Volatilidade no mercado de ações ajuda a prever a atividade econômica? Uma comparação entre processos lineares e não lineares para Brasil e Estados Unidos

• Ano de defesa: 2021
• Autor(a) principal: Kajiya, Daniel Yuzo Shimada
• Orientador(a): Marçal, Emerson Fernandes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Macroeconomia aplicada; Produção industrial; Autometrics; Volatilidade realizada
• Palavras-chave em Inglês: Industrial production; Realized volatility; Diebold-Mariano; Autometrics; Impulse indicator saturation
• Link de acesso: https://hdl.handle.net/10438/31326

The dissertation aims to explore the stock market volatility as a potential source of relevant information when modelling and forecasting the industrial production series for Brazil and the United States. Furthermore, the study also tries to investigate whether the introduction of an approximation of a non linear estimation processes (STAR) is able to generate more robust results compared to traditional linear models such as the AR. For the brazilian economy, the analysis is focused in 3 recent episodes where the stock market experienced an unexpected surge in realized volatility: Great Financial Crisis of 2008, Brazilian economical crisis of 2014-2015 and the most recent global crisis of 2020 caused by the COVID-19 pandemic. The US economy will be analyzed for the GFC and COVID-19 episodes only since the the remaining one was just a local event particular to the brazilian economy