Causality in stochastic and chaotic processes

Abstract

Identifying causal relations in complex systems such as economic systems is important for effective policy, management recommendations, research, etc. The tools for detecting causality may vary according to the nature of the system analized. This study analyzes the performance of a complete algorithm of causality that takes into account different dynamics. The causality tests analyzed are the Convergent Cross Mapping and a nonlinear extension of the Granger-causality test. These tests determine causation in two time series only after being filtered by the Motifs Profile test, a test used for classifying processes which enabled us to determine which causality test to use in each case.