Remote Work Has Two Different Problems to Solve
Remote work is not only a question of where an employee sits. It is a way of organizing access: access to colleagues, managers, and the informal knowledge of a firm, but also access to employers and career paths that may lie beyond a worker’s local labor market. This series examines those two forms of access together. It asks what happens when routine co-location disappears, what a small amount of coordinated presence can restore, and whether remote-eligible jobs change who can reach upward mobility. The practical judgment is not whether remote work is inherently good or bad, but which functions the arrangement preserves, which opportunities it opens, and how confidently the evidence supports the conclusion.
Remote work changes access in two directions
Inside an organization, much of the work that makes formal tasks possible is not itself a formal task. A colleague explains a faster way to handle an unusual problem. A manager notices confusion early enough to correct it. Someone learns the unwritten standard for a judgment that cannot be reduced to a procedure. A new employee becomes comfortable asking for help. These interactions build what the EBRD working paper on monthly office days calls organizational capital: the practical ties and shared understanding that support coordination, learning, mentoring, problem solving, and attachment to the firm.
Remote work does not eliminate those needs. It changes the conditions under which they occur. A scheduled video call can replace a scheduled meeting, but it does not automatically replace the brief exchange that would have happened before or after one. A messaging system can preserve a channel while still making it less likely that someone will ask a question, notice a weak signal, or encounter a colleague who knows the answer. The important point is not that face-to-face contact is always superior. It is that a work arrangement can remove some forms of contact without automatically supplying an equivalent mechanism.
The second change occurs outside the firm. A job that can be performed remotely may allow a worker to consider an employer in another metropolitan area without relocating or commuting there every day. In the arXiv study of U.S. job transitions, remote eligibility is treated as a feature of the job structure, inferred from employer postings before a worker enters a role. That is different from asking workers how often they actually worked from home. It is an opportunity measure: does the job, as offered by the employer, make a wider set of geographic matches feasible?
These two changes can point in different directions. A remote-capable job may give someone access to a stronger career ladder while making it harder to build the relationships that support learning after they arrive. Conversely, a firm may invest in better coordination while keeping its hiring market geographically narrow. Remote work therefore has to be examined at two levels. At the team level, the question is whether people can exchange feedback, knowledge, and support. At the labor-market level, the question is whether workers can reach opportunities that were previously blocked by distance, relocation costs, or the concentration of high-growth jobs in a few places.
This distinction also prevents a common measurement error. “Remote work” can describe a location, an employer’s eligibility rule, a hybrid schedule, or a worker’s realized behavior. Those are related but not interchangeable. The monthly-office-day experiment studies a specific change in actual working conditions: employees otherwise worked remotely but met together at the office one fixed day each month. The career study examines entry into jobs whose surrounding postings indicate remote or hybrid eligibility. Neither source measures every possible form of remote work, and neither should be used as if it did.
Two studies, two kinds of evidence
The first source is the randomized trial reported in The Value of One Office Day a Month. It took place in a large customer-service operation in Türkiye. The firm assigned 248 eligible volunteers to remain fully remote or to work together at the office one day each month for nine months. The office day did not change pay, shift length, workload, call routing, production technology, incentives, formal training, or evaluation criteria. Outside that day, treated employees returned to remote work. Because the assignment was randomized and the work was standardized, the study can give a relatively strong answer to a narrow question: what was the effect of this coordinated monthly contact in this setting?
That strength has a boundary. The participants were volunteers who lived within commuting distance of the office and had sufficient tenure to qualify. They performed a standardized customer-service task in stable teams. The result is therefore not a general estimate of “the effect of offices,” nor a proof that every team should meet once a month. It is evidence about a deliberately small coordination intervention among a particular population. Its value lies partly in that narrowness: the study makes it possible to see what changed when most other features of the job did not.
The second source asks a different question at a much larger scale. It links longitudinal career histories to employer job postings for roughly 48 million U.S. job-entry events between 2020 and 2024. A destination job is classified as remote-eligible when the matched postings for the same employer, occupation, country, and location show a majority share of remote or hybrid forms within a three-month pre-entry window. The researchers then compare remote-eligible and fully on-site entries after reweighting them on observable pre-entry characteristics and accounting for metropolitan area and entry-year differences. They also report within-individual comparisons and other robustness checks.
This design provides breadth, not random assignment. Workers and employers are not randomly sorted into remote-eligible jobs. The weighting and fixed-effects strategy makes the comparison more informative, and the robustness results strengthen the case that the pattern is not a simple artifact of stable worker differences. But the study’s estimates remain adjusted associations. The authors themselves note that time-varying factors at the moment of a job transition cannot be fully ruled out, and that eligibility at entry does not reveal how much remote work the employee actually performed later.
The difference matters because the two studies support different kinds of confidence. The trial can more directly attribute a change in productivity or attrition to the assigned monthly office-day policy, within its sample and setting. The career study can reveal a broad pattern across places, employers, occupations, and workers that a small experiment could never observe, but its interpretation depends on measurement and statistical adjustment. Reading them together is useful only if their strengths remain distinct.
The design question is not office or home
A location debate asks which place is best. A design question asks what the arrangement must make possible. That shift is more than a change in vocabulary. It changes what a manager measures, what a worker should evaluate in a job, and what a researcher should count as evidence.
For internal coordination, the relevant outcomes may include more than immediate output. The monthly office-day trial tracks calls per hour, customer ratings, audit scores, communication, manager feedback, and retention. That combination recognizes that a coordination failure can appear first as weaker learning or higher turnover and only later as lower performance. It also avoids treating any single productivity measure as a complete account of organizational health. The study eventually finds higher output and lower attrition, but its mechanism evidence points to several channels—peer interaction, managerial feedback, learning, and a stronger sense of connection—without isolating one as the sole cause.
For external mobility, the relevant outcomes are different. The career study follows wage growth, upward seniority moves, cross-metropolitan transitions, and the characteristics of destination employers. It finds that remote-eligible job entry is associated with higher wage growth and a greater probability of upward seniority movement, while destinations are often smaller and less prestigious than those associated with fully on-site entries. The gains are largest among workers with lower prior earnings and among those coming from regions with fewer high-skill opportunities. This suggests that the value of remote eligibility may be greatest where a local career ladder is thin, not only where remote work is already common.
That finding does not mean remote work automatically equalizes opportunity. Remote-capable jobs are not available in every occupation, and workers who never encounter or cannot pursue such jobs are absent from the transition being measured. Nor does higher career mobility eliminate the need for internal coordination after someone is hired. Access to a job and the ability to develop within it are separate design problems. A worker can cross a geographic boundary without acquiring the feedback, mentorship, or social connection needed to thrive in the new organization.
The central tension is therefore productive rather than ideological. Remote flexibility can reduce the importance of physical proximity in job matching, while also making some forms of organizational support less automatic. Firms that respond only by restoring presence may give up part of the geographic and scheduling flexibility that makes remote work valuable. Firms that respond only by defending flexibility may overlook the interaction infrastructure on which learning, retention, and performance depend. The evidence points toward a more demanding standard: identify the access that is constrained, choose a proportionate way to supply it, and measure whether the intended function actually improves.
The next question is consequently narrow and concrete. In a fully remote team, what exactly did one coordinated office day change? The randomized case provides a way to examine that question without turning it into a referendum on office culture. It begins with the setting, the counterfactual, and the small set of interactions that the intervention was designed to restore.