Working Paper: NBER ID: w31193
Authors: Erik Brynjolfsson; John J. Horton; Christos Makridis; Alexandre Mas; Adam Ozimek; Daniel Rock; Hongyi Tuye
Abstract: Remote work surged during the Covid pandemic but there is disagreement about the extent of the change. To address this question, we field a new, nationally-representative survey: the Remote Life Survey (RLS). We find that in October 2020, 31.6 percent of the continuously employed workforce always worked from home (WFH) and 21.9 percent sometimes or rarely WFH, totaling 53.5 percent. We compare our results with alternative measurement approaches, with a focus on government surveys and provide estimates on the impact of four factors: (a) differences among mail versus web-based survey respondents, (b) differences in the inclusion of self-employed workers, (c) the industry mix of the sample, and (d) the exclusion of people who were already remote pre-pandemic. We find that the last explanation (d) explains the bulk of the difference in estimates between the Current Population Survey (CPS) and other measures of remote work. Policymakers and researchers who turn to the BLS-CPS data series for an estimate of remote work prevalence in the American economy should note that it might be underestimating WFH levels by up to 25 percentage points. Under our preferred estimates, we find that about half of the U.S. workforce worked remotely at least one day each week as of December 2020.
Keywords: No keywords provided
JEL Codes: I15; J21; L23; M15; M5
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Exclusion of individuals who were already working remotely before the pandemic (J63) | Underestimation of the prevalence of remote work (J29) |
| Differences in survey modalities (mail vs. web) (C83) | Overestimation of remote work (J29) |
| Industry mix (L69) | Measurement gap between estimates and CPS data (C80) |
| Design and administration of surveys (C83) | Impact on measurement of remote work (F66) |
| Accurate measurement of remote work (J29) | Understanding its economic implications (F69) |