The assumption of habit formation in preferences induces two effects o
n time series of agents' marginal utility of consumption: greater vola
tility relative to standard time-separable preferences and negative se
rial correlation. This paper examines whether the second property can
help explain the behavior of the nominal term premium. A cash-in-advan
ce model of interest rates is appended with a model cf habit persisten
ce and calibrated to U.S. data. Using yields on three- and six-month U
.S. Treasury Bills for comparison, we find the model can indeed duplic
ate the observed average term premium, but cannot account for the term
premium's volatility.