Modeling and predicting oil VIX: Internet search volume versus traditional macro-finance variables

Abstract

As a key variable in option pricing models and monetary policy decisions, volatility is an important factor in valuing and hedging investments. This paper models and predicts the CBOE Crude Oil Volatility Index using Heterogeneous Autoregressive (HAR) models that include traditional macro-finance variables as well as abnormal search volume from Google (ASVI). We find that a pure HAR model fits oil volatility remarkably well. When adding ASVI, we discover that this variable has a significant and positive relationship with oil volatility. This relationship remains statistically significant when traditional financial and macroeconomic variables are accounted for; therefore, ASVI is not only a good proxy for traditional macro-finance variables, but also carries additional information. More importantly, out-of-sample predictions show that ASVI has high economic value, allowing traders of volatility-exposed portfolios to significantly increase returns.