Existing Real Business Cycle (RBC) models assume that the key impulses to business cycles are stochastic technology shocks. RBC analysts typically measure these technology shocks by the Solow residual. This paper assesses the sensitivity of inference based on Solow residual accounting to labor hoarding behavior. Our main results can be summarized as follows. First, the quantitative implications of RBC models are very sensitive to the possibility of labor hoarding. Allowing for such behavior reduces our estimate of the variance of technology shocks by 50%. Depending on the sample period investigated, this reduces the ability of technology shocks to account for aggregate output fluctuations by 30% to 60%. Second, our labor hoarding model is capable of quantitatively accounting for the observed correlation between government consumption and the Solow residual. Third, unlike standard RBC models, our labor hoarding model is consistent with three important qualitative features of the joint behavior of average productivity and hours worked: (i) average productivity and hours worked do not display any marked contemporaneous correlation, (ii) average productivity is positively correlated with future hours worked, and (iii) average productivity is negatively correlated with lagged hours worked.
