Estimating and Forecasting a Term Structure of Interest Rates with State-Space Nelson-Siegel Model

Joonhyuk Song (Hankuk University of Foreign Studies)

Journal of Derivatives and Quantitative Studies: 선물연구

ISSN: 1229-988X

Article publication date: 31 August 2011



This paper estimates a Nelson-Siegel model under the state-space representation in order to circumvent the shortcomings of the conventional Nelson-Siegel model and evaluates the predictive ability of the estimated model. The results indicate that the estimated Nelson-Siegel time-varying three factors have close relations to their counterparts : level, slope and curvature and the inflection of the Korean yield curve is located around the maturity of 55-month. Meanwhile, each factor is found to have unit-root but differenced-factors do not show signs of unit-roots, hence proved I (1) series. In order to assess the efficacy of the estimated model, we compare the yield prediction from our model with several natural competitors : random walk, Fama-Bliss, and Cochrane-Piazzesi. With respect to out-of-sample performance, Fama-Bliss model proves to be the worst in term structure forecasts in Korea. The predictive performance differs between the random walk and the state-space Nelson-Siegel model depending on the forecast horizon lengths. At the shorter horizon, the state-space Nelson-Siegel model outperforms the random walk, but the table is turned in the longer horizon



Song, J. (2011), "Estimating and Forecasting a Term Structure of Interest Rates with State-Space Nelson-Siegel Model", Journal of Derivatives and Quantitative Studies: 선물연구, Vol. 19 No. 3, pp. 309-334.



Emerald Publishing Limited

Copyright © 2011 Emerald Publishing Limited


This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at

Related articles