Back to the Office. But Who Can Follow?

The sentence arrives before the practical choice

A return-to-office instruction may alter a contractual expectation, introduce a workplace policy or enforce a requirement that already existed. The label is less important than the documents and implementation: the employment contract, assigned workplace, written policy, notice period, role requirements, exceptions and consequences for non-compliance. The legal effect is jurisdiction- and fact-specific; UK guidance cannot be treated as a universal rule.

The evidence is narrower than the argument

The clearest randomized evidence concerns a specific hybrid arrangement, not a five-day mandate. A six-month trial of 1,612 graduate employees in Trip.com’s Airfare and IT divisions in China assigned people to work from home two days a week. Quit rates fell by one-third, job satisfaction improved, and the researchers found no evidence of lower performance reviews, promotions or lines of code written by engineers.

The study ran from August 2021 to January 2022 at one Chinese technology multinational, mainly among graduate workers in software engineering, marketing, accounting and finance. Retention effects were significant for non-managers and women. For workers with above-median two-way commutes, the reduction did not meet conventional statistical significance; under the stricter definition of a commute exceeding two hours round trip, it did. That distinction matters. “Long commute” is not one uniform statistical category.

The OECD evidence is different. Its COVID-19-era Global Forum on Productivity survey of online respondents across 25 countries reported a preferred arrangement of roughly two to three teleworking days per week, with less commuting and fewer distractions balanced against weaker communication and knowledge flow. That is survey evidence about preferences and perceived trade-offs, not a current universal optimum or a randomized estimate of what every occupation needs.

Employers may have operational reasons for wanting people together. Meta’s March 14, 2023 statement said its preliminary internal analysis associated in-person work with better average performance among some engineers, particularly earlier-career engineers working in person with teammates at least three days a week. Meta also said the analysis required further study. That is evidence of management’s stated rationale, not independent causal proof.

Who pays when the address changes?

The cost is not simply a train ticket. A longer commute consumes time, narrows flexibility and can disrupt school or dependent care. The International Labour Organization estimated that 748 million people aged 15 and over were outside the global labour force because of care responsibilities in 2023, including 708 million women. The estimate drew on data from 125 countries and is not directly comparable with the ILO’s 2018 estimate because of methodology changes. It does not measure return-to-office effects, but it shows the scale of constraints that a location rule can encounter.

Teleworking has its own trade-offs. The ILO describes potential gains in commuting time and flexibility alongside blurred work and family boundaries, intensified hours and difficulty treating care responsibilities as compatible with uninterrupted paid work. A serious policy therefore needs clear hours, workload expectations and support. Neither the office nor the home automatically solves the underlying constraint.

Telework is also unevenly distributed. U.S. Current Population Survey analysis published by the Bureau of Labor Statistics in March 2025 reports differences by occupation, education, disability status and other worker characteristics. The analysis concerns people who were at work during the survey reference week; it is descriptive U.S. evidence, not proof that a particular mandate caused resignations or discrimination. It does, however, make it difficult to treat attendance as a burden shared equally across a workforce.

Exceptions are routes to assessment, not automatic escape clauses

In England, Wales and Scotland, employees can make a statutory flexible-working request from their first day, including a request to change where they work. The framework creates a right to request and a process for handling the request, not an automatic right to work from home. An employer may refuse for a good business reason. GOV.UK expressly says that flexible-working rules are different in Northern Ireland.

The relevant UK statutory change took effect on April 6, 2024, under commencement regulations extending to England and Wales and Scotland. Acas guidance says employers should consider role requirements, contracts, caring responsibilities, consultation and reasonable adjustments when handling home and hybrid arrangements. Those are jurisdiction-specific routes, not universal protections for every worker facing a longer commute.

Disability-related accommodation is a separate question from a general preference for hybrid work. U.S. Equal Employment Opportunity Commission guidance says working from home may be a reasonable accommodation where a disability prevents successful on-site performance and the essential work can be done at home without significant difficulty or expense. The employer may offer another effective accommodation, and the Americans with Disabilities Act does not require a telework programme for everyone.

In the UK, Acas likewise lists home or hybrid working among possible reasonable adjustments where it removes or reduces a disability-related disadvantage. It also cautions against assumptions: its guidance gives the example of a wheelchair user who preferred the workplace because working from home increased isolation. A parent, carer or disabled worker may need a particular arrangement, but no responsible policy should presume which one.

A mandate can change who stays, even without proving why it was made

There is research on what happened after some real mandates. A May 2024 University of Chicago Becker Friedman Institute working paper used 260 million resumes matched to company data and a distributional synthetic-controls method to study Microsoft, Apple and SpaceX. The authors estimated reductions in counterfactual tenure and a shift towards a less senior workforce after return-to-office mandates, with the strongest effects among longer-tenured and more senior employees. They also reported evidence that departures flowed towards larger direct competitors.

That paper is a working paper, not a universal estimate for all employers or occupations. Its cases are three large U.S. technology companies, and its conclusions concern workforce tenure and seniority rather than proof that every departure was caused by the policy or that management intended to force resignations. The result matters because it separates two questions often collapsed into one: whether a rule changes who remains, and whether it was designed to make people leave.

A different picture appears in the Richmond Fed’s March 2024 survey of business leaders in the Fifth Federal Reserve District. Among 45 employers that had articulated a return-to-office policy, 67% expected no net employment impact and 82% said the policy had actually produced no impact. The sample was small, employer-reported and did not separate hiring from firing. The researchers found muted average effects, with more negative outcomes in some service-sector firms.

Taken together, the evidence does not support calling every mandate a covert layoff. Some mandates may coincide with departures or change workforce composition; others may have little measured net employment effect. Evidence that an employer deliberately designed a rule to induce resignations would require something more direct: policy language, contemporaneous executive statements, restructuring records, selective enforcement or documented consequences aimed at particular groups.

Read the policy like an implementation document

A headline such as “back to the office” is incomplete until its machinery is visible. Check the effective date, designated workplace, number of days, required hours, travel expectations, role-based differences, approval route, treatment of existing arrangements, monitoring method and consequences for non-compliance. Also check whether the document preserves approved accommodations, exceptions, collective agreements, mission needs or business-need decisions.

The Federal Aviation Administration’s February 7, 2025 bulletin illustrates the distinction between announcement and machinery. It directed executives to return by February 24, 2025 and other employees, for planning purposes, to return full-time from March 10, subject to accommodations, exceptions, mission or operational needs and applicable policy or collective-bargaining arrangements.

The FAA’s February 24 follow-up then confirmed that executives were reporting full-time from that date, that other employees were to begin full-time in-person work on March 10, and that further guidance would follow for mobile workers, remote workers, employees teleworking as a reasonable accommodation and employees without designated workspace. The example is a U.S. federal-agency policy, not a private-sector template, but it shows why the implementation record matters.

Before deciding that compliance is impossible, a worker can assemble the relevant documents, ask which rule applies to the role and location, request clarification in writing and use the applicable flexible-working or accommodation process. The outcome may still be refusal, relocation, unreimbursed cost, reduced caring capacity or an eventual decision to leave. The useful distinction is between a policy disagreement, a documented request and an enforcement decision.

For employers, the practical test is narrower than “office good, home bad” or the reverse. What work demonstrably requires presence? Which roles differ? Are exceptions real and accessible? Is the rule applied consistently? Are retention, collaboration and performance being measured after implementation rather than assumed in advance? Those are decision questions, not universal legal tests, and the answer will vary with local employment law, collective bargaining and the contract.