Term Premium Variability and Monetary Policy

Abstract

Two traditional explanations for the mean and variability of the term premium are: (i) time-varying risk premia on long bonds, and (ii) segmented markets between long- and short-term bonds. This paper integrates these two approaches into a medium-scale DSGE model. We consider two sources of business cycle variability: shocks to total factor productivity (TFP), and shocks to the marginal efficiency of investment (MEI). The ability of the risk approach to match the first moment of the term premium depends upon the relative importance of these two shocks. If MEI shocks are an important driver of the business cycle, then long bonds are a hedge against the business cycle so that the average term premium is negative. The opposite is the case for the TFP shocks. But for either source of shocks, the risk approach to the term premium predicts a trivial amount of variability in the term premium. In contrast, the segmented markets model can easily match both moments. The market segmentation reflects a real distortion, so that smoothing the term premium is typically welfare-improving. There are two difficulties with such a policy. First, the mean level of the term premium will not properly reflect the segmentation distortion because of the risk adjustment. Second, if the term premium is measured with error, the welfare gain of a term premium peg is naturally reduced. The paper demonstrates that both of these effects are quantitatively modest so that the welfare advantage to a term premium peg survives.