Content uploaded by Md Ismail Haidar
Author content
All content in this area was uploaded by Md Ismail Haidar on Oct 09, 2024
Content may be subject to copyright.
Corporate Governance: An International Review, 2024; 0:1–26
https://doi.org/10.1111/corg.12617
1 of 26
Corporate Governance: An International Review
ORIGINAL ARTICLE
Board Gender Diversity and Workplace Safety: Evidence
From Quasi- Natural Experiments
MdIsmailHaidar1 | SahaIqbalHossain2
1Business Administration, University of Texas Rio Grande Valley, Edinburg, Texas, USA | 2International Business Administration, Texas A&M
International University, Laredo, Texas, USA
Correspondence: Saha Iqbal Hossain (sahahossain@dusty.tamiu.edu)
Received: 4 November 2023 | Revised: 20 August 2024 | Accepted: 21 August 2024
Funding: The authors received no specific funding for this work.
Keywords: board gender diversity | corporate governance | difference- in- differences | gender behavior theory | gender quotas | gender socialization theory |
social role theory | workplace safety
ABSTRACT
Research Questions/Issue: This paper examines the impact of board gender diversity on workplace safety in an international
setting from 2002 to 2019. Utilizing gender quotas on corporate boards as a quasi- natural experiment, we explore the causal
effects of gender diversity on workplace safety, highlighting the significant inf luence of corporate board gender quotas on cor-
porate policies.
Research Findings/Results: Using a sample of 13,124 firm- year observations from 48 countries, our findings reveal that
gender quotas on corporate boards are linked to higher rates of work- related injuries. The association is particularly strong in
financially constrained firms, as well as in those with high workloads and low investments in workplace safety. In contrast, this
effect is mitigated in firms located in countries with strong institutional frameworks and high union representation. The results
hold across various regression specifications and remain robust when considering different measures of workplace injuries,
board gender diversity, and additional control variables.
Theoretical Implications: Our study draws on gender behavior theory, social role theory, and gender socialization theory, with
findings particularly aligning with gender behavior theory. It advances understanding of how changes in corporate governance
structures related to board diversity influence workplace safety. This research underscores the importance of further empirical
work to expand this area of study, suggesting valuable directions for future research through cross- sectional and longitudinal
analyses.
Practical Implications: The findings provide critical insights for policymakers considering the implementation of gender quo-
tas in boardrooms. Additionally, this research highlights the need for further exploration of unexamined aspects within govern-
ance studies, which will significantly influence future policy debates, legislative processes, and corporate practices in promoting
gender diversity and enhancing workplace safety.
© 2024 J ohn Wiley & Sons Ltd .
We appreciate c omments from two a nonymous reviewer s, Nasim Sabah ( Discussant), A ndré Varella Mollick , and conference par ticipant at the 202 3 Southwestern Fi nance Assoc iation Annual
Meeting a nd 2023 Decision S cience Institute A nnual Meeting . All remaini ng errors are our ow n.
2 of 26 Corporate Governance: An International Review, 2024
1 | Introduction
Previous research in economics and finance supports the view
that board structure influences corporate performance (e.g.,
Chen2015; Lawrence and Raithatha2023). Moreover, emerging
research suggests that board gender diversity impacts several
aspects of corporate behavior and outcomes, including corpo-
rate innovation (Griffin, Li, and Xu 2021), firm performance
(Bennouri etal.2018; Li and Chen2018), dividend payout pol-
icy (Ye et al. 2019), cost of bank loan (Karavitis, Kokas, and
Tsoukas 2021), managerial ability (Baghdadi, Safiullah, and
Heyden 2023), and many more. Despite the extensive research
linking board gender diversity to these critical performance in-
dicators, little attention has been given to how board gender di-
versity impacts workplace safety, a crucial area for both societal
welfare and organizational success.
Board gender diversity is paramount for addressing workplace
safety issues due to its multifaceted impact on decision- making
processes and organizational culture. First, workplace accidents
and injuries not only cause significant physical harm and emo-
tional distress but also entail substantial financial costs through
direct and indirect expenses such as medical payments, legal
fees, and productivity losses (Bradley, Mao, and Zhang 2022).
Furthermore, workplace fatalities carry profound economic
consequences. For example, International Labor Organization
(ILO) reports that 2.3 million men and women are injured or
exposed to diseases while working.1 Approximately 6000 deaths
occur every day because of workplace accidents. Carrivick
et al. (2005) further note that manual labor settings contrib-
ute significantly to national injury counts. By fostering diverse
perspectives and experiences at the highest levels of decision-
making, diverse boards can facilitate more comprehensive dis-
cussions and considerations regarding workplace safety policies
and practices (Siciliano1996). This inclusive approach not only
promotes a culture of safety and accountability within organi-
zations but also contributes to creating an environment where
employees feel empowered to raise safety concerns without fear
of discrimination or reprisal. Therefore, board gender diversity
plays an instrumental role in addressing workplace safety chal-
lenges and promoting a safer and healthier work environment
for all stakeholders.
Second, board gender diversity and workplace safety have im-
portant policy implications (Kowalewska 2020). If gender-
diverse boards are indeed associated with improved workplace
safety outcomes, this correlation could lend support to policies
advocating for corporate board gender quotas or diversity tar-
gets within corporate boards. Such initiatives could further bol-
ster efforts to promote gender diversity at the highest levels of
corporate governance.
Finally, the relationship between board gender diversity and
workplace safety can shed light on broader questions related to
diversity and inclusion. By exploring the impact of board gender
diversity on workplace safety, firms can design more effective
diversity and inclusion strategies and initiatives. Such initiatives
can help foster a more equitable and inclusive work environ-
ment. Although workplace safety has been broadly studied in
various fields, such as industrial relations, operations manage-
ment, and industrial- organizational psychology, its connection
with corporate governance is largely unknown. Therefore,
delving into this intersection could yield valuable insights into
enhancing both workplace safety and corporate governance
practices.
We reason that board gender diversity could increase workplace
injury rates for the following reasons. First, greater board gen-
der diversity might lead to an increase in workplace injury rates
due to poor firm performance, which can impede investments
in workplace safety. Ahern and Dittmar(2012) and Adams and
Ferreira(2009) demonstrate that board gender quotas decrease
firm value. Cohn and Wardlaw(2016) argue that firms with lim-
ited financial resources tend to have higher workplace injury
rates. Taken together, the poor firm performance associated
with increased board gender diversity could contribute to ele-
vated workplace injury rates.
Second, increased board gender diversity could lead to higher
workplace injury rates by causing conf licts or misunderstand-
ings among board members and employees. Such internal strife
can lead to suboptimal decision- making, lowered morale, and
elevated stress levels, which negatively affect worker safety and
well- being. In addition, a prevalent economic theory suggests
that limiting shareholders' freedom to choose directors reduces
board effectiveness (Eckbo, Nygaard, and Thorburn2022). Ahern
and Dittmar (2012) note that boards play a vital role in strategic
decision- making and financial policy changes and argue that
incorporating potentially younger and less experienced women
could weaken the board's effectiveness. In support of this argu-
ment, Zhu and Westphal (2014) argue that executives might be
reluctant to share vital information with directors from varied de-
mographic backgrounds, thereby reducing board efficiency. This
dynamic suggests that greater gender diversity on boards might be
associated with higher workplace injury rates.
Third, gender behavior theory suggests that risk- taking behav-
iors differ between men and women, with women typically
being more risk- averse (Croson and Gneezy2009; Apesteguia,
Azmat, and Iriberri2012; Booth and Nolen2012; Charness and
Gneezy2012). This aversion to risk could impact corporate de-
cisions, especially concerning investments in workplace safety.
Hope etal.(2022) note that higher safety investments can reduce
workplace accidents, yet women, prioritizing financial stability,
may be hesitant to sanction such investment (Rosenthal and
Morith1993). Paradoxically, this cautious approach could lead
to increased workplace injury rates in firms with considerable
gender diversity on their boards, presenting a complex dynamic
in how gender diversity impacts organizational safety outcomes.
However, another body of literature suggests that board gen-
der diversity might improve workplace safety for the fol-
lowing reasons. First, social role theory argues that gender
differences stem from distinct social roles, experiences, and
backgrounds (Eagly and Wood 2016; Ardito, Dangelico, and
Messeni Petruzzelli2021). In the context of corporate boards,
male directors typically possess more technical expertise and
board experience, whereas female directors often have higher
educational qualifications and less board experience and
come from varied non- business backgrounds (Singh, Terjesen,
and Vinnicombe2008; Hillman, Cannella, and Harris2002).
These varied backgrounds lead to a broader range of strategic
3 of 26
preferences, with men typically favoring conventional business
tactics and women more likely to adopt community- focused
strategies (Glass, Cook, and Ingersoll2 016; Terjesen, Sealy, and
Singh 2009). The resulting diversity in perspectives and ap-
proaches enhances the development of workplace safety mea-
sures, combining traditional and community- based strategies.
This variety helps in effectively addressing a broader array of
risks and vulnerabilities, potentially leading to improved safety
outcomes in organizations.
Second, gender socialization theory indicates that women typ-
ically exhibit traits such as empathy, a focus on stakeholders,
long- term orientation, corporate social responsibility (CSR)
friendliness, employee orientation, and ethical behavior,
whereas men are often seen as independent, assertive, and au-
thoritative (Au, Tremblay, and You2022; Briano- Turrent2022).
Women's interactional leadership style is also more effective in
managing organizational challenges such as sexual harassment
(Eagly and Carli2007; Au, Tremblay, and You2022). Moreover,
female board members often prioritize socially responsible con-
cerns, which helps foster an inclusive workplace and enhances
social performance (Shaukat, Qiu, and Trojanowski 2016; Au,
Tremblay, and You 2022). Thus, increasing board gender di-
versity might reduce workplace injury rates. Consequently, the
impact of the board gender quotas on workplace accidents is ul-
timately an empirical question.
We use firm- level data on work- related injuries and illnesses
from the Refin itiv Environmental, Soc ial, and Governance (E SG)
database (previously known as ASSET4 ESG) to study the im-
pact of board gender quotas on workplace safety. Measurement
of workplace safety is imperative. Literature has used various
ways to measure workplace safety. A standard measure, and one
that we use throughout the paper, is the total case rate (TCR),
which is the number of injuries and fatalities from all establish-
ments in a firm divided by the total number of working hours
multiplied by 200,000 (Cohn and Wardlaw 2 016; Amin, Kim,
and Lee 2021). An increase in the injury and illness rates, as
captured by the TCR, typically indicates a decline in the overall
workplace safety environment.
We leverage the staggered implementation of board gender quo-
tas across countries during the sample period to identify exog-
enous shifts in gender diversity on corporate boards. Typically,
a quotas policy mandates firms to attain a specified gender bal-
ance on their corporate boards. Noncompliance with this quotas
policy may lead to repercussions such as delisting or annulment
of board elections. As board gender quota are adopted at the
country level, their adaptation is unlikely to be endogenous to
changes in workplace injury rates within individual firms. This
framework enables us to use the difference- in- differences (DID)
methodology to estimate the effect of board gender quotas on
firm- level injury rates.
Using a comprehensive sample of 13,124 firm- year observations,
we find that the adoption of board gender quotas is associated
with a significant increase in workplace injury rates, and our
results are robust to various model specifications, such as in-
cluding different fixed effects and using alternative proxies for
workplace safety and board gender diversity. These findings
challenge the conventional wisdom that board gender diversity
improves firm outcomes and suggest that board gender quotas
may have unintended negative consequences on employee wel-
fare. Our results are also robust to additional firm- level, corpo-
rate governance- level, and country- level control variables.
We further explore cross- sectional variations in the effects of
board gender diversity policies on workplace safety. First, we
find that the adverse effects of board gender quotas on work-
place safety are stronger for financially constrained firms, con-
sistent with the argument that such firms may be more likely to
prioritize cost- cutting measures over investments in employee
welfare. Second, the effect of board gender quotas in increas-
ing workplace injury rate is less pronounced for firms in coun-
tries with strong institutional environments, possibly due to the
presence of strong regulations and enforcement mechanisms
that promote workplace safety regardless of board composition.
Third, the influence of board gender diversity on workplace
injury rates is reduced in firms in countries with high union
membership. This is consistent with the idea that the collective
bargaining of unionized workers leads to better working con-
ditions (Kaufman2005). Finally, the influence of board gender
diversity on escalating workplace accidents is more pronounced
in firms burdened with high workloads and low investments
in workplace safety. This situation highlights the need for tar-
geted safety investments and balanced workload management
to mitigate the potential risks associated with diverse board
compositions.
This paper is related to several strands of literature. First, our
results shed light on the debate over the real effect of the board
gender quotas. Ahern and Dittmar (2012) find that women's
representation on corporate boards negatively impacts firm
performance. Adams and Ferreira (2009) find that higher fe-
male presence on corporate boards increase board attendance
rate and monitoring effort. Ginglinger and Raskopf(2023) find
that the presence of women on boards increases firms' environ-
mental and social (E&S) performance. Our research introduces
a new dimension by highlighting a potential adverse effect of
board gender diversity on workplace safety, a critical concern for
both business and society.
Second, our paper contributes to the growing literature on
the determinants of workplace safety, which has explored
factors such as financing constraints, analyst forecasts, local
religiosity, information quality, and many more. Cohn and
Wardlaw(2016) show that financial constraints adversely im-
pact investment in workplace safety. According to Caskey and
Ozel (2017), firms that beat or meet analyst forecasts reduce
discretionary expenditures for employee welfare, contribut-
ing to increased workplace injuries and illnesses. Amin, Kim,
and Lee(2021) report that local religiosity improves workplace
safety. Hope etal.(2022) show that higher information quality
is associated with significantly decreased work- related injury
rates. We contribute to this literature by demonstrating that
board gender diversity is another important determinant of
employee health and safety.
This paper is structured as follows: Section2 reviews the liter-
ature and develops hypotheses. Section3 outlines the research
design and sample selection and presents summary statistics.
Sect ion 4 discusses the empirical results. Section 5 examines
4 of 26 Corporate Governance: An International Review, 2024
cross- sectional tests, and Section 6 explores tests of various
channels. Finally, Section7 concludes the paper.
2 | Literature Review and Hypothesis
Development
2.1 | Board Gender Quotas and Economic Outcome
Theoretically, quotas operate as a quantity- based mechanism.
Quotas set a minimum gender ratio on corporate boards, a
quantity regulation much like zoning regulation on land use
or fishing restrictions (Glaeser and Shleifer2001). In the con-
text of business, the examination of the effects of board gender
quotas policy is particularly valuable for identifying the causal
impact of women's representation on economic outcomes and
overcoming the typical endogeneity concerns involved in ad-
dressing this relationship. The case of Norway, a well- known
example of such a policy, highlights the dynamic nature of
this topic. Several studies examine whether the increased fe-
male representation in top positions due to the quotas had any
impact on firms' economic performance. The evidence is not
conclusive. Ahern and Dittmar(2012) demonstrate that stock
prices initially decreased when the quotas law was proposed
in 2002. Firms lacking female directors experienced a 3.54%
drop, contrasting with a mere 0.02% increase for firms with at
least one female director. Additionally, they observed a sub-
sequent decline in board qualifications and a deterioration in
firm operating performance. These findings align with their
hypothesis that firms adjusted their board structure to max-
imize firm value prior to the implementation of the gender
quotas policy. Matsa and Miller(2011) compare firms before
and after 2006, when the quotas rule was written into law,
and find that affected firms increased employment levels and
labor costs and reduced workforce reduction during the 2008
financial crisis. They suggest that gender diversity on boards
affects corporate strategy in the labor dimension.
In a recent study, Bertrand et al.(2019) revisited the Norwegian
quotas policy, examining its long- term impact 7 years after it be-
came mandatory. They discovered a reduction in the gender pay
gap within boards but did not observe an increase in female rep-
resentation or earnings in other positions outside the boardroom.
Although the quotas rule benefited female directors, its effects did
not extend to other female leadership roles as intended. The au-
thors also analyzed the characteristics of newly appointed female
directors and found an improvement in their qualifications, con-
tradicting the findings of Ahern and Dittmar(2012).
Several papers have examined the impact of board gender pol-
icies in specific countries like France and the United Kingdom
(Chandler2016), Spain (Reguera- Alvarado, De Fuentes, and
Laffarga 2017), and Italy (Ferrari et al.2022). Despite these
detailed country- specific insights, comprehensive cross-
country analyses remain relatively rare. A recent survey paper
revealed that among 200 studies on gender quotas published
between 1990 and 2005, only 7% adopted a cross- regional or
global perspective (Hughes etal.2019). Among studies focus-
ing on board- specific gender policies, Terjesen, Aguilera, and
Lorenz (2015) compared countries with quotas versus those
with a non- binding “comply or explain” principle and explored
the institutional factors influencing policy choices. Another
study involving multiple countries is a collection of case stud-
ies published in the book Gender Diversity in the Boardroom,
where each case study provides a descriptive analysis of a spe-
cific country's gender policy. Additionally, Fauver, Hung, and
Taboada(2019) investigated cross- country variations in board
gender policies, analyzing the valuation impact of these pol-
icies. They found that investors reward firms for board gen-
der diversity while penalizing firms for lacking it. This body
of work, though informative, points to a significant need for
more robust, comparative research that could further illumi-
nate the varied impacts of gender diversity policies across dif-
ferent economic and cultural landscapes.
2.2 | Determinants of Workplace Safety
Workplace safety is an important ethical, social, and eco-
nomic concern for the employees, firm managers, and society
(Hart 2010). Firms invest significant resources in improving
workplace safety environment each year. However, they also
face huge losses due to safety- related incidents. For instance,
Leigh(2011) claims that the estimated losses from workplace ac-
cidents and illnesses are approximately $250 billion every year,
which is more than the total costs of cancer treatment in the
United States. Kaptein(2008) argues that unethical practices are
associated with firms with inadequate sa fety measures that indi-
rectly cause accidents and incur socioeconomic costs. Previous
literature shows that poor workplace safety leads to a significant
deterioration in firm performance (Cohn and Wardlaw 2016;
Christensen et al. 2017). Given the negative consequences of
workplace injuries and illnesses on firms, employees, and soci-
ety, researchers have begun investigating various determinants
that influence employee workplace safety.
In exploring these determinants, it is essential to highlight the piv-
otal role of board involvement in setting strategic safety directives.
Although operational safety tasks are often delegated to executives
such as CEOs or HR directors, the overarching responsibility of
boards in overseeing and intervening in significant safety issues
is crucial. This involvement not only helps mitigate risks but also
aligns w ith broader corporate governance and sustainability goals.
For instance, under CEO Paul O'Neill's leadership, Alcoa priori-
tized safety—a strategy fully supported by the board—that led to
substantial improvements in both safety and operational perfor-
mance, illustrating the critical role of board involvement in achiev-
ing comprehensive business success (Duhigg2012).
In the operations management and organizational psychology
literature, researchers have focused on the operational deter-
minants of workplace safety. For example, Hofmann, Burke,
and Zohar (2017) review the literature on the application of
psychology to occupational safety. Their analysis indicates that
research initially concentrated on safety- related training during
the mid- 20th century but later shifted toward examining the in-
fluence of leadership and organizational climate on employee
safety. Meanwhile, in the realm of operations management,
Wiengarten et al. (2017) find that reducing operational slack
correlates with a decline in employee workplace safety. Kim
etal.(2019) find that implementing safety management systems
improves firms' safety records, emphasizing the significance of
5 of 26
a strong safety climate or culture for effective safety manage-
ment system implementation.
Within the accounting and finance literature, workplace safety
has emerged as a novel area of investigation. For example, Cohn
and Wardlaw(2016) argue that firms with financial constraints
have limited financial resources for workplace safety–related
investments. Their study reveals a positive correlation between
injury rates and leverage, alongside a negative correlation with
cash- flow shocks. Li kewise, C askey and Ozel(2017) demonstrate
that firms that strive to meet or exceed market expectations also
have higher injury rates as they may curtail safety- related ex-
penditures to outperform benchmarks. These two papers sug-
gest that required workplace safety–related investments are
important factors of workplace safety.
Previous literature also investigates the roles of different stake-
holders as determinants of workplace safety. Cohn, Nestoriak,
and Wardlaw (2021) demonstrate that firms with private equity
buyouts experience lower injury rates. They reason that private
equity firms have a long- term view and the incentive to invest in
safety improvements that primarily generate long- run returns.
Liang et al. (2023) find that workplace injuries in publicly listed
firms are lower than those in comparable private firms, and this
effect relates to heightened monitoring by the media and regula-
tors. Bradley, Mao, and Zhang(2022) find that workplace safety is
positively associated with analyst coverage due to the monitoring
role of financial analysts. Gillen etal.(2002) find that union work-
ers are less likely to perceive risk- taking as a part of their job, and
Morantz(2013) claims that unionization leads to fewer workplace
injuries. Local religiosity plays an important role in a firm's work-
place safety policies. Amin, Kim, and Lee(2021) demonstrate that
firms with establishments located in religious countries are less
likely to violate workplace conduct and more likely to implement
workplace safety measures. Furthermore, internal organizational
dynamics shape workplace safety outcomes. Hope et al. (2022)
find that higher internal information quality is associated with
significantly lower work- related injury rates.
Executive compensation structures also influence workplace
safety. Wu, Li, and Yu(2023) show that CEOs inside debt hold-
ings are negatively related to employee workplace injury and
illness. Therefore, our understanding of the determinants of
workplace safety is not only limited but also inconclusive based
on the recent study, which warrants further empirical investi-
gation on the determinants of workplace safety. We provide evi-
dence that board gender quotas affect workplace safety.
2.3 | Board Gender Diversity and Workplace Safety
As previously discussed, we identif y three potential reasons why
board gender quotas might lead to increased workplace injury
rates. First, board gender quotas could lead to an increase in
work- related injury rates by impairing a firm's profitability and
performance, which in turn may necessitate cost- cutting mea-
sures that compromise safety equipment, training, and supervi-
sion. Supporting this view, Ahern and Dittmar(2012) document
that firms mandated to have female board representation expe-
rience a reduction in firm value. In the related study, Adams and
Ferreira(2009) and Lu(2019) demonstrate that increased board
gender diversity is associated with poorer firm performance.
Cohn and Wardlaw (2016) also reveal that firms with limited
financial resources exhibit higher workplace injury rates. Taken
together, the poor firm performance associated with a greater
board gender diversity could potentially translate into elevated
workplace injury rates.
Second, increased board gender diversity might exacerbate con-
flicts and misunderstandings among board members, impairing
decision- making and negatively affecting the organization's safety
culture. A common economic hypothesis suggests that restricting
shareholders' choice in director selection lowers board effective-
ness (Eckbo, Nygaard, and Thorburn 2022). As boards are cru-
cial in strategic decision- making and significant financial policy
changes, the inclusion of potentially younger and less experienced
women could undermine their effectiveness (Fahlenbrach, Low,
and Stulz 2010; Ahern and Dittmar2012). It is also essential to
consider additional dynam ics inf luenced by the quotas system. For
instance, retained male directors may adjust their behavior in re-
sponse to the appointment of new female members, or the newly
appointed women may have different priorities compared to men.
In certain boardrooms, women may face limitations in wielding
the authority necessary for impactful decision- making, potentially
resulting in diminished performance. Communication issues may
arise if company executives are reluctant to share crucial infor-
mation with directors from different demographic backgrounds,
which could potentially compromise the efficiency of the board.
Adams and Ferreira (2009) and Zhu and Westphal (2014) also
highlight that executives might withhold critical information from
demographically diverse directors.
Third, gender behavior theory suggests that men and women often
exhibit distinct behavioral tendencies that can significantly in-
fluence their decision- making processes in professional settings.
Research indicates that women generally exhibit greater risk aver-
sion than men (Croson and Gneezy2009; Booth and Nolen2012;
Carter, Franco, and Gine2017) and are more cautious in scenarios
involving risk (Apesteguia, Azmat, and Iriberri2012). These be-
havioral disparities can influence decision- making across various
domains, including financial matters, investment in workplace
safety initiatives, and project management. Moreover, women may
display a preference for short- term orientation, opting for safer
alternatives even if they offer lower returns or benefits (Powell
and Ansic1997; Croson and Gneezy2009). In contrast, men may
demonstrate a greater propensity for risk- taking and possess more
confidence in their decisions, potentially impacting their perfor-
mance and outcomes in work, business, and leadership contexts
(Byrnes, Miller, and Schafer1999; Barber and Odean2001; Dreyer
etal.2022).
Hope etal.(2022) suggest that enhanced safety investments are
typically associated with a reduction in workplace accidents.
However, Rosenthal and Morith (1993) highlight a paradox
where women, possibly prioritizing financial stability, may be
reluctant to commit significant resources to safety measures.
This financially conservative approach might inadvertently lead
to increased workplace injury rates in companies with consider-
able gender diversity on their boards.
Based on the above argument, we offer our main hypothesis in
the alternative form as follows:
6 of 26 Corporate Governance: An International Review, 2024
Hypothesis 1. Board gender quotas are associated with
higher workplace injury rates.
An alternative perspective posits that board gender diversity
will lead to better employee welfare treatment for several rea-
sons. First, social role theory, as articulated by Eagly (1987),
posits that men and women conform to the societal expectations
and stereotypes associated with their respective social roles.
These roles are shaped by cultural norms, cognitive frames, and
beliefs, which significantly influence managerial behaviors and
decisions within firms (Galaskiewicz1991). According to Eagly
and Kite (1987), these gender- specific roles dictate behaviors
that are typically descriptive, highlighting what is commonly
done by each gender, and prescriptive, suggesting what should
be done within a cultural context.
Incorporating this theory, research by Boulouta (2013) illus-
trates how “empathic caring,” a trait strongly associated with
female directors, can lead to enhanced corporate social perfor-
mance. This finding underscores the potential of female board
directors to positively impact firms through their distinct ap-
proach to governance and interaction, driven by communal
qualities such as empathy and compassion (Eagly 2009). Such
qualities are highly valued in nurturing relationships and foster-
ing a supportive work environment, whereas men are often as-
sociated with agentic traits like assertiveness and independence
(Dobbins 1985; Eagly and Karau 1991), which contribute to a
different style of leadership and decision- making.
The implications of these dynamics are substantial, especially in
the context of discussions about board gender quotas and work-
place safety. A gender- diverse board offers a wider range of per-
spectives and experiences, leading to more comprehensive and
inclusive policies that can improve employee welfare and safety.
This diversity enriches board dynamics and decision- making
processes, potentially transforming organizational culture and
enhancing overall corporate governance.
Moreover, studies have shown that these gender- based norms
and perceptions significantly influence the management styles
of directors, highlighting the benefits of a balanced board com-
position (Galaskiewicz 1991; Schwartz and Rubel2005). By le-
veraging the unique strengths associated with both communal
and agentic traits, organizations can achieve a more effective
governance structure that better addresses the complexities of
modern business environments. Therefore, integrating women
into corporate boards not only diversifies perspectives but also
embeds a richer blend of empathic and decisive leadership styles.
This combination can lead to improved decision- making pro-
cesses and enhance workplace safety, supporting the hypothesis
that board gender diversity is likely to lead to better employee
welfare outcomes.
Second, from the perspective of gender socialization theory,
women are often characterized as being more sympathetic,
stakeholder- fo cused, and long- term oriented th an men. Research
indicates that women are more likely to emphasize CSR, prior-
itize employee benefits, and oppose unethical practices (Au,
Tremblay, and You 2022; Briano- Turrent 2022). Men, on the
other hand, are typically described as independent, assertive,
competitive, and authoritarian (Eagly and Johnson 1990; Au,
Tremblay, and You2022; Briano- Turrent2022). Consequently,
women's leadership style, where female focus more on partic-
ipation and cooperation, may make them better equipped to
address organizational challenges such as sexual harassment
(Eagly and Carli 2007; Au, Tremblay, and You2022). Studies
have shown that women board members tend to prioritize
long- term goals and socially responsible initiatives, creating
inclusive workplaces and enhancing organizations' social per-
formance (Shaukat, Qiu, and Trojanowski2016; Au, Tremblay,
and You2022). Moreover, increasing the presence of women in
leadership roles not only brings diversity to corporate boards but
also empowers junior female staff through mentorship and sup-
portive work environments (Dezsö and Ross2012). Thus, firms
with greater gender diversity on their boards may be more in-
clined to implement strategies that effectively reduce workplace
injury rates, underscoring the broader benefits of diverse leader-
ship in promoting employee welfare.
Now, we develop four cross- sectional hypotheses to explore
cross- sectional variations in how board gender diversity af-
fects workplace safety (if any). First, we consider the financial
resources of the firm. Financial constraints within a company
may influence the extent to which board gender diversity af-
fects workplace safety practices. For example, companies with
limited financial resources may prioritize cost- cutting mea-
sures over safety initiatives, regardless of the composition of
their board. Cohn and Wardlaw(2016) suggest that firms tend
to reduce safety investments when confronted with financial
constraints, as managers often prioritize short- term goals over
long- term benefits. Consequently, maintaining a high level of
workplace safety might be perceived as a luxury that financially
constrained firms cannot afford. Based on the above arguments,
we propose the following hypothesis:
Hypothesis 2. The effect of board gender diversity in
increasing workplace injury is more pronounced when firms are
financially constrained.
Second, our cross- sectional prediction relies on the quality of for-
mal institutions at the country level. Institutional quality refers
to the efficiency and effectiveness with which formal institutions
support firms' engagement and success in market transactions
(North 1991). These institutions are fundamental in enforcing
laws and protecting property rights, which are crucial for ensuring
dependable market transactions and facilitating broad, impersonal
exchanges (North 1991). The effectiveness of these institutions is
reflected in their capability to uphold property rights, manage
corruption levels, and maintain the impartiality and efficiency
of government officials. It also involves ensuring public security
and promoting corporate ethics and accountability (Kafouros and
Aliyev2016; Shaner and Maznevski2011).
High- quality institutional frameworks, characterized by fac-
tors like political stability, the rule of law, regulatory qual-
ity, and government effectiveness, are linked to improved
corporate governance standards (Filatotchev, Jackson, and
Nakajima2013; Nguyen et al. 2021) and tighter enforcement
of regulations, including those related to workplace safety. In
environments with strong legal protections for shareholders
and rigorous regulatory oversight, the role of board gender di-
versity in enhancing workplace safety might be less critical.
7 of 26
With lower agency costs and reduced emphasis on workplace
safety as a mechanism to address agency problems, female di-
rectors may allocate their attention and efforts toward other
areas of corporate governance. Therefore, in countries with
high institutional quality, where governance structures are
robust and regulatory compliance is rigorously enforced, the
impact of board gender diversity on workplace safety could be
less significant.
These analyses lead us to our following hypothesis:
Hypothesis 3. The effect of board gender diversity in increas-
ing workplace injury is weaker when firms are in a good institu-
tional qualit y environment.
Existing empirical research lacks information on union ac-
cess to effective participation in workplace health and safety
and their influence over its management (Robinson and
Smallman 2013; Shearn 2004). Our third cross- sectional
prediction is based on union bargaining power. Freeman
and Medoff (1984) identify two fundamental roles of labor
unions: first, as a collective voice advocating for the interests
and rights of their members and, second, as entities that may
seek monopoly rents, which can benefit unionized workers at
the expense of others (Farber1986). In practice, labor unions
initiate various actions to improve workplace safety, which
include demanding employers maintain safe working condi-
tions, educating workers about potential hazards, influencing
regulatory scrutiny levels, and promoting safety innovations
through economies of scale (Morantz2009).
The effect of board gender diversity on workplace safety will be
less pronounced when employees have more bargaining power.
Employees naturally prioritize their own safety, and those with
more bargaining power are generally more successful in advocat-
ing for safer workin g conditions. Thi s dynamic could dimini sh the
impact of gender diversity on boards in such contexts. Research
indicates that unionized workers, who often negotiate safety is-
sues, tend to report fewer workplace injuries (Kaufman 2005;
Morantz2013). Therefore, we expect a diminished effect of board
gender diversity on workplace safety in environments where em-
ployees wield significant bargaining power.
Hypothesis 4. The effect of board gender diversity in increas-
ing workplace injury is weaker when firms are in a country with a
high proportion of union membership.
Finally, we examine the impact of employee workload and
workplace safety investment. Previous research has shown that
high workloads, combined with insufficient safety measures
such as lack of personal protective equipment (PPE) or inade-
quate training, are linked to an increased risk of workplace in-
juries and labor law violations (Cohn and Wardlaw2016; Caskey
and Ozel 2017). The risks are magnified under conditions in-
volving extended hours, physically demanding tasks, or haz-
ardous environments. Although physical jobs are often linked
to accidents, non- physical settings also pose significant dan-
gers. Factors such as tight deadlines, job insecurity, and related
stressors can lead to mental strain, potentially causing anxiety
and depression (Kurniawaty, Ramly, and Ramlawati 2019).
Such psychological strain not only impacts the well- being and
performance of employees but also affects their personal rela-
tionships (Fletcher1988; Galambos and Walters1992; Morrison
and Clements 1997; Domenighetti, D'Avanzo, and Bisig 2000;
Sparks, Faragher, and Cooper2001). In addition, extended work-
ing hours without sufficient breaks can lead to fatigue, raising
the likelihood of accidents and cognitive decline (Goode2003;
Caldwell etal.2019). In sedentary or computer- based roles, poor
ergonomics and a lack of health awareness can cause physical
injuries, whereas digital eye strain from excessive screen use
may lead to headaches, vision issues, and dry eyes, all detri-
mental to productivity and overall health (Dainoff etal.2012;
Akinbinu and Mashalla 2 014). Effective workload manage-
ment is essential for maintaining safety standards, yet boards
with gender diversity might not prioritize measures to prevent
overworking, possibly due to external pressures such as market
competitiveness and a focus on short- term financial returns that
often overshadow the long- term benefits of safety investments.
Investments in workplace safety are essential for mitigating
the risks associated with high workloads and ensuring a safe
working environment. These investments include maintaining
equipment, replacing outdated machinery, optimizing work-
flows, and providing employee training (Hope et al. 2022).
These measures ensure that machinery is kept up- to- date and
safe, workf lows are streamlined for efficiency, and employees
are adequately trained, collectively reducing the likelihood of
workplace accidents.
However, the impact of female board members on these invest-
ments might be nuanced. Although the increasing presence of
women on corporate boards garners significant interest, it may
not significantly influence investments in workplace safety.
Women on boards are often found to be more risk- averse, po-
tentially leading to a cautious approach toward capital- intensive
safety investments (Rosenthal and Morith 1993). This conser-
vative approach might limit their support for aggressive invest-
ments in safety measures, potentially affecting the scope and
scale of such initiatives (Priddy Patterson1994).
Based on the above argument, we develop the following
hypothesis:
Hypothesis 5. The effect of board gender diversity in increas-
ing workplace injury is stronger in firms with high workloads and
low workplace safety investment.
3 | Research Design, Sample Selection, and
Summary Statistics
This section defines the variables used in the study, describes
the econometric methods and sample selection processes em-
ployed, and presents an analysis of the descriptive statistics.
3.1 | Measures of Workplace Safety
Building on prior research (Hope etal.2022), we use firm- level
total case rate (TCR) as a primary measure for workplace safety.
Firm TCR is calculated by adding the total number of deaths, in-
juries, and illnesses resulting in days away from work or with job
8 of 26 Corporate Governance: An International Review, 2024
restriction or transfer, along with other recordable cases within
a firm. This sum is then divided by the total number of hours
worked by all employees in a specific firm- year and multiplied
by 200,000. To ensure the robustness of our findings, we use
four alternative measures of workplace safety, as suggested by
prior literature (i.e., Cohn and Wardlaw2016; Amin, Kim, and
Lee2021). First, LOG_TOTAL_CASE is the natural logarithm of
the total number of injuries and fatalities reported by employees
and contractors while working for a company. Second, DART
is the total number of injuries that caused the employees and
contractors to lose at least a working day relative to 200,000 h
worked. Third, L OG_ DAWF is the natural logarithm of num-
ber of lost working days of the employees and contractors. Lost
working days are defined as absences from work due exclusively
to occupational injuries or diseases, which is typical ly referred to
as the severity rate. Finally, industry- adjusted TCR (IND_TCR)
is the difference between firm i's TCR in year t and the sample
average of all firm's TCR in industry j, defined by two- digit SIC
codes in year t.2
3.2 | Econometric Strategy
3.2.1 | DID Regression
We employ a DID approach to test the relationship between
board gender quotas policy and workplace safety. Our treat-
ment group includes firms incorporated in a country that has
passed board gender quotas. The control group contains firms
incorporated in a country that does not pass the board gender
quotas. We have developed the following model specification,
which is grounded in the extensive literature on workplace
safety (e.g., Cohn and Wardlaw 2016; Bradley, Mao, and
Zhang2022; Hope etal.2022), but augmented with board gen-
der quotas to provide a comprehensive understanding of the
topic. More specifically, we use firm- level workplace model in
the following form:
where
i
,
c
, and
t
represent firms, country, and time, respec-
tively. Our dependent variable,
TCRict
, is the measures of
workplace safety of firm
i
incorporated in country c in year
t
. Our main explanatory variable,
GENDER_QUOTASict
, is an
indicator variable that equals 1 if firms in country
c
passed
board gender quotas in year
t
and 0 otherwise. The primary
coefficient of interest is
𝛽
, which estimates the response of
workplace safety to board gender quotas. If
𝛽<0
, there is a
negative estimate for gender quotas: Firm incorporated in
countries adopted gender quotas law decrease workplace in-
jury rate and vice versa.
Xi,t
represents firm, corporate gov-
ernance, and country- level control variables. We use several
firm- level control variables following previous literature (i.e.,
Cohn and Wardlaw2016; Caskey and Ozel2 017; Amin, Kim,
and Lee2021; Hope etal.2022). Those are firm size, leverage,
cash flow, cash, dividend dummy, tangibility, employee size,
and strike. The variable FIRM_SIZE is the natural logarithm
of one plus total asset in millions of US dollars. The variable
LEVERAGE is the total debt scaled by total assets. CASH_
FLOW is the sum of earnings before interest and taxes and
depreciation divided by total assets. CASH represents cash
and short- term investment divided by total assets. DIV_DUM
is an indicator variable equal to 1 if firms pay cash dividends
and 0 otherwise. TANGIBILITY is computed as the ratio of net
property, plant, and equipment to total assets. LOG_ EMPLY is
the natural logarithm of the total number of employees at the
firm level. STRIKE is a dummy variable equal to 1 if a strike or
an industrial dispute led to lost working days and 0 otherwise.
We further control for corporate gover nance factors. Speci fically,
we incorporate an indicator variable, ATD, to account for man-
agerial entrenchment, which takes a value of 1 if the number
of anti- takeover devices in place exceeds two and 0 otherwise.
Board size (BOARDSIZE) is used to control board structure.
BOARDSIZE represents the total number of directors on the
board. Performance- based compensation (PBCP) is used to
control the executive compensation system. PBCP is a dummy
variable equal to 1if firms have a performance- based compen-
sation system and 0 otherwise. Finally, we use the legal origin
of a country to control for the legal system. COMMON_ L AW is
an indicator variable equals 1 if the origin of the commercial
law of a country is common law and 0 otherwise. All contin-
uous variables are winsorized at the 1st and 99th percentiles.
Importantly, our regressions also control for industry and year
fixed effects to address the correlation between workplace safety
and unobserved industry characteristics, as well a s time- va rying
macroeconomic conditions. Standard errors have been adjusted
for heteroskedasticity and are clustered by firm and year to ad-
dress serial correlations in the residuals.3
3.2.2 | Dynamic Workplace Safety Model
The parallel trend assumption is the backbone of the DID ap-
proach. Under this assumption, if the exogenous shock did not
occur, the response of treated and control firms would be par-
allel. Because the counterfactual situation is unobservable, we
cannot directly check the underlying assumption. As a prag-
matic alternative, researchers often examine pretreatment
trends in outcome variables to assess the plausibility of the par-
allel trends assumption.
In this study, we employ a dynamic DID regression model to in-
vestigate the trends in outcome variables before and after the
implementation of board gender quotas. The model specification
encompasses a timeline of 3 years preceding and 3 or more years
following the adoption of gender quotas. The following dynamic
DID regression is estimated:
The dependent variable in Equation(2) is firm- level total case rate
(TCR). We set the seven indicator variables
GENDER
_QUOTAS−
3
c
,
GENDER_QUOTAS−2
c
,
GENDER_QUOTAS−1
c
,
GENDER_QUOTAS0
c
,
GENDER_QUOTAS1
c
,
GENDER_QUOTAS2
c
, and
GENDER
_QUOTAS≥
3
c
to 1 if the firm is incorporated in country that passed the board
(1)
TCR
ict =
𝛼
+
𝛽GENDER_QUOTAS
ict +
𝛾X
i,
t
+Year FE+Industry FE+𝜀
i,t
(2)
TCR
ict =𝛼+𝛽1GENDER_QUOTAS−
3
c
+𝛽2GENDER_QUOTAS−2
c+𝛽3GENDER_QUOTAS−
1
c
+𝛽4GENDER_QUOTAS0
c+𝛽5GENDER_QUOTAS1
c
+𝛽6GENDER_QUOTAS2
c+𝛽7GENDER_QUOTAS≥3
c
+𝛾X
ict
+Year FE+Industry FE+𝜀
ict
9 of 26
gender quotas law 3 years ago, passed the board gender quotas law
2 years ago, passed the board gender quota law 1 year ago, passes
the board gender quotas law this year, and will pass board gen-
der quota law next year, will pass board gender quotas law next
2 years, and will pass board gender quotas law next 3 or more
years, respectively. We also use the same set of control variables
as in Equation(1).
3.3 | Sample Selection and Summary Statistics
Data on firm- level workplace safety come from Refinitiv
Environmental, Social, and Corporate Governance (ESG) da-
tabase (formerly ASSET4), a comprehensive database that
has been providing detailed information on workplace safety
since 2002. Our data on board gender quotas law are derived
from Catalyst(2021) and Deloitte(2019), as well as earlier stud-
ies (Ahern and Dittmar 2012; Smith 2014; Liao, Loureiro, and
Taboada2022). The annual financial accounting data we utilize
are obtained from Worldscope. We use corporate board charac-
teristics data from Refinitiv ESG database. We filter out firms
with missing International Securities Identification Number
(ISIN) codes and those lacking essential accounting variables
specified in Equation (1). In line with extant literature (Cohn
and Wardlaw 2016; Caskey and Ozel 2 017; and Amin, Kim,
and Lee 2021), we further exclude financial institutions (SIC
[Standard Industrial Classification] code 6000 –6999) and utility
firms (4900–4999) from the sample. We use ISIN code to merge
workplace safety and board data with financial accounting data.
Following this screening process, we find 13,124 firm- year ob-
servations from 2662 unique firms in 48 countries from 2002 to
2019. Twenty- seven percent of the firm- year observations come
from the United States (see Table1). However, our sample size
may vary in the additional analysis due to data restrictions. Our
sample period is determined by the staggered adoption of board
gender quotas policy as presented in Column 3 of Table1 and
availability of workplace safety data. Norway was the pioneer
in implementing the board gender quotas law in 2003, whereas
Germany followed suit in 2015.
Table2 presents summary statistics of all variables used in our
baseline regression. The average FIR M_SIZE in our study is
8.951, which translates to total assets of $7712 million. This in-
dicates that the sample firms are relatively large. The average
LOG_ EM PLY is 9.252, suggesting that a firm, on average, has
10,420 employees. The average LEVER AGE is 0.258, and the
average CASH_FLOW is 0.182. On average, firms hold cash
(CASH) equivalent to 7.4% of their assets, and 29.5% of the sam-
ple firms pay cash dividends (DIV_DUM). These firm- level fi-
nancial characteristics are consistent with those recorded by
Cohn and Wardlaw(2016). The average BOARDSIZE is 10.535,
and 47.4% of the sample firms have more than two anti- takeover
devices (ATD). On average, 73.4% of the firm- year observations
come from common law countries (COM MO N_LAW), and
86.9% of the sample firms have performance- based compensa-
tion systems for executives (PBCP).
Table 3 presents the pairwise correlations among the main
variables used in our baseline regression mode. It is crucial to
understand that correlation does not imply causation, and we
must conduct further analysis to establish causal relationships.
Notably, we find that board gender quotas are positively cor-
related with the workplace injury rate. Additionally, variables
such as firm size, cash flow, cash holding, and dividend dummy
exhibit negative correlations with workplace safety. Conversely,
leverage and tangibility show positive correlations with work-
place safety. In conclusion, the correlation between board gender
diversity and workplace safety is consistent with expectations,
suggesting that increased board gender diversity may enhance
workplace safety. This hypothesis will be explored further in the
multivariate analysis that follows.
4 | Workplace Safety and Board Gender Diversity
4.1 | Baseline Regression Results
We first investigate the relationship between board gender di-
versity and workplace safety using DID analysis based on the
staggered adoption of board gender quotas in multiple countries
in recent decades. The right- hand side variables in Equation(1)
are contemporaneous with the dependent variable. We estimate
different regression models, including diverse types of control
variables. Table 4 shows results from estimating Equation(1).
The dependent variable is the TCR in all models. Our primary
variable of interest is the board gender quotas law (GENDER_
QUOTAS ). In Columns 1–4, we include the year and indus-
try fixed. We begin with a model without control variables in
Column 1. The estimated coefficient on GENDER_QUOTAS is
positive and statistically significant at the 1% level. In Column
2, we add firm characteristics. The result holds: The coefficient
estimate on GENDER_QUOTAS is significantly and positively
related to TCR. We include both firm and corporate gover-
nance characteristics in Column 3. The estimated coefficient
on GENDER_QUOTAS remains positive and statistically signif-
icant at the 1% level. With the complete set of control variables in
Column 4, GENDER_QUOTAS are statistically significant at the
1% level (t- statistics = 3.02).
These results suggest that workplace injury rates increase fol-
lowing the adaptation of board gender quotas law for firms in
countries introduced that legislation. Economically, regard-
less of the specifications, the size of the coefficient estimates
on GENDER_QUOTAS vary from 0.2104 to 0.3950. In Column
4, for example, an increase of 1 SD in GENDER_QUOTAS will
increase TCR by 0.0 825 = (0.2104 × 0.392), wh ich represents
5.79% = (0.0825/1.425) of its mean, following the passage of
board gender quotas. These results support our Hypothesis 1
that board gender quotas law increase workplace injury rates.
Furthermore, all adj. R- squared values increase to different de-
grees, indicating the incremental explanatory power of board
gender quotas on workplace injury rates.
Our baseline model, as presented in Table 4, not only aligns
with the literature on how firm characteristics are associated
with workplace safety but also provides a strong validation of
prior studies. We find that firm SIZE, CASH_FLOW, and CASH
are negatively and significantly related to injury rates, a find-
ing that is in perfect harmony with the research of Cohn and
Wardlaw(2016). Similarly, our findings on TANGIBILITY being
positively related to injury rates further confirm these papers.
The positive coefficient on STRIKE, indicating more cases in
10 of 26 Corporate Governance: An International Review, 2024
TABLE | Sample distribution by country and year of gender quotas law introduction.
Country name Frequency Percent Year of quotas law
Argentina 15 0.11 —
Australia 1050 82012
Austria 66 0.5 2011
Belgium 70 0.53 2011
Bermuda 18 0.14 —
Brazil 146 1.11 —
Canada 786 5.99 —
Chile 83 0.63 —
China 183 1.39 —
Colombia 58 0.44 2000
Denmark 36 0.27 2012
Finland 121 0.92 2005
France 405 3.08 2011
Germany 217 1.65 2015
Greece 52 0.4 2000
Hong Kong 292 2.22 —
Hungary 25 0.19 —
India 216 1.65 2013
Indonesia 76 0.58 —
Ireland 49 0.37 —
Israel 41 0.31 1999
Italy 196 1.49 2011
Japan 1127 8.58 —
Korea (South) 353 2.69 —
Luxembourg 27 0.21 —
Malaysia 173 1.32 2011
Mexico 71 0.54 —
Netherlands 241 1.84 2011
New Zealand 86 0.65 —
Norway 173 1.32 2003
Peru 14 0.11 —
Philippines 30 0.23 —
Poland 112 0.85 —
Portugal 43 0.33 —
Qatar 15 0.11
Russian Federation 198 1.51 —
Saudi Arabia 18 0.14 —
Singapore 128 0.97 —
(Continues)
11 of 26
firms where employees go on strike, is also a reassuring align-
ment with the research of Hope etal.(2022).
We also find that many corporate governance measures sig-
nificantly relate to workplace injury rates. In Column 4, we
observe that managerial entrenchment, as measured by anti-
takeover devices (ATD), exhibits a positive association with
workplace injury rates. This suggests that firms with more
entrenched executives tend to experience higher rates of work-
place injuries. Similarly, the presence of a performance- based
compensation system for executives is positively correlated
with workplace injury rates. This indicates that firms pri-
oritizing financial performance over workplace safety may
experience higher injury rates. However, the coefficient esti-
mate for board size is negative and statistically insignificant.
Finally, the negative relationship between workplace injury
rates and firms in countries with a common legal system high-
lights the importance of regulatory frameworks and their im-
pact on workplace safety. In summary, our baseline regression
findings presented in Table4 suggest that the implementation
of board gender quotas lead to higher firm- level injury rates,
aligning with our hypothesis (Hypothesis1).
Country name Frequency Percent Year of quotas law
South Africa 436 3.32 —
Spain 279 2.12 2007
Sweden 76 0.58 —
Switzerland 112 0.85 —
Taiwan 217 1.65 —
Thailand 55 0.42 —
United Arab Emirates 26 0.2 2012
Turkey 121 0.92 —
United Kingdom 1246 9.49 —
United States 3552 27.05 —
Total 13,124 100.00
Note: This t able presents the sample distribution of 48 cou ntries across the globe and the first- time board gender quot as introduction year. The sample period is f rom
2002 to 2019.
TABLE | (Continued)
TABLE | Summary statistics.
Var iable Observations Mean Std . d ev. Min Max
TCR 13,124 1.425 2.033 0.000 44.926
GENDER _QUOTAS 13,124 0.190 0.392 0.000 1.000
FIRM_SIZE 13,124 8.951 1.572 5.407 13.174
LEVERAGE 13,124 0.258 0.159 0.000 0.705
CASHFLOW 13,124 0.182 0.153 −0 .217 0.693
CASH 13,124 0.074 0.082 0.000 0.982
DIV_DUM 13,124 0.295 0.456 0.000 1.000
TANGIBILITY 13,124 0.364 0.262 0.002 0.936
LOG_ EM PLY 11,947 9.252 1.598 1.946 14.648
STRIK E 13,124 0.706 0.456 0.000 1.000
ATD 13,124 0.474 0.499 0.000 1.000
BOARDSIZE 13,094 10.535 3.378 1.000 34.000
PBCP 13,124 0.869 0.337 0.000 1.000
COMMON_ L AW 13,124 0.73 4 0.442 0.000 1.000
Note: This t able reports the summary statistics for the variables used in the baseline regression. The sample per iod is from 2002 to 2019. All continuous variables are
winsor ized at the 1st and 99th percentiles. The detailed variable definitions are reported in AppendixA.
12 of 26 Corporate Governance: An International Review, 2024
TABLE | Correlation matrix.
Var iables 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 TCR 1
2 GENDER_QUOTAS 0.04*** 1
3 FIRM_SIZE −0.19*** −0.09*** 1
4 LEVERAGE 0.05*** 0.01 0.11*** 1
5 CASHFLOW −0.08*** 0.01 0.11*** 0.01 1
6 CASH −0.10*** −0.01*−0.25*** −0.20*** −0.12*** 1
7 DIV_DUM −0.03*** 0.14*** 0.15*** 0.01 0.02*−0.08*** 1
8 TANGIBILITY 0.06*** −0.04*** −0.07*** 0.19*** 0.35*** −0.24*** −0.05*** 1
9 LO G_ EMPLY −0.04*** −0.10*** 0.58*** 0−0.30*** −0.03*** 0.07*** −0.29*** 1
10 STRIKE −0.02** 0.03*** 0.06*** −0.01 −0.02*** 0.04*** 0.08*** 00.09*** 1
11 ATD −0.04*** −0.08*** 0.10*** 0.09*** 0.03*** 0.03*** −0.11*** 0.01 0.10*** 0.04*** 1
12 BOARDSIZE −0.07*** −0.06*** 0.48*** 0.07*** −0.01 −0.12*** 0.12*** −0.07*** 0.34*** 0.06*** 0 1
13 PBCP 0.10*** 0.06*** −0.04*** 0.02*** 0.07*** −0.03*** −0.06*** 0.05*** −0.02** 0.07*** 0.28*** −0.05*** 1
14 COM MON_LAW 0−0.35*** −0.09*** −0.05*** −0.03*** 0.02** −0.24*** 0.10*** −0.03*** −0.01 0.24*** −0.13*** 0.13*** 1
Note: This t able presents pairw ise correlations bet ween dependent variable and independent variables used in the ba seline regression models. The sample period is 2002–2019. All continuous variables are winsorized at the 1st and
99th percentile. All variables are defined in AppendixA. The symbols ***, **, and * indicate statistical significance at 1%, 5% , and 10% level, respectively.
13 of 26
Although we use several firm attributes and country charac-
teristics in Column 4, it might still omit some unknown char-
acteristics that affect board gender diversity and workplace
safety. To ease this concern, we further use country- fixed
effect regressions to control for the influence of unknown time-
invariant country- level factors (i.e., culture, regulations, or other
is sues).4 We report the corresponding results in Column 5. We
continue to estimate the baseline model (Column 4) except that
TABLE | The ef fect of board gender diversity on workplace safety.
Var iables
(1) (2) (3) (4) (5)
TCR TCR TCR TCR TCR
GENDER _QUOTAS 0.3950*** 0.3172*** 0.3057*** 0.210 4*** 0.1319**
(5.98) (4 .9 2) (4.67) (3.02) (2.15)
FIRM_SIZE −0.2787*** −0.2758*** −0.2689*** −0.3579***
(−10.58) (−11.5 4) (−11.22) (−21.16)
LEVERAGE 0.1846 0.1329 0.0849 0.3557***
(1.59) (1.23) (0.79) (2 .95)
CASHFLOW −0.1291 −0.2007 −0.2619*−0.8403***
(−0.81) (−1.35) (−1.77) (−6.36)
CASH −2.5353*** −2.4574*** −2.4892*** −2.5231***
(−10. 51) (−10.61) (−10.77) (−12.01)
DIV_DUM −0.0538 −0.0464 −0.0759 −0.0448
(−0.94) (−0.88) (−1.50) (−0.86)
TANGIBILITY 0.2082*0.2347** 0.2742** 0.6050***
(1.87) (2.10) (2.53) (8 .10)
LOG_ EM PLY 0.0366 0.0329 0.0234 0.1969***
(1. 51) (1.42) (0.99) (11.44)
STRIK E 0.1512*** 0.1263*** 0.1310*** 0.0330*
(3. 56) (3.14) (3.23) (1.86)
ATD 0.0004 0.0362 0.1998
(0.01) (0.62) (0.26)
BOARDSIZE −0.0003 −0.0019 −0.0054
(−0.04) (−0.26) (−0. 89)
PBCP 0.5256*** 0.5478*** 0.0195*
(6. 63) (7.14) (1.76)
COMMON_ L AW −0.2335*** −0.2328*
(−3.37) (−1.9 4)
Constant 5.3313*** 7.1 385*** 6.7213*** 7.0451*** 3.0627***
(6.77) (6. 81) (6.51) (6.8 8) (8.29)
Yea r FE Yes Yes Yes Yes Ye s
Industry FE Yes Yes Yes Yes No
Country FE No No No No Yes
Observations 13,124 11,943 11,910 11,910 11,910
Adjusted R20.183 0.214 0.220 0.222 0.178
Note: This t able reports the baseline regression results on the relationship bet ween board gender diversity and workplace safety. The dependent variable is the
firm- level injur y rates (TCR). GENDER_QUOTAS is an indicator var iable that takes a value of 1 for the years in which board gender quotas law has adopted in firm's
country of incorporation and 0 otherw ise. The t- statistics, which are based on heteroscedasticity- robust standard errors clustered by firm and year, are shown in
parentheses. The sample period is 20 02–2019. All variables are defined in AppendixA. The symbols ***, **, and * represent statistical significance at 1%, 5%, and 10%
level, respectively.
14 of 26 Corporate Governance: An International Review, 2024
we replace industry dummies with country dummies. We find
that the coefficient estimate for GENDER_QUOTAS remains
positive and statistically significant at the 5% level or better, sug-
gesting that our results are robust to the inclusion of this alter-
native fixed effect.
4.2 | Dynamic Workplace Safety Model
In this section, we present a trend analysis for our variable of
interest, GENDER_QUOTAS. If GENDER_QUOTAS, are in-
deed capturing a significant change in a firm's future work-
place safety that relates to a country's adoption of board gender
quotas, we should theoretically find that introducing a lagged
GENDER _QUOTAS indicator variable into our regression pro-
duces insignificant results. If this is not the case, then it puts the
validity of our baseline results into question, as there would be
evidence of a pretreatment trend in workplace safety just before
the adoption of the board gender quotas. So, the DID parallel
trend assumption would not be valid. To conduct this test, we
replace GENDER_QUOTAS in Equation(1) with new indicator
variables in Equation(2) (see Section3.2.2).
The estimation results presented in Table5 demonstrate that the
coefficients of
GENDER_QUOTAS−3
c
,
GENDER_QUOTAS
−
2
c
,
GENDER_QUOTAS
−
1
c
, and
GENDER
_QUOTAS
0
c
are stati stically
insignificant, whereas the coefficients of
GENDER_QUOTAS1
c
,
GENDER_QUOTAS2
c
, and
GENDER
_QUOTAS
≥3
c
are positive
and statistically significant at the 10% level or better, suggesting
that the increase in TCR occurs after the adoption of board gender
quotas law but not before.5 This evidence supports our assump-
tion of pretreatment parallel trends and eliminates time trends as
a possible explanation for our finding.
4.3 | Sensitivity Test
We conduct an array of additional robustness tests. The results
are reported in Data S1. First, we use four alternative measures
of workplace safety: LOG_TOTA L_CASE, DA RT, LOG _ DAW F,
and IND_TCR. Second, we use six alternative measures of board
gender diversity. Third, we use five additional control variables
in the baseline regressions: ROA, TUR NOVER, DUALITY,
BOARD_TENURE, and PRIVATE_CREDIT. The findings re-
main consistent across these tests.
5 | Cross- Sectional Tests
In this section, we execute four cross- sectional analyses of the
relation between board gender diversity and workplace safety.
5.1 | Board Gender Diversity, Financial Constraint,
and Workplace Safety
In this section, we investigate how financial constraints might
influence the relationship between board gender diversity and
workplace safety. We hypothesize that in f irms with financial con-
straints, board gender diversity could inadvertently worsen work-
place safety. This is because financially constrained firms often
struggle to allocate resources effectively and may lack sufficient
funds for crucial safety improvements, opting instead to focus
on areas critical for short- term survival. The diverse perspectives
brought by gender- diverse boards, though valuable, could com-
plicate decision- making around these limited resources, poten-
tially leading to poorer safety outcomes. To rigorously test this
hypothesis, we divide firms into financially constrained and un-
constrained subgroups using four different measures of financial
constraints. The first measure is the size- age (SA) index (Hadlock
and Pierce2010), which we calculate as follows:
where AT is the natural logarithm of total assets and Age is the
difference between the last year in sample and the firm's date
of incorporation. HIGH_SA is an indicator variable equal to 1 if
(3)
SA
=−0.737 ×
AT
+0.043×
AT2
−0.040×
Age
TABLE | Dynamic workplace safety model.
Var iables
(1) (2)
TCR TCR
GENDER_QUOTAS−3
c
0.5367
(1.26)
GENDER
_QUOTAS−
2
c
0.7257
(1.00)
GENDER_QUOTAS
−
1
c
0.1065 0.6136
(1.05) (0.80)
GENDER_QUOTAS0
c
0.2310 0.2767
(1.52) (1. 55)
GENDER
_QUOTAS+
1
c
0.5330** 0.3463**
(2.11) (2.20)
GENDER
_QUOTAS+
2
c
0.3641**
(2.29)
GENDER_QUOTAS≥+3
c
0.3106*
(1.68)
Controls Yes Yes
Yea r FE Yes Yes
Industry FE Yes Ye s
Observations 11,910 11,910
Adjusted R20.244 0.092
Note: Th is table presents the re sults from the dyna mic workplace safet y
model, where the dep endent variable is the fir m- le vel injury rate (TCR). The
t- statistics are shown in parentheses and ar e based on heteroskedasticity-
robust standard errors clustered by fi rm and year. The anal ysis covers
the period f rom 2002 to 2019 and i ncludes two- digit SIC industr y and
year fi xed effects. T he seven indicator var iables
GENDER_QUOTAS
−
3
c
,
GENDER_QUOTAS
−
2
c
,
GENDER_QUOTAS
−
1
c
,
GENDER_QUOTAS0
c
,
GENDER_ QUOTAS1
c
,
GENDER_ QUOTAS2
c
, and
GENDER_QUOTAS
≥
3
c
are set
to 1 if the firm is incor porated in a country that passe d the gender quotas policy
in 3 years ago, passed the board gender quotas law 2 years ago, pas sed the board
gender quotas law 1 year ago, pa sses the board gender quota s law this year, and
will pa ss board gender quotas law nex t year, will pass board gender quotas law
next 2 years, and wi ll pass board gender quota s law next three or more yea r,
respect ively. All variables are def ined in AppendixA. T he symbols ***, **, and *
indicate st atistical significance at 1%, 5%, and 10% level, respectively.
15 of 26
firm's SA index is above the sample median in a given year and
0 otherwise. The second measure is the KZ index (Kaplan and
Zingales1997), which we calculate as follows:
where Cash flow is the income before extraordinary items plus
depreciation; K is property, plant, and equipment; Cash is the
cash and short- term investment; Dividends is the c ash dividend
paid; and Debt is the total debt. HIGH_ KZ is a dummy variable
equal to 1 if firm's KZ index is above the sample median in a
given year and 0 otherwise. Firms are classified as financially
constrained if they have either HIGH_KZ = 1 or HIGH_SA = 1,
indicating higher financial constraints. Firms not meeting
these criteria are considered financially unconstrained.
Following previous literature (Gatchev, Pulvino, and
Tarhan 2010; Arslan- Ayaydin, Florackis, and Ozkan2014), we
also use cash holding and leverage as proxy for financial con-
straint.6 Cash holding (CASH) is defined as the ratio of cash and
cash equivalents to total assets. LEVERAGE is the ratio of total
debt to total assets. Firms with low CASH and high LEV ER AGE
are likely to have financial constraints. The remaining firms are
considered financially unconstrained. We adopt the regression
model in Equation (1) by including the financial constraints
(4)
KZ index =−1.001909×
Cash flow
K+0.2826389
×Tobin Q+3.139193 ×Debt
Total Capital
+(−39.3678×Dividends
K)
+
(
−1.314759×Cash
K)
TABLE | Conditioned on the financial constraints.
Var iables
(1) (2) (3) (4)
TCR TCR TCR TCR
GENDER _QUOTAS 0.1520** 0.2110*** 0.1567** 0.2794***
(2.05) (2 .72) (1.99) (3 .31)
HIGH_ SA 0.1033**
(2. 27)
GENDER _QUOTAS × H IGH_S A 0.1522*
(1.84)
HIGH_ KZ 0.0363*
(1.84)
GENDER _QUOTAS × H IGH_K Z 0.0061**
(2.07)
CASH −0.6281***
(−10.49)
GENDER _QUOTAS × C ASH −0.0639**
(−2.16)
LEVERAGE 0.0953
(0. 89)
GENDER _QUOTAS × L EVER AGE 0.2622*
(1.82)
Controls Yes Yes Yes Ye s
Yea r FE Yes Ye s Yes Yes
Industry FE Yes Ye s Yes Yes
Observations 11,910 11,910 11,910 11,910
Adjusted R20.223 0.222 0.222 0.224
Note: This t able report the results of the relation between board gender diversity and workplace safet y conditional on financial constraints. The dependent variable
is the firm- level injury rate (TCR). GENDER_QUOTAS is an indicator var iable that takes a value of one for the years in which board gender quota s law has adopted
in firm's country of incorporation and 0 ot herwise. HIGH_SA is an indicator variable equal to 1 i f firm's SA index is above the sample median in a given year and 0
otherwise. HIGH_ KZ is a dummy variable equal to 1 if firm's KZ index is above the sample median in a given year and 0 other wise. CASH is the ratio of cash to tota l
assets . We include two- digit SIC industry and year fixed effects. Other control variables are included in the regression models but omitted for brevit y. The t- statistics
(in parentheses) are based on robust standard errors clustered by firm and year to control for time- series correlation in the error term. All variables are defined in
AppendixA. F ull version of the table is presented in Data S1. The symbols ***, **, and * represent signif icance at the 1%, 5%, and 10% levels, respectively.
16 of 26 Corporate Governance: An International Review, 2024
proxies and the interaction between GENDER_QUOTAS and fi-
nancial constraints proxies. The results are reported in Table6.
We observe that the estimated coefficients for the interactions
GENDER _QUOTAS×HIGH_SA, GENDER _QUOTAS×HIGH_
KZ, and GENDER _QUOTAS × LE VER AGE are positive and sta-
tistically significant (p value at 10% or better). Conversely, the
coefficient for GENDER_QUO TAS × CASH is negative and sta-
tistically significant at the 5% level. These results indicate that
the impact of board gender diversity on workplace safety is more
pronounced in firms experiencing financial constraints, align-
ing with Hypothesis2.
5.2 | Board Gender Diversity, Institutional Quality,
and Workplace Safety
We examine the potential effects of institutional quality on the
relationship between board gender diversity and workplace
safety. Country- level institutional quality can significantly
influence the relationship between board gender quotas and
workplace safety. Higher institutional quality typically entails
stronger legal frameworks, regulatory enforcement, and gov-
ernance structures, which may promote greater accountability
and transparency within organizations. In such contexts, boards
of directors, including those with gender diversity, may operate
more effectively in overseeing workplace safety measures and
ensuring compliance with regulations.
We proxy for institutional quality, which is a measure of the ef-
fectiveness and efficiency of a country's institutions using three
key indicators: regulatory quality, the rule of law, and political
stability. Regulator y quality (RQ) ref lects perceptions of the abil-
ity of the government to formulate and implement sound policies
and regulations that permit and promote private sector develop-
ment. RQ scores range from approximately −2.5 (weak) to 2.5
(strong) governance performance. The rule of law (RL) refers to
the extent to which agents believe in and adhere to social rules,
especially in relation to contract enforcement, property rights,
the police, and the courts, as well as the likelihood of crime and
violence. For each country, RL scores are assigned from −2.5
to 2.5, where a higher score indicates greater adherence to the
rule of law. Political stability (PS) captures perceptions of the
TABLE | Conditioned on institutional quality.
Var iables
(1) (2) (3)
TCR TCR TCR
GENDER _QUOTAS 0.2306*0.2088*0.0552*
(1.76) (1.86) (1.70)
RQ −0.6109***
(−9.93)
GENDER _QUOTAS × RQ −0.2565**
(−2.15)
RL −0.5038***
(9.29)
GENDER _QUOTAS × R L −0.2556**
(−2.22)
PS −0.0027
(−1.33)
GENDER _QUOTAS × PS −0.0005*
(−1.92)
Controls Yes Ye s Yes
Yea r FE Yes Yes Yes
Industry FE Yes Yes Ye s
Observations 11,868 11,868 11,868
Adjusted R20.108 0.104 0.088
Note: This t able reports the results of the relation bet ween board gender diversit y and workplace safety conditional on in stitutional quality. The dependent variable is
the firm- level injury rate (TCR). GENDER_QUOTAS is an indicator variable that takes a value of 1 for the years in which board gender quotas law has adopted in firm's
country of incorporation and 0 otherw ise. Regulatory quality (RQ ) reflects perceptions of the ability of the govern ment to formulate and implement sound policies and
regulations that perm it and promote private sector development. Rule of law (RL) refers to the extent to which agents believe in and adhere to social rules, especially in
relation to contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence. Political stability (PS) index is computed by
considering four factors: government stability, internal conf lict (i.e., violent demonstrations and riots), war, violent crimes, and ethnic conflict (e.g., racial tension and
riots). We include two- digit SIC industry and year fi xed effects. Other control variables are included in the regression models but omitted for brevity. The t- statistics
(in parentheses) are based on robust standard errors clustered by firm and year to control for time- series correlation in the error term. All variables are defined in
AppendixA. F ull version of the table is presented in Data S1. The symbols ***, **, and * represent signif icance at the 1%, 5%, and 10% levels, respectively.
17 of 26
possibility that the government will be destabilized or over-
thrown by unconstitutional or violent means, including polit-
ically motivated violence and terrorism. The political stability
(PS) index is computed by considering four factors: government
stability, internal conflict (i.e., violent demonstrations and riots),
war, violent crimes, and ethnic conflict (e.g., racial tension and
riot s). PS scores range from −2.5 to 2.5, with a higher (lower)
score indicating greater (lesser) political stability.
We obtain data on country- level institutional quality from World
Bank's Worldwide Governance Indicators database.7 We adopt
the regression model in Equation(1) by including the proxy of
institutional quality and an interaction between institutional
and board gender quotas. This analysis aims to examine the
potential effect of institutional quality on the relation between
board gender diversity and workplace safety. The results are re-
ported in Table7. Column 1 in Table7 shows the results when
we interact GENDER_QUOTAS with RQ. Column 2 shows the
results when GENDER_QUOTAS interact with RL. Column 3
shows the results when GENDER_QUOTAS interact with PS.
We find that the coefficients on the interaction terms are nega-
tive and statistically significant (p value at 10% or better), which
suggests that the positive relation between board gender diver-
sity and workplace injury rate is weaker for firms in a country
with strong institutional quality environment, consistent with
the view that a strong institutional environment may already
provide sufficient protection and regulatory frameworks for
workers.
5.3 | Board Gender Diversity, Labor Union,
and Workplace Safety
In this section, we explore the potential effect of employees' bar-
gaining power in the relationship between board gender quotas
and workplace safety. Labor unions can act as a critical inter-
mediary between employees and management in safeguarding
workplace safety, potentially mitigating any adverse effects
that may arise from decisions related to board gender quotas.
Kaufman (2005) contends that labor unions have strong bar-
gaining power and usually negotiate workloads and workplace
safety issues. Morantz(2013) argues that unionized firms have
fewer workplace injuries. Therefore, we predict that the effect
of board gender quotas on workplace safety is weaker when em-
ployees have more bargaining power.
As it is difficult to obtain firm- level unionization data from in-
ternational markets, we collect country- level trade union den-
sity data from OECD library to test this hypothesis.8 The trade
union density is characterized as the number of net union mem-
bers (i.e., excluding those not in the labor force, unemployed,
and self- employed) as a percentage of the number of employees.
We define HIGH_UNION_MEMBERSHIP as a dummy variable,
taking the value of 1 if a firm is located in a country with union
membership above the sample median in a given year and 0
otherwise. Our regression model in Equation(1) incorporates a
proxy of bargaining power and an interaction between bargain-
ing power and board gender quotas. This approach allows us to
assess the differential effect of board gender quotas in firms with
low versus high labor union bargaining power.
We report the result of this test in Table 8. The coefficient
on interaction term GEN DER_QUOTAS × HIGH_U NION_
MEMBERSHIP is negative and significant at the 10% level,
suggesting that the effect of board gender quotas on workplace
injury rates is weaker for firms in a country with high union
membership. This finding supports the argument that strong
employee bargaining power may play a more substantial role in
enhancing workplace safety, regardless of the gender composi-
tion of the board.
TABLE | Conditional on union membersh ip.
Var iables
(1)
TCR
GENDER _QUOTAS 0.1782**
(2.31)
HIGH_UNION_MEMBERSHIP −0.0387
(0.60)
GENDER _QUOTAS × HIGH_UNION_ MEMBERSH IP −0.1842*
(−1.77)
Controls Yes
Yea r FE Yes
Industry FE Yes
Observations 10,109
Adjusted R20.248
Note: This t able presents the results f rom estimating the ef fect of board gender diversity on workplace safety conditional on union membership. The dependent variable
is the total case rates (TCR). GENDER_QUOTAS is an indicator variable that takes a value of 1 for the years in which board gender quotas law has adopted in f irm's
country of incorporation and 0 otherw ise. HIGH_UNION_MEMBERSHIP is a dummy variable that equals 1 i f a firm is from a country with union membership above
the sample median in a given year and 0 otherwise. We include two - digit SIC industr y and year fixed effects. Other control variables are included in the regression
models but omitted for brevity. The t- statistics (in parentheses) are based on robust standard errors clustered by firm and year. All variables are defined in AppendixA.
Full version of the table is presented in Data S1. T he symbols ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.
18 of 26 Corporate Governance: An International Review, 2024
5.4 | Board Gender Diversity, Employee Workload
and Workplace Investments, and Workplace Safety
Previous research suggests that workplace injuries are associ-
ated with high employee workloads and inadequate investment
in workplace safety (Cohn and Wardlaw 2016; Caskey and
Ozel 2017). In this context, we examine whether board gender
quotas can potentially influence workplace injury rates through
changes in employee workload and investment in safety mea-
sures. Following the methodology of Caskey and Ozel (2017),
we utilize net sales per employee as a measure of employee
workload. We define HIWL as an indicator variable set to 1
when a firm's net sales per employee exceed the median for a
given year and 0 otherwise.
In addition, considering that firms often classify safety- related
expenditures and training costs as part of SG&A (Cohn and
Wardlaw2016; Caskey and Ozel2 017), we use selling, general,
and administrative expenses (SG&A) per employee and SG&A
to sales ratios to measure workforce investment. WFINV1_LOW
is an indicator variable equal to 1 if firms' SG&A expenses per
employee are below the sample median in a given year and 0
TABLE | Conditional on labor force workload and workplace investment.
Var iables
(1) (2) (3) (4) (5)
TCR TCR TCR TCR TCR
GENDER _QUOTAS 0.2894*** 0.2117*** 0.1951** 0.1752** 0.2034**
(3.02) (3.05) (2.49) (2 .3 0) (2.57)
HIWL 0.0152
(0.34)
GENDER _QUOTAS × H IWL 0.1498*
(1.74)
IND_HIWL −0.0376
(−1.0 4)
GENDER _QUOTAS × I ND_H IW L 0.0911**
(2.08)
WFINV1_LOW 0.4779***
(7.27)
GENDER _QUOTAS × W FIN V1_LOW 0.0528*
(1.75)
WFINV1_INDLOW −0.2463
(−0.92)
GENDER _QUOTAS × W FIN V1_IN DLOW 0.3260*
(1.71)
WFINV2_LOW 0.2603***
(5.53)
GENDER _QUOTAS × * WFIN V2_ LOW 0.0149*
(1.87)
Control Yes Yes Yes Yes Yes
Yea r FE Yes Yes Yes Yes Yes
Industry FE Yes Yes Yes Yes Yes
Observations 11,910 11,910 11,910 11,910 11,910
Adjusted R20.222 0.222 0.230 0.225 0.225
Note: This t able reports the results from estimating the effect of board gender diversity on workplace safety conditional on employee workload and workplace
investment. The dependent variable is the total case rates (TCR). GENDER_QUOTAS is an indicator variable that takes a value of 1 for the years in which board
gender quotas law has adopted in f irm's country of incorporation and 0 other wise. We include two- digit SIC industry and year fixed effects. Other control variables are
included in the regression models but omitted for brev ity. The t- statistics (in parentheses) are based on robust standard error s clustered by f irm and year. All variables
are defined in AppendixA. Full version of the table is presented in Data S1. The symbols * **, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.
19 of 26
TABLE | Channel test s.
Var iables
Firm performance Board effectiveness Risk aversion
(1) (2) (3) (4) (5) (6)
ROA Tobin's Q GSCORE Board skills
Health and
safety training
Capital
expenditure
GENDER _QUOTAS −0.0161*** −0.0566** −2.4918*** −2.3127*** −0.3341** −0.0020*
(−5.02) (−2.0 8) (−4.91) (−4.06) (−2.16) (−1.79)
FIRM_SIZE −0.0153*** −0.4319*** 4.1413*** −0.6431*** −0.0803 −0.0065***
(−6.19) (−18.89) (25.86) (−3.56) (−1.2 4) (−17.78)
LEVERAGE −0.1991*** −0.2323*** −0.8884 −2.9727** 0.7064** −0.0240***
(−23.73) (−3.2 8) (−0.78) (−2.32) (2.13) (−9.34)
CASHFLOW 0.2791*** 1.1119*** 4.8940*** 9.9136*** −0.1872 0.0639***
(33.37) (15.20) (3.61) (6.51) (−0.47) (20.76)
CASH 0.0771*** 0.6501*** 3.2207 13.3364*** −0.9571 0.0211***
(6.12) (6. 22) (1.42) (5.17) (−1.47) (4.10)
DIV_DUM −0.0065*** −0.0678*** −1.0875*** −3.4191*** −0.0360 −0.0036***
(−2.60) (−3.16) (−2.73) (−7.63) (−0.36) (−3.95)
TANGIBILITY −0.0489*** −0.2215** 3.1074*** −3.3288*** 0.4712*0.1204***
(−4.87) (−2.48) (4.06) (−3.87) (1.68) (69.45)
LOG_ EM PLY 0.0075*** 0.1204*** 0.7545*** −0.0526 0.9158*** 0.0049***
(3.19) (5.83) (4.96) (−0.31) (16.03) (14. 39)
STRIK E −0.0091*** −0.0048 3.8609*** 1.1956*** 0.4070*** −0.0 014
(−4.85) (−0.29) (9.81) (2.68) (3.03) (−1.60)
ATD −0.0109*** −0.0251 −0.5878 −1.8926*** 0.2691*** −0.0018**
(−5.53) (−1.31) (−1.51) (−4.35) (2.62) (−2.18)
BOARDSIZE 0.0007** 0.0146*** −1.1554*** −0.5460*** −0.0212 0.0001
(1.9 6) (4.89) (−19.93) (−8.12) (−1.56) (0.77)
PBCP 0.0013 0.0004 13.6689*** −4.5097*** 0.5523*** −0.0005
(0.40) (0 .01) (25.02) (−6.98) (3.90) (−0.3 8)
COMMON_ L AW 0.0147*0.8331*** 5.0210*** 19.2580*** 0.0146 0.0004
(1.84) (19.63) (11.0 8) (37.41) (0.12) (0.40)
Constant 0.1662*** 3.9334*** 8.3241*** 34.0714*** 1.2551 0.0235***
(7.52) (19.63) (2.81) (6.53) (0.90) (6. 85)
Yea r FE Yes Yes Ye s Yes Yes Ye s
Industry FE Yes Yes Ye s Ye s Yes Yes
Observations 11,910 9954 11,914 11,478 1954 11,914
Adjusted R20.649 0.138 0.182 0.204 0.631 0.389
Note: This table presents the result of dif ference- in- differences regression that examines the impact of board gender quotas on firm per formance (Columns 1 and 2),
board effectiveness (Columns 3 and 4), and risk aversion (Columns 5 and 6). GENDER_QUOTAS is an indicator variable that takes a value of 1 for the years in which
board gender quotas law has adopted in firm's country of incorporation and 0 ot herwise. We include two- dig it SIC industr y and year fixed ef fects. The t statistics
(in parentheses) are based on robust standard errors clustered by firm and year. All variables are def ined in AppendixA. The symbols * **, **, and * represent
significance at the 1%, 5%, and 10% levels, respectively.
20 of 26 Corporate Governance: An International Review, 2024
otherwise. WFINV2_LOW is a dummy variable equal to 1 if the
firm's SG&A to sales ratio is below the median in a given year
and 0 otherwise. Firms characterized by W FINV1_LOW = 1 or
WFINV2_LOW = 1 are more likely to underinvest in workplace
sa fety.
We have modified the regression model in Equation(1) to in-
clude proxies for employee workload and workplace invest-
ments, along with an interaction between these variables and
board gender quotas. This approach allows us to assess how
board gender quotas impact firms differently depending on
their levels of employee workload and investment in workplace
safety. The results are reported in Table 9. In Column 1, the
interaction term GENDER _QUOTAS × H IWL is shown to be
positive and statistically significant at the 10% level, indicating
that firms with greater gender- diverse boards experience higher
workplace injury rates, especially when employee workload is
high. In Columns 2 and 3, we observe that the estimated coef-
ficients of GENDE R_QUOTAS × W FIN V1_LOW and GENDER_
QUOTAS × WFIN V2_ LOW are both positive and statistically
significant at the 10% level. These findings suggest that the in-
jury rate tends to increase when workplace safety investment is
low relative to either the number of employees or sales.
We also employ industry- adjusted measures to ensure the ro-
bustness of our findings.9 We measure industry- adjusted em-
ployee workload as the difference between a firm's net sales
per employee and the average net sales per employee for all
firms within the same industry. We define IND_HIWL set to
1 if industry adjusted employee workload is above the sam-
ple median in a given year and 0 otherwise. Similarly, we
adjusted SG&A per employee by calculating the difference be-
tween a firm's SG&A per employee and the industry average.
The WFINV1_INDLOW dummy is set to 1 if this industry-
adjusted figure falls below the sample median and 0 other-
wise. These two industry- adjusted measures are interacted
with GENDER_QUOTAS, and the results are presented in
Table9. We find that the coefficients of the interaction terms
are positive and statistically significant at the 5% level. This
suggests that even after accounting for industry- specific ad-
justments in workload and spending on SG&A, the presence
of gender quotas continues to be associated with higher injury
rates, lending further credence to our findings.
Overall, the results presented in Table9 reveal that the relation-
ship between board gender quotas and workplace safety is more
pronounced in firms with higher employee workloads and lower
investments in workplace safety.
6 | Channel Tests
In formulating our hypothesis, we suggest that board gender
quotas increase workplace injury rate through three distinct
channels: firm performance, board effectiveness, and risk aver-
sion. This section offers empirical evidence to explore these eco-
nomic channels.10
First, drawing on prior research (Fauver et al. 2 017; Huang,
Kerstein, and Wang 2018; Mollick and Haidar 2024), we
use return on assets (ROA) and Tobin's Q as proxies for firm
performance. In Columns 1 and 2 of Table10, our findings reveal
a negative association between board gender quotas and firm
performance, suggesting that board gender quotas adversely im-
pact firm performance. Second, we evaluate board effectiveness
using a corporate governance score (GSCORE) and the percent-
age of board members with either an industry- specific or strong
financial background (BOARD_SKILLS), as per the Refinitiv
ESG database. We argue that firms with higher governance
scores or specialized board skills are more effective. In Columns
3 and 4 of Table10, our results show that board gender quotas
are negatively associated with these measures of effectiveness.
This suggests that firms with higher board gender diversity may
experience reduced board effectiveness. Third, we assess risk
aversion using two indicators: employee health and safety train-
ing (Health_Safety_training) and capital expenditure intensity
(CPX). We consider firms that offer less employee health and
safety training and have lower capital expenditure intensity to
be more risk averse. In Columns 5 and 6 of Table10, we find
that both employee and safety training and capital expenditures
intensity are negatively associated with the board gender quotas.
This suggests that firms with higher board gender diversity tend
to invest less in employee health and safety training and capital