Between 1890 and 2004 total factor productivity (TFP) growth in the United States has been strongly procyclical, while labor productivity growth has been mildly so. This article argues that these results are not simply a statistical artifact, as Mathew Shapiro and others have argued. Procyclicality resulted principally from demand shocks interacting with capital services which are relatively invariant over the cycle. This account contrasts with explanations emphasizing labor hoarding as well as those offered by the real business cycle (RBC) program, in which TFP shocks (deviations from trend) are themselves the cause of cycles.