The time-series literature reports two stylized facts about output dyn
amics in the United States: GNP growth is positively autocorrelated, a
nd GNP appears to have an important trend-reverting component. This pa
per investigates whether current real-business-cycle (RBC) models are
consistent with these stylized facts. Many RBC models have weak intern
al propagation mechanisms and must rely on external sources of dynamic
s to replicate both facts. Models that incorporate labor adjustment co
sts are partially successful. They endogenously generate positive auto
correlation in output growth, but they need implausibly large transito
ry shocks to match the trend-reverting component in output.