Identify the characteristic in the evolution of the causality between the gold and dollar

Abstract

The causal inference method based on the time-series analysis has been subject to intense scrutiny, by which the interaction has been revealed between gold and the dollar. The positive or negative causality between them has been captured by the existing methods. However, the dynamic interactions are time-varying rather than immutable, i.e., the evolution of the causality between gold and the dollar is likely to be covered by the statistical process. In this article, a method which combines the pattern causality and the state-transition network is developed to identify the characteristics of the causality evolution between gold and the dollar. Based on this method, we can identify not only the causality intensity but also the causality type, including the types of positive causality, negative causality and the third causality (dark causality). Furthermore, the patterns of the causalities for the segments of the bivariate time series are transformed to a state-transition network from which the characteristics in the evolution of the causality have also been identified. The results show that the causality has some prominent motifs over time, that are the states of negative causality. More interestingly, the states that act as a bridge in the transition between states are also negative causality. Therefore, our findings provide a new perspective to explain the relatively stable negative causality between gold and the dollar from the evolution of causality. It can also help market participants understand and monitor the dynamic process of causality between gold and the dollar.

Keywords

• causal inference
• time-series analysis
• pattern causality
• state-transition network
• motif