The Role of Intra-Day and Inter-Day Data Effects in Determining Linear and Nonlinear Granger Causality Between Australian Futures and Cash Index Markets

In order to explain the incidence of Granger causality between indices from the futures and the underlying cash market, as reported by numerous empirical studies in the literature, it is important to account for mean and volatility (second-order) persistence effects in the data. Further, there is need to control for inter-day and intra-day effects by imposing an appropriate autocorrelation structure upon each of the index returns from both markets. Once all these effects are controlled for, then linear Granger causality ceases to be statistically significant and the associated lead-lag phenomenon is no longer observable when the information flow between the spot and futures markets is completed within a five-minute observation interval. Additionally, nonlinear Granger causality testing indicates no compelling need to account for nonlinear effects (beyond the second-order moment condition) in order to explain causality. This result supports the price discovery role of futures markets.