The Estimation of Svensson Model Term Structures and Their Volatilities
33 Pages Posted: 8 May 2012
Date Written: May 4, 2012
The Nelson-Siegel and the Svensson models are widely used in practice for fitting the term structure of interest rates. However, due to their highly non-linear nature and the potential danger of multicollinearity, numerical difficulties in estimating these models hamper their implementation. In this paper, a conditional ridge regression based approach is proposed to remedy the reported problems. The method we suggest produces – even in conditions of financial crises - non-negative and bounded theoretical long and short rates. Both its in-sample and out-of-sample performance are not dominated by other procedures. The out-of-sample performance is judged by the MAE between the rates and the volatilities of the EONIA or the 30-year swap rate and their model-extrapolated counterparts. Also the one-day GARCH(1,1) with student-t innovations forecasting ability of the short and long rates is among the best.
Keywords: Term Structure, Nelson-Siegel Model, Svensson Model, Ridge Regression
JEL Classification: E43, C51
Suggested Citation: Suggested Citation