Construction of Commodity Portfolio and Its Hedge Effectiveness Gauging – Revisiting DCC Models

This paper examines how various types of dynamic conditional correlation (DCC) models performs in the construction of risk-minimizing portfolio. Our portfolios consist of SPY-ETF instrument as a primary asset and four commodities – Brent oil, gold, silver and platinum. In the process of hedge effectiveness measurement, we utilize three different performance metrics – Hedge Effectiveness Index in terms of variance, Value at Risk and Conditional Value at Risk, which target different risk minimizing goals. The additional objective is to test whether minimum-variance hedging portfolio yields a similarly large reduction in portfolio VaR and portfolio CVaR. The hedge effectiveness performances are scrutinized via portfolios that are designed with help of three different types of DCC models – DCC-GARCH, DCC-APARCH and DCC-FIAPARCH. In order to gauge how hedge effectiveness of the portfolios alters across periods of different market turbulences, we split full sample into three subsamples applying modified ICSS algorithm. The research determines that minimum-variance targeting portfolios with accounted long memory in volatility demonstrated considerable robustness when it comes to the best performing models, taking into account three different risk metrics. However, in cases of risk/return targeting goals as well as in the process of out-of-sample forecast, the best solution turns out to be the simplest DCC model.