Global yield curve macroeconomic dynamics and interactions: A dynamic Nelson Siegel approach

The Dynamic Nelson Siegel (DNS) model is a latent factor model frequently fitted to the government bond yield curves of individual countries. In this paper, I apply the methodology of Diebold et al. (2008) who extend it to a global setting. Of central interest is the existence of global factors that drive country specific latent factors. I extend the work of Diebold et al. (2008) by including local macroeconomic factors, considering the effects of persistent oversimplification of the DNS and generating one step ahead forecasts. My results reaffirm the existence of global latent factors. In addition, I show that macroeconomic factors have a strong effect only for the UK and Japan. Finally, I propose a model that ameliorates the effects of oversimplification and improves forecast performance.