5050 Fifty Fifty Politics ← Go to Homepage
Labor

Offices Are Emptier After Five Years of Return-to-Office Orders Than They Were in the Pandemic

Fifty Fifty Politics · Background & Data
The share of Americans working from home has barely moved in three years, despite a federal mandate ordering roughly two million people back to a desk. Office vacancy has meanwhile hit an all-time record. Here is what the numbers show, and why the evidence on productivity is more honestly described as mixed than either side admits.

The number that did not move

In 2025, 35 percent of employed Americans did some or all of their work at home on days they worked. That was the year the federal government ordered its entire workforce back to the office five days a week. The number did not fall.

The broader time series tells the same story. The share of paid full working days done from home was 7.2 percent in December 2019, peaked at 61.7 percent in May 2020, and has settled into a slow drift: 32.8 percent in June 2021, 28.3 percent in 2023, 27.9 percent in 2025 and 26.0 percent in mid-2026.

That is a decline of about two points a year against a pre-pandemic baseline four times lower. Whatever return-to-office mandates have achieved, reversing the shift is not it. The arrangement has stabilised at roughly a quarter of all work, which is a permanent change in how the country works rather than a hangover from an emergency.

Who gets to work from home, by education. Source: Bureau of Labor Statistics, Beyond the Numbers vol. 14, March 2025, first quarter 2024 data Who gets to work from home, by education Advanced degree 43.6% Bachelor's 38.4% Some college 18.3% High school 8.5% No diploma 3.5%
Source: Bureau of Labor Statistics, Beyond the Numbers vol. 14, March 2025, first quarter 2024 data

Meanwhile the buildings emptied anyway

National office vacancy reached 21 percent in the first quarter of 2026, an all-time record. Through 2019 it sat between 16.2 and 16.8 percent.

That is the fact that makes the downtown argument serious rather than sentimental. Office buildings are not just landlord property. They are a commercial tax base that funds buses, libraries, fire crews and schools, and falling values feed directly into municipal budgets.

The people affected first are not the executives arguing about hybrid schedules. They are the cleaners, security guards, cooks and drivers whose jobs existed because offices were full, and none of that work can be done from a spare bedroom. This is the strongest argument on the left for caring about office occupancy, and it is not an argument about productivity at all.

Share of paid full days worked from home. Source: WFH Research, Survey of Working Arrangements and Attitudes, monthly series Share of paid full days worked from home 0% 20% 40% 60% 80% Dec 19 May 20 Jun 21 Jun 22 Jun 23 Jun 24 Jun 25 Jun 26
Source: WFH Research, Survey of Working Arrangements and Attitudes, monthly series

The productivity evidence, stated honestly

Both sides cherry-pick here, so it is worth laying out the whole picture.

A randomised trial at a large travel company, published in Nature, found hybrid work with two days at home had no effect on performance grades or promotions and cut quit rates by about a third. That is real, and it is a study of hybrid work, not fully remote work.

A randomised trial of data entry workers found output 18 percent lower at home. A Federal Reserve Bank of New York study found remote call handlers took 12 percent fewer calls an hour, but decomposed that into roughly eight points of selection, meaning who chooses remote work, and only about four points of actual causal effect. Both sides quote that paper, and both quote it correctly.

A separate study of a call centre that went fully remote found productivity rose about 10 percent while the applicant pool widened. Federal Reserve Board work argues aggregate productivity gains from remote work take three to six years to show up at all.

The defensible summary is that hybrid work has good evidence behind it, fully remote has genuinely mixed evidence, and results vary enormously by task, by firm and by who selects in. Neither side gets to say the literature settles it.

Who actually gets the choice

Remote work sorted sharply by education. Among workers with an advanced degree, 43.6 percent teleworked. For those with a bachelor's degree it was 38.4 percent, for some college 18.3 percent, for a high school diploma alone 8.5 percent, and for those without one 3.5 percent.

Nurses, warehouse pickers, bus drivers and line cooks got no version of this benefit and no compensation for its absence. A debate that treats remote work as the central labour question of the decade quietly writes off most of the workforce, and that silence is itself a political choice.

The same data supports a different conclusion on the other side. Remote hiring carried jobs out of a few expensive coastal metros: the average distance between a worker's home and their employer's worksite rose from 15 miles in 2019 to 26 miles in 2024. Rural residents, military spouses and disabled workers gained access to jobs that were never going to relocate to them. Blanket return orders take that back.

The core disagreement

This one splits oddly. Unions have defended telework as a negotiated contract term, and arbitrators at several federal agencies ruled through 2026 that a presidential memo cannot cancel what both sides signed. Some progressives simultaneously worry about downtown decline and about the class gap between remote-capable and in-person workers.

On the other side, some conservatives argue that where work happens is an ordinary term of employment and a statutory right to remote work would be a sweeping mandate on every employer in the country. Others point out that a government directive telling two million people where to sit is central workforce planning rather than conservatism, and that agencies discovered they lacked the desks.

Underneath it sits a question neither camp has answered cleanly. Commuting is unpaid time and unpaid cost. When a job goes remote, the worker keeps both and prices it into what the role is worth. Ordering a return without raising pay is therefore a pay cut in everything but name, and 46 percent of remote workers say they would be unlikely to stay if the option disappeared. Whether an employer owes anything for taking that back is the argument that has not really been had.

Want the core arguments from both sides, side by side?

See both sides of the debate on Remote Work & Return to Office →
Sources
Browse All Blogs