Causal decomposition on multiple time scales: Evidence from stock price-volume time series

Abstract

This paper proposes a new causal decomposition method and further applies it to the study of information flow on different time scales between two financial time series. The method of causal decomposition mainly includes three steps: decomposition, reconstruction, and causality test. Firstly, the original signal is decomposed into information components of different frequencies by the complete ensemble empirical mode decomposition with adaptive noise, and then the information components of different frequencies are effectively divided into high-frequency components, low-frequency components and long-term trend with the help of the Fine-to-coarse reconstruction. Finally, counterfactual series are constructed and statistical causality test is conducted. In this way, the driving factors of causality can be tracked from the perspective of information frequency. This paper re-examined the causality between stock price and volume from the time-frequency perspective with the application of causal decomposition, and found strong evidence to support whether the statistical causality between them is significant or not related to the information components of different frequencies.