How do current term structure model behave beyond the last liquid point?: A comparison of the DNS and Smith-Wilson methods

A Work Project, presented as part of the requirements for the Award of a Master’s Double Degree in Finance from Maastricht University and NOVA – School of Business and Economics

This paper compares the popular Dynamic Nelson-Siegel (DNS) model with the Smith-Wilson (SW) method for the extrapolation of yield curves within the scope of the new regulation for pension funds and insurance companies, Solvency II. I have focused particularly on the behavior of the models after the last liquid point (LLP) of observable data. My main research shows that a longer LLP is beneficial at extrapolating the yield curve as well as using a onvergence period that relies on the available data. I also found that the DNS model is more market consistent whereas the SW method performs better fitting the available data and disregards the information they provide at the long-end of the curve.

UNL - NSBE