The 27th Conference on the Theories and Practices of Securities and Financial Markets (SFM 2019)

3 / 2019-08-14 14:34:59

A comprehensive look at commodity volatility forecasting

Realized volatility,High frequency data,Samuelson effect,term structure,supply and demand,commodity,derivatives,risk management,financial econometrics

This research combines recent advances in the Realized Volatility (RV) literature and
three specific commodity futures factors to improve the forecasts of commodity volatility. The three forecasting variables are the term structure slope, the time to maturity and a measure of supply and demand uncertainty. I first assess these variables’ empirical contribution to commodity futures volatility, in adding them in RV forecast models. First in the univariate HAR-RV of Corsi (2009) and second in the multivariate VAR-RV of Andersen, Bollerslev, Diebold, and Labys (2003). The long-term memory of assets RV justifies the former, whereas the "financialization" of commodities and the resulting commodity connectedness, supports the latter. I evaluate the out of sample validity of these forecast models and propose one risk management application. Hence, this research is important for both economic understanding and risk management purposes.