Causality Test in Time Series
Yuji Nakano · 2025
We introduce the local causal value, which measures the degree of influence of time series Y to time series X. When X and Y are stationary processes, we propose the Local Causality Test, which tests the existence of local causality between X and Y . If a stationary time series Z influences X and Y , we can constructε_{ X,Z } and ε_{ Y,Z } by removing the influence of Z , respectively. Applying the Local Causal Test, we propose the Partial Local Causality Test, which tests that Z is a hidden factor of causality between X and Y .