Authors
Abstract
Yield curve dynamics are usually analysed in terms of the unobservable components-level, slope and curvature. The factor-augmented vector autoregression (FAVAR) framework applied in this article has become increasingly popular for forecasting interest rates. To link these two strands of research, the authors operate in the space of latent yield and latent macroeconomic factors to analyse the relationship between the term structure of interest rates and the macroeconomic aggregates. They predict the yield curve dynamics by directly forecasting the unobservable yield curve factors. They use the FAVAR methodology to encompass a data-rich environment and to identify dynamic responses of the yield curve to the macroeconomic variables. The empirical results suggest that parsimonious FAVAR models with a few latent macroeconomic factors and a reduced lag order show superior short-horizon forecast performance over simple VAR systems and univariate autoregressions.
More information