Improving the Predictive Power of Spreads for Economic Activity: Decomposition Methods

In this paper, we examine whether and to what extent the predictive power of credit and/or term spreads for real economic activity can be enhanced by using additional information via decomposition. In doing so, we first apply the wavelet analysis to present evidence that the business-cycle component of the credit spread can better predict the probability of a recession than the usual time-domain analysis. In particular, we investigate the predictive power of the credit spread, given the recent empirical findings that it has a useful explanatory power for future economic fluctuations. We suggest that the wavelet decomposition can enhance the predictive power of the credit spread compared to the usual regression model. We also consider a decomposition of the term spread into the expectations effect and the term premium, based on the liquidity premium theory, and discuss evidence that the decomposition might lead to a better prediction for business-cycle fluctuations than the usual term spread.