[pt] No Brasil, grandes consumidores podem estabelecer seus contratos de energia elétrica em dois ambientes: Ambiente de Contratação Regulado e Ambiente de Contratação Livre. Grandes consumidores são aqueles que possuem carga igual ou superior a 2 MW e podem ser atendidos sob contratos firmados em quaisquer um desses ambientes. Já os consumidores com demanda contratada inferior a 2 MW e superior a 500 kW podem ter seu contrato de energia estabelecido no Ambiente de Contratação Livre proveniente de geração de energia renovável ou no Ambiente de Contratação Regulada através das distribuidoras de energia. A principal vantagem do Ambiente de Contratação Livre é a possibilidade de negociar contratos com diferentes parâmetros, como, por exemplo, preço, quantidade de energia e prazo. Eventuais diferenças entre a energia contratada e a consumida, são liquidadas ao preço de energia de curto prazo, que pode ser bastante volátil.Neste caso o desafio é estabelecer uma estratégia de contratação que minimize os riscos associados a este ambiente. Esta dissertação propõe uma metodologia que envolve a simulação estatística de cenários correlacionados de energia, demanda máxima e preço de curto prazo (também chamado de PLD – Preço de Liquidação das Diferenças) para serem inseridos em um modelo matemático de otimização estocástica, que define os parâmetros ótimos da contratação de energia e demanda. Na parte estatística, um modelo Box e Jenkins é usado para estimar os parâmetros das séries históricas de energia e demanda máxima com o objetivo de simular cenários correlacionados com o PLD. Na parte de otimização, emprega-se uma combinação convexa entre Valor Esperado (VE) e Conditional Value-at-Risk (CVaR) como medidas de risco para encontrar os valores ótimos dos parâmetros contratuais, como a demanda máxima contratada, o volume mensal de energia a ser contratado, além das flexibilidades inferior e superior da energia contratada. Para ilustrar a abordagem proposta, essa metodologia é aplicada a um estudo de caso real para um grande consumidor no Ambiente de Contratação Livre. Os resultados indicaram que a metodologia proposta pode ser uma ferramenta eficiente para consumidores no Ambiente de Contratação Livre e, dado à natureza do modelo, pode ser generalizado para diferentes contratos e mercados de energia. / [en] In Brazil, big consumers can choose their energy contract between two different energy environments: Regulated Contract Environment and Free Contract Environment. Big consumers are characterized by installed load capacity equal or greater than 2 MW and can firm an energy contract under any of these environments. For those consumers with installed load lower than 2 MW and higher than 500 kW, their energy contracts can be firmed in the Free Contract Environment using renewable energy generation or in the Regulated Contract Environment by local distribution companies. The main advantage of the Free Market Environment is the possibility of negotiating contracts with different parameters such as, for example, price, energy quantity and deadlines. Possible differences between contracted energy and consumed energy are settled by the spot price, which can be rather volatile. In this case, the challenge is to establish a contracting strategy that minimize the associated risks with this environment. This thesis proposes a methodology that involves statistical simulation of correlated energy, peak demand and Spot Price scenarios to be used in a stochastic optimization model that defines the optimal energy and demand contract parameters. In the statistical part, a Box and Jenkins model is used to estimate parameters for energy and peak demand in order to simulate scenarios correlated with Spot Price. In the optimization part, a convex combination of Expected Value (EV) and Conditional Value-at-Risk (CVaR) is used as risk measures to find the optimal contract parameters, such as the contracted peak demand, the seasonal energy contracted volumes, in addition to the upper and lower energy contracted bound. To illustrate this approach, this methodology is applied in a real case study for a big consumer with an active Free Market Environment contract. The results indicate that the proposed methodology can be a efficient tool for consumers in the Free Contract Environment and, due to the nature of the model, it can be generalized for different energy contracts and markets.

[pt] No Brasil, a demanda de gás natural possui um comportamento estocástico devido ao consumo das usinas termelétricas, as quais operam em regime de complementariedade ao sistema hidrelétrico. O suprimento de gás natural para estas usinas depende em grande parte do fornecimento de Gás Natural Liquefeito (GNL) spot, importado através de navios metaneiros. Em função do tempo de trânsito dos navios, as compras de GNL devem ocorrer com antecedência em relação ao despacho hidrotérmico. Este descasamento de tempo incentiva a utilização de mecanismos de compatibilização da dinâmica do setor elétrico com a dinâmica da cadeia do gás natural. Uma possibilidade de aumentar a sinergia entre estes domínios é utilizar uma estocagem de gás natural para inserir flexibilidade no sistema. A viabilidade da estocagem dependerá do preço do gás e da demanda ao longo do horizonte de análise. O objetivo deste trabalho é a construção de um modelo de programação linear para dimensionar a capacidade de uma estocagem de gás natural sob incerteza de demanda e de preço de GNL. O modelo apresentado é um híbrido de otimização estocástica, construído para considerar a incerteza do consumo de gás, com otimização robusta, construído para levar em conta a incerteza relacionada aos preços do GNL. O modelo caracteriza o perfil de risco do supridor de gás natural pela utilização do Conditional Value-at-Risk (CVaR) e utiliza um critério de segurança que reproduz um processo de suprimento avesso a risco de déficit. Ao final do trabalho é apresentado um estudo de caso hipotético, utilizando dados públicos do setor elétrico e de gás natural, para avaliar a implantação da estocagem para 2.000 cenários de demanda e patamares distintos de robustez à variação do preço do GNL. / [en] In Brazil, natural gas demand has stochastic behavior since gas-fired power plants operate in conjunction with the hydroelectric system. Natural gas supply to these plants relies upon Liquefied Natural Gas (LNG), imported through cryogenic ships. LNG acquisitions must occur before the natural gas demand is known because of the time of displacement of the ships. This lack of synchronism stimulates the use of harmonizing mechanisms between the electric sector and the natural gas sector. In this context, natural gas storage could be used to introduce flexibility into the system and increase synergy between natural gas supply and demand dynamics. However, the economic performance of the storage will depend on actual gas prices and demand behavior during the period of analysis. This study aims to construct a linear programming model to determine the size of a natural gas storage under demand and LNG price uncertainty. The model is a hybrid of a stochastic optimization algorithm – developed to consider gas demand uncertainty – and a robust optimization algorithm – built to take into account LNG price uncertainty. A convex combination between Conditional Value-at-Risk (CVaR) and expected value is also used to indicate the supplier risk profile as well as a security criterion, introduced to represent a deficit-averse supply process. At the end, a hypothetic case is presented to evaluate the implementation of a natural gas storage. The case presented uses public data from the Brazilian electric and gas natural sectors and considers 2.000 demand scenarios and various levels of robustness to LNG price variation.

[pt] Nos últimos anos, observa-se crescente penetração da energia eólica na matriz energética mundial e brasileira. Em 2015, ela já representava (seis por cento) da capacidade total de geração de energia do país, colocando-o na (décima) posição entre os países com capacidade eólica instalada. A crescente penetração dessa fonte de energia e suas características de intermitência e forte sazonalidade, passaram a demandar modelos de otimização capazes de auxiliar tanto a gestão dos sistemas elétricos com geração intermitente de energia eólica, quanto a comercialização dessa energia. Avançaram, assim, os estudos de previsões de médias a cada (dez) minutos, horárias e diárias de geração eólica, para atender a sua inserção na programação dos sistemas elétricos e a sua comercialização em mercados diários e horários. Contudo, poucos estudos deram atenção à previsão e simulação de médias mensais de geração eólica, imprescindíveis para gestão e otimização da comercialização dessa energia no Brasil, visto que esta ocorre essencialmente em base mensal. Neste contexto, insere-se esta tese, que busca avaliar a otimização comercial de um parque eólico no mercado livre de energia brasileiro, considerando diferentes modelos de simulação da incerteza de geração eólica e níveis de aversão ao risco do gestor. Para representar diferentes níveis de aversão ao risco do gestor, desenvolveu-se uma nova função de preferência, capaz de modelar a variação do nível de aversão ao risco de um mesmo gestor, para diferentes faixas de preferência, definidas a partir de percentis αs de VaRα. A função de preferência desenvolvida é uma ponderação entre o valor esperado e níveis de CVaR dos resultados. De certo modo, ela altera as probabilidades dos resultados, de acordo as preferências do gestor, similar ao efeito dos pesos de decisão na Teoria do Prospecto. Para simulação da geração eólica são adotados modelos autorregressivos com sazonalidade representada por dummies mensais (ARX-11) e periódicos (PAR). Considera-se, ainda, a inclusão de variáveis climáticas exógenas no modelo ARX-11, com ganho de capacidade preditiva. Observou-se que, para um gestor neutro ao risco, as diferentes simulações de geração eólica não alteraram a decisão ótima. O mesmo não é válido para um gestor avesso ao risco, especialmente ao ser considerado o modelo de simulação com variáveis climáticas exógenas. Portanto, é importante a definição de um único modelo de simulação a ser considerado pelo gestor avesso ao risco ou, a adoção de alguma técnica multicritério para ponderação de diferentes modelos. O perfil de risco também altera as decisões ótimas do gestor, observando-se redução do desvio-padrão e da média da distribuição dos resultados e, aumento dos CVaRs e prêmio de risco, à medida que aumenta a aversão ao risco. Assim, é importante a especificação de uma única função de preferência, que represente adequadamente o perfil de risco do gestor ou da empresa, para otimização da comercialização. A flexibilidade da função de preferência desenvolvida, ao permitir a definição de diferentes níveis de aversão ao risco do gestor, para diferentes faixas de preferência, contribui para essa especificação. / [en] In recent years, we have seen an increased penetration of wind power in the Brazilian energy matrix and also worldwide. In 2015, wind power already accounted for (six percent) of the Brazilian total power capacity and the country was the (tenth) in the world raking of wind power installed capacity. Due to the growing penetration of the source, its intermittency and strong seasonality, optimization models able to deal with the management of wind power, both in electrical systems operation and in trading environment, are necessary. Thus, we see the growth in the number of studies concerned about wind power forecasts for every (10) minutes, hours and days, meeting the electrical systems and international trading schedules. However, few studies have given attention to the forecasting and simulation of wind power monthly averages, which are essential for the management and optimization of energy trading in Brazil, since its occurs essentially on a monthly basis. In this context, we introduce this thesis, which seeks to assess the commercial optimization of a wind farm in the Brazilian energy free market, considering different simulation models for the wind power production uncertainty and different levels of manager s risk aversion. In order to represent the manager s different levels of risk aversion, we developed a new preference function, which is able to model the variation of risk aversion level of the same manager, for different preference groups. These groups are defined by α s percentiles of VaRα. The developed preference function is a weighted average between expected value of results and CVaR levels. In a way, it changes the odds of the results, according to the manager s preference, similar to the effect of the decision weights on Prospect Theory. We adopted autoregressive models to simulate wind power generation, with seasonality represented by monthly dummies (ARX -11) or periodic model (PAR). Furthermore, we consider the inclusion of climate exogenous variables in the ARX-11 model and obtain predictive gain. We observed that for a risk neutral manager, different simulations of wind power production do not change the optimal decision. However, this does not apply for risk averse managers, especially when we consider the simulation model with climate exogenous variables. Therefore, it is important that the risk averse manager establishes a single simulation model to consider or adopts some multi-criteria technique for weighting different models. The risk profile also changes the manager optimal decision. We observed that increasing risk aversion, the standard deviation and mean of the results distribution decrease, while risk premium and CVaRs increase. Therefore, to proceed the optimization, it is important to specify a single preference function, which represents adequately the manager or company risk profile. The flexibility of the developed preference function, allowing the definition of different manager s risk aversion levels for different preference groups, contributes to this specification.

Foreign aid is a major flow of income into sub-Saharan African (SSA) countries, averaging roughly 12% of GDP over the last four decades. Yet, SSA countries are characterized by very low per capita output, low human capital attainment, and widespread poverty. This dissertation investigates the macroeconomic and welfare effects of foreign aid to SSA countries. The empirical part of the dissertation studies 22 SSA countries, and uses a cointegrated vector autoregressive analysis (CVAR). This methodology identifies long-run effects without imposing strong statistical priors. I introduce tradable and non-tradable sectors into the analysis to determine if the so-called “Dutch Disease” is the reason for the plight of SSA countries. “Dutch Disease” occurs when a positive shock to foreign aid perversely reduces GDP, by decreasing the relative price of tradable to nontradable goods, thus reducing the size of the tradable sector. While I find that aid reduces GDP in eight countries, this result is inconsistent with the “Dutch Disease” as it is not accompanied by large relative price changes. The analysis controls for a number of country-specific characteristics including extraordinary events. Overall, I find non-positive impacts of foreign aid on GDP and the tradable sector, with a few exceptions. I also consider the reverse causal channel and test whether country-specific macroeconomic variables drive foreign aid flows. I find that GDP, tradable output, and tradable and non-tradable goods prices do affect the amount of aid a country receives in 15 countries. These variables have no impact on foreign aid (aid is considered as weakly exogenous) in six countries. The theoretical part of the dissertation develops two dynamic stochastic general equilibrium — real business cycle — (DSGE-RBC) models to analyze the effects of foreign aid on human capital investment and the business cycle. The distinguishing feature of the models is to embed a human capital investment in a small open economy model of Mendoza (1991). The first model considers one-sector DSGE model, which is followed by two-sector (tradable and non-tradable) DSGE model. Both models distinguish between physical and human capital investment and allow for labor-leisure choice. In the analysis, labor supply and time spent studying or acquiring skills are optimally chosen. The models are calibrated to match the key features of the Kenyan economy. In both models, a positive aid shock initially has a negative impact on labor supply and output. However, the shock subsequently has a positive effect on physical and human capital investment, and time spent studying. This is due to a positive income effect from the shock. A rise in foreign aid increases consumption; consumption smoothing across periods raises physical and human capital investment, labor productivity, and output. I also find that reducing the volatility of aid has a significant positive effect on human capital investment and welfare. Policymakers should focus on reducing the volatility of foreign aid and not solely concentrate on the average level of aid. The analysis of the two-sector DSGE-RBC model incorporates the role for the “Dutch Disease” mechanism. Consistent with the “Dutch Disease”, I find that a shock to foreign aid appreciates the relative price of non-tradable goods that causes the factors of production to reallocate from the tradable sector to the non-tradable sector, leading to a decline in GDP and the tradable output. Finding the “Dutch Disease” result here is not necessarily at odds with the CVAR estimation results as the DSGE-RBC simulation is a short-run analysis and the CVAR estimation is a long-run analysis. / Graduate

Value at Risk (VaR) and Expected Shortfall (ES) are methods often used to measure market risk. Inaccurate and unreliable Value at Risk and Expected Shortfall models can lead to underestimation of the market risk that a firm or financial institution is exposed to, and therefore may jeopardize the well-being or survival of the firm or financial institution during adverse markets. The objective of this study is therefore to examine various Value at Risk and Expected Shortfall models, including fatter tail models, in order to analyze the accuracy and reliability of these models. Thirteen VaR and ES models under three main approaches (Parametric, Non-Parametric and Semi-Parametric) are examined in this study. The results of this study show that the proposed model (ARMA(1,1)-GJR-GARCH(1,1)-SGED) gives the most balanced Value at Risk results. The semi-parametric model (Extreme Value Theory, EVT) is the most accurate Value at Risk model in this study for S&P 500. / October 2014

Made available in DSpace on 2010-04-20T21:00:12Z (GMT). No. of bitstreams: 4 Lucas Machado Braga de Araujo.pdf.jpg: 18740 bytes, checksum: e39415b36953ced37a6d87a19c381b38 (MD5) Lucas Machado Braga de Araujo.pdf.txt: 95713 bytes, checksum: a95ad7b44ec22b2f5ba2dbfe4f6798bb (MD5) Lucas Machado Braga de Araujo.pdf: 758972 bytes, checksum: c20d926964cc37d2ee7bdfc1bdf6aafc (MD5) license.txt: 4886 bytes, checksum: f5ef202e66d94f51fd6c9a09505731a6 (MD5) Previous issue date: 2009-02-03T00:00:00Z / The aim of this work is to show that the optimization of a portfolio composed of Brazilian hedge funds presents better results when the risk measure considered is Conditional Value-at-Risk. Portfolio optimization models aim to select assets that maximize the investor‘s return for a given level of risk. Therefore the definition of an appropriate measure of risk is of fundamental importance to the allocation process. The traditional methodology of portfolio optimization, developed by Markowitz, uses the variance of assets returns as risk measure. However variance is a measure appropriate only for cases where the returns are normally distributed or that the investor utility function is quadratic. Nevertheless it will be shown that the returns of Brazilian hedge funds usually do not have a Normal distribution. Consequently, to perform the optimization of a portfolio composed by Brazilian hedge funds is necessary to use an alternative risk measure. / O objetivo do trabalho é demonstrar que a otimização de uma carteira composta por fundos multimercados brasileiros gera melhores resultados quando a medida de risco utilizada é o Conditional Value-at-Risk. Modelos de otimização de carteira têm como objetivo selecionar ativos que maximizem o retorno do investidor para um determinado nível de risco. Assim, a definição de uma medida apropriada de risco é de fundamental importância para o processo de alocação. A metodologia tradicional de otimização de carteiras, desenvolvida por Markowitz, utiliza como medida de risco a variância dos retornos. Entretanto, a variância é uma medida apenas apropriada para casos em que os retornos são normalmente distribuídos ou em que os investidores possuem funções de utilidade quadrática. Porém, o trabalho mostra que os retornos dos fundos multimercados brasileiros tendem a não apresentar distribuição normal. Logo, para efetuar a otimização de uma carteira composta por fundos multimercados brasileiros é necessário utilizar uma medida de risco alternativa.

The financial crisis of 2007-2008 highlighted the societal impacts of bank risk-taking. A strong focus on maximizing profits for shareholders combined with a disregard for, and  underestimation of risks led to the downfall of large banks such as Lehman Brothers and multiple other banks getting bailed out by several governments and other banks. The financial crisis spread and impacted all major financial markets across the globe, which highlights the importance of investigating the banking sector from a global perspective. In addition, the influences of corporate social responsibility (CSR) on financial performance and risk have been a growing topic in research as well as in practice. Most banks today invest large amounts of money in CSR activities. The question of how bank spending in CSR activities impacts market risk is important. There are two contradicting views on CSR activities and market risk. The risk mitigation view suggests that banks that focus on stakeholder satisfaction have lower risk due to increased moral capital with stakeholders. The overinvestment view suggests that managers waste scarce resources by overinvesting in CSR activities to further selfish goals and therefore increase risk. This study examines the relationship between sustainability performance (ESG Combined score) and market risk (VaR/CVaR) using a deductive approach. The authors sample 159 banks from 39 countries and all 7 economic regions from 2011-2022. Data is used for testing hypotheses. Results find high ESG Combined Scores are associated with lower VaR/CVaR and results are robust to modifications in VaR/CVaR calculation assumptions. Disaggregation of ESG pillars shows that social pillar scores decrease VaR/CVaR in banks while environmental and governance pillars are insignificant. Results primarily lend support to the risk mitigation view and stakeholder theory stating that firms should focus more on stakeholder satisfaction than maximizing shareholder value. Complementing theories such as legitimacy theory and resources-based view are also considered important theories for explaining the results.

L'approche traditionnelle des produits dérivés consiste, sous certaines hypothèses bien définies, à construire des stratégies de couverture à risque strictement nul. Cependant,dans le cas général ces stratégies de couverture "parfaites" n'existent pas,et la théorie doit plutôt s'appuyer sur une idée de minimisation du risque. Dans ce cas, la couverture optimale dépend de la quantité du risque à minimiser. Dans lecadre des options, on considère dans ce travail une nouvelle mesure du risque vial'approche de l'utilité espérée qui tient compte, à la fois, du moment d'ordre quatre,qui est plus sensible aux grandes fluctuations que la variance, et de l'aversion aurisque de l'émetteur d'une option vis-à-vis au risque. Comparée à la couverture endelta, à l'optimisation de la variance et l'optimisation du moment d'ordre quatre,la stratégie de couverture, via l'approche de l'utilité espérée, permet de diminuer lasensibilité de la couverture par rapport au cours du sous-jacent. Ceci est de natureà réduire les coûts des transactions associées / The traditional approach of derivatives involves, under certain clearly defined hypothesis, to construct hedging strategies for strictly zero risk. However, in the general case these perfect hedging strategies do not exist, and the theory must be rather based on the idea of risk minimization. In this case, the optimal hedging strategy depends on the amount of risk to be minimized. Under the options approach, we consider here a new measure of risk via the expected utility approach that takes into account both, the moment of order four, which is more sensitive to fluctuations than large variance, and risk aversion of the investor of an option towards risk. Compared to delta hedging, optimization of the variance and maximizing the moment of order four, the hedging strategy, via the expected utilitiy approach, reduces the sensitivy of the hedging approach reported in the underlying asset price. This is likely to reduce the associated transaction costs.

The research presented in this thesis addresses different aspects of dynamic portfolio construction and portfolio risk measurement. It brings the research on dynamic portfolio optimization, replicating portfolio construction, dynamic portfolio risk measurement and volatility forecast together. The overall aim of this research is threefold. First, it is aimed to examine the portfolio construction and risk measurement performance of a broad set of volatility forecast and portfolio optimization model. Second, in an effort to improve their forecast accuracy and portfolio construction performance, it is aimed to propose new models or new formulations to the available models. Third, in order to enhance the replication performance of hedge fund returns, it is aimed to introduce a replication approach that has the potential to be used in numerous applications, in investment management. In order to achieve these aims, Chapter 2 addresses risk measurement in dynamic portfolio construction. In this chapter, further evidence on the use of multivariate conditional volatility models in hedge fund risk measurement and portfolio allocation is provided by using monthly returns of hedge fund strategy indices for the period 1990 to 2009. Building on Giamouridis and Vrontos (2007), a broad set of multivariate GARCH models, as well as, the simpler exponentially weighted moving average (EWMA) estimator of RiskMetrics (1996) are considered. It is found that, while multivariate GARCH models provide some improvements in portfolio performance over static models, they are generally dominated by the EWMA model. In particular, in addition to providing a better risk-adjusted performance, the EWMA model leads to dynamic allocation strategies that have a substantially lower turnover and could therefore be expected to involve lower transaction costs. Moreover, it is shown that these results are robust across the low - volatility and high-volatility sub-periods. Chapter 3 addresses optimization in dynamic portfolio construction. In this chapter, the advantages of introducing alternative optimization frameworks over the mean-variance framework in constructing hedge fund portfolios for a fund of funds. Using monthly return data of hedge fund strategy indices for the period 1990 to 2011, the standard mean-variance approach is compared with approaches based on CVaR, CDaR and Omega, for both conservative and aggressive hedge fund investors. In order to estimate portfolio CVaR, CDaR and Omega, a semi-parametric approach is proposed, in which first the marginal density of each hedge fund index is modelled using extreme value theory and the joint density of hedge fund index returns is constructed using a copula-based approach. Then hedge fund returns from this joint density are simulated in order to compute CVaR, CDaR and Omega. The semi-parametric approach is compared with the standard, non-parametric approach, in which the quantiles of the marginal density of portfolio returns are estimated empirically and used to compute CVaR, CDaR and Omega. Two main findings are reported. The first is that CVaR-, CDaR- and Omega-based optimization offers a significant improvement in terms of risk-adjusted portfolio performance over mean-variance optimization. The second is that, for all three risk measures, semi-parametric estimation of the optimal portfolio offers a very significant improvement over non-parametric estimation. The results are robust to as the choice of target return and the estimation period. Chapter 4 searches for improvements in portfolio risk measurement by addressing volatility forecast. In this chapter, two new univariate Markov regime switching models based on intraday range are introduced. A regime switching conditional volatility model is combined with a robust measure of volatility based on intraday range, in a framework for volatility forecasting. This chapter proposes a one-factor and a two-factor model that combine useful properties of range, regime switching, nonlinear filtration, and GARCH frameworks. Any incremental improvement in the performance of volatility forecasting is searched for by employing regime switching in a conditional volatility setting with enhanced information content on true volatility. Weekly S&P500 index data for 1982-2010 is used. Models are evaluated by using a number of volatility proxies, which approximate true integrated volatility. Forecast performance of the proposed models is compared to renowned return-based and range-based models, namely EWMA of Riskmetrics, hybrid EWMA of Harris and Yilmaz (2009), GARCH of Bollerslev (1988), CARR of Chou (2005), FIGARCH of Baillie et al. (1996) and MRSGARCH of Klaassen (2002). It is found that the proposed models produce more accurate out of sample forecasts, contain more information about true volatility and exhibit similar or better performance when used for value at risk comparison. Chapter 5 searches for improvements in risk measurement for a better dynamic portfolio construction. This chapter proposes multivariate versions of one and two factor MRSACR models introduced in the fourth chapter. In these models, useful properties of regime switching models, nonlinear filtration and range-based estimator are combined with a multivariate setting, based on static and dynamic correlation estimates. In comparing the out-of-sample forecast performance of these models, eminent return and range-based volatility models are employed as benchmark models. A hedge fund portfolio construction is conducted in order to investigate the out-of-sample portfolio performance of the proposed models. Also, the out-of-sample performance of each model is tested by using a number of statistical tests. In particular, a broad range of statistical tests and loss functions are utilized in evaluating the forecast performance of the variance covariance matrix of each portfolio. It is found that, in terms statistical test results, proposed models offer significant improvements in forecasting true volatility process, and, in terms of risk and return criteria employed, proposed models perform better than benchmark models. Proposed models construct hedge fund portfolios with higher risk-adjusted returns, lower tail risks, offer superior risk-return tradeoffs and better active management ratios. However, in most cases these improvements come at the expense of higher portfolio turnover and rebalancing expenses. Chapter 6 addresses the dynamic portfolio construction for a better hedge fund return replication and proposes a new approach. In this chapter, a method for hedge fund replication is proposed that uses a factor-based model supplemented with a series of risk and return constraints that implicitly target all the moments of the hedge fund return distribution. The approach is used to replicate the monthly returns of ten broad hedge fund strategy indices, using long-only positions in ten equity, bond, foreign exchange, and commodity indices, all of which can be traded using liquid, investible instruments such as futures, options and exchange traded funds. In out-of-sample tests, proposed approach provides an improvement over the pure factor-based model, offering a closer match to both the return performance and risk characteristics of the hedge fund strategy indices.

[pt] A seleção de variáveis em modelos estatísticos é um problema importante, para o qual diferentes soluções foram propostas. Tradicionalmente, pode-se escolher o conjunto de variáveis explicativas usando critérios de informação ou informação à priori, mas o número total de modelos a serem estimados cresce exponencialmente a medida que o número de variáveis candidatas aumenta. Um problema adicional é a presença de mais variáveis candidatas que observações. Nesta tese nós estudamos diversos aspectos do problema de seleção de variáveis. No Capítulo 2, comparamos duas metodologias para regressão linear: Autometrics, que é uma abordagem geral para específico (GETS) baseada em testes estatísticos, e LASSO, um método de regularização. Diferentes cenários foram contemplados para a comparação no experimento de simulação, variando o tamanho da amostra, o número de variáveis relevantes e o número de variáveis candidatas. Em uma aplicação a dados reais, os métodos foram comparados para a previsão do PIB dos EUA. No Capítulo 3, introduzimos uma metodologia para seleção de variáveis em modelos regressivos e autoregressivos de transição suave (STR e STAR) baseada na regularização do LASSO. Apresentamos uma abordagem direta e uma escalonada (stepwise). Ambos os métodos foram testados com exercícios de simulação exaustivos e uma aplicação a dados genéticos. Finalmente, no Capítulo 4, propomos um critério de mínimos quadrados penalizado baseado na penalidade l1 do LASSO e no CVaR (Conditional Value at Risk) dos erros da regressão out-of-sample. Este é um problema de otimização quadrática resolvido pelo método de pontos interiores. Em um estudo de simulação usando modelos de regressão linear, mostra-se que o método proposto apresenta performance superior a do LASSO quando os dados são contaminados por outliers, mostrando ser um método robusto de estimação e seleção de variáveis. / [en] Variable selection in statistical models is an important problem, for which many different solutions have been proposed. Traditionally, one can choose the set of explanatory variables using information criteria or prior information, but the total number of models to evaluate increases exponentially as the number of candidate variables increases. One additional problem is the presence of more candidate variables than observations. In this thesis we study several aspects of the variable selection problem. First, we compare two procedures for linear regression: Autometrics, which is a general-to-specific (GETS) approach based on statistical tests, and LASSO, a shrinkage method. Different scenarios were contemplated for the comparison in a simulation experiment, varying the sample size, the number of relevant variables and the number of candidate variables. In a real data application, we compare the methods for GDP forecasting. In a second part, we introduce a variable selection methodology for smooth transition regressive (STR) and autoregressive (STAR) models based on LASSO regularization. We present a direct and a stepwise approach. Both methods are tested with extensive simulation exercises and an application to genetic data. Finally, we introduce a penalized least square criterion based on the LASSO l1- penalty and the CVaR (Conditional Value at Risk) of the out-of-sample regression errors. This is a quadratic optimization problem solved by interior point methods. In a simulation study in a linear regression framework, we show that the proposed method outperforms the LASSO when the data is contaminated by outliers, showing to be a robust method of estimation and variable selection.