The shift to work from home (WFH) has been a large and persistent consequence of the pan- demic. To quantify the effect of WFH on the macroeconomy, researchers have exploited the fact that local labor markets are differentially exposed to this shock, in either empirical or quantitative spatial settings. These analyses require a measure of WFH at disaggregated levels. In this paper, we compare several important measures used in the literature: Barrero, Bloom, and Davis (2021); Bick, Blandin, and Mertens (2022); Dingel and Neiman (2020); and the American Community Survey (ACS). While these measures differ in how comprehensively they measure WFH (e.g., they may or may not include hybrid work), we show that they are highly correlated in the cross section. Therefore, these measures will yield similar causal effects once appropriately scaled by the average level of WFH. We argue that when choosing a particular measure, researchers should carefully consider the trade-off between how comprehensively WFH is measured and measurement error in the survey at the particular level of geographic aggregation.

The shift to work from home (WFH) has been a large and persistent consequence of the pan- demic. To quantify the effect of WFH on the macroeconomy, researchers have exploited the fact that local labor markets are differentially exposed to this shock, in either empirical or quantitative spatial settings. These analyses require a measure of WFH at disaggregated levels. In this paper, we compare several important measures used in the literature: Barrero, Bloom, and Davis (2021); Bick, Blandin, and Mertens (2022); Dingel and Neiman (2020); and the American Community Survey (ACS). While these measures differ in how comprehensively they measure WFH (e.g., they may or may not include hybrid work), we show that they are highly correlated in the cross section. Therefore, these measures will yield similar causal effects once appropriately scaled by the average level of WFH. We argue that when choosing a particular measure, researchers should carefully consider the trade-off between how comprehensively WFH is measured and measurement error in the survey at the particular level of geographic aggregation.

The shift to work from home (WFH) has been a large and persistent consequence of the pan- demic. To quantify the effect of WFH on the macroeconomy, researchers have exploited the fact that local labor markets are differentially exposed to this shock, in either empirical or quantitative spatial settings. These analyses require a measure of WFH at disaggregated levels. In this paper, we compare several important measures used in the literature: Barrero, Bloom, and Davis (2021); Bick, Blandin, and Mertens (2022); Dingel and Neiman (2020); and the American Community Survey (ACS). While these measures differ in how comprehensively they measure WFH (e.g., they may or may not include hybrid work), we show that they are highly correlated in the cross section. Therefore, these measures will yield similar causal effects once appropriately scaled by the average level of WFH. We argue that when choosing a particular measure, researchers should carefully consider the trade-off between how comprehensively WFH is measured and measurement error in the survey at the particular level of geographic aggregation.