On the other hand, the predictive power of the spread has been also examined in respect of its
ability to anticipate future inflation. In 1977 Robert Shiller and Jeremy Siegel provide an empirical
assessment of the so-called Gibson paradox. Using British data, Gibson (1923) found a strong
positive correlation between the (log) series of a price index and long term interest rates. Using
spectral techniques, Shiller and Siegel confirm the aforementioned relation.
Mishkin (1990) investigates whether the term structure of interest rates helps to forecast future
inflation. He finds that interest rates on bonds with maturities less than or equal to six months do
not provide any relevant information about the future path of inflation. However, the short end of
the term structure seems to contain substantial information about the term structure of real interest
rates. Conversely, for maturities between nine and twelve months, the slope of the term structure
appears to carry information about future inflation but not about the real term structure. In line with
the prevalent view, Mishkin provides also evidence that an inverted yield curve reflects
expectations of a declining rate of inflation. Estrella and Mishkin (1997) analyse the predictive
content of the spread regarding the future level of both output and inflation. Estimates for U.S.
indicate that at very short horizon the ability of the spread to anticipate future movements in
inflation is either negligible or absent. The predictive accuracy of the spread increases at long
horizons; in particular the slope of the term structure is informative about future inflation over
horizons of three to five years. Estrella and Mishkin have also documented the ability of the spread
to predict future economic growth.
A recent strand of research has focused on the ability of term premium, rather than of the spread, to
predict future movements in economic activity. The term premium is derived from a decomposition
of the spread into an expectations-based factor and a risk premium. Hamilton and Kim (2002) show
that both components are informative for predicting real GDP growth. Their findings suggest that a
decrease in the term premium predicts slower GDP growth. Favero, Kaminska, and Soderstrom
(2005) obtain similar results. Ang, Piazzesi, and Wei (2006) find that the coefficient of the
expectational component is larger in magnitude than the estimated coefficient of the term premium,
which is not statistically significant. Finally, in spite of existing evidence, Rudebusch, Sack, and
Swanson (2007) provide evidence of inverse relation between term premia and business cycle, in
that respect a decline in term premia tends to be a stimulus for economic activity. They rationalize
this atypical result concluding “we only speculate that our empirical findings may reflect a
heterogeneous population in which a decline in the term premium makes financial markets
conditions more accommodative for certain classes of borrowers”.