Causal diagrams for I(1) structural VAR models
Granville Tunnicliffe Wilson and Marco Reale
View Report [PDF - 160 KB]
Abstract
Structural Vector Autoregressions allow dependence among contemporaneous variables. If such models have a recursive structure, the causal relation among the variables can be represented by directed acyclic graphs. The identi.cation of these relationships for stationary series may be enabled by the examination of the conditional independence graph constructed from sample partial autocorrelations of the observed series. In this paper we extend this approach to the case when the series follow an I(1) vector autoregression. We show that, even though the theoretical partial autocorrelations are unde.ned for integrated processes, exactly the same data procedures and sampling properties may be applied. The theoretical reasoning is supported by the empirical results of simulation, and applications from banking series and term interest rates are used to illustrate the procedure.
Back to Research Reports
