In this dissertation, three related issues concerning empirical time series models for energy financial markets and the stock market were investigated. The purpose of this dissertation was to analyze the interdependence of price movements, focusing on the forecasting models for crude oil prices and the hedging models for gasoline prices, and to study the change in the contemporaneous causal relationship between investors' activities and stock price movements in the Korean stock market. In the first essay, the nature of forecasting crude oil prices based on financial data for the oil and oil product market is examined. As crack spread and oil-related Exchange-Traded Funds (ETFs) have enabled more consumers and investors to gain access to the crude oil and petroleum products markets, I investigated whether crack spread and oil ETFs were good predictors of oil prices and attempted to determine whether crack spread or oil ETFs were better at explaining oil price movements. In the second essay, the effectiveness of diverse hedging models for the unleaded gasoline price is examined using futures and ETFs. I calculated the optimal hedge ratios for gasoline futures and gasoline ETF utilizing several advanced econometric models and then compared their hedging performances. In the third essay, the contemporaneous causal relationship between multiple players' activities and stock price movements in the Korean stock market was investigated using the framework of a DAG model. The causal impacts of three players' activities in regard to stock return and stock price volatility are examined, concentrating on foreign investor activities. Within this framework, two Korean stock markets, the KSE and KOSDAQ markets, are analyzed and compared. Recognizing the global financial crisis of 2008, the change in casual relationships was examined in terms of pre- and post-break periods. In conclusion, when a multivariate econometric model is developed for multi-markets and multi-players, it is necessary to consider a number of attributes on data relations, including cointegration, causal relationship, time-varying correlation and variance, and multivariate non-normality. This dissertation employs several econometric models to specify these characteristics. This approach will be useful in further studies of the information transmission mechanism among multi-markets or multi-players.

[pt] O objetivo deste trabalho é avaliar o investimento em uma refinaria de petróleo no Brasil, país importador de derivados de petróleo, através da Teoria de Opções Reais, abordagem de grande valor prático e acadêmico, que realiza uma valoração mais justa da refinaria, ao incluir a modelagem das incertezas e incorporar flexibilidades gerenciais, negligenciadas pela análise tradicional. As incertezas, modeladas através de processos estocásticos, são o câmbio e o crack spread, adaptando-se o crack spread para o refino brasileiro. São avaliadas as opções de postergação, de parada temporária e sua interação. É apresentada a melhor decisão para o investimento remanescente no projeto e como ela seria alterada se o projeto ainda fosse ser iniciado. Para o investimento remanescente, a melhor decisão é postergar o projeto se a opção for perpétua e investir imediatamente se expirar em até 5 anos. Sensibilidades alterando volatilidade e taxa de conveniência apresentam recomendações diferentes, reduzindo a robustez dos resultados. Quando se considera o valor completo do investimento, a postergação é recomendada em todos os cenários analisados. A opção de parada temporária aumenta o valor da refinaria e reduz o valor da opção de postergação. O modelo desenvolvido permitiu aperfeiçoar a análise de investimento da refinaria em questão e pode ser replicado para análise de investimento em outras refinarias pertencentes à mesma empresa ou a outras. Para trabalhos futuros, recomenda-se calcular a opção de troca input–output, aperfeiçoar a modelagem do crack spread e incorporar o preço do gás como incerteza. / [en] The main objective of this study is to provide an investment analysis of a crude oil refinery in Brazil, an oil products importer country, under the real option theory, an approach with great academic and practical use, allowing a fairer refinery valuation by modeling uncertainty and including managerial flexibility, neglected in traditional analysis. The uncertainties considered, described by stochastic process, are the exchange rate and the crack spread, adapting the crack spread to the Brazilian refining. The options to defer, to shut down and their interaction are analyzed. It is shown the best decision considering the remaining investment and how this decision would be changed if the project hadn t been started. Considering the remaining investment, the best decision is to defer the investment if it is an infinitely-lived option and invest immediately if the right to invest expires upon 5 years. The influence of volatility e convenience yield was also taken into account and shows different recommendations, reducing the results robustness. When the full investment is analyzed, postponement is recommended in all analyzed scenarios. The shutdown option increases the refinery value and reduces the option to defer value. The developed model allowed an improvement in the investment analysis of this refinery and can be replicated to other refineries investment analysis, owned by the same company or not. For further works, we propose to include the switch input-output option, improve the crack spread modeling and consider the gas price as uncertainty.