Modeling Swedish government yields with the Dynamic Nelson Siegel and the Dynamic Nelson Siegel Svensson Model

Abstract: The purpose of this thesis is to model and forecast Swedish government yields by using three classes of the Nelson Siegel Model Family. The three models considered are the Dynamic Nelson Siegel Model, Arbitrage-Free Nelson Siegel Model and Dynamic Nelson Siegel Svensson Model. A brief introduction to interest rate theory is given with emphasis on coupon bonds and yield curves. To introduce the concepts needed for the arbitrage-free model, arbitrage theory is introduced. The modeling framework used in this thesis implements the Kalman Filter, thereby necessitating introduction of State Space modeling and the derivation of the Kalman Filter. The Nelson Siegel model classes under study are introduced and an estimation procedure for each model is detailed. In general, all model parameters are estimated by both cross-sectional and time-series optimization. The method of estimation employed ensures that we have stable and meaningful estimates. Our modeling procedure, shows that indeed the independent three factor Dynamic Nelson Siegel model do represent well Swedish government bonds both in-sample