Causal Reasoning in Graphical Time Series Models

TL;DR

This paper introduces a formal definition of causality in multivariate time series, establishing conditions for identifiability and methods for computing causal effects, with illustrations for linear models.

Contribution

It provides a novel causality framework for time series, including graphical criteria and computational methods, extending causal inference to dynamic data.

Findings

Conditions for causal effect identifiability are established.

Graphical criteria for causality are verified.

Causal effect computation is demonstrated for linear models.

Abstract

We propose a definition of causality for time series in terms of the effect of an intervention in one component of a multivariate time series on another component at some later point in time. Conditions for identifiability, comparable to the back-door and front-door criteria, are presented and can also be verified graphically. Computation of the causal effect is derived and illustrated for the linear case.