Estimation of the Nelson‐Siegel parsimonious modeling of yield curves using an exponential GARCH process

Abstract

Refers to previous research on the yields of default‐free securities and uses the Nelson‐Siegel model for estimating yield curve as a basis for developing a model which decomposes the risk premium into long‐term risk, two factors influencing the rate of decay (curvature) and a feed back factor. Applies this to 1984‐1993 data for treasury bills to test for predictive validity and shows that the feedback factor (prediction error of the most recent period) improves this by around 10 per cent. Goes on to apply a multivariate exponential GARCH process to the components to produce a prediction model for the term structure of interest rates. Promises further research to refine this estimation and compare it with the expectations hypothesis as a basis for strategy.