Vantagens e desvantagens do modelo dinâmico de Nelson-Siegel: aplicação ao mercado brasileiro

• Ano de defesa: 2015
• Autor(a) principal: Franciscangelo, João Gabriel Costa
• Orientador(a): Ruilova Terán, Juan Carlos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Modelo dinâmico de Nelson-Siegel; Espaço de estados
• Link de acesso: http://hdl.handle.net/10438/14076

Modeling the term structure of interest rates has high relevance to the financial market, due to the fact of its utilization for pricing bonds and derivatives, being a fundamental component in the economic policies and assisting in the development of trading strategies. The class of models created by Nelson-Siegel(1987), was extended by many authors and currently is largely used by several centel banks around the world. In this work the extension proposed by Diebold and Li (2006) was applied to the brazilian market, the parameters were calibrated using the Kalman Filter and the Kalman Filter Extended, the last method allowing the estimation with dinamism of all the four parameters used in the model. As mentioned by Durbin and Koopman (2012), the equations contained in the Kalman filter and its extended version do not enforce conditions of constant dimensions in the observations vector. Based on this concept, the filters implementation were made allowing its application independent on the number of observations on each time instant, avoiding the need of previous interporlation of data. It helps the model to reflect more faithfully the markets reality and relax the assumed hypotheses to obtain fixed vértices through interpolation. A new propose of adaptation will be tested in the Nelson-Siegel model, where the level parameter will be conditioned to the bond’s maturities happened before or after the next Copom’s meeting. The objective is to compare the prediction quality across the methods, pointing the benefits and drawbacks observed on each one of them.