Diverse Causality Inference In Foreign Exchange Markets

The relationship between currencies in foreign exchange markets has been a topic of significance in economics. Previous studies have focused more on correlations between currencies. However, the detection of causality can reveal their inherent laws. Although the traditional Granger causality test can identify causality, it cannot take into account the nature and intensity of the causality. Thus, the objective of this paper is to identify the causalities of currencies from the perspective of dynamics. In this paper, we select 25 currencies (with the US dollar (USD) as the numeraire) from foreign exchange markets, as they occupy large shares in their regions. To detect the causalities of the foreign exchange markets, we combine PC (pattern causality) theory and complex networks to construct directed and weighted causality networks, in which the nodes represent the currencies and the directed edges represent the causal intensities. Furthermore, we study the symmetry of each causality and quantify the symmetry degree. The results demonstrate that causalities exist between currencies that differ in terms of nature and intensity. The positive causality network exhibits substantial robustness, which can be regarded as the dominant causal relationship in the foreign exchange markets, although a few exceptions are encountered, such as the dominant negative and disordered causalities between currency pairs. In addition, the dominant causalities between most currencies are symmetric in terms of nature, and they also exhibit symmetry in terms of intensity. Furthermore, by gradually deleting the network by thresholding according to the edge weights, we identify the important driving currencies of the markets. This paper may provide valuable information for investors and supervisory departments.