Time-varying Granger causality tests for applications in global crude oil markets

This paper proposes time-varying Granger causality tests based on the tests developed by Hong (2001) and two dynamic correlation estimators (i.e., rolling correlation and dynamic conditional correlation multivariate GARCH), here called the rolling Hong and DCC-MGARCH Hong tests, respectively. The proposed tests are used to examine time-varying information spillover among global crude oil markets. The results provide empirical evidence of time-varying information spillover. In particular, the instantaneous causal effects of Dubai and Tapis crudes on Brent and WTI become stronger when a major event or events occur in major oil-producing countries. Such events include the Iraq War in March 2003, OPEC's announcement of a record production cut in December 2008, and the Libyan civil war in early 2011. And consistent with previous studies, WTI and Brent play dominant roles in global crude markets. Impulse response analysis shows that market information has a positive influence on the spillover effect in global crude oil markets. Moreover, the DCC-MGARCH Hong test consistently leads the rolling Hong test, which indicates that the former performs better.