What Science Says About Remote Work Productivity
The answer is not mandatory office work, and it is not remote work on principle. It depends on the task, the metric, and the coordination cost.
Remote-work productivity has become an argument about taste. One side treats home working as proof that the office was unnecessary. The other treats it as a loss of control. The research is narrower, which makes it more useful: remote work performs best when the work is measurable, dependencies are limited, and the organization knows which work belongs away from the office.
Start with the awkward part. "Productivity" does not mean the same thing in a call center, a product team, an architecture studio, or a small ecommerce brand. Some studies measure calls handled. Others use performance reviews, lines of code, labor productivity, total factor productivity, employee retention, or collaboration networks. A productivity claim is only as good as the metric behind it.
01. The strongest early result came from very measurable work
The classic study is the Ctrip experiment, published in the Quarterly Journal of Economics. Volunteers from a Chinese travel-agency call center were randomly assigned to work either from home or from the office for nine months.
The result was clear: home workers delivered a 13% performance increase. The detail matters. Nine points came mostly from working more minutes, with fewer breaks and sick days. Four points came from more calls per minute, attributed to a quieter and more convenient working environment. Attrition also halved, while promotion rates, conditional on performance, fell.
That finding is useful because it is narrow. It does not prove that every job becomes more productive at home. It shows that individual, measurable work, done by volunteers, can transfer well to the home when the environment is calmer and the output remains visible.
Does the task produce an observable result without constant meetings? If yes, remote work has a real chance of protecting focus. If the result depends on collective decisions or implicit information, the answer changes.
02. The better current case is usually hybrid
A newer experiment, published in Nature in 2024, studied 1,612 graduate employees at Trip.com. Employees in the hybrid group worked from home on Wednesday and Friday and came into the office on the other three days.
After a six-month trial and two years of follow-up, the researchers found no damage to performance grades, promotions, or lines of code written by engineers. They did find that quit rates fell by one-third, especially among non-managers, women, and employees with long commutes.
The lesson is not that "two days" is a universal law. It is more restrained: in this company, with this structure, hybrid work did not damage measured performance and improved retention. Even the managers in the study revised their view of hybrid productivity, moving from a mildly negative expectation before the trial to a mildly positive view after it.
03. In France, the gains look organizational
For French readers, the most relevant recent source is the May 2026 Insee working paper on telework and productivity in France. It does not ask whether employees feel more efficient. It links the share of teleworkers in 2022 to productivity growth between 2019 and 2022, using matched Dares employer survey data and Insee administrative data.
The simple estimates show a positive but moderate relationship: a ten-point increase in the share of teleworkers is associated with a 0.7% to 1.0% relative gain in labor productivity in 2022. With an instrumental-variable strategy, the estimated local effect rises to about 2.7 percentage points of productivity growth for a ten-point increase in teleworkers.
That caveat matters. The gains appear concentrated among firms that were already better positioned to adopt telework, especially because of their pre-existing organization. Remote work mainly amplifies organizations that already make work legible and manageable.
04. The trap: longer days with less output per hour
Not every study points upward. Michael Gibbs, Friederike Mengel, and Christoph Siemroth studied more than 10,000 professionals at a large Asian IT services company during an imposed shift to home working.
The result cuts against the easy story. Hours worked increased, including outside normal working time. Average output changed little or declined slightly. Output per hour fell by 8% to 19%, depending on the measure. The researchers point to a concrete cost: more time in coordination and meetings, and fewer long, uninterrupted blocks for actual work.
That is one of the best cautions in the literature. A packed day is not necessarily a productive day. If remote work turns every question into a call, every approval into waiting time, and every focus block into notifications, it can stretch the day without improving the result. This is exactly what the remote-work well-being station is meant to prevent: a setup that protects long stretches of time instead of dissolving them.
05. What the office still does better
The physical office is more than a place of supervision or attendance. It still helps weak information move: learning who knows what, resolving ambiguity in two minutes, noticing that a project is drifting before it becomes a formal meeting.
A Nature Human Behaviour study using data from 61,182 Microsoft employees points in that direction. Firm-wide remote work made collaboration networks more static and siloed, with fewer bridges between groups. Synchronous communication decreased, while asynchronous communication increased. The authors write that these effects may make it harder for employees to acquire and share new information across the organization.
It is a boundary, not a case against remote work. Deep work can happen very well at home. Work that depends on weak signals, trust formation, learning, or collective judgment often needs better-designed shared time.
06. The right unit is the task, not the location
The macro evidence also argues for caution. The U.S. Bureau of Labor Statistics finds positive associations between the rise of remote work and total factor productivity across several industry-level windows. The Federal Reserve Bank of San Francisco, looking at pre-pandemic industry trends, finds little statistical relationship between telework and pandemic productivity performance after controls.
That disagreement is not a problem for a worker or a small team. It simply shows that location alone explains productivity badly. The better decision starts lower, at task level.
Use home days for tasks that need a quiet block, a visible output, and few dependencies. Use office days for tradeoffs, launches, fast learning, and decisions that benefit from everyone being present at the same time.
07. The home workstation has to make work visible
One mistake is to treat remote work only as permission to choose a place. It is also a scene problem. At home, nobody naturally corrects the work surface, visual noise, lighting, screen height, or the place where documents land. The workstation has to carry more discipline than it does in a shared office.
That discipline does not need to be dramatic. A stable screen, a separate keyboard, a notebook that stays in one place, an active zone that is distinct from the rest of the table: small choices make the task visible. When the day starts, the desk already says what is being done. This is also the whole point of making the order visible, and the reason a full ergonomics guide beats a string of improvised adjustments.
The science does not say that remote work makes people productive by itself. It says something less convenient and more usable: remote work can work very well when the task is clear, the metric is honest, coordination days still exist, and the workstation is not fighting attention.
