We provide a few new empirical facts that theoretical models should feature in order to be consistent with US data. 1) There are two classes of shocks: demand and supply. Supply shocks have long-run effects on economic activity, demand shocks do not. 2) Both supply and demand shocks are important sources of business cycles fluctuations. 3) Supply shocks are the primary driver for consumption fluctuations, demand shocks for investment. 4) The demand shock is closely related to the credit spread, while the supply shock is essentially a news shock. The results are obtained using a novel approach which combines frequency domain identification and Dynamic Factor Model analysis.