Abstract
We introduce a discrete-time multi-component model with ๐‘€ stochastic volatility (SV) factors to jointly model equity returns and the volatility index term structure (VIXTS). By conveniently inserting stochastic shocks into volatility components, we decouple volatility innovations from return shocks. This creates a rich dependence structure while producing closed-form moment generating functions, which facilitates derivative pricing on the full term structure (TS). While the framework allows for exact filtering of the factors, our empirical analysis demonstrates that a parsimonious specification with ๐‘€ = 2 factors and ๐‘ = 1 component is sufficient to ensure key statistical and economical objectives: a precise estimation of the parameters, a superior in and out-of-sample goodness of fit of the full VIXTS on several measures, and a close mimicking of contango/backwardation and convexity of not only the VIXTS, but also the variance risk premium TS.

This publication has 27 references indexed in Scilit: