Australian and U.S. growth cycle linkages, 1967–1983

Abstract

In this paper the hypothesis that the U.S. growth cycle is “causal” to the Australian growth cycle is tested within the Granger-causality context. A dynamic transfer function model linking the Australian cycle to that of the U.S. is empirically specified, using the approach of Box and Jenkins (1976) , with the finding that U.S. cyclical activity leads on average by about eight months during the estimation period considered. In terms of MSE the resulting model outperforms a corresponding ARIMA model for the Australian growth cycle in a post-estimation forecasting evaluation analysis, but the reduction in MSE is found to be only marginal and statistically insignificant. These results provide little empirical support for the existence of a systematic linkage between cyclical movements in the two economies.