Weak Exogeneity, Driving Trends and Non‐Causality. Is Dynamic Adjustment Informative About Causality?

ABSTRACT In VECMs (or CVARs), the weak exogeneity of some of the variables (for the long‐run parameters) is related to the equilibrium adjustment dynamics. This has sometimes led to the statistical concept of weak exogeneity being given a causal interpretation. This paper analyses the causal informational value of weak exogeneity, using graphical representations of causal structures, to show when such an interpretation is valid. We specifically consider the identification of the driving variables, and their relationship to non‐causality. We demonstrate that causal interpretations of weak exogeneity may not hold when relevant variables are omitted. Our exploration of the conditions under which weak exogeneity conveys causal information helps further the study of long‐run causal inference.