The Study of the Applicability of Stochastic Simulation-Based Granger Causality Testing Methods
Xiaodi Zhang, Shushan Li · 2023
This study investigates the applicability of traditional Granger causality tests under different conditions by establishing various causal relationship models and generating corresponding two-dimensional time series through Monte Carlo simulation. Traditional Granger causality tests are then used to explore the causal relationships from the perspectives of normality, independence, and homoscedasticity. The research findings highlight that when the time series does not meet the preset conditions, the accuracy of the Granger causality test in determining causal relationships tends to be low, and particularly weak in the case of nonlinear relationships. These results emphasize the impact of conditions on the accuracy of the test outcomes.