Abstract:
Aggregate productivity suffers when workers and machines are not matched with their most productive uses. This paper uses restricted U.S. Census microdata and a model that features industry-specific markups, industry-specific returns to scale, and establishment-specific distortions, to show that the extent of this misallocation in U.S. manufacturing declined by 13% between 1977 and 2007. This finding starkly contrasts with the 29% increase in misallocation implied by the widely-used Hsieh-Klenow (2009) model, which assumes that all establishments charge the same markup and have constant returns to scale. We jointly estimate industry markups and returns to scale parameters, and show that these characteristics vary substantially across industries and over time. Furthermore, while the average markup has been relatively constant, the average returns to scale declined over this period. Accounting for these features of the data leads to the strong divergence of our estimates from those implied by the Hsieh-Klenow model.