US work-from-home share roughly tripled between 2019 and 2021

The signal

Signal Atlas's 02 Enterprise sector logs one of the sharpest survey-based shifts in its archive here. A US Census Bureau press release published 15 September 2022, drawing on the American Community Survey (ACS) 1-year estimates, states that the share of workers whose primary means of getting to work was "worked at home" rose from 5.7% in 2019 (about 9 million people) to 17.9% in 2021 (about 27.6 million people), roughly a threefold increase across two survey years. The release also reports wide geographic variation: the District of Columbia led at 48.3%, and the San Francisco and San Jose metro areas each recorded rates near 35%, which the release attributes to their information-technology employment base.

The evidence

The ACS measures this through its journey-to-work question, which asks a respondent's primary means of transportation and includes "worked at home" as an answer category, a proxy for remote work rather than a direct measure of hours worked remotely. A different federal series measures a related but distinct concept: a Bureau of Labor Statistics release published 24 September 2019, drawing on the American Time Use Survey's Leave and Job Flexibilities Module for 2017-2018, found "about 36 million wage and salary workers (25 percent) worked at home at least occasionally," and 15% "had days they only worked at home," a higher pre-pandemic baseline than the ACS's 5.7% commute-mode figure for 2019, because the two surveys ask different questions of different populations over different periods. Reading the two series side by side, rather than treating either alone as the remote-work number, is the more defensible practice.

Timeframe and confidence

This is an editorial reading: the ACS is a large, methodologically documented annual survey, per the Census Bureau's ACS programme page, current as retrieved on 16 September 2026, so confidence in the 5.7%-to-17.9% comparison, as a measure of primary commute mode, is high within that series. Confidence that this single figure represents "the" US remote-work rate is lower, given the BLS series shows a different, broader figure measuring a related but non-identical concept over a different period.

What would change the reading

A subsequent ACS release for 2022 or later years showing the "worked at home" share holding near the 2021 level, rather than reverting toward 2019's, would support reading 2021 as a durable shift rather than a pandemic-year peak; this review did not open a later ACS release, so that question remains open rather than answered here.

• Does a remote-work figure come from a commute-mode question, a time-use survey, or a payroll record, and do those measure the same thing?
• Is a 2021 figure being compared to its correct 2019 baseline from the same survey series, rather than to a different series' baseline?
• Has a later release from the same survey shown the 2021 level persisting, reverting, or continuing to rise?

The 2021 ACS figure documents a survey-year peak with a precise, sourced comparison to 2019; whether it marks a lasting reweighting of how Americans get to work, or a pandemic-year high later partly reversed, is a question for the survey's own subsequent releases to answer, not one this record resolves.

Source trail

1. Around 1 in 5 Home Workers in U.S. Worked Fully Remotely in 2021www.census.gov · Source publication: 2022-09-15 · Retrieved 2026-09-16

   States the 5.7% (2019) to 17.9% (2021) primary-work-from-home share from the ACS and its geographic breakdown.

2. American Time Use Survey — Leave and Job Flexibilities News Releasewww.bls.gov · Source publication: 2019-09-24 · Retrieved 2026-09-16

   Gives a pre-pandemic baseline from a different survey question: 25% worked at home at least occasionally in 2017-2018.

3. American Community Survey (ACS)www.census.gov · Source publication: not established · Retrieved 2026-09-16

   Living programme page describing the ACS's annual methodology used to produce the work-from-home estimates.

Read across the evidence

• US office vacancy climbed to a series record in 2023
• Retail sales data is also a survey, just of firms not people
• Teams daily users passed 44 million as lockdowns began

Platforms & distribution · All 100 historical records