Análise estatística do modelo de Nelson e Siegel

• Ano de defesa: 2013
• Autor(a) principal: Brocco, Marcelo Bertini
• Orientador(a): Diniz, Márcio Alves
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Estatística - PPGEs
• País: BR
• Palavras-chave em Português: Estatística; Inferência bayesiana; Taxas de juros; Métodos de estimação; Restrições de desigualdade; Estrutura a termo das taxas de juros
• Palavras-chave em Inglês: Nelson and Siegel Model; Bayesian Inference; Inequality Restrictions; Yield Curve; Term Structure of Interest Rates; Parameter Estimation
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::PROBABILIDADE E ESTATISTICA::ESTATISTICA
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/4566

The present paper studies the yield curve, an important tool for financial decisions, due to its fundamental role in the implementation and evaluation of monetary policies by the central banks. It also shows market perspectives in relation to the future development of interest rates, inflation and economical activities. Using an adequate model and a reasoned assessment of its parameters enables us to adjust the curve as far as possible to the real curve and hence obtain most precise and trustful results. These results were acquired by studying a model which was developed in 1987 by Nelson and Siegel and used to draw up the yield curve. Considering the model s limitations, diferent methods were used to attain the estimated parameters, such as Ordinary Least Squares, Maximum Likelihood and Bayesian Inference in the static version. The Nelson-Siegel model is widely used in Brazil and in the rest of the world, due to its economical idea, easy implementation and eficient adjustment into diferent formats that the yield curve is able to deal with. By considering the restrictions of the model, we found estimations for the parameters of the model safer than other and besides, the main point of this work is an estimation form of parameters of time together with others parameters of the model without considering one fixed value for it.