Conditional versus unconditional forecasting with the New Area-Wide Model of the euro area

In this paper we examine conditional versus unconditional forecasting with a version of the New Area-Wide Model (NAWM) of the euro area designed for use in the context of the macroeconomic projection exercises at the European Central Bank (ECB). We first analyse the out-of-sample forecasting properties of the estimated model from 1999 to 2005 by comparing its unconditional forecasts with those obtained from a Bayesian VAR with a steady-state prior as well as na¨ıve forecasts. Model-based forecasts that are conditioned on differing information sets are then studied and evaluated through, for instance, modesty statistics to assess the relevance of the Lucas critique. In contrast to other studies in the literature, we condition on a fairly large set of policy-relevant variables. Furthermore, we consider conditioning information that partially, albeit not fully determine the future path of the observed variables, but which restrict the channels through which they can be affected.