Granger causality for state space models

Abstract

Granger causality, a popular method for determining causal influence between stochastic processes, is most commonly estimated via linear autoregressive modeling. However, this approach has a serious drawback: if the process being modeled has a moving average component, then the autoregressive model order is theoretically infinite, and in finite sample large empirical model orders may be necessary, resulting in weak Granger-causal inference. This is particularly relevant when the process has been filtered, downsampled, or observed with (additive) noise - all of which induce a moving average component and are commonplace in application domains as diverse as econometrics and the neurosciences. By contrast, the class of autoregressive moving average models - or, equivalently, linear state space models - is closed under digital filtering, downsampling (and other forms of aggregation) as well as additive observational noise. Here, we show how Granger causality, conditional and unconditional, in both time and frequency domains, may be calculated simply and directly from state space model parameters, via solution of a discrete algebraic Riccati equation. Numerical simulations demonstrate that Granger causality estimators thus derived have greater statistical power and smaller bias than pure autoregressive estimators. We conclude that the state space approach should be the default for (linear) Granger causality estimation.