Return-to-Office Mandates

Yuye Ding
[email protected]
Katz Graduate School of Business
University of Pittsburgh

Mark (Shuai) Ma*

[email protected]
Katz Graduate School of Business
University of Pittsburgh

Current Version: December 2023

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

Acknowledgements: We appreciate comments and suggestions from workshop participants at the

University of Pittsburgh. All errors are our own.
 “The RTO push is eyewash for investors to prove that drops in revenue and profitability
aren't a result of poor managerial decisions but the result of lazy workers sitting at home in
their pajamas. In some ways, it's a genius move for executives — a way to establish control
over workers during an unprecedented societal awareness of labor rights (thanks to the
striking workers of the Writers Guild of America, SAG-AFTRA, and the United Auto
Workers) while also shifting the blame and consequences of poor stock performance onto
those least responsible.” --Ed Zitron, the CEO of EZPR, a public-relations company.

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

performance (e.g., Davey, 2023; Business Insider 2023a).

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

consequences of U.S. firms’ RTO mandates.

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

mandates after the pandemic.

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.

Therefore, the effect of mandatory RTO on employee satisfaction is an empirical question.

Using difference-in-difference (DiD) regressions, we find significant declines in

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

being in general more negative about their employers.

Finally, we assess the consequences of RTO mandates to another important stakeholder,

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

consequences of remote work, limited attention is paid to determinants of workplace flexibility.

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

employee work arrangements. Due to government-mandated measures to enforce physical

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

of telecommuting and changed the traditional working mode (Savić, 2020).

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

they can focus on doing their life’s work” (Fortune, 2023c).

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).

2.2 Research on Remote Work Prior to the Pandemic

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

documented a substantial increase in satisfaction and a decrease in turnover intentions among

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.

2.3 Research on Remote Work during 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

in employee performance. Using self-reported productivity of remote workers during lockdown in

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

reported higher productivity.

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

employee mood and satisfaction (Giménez-Nadal et al., 2020).

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

distracting environment, were reported to decrease employee productivity, while self-autonomy

and self-leadership were reported as contributors to increased productivity.

3. Empirical Research Design

3.1 Data and Sample Selection

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

date. We collect data on effective dates in a similar way.

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

respective employers from Glassdoor.com. Financial information, stock-return data, CEO

information, institutional ownership, and industry-related factors are sourced from Compustat,

BoardEx, and Thomson/Refinitiv, respectively. Political-affiliation data are manually collected

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

Community Survey (ACS) 5-year estimates of travel time to work data.8

3.2 Determinants of RTO Mandates

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.

𝑅𝑇𝑂 = 𝛽0 + 𝛽1 𝑆𝑖𝑧𝑒_𝐿𝑎𝑔 + 𝛽2 𝐾𝑍_𝐿𝑎𝑔 + 𝛽3 ∆𝑅𝑂𝐴_𝐿𝑎𝑔 + 𝛽4 𝑅𝐸𝑇_𝐿𝑎𝑔 +

𝛽5 𝐼𝑛𝑠𝑡𝑖𝑡_𝑂𝑤𝑛_𝐿𝑎𝑔 + 𝛽6 𝐶𝐸𝑂_𝑃𝑜𝑤𝑒𝑟 + 𝛽7 𝐶𝐸𝑂_𝑆𝐻𝑅𝑂𝑊𝑁 +
𝛽8 𝐶𝐸𝑂_𝐺𝑒𝑛𝑑𝑒𝑟 + 𝛽9 𝐶𝐸𝑂_𝑃𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙 + 𝛽10 𝐻𝑖𝑔ℎ_𝑡𝑒𝑐ℎ + 𝛽11 𝐻𝐻𝐼 +
𝛽12 𝐶𝑜𝑚𝑚𝑢𝑡𝑒_𝑇𝑖𝑚𝑒 + 𝐹𝑖𝑟𝑚 𝐹𝐸 + 𝑄𝑢𝑎𝑟𝑡𝑒𝑟 𝐹𝐸 (1)

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

(∆ROA_Lag), stock returns (RET_Lag), and institutional ownership indicator (Instit_Own_Lag).

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

return. Details of these variable definitions are in the Appendix.

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

is expected to reduce the probability of RTO.

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

the short term.

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.

3.3 Consequences of RTO Mandates

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

Model (2) with two-way fixed effects (TWFE).12

𝑅𝑎𝑡𝑖𝑛𝑔𝑠 = 𝛽1 𝑅𝑇𝑂 + 𝛽2 𝑆𝑖𝑧𝑒_𝐿𝑎𝑔 + 𝛽3 𝑅𝑂𝐴_𝐿𝑎𝑔 + 𝛽4 𝑇𝑜𝑏𝑖𝑛_𝐿𝑎𝑔 +

𝛽5 𝐾𝑍_𝐿𝑎𝑔 + 𝛽6 𝑅𝐸𝑇_𝐿𝑎𝑔 + 𝛽7 𝐻𝐻𝐼_𝐿𝑎𝑔 + 𝐹𝑖𝑟𝑚 𝐹𝐸 +
𝑄𝑢𝑎𝑟𝑡𝑒𝑟 𝐹𝐸 (2)

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

relative to that of non-RTO firms. If employee satisfaction increases (decreases), we expect a

positive (negative) coefficient on this term.

Following prior research, we have incorporated several control variables designed to

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

observations between January 1, 2020 and November 30, 2023.

𝑅𝑂𝐴 = 𝛽1 𝑅𝑇𝑂 + 𝛽2 𝑆𝑖𝑧𝑒_𝐿𝑎𝑔 + 𝛽3 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝐿𝑎𝑔 + 𝛽4 𝐶𝑎𝑠ℎ𝑓𝑙𝑜𝑤_𝐿𝑎𝑔 +

𝛽5 𝑇𝐴_𝐿𝑎𝑔 + 𝛽6 ∆𝑅𝑂𝐴_𝐿𝑎𝑔 + 𝛽7 𝑆𝑎𝑙𝑒𝑠_𝐺𝑟𝑜𝑤𝑡ℎ_𝐿𝑎𝑔 + 𝐹𝑖𝑟𝑚 𝐹𝐸 +
𝑄𝑢𝑎𝑟𝑡𝑒𝑟 𝐹𝐸 (3)

𝑇𝑜𝑏𝑖𝑛𝑄 = 𝛽1 𝑅𝑇𝑂 + 𝛽2 𝑆𝑖𝑧𝑒_𝐿𝑎𝑔 + 𝛽3 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝐿𝑎𝑔 + 𝛽4 𝐶𝑎𝑠ℎ𝑓𝑙𝑜𝑤_𝐿𝑎𝑔 +

𝛽5 𝑇𝐴_𝐿𝑎𝑔 + 𝛽6 ∆𝑇𝑜𝑏𝑖𝑛_𝐿𝑎𝑔 + 𝛽7 𝑆𝑎𝑙𝑒𝑠_𝐺𝑟𝑜𝑤𝑡ℎ_𝐿𝑎𝑔 + 𝐹𝑖𝑟𝑚 𝐹𝐸 +
𝑄𝑢𝑎𝑟𝑡𝑒𝑟 𝐹𝐸 (4)

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)

coefficient on this term.

Following prior research, we have incorporated several control variables designed to

capture firm characteristics relevant to shareholder values, including firm size (Size_Lag), leverage

ratio (Leverage_Lag), net cashflow-to-asset ratio (Cashflow_Lag), total accruals (TA_Lag),

quarterly change of return on assets ( ∆ ROA_Lag), quarterly change of Tobin’s q ratio

(∆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

4.1 Determinant 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

make up 28 percent of our sample.

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

retail industry following.

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

probability of RTO. However, the coefficient on Instit_Own_Lag (coefficient = -0.132, z-statistic

= -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

performance. We do not find a significant effect of financial constraints.

In Columns 2, we introduce CEO characteristics into the analysis. The coefficients for

CEO_Power (coefficient = 0.082, t-statistic = 1.65) and CEO_Gender (coefficient = 0.187, z-

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.

In Column 3, we introduce additional industry-level factors and county-level commuting

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

mandate their employees to return to the office.

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.

4.2 Employee Satisfaction

4.2.1 DiD Estimators for Employee Ratings Change

We next report analyses of consequences to employees. We provide summary statistics for

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

on RTO is statistically negative (coefficient = -0.057, t-statistic = -2.79) in Column 1. Thus,

employees’ overall perception of their employer is negatively influenced by RTO mandates.

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

together suggest that RTO mandates negatively affect employee satisfaction.

4.2.2 Parallel Trends Analyses

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

concerns that our findings are due to pre-existing trend.

4.2.3 Alternative Explanation

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.

4.3 Firm Performance Analysis

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

firm performance and values.

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

working from home.

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,

helping them make more informed decisions going forward.

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

explanations based on potential endogenous issues.

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

(2)Size_Lag 0.204*** 1.000

(0.000)
(3)KZ_Lag -0.049*** -0.227*** 1.000
(0.001) (0.000)
(4)∆ROA _Lag -0.004 0.000 -0.016 1.000
(0.813) (0.982) (0.288)
(5)RET_Lag -0.030** -0.012 -0.061*** 0.008 1.000
(0.047) (0.436) (0.000) (0.585)
(6)Instit_Own_Lag -0.069*** -0.231*** -0.073*** 0.009 -0.003 1.000
(0.000) (0.000) (0.000) (0.565) (0.844)
(7)CEO_Power -0.010 -0.157*** 0.040*** 0.000 0.039*** 0.064*** 1.000
(0.487) (0.000) (0.008) (0.975) (0.010) (0.000)
(8)CEO_SHROWN -0.021 -0.152*** 0.008 0.005 0.031*** 0.006 -0.137*** 1.000
(0.156) (0.000) (0.586) (0.737) (0.041) (0.699) (0.000)
(9)CEO_Gender 0.011 -0.067*** 0.084*** 0.009 -0.004 0.047*** -0.004 0.038*** 1.000
(0.463) (0.000) (0.000) (0.551) (0.772) (0.002) (0.773) (0.010)
(10)CEO_Political 0.022 0.030*** 0.056*** 0.006 0.007 -0.017 0.017 0.115*** 0.084*** 1.000
(0.137) (0.047) (0.000) (0.670) (0.634) (0.243) (0.261) (0.000) (0.000)
(11)High_Tech -0.057*** -0.190*** -0.031** -0.004 0.023 0.027* 0.077*** 0.083*** 0.084*** -0.111*** 1.000
(0.000) (0.000) (0.038) (0.775) (0.117) (0.076) (0.000) (0.000) (0.000) (0.000)
(11)HHI 0.003 0.138*** -0.092*** -0.007 -0.014 -0.109*** -0.055*** 0.040*** -0.079*** -0.001 -0.180*** 1.000
(0.854) (0.000) (0.000) (0.659) (0.364) (0.000) (0.000) (0.007) (0.000) (0.959) (0.000)
(13)Commute_Time 0.020 0.034** -0.089*** -0.001 0.033** 0.073*** -0.011 -0.055*** 0.045*** -0.026* 0.023 -0.089***
(0.178) (0.022) (0.000) (0.961) (0.027) (0.000) (0.471) (0.000) (0.003) (0.088) (0.133) (0.000)
This table provides the descriptive statistics for the determinant analyses. Panel A reports the industry distribution of firms included in the
determinant analysis. Panel B presents the summary statistics for determinant factors used in the determinant analysis, including the number of
observations (N), mean, the bottom quartile, the median, the top quartile and standard deviation. Panel C reports Pearson correlations among the
dependent variable RTO and the determinant variables. Variables are defined in the Appendix. P-values in parentheses: ***, **, and * indicate
significance at the 1 percent, 5 percent, and 10 percent levels (two-tailed).

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