Long-Run Causal Order: A Preliminary Investigation
37 Pages Posted: 23 May 2018 Last revised: 24 Sep 2019
Date Written: May 11, 2018
The concept of long-run causal order and its relationship to nonstationarity in time series data is explored using some ideas drawn from the literature on graphical causal modeling. Ordinary variables, which are nonstationary only because they are caused by a distinct nonstationary variable, are are distinguished from fundamental trends, which are nonstationary owing to their own-dynamics. Although systems of equations in which the own dynamics appears stationary may, in principle, generate nonstationary behavior through their interaction, it is argued that such behavior is nongeneric and that typically ordinary variables trend because an (often latent) fundamental trends is among their causes. The possibility of inferring the long-run causal structure almong a set of time-series variables from an exhaustive examination of weak exogeneity in irreducibly cointegrated subsets of variables is explored and illustrated.
Keywords: graphical causal modeling, causal search, cointegrated vector autoregression (CVAR), weak exogeneity, irreducible cointegrating relations
JEL Classification: C32, C51, C18
Suggested Citation: Suggested Citation