Preprint,
Bond Risk Premiums at the Zero Lower Bound
Affiliations
- [1] Aarhus University [NORA names: AU Aarhus University; University; Denmark; Europe, EU; Nordic; OECD];
- [2] European Central Bank [NORA names: Germany; Europe, EU; OECD];
- [3] Federal Reserve Board of Governors [NORA names: United States; America, North; OECD]
Abstract
We document that the spread between long- and short-term government bond yields is a stronger predictor of excess bond returns when the U.S. economy is at the zero lower bound (ZLB) than away from this bound. The Gaussian shadow rate model with a linear or quadratic shadow rate is unable to explain this change in return predictability. The same holds for the quadratic term structure model and the autoregressive gamma-zero model that also enforce the ZLB. In contrast, the linear-rational square-root model explains our new empirical finding because the model allows for unspanned stochastic volatility as seen in bond yields.
Keywords
Lower,
U.S.,
U.S. economy,
ZLB,
Zero Lower,
bond returns,
bond risk premium,
bond yields,
bonds,
economy,
empirical findings,
excess bond returns,
findings,
government bond yields,
longer,
model,
prediction,
predictors,
premium,
quadratic term structure models,
rate,
rate model,
return,
return predictability,
risk premium,
shadow rate,
shadow rate model,
spread,
square-root model,
stochastic volatility,
structural model,
term structure model,
unspanned stochastic volatility,
volatility,
yield,
zero