Is the Business Cycles a Necessary Consequence of Stochastic Growth?

We compute the forecastable changes in output, consumption, and hours implied by a VAR that includes the growth rate of private value added, the share of output that is consumed, and the detrended level of private hours. We show that the size of the forecastable changes in output greatly exceeds that predicted by a standard stochastic growth model, of the kind studied by real business cycle theorists. Contrary to the model's implications, forecastable movements in labor productivity are small and only weakly related to forecasted changes in output. Also, forecasted movements in investment and hours are positively correlated with forecasted movements in output. Finally, and again in contrast to what the growth model implies, forecasted output movements are positively related to the current level of the consumption share and negatively related to the level of hours. We also show that these contrasts between the model and the observations are robust to allowance for measurement error and a variety of other types of transitory disturbances.