Predicting bond returns using the output gap in expansions and recessions

In this paper new evidence will be provided about the consensus that the output gap can predict excess bond returns. Also I investigate how the predictability of the output gap changes across economic expansions and recessions. Because of the persistence of the variables and the serial correlation of the error term, I find no evidence for the statistical significance of the output gap, even when the regression is conditioned on the business cycle. In general, the business cycle gives different coefficients for the predictors with a different statistical significance. Because of the lack of observations in recessions, it is hard to make conclusions about the significance of the business cycle on the predictability of excess bond returns.