Multivariate Granger Causality and Generalized Variance

Abstract

Granger causality analysis is a popular method for inference on directed interactions in complex systems of many variables. A shortcoming of the standard framework for Granger causality is that it only allows for examination of interactions between single (univariate) variables within a system, perhaps conditioned on other variables. However, interactions do not necessarily take place between single variables, but may occur among groups, or "ensembles", of variables. In this study we establish a principled framework for Granger causality in the context of causal interactions among two or more multivariate sets of variables. Building on Geweke's seminal 1982 work, we offer new justifications for one particular form of multivariate Granger causality based on the generalized variances of residual errors. Taken together, our results support a comprehensive and theoretically consistent extension of Granger causality to the multivariate case. Treated individually, they highlight several specific advantages of the generalized variance measure, which we illustrate using applications in neuroscience as an example. We further show how the measure can be used to define "partial" Granger causality in the multivariate context and we also motivate reformulations of "causal density" and "Granger autonomy". Our results are directly applicable to experimental data and promise to reveal new types of functional relations in complex systems, neural and otherwise.

added 1 reference, minor change to discussion, typos corrected; 28 pages, 3 figures, 1 table, LaTeX

References in corpus (8)

• Measuring Information Transfer
• Granger causality and transfer entropy are equivalent for Gaussian variables
• Kernel method for nonlinear Granger causality
• Analyzing Multiple Nonlinear Time Series with Extended Granger Causality
• Automatic Filters for the Detection of Coherent Structure in Spatiotemporal Systems
• On the Dynamical Complexity of Small-World Networks of Spiking Neurons
• What Is a Macrostate? Subjective Observations and Objective Dynamics
• Redundant variables and Granger causality

Cited by in corpus (24)

• Exploration of synergistic and redundant information sharing in static and dynamical Gaussian systems
• Granger causality for state space models
• A unified framework for information integration based on information geometry
• Transfer Entropy as a Log-likelihood Ratio
• The relation between Granger causality and directed information theory: a review
• Synergy and redundancy in the Granger causal analysis of dynamical networks
• Multiscale Information Decomposition: Exact Computation for Multivariate Gaussian Processes
• Efficient Algorithms for Searching the Minimum Information Partition in Integrated Information Theory
• Dynamical independence: discovering emergent macroscopic processes in complex dynamical systems
• Identifying Coupling Structure in Complex Systems through the Optimal Causation Entropy Principle
• Understanding causation via correlations and linear response theory
• Multiscale Wavelet Transfer Entropy with Application to Corticomuscular Coupling Analysis
• Causation entropy from symbolic representations of dynamical systems
• The Global Dynamical Complexity of the Human Brain Network
• Non-Parametric Causality Detection: An Application to Social Media and Financial Data
• Influence of time delay on information exchanges between coupled linear stochastic systems
• Geometry of Information Integration
• The Many Faces of Fluctuation-Dissipation Relations Out of Equilibrium
• Solved problems and remaining challenges for Granger causality analysis in neuroscience: A response to Stokes and Purdon (2017)
• Sampling distribution for single-regression Granger causality estimators
• Linear scaling causal discovery from high-dimensional time series by dynamical community detection
• Robust Non-linear Wiener-Granger Causality For Large High-dimensional Data
• Multiscale Partial Information Decomposition of Dynamic Processes with Short and Long-range correlations: Theory and Application to Cardiovascular Control
• Causal information approach to partial conditioning in multivariate data sets