Financialization, Crisis and Commodity Correlation Dynamics

We study bi-variate conditional volatility and correlation dynamics for individual commod-ity futures and \u85nancial assets from May1990-July 2009 using DSTCC- GARCH (Silvennoinen and Teräsvirta 2009). These models allow correlation to vary smoothly between extreme states via transition functions driven by indicators of market conditions. Expected stock volatility and money manager open interest in futures markets are relevant transition variables. Results point to increasing integration between commodities and \u85nancial markets. Higher commodity returns volatility is predicted by lower interest rates and corporate bond spreads, US dollar depreciations, higher expected stock volatility and \u85nancial traders open positions. We observe higher and more variable correlation, particularly from mid-sample, often predicted by higher expected stock volatility. For many pairings, we observe a structural break in the conditional correlation processes from the late 1990s