Does curvature enhance forecasting?

Resumo

In this paper, we analyze the importance of curvature term structure movements on forecasts of interest rates. An extension of the exponential three-factor Diebold and Li (2006) model is proposed, where a fourth factor captures a second type of curvature. The new factor increases model ability to generate volatility and to capture nonlinearities in the yield curve, leading to a significant improvement of forecasting ability. The model is tested against the original Diebold and Li model and some other benchmarks. Based on a forecasting experiment with Brazilian fixed income data, it obtains significantly lower bias and root mean square errors for most examined maturities, and under three different forecasting horizons. Robustness tests based on two sub-sample analyses partially confirm the favorable results

Ficha do documento

Tipo

Artigo científico

Ano

2008

Instituição

Banco Central do Brasil

Fonte

Repositório da FGV

Idioma

Português

Acesso

Não informado

Identificador

oai:repositorio.fgv.br:10438/24227

Temas

Economia

Palavras-chave

Parametric term structure modelsPrincipal componentsVector autoregressive modelsInterest rate mean forecastingEconomiaRisco (Economia)Taxas de jurosTaxas de juros futuras