Previsão do preço e da volatilidade de commodities agrícolas, por meio de modelos ARFIMA-GARCH

• Ano de defesa: 2008
• Autor(a) principal: Bayer, Fabio Mariano
• Orientador(a): Souza, Adriano Mendonça
• Banca de defesa: Silva, Wesley Vieira da , Rabenschlag, Denis Rasquin
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria
• Programa de Pós-Graduação: Programa de Pós-Graduação em Engenharia de Produção
• Departamento: Engenharia de Produção
• País: BR
• Palavras-chave em Português: Séries temporais; Memória longa; Volatilidade; Preço de commodities
• Palavras-chave em Inglês: Time series; Long-memory; Volatility; Price of commodities
• Área do conhecimento CNPq: CNPQ::ENGENHARIAS::ENGENHARIA DE PRODUCAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/8078

This research aims to analyze and predict the prices and volatility of the two major agricultural commodities traded on the market of the Rio Grande do Sul state through ARFIMA-GARCH models. Such models are heteroscedasticity conditional to the volatility, with modeling of integration fraction for the mean conditional. The commodities under study are soy and corn, which represent the two main crops standing of the state of Rio Grande do Sul, in terms of quantity produced in the period, which includes January 1995 to May 2007. The models found to the series of price of soy and corn were ARFIMA (1, d, 0)-GARCH (0, 1) and ARFIMA (1, d, 2)-GARCH (0, 2), respectively. These models are capable of modeling the data satisfactorily, allowing an analysis of their behavior and conduct of forecasts in the short term, signaling possible positions of buying and selling in the market future. Given that the decisions in the context of agribusiness, involving the administration of risk in the purchase and sale in the future market, where risks are related to the volatility of prices, a prediction consistent becomes an important tool in decision-making of the participants of this production process.