Affine term structure models generate sharp predictions about the time series evolution of bond yields. In
special, they tie the quadratic variation of yields to the cross-section of average yields. I derive these conditions in
a flexible jump-diffusion setting and use jump-robust estimators to formally test these restrictions. My approach
is more general than previous techniques because it does not constrain the dynamics of underlying factors under
the physical measure. I also show that two thirds of all unspanned volatility can be captured by a single factor. I
investigate if this factor is related to monetary policy surprises. I find that only forward-guidance-type shocks fuel
unspanned volatility, although such surprises can explain less than 10% of the unspanned volatility factor.
Complete tool for submission, evaluation and publication of proceedings for scientific events.
