Online Appendix to Identifying News Shocks with Forecast Data

This section provides another example to demonstrate that adding expectational variables to the set of observables is informative for identifying news shocks. The example in Section 2 of the paper is simple enough to see the basic idea of our approach for identifying news shocks. In the empirical analysis presented in the subsequent sections of the paper, however, each exogenous disturbance is assumed to follow a more complicated stochastic process. Specifically, it is governed by a first-order autoregressive process incorporated with news shocks up to five quarters ahead. To analyze the role of observed expectational variables in the identification of news shocks with longer forecast horizon, the model presented in Section 2 is extended to