Identifying the Sources of the Predictive Ability of the Yield Spread

Abstract

This paper aims at identifying the sources of the predictive power of the nominal yield spread for economic activity. For this purpose, we analyse the relationship between the yield spread and future GDP growth in an economy subject to three types of structural disturbances: Supply shocks, monetary shocks and shocks to term premia. The theoretical model extends the work of Harvey (1988) and Hu (1993), which attributes the correlation between the real term structure and future output growth to intertemporal consumption smoothing, by allowing short-term price stickiness in the economy. Sticky prices imply that economic disturbances generate expectations of future changes in output and inflation, thus allowing for intertemporal substitution effects and changes in both the real and the nominal term structure. We derive analytical solutions of the covariance between the nominal yield spread and future output growth and show that this covariance is not independent of the type of economic disturbances. Using data for the U.S. over the period 1957:1 - 2002:4, we find empirical evidence that the predictive power of the yield spread is due to supply shocks. In contrast, monetary shocks and term premia shocks generate a negative correlation between the yield spread and future economic activity, explaining why, at times, the predictive power of the yield spread may break down.