The role of investors' fears on crude oil volatility forecasting

Abstract

We study whether investors’ fears can predict oil price volatility. The proxies used for investors’ fears are CBOE Crude Oil Volatility Index (OVX), Google searches for “oil price”, and United States Oil Fund (USO) trading volume. In the in-sample analysis we find that increased OVX, increased Google searches, and increased trading volume predict increased oil price volatility. Additionally, we find bidirectional Granger-causalities between volatility and OVX, volatility and Google searches, and between volatility and trading volume. However, results are very different for the out-of-sample forecasts. We incorporate OVX, Google searches for “oil price”, trading volume, and their combinations into commonly used volatility models, but find that none of the models is improved by these variables or their combinations in terms of out-of-sample forecasting.