Return-to-office mandates are usually announced in the language of outcomes. Leaders talk about productivity, collaboration, culture and the creative energy that supposedly appears when people share a building.

Those benefits may exist in particular teams. Yet when researchers examined mandates across America’s largest listed companies, the broad economic return was remarkably difficult to find.

Yuye Ding and Mark Ma’s study of S&P 500 companies found no significant improvement in profitability or market valuation after return-to-office announcements. Employee ratings, meanwhile, moved clearly downwards. Overall satisfaction, work-life balance, views of senior management and perceptions of corporate culture all declined.

The most measurable result of the policy was not the benefit promised by management. It was the cost experienced by workers.

The researchers built a sample of 137 mandate firms

Ding and Ma searched Google and Factiva for public return-to-office announcements from companies in the S&P 500. They classified 137 companies as treatment firms because each had publicly announced a policy requiring employees to spend at least several days a week in the office.

The team excluded another 43 companies that had no public mandate but where Indeed data suggested more than 70 per cent of employees had no work-from-home option. That choice reduced the risk of placing quiet mandate firms in the control group.

They combined the policy dates with employee reviews from Glassdoor, financial data from Compustat and market information from Thomson Reuters. Their main tool was a difference-in-differences design, which asks whether the change after a mandate differs from the change over the same period among comparison firms.

This is much stronger than looking at one company before and after a policy. It still is not a randomised experiment. Companies chose whether and when to issue mandates, and those choices may be related to other forces affecting performance and morale.

The promised financial lift did not appear

The researchers tested profitability using return on assets and assessed market value using Tobin’s Q, a standard ratio comparing a company’s market valuation with the replacement cost of its assets.

After controlling for firm characteristics and time effects, the coefficients associated with RTO mandates were not statistically significant for either measure. Put plainly, mandate firms did not show a detectable improvement in profitability or firm value relative to non-mandate companies.

That does not prove the effect was exactly zero. A statistically insignificant result can reflect an effect too small or noisy for the study to detect. It does mean the data did not support a confident claim that mandates improved the economic outcomes managers cited.

There is another important boundary: the researchers did not have a direct measure of individual employee productivity. Firm-level profit and market value are influenced by sales, prices, investment, competition and macroeconomic conditions. The study tested whether any productivity benefit became visible in the broader outcomes companies care about. It did not time keystrokes or compare each worker’s output at home and in the office.

Employee ratings did change significantly

The contrast was clearest in Glassdoor reviews. Following a mandate, a company’s average overall employee rating fell by 0.057 points on a five-point scale. Ratings of work-life balance, senior management and culture also declined significantly.

Each change was modest in absolute size. Across a large sample, however, the direction was consistent enough to distinguish from ordinary variation. The researchers also checked compensation-and-benefits and diversity-and-inclusion ratings, which were less directly connected to work location. Those measures did not show the same significant shift, weakening the idea that reviewers had simply become more negative about everything.

A review platform is not a random sample of every employee. People choose whether to post, and a disliked mandate may prompt unhappy workers to speak. The authors tested pre-policy trends and some alternative explanations, but self-selection cannot be made to disappear completely.

Still, employee dissatisfaction was measurable where the financial gain was not. That asymmetry deserves more attention than it usually receives.

The evidence does not say offices are useless

The useful comparison is not office good, home bad, or the reverse. Different jobs contain different mixtures of focused work, mentoring, customer contact, equipment use and creative coordination. An RTO mandate is a policy about who decides the mixture.

A six-month randomised trial involving 1,612 Trip.com employees illustrates the distinction. Workers assigned the option to work from home on Wednesday and Friday had higher satisfaction and one-third lower attrition than colleagues in the office five days a week. Performance reviews, promotions and lines of code showed no evidence of damage from the hybrid schedule.

That study covered graduate employees at one Chinese technology company, so it is not a universal template either. Together, the two projects suggest that flexibility can preserve performance while improving outcomes workers value. They do not establish that every team should work remotely.

I recently wrote about how visible busyness can be confused with useful output. Office attendance risks becoming another visibility measure that feels concrete because it is easy to count.

Later research found a cost in lost talent

Employee ratings are not merely soft sentiment if they influence who stays. A later study of 54 technology and financial companies used more than three million employee profiles to examine turnover after RTO announcements.

The researchers reported a 13 to 14 per cent increase in abnormal turnover, with larger effects among women, senior managers and highly skilled staff. Vacancy duration rose from 51 to 63 days on average, while hiring rates fell. The Baylor University summary describes a labour-market penalty that ordinary quarterly profit measures may take time to reveal.

This is still observational evidence. Companies changing location policies may simultaneously restructure, change leadership or respond to weakening demand. But it gives the satisfaction finding a plausible business consequence: people with valuable outside options can act on their dislike.

A mandate should have to pass its own performance review

Ding and Ma acknowledged important limits. Their window was not long enough to assess distant financial consequences. The sample covered unusually large American companies, not small firms or public services. The post-pandemic labour shortage may also have shaped employee reactions, and the design could not rule out every source of endogeneity.

Those caveats are reasons to measure better, not permission to rely on intuition. If a company removes flexibility to improve collaboration, it can define the collaboration problem in advance. If the target is mentoring, it can measure development and promotion. If the target is innovation, it can track project quality and speed. If the promised gain never appears while retention and satisfaction deteriorate, the policy is failing on its own terms.

There may be sound reasons for certain people to meet in person at certain times. The research challenges a blunter proposition: that compelling everyone back is self-justifying because an office is where serious work visibly happens.

Presence is easy to observe. Value is harder. A serious workplace policy should know the difference.