Volatility transmission and volatility impulse response functions in crude oil markets

Using daily data from July 2005 to February 2011 for WTI, Dubai and Brent futures contracts, we employ a VAR-BEKK model to investigate crude oil markets integration on the second moment. We also quantify the size and persistence of these connections through the analysis of Volatility Impulse Response Functions (VIRF) for two historical shocks, namely the 2008 Financial Crisis and the BP Deepwater Horizon oil spill. We observe that Brent and Dubai crude are highly responsive to market shocks, whereas WTI crude shows the least responsiveness of the three benchmarks, which creates questions about its predominance as a benchmark crude oil. Furthermore, we fit the density of the VIRF at different forecast horizons. These fitted distributions are asymmetric, showing that the probability of observing a large impact of a shock is lower while the probability of a relatively smaller impact is much higher. Finally, we simulate the VIRF for a given probability of a random shock. The VIRF shows that only a “large” shock (derived from a smaller probability) will result in an increase in expected conditional volatilities. These results provide useful insights into the volatility transmission mechanism in crude oil markets and their associated risk estimation, and may have significant implications for various market participants and regulators. ► We examined time series volatility models of major oil benchmarks from 2005 to 2011. ► Oil benchmarks are highly responsive to shocks, with WTI being the least responsive. ► Probability distribution of shocks is asymmetric. ► Large shocks of small possibilities may result in increases in expected volatilities. ► VIRF for a given possibility of a random shock can be used as a risk measure.