet. Basierend aus den Produktionsdaten, dem Konsum, inländischen Preis, Transportkosten, der Elastizität der Nachfrage und der Angebotsfunktion der vietnamesischen und thailändischen Reisbranche zeigt dieses Modell Resultat: Mit kleinen Anpassungen in der Preispolitik wächst der Wettbewerb in Bezug auf der vietnamesischen Exportmenge auf dem internationalen Markt ohne negative Auswirkungen auf die Ziele der nationalen Ernährungssicherung. Nicht-Reisbauern werden in eine nachteilige Position gestellt aufgrund von höheren Preisen auf dem inländischen Markt. Jedoch Bevölkerung, welche Reisbauern sind, werden von den höheren Handelsmengen, den steigenden Umsätzen und ebenfalls von dem steigenden Nettoeinkommen der Reisexporteure profitieren. Deshalb empfehlen wir nicht die staatliche Preispolitik regelmäßig zu überarbeiten, um das Ziel einer besseren Wetbewerbsfähigkeit von Vietnam zu erreichen / Vietnam and Thailand are the top two rice exporters who contribute more than 50 per cent of market shares in the international market. Therefore, any changes in their rice policies have a strong influence on the world market. Currently, one of the strongest and most often impacts on the Vietnamese rice industry is the Price Controls Policy that regulates the competition in rice market. It restricts the export amount in order to ensure national food security. This policy keeps Vietnamese rice production under its full potential compared with the rice industry in Thailand. Many experts suggest that Vietnam pays more attention on national food security than necessary. This causes market distortion and weakens the country’s competitiveness with Thailand. In order to examine the effects of the Price Controls Policy with a quantitative method, we build and run a spatial equilibrium model with 3 different scenarios: (1) Price Controls Policy updated every week; (2) Price Controls Policy updated every month; (3) Price Controls Policy updated quarterly. Base on the available data of production, consumption, domestic price, transportation cost and elasticities of demand and supply function of Vietnamese and Thai rice industry, the model shows the following result. With less changes in the Price Control policy, the competitiveness in terms of export quantity of the Vietnamese rice on international markets rises without having negative effects on the national food security goal. Non-rice farmers will be in disadvantage position due to higher domestic prices for rice, but the majority of the population who are rice farmers, accounting for more than 70 per cent of the population, will benefit from the higher volume and turnover of rice export, and so the net social revenue will increase also. Therefore, we highly recommend that the Price Controls Policy should not be revised on regular to serve the purpose of achieving better competitiveness of Vietnamese rice

O objetivo deste trabalho é investigar o desempenho de estimadores baseados em momentos das famílias verossimilhança empírica generalizada (GEL) e mínimo contraste generalizado (GMC) na estimação de modelos de equilíbrio geral dinâmico e estocástico (DSGE), com enfoque na análise de robustez sob má-especificação, recorrente neste tipo de modelo. Como benchmark utilizamos método do momentos generalizado (GMM), máxima verossimilhança (ML) e inferência bayesiana (BI). Trabalhamos com um modelo de ciclos reais de negócios (RBC) que pode ser considerado o núcleo de modelos DSGE, apresenta dificuldades similares e facilita a análise dos resultados devido ao menor número de parâmetros. Verificamos por meio de experimentos de Monte Carlo se os estimadores estudados entregam resultados satisfatórios em termos de média, mediana, viés, erro quadrático médio, erro absoluto médio e verificamos a distribuição das estimativas geradas por cada estimador. Dentre os principais resultados estão: (i) o estimador verossimilhança empírica (EL) - assim como sua versão com condições de momento suavizadas (SEL) - e a inferência bayesiana (BI) foram, nesta ordem, os que obtiveram os melhores desempenhos, inclusive nos casos de especificação incorreta; (ii) os estimadores continous updating empirical likelihood (CUE), mínima distância de Hellinger (HD), exponential tilting (ET) e suas versões suavizadas apresentaram desempenho comparativo intermediário; (iii) o desempenho dos estimadores exponentially tilted empirical likelihood (ETEL), exponential tilting Hellinger distance (ETHD) e suas versões suavizadas foi bastante comprometido pela ocorrência de estimativas atípicas; (iv) as versões com e sem suavização das condições de momento dos estimadores das famílias GEL/GMC apresentaram desempenhos muito similares; (v) os estimadores GMM, principalmente no caso sobreidentificado, e ML apresentaram desempenhos consideravelmente abaixo de boa parte de seus concorrentes / The objective of this work is to investigate the performance of moment-based estimators of the generalized empirical likelihood (GEL) and generalized minimum contrast (GMC) families in the estimation of dynamic stochastic general equilibrium (DSGE) models, focusing on the robustness analysis under misspecification, recurrent in this model. As benchmark we used generalized method of moments (GMM), maximum likelihood (ML) and Bayesian inference (BI). We work with a real business cycle (RBC) model that can be considered the core of DSGE models, presents similar difficulties and facilitates the analysis of results due to lower number of parameters. We verified, via Monte Carlo experiments, whether the studied estimators presented satisfactory results in terms of mean, median, bias, mean square error, mean absolute error and we verified the distribution of the estimates generated by each estimator. Among the main results are: (i) empirical likelihood (EL) estimator - as well as its version with smoothed moment conditions (SEL) - and Bayesian inference (BI) were, in that order, the ones that obtained the best performances, even in misspecification cases; (ii) continuous updating empirical likelihood (CUE), minimum Hellinger distance (HD), exponential tilting (ET) estimators and their smoothed versions exhibit intermediate comparative performance; (iii) performance of exponentially tilted empirical likelihood (ETEL), exponential tilting Hellinger distance (ETHD) and its smoothed versions was seriously compromised by atypical estimates; (iv) smoothed and non-smoothed GEL/GMC estimators exhibit very similar performances; (v) GMM, especially in the over-identified case, and ML estimators had lower performance than their competitors

La scelta del modeller riguardo alla chiusura del modello CGE influenza i suoi risultati finali e le sue prescrizioni di policy. In questa tesi, lo scopo è l’analisi e l’identificazione del problema, sia attraverso una discussione teorica che un’ applicazione pratica. Dall’articolo del 1963 di Amartya Sen in poi, la letteratura ha presentato vari articoli sull’argomento. Attualmente, il problema delle chiusure del modello non appare più centrale nel dibattito. Dopo una breve introduzione su cosa siano i CGE, il loro sviluppo e la loro struttura, è presentata una serie di esemplificative maquette con lo scopo di introdurre al concetto di chiusura, come essa influenzi i risultati e come questa scelta del modeller sia intimamente connessa ai fondamenti macroeconomici del sistema. Dopo la teoria, ci si sposta nel mondo reale analizzando con differenti modelli (Neoclassico, “Bastardo Keynesiano”, Strutturalista/Post- Keynesiano) e diverse chiusure per gli aggregati macroeconomici (risparmi privati, pubblici, e stranieri) l’impatto dell’accordo regionale SADC sull’economia mozambicana. I modelli CGE per il Mozambico sono calibrati su una SAM del 2003 e sono svolti con l’ausilio di GAMS/MPSGE. I risultati dimostrano che la chiusura influenza i risultati stessi del modello per cui ognuno presenta una serie di raccomandazioni politiche per l’applicazione dell’accordo SADC. / Modellers’ choice on closure rules affects a CGE model results and consequently its policy prescriptions. In this thesis, the aim is to detect and assess this issue, both through a theoretical discussion and an empirical application. Starting from Amartya Sen’s 1963 paper, literature presents many contributions on this topic. Currently, the closure rule problem is not central in the CGE debate. After a brief introduction on CGEs, their development and their structure, a series of simple maquette is presented. They have the exemplary role of introducing the concept of closures, explain how they affect final outcomes and how this modeller’s choice is strictly connected to the macroeconomic foundation of the economic system. After theory, we move into the real World analyzing through different models (Neoclassical, “Bastard Keynesian”, and Structuralist/ Post- Keynesian), and through different closure rules for macro- aggregates (private, public and foreign savings) the impact of the Regional Trade Agreement of SADC with respect to the Mozambican economy. The Mozambican CGE models are calibrated on a 2003 Social Accounting Matrix (SAM) and they are performed using GAMS/ MPSGE. Outcomes show that closure rules affect them and each model presents a set of policy prescription to implement the SADC agreement.

Trade liberalization in Kenya started in the early 1980s with the structural adjustment programmes, and continued under the multilateral framework of the WTO. During the same period, the incidence of poverty and level of inequality also worsened. The government’s focus on trade negotiations has been to ensure that there is policy space for the daily running of the economy even though welfare impacts are also important. Non-state actors have argued that trade liberalization has negatively affected the poor; particularly the farmers, since they cannot compete with the developed countries whose farmers enjoy significant government support through subsidies, making their products much cheaper in the world market. Government officials, on the other hand, contend that trade liberalization is good as it brings in competition and transfer of technology which is good for an economy. It is important to examine how trade liberalization has affected household’s welfare in Kenya, given that this kind of analysis has not been conducted in Kenya. This study is unique because it does not assume the existence of a trade liberalization– poverty relationship, unlike most studies. It uses a multi-method approach to first test the hypothesis that there is no statistically significant relationship between trade liberalization and poverty, it further tests for multiplier effects of trade liberalization on poverty determinants. Trade Liberalization and poverty is found to have a stochastic relationship, furthermore investments and capital stock were found to significantly affect poverty determinants in the stochastic model. Due to unavailability of household welfare measure data in time series, a CGE model was used to establish the dynamics of trade liberalization on poverty at a point in time using the 2003 Social Accounting Matrix Data for Kenya. Overall, trade liberalization accompanied by FDI had the greatest impact on household welfare. Trade liberalization had a positive impact on household welfare since household incomes and consumption increased. Micro simulations results, based on changes in consumption, also showed that poverty incidence reduced for all households, even though the urban households experienced higher decreases. The study found that there was little difference in protecting sensitive products and not protecting them; secondly, trade liberalization accompanied by foreign direct investment had greater impact on improving the household welfare. Consumption and incomes increased, resulting in overall poverty reduction. The welfare of urban households was much higher than rural households in terms of income and consumption increases. However, income inequality was much higher in urban than rural areas. / Economics / D. Litt. et Phil. (Economics)

A absorção é a etapa determinante da recuperação de acetona no Filter Tow, por reduzir a emissão de acetona e trazer melhorias à economia. Por isso, ela constitui o objeto deste estudo, que inclui desde a revisão dos conceitos fenomenológicos, considerando a escolha do melhor modelo de equilíbrio líquido-vapor, passando pela execução dos balanços coerentes de massa e o estabelecimento da Fotografia da situação original, relatando um caso complexo de transposição de pratos para recheios. Contudo, a operação das colunas de absorção é influenciada por outras variáveis de caráter desconhecido, ruídos em relação à fenomenologia, o que se pretende avaliar, justificando-se desenvolver estudo para avaliar seus efeitos. Com o objetivo geral de otimizar absorção, uma análise estatística foi executada a partir do levantamento de dados operacionais, utilizando todas suas variáveis, sejam as fenomenológicas, como os ruídos, com o objetivo específico de obter modelos empíricos que complementem as simulações fenomenológicas, aumentando sua abrangência. Para a execução da análise estatística, os conjuntos de dados históricos foram levantados e validados pelos balanços coerentes de massa e pela Fotografia, o que tornou possível sua evolução, desde a seleção das variáveis, até estabelecer os modelos de regressão, com os quais pode-se obter um novo modo de controle, que estabiliza a operação, possibilitando a otimização. Em termos ambientais, a utilização destes modelos resulta em redução de até 15% das perdas de acetona para o ambiente, como também de consumo energético, com uma economia da ordem de 1 milhão de reais por ano, sem quaisquer custos adicionais. / Absorption is the key step of the acetone recovery at Filter Tow production, for reducing the acetone emission and improving economics. For this reason it is the subject of this study, including the revision of phenomenological concepts, considering the choice of the best vapor liquid equilibrium model, passing by the improvement of coherent mass balances and establishing the Photography of the original situation, detailing a complex case of transposition of sieve trays to structured packing. However, the operation of the absorption columns is influenced by other variables, with unknown impacts, noises in relation to the phenomenology, justifying the development of this study, for evaluating them. With the general objective of optimizing the absorption, a statistical analysis is performed from collecting operating data, considering all variables, phenomenological and noises, with the specific objective of getting empirical models complementing the phenomenological simulations, increasing their comprising. For performing the statistical analysis sets of historical data have been collected and validated by coherent mass balances and the Photography, making possible its evolution, from the selection of the variables till establishing the regression models, and with them getting a new control way, which stabilizes the operation, allowing the optimization. In environmental terms, the use of these models results in up to 15% decrease in acetone losses to the environment, as well as power consumption with a saving of approximately 1 million reais per year, without any additional costs.

O objetivo deste trabalho é investigar o desempenho de estimadores baseados em momentos das famílias verossimilhança empírica generalizada (GEL) e mínimo contraste generalizado (GMC) na estimação de modelos de equilíbrio geral dinâmico e estocástico (DSGE), com enfoque na análise de robustez sob má-especificação, recorrente neste tipo de modelo. Como benchmark utilizamos método do momentos generalizado (GMM), máxima verossimilhança (ML) e inferência bayesiana (BI). Trabalhamos com um modelo de ciclos reais de negócios (RBC) que pode ser considerado o núcleo de modelos DSGE, apresenta dificuldades similares e facilita a análise dos resultados devido ao menor número de parâmetros. Verificamos por meio de experimentos de Monte Carlo se os estimadores estudados entregam resultados satisfatórios em termos de média, mediana, viés, erro quadrático médio, erro absoluto médio e verificamos a distribuição das estimativas geradas por cada estimador. Dentre os principais resultados estão: (i) o estimador verossimilhança empírica (EL) - assim como sua versão com condições de momento suavizadas (SEL) - e a inferência bayesiana (BI) foram, nesta ordem, os que obtiveram os melhores desempenhos, inclusive nos casos de especificação incorreta; (ii) os estimadores continous updating empirical likelihood (CUE), mínima distância de Hellinger (HD), exponential tilting (ET) e suas versões suavizadas apresentaram desempenho comparativo intermediário; (iii) o desempenho dos estimadores exponentially tilted empirical likelihood (ETEL), exponential tilting Hellinger distance (ETHD) e suas versões suavizadas foi bastante comprometido pela ocorrência de estimativas atípicas; (iv) as versões com e sem suavização das condições de momento dos estimadores das famílias GEL/GMC apresentaram desempenhos muito similares; (v) os estimadores GMM, principalmente no caso sobreidentificado, e ML apresentaram desempenhos consideravelmente abaixo de boa parte de seus concorrentes / The objective of this work is to investigate the performance of moment-based estimators of the generalized empirical likelihood (GEL) and generalized minimum contrast (GMC) families in the estimation of dynamic stochastic general equilibrium (DSGE) models, focusing on the robustness analysis under misspecification, recurrent in this model. As benchmark we used generalized method of moments (GMM), maximum likelihood (ML) and Bayesian inference (BI). We work with a real business cycle (RBC) model that can be considered the core of DSGE models, presents similar difficulties and facilitates the analysis of results due to lower number of parameters. We verified, via Monte Carlo experiments, whether the studied estimators presented satisfactory results in terms of mean, median, bias, mean square error, mean absolute error and we verified the distribution of the estimates generated by each estimator. Among the main results are: (i) empirical likelihood (EL) estimator - as well as its version with smoothed moment conditions (SEL) - and Bayesian inference (BI) were, in that order, the ones that obtained the best performances, even in misspecification cases; (ii) continuous updating empirical likelihood (CUE), minimum Hellinger distance (HD), exponential tilting (ET) estimators and their smoothed versions exhibit intermediate comparative performance; (iii) performance of exponentially tilted empirical likelihood (ETEL), exponential tilting Hellinger distance (ETHD) and its smoothed versions was seriously compromised by atypical estimates; (iv) smoothed and non-smoothed GEL/GMC estimators exhibit very similar performances; (v) GMM, especially in the over-identified case, and ML estimators had lower performance than their competitors

La nouvelle stratégie d'intégration adoptée en 2000 par les pays Sud-Américains, après trois décennies d'instabilité économique et de crises récurrentes, est un jalon de l'histoire économique de la région. En effet, la volatilité du cycle économique de ces pays s'est réduite significativement à partir de cette date, atteignant son niveau le plus bas depuis 1950. L'analyse d'un tel phénomène est particulièrement intéressante en particulier lorsque l'on se place dans le contexte de turbulences et de crises des années 2000, à savoir, la crise financière mondiale (2008-2009) et, dans son sillage, la crise des dettes souveraines en zone euro. Dans cette thèse, l'objectif est d'étudier le projet d'intégration régionale d'Amérique du Sud, institutionnalisé en 2008 avec la création de l'Union des Nations Sud-Américaines Unasur, en tant que vecteur de stabilisation de ces économies. De ce fait, il s'agit de concentrer l'analyse sur les interactions entre les douze pays du continent Sud-Américain – Argentine, Bolivie, Brésil, Chili, Colombie, Équateur, Guyana, Paraguay, Pérou, Uruguay, Suriname et Venezuela – qui forment un groupe hétérogène autour d'un objectif commun l' "… intégration culturelle, sociale, économique et politique …" et la "… réduction des asymétries de la qualité de vie de ses citoyens … ". La thèse s'intéresse exclusivement aux aspects économiques d'un tel projet d'intégration régionale. À partir d'outils empiriques et théoriques, nous cherchons à évaluer le niveau de convergence et de vulnérabilité des économies concernées. Plus particulièrement une analyse des impacts des politiques d'intégration dans court terme et une étude de leurs performances macroéconomiques de long terme. La thèse se divise en quatre chapitres et s'appuie sur des modèles qui intègrent diverses sources de diffusion des chocs asymétriques. Le premier chapitre présente l'état de l'art de la théorie d'intégration économique régionale en soulignant le cas Sud-Américain. Le deuxième chapitre analyse, à l'aide de modèles vectoriels autorégressifs structurels et de mesures de corrélation, l'impact de chocs externes sur les secteurs réel, monétaire et budgétaire des pays membres de l'Unasur. L'analyse montre que : (i) même les pays les plus fermés (Argentine et Venezuela) et les plus industrialisées (Brésil) présentent une forte vulnérabilité aux perturbations internationales, (ii) cette vulnérabilité individuelle se traduit en une convergence de court terme des trajectoires des principales variables macroéconomiques des pays concernés. Dans le troisième chapitre, on cherche à mesurer le degré de convergence de long terme des niveaux de vie des citoyens Sud-Américains à l’aide de modèles empiriques vectoriels à correction d'erreur et de techniques de cointégration. Les résultats montrent l'existence de tendances stochastiques communes à long terme. Cela signifie que les pays sont engagés dans un processus d'évolution vers un objectif commun, autrement dit, que les conditions de vie des citoyens Sud-Américains ne divergent pas à long terme. En fin, le troisième chapitre vise à analyser l'impact de l'investissement dans la construction de réseaux régionaux de transport, de communication et d'énergie, sur la réduction de l'hétérogénéité structurelle des pays de l'Unasur (projet IIRSA). En effectuant un certain nombre d'expériences de politique dans un cadre théorique, cette analyse constate que : (i) une accroissement d'investissement public en infrastructure suscite une augmentation du commerce intra-intra-régional mais pas forcément une réduction de l'écart de production entre les pays, (ii) l'écart de production à long terme entre l'Argentine et le Brésil diminue, dans un scénario gagnant-gagnant, en termes de croissance économique, seulement si les gouvernements de ces deux pays coordonnent leur augmentation d'investissement en infrastructure, comme proposé par l'IIRSA. / Economic integration seems to be a new global trend. The past two decades have witnessed the formation of several economic unions in Asia (ASEAN+3 in 1997), Europe (Eurozone in 1999), Africa, and America (Union of South American Nations, Unasur in 2008). The South American case deserves special attention because, unlike the other blocs, the Unasur emerged as a political alliance and not as an economic one. Furthermore, Unasur is conceived as a strategy for improving the socioeconomic conditions of nations that have a common history of economic instability and external dependence. However, while common concerns and political willingness exist among group members, the question of whether that consensus is sufficient to ensure economic integration remains unanswered. For instance, economic integration as a strategy for macroeconomic stability has seemed to work well in Europe after the euro was launched in 1999 (Sapir, 2011), until the breakdown of the European sovereign debt crisis in recent years has revealed the inherent weaknesses of an economic union that lacks a political union (Fligstein et al., 2012, Issing, 2011). This development suggests that the Unasur project is likely to fail if the concerned economies do not converge economically. This is the reason why, this thesis assesses the Unasur project from an economic integration perspective, thus, complementing the huge body of political literature that has been developed on the issue (Briceño-Ruiz, 2014, Sanahuja, 2012). The first chapter describes the theory of economic integration' state of art focusing on South America. The second chapter examines the reactions of the Unasur economies to external shocks. By using a structural vector autoregression approach, it measures the impact of three external shocks (monetary, commercial, and financial) in the real, monetary, and fiscal economic sectors of Unasur economies and investigates co-movement paths. The results show (i) a non-negligible degree of synchronization across the studied economies, confirming their high external vulnerability, (ii) irrespective of size or integration degree, all Unasur members share mutual weaknesses, which they must fight to overcome. The third chapter evaluates the convergence in real GDP per-capita, as a suitable proxy measure, of the concerned economies for the period 1951-2011. By relying on cointegration techniques and applying Bernard and Durlauf's (1995) stochastic definitions of convergence and common trends, the presented evidence supports the existence of common long-run trends driving output in South America, meaning that the region is involved in a dynamic process of convergence in living standards. Finally, the fourth chapter studies the economic spillovers of the most advanced structural project of the group: the Initiative for the Integration of Regional Infrastructure in South America (IIRSA). A micro-founded two-country general equilibrium model is constructed to evaluate potential gains or losses (in terms of output convergence and trade integration) of raising publicly provided transportation infrastructure in a coordinated and uncoordinated manner. The model is solved using data from Argentina and Brazil. Results show that: (i) rising public investment in infrastructure boost commercial integration but not necessarily generates output converge, (ii) the only way for the Argentina and Brazil to achieve output convergence is to coordinate their increments on public infrastructure as proposed by the IIRSA.

Concorder un modèle à certaines observations, voilà qui résume assez bien ce que l’apprentissage machine cherche à accomplir. Ce concept est maintenant omniprésent dans nos vies, entre autre grâce aux percées récentes en apprentissage profond. La stratégie d’optimisation prédominante pour ces deux domaines est la minimisation d’un objectif donné. Et pour cela, la méthode du gradient, méthode de premier-ordre qui modifie les paramètres du modèle à chaque itération, est l’approche dominante. À l’opposé, les méthodes dites de second ordre n’ont jamais réussi à s’imposer en apprentissage profond. Pourtant, elles offrent des avantages reconnus qui soulèvent encore un grand intérêt. D’où l’importance de la méthode du col, qui unifie les méthodes de premier et second ordre sous un même paradigme. Dans ce mémoire, nous établissons un parralèle direct entre la méthode du col et le domaine du contrôle optimal ; domaine qui cherche à optimiser mathématiquement une séquence de décisions. Et certains des problèmes les mieux compris et étudiés en contrôle optimal sont les commandes linéaires quadratiques. Problèmes pour lesquels on connaît très bien la solution optimale. Plus spécifiquement, nous démontrerons l’équivalence entre une itération de la méthode du col et la résolution d’une Commande Linéaire Quadratique (CLQ). Cet éclairage nouveau implique une approche unifiée quand vient le temps de déployer nombre d’algorithmes issus de la méthode du col, tel que la méthode du gradient et celle des gradients naturels, sans être limitée à ceux-ci. Approche que nous étendons ensuite aux problèmes à horizon infini, tel que les modèles à équilibre profond. Ce faisant, nous démontrons pour ces problèmes que calculer les gradients via la différentiation implicite revient à employer l’équation de Riccati pour solutionner la CLQ associée à la méthode du gradient. Finalement, notons que l’incorporation d’information sur la courbure du problème revient généralement à rencontrer une inversion matricielle dans la méthode du col. Nous montrons que l’équivalence avec les CLQ permet de contourner cette inversion en utilisant une approximation issue des séries de Neumann. Surprenamment, certaines observations empiriques suggèrent que cette approximation aide aussi à stabiliser le processus d’optimisation quand des méthodes de second-ordre sont impliquées ; en agissant comme un régularisateur adaptif implicite. / Machine learning entails training a model to fit some given observations, and recent advances in the field, particularly in deep learning, have made it omnipresent in our lives. Fitting a model usually requires the minimization of a given objective. When it comes to deep learning, first-order methods like gradient descent have become a default tool for optimization in deep learning. On the other hand, second-order methods did not see widespread use in deep learning. Yet, they hold many promises and are still a very active field of research. An important perspective into both methods is steepest descent, which allows you to encompass first and second-order approaches into the same framework. In this thesis, we establish an explicit connection between steepest descent and optimal control, a field that tries to optimize sequential decision-making processes. Core to it is the family of problems known as Linear Quadratic Regulation; problems that have been well studied and for which we know optimal solutions. More specifically, we show that performing one iteration of steepest descent is equivalent to solving a Linear Quadratic Regulator (LQR). This perspective gives us a convenient and unified framework for deploying a wide range of steepest descent algorithms, such as gradient descent and natural gradient descent, but certainly not limited to. This framework can also be extended to problems with an infinite horizon, such as deep equilibrium models. Doing so reveals that retrieving the gradient via implicit differentiation is equivalent to recovering it via Riccati’s solution to the LQR associated with gradient descent. Finally, incorporating curvature information into steepest descent usually takes the form of a matrix inversion. However, casting a steepest descent step as a LQR also hints toward a trick that allows to sidestep this inversion, by leveraging Neumann’s series approximation. Empirical observations provide evidence that this approximation actually helps to stabilize the training process, by acting as an adaptive damping parameter.