Predictive Systems, Real Economy, and Bond Risk Premia
43 Pages Posted: 23 Sep 2019 Last revised: 18 Dec 2019
Date Written: December 31, 2018
This paper investigates bond risk premia in the framework of predictive systems. Different from the traditional linear predictive models, predictive systems allow predictors to be imperfectly correlated with conditional expected returns, and could incorporate prior beliefs on the negative correlation between unexpected and expected returns. We find that predictive systems can deliver stronger evidence of predictability than linear predictive models. Furthermore, bond risk premia inferred by predictive systems are countercyclical and increase with inflation risk, and this is consistent with what consumption-based asset pricing models imply.
Keywords: Bond Risk Premia, Imperfect Predictors, Predictive Systems, Real Economy
JEL Classification: C11, G11, G12, G17
Suggested Citation: Suggested Citation