Quarterly Labor Input Measurement: Extending the Industry-Level Production Account to Business Cycle Frequencies

Industry-level labor input is conventionally measured annually, which averages away the sharpest movements in hours and workforce composition during rapid recessions. This paper extends the U.S. Bureau of Economic Analysis annual Industry-Level Production Account to a quarterly frequency for 63 industries over 2003–2024. To overcome the sparsity of Current Population Survey microdata, we combine state-space smoothing of marginals, Denton-Cholette benchmarking, and iterative proportional fitting to recover a reconciled distribution. The resulting quarterly quality-adjusted labor input series demonstrates that annual aggregation cannot resolve short-duration shocks. Because the COVID–19 contraction and early recovery occurred within a single calendar year, the peak-to-trough raw-hours decline measured quarterly is 6.7 percentage points deeper than its annual counterpart, and the labor composition spike is 1.3 points larger. Furthermore, quarterly data reveal that COVID–19’s composition shocks were concentrated and transitory, whereas the Great Recession’s were broad and persistent. At the industry level, the COVID–19 quarterly-annual measurement differences are near-uniform shifts across industries, and the industries with the largest composition spikes reverse the most through the recovery.