Supplement to ``A New Perspective on Gaussian DTSMs''


JSZ shows that restrictions that only affect Q-parameters are irrelevant for forecasting the portfolio of yields P. An immediate implication of this observation is that forecasts of P using an arbitrage-free Nelson-Siegel (AFNS) model are equivalent to forecasts based on an unconstrained VAR(1) representation of P. To see this, we show that the AFNS model of Christensen, Diebold, and Rudebusch (2010) is an invariant transformation of a special case of the JSZ normalization (and indeed of the DS normalization) with the additional constraint that λQ = (0, λ, λ) and rQ ∞ = 0. 1 The AFNS(3) model with latent state vector Xt = (X 1 t , X 2 t , X 3 t ) ′ has a feedback matrix KQ 1X of the form

Cite this paper

@inproceedings{Joslin2010SupplementT, title={Supplement to ``A New Perspective on Gaussian DTSMs''}, author={Scott Joslin and Kenneth J. Singleton and Haoxiang Zhu}, year={2010} }