The predictive power of value-at-risk models in commodity futures markets

Applying standard value-at-risk (VaR) models to assets with non-normally distributed returns can lead to an underestimation of the true risk. Commodity futures returns are driven by continuous supply and demand shocks that lead to a distinct pattern of time-varying volatility. As a result of these specific risk characteristics, commodity returns create the ideal environment for testing the accuracy of VaR models. Therefore, this article examines the in- and out-of-sample performance of various VaR approaches for commodity futures investments. Our results suggest that dynamic VaR models such as the CAViaR and the GARCH-type VaR generally outperform traditional VaRs. These models can adequately incorporate the time-varying volatility of commodity returns, and are sensitive to significant changes in the series of commodity returns. This has important implications for the risk management of portfolios involving commodity futures positions. Risk managers willing to familiarize themselves with these complex models are rewarded with a VaR that shows the adequate level of risk even under extreme and rapidly changing market conditions, as well as under calm market periods, during which excessive capital reserves would lead to unnecessary opportunity costs.