Predicting inflation with the term structure spread

It is tempting to interpret empirical evidence in a number of recent studies as suggesting that term structure spreads help predict future inflation over moderate horizons of 3 to 5 years. This paper argues that common measures of the predictive power of the term structure spread for future inflation are misleading. In particular, R2s for estimated inflation-change equations can drastically overstate the predictive power of spreads. The paper explains why the overstatement is likely to be particularly large in countries whose monetary authorities have strong reputations for credibly targeting a stable inflation rate. Results from an empirical analysis of data from eleven industrialized countries suggest that the level of the short-term real rate may be more useful for predicting inflation than the term structure spread, possibly because changes in short-term real rates provide clearer measures of changes in the stance of monetary policy.