Robust Bond Risk Premia

Recent studies appear to have found evidence that information not reflected in the yield curve helps predict interest rates and excess bond returns. These studies reject the Markov property of the yield curve and conclude that there is unspanned or hidden information that should be used in forecasting. We revisit the evidence of these papers using novel econometric techniques that address the difficult problems surrounding inference about predictability of highly persistent series. We reach the opposite conclusion: only the level and the slope of the yield curve are robust predictors of excess bond returns, and there is no robust and convincing evidence for unspanned macro risk. In other words, the Markov property of the yield curve seems alive and well.