Volatility Forecasting of Strategically Linked Commodity ETFs : Gold - Silver

We apply heterogenous autoregressive (HAR) models – including nine univariate, two multivariate and three combination models – to high-frequency data to predict the one-day forward volatilities of two strategically linked commodities, gold and silver. We provide evidence that it is difficult to beat the benchmark model of Corsi (2009) using univariate models and that a much better strategy is to average the forecasts from many models. In addition, the forecasts are not improved by using volatilities from strategically linked commodities; thus, no volatility spillovers are detected. Interestingly, when the two strategically linked commodities are modeled together using the generalized HAR (GHAR) model of Baruník and Čech (2016), the forecasts are comparable to those of combination forecast models