Abstract:
This paper studies the propagation and properties of a confidence shock in a structural vector autoregression (VAR) model with and without financial variables. The addition of a financial block does not considerably change the propagation and the contribution to the forecast error variance by the confidence shock. Nevertheless, for specific historical episodes, the inclusion of a financial block plays a role. In several recessions, the VAR with the financial block assigns a smaller role to confidence shocks for the fall in GDP. This suggests that the confidence shock may not be properly identified in a structural VAR when financial variables are omitted. Further, I identify a financial channel by which the confidence shock affects economic activity.