Previsão da estrutura a termo brasileira utilizando de fatores macroeconômicos

• Ano de defesa: 2013
• Autor(a) principal: Faria, Adriano Augusto de
• Orientador(a): Almeida, Caio Ibsen Rodrigues de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Estrutura a termo brasileira; FAVAR; Modelo afim de estrutura a termo
• Link de acesso: https://hdl.handle.net/10438/10964

This article studies the prediction of the Brazilian interest rate term structure employing the use of common factors extracted from a vast database of macroeconomic series. The estimation and prediction periods analyzed are between January of 2000 and May of 2012. The rst approach is based on the model proposed by Moench (2008) in which the short term interest rate dynamics is modeled in a FAVAR framework and its term structure is derived through the use of restrictions implicated by no-arbitrage conditions . The choice of this model is justi ed by the results it obtained in the original study. It has presented the best predictive performance for intermediary and long horizons when compared to usual benchmarks. Nonetheless, such results also presented a progressive deterioration when subject to expansion of maturity periods. This suggests a possible failure from the latter to the estimation of the intermediary and long parts of the curve. When implemented to the Brazilian term structure, the model achieved similar results to Moench's study. In an attempt to overcome the previously mentioned deterioration, we propose an alternative modeling approach in which the dynamics of each rate is modeled alongside with the macroeconomic factors, therefore eliminating the restrictions implicated by the no-arbitrage condition .This approach led to fairly superior prediction results and also made possible to con rm the acknowledged inadequacy. Lastly, we have also inserted the macro factors in the dynamic of the factors from Diebold e Li (2006) model. There was also a predictive capacity gain when comparing to the article's dynamics especially to greater prediction horizons.