Can Linear-Rational Term Structure Models Capture Conditional Volatility in the Treasury Yield Market?

Abstract

We show that the class of linear-rational square-root (LRSQ) model is able to match the cross section of yields and the time variability of conditional yield volatility simultaneously. Models in this class are, in this regard, able to break the tension noted for the affine term structure models from matching the conditional first and second moments of yields. Using a panel data set of US Treasury yields and realized yield volatilities, we evaluate the performance of various LRSQ model specifications based on in-sample and out-of-sample exercises and find that the preferred specification relies on three unspanned stochastic volatility factors, which, correlate strongly with the level and slope factor of conditional yield volatility.