Noisy News in Business Cycles

We investigate the role of "noise" shocks as a source of business cycle fluctuations. To do so we set up a simple model of imperfect information and derive restrictions for identifying the noise shock in a VAR model. The novelty of our approach is that identification is reached by means of dynamic rotations of the reduced-form residuals. We find that noise shocks generate hump-shaped responses of GDP, consumption and investment, and account for a sizable fraction of their prediction error variance at business cycle horizons.