Return-to-Office Orders: A Survey Analysis of Employment Impacts

How did employers expect return-to-office (RTO) orders to affect employment? Were those expectations correct? We use special questions from the Richmond Fed business surveys to shed light on these questions. Overall, RTO orders were expected to reduce employment, but there was both substantial uncertainty and heterogeneity in expectations. Some employers even expected that RTO would increase employment. Ex post, employers believe RTO orders had a muted effect on employment. We find that the service sector was more likely to both issue RTO orders and expect and experience a reduction in employment.

The COVID-19 pandemic changed the way that both employers and employees think about the location of work.¹ The advent of remote work en masse in 2020 has been followed by a gradual implementation of requiring workers to work from the office, at least for some of their workweek. These forced return-to-office (RTO) orders have come with controversy: Many employers have implemented these policies, while many employees have resisted.

In this article, we attempt to shed light on the effects of RTO by reporting on special questions we asked in the March Richmond Fed business surveys. Specifically, these questions shed light on both the anticipated and realized employment outcomes of RTO orders from the employer's perspective. We find that uncertainty in the decision-making process was prevalent, but also that realized outcomes were generally muted. RTO did have an expected and actual negative effect on employment, but only in some sectors and for some employers. For others, RTO was a means of increasing employment. Our results highlight the large uncertainty in the pandemic, the heterogeneity of firms and the large heterogeneity of workers.

Why Examine the Impacts of RTO Orders?

This survey builds on a recent literature investigating the implications of remote work for workers, businesses and local economies. Uniquely, it attempts to discern how business leaders anticipated RTO policies would impact their firms as well as the actual impact on employment within their firms. Although there is work evaluating the benefits and costs to employers in terms of productivity or labor/non-labor costs,² there has been little work to understand the firm-by-firm implication of articulating and enforcing an RTO order.

Research indicates that hybrid options are highly valued by employees,³ but how many separations can be attributed to an RTO policy? There is evidence that managers value in-person work more than employees,⁴ but does that result in actual separations when RTO orders are implemented? Our results suggest the effects of these policies were muted.

There is also evidence of wide variation in employee hybrid-work preferences and in their willingness to pay for the option to work from home⁵ as well as evidence that the value workers place on the "amenity" of remote or hybrid work has implications for aggregate wage changes in the macroeconomy.⁶ Our work indicates this heterogeneity in preferences may have dampened the effect of RTO orders on employment. Our results are consistent with a literature that is still relatively mixed about the net effect on employers and workers of remote or on-site policies.

Methodology

The Federal Reserve Bank of Richmond has surveyed CEOs and other business leaders across the Fifth Federal Reserve District⁷ for almost 30 years, currently gathering around 200-250 responses per month. The survey panel underweights the smallest firms and, due to the history of the survey, manufacturing firms make up about one-third of respondents even though they make up a much smaller share of establishments in the Fifth District or the nation.

In addition to a series of questions about variables such as demand, employment and prices, respondents are commonly asked a set of ad hoc questions. Here, we focus on a set of questions asked in March 2024 regarding the extent to which respondents articulated and enforced a mandatory RTO policy and what they expected upon its implementation. Emily Corcoran reported on employers' on-site general expectations for employees and how those have changed. But here, we focus on business leaders' expectations of RTO policy effects, providing insight into the anticipated and unanticipated employment effects of RTO orders. We begin by assessing whether the establishment implemented RTO. These results are tabulated in Table 1.

Overall, explicit RTO orders were relatively rare, with only 20 percent of respondents articulating RTO orders in the last three years. This small percentage is partly because 37 percent of respondents — many of them manufacturing firms — were fully on-site before the end of 2020, and an additional 26 percent of respondents said RTO wasn't applicable for their companies.⁸ Of the remaining companies, there is a roughly equal split between firms that have an explicit RTO policy (20 percent of the full sample) and those that do not (16 percent of the full sample).

We asked these 20 percent of employers about the expected consequences of issuing RTO orders. Did they expect workers to quit because of these policies? Were they sure about the effect on employment? We also asked employers about their assessment of realized outcomes. Did workers quit as anticipated? Did RTO help the firm recruit workers?

What Did Employers Expect, and What Actually Happened?

Perhaps surprisingly, we found two-thirds of employers expected no impact on (net) employment from RTO orders, while 16 percent were too unsure of the impact to answer (Table 2). Among the 18 percent that expected some impact, the anticipated outcome was split between those that expected a decrease in employment (11 percent) and those that expected an increase (7 percent).

Why might employment increase? One possibility derives from employees feeling more connected to their co-workers with greater mentoring opportunities when in the office.⁹ This could reduce quitting and improve hiring, as one survey respondent reported that, "...the employees that [formerly] chose to work remotely decided that they were more productive in the office. We are [now] 90+ percent in the office."

Additionally, RTO orders have often been hybrid,¹⁰ potentially allowing the benefits of office culture to be obtained without sacrificing all of the flexibility associated with remote work.

We also asked employers about their evaluation of outcomes, and the results are given in Table 3. Here, a greater percentage reported no impact (82 percent), while 4 percent assessed that RTO had decreased employment, and 4 percent assessed that RTO had actually increased employment. (Nine percent were still unsure.)

Sectoral level analysis reveals employment impacts (both expected and realized) were concentrated in the service sector. In manufacturing, no firms concretely expected a change in employment (though some were unsure), and ex post they believe RTO did not cause them to lose workers. In services, however, only 59 percent expected no impact, while 16 percent expected a negative impact on employment. Ex post, impact on employment was less than expected.

Conclusion

While our analysis is suggestive, there are a few limitations. Foremost, our effective sample size was small, meaning some of these results could be driven by sampling error. Second, it has been years since some employers implemented RTO policies, so their memories of their expectations could be inaccurate. Third, our survey did not control for any other firm changes — such as changes in wages or product demand — that could confound our findings. Fourth, although our findings provide insight into net employment gains and losses, they do not speak to hiring and firing separately.¹¹

With these caveats in mind, however, our results show that RTO — while still a common topic of conversation — is not necessarily important to employers' and workers' employment decisions. Concerns about employment effects ex ante mostly did not materialize. Employment effects that did materialize were concentrated in services and resulted in a net gain of employees in some cases, rather than a loss.

Grey Gordon is a senior economist and Sonya Ravindranath Waddell is a vice president and economist, both in the Research Department of the Federal Reserve Bank of Richmond. The authors thank Jason Kosakow for helping to develop and execute the survey and for providing the tabulations underlying this analysis and thank RC Balaban, Zach Edwards and Claudia Macaluso for providing feedback on an earlier draft.

This article may be photocopied or reprinted in its entirety. Please credit the authors, source, and the Federal Reserve Bank of Richmond and include the italicized statement below.

Views expressed in this article are those of the authors and not necessarily those of the Federal Reserve Bank of Richmond or the Federal Reserve System.