ABSTRACT
In December 2021, with COVID-19 cases about to reach their peak, US employment still remained below its pre-pandemic level, with data indicating a strong labour demand and a weak labour supply. Many US individuals did not work, out of a concern about Covid-19. At the same time, the pandemic affected businesses of different sizes differently. At first, the pandemic hit especially hard small businesses. But later in 2020 the recovery was stronger at smaller businesses than at larger ones. These developments call for a renewed attention to how the labour supply varies across employers of different sizes. I focus on two important determinants of labour supply during a pandemic: the vulnerability to the severe effects of the virus because of pre-existing health conditions and the ability to work from home. I study how these two factors vary across establishments of different sizes. I find that the ability to work from home is higher at larger establishments, for both workers with and without COVID-19 health risk. However, the difference in ability to work from home across these two types of workers does not decrease at larger employers.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 An early unpublished paper in the same vein is Kahn, Lange, and Wiczer (Citation2020). They use the same data that I use, but they do not stratify the analysis by establishment size.
2 MPW’s data is at the 4-digit occupation level. With a crosswalk, I aggregate the LWFH measure at the NHIS’ coarser 3-digit level, using national employment weights from MPW. LWFH still remains a continuous – not binary – measure. The aggregation does not lose too much information: in a weighted (by employment) regression of the 4-digit LWFH on 3-digit occupation dummies, the adjusted R2 is 83.8%.
3 Indeed, in unreported results, the mean LWFH is 0.27 (0.21) for workers with (without) COVID-19 health risks.
4 E.g., for at-risk workers in the largest employers (1000+), relative to the baseline, the difference in LWFH is −0.054 (i.e. −0.080 + 0.026) with a standard error of 0.01. Full test results available from the author.