Lorenz Schneider (EM Lyon)
Abstract: We propose a stochastic volatility model for crude oil markets that features a regime-switching price of variance-risk.
This tractable model allows us to capture episodes of negative and positive variance risk premium.
A two-state version of the model is estimated using the Hamilton-Lam-Kim filter on CBOE OVX volatility data.
The model characterizes two states: a normal state with low volatility and negative variance premium, and a crisis state with high volatility and positive variance risk premium.
The estimated states are consistent with GDP data and anecdotal evidence.
Incorporating regime information improves the performance of CAPM regressions.
Finally, we study the performance of variance trading strategies that use the estimated state as a trading signal.dden