This paper shows how to estimate a Bayesian VAR with drifting parameters and nonlinear cross-equation restrictions. The restrictions promote parsimony by reducing the dimension of the drifting component in conditional mean parameters. As an application, the paper investigates an anticipated-utility version of the expectations model of the term structure. The estimates suggest that changing beliefs matter for understanding the yield curve and point to an intriguing clue about risk premia. Local approximations to the mean yield spread are highly correlated with the variance of the trend short rate, suggesting a connection between uncertainty about the long-run target of monetary policy and risk premia on long-term bonds.
- Bayesian analysis
- Monte Carlo methods
- Term structure of interest rates
ASJC Scopus subject areas
- Economics and Econometrics