DP9601 Noisy News in Business cycles
In a situation where agents can only observe a noisy signal of the shock to future economic fundamentals, SVAR models can still be successfully employed to estimate the shock and the associated impulse response functions. Identification is reached by means of dynamic rotations of the reduced form residuals. We use our identification approach to investigate the role of the "noise" shock the component of the signal observed by agents which is unrelated to economic fundamentals as a source of business cycle
fluctuations. We find that noise shocks generate hump-shaped responses of GDP, consumption and investment and account for about a third of their prediction error variance at business cycle horizons.