SSRN Id4675401
SSRN Id4675401
SSRN Id4675401
Yuye Ding
[email protected]
Katz Graduate School of Business
University of Pittsburgh
Abstract
Using a sample of Standard and Poor’s 500 firms, we examine determinants and consequences of
U.S. firms’ return-to-office (RTO) mandates. Results of our determinant analyses are consistent
with managers using RTO mandates to reassert control over employees and blame employees as a
scapegoat for bad firm performance. Also, our findings do not support the argument that managers
impose mandates because they believe RTO increases firm values. Further, our difference in
differences tests report significant declines in employees’ job satisfactions mandates but no
significant changes in financial performance or firm values after RTO mandates. In summary, our
research contributes to the ongoing debate over RTO versus working from home and has important
implications for practitioners.
Keywords: Employee Satisfaction, Firm Performance, Return to Office, Work from Home, and
Work-Life Balance.
*Corresponding author
1. Introduction
Since early 2020, the COVID-19 pandemic has dramatically reshaped the landscape of work.
Many individuals transitioned to remote work to mitigate the risk of virus transmission in crowded
office environments. This shift also granted workers greater flexibility and eliminated the need for
time-consuming commutes (e.g., Business Horizons, 2023). After the threat of COVID-19 began
to recede since mid-2022, many companies reconsidered their workplace policies. While many
firms, such as Nvidia and Airbnb, are sticking to the working from home (WFH) mode, a lot of
others, such as Amazon and Disney, have implemented or are considering return-to-office (RTO)
mandates (CNBC, 2020; Fortune, 2023a; CNBC, 2023a).1 Numerous company top managers argue
that working from home reduces employee productivity and hurts firm performance and values
(Fox Business, 2023). However, a large number of employees vehemently oppose this viewpoint
and argue that the elimination of arduous commutes and enhanced flexibility actually contribute
to higher work efficiency and better overall well-being (Fonner & Roloff, 2010; Chatterjee et al.,
2020; Putri & Amran, 2021; BBC 2023). Many experts believe that the managers use RTO
mandates to regain control over employees and blame employees as a scapegoat for bad firm
1
By RTO mandates, we refer to firms’ requirements that employees must work in the office for at least several days
in a week. Unfortunately, most firms with RTO mandates in our sample do not provide details about how many days
their employees are required to work in the office. Therefore, we do not further distinguish RTO mandates based on
the number of days in the office required by a firm’s mandate.
1
The discord between employees and managers regarding the necessity of RTO reached a
boiling point recently and led to significant conflicts within firms, which garnered significant
social attention. For example, over 5,000 Amazon employees signing a petition in 2023 to express
their opposition to CEO Andy Jassy's mandate for most employees to return to the office (CNBC,
2023b). As the tension caused by RTO mandates has become evident, many CEOs are expressing
regret over their RTO mandates. A recent survey reveals that a large number of managers admit
that, with a more comprehensive understanding of workplace data, they would have pursued
significantly different strategies for their return-to-office requirements (Fortune, 2023b). Further,
nearly a quarter of managers acknowledge that they made RTO decisions primarily based on
intuition rather than facts, leading to employee resentment and disillusionment. This anecdotal
evidence underscores the complexities of postpandemic work arrangements and the importance of
understanding firms’ RTO mandates. In this study, we empirically examine determinants and
Prior research has investigated the pros and cons of remote working prior to or in the early
stage of the COVID-19 pandemic and provided very mixed evidence on the efficiency of WFH vs.
working in the office. However, these studies have limited implications for understanding RTO
mandates for several reasons.2 First, as discussed above, remote work was limited to a very select
group of workers before the pandemic. However, many more employees with more diverse
backgrounds have WFH experiences since the pandemic. Second, employees who have never
worked from home before the pandemic may not fully understand the benefits and costs of WFH.
Thus, they are less likely to have negative reactions to working in the office, which is the traditional
workplace model. Even if some employees are more productive in the office before the pandemic,
2
Most RTO mandates were announced after the threat of the pandemic significantly decreases from mid-2022.
However, most prior studies on WFH using data from the period before 2022. Please see section 2 for details.
2
it does not necessarily mean that these employees would become more productive when they return
to office after the pandemic. This is because they are more likely to react negatively to being forced
to work in the office after enjoying the benefits of WFH during the pandemic. Third, during the
early stage of the pandemic, employees’ performance may be negatively affected by isolations
(due to social distancing policies) and day care responsibilities. Also, employees needed to spend
significant time adjusting to remote work and learning necessary technologies. This adjustment
process could also negatively affect their performance. Such constraints and negative effects are
less likely to exist after the pandemic. Therefore, it is important to empirically examine RTO
We manually identify Standard and Poor’s (S&P) 500 firms with RTO mandates based on
news search.3 Our empirical analyses begin with investigating determinants of RTO mandates.
Specifically, we consider three possible explanations for RTO mandates. First, managers suggest
that their RTO mandates aim to improve employee productivity, firm performance and ultimately
firm values. Under this explanation, we would expect a higher probability of RTO mandates when
the manager has stronger incentives to maximize shareholder values, such as when the manger’s
wealth is more closely tied to firm values or when the firm has higher institutional ownership. 4
Also, if managers indeed believe RTO improves performance and firm values, the probability of
RTO mandates is expected to be higher for firms with poor prior performance. Second, RTO
mandates may be used by managers to blame employees’ low productivity rather than poor
management decisions as the reason for poor firm performance (e.g., Business Insider 2023a). This
explanation would also expect a higher probability of RTO mandates for firms with poor
3
We focus on S&P 500 firms because these large firms have more media coverage and their RTO mandates are thus
less likely to be omitted in news reports.
4
Prior research suggests that institutional investors are found to improve corporate governance and encourage
managers to adopt more value-enhancing policies in prior literature (e.g., Chung and Zhang 2011).
3
performance. However, firms with higher institutional ownership are expected to have a lower
probability of RTO mandates because institutional investors are more likely to see through
managers’ “blame game”. Third, many employees believe that managers lack trust in employees
and worry about losing their control over employees who work remotely from home rather than
physically in the office (e.g., BBC, 2023; Davey, 2023; Business Insider 2023d). Thus, managers
use RTO mandates to reassert control over their workers and grab power back from employees.
This argument predicts a higher likelihood of RTO mandates for managers who like to hold power
over others.
Our empirical analyses find that the probability of RTO mandates is higher for firms with
poor prior stock market performance. However, institutional ownership significantly decreases the
probability of RTO, and CEO stock ownership does not have a significant effect on RTO mandates.
Further, the probability of RTO mandates is significantly higher for firms with male and powerful
CEOs, who are more likely to grab power back from employees through RTO (Cragun et al., 2020;
Business Insider, 2023a).5 Overall, our results do not support the argument that managers impose
these mandates to increase firm values. Instead, these findings are consistent with managers using
RTO mandates to reassert control over employees and blame employees as a scapegoat for bad
firm performance.
Next, we examine the impact of RTO mandates on employees, one key stakeholder in the
debate. We collect employee job satisfaction data from Glassdoor and focus on employees’ ratings
of overall job satisfaction, work-life balance and senior management because these three ratings
are most likely to be influenced by RTO mandates. Many employees suggest that daily commutes
5
In addition, we also consider several other firm-level attributes, CEO characteristics, industry-level factors, and
local transportation conditions as potential determinants for RTO mandates. We find that the probability of RTO is
higher for firms in high-tech industries but lower for firms in industries with high competition and firms
headquartered in areas with longer commuting times.
4
and decreased flexibility lead to worse work-life balance (e.g., Chatterjee et al., 2020; Ipsen et al.,
2021; Business Insider, 2023b). Also, as discussed above, RTO is viewed as a signal of
management power grabbing and lack of trust in employees (Business Insider, 2023a; Davey,
2023). As a result, employees may have worse perceptions of senior management. Nonetheless, it
is also possible that a significant portion of employees may agree with managers’ RTO decisions
and believe working in the office can enhance collaboration (Palumbo 2020; Business Horizons,
2023; Gibbs et al., 2021). Further, WFH may blur the separation between work and home and
reduce work-life balance. Thus, employees may become more satisfied when they return to office.
employees’ ratings of overall job satisfaction, work-life balance and senior management after a
firm announced an RTO mandate. Also, we show that employees’ other ratings that are not closely
related to RTO do not significantly change. Therefore, the negative impact of RTO mandates on
employee satisfaction is not simply due to employees who posted reviews after the RTO mandates
shareholders. Specifically, we test how financial performance and firm values are affected by RTO
mandates.6 On the one hand, as discussed earlier, many managers suggest that RTO would improve
employee productivity and firm performance. On the other hand, as employee satisfaction is
negatively affected by RTO mandates, their productivity may become lower. Also, RTO may
increase firms’ spending for leasing offices and other operational costs (Business Insider, 2023c).
6
Many managers’ arguments are related to employee productivity. Unfortunately, we do not have a good measure of
employee productivity in our sample. We focus on firm performance and firm values because higher employee
productivity is ultimately expected to translate into better firm performance and the ultimate goal of managers is to
improve firm values.
5
Therefore, ex ante, it is less clear how RTO mandates affect firm performance and values. Our
DiD regressions find that firms with mandatory RTO plans do not experience significant changes
in profitability and market values relative to non-RTO firms. These findings do not support that
managers' argument for RTO mandates are based on firm performance or shareholder values.
In conclusion, our study delivers a timely analysis of U.S. firms’ RTO mandates. Our
findings are consistent with employees’ concerns that managers use RTO for power grabbing and
blaming employees for poor performance. We provide evidence that RTO mandates hurt employee
satisfaction but do not improve firm performance. We believe our empirical evidence can better
assist managers and shareholders in assessing the value of adopting an RTO mandate and offers
guidance for firms in crafting effective workplace policies after the pandemic.
Our study makes several important contributions to the economics and management
literature. First, prior research examined how remote work affects employee satisfaction and
performance before the pandemic and provided mixed findings (see section 2 for details). We add
to the literature by offering empirical evidence that employees react negatively when they are
required to return to the office postpandemic. Second, while prior research has investigated
Our study provides empirical evidence on how other firm-level attributes, CEO characteristics,
industry-level factors, and local transportation conditions affects RTO mandates. Third, we add to
the literature on the role of CEOs in determining corporate policies. Prior research shows that
CEOs’ seek for power affect many different corporate policies, such as corporate social
responsibility (CSR) focus, disclosure practice, and investment (Daily & Johnson, 1997; Pathan,
2009; Friedman, 2014; Li et al., 2018). We further show that CEOs’ preference to hold power over
employees have significant impacts on RTO mandates and, more generally, workplace flexibility.
6
2. Background and Literature Review
2.1 Background
The COVID-19 pandemic posed significant challenges to traditional business models and
distancing and curb the virus's spread, the percentage of the U.S. workforce working exclusively
from home surged from 8.2 percent in February 2020 to over 35 percent in May 2020 (Bick et al.,
2022). Even before the pandemic, the digital transformation was already a hot topic. But
organizations were generally moving slowly on the trend. The pandemic accelerated the adoption
Following the wane of the pandemic's impact, the proportion of the U.S. workforce
exclusively working from home began to decline. A considerable segment of employees was
gradually returning to work on-site (Bick et al., 2022; Business Horizons, 2023). According to a
survey by the Chief Executive Group, there was a notable increase in the proportion of firms
requiring their employees to be fully present on-site, with 31 percent of U.S. companies conducting
on-site operations in May 2022, escalating to nearly 50 percent in 2023 (Nolen, 2023). This
transition also resulted in a decrease in the number of companies embracing hybrid work models,
declining from 61 percent in 2022 to 48 percent in 2023. However, many firms, including Nvidia
and Airbnb, stick to WFH and commit to a workplace model without mandatory office attendance.
Many employees hold a favorable view of remote work. As highlighted in a Wall Street
Journal report, employees refuse to go back to office because of terrible commutes; escalated
expenses of commutes, meals, and childcare; and lost productivity (WSJ, 2023). However,
managers hold different views about WFH. Both Citi CEO Michael Corbat and JPMorgan Chase
CEO Jamie Dimon have expressed worries that WFH could hurt employee productivity in the long
7
run (Bloomberg, 2020; WSJ, 2021). Some managers are also in favor of WFH. For example, Beau
Davidson, Nvidia’s vice president of employee experience, suggests that WFH serves “as a way
for employees to balance their personal and work obligations, while preparing for the future, so
Forcing employees who prefer to work from home to return to the office could exacerbate
conflicts between managers and employees and result in job dissatisfaction among the workforces.
For example, Amazon workers protested the firm’s return-to-office mandate in June 2023. In
response, Amazon CEO Andy Jassy insisted that employees should return to the office and warned
employees that “it’s probably not going to work out for you” if employees ignored a return-to-
office mandate (Fortune, 2023a). Notably, a recent survey conducted by Envoy has garnered
significant media attention. According to this survey, Envoy found that 80 percent of employers
who hastily decided to bring their employees back to the office regretted their choices. 7 Those
companies that insisted on an early return to the office have reported low employee morale, a
decline in their hiring rates, and a loss of top talent (Barrero et al., 2021; Business Horizons, 2023).
Remote work has become a prominent subject of discussion over the past several decades
thanks to the technological revolutions. Prior literature has examined the impacts of working from
home on employee productivity, work-life balance, and job satisfaction. Before the pandemic,
remote work was primarily limited to a small percentage of individuals, often male professionals
with higher education and income levels, or female clerical workers (Felstead et al., 2002; Bailey
& Kurland, 2002). Early research studies generally documented an increase in productivity and
7
See “Without accurate data, the physical workplace won’t survive“
https://assets-global.website-
files.com/6509fe179d7033a278a05268/652cb5d2e4e95fb82f8da955_Workplace_Data_Report_Final-Document-
Envoy-230808-1.pdf
8
job satisfaction among remote workers before the COVID pandemic by using self-reported survey
data (DuBrin, 1991; Hartman et al., 1992; Apgar, 1998; Fonner & Roloff, 2010; Brittany &
MacDonnell, 2012; Church, 2015; Gajendran et al., 2015; Kröll & Nüesch, 2019; Schall, 2019).
Kröll and Nüesch (2019) utilized panel survey data reported to the GSOEP, a platform collecting
information about employee job satisfaction and leisure satisfaction in Germany. They
employees when flexible work practices were offered. However, Bailey and Kurland (2002)
pointed out that the percentage of employees reporting increased productivity and satisfaction in
the abovementioned research closely aligns with the percentage of employees who volunteered to
work at home, raising the concern that these voluntary remote workers may be biased to claim
success of WFH.
Several studies further use natural or controlled experiments to identify causal effects of
WFH and workplace flexibility. Bloom et al. (2015) randomly assign call center employees at
Ctrip, a NASDAQ-listed Chinese travel agency, either to work from home or in the office for a
nine-month period. Those who WFH reported a 13% enhancement in performance and increased
work satisfaction. Choudhury et al. (2021) examine an even more flexible workplace arrangement,
work from anywhere (WFA). In Bloom et al. (2015), workers who work from home are still located
close to the company. WFA further offer even greater geographic flexibility that employees can
live in any location that they like. Choudhury et al. (2021) reported an approximate 4.4 percent
increase in output (number of patents examined) when employees of by the United States Patent
and Trademark Office (USPTO) have the flexibility to work from any location. Angelici et al.
(2020) further look at “smart-working”, which allow employees to choose not only their locations
but also work time schedule. By randomly assigning workers in a large Italian company to either
9
“smart-working” or traditional work modes, Angelici et al. (2020) also observed an increase in
worker productivity, well-being, and work-life balance in the flexible work mode group.
Studies above generally support WFH. However, a few studies express a contrasting view.
Using data from the 2010, 2012, and 2013 American Time Use Survey Well-Being Modules, Song
and Gao (2020) find that working from home significantly decreases employee happiness and
elevates stress levels. Also, the effect of WFH on well-being varies by parental status and gender.
Parents and females are more likely to be affected by working at home. Additionally, Barber et al.,
(2019) suggest that workplace telepressure could negate the benefits of remote work. Song and
Gao’s (2020) study was conducted before the pandemic when WFH was not the norm. In their
sample, employees still primarily work in the office and only occasionally work from home due
to day care and other responsibilities at home, which could result in employees’ dissatisfaction
with WFH. Thus, their findings may not be applicable to more general settings after the pandemic.
The pandemic triggered a surge in research studies examining the impact of WFH on
productivity, job satisfaction, and work-life balance under the new working mode during the early
stage of the pandemic. Their findings are also mixed. On the one hand, many survey studies
document benefits of WFH. Drawing on survey data from over 5,700 workers who WFH during
the initial phases of lockdown, Ipsen et al. (2021) documented that employees rated the advantages
of WFH higher than its disadvantages and that rank-and-file employees rate WFH higher compared
with how their managers do. Similarly, using survey data, WFH was documented to improve
employee satisfaction in Portugal, Romania, and Indonesia (Davidescu et al., 2020; Petcu et al.,
2021; Sousa-Uva et al., 2021; Irawanto et al., 2021). A few studies also suggest an improvement
10
the United Kingdom from the early months of the pandemic and before the pandemic, Etheridge
et al. (2020) found that workers in industries and occupations deemed suitable for remote work
The positive correlation between WFH and employee satisfaction and performance during
the pandemic can be attributed to various factors, including enhanced work-life balance, improved
work efficiency, and increased work autonomy (Ipsen et al., 2021; Nugraha et al., 2022; Schade et
al., 2021). The substantial time saved from commuting also significantly contributes to the positive
outlook on WFH (Smite et al., 2023). Using survey data, Giménez-Nadal et al. (2020) found that
remote workers reported a 40 percent decrease in commuting time. Commuting often links to lower
mood due to stress induced by factors like congestion, crowding, and unpredictability, and these
negative experiences can spill over into people's work (Chatterjee et al., 2020). Therefore, the
saved time on commute can be redirected to leisure and relaxation and significantly improve
Nevertheless, some other studies which also use survey data during the pandemic found
that remote work could have adverse effects on employee work-life balance due to increased work-
related fatigue, blurred boundaries between work and personal life, and family dynamics (Palumbo
2020; Barriga et al., 2021; Bellmann & Hübler 2021; Möhring et al., 2021; Palumbo et al., 2021).
This issue can be especially pertinent for professional women with children (Bahn et al., 2020;
Möhring et al., 2021; Hjálmsdóttir & Bjarnadóttir, 2021; Pathak, 2021; Matthews et al., 2022;
Kotini-Shah et al., 2022). Concerns have emerged regarding the potential adverse effects of
working from home on employee health and well-being. While WFH holds the potential to foster
a healthier lifestyle, it may also contribute to increased sedentary behavior, and the impact of
11
remote work on mental health hinges on individual and social factors (De Sio et al., 2021; Furuya
et al., 2022).
A few studies further provide mixed empirical evidence on employee productivity and
performance when working from home during the early stage of the pandemic. Gibbs et al. (2021)
look at an Indian IT services company HCL Technologies, which moved abruptly to WFH in
March 2020. Utilizing personnel and analytics data from over 10,000 skilled professionals before
August 2020, Gibbs et al. (2021) documented that while employees do not have lower overall
output after working from home, they report a roughly 10 percent increase in working hours,
suggesting a productivity decline ranging from 8 percent to 19 percent. Their study suggests that
the decline is likely due to increased time spent on coordination activities among employees
working from home during the pandemic. They also pointed out that other confounding factors,
such as day care responsibilities, could also contribute to this decline in productivity. Similar
declines in productivity during the early stage of the Pandemic were also reported in Japan, but
there was significant variation in the effect of WFH on productivity (Morikawa, 2022).
Specifically, highly educated, high‐wage employees, long‐distance commuters, and those who
work in the information and communications industry exhibited smaller reductions in productivity.
Further, Yang et al. (2022) analyzed data of Microsoft employees over the first six months of 2020
and found that firm-wide remote work caused the collaboration network to become more static
and siloed.
However, examining IT workers in one of China’s largest IT firms, Baidu, Bao et al. (2022)
identified both positive and negative impacts of remote work on developer productivity across
various metrics, including the number of builds, commits, and code reviews. Bao et al. (2022)
noted that the effects of WFH on employee productivity varied among projects and individual
12
developers. Harrington and Emanuel (2023) analyze the performance of call center workers in a
U.S. Fortune 500 firm. They found that remote workers have lower performance than on-site
workers before the pandemic. However, during the early stage of the pandemic, already remote
workers increased their productivity relative to those who are formerly on-site. They suggest that
the lower productivity by remote workers before the pandemic is mainly due to negative worker
selection into remote work rather than WFH itself. Using data from an online survey, Galanti et
al., (2021) find that individual factors, such as family-work conflicts, social isolation, and a
We collect RTO information for S&P 500 firms through manual news searches on Google
and Factiva. Specifically, for each firm, we search by using the company’s name and the following
keywords: ‘Work from Home’, ‘return to office’, and ‘days required working in the office’. We
focus on S&P 500 firms because these large firms are closely followed by media, thus mitigating
concerns about media coverage bias. Among the S&P 500 firms, 137 publicly announce their RTO
policy and are classified as RTO firms in this study. These 137 firms become our treatment firms.
For these firms, we further collect data on announcement dates and effective dates of their RTO
mandate based on news reports. In cases where news reports do not provide an accurate
announcement date and use vague phrases such as 'earlier this month' or 'earlier this quarter,' we
consider the first day of the mentioned period as the announcement date. If the announcement date
is not available in news reports, we use the date of the earliest news report as the announcement
13
To mitigate concerns that some RTO firms may not publicly disclose their RTO policies,
we manually collect information about the percentage of employees working from home for each
firm on Indeed.com. We identify 43 S&P 500 firms that do not publicly announce a mandatory
RTO policy but have more than 70 percent of employees reporting no WFH option on Indeed. We
exclude these 43 firms from our sample. Remaining S&P 500 firms are designated as control firms.
We also manually collect employee ratings and comments about the pros and cons of their
information, institutional ownership, and industry-related factors are sourced from Compustat,
from the Federal Election Commission (FEC) website. County-level commute time is obtained
from the United States Census Bureau using the county subdivision–level 2021 American
To investigate the factors influencing the RTO decision, we analyze a sample of 4,455
firm-quarter observations of the 457 S&P 500 firms (after excluding the 43 firms lacking public
RTO news but exhibiting high 'NO WFH' rates on Indeed). The sample period is from June 2019
and January 2023.9 We use the following Probit regression Model (1) to assess the determinants
of RTO mandates.
8
See United States Census Bureau, American Community Survey, “Travel Time to Work,”
https://data.census.gov/table/ACSDT5YSPT2021.B08303?q=Travel+Time+to+Work+Indicators+for+&g=010XX0
0US$0500000
9
We have restricted our sample period of this determinant analysis to January 2023, because we don’t have CEO’s
compensation data to calculate their relative power measure in year 2023.
14
Our dependent variable, 𝑅𝑇𝑂, is a binary variable, taking the value of 1 if the firm-quarter
observation of an RTO firm is from the period after the RTO announcement (including the
announcement quarter), and 0 for observations of the firm before the RTO announcement or firms
that do not have RTO mandate. We include a variety of firm-level, CEO-level, industry-level, and
region-level factors in independent variables. The first set of controls are a number of firm
attributes, including firm size (Size_Lag), financial constraints (KZ_Lag), change in ROA
All these variables except RET_Lag are measured at the end of the prior quarter. Following prior
studies (Green et al., 2019), RET_Lag is calculated as cumulative returns from quarter q-3 to q-2,
where q is the current quarter. We do not include quarter q-1 in the calculation of this measure to
avoid any effect of information leak related to RTO before the official announcement on the stock
We expect that a larger firm is more likely to have RTO mandates, because it is likely for
these firms to coordinate among employees who work from home. More financially constrained
firms may be less likely to have an RTO mandate as working in the office may increase the firm’s
operational budget, such as spending on an office lease (Reuters, 2023). As discussed earlier, prior
poor financial or stock market performance could increase the pressure for managers to bring
workers back to the office for two reasons (Garmaise, 2008). First, managers may believe that
employees working from home have lower productivity, which hurts firm performance. Second,
managers may blame employees as a scapegoat for bad firm performance. Further, institutional
ownership is expected to increase the pressure to return to the office if RTO indeed increases
shareholder values, as institutional investors are more sophisticated and can better understand the
15
implications of WFH for firm values (Sundaramurthy, 1999; Hao, 2014; Chen et al., 2020).
However, if managers use RTO to blame employees for poor performance, institutional ownership
Our second set of dependent variables are CEO characteristics, including power, ownership,
gender, and political affiliation, which are found to have a significant influence on firm decision-
making in prior literature (Daily & Johnson, 1997; Friedman, 2014; Ho et al., 2015; Goel & Thakor,
2008; Arikan et al., 2023). CEO_Gender denotes the gender of the CEO, taking a value of 1 if the
CEO is male and 0 if female. CEO_Power is measured as CEO’s total compensation divided by
the average total compensations paid to the four highest-paid executives in the firm (Frydman &
Jenter, 2010). A higher value of CEO_Power indicates that the CEO is more powerful inside the
firm. CEO_SHROWN represents the percentage of firm shares owned by the CEO. CEO_Political
is calculated as the difference in individual contributions to political parties divided by the total
individual political contributions. CEO Political takes a positive value 1 if the CEO's political
contribution leans more towards the Republican Party and 0 if the CEO’s contribution leans more
towards the Democratic Party. These variables are measured annually, as they are more stable in
As discussed above, employees believe that managers use RTO to reassert controls over
employees. If so, we expect male and more powerful CEOs to have more RTO mandates. Prior
studies suggest that male CEOs are more likely to seek power for themselves (Ho et al., 2015;
Cragun et al., 2020). Similarly, more powerful CEOs are likely concerned about losing their power
to employees and thus are more likely to use RTO to grab power back from employees. On the
other hand, if managers impose RTO mandates due to concerns that WFH hurts firm values, CEO
ownership is expected to increase the probability of RTO mandates. That is, when CEOs’ wealth
16
is more closely tied to firm values, the CEOs would be more concerned about negative impacts of
WFH on firm performance. Further, Republican-leaning CEOs often have more strict employee
treatment and thus are more likely to have RTO mandates (Bayat et al., 2021; Weng & Yang,
2023).
The third set of determinants are two industry-level factors. High_tech is an indicator for
firms in high-tech industries. Following Kile and Phillips (2009), the following Standard Industrial
Classification (SIC) codes are used to classify an industry as high-tech: 283, 357, 366, 367, 382,
384, 481, 482, 489, 737, and 837. The high-tech industry is less likely to have RTO mandates for
two reasons. First, employees in high-tech firms are better trained and equipped with working from
home, as they are more familiar with advanced technologies (Bai et al., 2021). Second, these firms
are likely more concerned that a mandatory RTO policy could lead to the loss of highly intelligent
workers, which plays a central role in their business models. HHI (Herfindahl-Hirschman Index)
is computed as a measure of sales concentration within the 4-digit SIC industry. A higher HHI
index indicates greater concentration in industry sales and lower industry-level competition. Firms
in more competitive industries are more likely to have RTO mandates if WFH hurts firm
performance.
Finally, we consider the commute time that workers need to spend in the firm’s
headquarters location. Commute_Time is collected from the 2021 ACS 5-Year Estimates of travel
time data. We first identify the county where the company's headquarters is located.10 Next, we
define Commute_Time as 1 if the county ranks within the top 25 percent in terms of commute time,
and as 0 otherwise. Commute is a significant factor to consider when assessing the merits and
10
In cases where a firm is located in a city that spans multiple counties, we calculate the firm commute time by
considering the commute time within each county the firm city spans, weighted by the percentage of the population
in that county.
17
drawbacks of remote work (Giménez-Nadal et al., 2020; Smite et al., 2023). Given that time saved
on commuting is a key element contributing to the benefits of remote work, we anticipate that
companies in more congested areas will be less inclined to implement an RTO policy.
We next examine the effect of RTO mandates to two important stakeholders in the debate,
employees and shareholders. To analyze the effect of RTO mandates on employee satisfaction, we
manually collect data on Glassdoor employee ratings of our sample firms from January 1, 2020,
to November 30, 2023. The dependent variable is one of the firm-quarter average of employee
ratings on overall job satisfaction (Overall Satisfaction), work-life balance (Work-Life Balance),
or senior management (Senior Management). These ratings are made on a scale from 1 to 5, with
a higher value indicating higher employee satisfaction. To ensure the robustness of our findings,
we excluded firm-quarters with fewer than 5 employee reviews.11 Consequently, our analysis is
based on a sample comprising 5,053 firm-quarter observations for tests with overall job
satisfaction ratings, along with 5,051 firm-quarter observations for tests with ratings for work-life
balance and senior management. We use these samples to estimate the following DiD regression
The key independent variable is RTO, which is an indicator that is set to 1 for firm-quarter
observations after a firm’s RTO announcement (including the announcement quarter) and 0 for
11
We also use 10 reviews or 15 reviews as the minimum requirements and find similar results.
12
Baker, Larcker & Wang (2022) suggest that TWFE estimators could be biased in certain sample. However, since
we have a large enough controlled sample that accounts for 70% of the whole sample, the biases associated with our
TWFE staggered DiD are not problematic (Baker et al., 2022). In addition, in untabulated tests, we adopt an
alternative DiD estimator from Callaway and Sant'Anna (2021), which is recommended by Baker et al. (2022) and
still find a significant decline in employee job satisfaction.
18
observations of the firm before the RTO announcement or firms that do not have RTO mandate.13
The coefficient on the RTO shows the change in employee ratings of RTO firms after the mandate
capture firm characteristics relevant to employee review rates, including firm size (Size_Lag),
return on assets (ROA_Lag), Tobin’s Q (Tobin_Lag), financial constraints (KZ_Lag), stock returns
(RET_Lag), and industry concentration (HHI_Lag) (Green et al., 2019; Farhadi & Nanda, 2021;
Dube & Zhu, 2021). For each observation, we calculate these control variables using data from the
prior quarter. In addition, we control for firm and quarter fixed effects to control for time-invariant
factors or other time-variant common shocks that affect both treatment and control firms.
We further use the following DiD regression models to empirically analyze the effect of RTO
mandates on firm performance and shareholder values. The sample consists of 6,026 firm-quarter
The dependent variables in Models (3) and (4) are quarterly return on assets (ROA) and
Tobin’s q ratio (Tobin Q) at the end of the quarter, respectively. High values of both measures
13
We use RTO announcement date instead of effective date because employees are very likely to start expressing
their dissatisfaction once they are noticed of the RTO mandates. Some firms informed their employees of the RTO
mandates several quarters ahead, for example, NetApp Inc announced their RTO plan in Aug 2020, but their RTO
effective date was in July 2021. By that time, we think employees have already expressed their negative comments.
19
suggest better firm performance and higher firm values. ROA is used to measure a firm’s short-
term accounting profitability, and TobinQ measures firm values and long-term market expectations
of firm performance. These two measures complement each other, as RTO policy changes may
not yield immediate effects on financial performance. However, if the mandate is expected to
crease long-term benefit, the effect should be reflected in firms’ stock market values. The key
independent variable is again the RTO (an indicator for the post-period of RTO firms). If RTO
increases (decreases) firm performance and shareholder values, we expect a positive (negative)
capture firm characteristics relevant to shareholder values, including firm size (Size_Lag), leverage
(∆Tobin_Lag), and sales growth (Sales_Growth_Lag) (Gupta et al., 2010; Patatoukas, 2012;
Huang & Hilary, 2018). These controls are also calculated using data from the prior quarter. In
addition, we control for firm and quarter fixed effects to control for time-invariant factors or other
time-variant common shocks that affect both RTO and non-RTO firms.
4. Empirical Analyses
Table 1 provides summary statistics for our sample used for estimating Model (1). Panel
A shows industry distribution of sample firms. In our sample, about 60 percent of the S&P 500
firms belong to the manufacturing industry and the finance, insurance, and real estate sector. Firms
from the service industry and the transportation and public utilities industry take up another 28
percent. In Panel B, we present the descriptive statistics for the determinants. We winsorize all
20
continuous variables at the top and bottom 1 percent level to mitigate the influence of outliers.
Since we have chosen our sample from the S&P 500 firms, our sample firms are large and on
average less financially constrained. Our sample is generally comparable to those used in similar
prior studies. For example, institutional ownership has an average of 84 percent, which aligns with
findings in previous research (Harvard Business Review, 2019; Chen et al., 2020). Notably, on
average, CEOs of S&P 500 firms receive more than twice the compensation of the second highest-
paid executive within their company, and 93 percent of S&P 500 CEOs are male. High-tech firms
Table 1 Panel C shows Pearson correlations between variables. We find that the RTO
mandate is positively correlated with firm size and negatively correlated with financial constraints,
stock performance, institutional ownership, and high-tech industries. The highest correlation is
between firm size and institutional ownership. But this is not high enough to cause concerns about
multicollinearity. These bivariate tests are only suggestive, and we draw our inferences based on
the regression analyses after controlling for other determinant factors as well as firm and quarter
fixed effects.
Figure 1 shows the distribution of RTO announcements from 2020 to 2023 on a quarterly
basis. Many firms announced their RTO plans between the second quarter of 2021 and the first
quarter of 2022. Figure 2 demonstrates the industry breakdown of RTO firms. The manufacturing
industry and the finance, insurance, and real estate industries collectively represent about 60
percent of RTO firms, with the service industry, transportation and public utilities industry, and
Table 2 analyzes the determinants of the RTO decision by running Probit Model (1) and
presents the results. Column 1 includes only firm-level performance-related factors. The
21
coefficient on Size is significantly positive at the 1 percent level (coefficient = 0.263, z-statistic =
12.01). This finding suggests that larger firms are more inclined to adopt RTO policies. Larger
organizations tend to have more complex operations and employ a greater number of individuals,
which could contribute to this tendency. The coefficient on RET_Lag (coefficient = -0.133, z-
statistic = -1.96) is negative and significant at the 1 percent level. This finding indicates that firms
with weaker stock performance before RTO announcements are more likely to enforce an RTO
mandate. However, this result does not necessarily mean that the poor performance is due to WFH.
As discussed above, this result could be expected if either employees working from home have
lower productivity or managers use employees as a scapegoat for bad firm performance. If WFH
indeed drives the poor performance, we expect institutional ownership to further increase the
= -1.95) is negative and significant at the 10 percent level. This finding is aligned with institutional
ownership decreases the probability that managers use RTO to blame employees for poor
In Columns 2, we introduce CEO characteristics into the analysis. The coefficients for
statistic = 1.76) are all statistically significant. These results suggest that firms led by male CEOs
with greater power are more inclined to enforce an RTO policy. These are in line with employees’
concerns that CEOs are using RTO for power grabbing. CEO_SHROWN and CEO_Political do
not appear to have a substantial impact on the adoption of RTO decisions. The insignificant effect
of CEO ownership is also inconsistent with managers believing RTO would increase firm values.
time into our analysis. We find that high-tech firms or firms facing higher industry sales
22
concentration are less likely to recall their employees to the office. This trend may be attributed to
the flexibility inherent in knowledge-based work and concerns that RTO would lead to brain drain
in these firms. Further, we find that commuting time has a significant and negative effect of RTO
mandates. Thus, firms headquartered in counties with longer commuting time are less inclined to
In summary, Table 2 highlights the importance of firm size, stock performance, CEO
attributes, industry nature, and geographical commuting conditions in influencing the RTO
decisions. Our findings are more consistent both with CEO power grabbing and with RTO being
imposed due to CEOs using RTO to blame employees for poor performance. We do not find
evidence that managers impose mandate because they believe RTO increases firm values.
employee reviews in Table 3 Panel A. The average ratings for Overall Satisfaction, Work-Life
Balance, and Senior Management are 3.75, 3.53 and 3.31 out of 5, respectively. These are
consistent with prior studies (Dube & Zhu, 2021; Liu & Sun, 2021).
Table 3, Panel B presents the main results of employee review rate change analysis. The
variable RTO is the key independent variable of interest. In the first column, the dependent variable
is firm-quarter average employees’ overall ratings of their employer. We find that the coefficient
Further, in Column 2 and 3, the dependent variables are the firm-quarter average employees’
ratings of Work-life Balance and Senior Management. The coefficients on Post are significantly
23
negative for both Work-life Balance and Senior Management (coefficients = -0.041 and -0.061, t-
statistic = -1.71 and -2.24, respectively). So, employees’ ratings for Work-life Balance and Senior
Management also experience a statistically significant decrease after RTO mandates. These results
We next provide tests of the parallel trends assumption underlying the DiD analyses in
Panel C of Table 3. To do so, we set up 5 indicator variables that indicate the 5 quarters before the
RTO announcement quarter t of an RTO firms respectively (i.e., Pre1, Pre2, Pre3, Pre4 and Pre5).
Then we re-estimate the DiD model after including these 5 indicators. Coefficients on these 5
indicators show the difference between RTO and non-RTO firms in the preperiod before the RTO
announcement. Thus, the benchmark period are the quarters before quarter t-5. Across the three
columns, we find that the coefficients on these indicators are generally insignificant. The only
exception is the coefficient on Pre5 in column 3, which is significant at the 10 percent level.
However, the coefficients on the other four quarters that lead up to the announcement are all
insignificant. Therefore, these findings support the parallel trends assumption and mitigate
To mitigate the concern that employees who self-selected to post after the RTO mandates
generally hold more negative views of their employers, we examine changes in two other ratings
on Glassdoor related to Compensation Benefits and Diversity Inclusion, which are concurrently
posted with Work-life Balance and Senior Management by each reviewer. These two ratings are
less directly related to RTO mandates and thus are expected to remain similar after the mandate.
Table 3 Panel D displays insignificant coefficients of RTO (coefficients = -0.023 and -0.018, t-
24
statistics = -1.09 and -0.59, respectively), indicating no significant changes in employee review
ratings for these two aspects. These findings are inconsistent with the alternative explanation.
Table 4 Panel A provides summary statistics for the firm performance analyses. As our
sample comprises solely S&P 500 firms, we expected our firms to be financially strong. Consistent
with our expectation, the mean value of quarterly return on assets (ROA) is positive, and the mean
Tobin’s Q (TobinQ) value is 2.60. These metrics suggest that the S&P 500 firms are profitable and
are highly valued by the market. Also consistent with our expectation, the low leverage ratio
indicates that the sampled firms are well managed and financially sound.
In Table 4 Panel N, Column 1 and Column 2 present changes in ROA and TobinQ after the
RTO mandate with the inclusion of control variables using TWFE regressions. The independent
variable of interest is RTO. The coefficients on RTO are insignificant in both columns. These
results suggest that firm performance and firm values did not significantly improve as predicted
by managers. Given the significant results of the employee satisfaction tests in a similar sample,
these insignificant coefficients are unlikely due to the sample size or lack of power for the analyses.
In summary, these findings are inconsistent with managers’ argument that RTO mandates improve
5. Conclusion
The COVID-19 pandemic has brought about significant changes in working modes and
perceptions. Past experiences need to be reevaluated to adapt to this new era. Before the pandemic,
employees generally didn’t have the chance to fully WFH. During the pandemic, employees who
worked fully from home may suffered from distractions, such as family conflicts and childcare
responsibilities, and they could only compare their WFH experience to their pre-pandemic in-
25
office experiences. However, in the postpandemic era, employees’ attitudes towards WFH may
change, as they may be less distracted at home or have successfully adjusted their pace of life when
Using a sample of S&P 500 firms, we examine determinants and consequences of U.S.
firms’ return-to-office (RTO) mandates. Results of our determinant analyses are consistent with
managers using RTO mandates to reassert control over employees and blame employees as a
scapegoat for bad firm performance. Also, our findings do not support the argument that managers
impose mandate because they believe RTO increases firm values. Further, our study finds a
significant decline in employee job satisfaction. However, we do not find significant changes in
firm performance in terms of profitability and stock market valuation after the RTO mandates.
Overall, our study adds to the ongoing debate over RTO mandates and demonstrates that
mandating employees to return to the office after the COVID pandemic does not result in a
significant improvement in firm performance. Instead, such mandates hurt employee satisfaction.
We demonstrate that one of the most frequently cited motivations by managers for implementing
the RTO policy, namely, firm performance improvement, might not be a valid justification for
such a policy in the postpandemic era. These results serve as a valuable resource for managers who
have previously mandated or are thinking about mandating a return to the office based on intuition,
Our study is subject to a number of limitations, which also presents opportunities for future
research. First, due to the timely nature of this issue, we do not have a long-enough window to
examine the long-term consequences to firm performance. Second, our study focuses on S&P 500
firms, which are the largest firms in the United States. While there are no strong reasons to believe
that our results are not applicable to other firms, we encourage future research to further examine
26
this issue in an even broader sample. Third, the period following the pandemic features a labor
shortage. If such labor shortage eases, it will be interesting to examine how employees may react
differently. We still expect RTO to negatively affect employee satisfaction. It is less clear whether
employees are less likely to voice their dissatisfaction in a tight labor market. Finally, though we
do not have strong concerns about endogeneity in our setting, we can not fully rule out alternative
27
References
Angelici M, Profeta P (2020) Smart-Working: Work Flexibility Without Constraints. CESifo Working
Paper No. 8165
Apgar 4th (1998) The alternative workplace: Changing where and how people work. Harvard Bus. Rev.
76(3):121–136.
Arikan KM, Masli A, Xi Y (2023) Political euphoria and corporate disclosures: An investigation of CEO
partisan alignment with the president of the United States. J. Account. Econ. 75(2–3):101552–.
https://doi.org/10.1016/j.jacceco.2022.101552.
Bahn K, Cohen J, van der Meulen Rodgers, Y (2020) A feminist perspective on COVID‐19 and the value
of care work globally. Gender Work Organ. 27(5):695–699.
Bai JJ, Brynjolfsson E, Jin W, Steffen S, Wan C (2021) Digital resilience: How work-from-home
feasibility affects firm performance (No. w28588). National Bureau of Economic Research.
Bailey DE, Kurland NB (2002) A review of telework research: Findings, new directions, and lessons for
the study of modern work. J. Organ. Behav. 23(4):383–400. https://doi.org/10.1002/job.144.
Baker AC, Larcker DF, Wang CC (2022) How much should we trust staggered difference-in-differences
estimates?. Journal of Financial Economics, 144(2), 370-395.
Bao L, Li T, Xia X, Zhu K, Li H, Yang X (2022) How does working from home affect developer
productivity?—A case study of Baidu during the COVID-19 pandemic. Sci. China Inform. Sci.
65(4):142102.
Barber LK, Conlin AL, Santuzzi AM (2019) Workplace telepressure and work–life balance outcomes:
The role of work recovery experiences. Stress Health 35(3):350–362.
Barrero JM, Bloom N, Davis SJ (2021) Let me work from home, or I will find another job. University of
Chicago, Becker Friedman Institute for Economics Working Paper (2021-87).
Barriga Medina HR, Campoverde Aguirre R, Coello-Montecel D, Ochoa Pacheco P, Paredes-Aguirre MI
(2021) The influence of work–family conflict on burnout during the COVID-19 pandemic: The
effect of teleworking overload. Int. J. Env. Res. Pub. He. 18(19):10302.
Bayat A, Goergen M, Koutroumpis P, Wei X (2021) Between Scylla and Charybdis: CEO political
ideology, dividends and downsizing during the pandemic. Working paper, European Corporate
Governance Institute–Finance (802).
BBC, 2023, The double standard of the return-to-office.
Bellmann L, Hübler O (2021) Working from home, job satisfaction and work–life balance–robust or
heterogeneous links? Int. J. Manpower 42(3):424–441.
Bick A, Blandin A, Mertens K (2022) Work from home before and after the COVID-19 outbreak.
Working paper, online. https://doi.org/10.20955/wp.2022.008.
Bloom N, Liang J, Roberts J, & Ying ZJ (2015) Does working from home work? Evidence from a
Chinese experiment. The Quarterly Journal of Economics, 130(1), 165–218.
Bloomberg (2020) Citi CEO says long-term productivity may suffer with remote work. National Real
Estate Investor.
Brittany HM, MacDonnell R (2012) Is telework effective for organizations. Manag. Res. Rev. 35(7):602–
616.
Business Horizons (2023) The work-from-home trend revisited: An interview with Brian Kropp.
Business Insider (2023a) Return-to-office orders look like a way for elite, work-obsessed CEOs to grab
power back from employees.
28
Business Insider (2023b) The back-to-office backfire: Companies ending WFH perks lose out on top
talent, who view flexible work as equivalent to an 8% raise.
Business Inside (2023c) A business owner ditching the office explains how he came to trust his
employees to work remotely so he could save money on rent.
Business Inside (2023d) The fight over return-to-office is getting dirty.
Callaway, B., & Sant’Anna, P. H. C. (2021). Difference-in-Differences with multiple time periods.
Journal of Econometrics, 225(2), 200–230.
Chatterjee K, Chng S, Clark B, Davis A, De Vos J, Ettema D, ... Reardon L (2020) Commuting and
wellbeing: A critical overview of the literature with implications for policy and future research.
Transport Rev. 40(1):5–34.
Chen, T, Dong, H, Lin, C (2020). Institutional shareholders and corporate social responsibility. Journal of
Financial Economics, 135(2), 483–504.
Choudhury P, Foroughi C, Larson B (2021) Work‐from‐anywhere: The productivity effects of geographic
flexibility. Strateg. Manage. J. 42(4):655–683.
Church NF (2015) Gauging perceived benefits from'working from home'as a job benefit. International
Journal of Business & Economic Development, 3(3), 81–89.
CNBC (2020) Drugmaker AbbVie pressures reluctant employees to return to work, raising safety
questions.
CNBC (2023a) Amazon employees express dismay, anger about sudden return-to-office policy.
CNBC (2023b) Bob Iger tells Disney employees they must return to the office four days a week.
Cragun OR, Olsen KJ, Wright PM (2020) Making CEO narcissism research great: A review and meta-
analysis of CEO narcissism. J. Manage. 46(6):908–936.
Chung, K. H., Zhang, H. (2011). Corporate Governance and Institutional Ownership. J Financ Quan
Anal. 46(1), 247–273.
Daily CM, Johnson JL (1997) Sources of CEO power and firm financial performance: A longitudinal
assessment. Journal of Management, 23(2), 97–117.
Davey L (2023) The return to office fight is about control not productivity.
https://www.lianedavey.com/the-return-to-office-fight-is-about-control-not-productivity/.
Davidescu AA, Apostu SA, Paul A, Casuneanu I (2020) Work flexibility, job satisfaction, and job
performance among Romanian employees—Implications for sustainable human resource
management. Sustainability 12(15):6086.
De Sio S, Cedrone F, Nieto HA, Lapteva E, Perri R, Greco E, Mucci N, Pacella E, Buomprisco G (2021)
Telework and its effects on mental health during the COVID-19 lockdown. Eur. Rev. Med.
Pharmaco. 25:3914–3922. https://doi.org/10.26355/eurrev_202105_25961.
Dube S, Zhu C (2021) The disciplinary effect of social media: Evidence from firms’ responses to
Glassdoor reviews. J. Account. Res. 59(5):1783–1825. https://doi.org/10.1111/1475-679X.12393.
DuBrin AJ (1991) Comparison of the job satisfaction and productivity of telecommuters versus in-house
employees: A research note on work in progress. Psychol. Rep. 68(3_suppl):1223–1234.
Etheridge B, Wang Y, Tang L (2020) Worker productivity during lockdown and working from home:
Evidence from self-reports (No. 2020-12). ISER Working Paper Series.
Farhadi R, Nanda V (2021) What do employees know? Quality perception and “over-satisfaction” in
firms going public. J. Corp. Financ. (Amsterdam, Netherlands) 66:101779–.
https://doi.org/10.1016/j.jcorpfin.2020.101779.
29
Felstead A, Jewson N, Phizacklea A, Walters S (2002) Opportunities to work at home in the context of
work‐life balance. Hum. Resour. Manag. J. 12(1):54–76.
Friedman HL (2014) Implications of power: When the CEO can pressure the CFO to bias reports. Journal
of Accounting & Economics, 58(1), 117–141.
Fonner KL, Roloff ME (2010) Why teleworkers are more satisfied with their jobs than are office-based
workers: When less contact is beneficial. J. Appl. Commun. Res. 38(4):336–361.
Fortune (2023a) Nvidia, a $1 trillion AI powerhouse, is fine with remote work—and is ignoring the
return-to-office trend. Fortune, 14 Sep 2023.
Fortune (2023b) Most bosses regret how they mandated workers return to the office. They blamed it on
not having enough data.
Fortune (2023c) San Francisco CEO who’s closing his $215 million company’s upscale office lounge
says return-to-work push is ‘fighting a losing war’.
Fox Business (2023) JPMorgan Chase CEO Jamie Dimon speaks up about remote work: 'there are real
flaws'.
Frydman C, Jenter D (2010) CEO compensation. Annu. Rev. Financ. Econ., 2(1), 75-102.
Furuya Y, Nakazawa S, Fukai K, Tatemichi M (2022) Health impacts with telework on workers: A
scoping review before the COVID-19 pandemic. Front. Public Health 10:981270–981270.
https://doi.org/10.3389/fpubh.2022.981270.
Gajendran RS, Harrison DA, Delaney-Klinger K (2015) Are telecommuters remotely good citizens?
Unpacking telecommuting’s effects on performance via i-deals and job resources. Pers. Psychol.
68(2):353–393. https://doi.org/10.1111/peps.12082.
Galanti T, Guidetti G, Mazzei E, Zappalà S, Toscano F (2021) Work from home during the COVID-19
outbreak: The impact on employees’ remote work productivity, engagement, and stress. J. Occup.
Environ. Med. 63(7):e426.
Garmaise MJ (2008) Production in entrepreneurial firms: The effects of financial constraints on labor and
capital. Rev. Financ. Stud. 21(2):543–577. https://doi.org/10.1093/rfs/hhm081.
Gibbs M, Mengel F, Siemroth C (2023) Work from home & productivity: Evidence from personnel &
analytics data on IT professionals. Journal of Political Economy Microeconomics, volume 1,
number 1, https://doi.org/10.1086/721803.
Giménez-Nadal JI, Molina JA, Velilla J (2020) Work time and well-being for workers at home: Evidence
from the American Time Use Survey. Int. J. Manpower 41(2):184–206.
Goel A, Thakor AV (2008) Overconfidence, CEO selection, and corporate governance. J. Financ. (New
York) 63(6):2737–2784. https://doi.org/10.1111/j.1540-6261.2008.01412.x.
Green TC, Huang R, Wen Q, Zhou D (2019) Crowdsourced employer reviews and stock returns. J.
Financ. Econ. 134(1):236–251. https://doi.org/10.1016/j.jfineco.2019.03.012.
Gupta M, Pevzner M, Seethamraju C (2010) The implications of absorption cost accounting and
production decisions for future firm performance and valuation. Contemp. Account. Res.
27(3):889–922. https://doi.org/10.1111/j.1911-3846.2010.01030.x.
Hao Q “Grace” (2014) Institutional shareholder investment horizons and seasoned equity offerings.
Financ. Manage. 43(1):87–111. https://doi.org/10.1111/fima.12038.
Harrington E, Emanuel N (2020) Working remotely? Selection, treatment, and market provision of
remote work. Unpublished Manuscript. Harvard University. doi, 10.
Hartman RI, Stoner CR, Arora R (1992) Developing successful organizational telecommuting
arrangements: Worker perceptions and managerial prescriptions. SAM Adv. Manage. J. 57(3):35.
30
Harvard Business Review. (2019). How Big a Problem Is It That a Few Shareholders Own Stock in So
Many Competing Companies?
Hjálmsdóttir A, Bjarnadóttir VS (2021) “I have turned into a foreman here at home”: Families and work–
life balance in times of COVID‐19 in a gender equality paradise. Gender Work Organ.
28(1):268–283.
Ho SSM, Li AY, Tam K, Zhang F (2015) CEO gender, ethical leadership, and accounting conservatism.
J. Bus. Ethics 127(2):351–370. https://doi.org/10.1007/s10551-013-2044-0.
Huang S, Hilary G (2018) Zombie board: Board tenure and firm performance. J. Account. Res.
56(4):1285–1329. https://doi.org/10.1111/1475-679X.12209.
Ipsen C, van Veldhoven M, Kirchner K, Hansen JP (2021) Six key advantages and disadvantages of
working from home in Europe during COVID-19. Int. J. Env. Res. Pub. He. 18(4):1826.
Irawanto DW, Novianti KR, Roz K (2021) Work from home: Measuring satisfaction between work–life
balance and work stress during the COVID-19 pandemic in Indonesia. Economies 9(3):96.
Kaplan SN, Zingales L (1997) Do investment-cash flow sensitivities provide useful measures of financing
constraints? Q. J. Econ. 112(1):169–215. https://doi.org/10.1162/003355397555163.
Kile CO, Phillips ME (2009) Using industry classification codes to sample high-technology firms:
Analysis and recommendations. J. Account. Audit. Finance 24(1):35–58.
Kotini-Shah P, Man B, Pobee R, Hirshfield LE, Risman BJ, Buhimschi IA, Weinreich HM (2022) Work–
life balance and productivity among academic faculty during the COVID-19 pandemic: A latent
class analysis. J. Womens Health 31(3):321–330.
Kröll C, Nüesch S (2019) The effects of flexible work practices on employee attitudes: Evidence from a
large-scale panel study in Germany. Int. J. Hum. Resour. Man. 30(9):1505–1525.
Li Y, Gong M, Zhang X-Y, Koh L (2018) The impact of environmental, social, and governance
disclosure on firm value: The role of CEO power. Brit. Account. Rev. 50(1):60–75.
Liu S, Sun K (2021) Work environment and employee performance: Evidence from sell-side analysts.
Rev. Bus. 41(2):85–117.
Matthews TA, Chen L, Omidakhsh N, Zhang D, Han X, Chen Z, ... Li J (2022) Gender difference in
working from home and psychological distress—A national survey of US employees during the
COVID-19 pandemic. Ind. Health 60(4):334–344.
Möhring K, Naumann E, Reifenscheid M, Wenz A, Rettig T, Krieger U, ... Blom AG (2021) The
COVID-19 pandemic and subjective well-being: Longitudinal evidence on satisfaction with work
and family. Eur. Soc. 23(sup1):S601–S617.
Morikawa M (2022) Work‐from‐home productivity during the COVID‐19 pandemic: Evidence from
Japan. Econ. Inq. 60(2):508–527.
Nolen M (2023) Remote/hybrid work on the decline as companies struggle to drive engagement, poll
finds. ChiefExecutive.net. https://chiefexecutive.net/remote-hybrid-work-on-the-decline-as-
companies-struggle-to-drive-engagement-poll-finds/.
Nugraha R, Wolor CW, Yohana C (2022) The effect of work from home, work-life balance, and job
satisfaction on employee performance. Account. Financ. (Kiev, Ukraine) 95(1):103–112.
Palumbo R (2020) Let me go to the office! An investigation into the side effects of working from home
on work-life balance. Int. J. Public Sect. Manag. 33(6/7):771–790.
Palumbo R, Manna R, Cavallone M (2021) Beware of side effects on quality! Investigating the
implications of home working on work-life balance in educational services. TQM J. 33(4):915–
929.
31
Patatoukas PN (2012) Customer-base concentration: Implications for firm performance and capital
markets. Account. Rev. 87(2):363–392. https://doi.org/10.2308/accr-10198.
Pathak D (2021) The work-life balancing act: A study on the mandatory work from home due to COVID-
19 on the IT and non-IT industry sectors. IJHCITP 12(3):1–20.
Pathan S (2009) Strong boards, CEO power and bank risk-taking. Journal of Banking & Finance, 33(7),
1340–1350.
Petcu MA, Sobolevschi-David MI, Anica-Popa A, Curea SC, Motofei C, Popescu AM (2021)
Multidimensional assessment of job satisfaction in telework conditions. Case study: Romania in
the covid-19 pandemic. Sustainability 13(16):8965.
Putri A, Amran A (2021) Employees work-life balance reviewed from work from home aspect during
COVID-19 pandemic. Int. J. Manag. Sci. Info. Tech. 1(1):30–34.
Reuters (2023) Hybrid-work trend may wipe out $800 billion from office property values by 2030,
McKinsey study says. (July 13).
Savić D (2020) COVID-19 and work from home: Digital transformation of the workforce. Grey J.
16(2):101–104.
Schade HM, Digutsch J, Kleinsorge T, Fan Y (2021) Having to work from home: Basic needs, well-
being, and motivation. Int. J. Env. Res. Pub. He. 18(10):5149.
Schall MA (2019) The relationship between remote work and job satisfaction: The mediating roles of
perceived autonomy, work-family conflict, and telecommuting intensity. Doctoral dissertation,
San Jose State University.
Smite D, Moe NB, Hildrum J, Gonzalez-Huerta J, Mendez D (2023) Work-from-home is here to stay:
Call for flexibility in post-pandemic work policies. J. Syst. Software 195, 111552.
Song Y, Gao J (2020) Does telework stress employees out? A study on working at home and subjective
well-being for wage/salary workers. J. Happiness Stud. 21(7):2649–2668.
Sousa-Uva M, Sousa-Uva A, Serranheira F (2021) Telework during the COVID-19 epidemic in Portugal
and determinants of job satisfaction: A cross-sectional study. BMC Public He. 21(1):1–11.
Sundaramurthy C (1999) Institutional investors and corporate governance: Best practices for increasing
corporate value [review of Institutional Investors and Corporate Governance: Best Practices for
Increasing Corporate Value]. Acad. Manage. Rev. 24(2):358–360.
Wall Street Journal. (2021) Jamie Dimon Says JPMorgan Won’t Go Remote After the Pandemic; Smaller
group of employees to work under hybrid model, CEO says in annual shareholder letter.
Wall Street Journal (Online) (2023) We asked workers why they're not coming back to the office; terrible
commutes. expensive child care. employees explain why they will keep working from home.
Weng DH, Yang H (2023) Is red or blue more likely to narrow the gap? The effect of CEO political
ideology on CEO‐employee pay disparity. J. Manage. Stud. https://doi.org/10.1111/joms.12917.
Yang L, Holtz D, Jaffe S, Suri S, Sinha S, Weston J, Joyce C, Shah N, Sherman K, Hecht B, Teevan J
(2022) The effects of remote work on collaboration among information workers. Nature Human
Behaviour, 6(1), 43–54.
32
Figure 1 Timeline for RTO Announcement Date
30
27
25
22
20 18
15
12
11
10 10
10
7
5
5 4 4
3
2
1 1
0
tr1 tr2 tr3 tr4 tr1 tr2 tr3 tr4 tr1 tr2 tr3 tr4 tr1 tr2 r3
Qt
2 0Q 2 0Q 2 0Q 2 0Q 2 1Q 2 1Q 2 1Q 2 1Q 2 2Q 2 2Q 2 2Q 2 2Q 2 3Q 2 3Q 2 3
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
Figure 1. Distribution of firms’ RTO mandates over our sample period. This figure shows the
distribution of RTO announcements at the firm level across different quarters from January 2020 to Aug
2023. A total of 137 S&P500 firms publicly announced their RTO plans. We summarize the frequency of
RTO announcements on a quarterly basis.
33
Fig 2. Industrial Distribution of RTO mandates. This figure shows the distribution of RTO mandates at
the industry level. The industry categorization is based on the first two digits of the SIC code. The figure is
constructed based on RTO plans publicly announced by a total of 137 S&P500 firms.
34
Table 1. Summary statistics for Determinant Analyses
Panel A: Industry Distribution of Sample Firms in Determinant Analysis
Industry # of Firms
Agriculture, Forestry, Fishing 1
Construction 4
Finance, Insurance, Real Estate 97
Manufacturing 177
Mining 16
Public Administration 3
Retail Trade 21
Services 69
Transportation & Public Utilities 59
Wholesale Trade 10
Total 457
Panel B: Descriptive Statistics of Determinants
Bottom Top
N Mean Quartile Median Quartile Std.Dev.
Size_Lag 4,455 10.18 9.29 10.09 10.99 1.28
KZ_Lag 4,455 -0.82 -0.99 1.05 1.89 5.40
∆ROA_Lag 4,455 0 -0.01 0 0.01 0.02
RET_Lag 4,455 0.10 -0.08 0.06 0.22 0.37
Instit_Own_Lag 4,455 0.84 0.73 0.82 0.90 0.24
CEO_Power 4,455 1.82 1.61 1.91 2.15 0.54
CEO_ SHROWN 4,455 0.59 0.04 0.1 0.32 1.87
CEO Gender 4,455 0.93 1.00 1 1 0.25
CEO Political 4,455 0.14 0.00 0 1 0.78
High_Tech 4,455 0.28 0.00 0 1 0.45
HHI 4,455 2.61 0.88 2 3.42 2.31
Commute_Time 4,455 13.08 12.65 13.08 13.67 1.03
35
Panel C: Correlation Matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
(1)RTO 1.000
36
Table 2. Determinant Analyses of RTO Decision
Dependent Variable = RTO (1) (2) (3)
Size_ Lag 0.263*** 0.275*** 0.281***
(12.01) (12.11) (12.11)
KZ_ Lag 0.002 0.002 0.002
(0.39) (0.34) (0.34)
∆ROA_ Lag -0.339 -0.394 -0.457
(-0.25) (-0.29) (-0.33)
RET_Lag -0.133** -0.142** -0.143**
(-1.96) (-2.08) (-2.09)
Instit_Own _Lag -0.132* -0.131* -0.126*
(-1.95) (-1.93) (-1.85)
CEO_Power 0.082* 0.083*
(1.65) (1.67)
CEO_ SHROWN 0.016 0.017
(1.15) (1.20)
CEO_Gender 0.187* 0.229**
(1.76) (2.12)
CEO_Political 0.036 0.022
(1.10) (0.67)
High_Tech -0.182***
(-2.91)
HHI -0.027**
(-2.44)
Commute_Time -0.223***
(-3.37)
Intercept -3.903*** -4.374*** -4.310***
(-16.65) (-14.66) (-14.28)
Pseudo R-squared 0.0616 0.0642 0.0714
Observations 4,455 4,455 4,455
This table reports the Probit regression results of Model (1). The dependent variable RTO takes a value
of 1 if the observation of an RTO firm is from the period after the RTO announcement date (including
the announcement period), and 0 otherwise. All variables are defined in the Appendix. z-statistics based
on standard errors clustered by firm in parentheses. ***, **, and * indicate significance at the 1 percent,
5 percent, and 10 percent levels (two-tailed).
37
Table 3. Employee Ratings Analyses
Panel A: Summary Statistics for Employee Ratings Analyses
N Mean p25 Median p75 Std.Dev
Overall Rating 5,053 3.75 3.50 3.77 4.02 0.41
Work-Life Balance 5,051 3.53 3.20 3.55 3.89 0.52
Senior Management 5,051 3.31 3.00 3.32 3.63 0.52
Size_Lag 5,053 10.33 9.45 10.28 11.12 1.25
ROA_Lag 5,053 0.02 0.01 0.01 0.03 0.02
Tobin_Lag 5,053 2.76 1.31 1.98 3.22 2.17
KZ_Lag 5,053 -0.63 -0.84 1.06 1.93 5.12
RET_Lag 5,053 0.09 -0.09 0.05 0.22 0.37
HHI_Lag 5,053 2.74 0.91 2.20 3.47 2.34
Panel B: DiD Estimators
(1) (2) (3)
Dependent Variable= Overall Salinification Work-life Balance Senior Management
RTO -0.057*** -0.041* -0.061**
(-2.79) (-1.71) (-2.24)
Size_Lag 0.014 0.015 -0.018
(0.28) (0.30) (-0.24)
ROA_Lag 0.210 -0.356 0.588
(0.63) (-0.73) (1.04)
Tobin_Lag 0.006 0.001 0.019
(0.69) (0.11) (1.49)
KZ_Lag -0.004 -0.004 -0.001
(-1.31) (-0.84) (-0.29)
RET_Lag 0.017 -0.010 0.003
(1.62) (-0.64) (0.14)
HHI_Lag 0.009 0.001 0.008
(0.60) (0.05) (0.40)
Firm FE Yes Yes Yes
Quarter FE Yes Yes Yes
Observations 5,053 5,051 5,051
R-squared 0.575 0.517 0.449
38
Panel C: Pre-Trend Analyses
(1) (2) (3)
Dependent Variable= Overall Salinification Work-life Balance Senior Management
Pre5 0.027 -0.052 -0.076*
(0.94) (-1.29) (-1.91)
Pre4 -0.043 -0.030 -0.068
(-1.24) (-0.79) (-1.41)
Pre3 -0.029 -0.022 -0.040
(-0.91) (-0.60) (-0.97)
Pre2 -0.043 -0.040 -0.034
(-1.30) (-1.22) (-0.89)
Pre1 -0.045 -0.045 -0.035
(-1.41) (-1.19) (-0.80)
Post -0.083** -0.073** -0.099**
(-2.57) (-2.11) (-2.31)
Firm FE Yes Yes Yes
Quarter FE Yes Yes Yes
Observations 5,053 5,051 5,051
R-squared 0.575 0.517 0.448
Panel D: DiD Estimators for Non-WFH Related Employee Ratings
(1) (2)
Dependent Variable= Compensation Benefits Diversity Inclusion
RTO -0.023 -0.018
(-1.09) (-0.59)
Controls Yes Yes
Firm FE Yes Yes
Quarter FE Yes Yes
Observations 5,041 4,731
R-squared 0.547 0.376
This table presents the summary statistics and results of employee rating tests. Panel A presents the
summary statistics for variables used in the employee ratings analyses, including the number of
observations (N), mean, the bottom quartile, the median, the top quartile and standard deviation. The
sample covers the period from June 2019 to Nov 2023 and consists of 5,053 firm-quarter level
observations. Panel B reports the difference-in-difference (DiD) regression with two-way fixed effects.
The dependent variables in the three columns are the firm-quarter average employee ratings of overall
job satisfaction (Overall Salinification), work-life balance (Work-life Balance), and senior management
(Senior Management), respectively. RTO is an indicator for observations of RTO firms in the post-RTO
period, including the RTO announcement quarter. Panel C reports tests of pre-existing time trends. PreN
is an indicator for N quarters before the announcement quarter for an RTO firm. Panel D reports the
TWFE DiD estimators for tests of two employee ratings that are not closely related to RTO. The
dependent variables in the two columns are the firm-quarter average employee ratings of compensation
benefits (Compensation Benefits) and diversity&inclusion (Diversity Inclusion), respectively. All
variables defined in the Appendix. Firm and quarter fixed effects are included but not tabulated. Robust
t-statistics based on standard errors clustered by firm are in parentheses. ***, **, and * indicate
significance at the 1 percent, 5 percent, and 10 percent levels (two-tailed).
39
Table 4. Performance Analyses
Panel A: Summary Statistics for Performance Analyses
N Mean p25 Median p75 Std.Dev.
ROA 6,026 0.02 0.01 0.01 0.03 0.02
TobinQ 6,026 2.6 1.31 1.95 3.17 1.76
∆ROA_Lag 6,026 0 -0.01 0 0.01 0.02
Size_ Lag 6,026 10.2 9.35 10.11 10.95 1.24
Leverage_Lag 6,026 0.36 0.24 0.35 0.47 0.2
Cashflow_ Lag 6,026 0 -0.02 0.02 0.04 0.06
∆Tobin_ Lag 6,026 0.01 -0.10 0.01 0.13 0.42
Sales_Growth_ Lag 6,026 0.03 -0.04 0.02 0.09 0.18
TA_Lag 6,026 -0.04 -0.07 -0.03 -0.01 0.05
Panel B: DiD Estimators
(1) (2)
Dependent Variable= ROA TobinQ
RTO -0.001 0.009
(-0.81) (0.16)
Size_Lag -0.010* -1.117***
(-1.90) (-5.88)
Leverage_Lag -0.013* -0.618*
(-1.84) (-1.91)
Cashflow_Lag -0.006 -0.479***
(-0.63) (-2.60)
TA_Lag -0.030* -0.973***
(-1.80) (-3.18)
∆ROA_Lag 0.066***
(3.29)
∆Tobin_Lag 0.251***
(7.81)
Sales_Growth_Lag 0.010*** 0.197***
(5.56) (4.31)
Firm Fixed Effects Yes Yes
Quarter Fixed Effects Yes Yes
Observations 6,026 6,026
R-squared 0.551 0.940
This table presents the summary statistics and results of firm performance tests. Panel A presents the
summary statistics for variables used in the firm performance analyses, including the number of
observations (N), mean, the bottom quartile, the median, the top quartile and standard deviation. The sample
covers the period from June 2019 to Nov 2023 and consists of 6,026 firm-quarter level observations. Panel
B reports results of firm performance tests using the difference-in-difference (DiD) regression with two-
way fixed effects. The dependent variables in the two columns are the quarterly return on assets (ROA) and
Tobin’s Q (TobinQ), respectively. RTO is an indicator for observations of RTO firms in the post-RTO
period, including the RTO announcement quarter. All variables defined in the Appendix. Firm and quarter
fixed effects are included but not tabulated. Robust t-statistics based on standard errors clustered by firm
are in parentheses. ***, **, and * indicate significance at the 1 percent, 5 percent, and 10 percent levels
(two-tailed).
40
Appendix. Variable Definitions
Indicator variable for observations in the post-RTO period, which equals 1
if an observation of an RTO firm is from the period after the RTO
RTO
announcement date (including the RTO announcement quarter), and 0 for
otherwise. Source: Manual collection.
Firm size at the end of the prior quarter, calculated as the natural logarithm
Size_Lag
of total assets (atq). Source: Compustat.
Financial constraints at the end of the prior quarter, calculated using
KZ_Lag
quarterly data following Kaplan & Zingales (1997). Source: Compustat.
Return on assets of the prior quarter, calculated as quarterly net income
ROA_Lag
(niq) deflated by average quarterly total assets (atq). Source: Compustat.
The quarterly change in ROA over the prior quarter, defined as ROA in
∆ROA_Lag
quarter t minus ROA in quarter t-1. Source: Compustat.
The cumulative returns from quarter q-3 to quarter q-2, where the current
RET_Lag
quarter is quarter q. Source: Compustat.
The percentage of ordinary shares owned by institutional shareholders at
the end of the prior quarter. The institutional holding numbers are
Instit_Own_Lag
extracted from Thomson Reuters Institutional Holding database. Source:
Thomson Reuters.
CEO relative power, calculated as CEO’s total compensation divided by
the average total compensations paid to the four highest-paid executives in
CEO_Power
the firm, following Frydman & Jenter (2010). Source: Compustat
Execucomp.
CEO ownership, measured as the percentage of common stock owned by
CEO_SHROWN
the CEO. Source: Compustat Execucomp.
Indicator variable that equals 1 if the CEO is male and 0 if female. Source:
CEO_Gender
Compustat Execucomp.
Indicator variable that equals 1 if the CEO's personal political contribution
CEO_Political to the Republican Party is more than half of the CEO’s total political
contribution from year 2012 to 2023, and 0 otherwise. Source: FEC.
Indicator variable that equals 1 if the firm is categorized as a high-tech
High_Tech company (first 3 digits being 283, 357, 366, 367, 382, 384, 481, 482, 489,
737, and 837). Source: Compustat.
Sales concentration within the 4-digit SIC industry, calculated by squaring
HHI the sales share of each firm competing in the industry and then summing
the resulting numbers. Source: Compustat.
Indicator variable that equals 1 if the firm’s headquarters county’s transfer
Commute_Time
time is in the top quintile. Source: United States Census Bureau website.
Employee overall job satisfaction, calculated as the firm-quarter average of
Overall Salinification employee ratings of their overall job satisfaction on Glassdoor. Source:
Manual collection.
Employee satisfaction with work-life balance, calculated as the firm-
Work-life Balance quarter average of employee ratings of their work-life balance on
Glassdoor. Source: Manual collection.
Employee satisfaction with firm top management, calculated as the firm-
Senior Management quarter average of employee ratings of top management on Glassdoor.
Source: Manual collection.
Indicator variable for observations in the pre-RTO period, which equals 1
PreN if an observation of an RTO firm is from N quarter(s) before the the RTO
announcement quarter. N ranges from 1 to 5. Source: Manual collection.
41
Employee satisfaction with compensation benefits, calculated as the firm-
Compensation Benefits quarter average of employee ratings of their compensation benefits on
Glassdoor. Source: Manual collection.
Employee satisfaction with diversity and inclusion, calculated as the firm-
Diversity Inclusion quarter average of employee ratings of their diversity and inclusion on
Glassdoor. Source: Manual collection.
Tobin’s Q, which is calculated as the market value (prccq*cshoq) plus
TobinQ
total liabilities (ltq) divided by the assets (atq). Source: Compustat.
Sales_Growth_Lag Sales growth over the prior quarter. Source: Compustat.
Financial leverage at the end of the prior quarter, which is measured as the
Leverage_Lag
ratio of total debt (dd1q + dlcq + dlttq) to assets (atq). Source: Compustat.
The quarterly change in Tobin’s Q over the prior quarter, defined as
∆Tobin_Lag
Tobin’s q in quarter t minus Tobin’s q in quarter t-1. Source: Compustat.
Cashflow from operating activities (oancfy) scaled by average total assets
Cashflow_Lag
(atq). Source: Compustat.
Total accruals, estimated as net income (niq) minus cashflow from
TA_Lag operating activities (oancfy), scaled by total assets (atq). Source:
Compustat.
42