Michael D. Bauer, James Hamilton, 07 July 2017

Several recent empirical papers have challenged the ‘spanning hypothesis’, which holds that the level, slope, and curvature of bond yield curves are sufficient to forecast returns and estimate bond risk premia This column argues that these studies suffer from a previously unrecognised standard error bias. Controlling for this bias, false positives are found to be between six and twelve times more likely, suggesting that the evidence against the spanning hypothesis is substantially less convincing than would appear from the studies.