∗Gary Koop is a Senior Fellow at the Rimini Center for Economic Analysis. Emails:

This appendix is divided into seven sections labelled A through G. Almost all details of the prior are specified in the paper itself, but the few remaining details about the prior are given in Appendix A. An outline of the MCMC algorithm was provided in the paper, but complete details and formulae are provided in Appendix B. Appendix C describes how we calculate the DIC which is one of the methods of model comparison used in our empirical section. Appendix D provides details about the transformation of the expanded form VARMA parameters to other ways of parameterizing the VARMA. Appendix E presents results on the performance of our algorithms in a variety of artificial data sets. Appendix F presents additional empirical results for our macroeconomic application. 2 Appendix A: Priors The empirical work in this paper uses relatively noninformative priors. The hierarchical SSVS priors for the VARMA coefficients are described in Section 3 of the paper. Recall that in terms of these, we specify uniform priors on the Kronecker indices κ. Moreover, for both the hard and soft SSVS priors, we set τ 21,j,ik = 1; for soft SSVS we set τ