Exploiting dependence: Day-ahead volatility forecasting for crude oil and natural gas exchange-traded funds
| Authors | |
|---|---|
| Year of publication | 2018 |
| Type | Article in Periodical |
| Magazine / Source | Energy |
| MU Faculty or unit | |
| Citation | |
| Web | https://www.sciencedirect.com/science/article/pii/S0360544218308193#! |
| Doi | http://dx.doi.org/10.1016/j.energy.2018.04.194 |
| Keywords | Oil; Natural gas; Volatility forecasting; High-frequency data; ETF |
| Attached files | |
| Description | This paper investigates volatility forecasting for crude oil and natural gas. The main objective of our research is to determine whether the heterogeneous autoregressive (HAR) model of Corsi (2009) can be outperformed by harnessing information from a related energy commodity. We find that on average, information from related commodity does not improve volatility forecasts, whether we consider a multivariate model, or various univariate models that include this information. However, superior volatility forecasts are produced by combining forecasts from various models. As a result, information from the related commodity can be still useful, because it allows us to construct wider range of possible models, and averaging across various models improves forecasts. Therefore, for somebody interested in precise volatility forecasts of crude oil or natural gas, we recommend to focus on model averaging instead of just including information from related commodity in a single forecast model. |
| Related projects: |