We use a production function approach in investigating the relationship between financial development and economic growth in 9 EU accession - mostly transition countries. These findings are compared with the results for the group of 18 developed countries, and separately, with the results for a group of less developed EU countries - structural fund recipients. We use aggregate measures of financial development as well as measures for single segments of financial sectors. In context of transition countries, bond markets are, to our knowledge, taken explicitly into account for the first time. We find that domestic credit and bond markets together with real capital stock growth stimulate economic growth in transition. With progress in cohesion, educational attainment becomes the next important factor that contributes to economic growth followed by labor participation in mature market economies. For the developed countries, financial sector did not play any positive role for growth over the period under study. We conclude that transfer mechanisms for growth differ over the development cycle. This is important to growth theory, to the sequencing of economic reforms and to financial sector development priorities. (author's abstract) / Series: EI Working Papers / Europainstitut

The objective of the study is to analyze the effects of two breakpoints on the relationships of Istanbul Stock Exchange with the European stock markets and on the relationships among these European stock markets to increase the economic integration. The breakpoints are the execution of the Customs Union Agreement of Turkey with the European Union in 1/1/1996 and the introduction of the Euro in 1/1/1999. While both breakpoints have effects on Turkey&rsquo / s economic relations, the European Union countries are expected to be influenced by only the introduction of the Euro. Stock market indices provided by DataStream is utilized. The statistical techniques used include the correlation and cointegration analysis. Results indicate that when examined on pair wise basis Turkish stock market has more liaisons with the European stock markets, in general, after the Customs Union / but less liaisons after the conversion to Euro. However, when examined as a group, the cointegration result finds the Euro as influential as the Customs Union. Alternatively, the European stock markets have decreasing integrations as a result of correlation analysis after the Euro, but it is an influential breakpoint according to cointegrating structures.

The thesis offers a framework for trading algorithm optimization and tests statistical and economical significance of its performance on American, Swedish and Russian futures markets. The results provide strong support for proposed method, as using the presented ideas one can build an intraday trading algorithm that outperforms the market in long term.

[pt] O Equity Premium Puzzle tem sido muito estudado no mundo desde 1985, ano da publicação do trabalho de Mehra e Prescott. O intuito desta dissertação foi fazer uma análise e avaliação do Equity Premium Puzzle utilizando diferentes contextos vividos na economia brasileira no período de 1990 até 2005. O modelo utilizado foi o do agente representativo com utilidade separável no tempo desenvolvido por Mehra e Prescott (1985). A fim de realizar comparações de resultados foi utilizado também o modelo revisado por Mehra (2003) e um modelo com utilidade tipo Kreps - Porteus com processo de dotação seguindo a cadeia de Markov. / [en] The Equity Premium Puzzle has been very studied in the world since 1985, year of the publication of the work of Mehra and Prescott. The intention of this dissertation was to make an analysis and valuation of the Equity Premium Puzzle being used different contexts lived in the Brazilian economy in the period of 1990 up to 2005. It was used the representative agent model with separable utility in the time developed for Mehra and the Prescott (1985). In order to carry through comparisons of results was used also the model revised for Mehra (2003) and a model with utility type Kreps - Porteus with endowment process having followed the Markov´s chain.

A previsão de séries temporais financeiras tem sido um tópico popular da literatura nos últimos anos. Contudo, embora muitos estudos de previsão de séries temporais foquem na previsão exata de valores futuros, defendemos que este tipo de previsão é de difícil aplicação em cenários reais, sendo mais vantajoso transformar este problema de previsão em um problema de classificação que indique se a série temporal irá subir ou descer no próximo período. Neste trabalho é proposto um método de compra e venda de ações baseado nas previsões feitas por dois ensembles de redes neurais adaptados para diferentes perfis de investimento: um para investidores moderados e outro para investidores mais agressivos. Os resultados desses ensembles preveem se determinada ação irá subir ou descer no próximo período ao invés de prever seus valores futuros, permitindo que se criem recomendações de operações de compra ou venda para o próximo período de tempo. A criação de tais ensembles, contudo, pode encontrar dificuldades no fato de que cada mercado se comporta de uma maneira diferente: fatores como a sazonalidade e a localidade da bolsa de valores são determinantes no desenvolvimento das redes neurais apropriadas. Para mostrar a eficiência do nosso método em diferentes situações, o mesmo é avaliado exaustivamente em dois conjuntos de dados diferentes: os mercados de ações norteamericano (S&P 500) e brasileiro (Bovespa). Operações reais foram simuladas nestes mercados e fomos capazes de lucrar em 89% dos casos avaliados, superando os resultados das abordagens comparativas na grande maioria dos casos. / Financial time series prediction has been a hot topic in the last years. However, although many time series prediction studies focus on the exact prediction for future values, we defend that this kind of prediction is hard to apply in real scenarios, being more profitable to transform the prediction problem into a classification problem that indicates if the time series is going to raise or fall in the next period. In this work we propose a stock buy and sell method based on predictions made by two neural network ensembles adjusted for different investment profiles: one for moderate investors and another for aggressive investors. The results of these ensembles predict if certain stock will raise of fall in the next time period instead of predicting its future values, allowing the creation of buy and sell operations recommendations for the next time period. The creation of such ensembles, however, can find difficulties in the fact that each market behaves in a different manner: factors as the seasonality and the location of the stock market are determinant in the development of the appropriate neural networks. To show the efficiency of our method in different situations, it is tested exhaustively in two differents datasets: the north american (S&P 500) and brazilian (Bovespa) stock markets. Real operations were simulated in these markets and we were able to profit in 89% of the tested cases, outperforming the results of the comparative approaches in most of the cases.

This thesis analyses financial contagion between a reference EU market - Germany and markets of five countries which are actively seeking to become a part of European Union - Montenegro, Serbia, Turkey, Bosnia and Macedonia in the period of March 2006 to March 2018. We apply quantile regression framework to analyse contagion which we base on the occurrence and degree of coexceedances between the reference and analysed market. The results indicate that contagion between stock markets exists, however in different degree for each of the analysed markets. In addition we apply the regression framework specifically for period of financial crisis of 2008 to demonstrate that contagion is stronger during turbulent market periods. JEL Classification G01, G14, G15 Keywords coexceedance, quantile regression, contagion, stock markets Author's e-mail 80605682@fsv.cuni.cz Supervisor's e-mail roman.horvath@fsv.cuni.cz

A previsão de séries temporais financeiras tem sido um tópico popular da literatura nos últimos anos. Contudo, embora muitos estudos de previsão de séries temporais foquem na previsão exata de valores futuros, defendemos que este tipo de previsão é de difícil aplicação em cenários reais, sendo mais vantajoso transformar este problema de previsão em um problema de classificação que indique se a série temporal irá subir ou descer no próximo período. Neste trabalho é proposto um método de compra e venda de ações baseado nas previsões feitas por dois ensembles de redes neurais adaptados para diferentes perfis de investimento: um para investidores moderados e outro para investidores mais agressivos. Os resultados desses ensembles preveem se determinada ação irá subir ou descer no próximo período ao invés de prever seus valores futuros, permitindo que se criem recomendações de operações de compra ou venda para o próximo período de tempo. A criação de tais ensembles, contudo, pode encontrar dificuldades no fato de que cada mercado se comporta de uma maneira diferente: fatores como a sazonalidade e a localidade da bolsa de valores são determinantes no desenvolvimento das redes neurais apropriadas. Para mostrar a eficiência do nosso método em diferentes situações, o mesmo é avaliado exaustivamente em dois conjuntos de dados diferentes: os mercados de ações norteamericano (S&P 500) e brasileiro (Bovespa). Operações reais foram simuladas nestes mercados e fomos capazes de lucrar em 89% dos casos avaliados, superando os resultados das abordagens comparativas na grande maioria dos casos. / Financial time series prediction has been a hot topic in the last years. However, although many time series prediction studies focus on the exact prediction for future values, we defend that this kind of prediction is hard to apply in real scenarios, being more profitable to transform the prediction problem into a classification problem that indicates if the time series is going to raise or fall in the next period. In this work we propose a stock buy and sell method based on predictions made by two neural network ensembles adjusted for different investment profiles: one for moderate investors and another for aggressive investors. The results of these ensembles predict if certain stock will raise of fall in the next time period instead of predicting its future values, allowing the creation of buy and sell operations recommendations for the next time period. The creation of such ensembles, however, can find difficulties in the fact that each market behaves in a different manner: factors as the seasonality and the location of the stock market are determinant in the development of the appropriate neural networks. To show the efficiency of our method in different situations, it is tested exhaustively in two differents datasets: the north american (S&P 500) and brazilian (Bovespa) stock markets. Real operations were simulated in these markets and we were able to profit in 89% of the tested cases, outperforming the results of the comparative approaches in most of the cases.

FundaÃÃo de Amparo à Pesquisa do Estado do Cearà / CoordenaÃÃo de AperfeiÃoamento de Pessoal de NÃvel Superior / A presente pesquisa tem como objetivo o estudo das representaÃÃes da pobreza para os considerados pobres inseridos no Programa Bolsa FamÃlia. O foco do trabalho à adentrar nas representaÃÃes que o âpobreâ, a partir de suas vivÃncias sociais, faz da sua condiÃÃo de pobreza, tendo em vista o contexto de transformaÃÃes sociais que, cada vez mais, tem levado milhares de indivÃduos à condiÃÃo de ârefugoâ social. Diante dessa realidade, a condiÃÃo de âpobreâ se configura numa condiÃÃo socialmente reconhecida e encarnada em vÃrias representaÃÃes relacionadas ao âser pobreâ. Tais questÃes sÃo o fio condutor desta pesquisa, como tambÃm nos instigaram a dimensionar as formas de representaÃÃes da pobreza, como elas refletem nas vivÃncias do âpobreâ, na maneira como tais pobres se apresentam a um programa voltado aos pobres, no modo de interagir com as diversas situaÃÃes e enfim nos espaÃos por eles ocupados a partir do lugar social que a sociedade do capital lhes reserva. / The present research has as objective the study of the poor considered representations of the poverty for inserted in the Program the Stock market Family. The focus of the work is to in the representations that the âpoor personâ, from its social experiences, makes of its condition of poverty, in view of the context of social transformations that, each time more, has taken thousand of individuals to âthe social rubbishâ condition. Ahead of this reality, the condition of âpoor personâ if configures in a recognized and socially incarnate condition in some representations related to the âpoor beingâ. Such questions are the conducting wire of this research, in had as well as instigated them to the forms of representations of the poverty, as they reflect in the experiences of the âpoor personâ, in the way as such poor persons if they present to a come back program the poor persons, in the way to interact at last with the diverse situations and in the spaces for busy them from the social place that the society of the capital them reserve.

[pt] O mercado de capitais das grandes economias mundiais já incorporou o conceito de rating, ou seja, a classificação de risco de crédito, utilizada amplamente nos Estados Unidos há muitos anos. A existência de agências de rating capazes de fornecer classificações de risco de crédito totalmente independentes é uma condição imprescindível para o desenvolvimento de qualquer mercado de dívida. Este estudo tem como objetivo avaliar se as variáveis definidas no contrato de um bônus corporativo afetam o rating determinado por estas instituições para determinada emissão. Foram analisadas as variáveis cupom, preço de emissão, volume de emissão, maturidade e a presença de garantia. Além disso, analisou-se também a influência do risco país representado pelo EMBI +. Quanto ao método de análise, optou-se por comparar o método dos mínimos quadrados ordinários (MQO) com o probit ordenado. Os resultados mostram que não houve diferença de desempenho entre os dois modelos. E quanto às variáveis analisadas, apenas o cupom demonstrou exercer influência sobre o rating da emissão. / [en] The financial markets in the largest economies of the world often utilize rating agencies as a tool for credit risk classification, following the concept introduced in the USA a long time ago. The existence of independent institutions capable of classifying credit risk is a vital condition for the development of debt market. This study aims at analysing whether the variables defined in the indenture of the bond issue affect the rating assigned by rating agencies. The following variables were investigated: coupon, price, issued amount, maturity and the existence of collateral. Furthermore, it was analysed if the country risk, represented by the EMBI+ index, also affected the bond rating. Regarding the methodology, the ordinary least square (OLS) and ordered probit were chosen as the method of analysis. A comparative study was performed and the results indicated no major differences between both models. Concerning the variables analysed, only the coupon has shown some influence on the bond rating.

[pt] Esta tese investiga a decisão por emissões públicas ou particulares de ações no Brasil. É apresentada uma revisão da literatura nacional e internacional sobre o assunto. É relatada a evidência empírica no Brasil. Dois modelos são construídos de forma a elucidar o problema investigado, à luz da literatura revista e da evidência empírica: Análise Discriminante e Regressão Logística. A legislação relativa às emissões e ações no Brasil é resumida. Os resultados apontam para maior propensão à emissão privada de empresas estatais ou holdings, com menor lucratividade (maiores prejuízos), que realizaram emissões de menor porte, apresentavam maior concentração de propriedade, endividamento e menor liquidez em Bolsa de Valores. / [en] This dissertation studies the decision for public issues or private placements of equity in Brazil as a source of capital. The national and international literature on the subject is reviewed. The empirical evidence of placements of the Brazilian capital market and the characteristics of the issuers are also presented. After the discussion of the evidence and literature, two models are constructed: Multiple Discriminant Analysis and Logistic Regression, relating the characteristics of firms to the decision of type of issue. The regulation of stock issues is also discussed. The results show that firms with less liquid stocks, more debt, lower returns on equity and profits, more concentration of control and smaller issues, more probably use private issues.