Industry evidence and the vanishing cyclicality of labor productivity

Aggregate labor productivity used to be strongly procyclical in the United States, but the procyclicality has largely disappeared since the mid-1980s. This paper explores the industry-level evidence in order to discriminate between existing explanations of the vanishing procyclicality of the labor productivity. I document the change in the cyclical properties of productivity in the U.S. using industry-level data and focus on a particularly puzzling feature, namely that the correlations of the industry productivity with industry output and labor input remained on average much more stable before and after the mid-1980s compared to the aggregate correlations. In other words, there is little evidence for the vanishing cyclicality of labor productivity at the industry level. I construct a simple industry-level RBC model that nests two leading explanations of the vanishing cyclicality of productivity that have been proposed in the literature. I show that the two explanations have qualitatively different predictions for the cyclical properties of industry-level variables. The mechanism based on a structural change in the composition of aggregate shocks is able to replicate the stability of industry-level moments across time. In contrast, the mechanism based on increased labor market flexibility is less successful in matching the industry-level evidence.