Do employee relation responsibility and culture matter for firm value?
International evidence

Jaeho Lee, Hakkon Kim

PII: S0927-538X(16)30201-3
DOI: doi: 10.1016/j.pacfin.2016.10.006
Reference: PACFIN 888

To appear in: Pacific-Basin Finance Journal

Received date: 29 March 2016

Revised date: 4 October 2016
Accepted date: 19 October 2016

Please cite this article as: Lee, Jaeho, Kim, Hakkon, Do employee relation responsibility
and culture matter for firm value? International evidence, Pacific-Basin Finance Journal
(2016), doi: 10.1016/j.pacfin.2016.10.006

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Do Employee Relation Responsibility and Culture

Matter for Firm Value? International Evidence

T
IP
Jaeho Lee

R
Associate Professor of International Business
Kyung Hee University - Department of International Business and Trade

SC
26 Kyunghee-daero, Dongdaemun-gu, Seoul, 02447, Korea
(jaeholee@khu.ac.kr)

NU
Hakkon Kim
(Corresponding Author)
Assistant Professor of International Business
MA
Chungbuk National University - Department of International Business, College of Business
Chungdae-ro 1, Seowon-Gu, Cheongju, Chungbuk, 28644, Korea
(kimhk0283@chungbuk.ac.kr)
D
P TE
CE
AC

1
 ACCEPTED MANUSCRIPT

Do Employee Relation Responsibility and Culture

Matter for Firm Value? International Evidence
Abstract Using a dataset on employee relation responsibility (ERR) from 30 countries, we

T
examine the effects of employee relationships on firm value. In addition, this paper attempts

IP
to understand the role of national cultural dimensions regarding their relation with ERR and
firm value. We find that firms with good employee relationships tend to demonstrate

R
significantly higher levels of firm value than firms whose employees exhibit poor rapport. We

SC
also find that the positive impact of ERR on firm value is significantly strengthened in
cultures with high degrees of power distance, collectivism, masculinity, and risk avoidance.
These results are robust even after mitigating endogeneity issues, adjusting sample

NU
composition bias and using alternative independent variables. Our findings extend the
previous CSR-related research into a new horizon, with the objective of investigating this less
explored subject within the sub-areas of CSR research. The findings of this study are
MA
noteworthy given that we incorporate the role of national culture into the ERR-firm value
link, a task that has not yet been undertaken.
D

1. Introduction
TE

Financial implications of corporate social responsibility (CSR) have captured academic

P

attention. As is well known in the CSR literature, stakeholder theory, as postulated by

CE

Freeman (1984), asserts that a company must consider and satisfy a variety of constituents,
such as workers, consumers, suppliers, and local community, all of whom can affect or be
AC

affected by the company1. Stakeholder theory implies that a company can secure support
from its related constituents by enhancing its engagement in certain CSR activities and that
its CSR-oriented activities connecting social dimensions with business models can offer
benefits that enhance the company’s performance and lead to long-term survival. Although
several studies have revealed either a negative link or no link between CSR and firm
performance or value, many studies have exhibited a positive relation between the two

1
CSR is posited to integrate social dimensions into economic activities by incorporating more employee-
friendly human resource management (HRM) practices (Bae et al., 2011; Jones et al., 2014; Collier and Esteban,
2007), accomplishing higher levels of environmental performance via pollution reduction and recycling
activities (Jo et al., 2015; Miles and Covin, 2000), pursuing mutually beneficial transactions with suppliers
(Andersen and Skjoett-Larsen, 2009), and considering the goals of the local community (Kapelus, 2002).

2
 ACCEPTED MANUSCRIPT

constructs2. In this paper, we aim to answer this widely explored CSR-linked question: do
companies do well by doing good? However, rather than focusing on the role and impact of
overall CSR for empirical analysis, we investigate a specific area of CSR from inside
stakeholder relationships, that of employee relationships, as such relationships have great

T
potential to influence firm value. Research indicates that good employee relationships within

IP
a company help the firm appear attractive to current employees who are and new job

R
candidates who will be more likely to commit themselves to the performance of the company.

SC
Hence, our research aims to contribute to academia by analyzing the impact of the
stakeholder relationship with employees on firm value, thus extending the previous related

NU
empirical research connected with CSR into a new horizon with the objective to fill this less
explored area of CSR and related finance research. We posit that a company’s employee
relation responsibility (ERR) is a sub-category of CSR and that it addresses the company’s
MA
comprehensive concerns for employees.
We also attempt to understand the role of national cultural dimensions in the
association between employee relation responsibility and firm value. National cultural
D

dimensions have recently been used in financial studies to examine their roles in investors’
TE

trading activities and stock price movements (Eun et al., 2015), cross-border merger volumes
and synergy gains (Ahern et al., 2015), trading volume, volatility and momentum profits
P

around the world (Chui et al., 2010), and allocation of funds in international portfolio
CE

holdings (Karolyi, 2016). Previous studies have also found that national culture has a
considerable influence on business ethics and CSR. Since Hofstede (1980; 1983; 1984; 1997)
AC

identified the cultural dimensions that differ across countries, several researchers have
attempted to understand the substantial extent to which national culture impacts companies’
attitudes and behaviors in terms of business ethics and CSR (Christie et al., 2003; Moon and

2
For instance, Waddock and Graves (1997) find that CSR leads to an increase in firm performance. Russo and
Fouts (1997) also observe that CSR activities focused on environment improvement demonstrate a positive
relation with financial performance. Gao et al. (2012) find a positive link between CSR activities and financial
performance by exploring the equity market impact of corporate donations. Hillman and Keim (2001) find that
stakeholder management CSR conducted in an effort to build constructive relations with primary stakeholders
such as employees, customers, suppliers, and communities can result in positive shareholder wealth creation
measured by market value added, while social CSR issues that are not directly related to primary stakeholders
may not create value for shareholders.

3
 ACCEPTED MANUSCRIPT

Franke, 2000; Blodgett et al., 2001; Vitell et al., 2003; Waldman et al., 2006). For example,
Christie et al. (2003), using survey data from the US, Korea, and India, find that cultural
dimensions of individualism and power distance have a strong influence on business
managers’ perceptions towards general aspects of business ethics and a variety of

T
questionable business practices, such as bribery and nepotism. We attempt to extend the

IP
previous research on the relationship between national culture and CSR by investigating the

R
combined effect of ERR and national culture on firm value and interpret the ERR-firm value

SC
association in view of cultural dimensions in different countries. To our best knowledge, this
empirical research is the first endeavor to examine relationships with employees in the setting

NU
of CSR-related stakeholder relationships, national cultural dimensions, and firm value.
This study first examines the impact employee relationships have on firm value and
provides empirical evidence that ERR is positively and significantly related to firm value. We
MA
also hypothesize that the impact of ERR on firm value will be strengthened by national
cultural dimensions across countries, and we obtain an empirical finding that ERR has
significantly greater effects on firm value in all cultural dimensions when measured as the
D

levels of power distance, collectivism, masculinity, and risk avoidance. In addition to the
TE

multivariate analyses, we perform supplementary tests that consider endogeneity issues from
the perspective of reverse causality, time-invariant omitted variables, and sample composition
P

bias. We also use alternative independent variables to ensure the robustness of our empirical
CE

results. After these tests, we still find that the impacts of ERR and its interaction with cultural
dimensions on firm value are still positive and significant.
AC

The remainder of this paper is organized as follows. In Section 2, we account for the
theoretical perspective on the association of employee relations with firm value in the CSR
framework and extend this theoretical perspective with cultural dimensions. We also
formulate testable hypotheses in this section. In Section 3, we describe the research methods
in relation to econometric specifications and data sources. Empirical results are presented in
Section 4, and we summarize our findings, implications and limitations in Section 5.

2. Literature Review and Hypothesis Development

4
 ACCEPTED MANUSCRIPT

2.1 CSR, Employee Relation Responsibility (ERR), and Firm Value

CSR involves the firm’s response to issues beyond the narrow technical and legal
requirements of the organization and considers the various stakeholders, i.e., those groups

T
and individuals who are related to the operation and performance of the organization

IP
(Freeman, 1984). By embracing CSR, a company provides benefits for society in general by

R
improving the environment and reducing social inequality. More particularly for the firm

SC
itself, CSR leads to an increase in corporate value by enhancing its reputation and, in turn,
increasing its potential to charge a premium price for products and recruit and retain high-

NU
quality workers. Accordingly, adopting CSR can be advantageous to both corporate
shareholders and stakeholders, which presents a potential win–win situation.
The literature on CSR proposes a number of theoretical implications regarding how
MA
CSR activities are linked to firm value. For example, CSR can promote a positive brand
image (Moskowitz, 1972), nurture good client relationships (Solomon and Hanson, 1985),
reduce company vulnerability to negative legal and regulatory actions as a result of the
D

reduction of information asymmetry with respect to stakeholders (Porter and van der Linde,
TE

1995; Freeman, 1984), and attract investments from socially responsible institutions and
individuals (Cheng et al., 2014). In this regard, CSR can increase firm value, and thereby help
P

the company identify several channels through which CSR can effect financial benefits.
CE

Examples include, but are not limited to, a strong recommendation from securities analysts
(Ioannou and Serafeim, 2015) and a greater likelihood to access finance (Cheng et al., 2014).
AC

Although CSR may raise related expenditures considerably, the potential financial benefits
will appear in the long run if a company carefully manages the CSR process (Henderson,
2002).
In this research, we state that a company’s endeavor to provide benefits for
employees, one group of company stakeholders, in the overall CSR context can also lead to
higher firm value. Little research has examined the association between a company’s social
responsibility for employees and firm value, Furthermore, there is no specific established
terminology representing the stakeholder relationship with employees with respect to CSR.
However, we contend that the stakeholder relationship with employees can be broadly
understood in the setting of employee relation responsibility (ERR) because it largely defines

5
 ACCEPTED MANUSCRIPT

the relationship with employees within the company in terms of how the company treats its
employees and what the company provides to ensure more employee-friendly working
environments to satisfy their needs and desires. ERR among stakeholder relationships has
considerable potential to influence firm value, as maintaining good employee relationships

T
increases the company’s attractiveness to its workers and job candidates, thus enhancing its

IP
capability to recruit high-quality workers, reduce hiring costs, and increase its employee

R
retention rate (Turban and Greening, 1997; Albinger and Freeman, 2000; Greening and

SC
Turban, 2000).
We argue that employee relation responsibility as a sub-category of CSR embodies a
company’s continuing and consistent endeavors and attempts to maintain and improve a

NU
constructive relationship with employees by providing both tangible and intangible benefits
and satisfactory working environments. Furthermore, ERR may include a range of high-
MA
performance workplace systems (HPWSs) that are devised to increase job satisfaction among
employees in the form of incentive compensation, working conditions, fringe benefits etc.,
but rather constitutes a company’s wide-ranging engagement and participation in socially
D

responsible behaviors to improve employee treatment. ERR includes a company’s actions and
TE

related investments to provide safe and sound working environments (health and safely); to
improve employee skills and abilities and offer an opportunity for career development
P

(training and development); to promote a more varied composition of employees (diversity);

CE

and to ensure employee freedom to express opinions and avoid forced and child labor in the
workplace (human rights). ERR can be estimated as an indicator of how a company addresses
AC

these areas.
Investing in employee relationship (and CSR) attributes and activities could be
important factors in the building of a firm’s reputation (Edmans, 2011) and in the
differentiation of products. ERR can lead to increased job satisfaction, which can be an
intermediate pathway for a company to achieve greater performance, thus resulting in
increased firm value. Within this discussion, we propose the following hypothesis.

Hypothesis 1: The improvement of employee relation responsibility (ERR) will have a positive
impact on firm value.

6
 ACCEPTED MANUSCRIPT

2.2 Employee Relation Responsibility (ERR), Cultural Dimensions, and Firm Value

Many cross-cultural studies attempt to capture the impact of national culture on a person’s

T
ethical attitudes and behaviors at the individual employee level (e.g., Allmon et al., 1997;

IP
Hood and Logsdon, 2002). Researchers use Hofstede’s cultural dimensions to operationalize

R
culture and apply them to the process of identifying their impacts on ethical behavior and

SC
decision making within organizations (e.g., Christie et al., 2003; Blodgett et al., 2001; Vitell
et al., 2003; Moon and Franke, 2000). A conceptual model developed by Vitell et al. (1993),
using Hofstede’s original four cultural dimensions, postulates that a strong relationship could

NU
exist between these dimensions and the managerial ethical decision making of companies.
Hofstede (1980; 1983; 1984; 1997) identifies the following four cultural dimensions: power
MA
distance, individualism/collectivism, masculinity/femininity, and uncertainty avoidance.
The power-distance dimension indicates that workers in countries with a high power
distance are more likely to accept hierarchies and authority within the organizations and
D

follow the opinions of their superiors (Hofstede, 1983; Vitell et al., 2003). While the
TE

employees in low power distance cultures easily approach their superiors and exhibit less fear
expressing disagreement with them, those in high power distance environments struggle to
P

challenge and contradict their superiors and tend to comply fully with their superiors’ orders.
CE

This suggests that when workers in high power-distance countries involve unethical
management issues, they are more likely to look to their superiors for guidance and may
AC

engage in questionable actions when they believe an action is in the best interest of the
company (Blodgett et al., 2001). Accordingly, it is also plausible that managers in high
power-distance countries may not be sensitive to employment treatment issues because their
autocratic leadership is generally based on unequally distributed power and because superiors
are likely to exert their authority unfavorably and unfairly over their subordinates. Given our
previous discussion about the potential for higher firm value from investing in employee
relationships, it may be more likely that companies operating in high power-distance
countries will gain more benefits by investing in employee relationships as the decision to
adopt a higher level of employee relationships in once-autocratic and unequally powered
companies and transform the company’s CSR policy will have greater marginal impacts on

7
 ACCEPTED MANUSCRIPT

the behaviors and performances of employees who have not previously experienced
appropriate employee treatment. Moreover, relatively faster decision making and strategy
implementation in high power-distance cultures may be more advantageous and more likely
to evoke higher effects of ERR, which will be subsequently reflected in firm value.

T
The individualism/collectivism dimension indicates that collectivists are dependent on

IP
others because they think of themselves as parts of organizations, endeavor to achieve group

R
success rather than seeking personal accomplishments, and tend to embrace the ideological

SC
identity of the organizations to which they belong, while individualists are emotionally
independent, exhibit a high need to fulfill their self-interests, and may not want to be

NU
confined to organizational rules and regulations (Hofstede, 1984; Peabody, 1985; Singelis et
al., 1995). Although collectivists may be likely to adopt an organizational code of ethics even
at the expense of personal interest (Akaah, 1992), Moon and Franke (2000) find that bribery
MA
is more acceptable in collectivist cultures than in individualistic ones. Christie et al. (2003)
also argue that managers in countries with low collectivism, and thus high individualism,
tend to be more sensitive to ethical issues and view business practices such as gift-giving,
D

nepotism, and software piracy as more unethical than do managers in countries with high
TE

collectivism. We argue that, given the lower sensitivity to employment treatment issues in a
collectivist society, the effect of investing in ERR is likely higher in the collectivist culture as
P

unit investment in ERR exhibits a higher marginal effect on firm value.

CE

The masculinity/femininity dimension suggests that cultures featured as masculine

compel individuals to be ambitious, assertive, competitive, and achievement-oriented, while
AC

individuals in feminine cultures are encouraged to be modest, humble, caring, and

relationship-oriented (Hofstede, 1983; 1984). The literature generally documents that highly
masculine individuals are less sensitive to the interests of other stakeholders and are less
likely to perceive ethical problems than the counterparts in feminine cultures due to the
tendency in the masculine culture to seek personal financial gain within a very competitive
atmosphere (Vitell and Festervand, 1987; Chang and Ding, 1995). This indicates that
companies dominated by masculine cultures are less likely to engage in a range of human
resource management practices that promote favorable employee treatment and that are
comprising of varied tools, techniques, and incentives. Conversely, companies dominated by
feminine cultures tend to embrace more employee-friendly human resource policies and take

8
 ACCEPTED MANUSCRIPT

proactive steps to engage in employee relationship-oriented actions, such as work-family

programs, affirmative action for the recruitment of minorities, etc. We expect that investing in
ERR may result in higher marginal effects on firm value in countries with masculine cultures
than in those with feminine cultures because the adoption of the ERR in companies

T
dominated by masculine cultures provides them an opportunity to undergo a dramatic

IP
transformation in CSR, which, in turn, will result in a greater increase in value creation.

R
The uncertainty avoidance dimension suggests that workers from societies with strong

SC
uncertainty avoidance tend not to accept any deviations from group/organizational norms
compared to those from countries with weak uncertainty avoidance (Ferrell and Skinner,

NU
1988). This may imply that organizations in high uncertainty-avoidance countries prefer to
have formal rules to regulate the ethical behaviors of the workers who, by their nature, would
not avoid or ignore internal norms/cultures (Vitell et al., 1993). However, despite this rigidity
MA
in adhering to the formal rules, it may take a great deal of time for organizations governed by
strong uncertainty avoidance to make a decision regarding the adoption of rules for business
ethics, let alone employee relationships. When organizations are dominated by high
D

uncertainty avoidance, they are risk averse, which indicates that they may not be aggressively
TE

searching for new investment opportunities or that they experience a deliberate and lengthy
process before reaching a final decision to relocate resources to an investment alternative.
P

With respect to companies with high uncertainty avoidance, investment in employee

CE

relationships may be perceived as highly risky as it may be considered to demonstrate the

potential for failure, and accordingly, it may require a long-term management perspective.
AC

Therefore, it is plausible that managers in high uncertainty-avoidance countries will hesitate

to invest in employee relation responsibility and may be insensitive to employment treatment
issues. Consistent with the aforementioned marginal effect concerns regarding the other
cultural dimensions, a firm’s effort to improve ERR could be more effective in the
enhancement of firm value in high risk-avoidance cultures, as they are more likely to lack
consideration for employees and less likely to invest in employees. Moreover, the emphasis
on ERR issues in legal terms in high uncertainty-avoidance cultures (Christie et al., 2003)
may gradually draw increasing public attention to ERR and thereby improve employee
treatment practices and positively affect the firm’s value in the long term.
All in all, our expectation regarding the impact of ERR on firm value in countries

9
 ACCEPTED MANUSCRIPT

with different cultural dimensions is that the marginal effect of ERR will be greater in
countries with higher power-distance, stronger collectivism, greater masculinity, and stronger
risk avoidance. We argue that ERR can produce much greater impact in these countries with
stronger potential for reputation building and that it can provide companies with opportunities

T
to decrease hiring costs, retain high-quality workers, and attract capable workers who have

IP
been searching for one of the rare companies that treats people well in that cultural context.

R
The marginal effects of investments in ERR can be greater for firms that have fewer and less

SC
restrictive policies, procedures, and regulations regarding employee treatment compared to
firms with numerous rigid policies and regulations whose investments in ERR may result in

NU
lower employee satisfaction and less impact on employees according to the economic theory
of the law of diminishing marginal utility3. Thus, the following hypotheses are proposed.
MA
Hypothesis 2: Cultural dimensions strengthen the positive relationship between employee
relation responsibility (ERR) and firm value.
D

Hypothesis 2-1: The impact of ERR on firm value will be more prominent in countries with
TE

high power-distance cultures.

Hypothesis 2-2: The impact of ERR on firm value will be more prominent in countries with
P

strong collectivist cultures.

CE

Hypothesis 2-3: The impact of ERR on firm value will be greater in countries with strong
masculine cultures.
AC

Hypothesis 2-4: The impact of ERR on firm value will be higher in countries with strong risk-
avoidance cultures.

3. Empirical Setting

3.1 Sample Construction

3
See Layard et al. (2008) and Finkelstein et al. (2013) to gain an understanding of how this concept applies to
the behaviors of rational economic agents.

10
 ACCEPTED MANUSCRIPT

Our empirical analysis centers on testing our hypotheses regarding the effects of employee
relation responsibility and its interactive effects of cross-cultural dimensions on firm value.
To test the hypotheses, we use the following four databases: (i) Thomson Reuters ASSET4,

T
which provides the score relating to employee relation responsibility at the firm level

IP
worldwide; (ii) Hofstede Centre, which reports Hofstede’s cultural dimensions; (iii) Capital

R
IQ, which provides financial statement data; and (iv) World Bank, which offers information

SC
on the logarithm of GDP per capita, GDP growth volatility, and governance indicators.
Our sample consists of a panel of 16,693 firm-year observations from 30 countries

NU
for the period 2002 to 2012, of which 3,342 are from seven Asia-Pacific countries, 6,347
observations are from nineteen European countries, and 7,004 observations are from four
countries in North/South America.4 Table 1 reports the descriptive statistics of the key
MA
variables used in our empirical tests. The mean (median) value of ERR is 54.24 (56.6). The
mean (median) values of power distance and collectivism are 44.65 (40.00) and 26.89
(11.00), respectively. The mean (median) values of masculinity and uncertainty are 62.25
D

(60.00) and 57.57 (46.00), respectively. All continuous variables are winsorized at the 1% and
TE

99% levels to consider potential outlier problems or data errors.

P
CE

3.2 Employee Relation Responsibility (ERR)

AC

For our empirical analysis, we use a panel dataset where the ERR index was obtained from
the Thomson Reuters ASSET4 database, a database that has been used in a number of recent
studies as a subdomain of the CSR index (McWilliams and Siegel, 2001; Cheng et al., 2014;
Ioannou and Serafeim, 2015; Roulet and Touboul, 2015). The Thomson Reuters ASSET4 is a
leading provider of the world’s largest objective, comparable, and auditable database of ESG
information. Cheng et al. (2014) also argue that the Thomson Reuters ASSET4 specializes in

4
Our data contain seven Asia-Pacific countries (Indonesia, Japan, Malaysia, Philippines, Singapore, South
Korea, and Thailand), nineteen European countries (Austria, Belgium, Denmark, Finland, France, Germany,
Greece, Ireland, Italy, Netherlands, New Zealand, Norway, Poland, Portugal, Spain, Sweden, Switzerland,
Turkey, and United Kingdom), and four North/South American countries (i Brazil, Chile, Mexico, and the
United States).

11
 ACCEPTED MANUSCRIPT

providing objective, relevant, auditable, and systematic ESG information to industries and
academia.
The ESG information is composed of environmental, social, and (corporate)
governance indexes that are collected by more than 130 specially trained analysts from

T
annual reports, company websites, proxy filings, and NGOs as well as from news reports of

IP
all major providers (see Thomson Reuter’s data collection and rating methodology). Among

R
the ESG information (i.e., environmental, social, and corporate governance indexes), the

SC
social index is aimed at measuring the quality of relationships between a firm and its
employees. Therefore, we use the social index as our main measure of ERR. In addition, for

NU
the robustness tests, we use an equally weighted ERR index (i.e., EW employee relation),
which is defined as the equal weighting of the (mostly) employee-related (five) sub-social
indexes, namely, (a) employment quality index, (b) training and development index, (c)
MA
health and safety index, (d) diversity index, and (e) human rights index.5 Our measures of
ERR (i.e., ERR and EW ERR) are normalized to a z-score (i.e., a standard score) and
multiplied by 100 to simplify the interpretation of regression coefficients (for more detail on
D

the employee relation index, see Appendix A).

P TE

3.3 Hofstede’s Cultural Dimensions

CE

Hofstede’s cultural dimensions are the most widely used cross-country measures to facilitate
AC

an understanding of national cultural differences in financial studies (Ahern et al., 2015; Eun
et al., 2015; Waldman et al., 2006). Furthermore, the cultural dimensions have been regarded
as one of the more important factors linked to business ethics and CSR decision making
(Moon and Franke, 2000; Blodgett et al., 2001; Christie et al., 2003; Vitell et al., 2003).
Hofstede (2001) first used the data obtained from the survey results of IBM employees in 40
countries, after which the database was expanded to more than 70 countries. Because
Hofstede’s database is collected within the IBM Corporation, McSweeney (2002) criticizes

5
We use the (original) social score as our main measure of ERR and the equally weighted ERR index as our
second measure of employee relation to reduce the potential measurement biases that can occur when we
arbitrarily calculate the equal weighting of the (mostly)employee-related (five) sub-social indexes.

12
 ACCEPTED MANUSCRIPT

Hofstede’s dimensions, claiming the samples in the studies are not nationally representative.
However, a large body of literature, including Williamson (2002), contradicts the concern of
McSweeney (2002) and argues that Hofstede’s model can effectively explain relative cultural
differences across countries because organizational cultures reflect the national culture.

T
Hofstede’s cultural dimensions are composed of four primary measures, namely,

IP
power distance, collectivism (/individualism), masculinity, and uncertainty avoidance, for a

R
maximum of 100 points. Power-distance measures the level of hierarchy in a society.

SC
Individuals in high power-distance societies are likely to admit the inequality of power and
differences between superiors and subordinates. Higher scores on the collectivist index

NU
indicate that individuals tend to think of themselves as part of the organization and that they
endeavor to achieve group success, while lower scores on the collectivist index indicate that
individuals tend to fulfill their self-interests, value their personal freedom, and avoid
MA
confining themselves to collectivist obligations. Although Hofstede distinguishes between the
power distance and collectivism dimensions, these two dimensions seem to share many
analogous aspects and co-move closely with each other in terms of their impacts on
D

employees and their similarity in meaning. Therefore, we interpret the effect of the
TE

collectivism dimension together with the effect of the power-distance dimension for the
analysis to avoid overlapping the application of these similar dimensions (see Panel B of
P

Table 2). Higher scores on the masculinity index indicate the dominance of traditional social
CE

male values, such as a preference for achievement, heroism, assertiveness, and material
success (Hofstede, 1984). However, lower scores on the masculinity index indicate the
AC

dominance of feminine values, which are characterized as tender, humble, caring, benevolent,
and relationship-oriented (Hofstede, 1984). Uncertainty avoidance refers to the society’s
tolerance for uncertain and unknown situations. Individuals in high-uncertainty societies are
likely to be risk-averse, resist change, place a premium on job security, and favor written
rules in contrast to individuals in cultures with low uncertainty avoidance societies (Hofstede,
1984). We connect Hofstede’s cultural dimensions with our ERR measures to examine the
combined effect of employee relation responsibility and national culture on firm value.

3.4 Residuals of Hofstede’s Cultural Dimensions

13
 ACCEPTED MANUSCRIPT

In cross-country literature, the logarithm of GDP per capita is an important variable as it

controls for economic development (Stulz and Williamson, 2003; Eun et al., 2015). However,
as Panel A of Table 2 reveals, the logarithm of GDP per capita is strongly correlated with
Hofstede’s cultural dimensions, measuring 0.75 (75%), thus raising the possibility of

T
multicollinearity problems. Thus, if we use both Hofstede’s cultural dimensions and the

IP
logarithm of GDP per capita simultaneously in the same regressions, our results can be biased

R
due to the approximate linear relationship.

SC
Thus, to mitigate multicollinearity concerns, we use the residuals of Hofstede’s
cultural dimensions as our key independent variables following Graham (2003). Furthermore,

NU
by using the residuals, we can examine the combined actual effect of cultural differences-
employee relationship and firm value without the impact of economic development on firm
value. Following Graham (2003), our residuals of Hofstede’s cultural dimensions are
MA
obtained from regressing each of Hofstede’s cultural dimensions on the logarithm of GDP per
capita every year. In addition, we confirm the variance inflation factor (VIF), which is a
widely used multicollinear indicator and find that the values of the VIF are greater than 30 if
D

we use both Hofstede’s cultural dimensions and the logarithm of GDP per capita
TE

simultaneously in the same regressions. The results indicate strong multicollinearity problems
and potential difficulties with least squares estimation. However, if we use the residuals of
P

Hofstede’s cultural dimensions instead of the (original) Hofstede’s cultural dimensions, the
CE

values of VIF are less than 7, indicating that there is a low possibility of multicollinearity
problems in our empirical analysis.
AC

3.5 Empirical Model and Variable

To ascertain whether a firm value is influenced by employee relation responsibility and its
interactive effects of cross-cultural differences, we regress Tobin’s q on the ERR index and its
interaction terms with the residuals of Hofstede’s cultural dimensions. The basic regression
model used is as follows:

Tobin’s q = α0 + α1 ERR + α2 ERR＊Hofstede’s residual cultural dimensions (Power distance

residual, Collectivism residual, Masculinity residual, and Uncertainty avoidance residual) +

14
 ACCEPTED MANUSCRIPT

α3 Hofstede’s residual cultural dimensions + j＊controlj +ε (1)

where the dependent variable, Tobin’s q, is widely used as a measure of firm value in studies

T
such as Chung and Pruitt (1994), Chung and Jo (1996), and Servaes and Tamayo (2013). We
employ the formula proposed by Chung and Pruitt (1994) to compute Tobin’s q, which is

IP
defined as (book value of assets + market value of equity - book value of equity) / (book

R
value of assets). ERR is the score of the equally weighted index of the following seven

SC
dimensions: employment quality, health and safety, training and development, diversity,
human rights, corporate community involvement, and product responsibility. Hofstede’s

NU
cultural dimensions are the indexes that measure cross-cultural differences around the world.
The dimensions range from 0 to 100 on various indexes, including power distance,
collectivism, masculinity, and uncertainty avoidance. Following Graham (2003), we use the
MA
residuals of Hofstede’s cultural dimensions for minimizing the multicollinearity problem and
find the combined actual effect of cultural differences–ERR on firm value without the
D

economic development impact on firm value (detailed explanations described in the ERR,
cultural dimensions, and firm value sections). The interaction terms between ERR and
TE

Hofstede’s residual cultural dimensions are used to indicate the differences in the effects of
ERR on firm value among different cultures.6
P

We control for the differential effects of firm- and country-level characteristics on

CE

firm value by including leverage, log (total assets), capital expenditure, R&D/total assets, log
(GDP per capita), GDP growth volatility, and governance indicators. Leverage is the ratio of
AC

the book value of debt divided by the total assets. Log (total assets) is the logarithm of the
firm’s total assets. Capital expenditure is calculated as the capital expenditure expense
divided by the total assets. R&D/total assets are calculated as the research and development
expenditures divided by the total assets. Log (GDP per capita) is the logarithm of GDP per
capita, which measures economic development of each country (Stulz and Williamson, 2003;
Eun et al., 2015). GDP growth volatility is used as a measure of GDP growth volatility and
estimated as the standard deviation of growth in GDP per capita over five years (Eun et al.

6
We employ one of Hofstede’s residual cultural dimensions in each regression model (i.e., they are not used
together in one regression model) to avoid the multicollinearity problem among the cultural dimensions,
following Christie et al. (2003).

15
 ACCEPTED MANUSCRIPT

2015). Governance indicators are used to control for the differential governance system of
each country (Morck et al., 2000). These indicators are estimated as the sum of six indexes
from Kaufmann et al. (2009), specifically, (i) voice and accountability, (ii) political stability
and absence of violence/terrorism, (iii) government effectiveness, (iv) regulatory quality, (v)

T
rule of law, and (vi) control of corruption. The governance indicators provided by Kaufmann

IP
et al. (2009) are only available for the period 2002 to 2008. Therefore, following Eun et al.

R
(2015), we assign the 2008 values for 2008 to 2012. We include year-, industry-, and

SC
country-level fixed effects in all specifications with the robust standard errors used in a study
by White (1980). We provide more detailed variable explanations in Appendix A.

4. Empirical Results NU
MA

4.1 Multivariate Analysis

D

To examine the relationship between firm value and ERR, in Models 1 to 5 of Table 3, we
TE

regress the firm value on the ERR index with control variables including country-, industry-,
and year-fixed effects. In all models of Table 3, we find that the ERR is strongly positively
P

associated with Tobin’s q, consistent with our expectation that the improvement of ERR leads
CE

to an increase in firm value.

AC

[Table 3 inserted]

We also test whether firm value is related to the interactive effects of cultural dimensions and
ERR. We include interaction terms between the residuals of Hofstede’s cultural dimensions
and the ERR index in Models 2 to 5 of Table 3. In Models 2 and 3 of Table 3, the coefficients
of ERR x Power distance residual and ERR x Collectivism residual are positive and
statistically significant at the 1% level. Thus, our results suggest that the impact of ERR on
firm value is more prominent in countries with more autocratic, unequal power distribution
with strong group and collectivistic norms (in higher power-distance cultures/collectivist
cultures). In Model 4 of Table 3, ERR x Masculinity residual is strongly positively related to

16
 ACCEPTED MANUSCRIPT

Tobin’s q, supporting the premise that the impact of ERR on firm value is greater in countries
with stronger masculine cultures where employees focus on the firm’s success and exhibit
competition-oriented attitudes and behaviors and have less consideration for other people.
The coefficient of ERR x Uncertainty avoidance residual is positive and significant,

T
indicating that the impact of ERR on firm value will be greater in countries that are unwilling

IP
to take a risk in investment activity in ERR due to the potential danger of failure (strong risk-

R
avoidance cultures). Overall, our findings consistently reveal that the firms with higher ERR

SC
scores tend to demonstrate higher firm values than firms with lower ERR scores. In addition,
the positive ERR-firm value association becomes stronger when firms are located in societies

NU
with high power distance, collectivism, masculinity, and uncertainty avoidance. The findings
of the empirical tests support our first, second, and sub-hypotheses.
MA
4.2 Adjusting Sample Composition Bias

The heterogeneity of the number of firm-year observations across 30 countries may influence
D

the associations among ERR, its interactive effects of cultural dimensions, and firm value.
TE

The weighted least squares (WLS) method addresses the potential problem of heterogeneity
of observations and maximizes the efficiency of parameter estimation. This can be made
P

possible by attempting to give each data point its suitable amount of influence over the
CE

parameter estimates. Thus, in Table 4, we present the results of the WLS regression, in which
the inverse of the number of firm-year observations per country is used for the weight,
AC

following Kim et al. (2015). Consistent with previous findings, the results reveal significant
and positive relationships between ERR and firm value in Models 1 to 5. We also find
significant and positive relations between firm value and interaction terms as follows: ERR x
Power distance residual; ERR x Collectivism residual; ERR x Masculinity residual; and ERR
x Uncertainty avoidance residual. These results indicate that the improvement of ERR
enhances firm value and that its marginal effect is greater in countries with higher power
distance, stronger collectivism, greater masculinity and stronger risk avoidance, thus
supporting our first, second, and sub-hypotheses.

[Table 4 inserted]

17
 ACCEPTED MANUSCRIPT

A possible concern with our analysis is that previous results could be mainly derived from
U.S. firms, which account for a large fraction of our sample. In Panel A of Table 5, we
therefore re-conduct our baseline regression model in Eq. (1) after removing the U.S. sample.

T
The findings indicate that all of the ERR-firm values as well as the interaction terms (ERR x

IP
Hofstede’s residual cultural dimensions) and firm value associations remain qualitatively the

R
same as those of our earlier findings. Thus, the robustness tests indicate that our empirical

SC
results are not driven by U.S. firms.

NU
[Table 5 inserted]

As previously mentioned, it is reasonable to suppose that power distance and collectivism co-
MA
move closely with each other in the same direction, which is also strongly supported by our
Pearson correlation result (see footnote 2). Although the co-movement between power
distance and collectivism is quite obvious in most countries in our sample, the positive
D

relationship between the two variables is relatively weak, though still positive, in five
TE

countries, specifically, Belgium, France, Japan, Poland, and Spain. Thus, in Panel B of Table
5, we verify that our cross-country results do not change if we exclude firms in the five
P

countries. We find that the results are unchanged across all of the ERR-firm values as well as
CE

the interaction terms firm value associations; however, excluding the five countries reduces
our observations from 16,693 to 12,926.
AC

4.3 Alternative Independent Variables

In this section, we re-estimate our tests by applying alternative measures of ERR to ensure
that our findings are not sensitive to alternative independent variables, including the equally
weighted ERR index (i.e., EW employee relation) and the first principal component of the
ERR index (i.e., PCA employee relation). As mentioned earlier, the EW ERR is defined as the
equal weighting of the (mostly) employee-related (five) sub-social indexes, namely, (a)
employment quality index, (b) training and development index, (c) health and safety index,
(d) diversity index, and (e) human rights index. Consistent with the earlier findings, Table 6

18
 ACCEPTED MANUSCRIPT

indicates that all coefficients of EW ERR and its interaction terms with the residuals of
Hofstede’s cultural dimensions are positive and statistically significant at the 1% level.

[Table 6 inserted]

T
IP
We also repeat our analysis using the PCA ERR instead of the (original) ERR index, which is

R
defined as the principal components of the five categories of employment quality, training

SC
and development, health and safety, diversity, and human rights. Abdi and Williams (2010)
argue that the PCA extracts critical information from data in which observations are inter-

NU
correlated by quantitative variables. Table 7 presents the regression results based on the
alternative measure (i.e., PCA employee relation) of our key independent variable. Our
results also indicate that the coefficients of the PCA ERR and its interaction terms with the
MA
residuals of Hofstede’s cultural dimensions are significant and positive.
Tables 6 and 7 indicate that the ERR effect combined with the effect of national
cultural dimensions on firm value is stronger with larger coefficients of key independent
D

variables than the coefficients in Tables 3, 4, and 5, which reflect the values when using the
TE

EW ERR index or the PCA ERR index. It is reasonable to expect our results to be stronger
when we use the key employee sub-five indexes as alternative ERR indexes. The evidence
P

supports our hypotheses that a firm’s concentrated effort to improve relations with employees
CE

would increase its value and that the positive relationship between employee relation
responsibility and firm value would be strengthened in countries with greater power distance,
AC

stronger collectivism, stronger masculinity, and stronger risk-avoidance cultures.

[Table 7 inserted]

4.4 Endogeneity Concerns

Endogeneity is unlikely to be a concern for the Hofstede’s cultural dimensions in our study.
Eun et al. (2015) argue that a reverse causality of the cultural dimensions seems implausible
because it is highly improbable that firm value changes the country’s cultural traits. In
addition, we also control for several country-level variables such as log GDP per capita, GDP

19
 ACCEPTED MANUSCRIPT

growth volatility, governance indicators, and country-fixed effects following Stulz and
Williamson (2003) and Eun et al. (2015). Thus, it is difficult to identify possible omitted
variables that are more exogenous than culture, which can affect firm value. However, a
potential criticism of the baseline regressions is that the ERR index might not be exogenously

T
determined. For example, it might happen that firms with high levels of value tend to have a

IP
better ERR, which creates a reverse causality.

R
SC
4.4.1. Two-stage Least Squares and Dynamic System GMM

NU
To mitigate the endogeneity concern, in this section, we re-estimate our earlier analysis using
two approaches, the two-stage least squares (2SLS) and the dynamic system GMM. First, we
perform a two-stage least squares (2SLS) approach and present the results in Table 8. We
MA
construct the instrumental variables (IVs), which include the initial ERR score recorded when
the firm enters the sample, the industry average of the ERR score in the first year of data, and
a dummy variable taking the value 1 if the previous year’s earnings represent a loss, similar
D

to Elsas et al. (2010), Ferreira et al. (2011), El Ghoul et al. (2011), and Cai et al. (2015).
TE

Although lagged variables are not fully exogenous, using lagged variables, such as the initial
ERR score and industry average ERR score in the first year of data, is a common procedure
P

in econometrics for instrumental variable methods because these variables are set before
CE

determining firm value, but are usually highly correlated with an endogenous variable (see
Elsas et al., 2010). El Ghoul et al. (2011) further use a negative earning dummy variable as an
AC

instrument variable because the investment in ERR (and/or CSR) can be negatively affected
by negative earnings in the previous year.7 In Table 8, we present the results of the F-tests
and Sargan’s over-identification tests to confirm the robustness of our instruments. According
to the F-tests, the null hypothesis that instrument variables can be excluded from the first-
stage regressions is strongly rejected, indicating that our instruments are not weak. The p-
values of Sargan’s over-identification tests range from 0.142 to 0.165, suggesting that our
instruments are valid as they are not related to the residual error term. The 2SLS results

7
In (untabulated) Pearson correlation tests, we confirm that our lagged variables are highly correlated with the
ERR index and that the previous year’s negative earnings are negatively correlated with one endogenous
variable.

20
 ACCEPTED MANUSCRIPT

reveal that the employee relation responsibility and all of its interactive effects of cultural
dimensions are strongly significant and positive throughout, thus supporting the first, second,
and sub-hypotheses.

T
[Table 8 inserted]

R IP
To further address the endogeneity concerns in our analysis, we perform the dynamic system

SC
GMM estimator following Blundell and Bond (1998) and Wintoki et al. (2012) and present
the results in Table 9. Blundell and Bond (1998) argue that the dynamic system GMM

NU
improves the efficiency of the estimator by mitigating endogeneity issues from the reverse
causality and time-invariant omitted variables and autocorrelations between the main
variables and the unobserved heterogeneity in the panel data by combining the moment
MA
conditions of the first differencing and the level GMM estimators.8

[Table 9 inserted]
D
TE

In Table 9, to verify the validity of our instruments, we perform the following two
specification tests: the second-order serial correlation test (i.e., AR (2)) and the Hansen J test
P

of over-identification. The p-values of the AR (2) range from 0.105 to 0.121, suggesting that
CE

the null hypothesis that there is no second-order serial correlation cannot be rejected. In
addition, the p-values of the Hansen J test range from 0.103 to 0.130. Therefore, we do not
AC

reject the null hypothesis that the instruments are uncorrelated with the residuals. We further
report the results of a difference-in-Hansen test of exogeneity, as suggested by Eichenbaum et
al. (1988) and Wintoki et al. (2012).9 Our results indicate that the p-values of the difference-
in-Hansen test of exogeneity range from 0.435 to 0.656, suggesting that we cannot reject the
null hypothesis that the additional subset of instruments is exogenous. These specification

8
Blundell and Bond (1998) argue that the estimators of Arellano and Bond (1991) can be biased when the ratio
of the variance of the panel-level effect divided by the variance of idiosyncratic error is high or the
autoregressive parameters are large.
9
A difference-in-Hansen test of exogeneity yields a J-statistic, and the null hypothesis is that the subsets of
instruments in the level equations are exogenous.

21
 ACCEPTED MANUSCRIPT

tests confirm the validity of our instruments, and we find a significant positive association
between ERR and firm value. Our results further support the preservation of the previous
results with interactive effects of cultural dimensions and the ERR index on firm value. Taken
together, the results in Tables 8 and 9 further support our arguments regarding the positive

T
effects of employee relation responsibility on firm value and the combined effect of cultural

IP
differences-ERR and firm value.

R
SC
4.4.2. Granger Causality Tests

NU
To alleviate the concern regarding the reverse causality between firm value and ERR, we
perform Granger causality tests to determine whether a significant bi-directional causal
relationship is present between two variables and present the results in Table 10. The 1-, 2-,
MA
and 3-year lagged Tobin’s q and employee relation variables, including EW ERR and PCA
ERR, are used to estimate the model, an approach that is similar to that of Kim et al. (2015).
In Models 1 and 3 of Table 10, we find positive and significant relationships between ERR
D

(and/or EW ERR and PCA ERR) and firm value when the 2-year lagged ERR variables are
TE

used as explanatory variables and Tobin’s q is used as a dependent variable. On the contrary,
our results reveal insignificant relationships between ERR (and/or EW ERR and PCA ERR)
P

and firm value in Models 2 and 4 when the 1-, 2-, and 3-year lagged Tobin’s q variables are
CE

used as explanatory variables and ERR is used as a dependent variable. The findings of our
empirical tests suggest that there is a positive and unidirectional Granger causality between
AC

ERR and firm value, suggesting a lower probability of reverse causality between the two
variables and indicating that the problem of reverse causality is not a serious factor in our
analysis.

[Table 10 inserted]

5. Conclusion

In this paper, our research interests lie in the association between companies’ investments in

22
 ACCEPTED MANUSCRIPT

the social responsibility of stakeholder relationships with employees and firm value. Our
study distinguishes employee relation responsibility, a sub-category of CSR, from other
related activities and examines an empirical analysis that links it to firm value by taking into
account a comprehensive list of employee relation responsibilities, thus providing an ERR

T
score in diverse areas at the firm level across countries. Our results indicate that ERR is

IP
significantly and positively associated with firm value and that this significant and positive

R
relationship is more pronounced when ERR is combined with four national cultural

SC
dimensions. Our results are robust even after mitigating endogeneity issues from reverse
causality and time-invariant omitted variables using 2SLS and dynamic system GMM as well

NU
as adjusting for sample composition bias and using alternative independent variables.
Our finding confirms the stakeholder maximization perspective that CSR activities
positively impact shareholder wealth (Deng et al., 2013). Greater effects of ERR on firm
MA
value in cultures with greater power distance, stronger collectivism, greater masculinity, and
stronger risk avoidance indicate that when a company increases its investment in ERR in
these types of cultures, its marginal effect on the firm value per unit cost of investment is
D

greater than it is in other cultures. A company that invests in ERR in these cultures benefits
TE

by improving its reputation and by attracting and retaining high quality, capable employees
who are eager to join a company that promotes employee relationships in cultures that are
P

characterized as autocratic, group oriented, highly competitive, and risk averse to investing in
CE

CSR. Our results provide evidence that cultural dimensions affect and reinforce the positive
relationship between ERR and firm value.
AC

We offer critical contributions that have implications for CSR and related finance
researchers. First, our research indicates that the CSR variable in the analysis with respect to
its link to firm value can be elaborated by more in-depth investigation into its sub-categories,
in particular, a company’s relationship with its stakeholders (i.e., employees). CSR is
composed of varied dimensions, each of which may have different impacts on firm value.
The indicators that measured CSR tend to be aggregated overall to determine the total
quantity of CSR activity in the empirical analysis. However, as a company may have different
interests in each CSR stakeholder relationship, when we separate CSR into different
dimensions with the characteristics of each dimension, we not only improve the explanatory
power for CSR and its effect on firm value, but we also obtain a better understanding of the

23
 ACCEPTED MANUSCRIPT

mechanism and procedure by which CSR activity in a sub-category can result in an increase
or a decrease in firm value.
Second, our analysis brings attention to the importance of the careful selection of a
relevant variable that can strengthen the relation between CSR and firm value. Though

T
previous studies examine relationships between these two constructs, they mainly attempt to

IP
identify the direct impact of CSR-related variables on firm value. There is scant research that

R
seeks to identify the combined effects of CSR and another variable on the increase or

SC
decrease of firm value. This study is, to the best of our knowledge, the first attempt to
investigate the role of cultural dimensions in the relation between employee relation

NU
responsibility and firm value. Thus, we contend that further research must use different
variables that can be combined with CSR measures to estimate the interaction effect on firm
value and that the use of cultural dimensions in this paper has contributed to expanding CSR
MA
research.
Third, this study suggests an improved econometric methodology in consideration of
residual values of independent variables when they are highly correlated with another
D

variable. In our case, cultural dimensions are strongly correlated with the logarithm of the
TE

GDP, thus raising concern related to serious multicollinearity. However, by converting

cultural dimension values into residuals, we investigate the combined effects of cultural
P

dimensions and employee relationships on firm value without causing biased results from
CE

problems of multicollinearity. This is another contribution of our research to academia as it

offers a better understanding of the relationship between CSR measures and firm value.
AC

Despite the important findings of this paper, the study also has some limitations.
First, the combined effect of national culture and CSR (including employee relation
responsibility) on firm value can be affected by other country-level factors, such as the
regulatory enforcement environment. Therefore, as in prior studies, we use various country-
level control variables, including country fixed effects, governance indicators, log (GDP per
capita), and GDP growth volatility. However, there is still a possibility that other formal and
informal institutional factors affect the combined effect. Therefore, future research should
consider the problem by controlling for other institutional variables. Second, although
Hofstede’s cultural dimensions provide the most widely used data to understand national
cultural differences, the data are, as mentioned previously, subject to concern due to the lack

24
 ACCEPTED MANUSCRIPT

of nationally representative samples given that the dimensions are obtained from a specific
company (i.e., IBM). Williamson (2002) contradicts the concern of representativeness and
argues that Hofstede’s model can effectively explain national cultural differences because
organizational cultures reflect the national cultures. Nonetheless, it would benefit future

T
studies to use other cultural variables, such as global leadership and organizational behavior

IP
effectiveness (GLOBE), for their robustness tests.

R
SC
References

NU
Abdi, H., Williams, L. J. (2010). Principal component analysis. Wiley Interdisciplinary
MA
Reviews: Computational Statistics, 2(4), 433-459.
Ahern, K. R., Daminelli, D., Fracassi, C. (2015). Lost in translation? The effect of cultural
values on mergers around the world. Journal of Financial Economics, 117(1), 165-189.
Akaah, I. (1992). Social inclusion as a marketing ethics correlate. Journal of Business Ethics,
D

11(8), 599-608.
Albinger, H. S., Freeman, S. J. (2000). Corporate social performance and attractiveness as an
TE

employer to different job seeking populations. Journal of Business Ethics, 28(3), 243-
253.
P

Allmon, D., Chen, H. K., Pritchett, T., Forrest, P. (1997). A multicultural examination of
business ethics perceptions. Journal of Business Ethics, 16(2), 183-188.
CE

Andersen, M., Skjoett‐Larsen, T. (2009). Corporate social responsibility in global supply

chains. Supply Chain Management: An International Journal, 14(2), 75-86.
AC

Arellano, M., Bond, S. (1991). Some tests of specification for panel data: Monte Carlo
evidence and an application to employment equations. Review of Economic Studies,
58(2), 277-297.
Bae, K. H., Kang, J. K., Wang, J. (2011). Employee treatment and firm leverage: A test of the
stakeholder theory of capital structure. Journal of Financial Economics, 100(1), 130-153.
Blodgett, J. G., Lu, L. C., Rose, G. M., Vitell, S. J. (2001). Ethical sensitivity to stakeholder
interests: A cross-cultural comparison. Journal of the Academy of Marketing Science,
29(2), 190-202.
Blundell, R., Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel
data models. Journal of Econometrics, 87, 115-143.
Cai, L., Cui, J., Jo, H. (2015). Corporate environmental responsibility and firm risk. Journal
of Business Ethics (in press).
Chang, K., Ding, C. G. (1995). The influence of culture on industrial buying selection criteria
in Taiwan and Mainland China. Industrial Marketing Management, 24(4), 277-284.
25
 ACCEPTED MANUSCRIPT

Cheng, B., Ioannou, I., Serafeim, G. (2014). Corporate social responsibility and access to
finance. Strategic Management Journal, 35(1), 1-23.
Christie, P. M., Kwon, I. W., Stoeberl, P. Baumhart, R. (2003). A cross-cultural comparison of
ethical attitudes of business managers: India, Korea and the United States. Journal of
Business Ethics, 46(3), 263-287.

T
Chui, A. C., Titman, S., Wei, K. J. (2010). Individualism and momentum around the world.

IP
The Journal of Finance, 65(1), 361-392.
Chung, K. H., Pruitt, S. W. (1994). A simple approximation of Tobin's q. Financial

R
management, 23, 70-74.

SC
Chung, K. H., Jo, H. (1996). The impact of security analysts' monitoring and marketing
functions on the market value of firms. Journal of Financial and Quantitative Analysis,
31(4), 493-512.

NU
Collier, J. Esteban, R. (2007). Corporate social responsibility and employee commitment.
Business Ethics: A European Review, 16(1), 19-33.
Deng, X., Kang, J. K., Low, B. S. (2013). Corporate social responsibility and stakeholder
MA
value maximization: Evidence from mergers. Journal of Financial Economics, 110(1),
87-109.
Edmans, A. (2011). Does the stock market fully value intangibles? Employee satisfaction and
equity prices. Journal of Financial Economics, 101(3), 621-640.
D

El Ghoul, S., Guedhami, O., Kwok, C. C., Mishra, D. R. (2011). Does corporate social
TE

responsibility affect the cost of capital? Journal of Banking & Finance, 35(9), 2388-2406.
Elsas, R., Hackethal, A., Holzhäuser, M. (2010). The anatomy of bank diversification.
Journal of Banking & Finance, 34(6), 1274-1287.
P

Eichenbaum, M. S., Hansen, L. P., Singleton, K. J. (1988). A time series analysis of

CE

representative agent models of consumption and leisure choice under uncertainty.

Quarterly Journal of Economics, 103(1), 51-78.
Eun, C. S., Wang, L., Xiao, S. C. (2015). Culture and R2. Journal of Financial Economics,
AC

115(2), 283-303.
Ferrell, O. C., Skinner, S. J. (1988). Ethical behavior and bureaucratic structure in marketing
research organizations. Journal of Marketing Research, 25(1), 103-109.
Ferreira, D., Ferreira, M. A., Raposo, C. C. (2011). Board structure and price informativeness.
Journal of Financial Economics, 99(3), 523-545.
Finkelstein, A., Luttmer, E. F. P. Notowidigdo, M. J. (2013). What good is wealth without
health? The effect of health on the marginal utility of consumption. Journal of the
European Economic Association 11, 221-258.
Freeman, R. E. (1984). Strategic management: A stakeholder perspective. Englewood Cliffs,
NJ, Prentice Hall.
Gao, F., Faff, R., Navissi, F. (2012). Corporate philanthropy: Insights from the 2008
Wenchuan earthquake in China. Pacific-Basin Finance Journal, 20(3), 363-377.
Graham, M. H. (2003). Confronting multicollinearity in ecological multiple regression.

26
 ACCEPTED MANUSCRIPT

Ecology, 84(11), 2809-2815.

Greening, D. W., Turban, D. B. (2000). Corporate social performance as a competitive
advantage in attracting a quality workforce. Business & Society, 39(3), 254-280.
Henderson, D. (2002). Misguided virtue: False notions of corporate social responsibility.
Institute of Economic Affairs, London.

T
Hillman, A. J., Keim, G. D. (2001). Shareholder value, stakeholder management, and social

IP
issues: What's the bottom line? Strategic Management Journal, 22(2), 125-139.
Hofstede, G. (1980). Culture's consequences: International differences in work-related values.

R
Beverly Hills, CA, Sage Publications.

SC
Hofstede, G. (1983). The cultural relativity of organizational practices and theories. Journal
of International Business Studies, 14(2), 75-89.
Hofstede, G. (1984). The cultural relativity of the quality of life concept. Academy of

NU
Management Review, 9(3), 389-398.
Hofstede, G. (1997). Culture and organizations: Software of the mind. New York, McGraw-
Hill.
MA
Hofstede, G. (2001). Culture's consequences: Comparing values, behaviors, institutions and
organizations across nations, Second Ed. Thousand Oaks, CA, Sage Publications.
Hood, J. N., Logsdon, J. M. (2002). Business ethics in the NAFTA countries: A cross-cultural
comparison. Journal of Business Research, 55(11), 883-890.
D

Ioannou, I., Serafeim, G. (2015). The impact of corporate social responsibility on investment
TE

recommendations: Analysts' perceptions and shifting institutional logics. Strategic

Management Journal, 36(7), 1053-1081.
Jones, D. A., Willness, C. R., Madey, S. (2014). Why are job seekers attracted by corporate
P

social performance? Experimental and field tests of three signal-based mechanisms.

CE

Academy of Management Journal, 57(2), 383-404.

Jo, H., Kim, H., Park, K. (2015). Corporate environmental responsibility and firm
performance in the financial services sector. Journal of Business Ethics, 131(2), 257-284.
AC

Kapelus, P. (2002). Mining, corporate social responsibility and the "community": The case of
Rio Tinto, Richards Bay Minerals and the Mbonambi. Journal of Business Ethics, 39(3),
275-296.
Karolyi, G. A. (2016). The gravity of culture for finance. Journal of Corporate Finance,
(forthcoming).
Kaufmann, D., Kraay, A., Mastruzzi, M. (2009). Governance matters VIII: aggregate and
individual governance indicators, 1996-2008. World bank policy research working paper.
Kim, H., Park, K., Ryu, D. (2015). Corporate environmental responsibility: A legal origins
perspective. Journal of Business Ethics, (in press).
Layard, R., Mayraz, G. Nickell, S. (2008). The marginal utility of income. Journal of Public
Economics 92(8–9), 1846-1857.
McSweeney, B. (2002). Hofstede’s model of national cultural differences and their
consequences: A triumph of faith – A failure of analysis, Human Relations, 55(1), 89-118.

27
 ACCEPTED MANUSCRIPT

McWilliams, A. Siegel, D. S. (2001). Corporate Social Responsibility: a Theory of the Firm

Perspective. Academy of Management Review 26(1), 117-127.
Miles, M., Covin, J. (2000). Environmental marketing: A source of reputational, competitive,
and financial advantage. Journal of Business Ethics, 23(3), 299-311.
Moon, Y. S., Franke, G. R. (2000). Cultural influences on agency practitioners' ethical

T
perceptions: A comparison of Korea and the U.S. Journal of Advertising, 29(1), 51-65.

IP
Morck, R., Yeung, B., Yu, W. (2000). The information content of stock markets: why do
emerging markets have synchronous stock price movements? Journal of Financial

R
Economics, 58(1), 215-260.

SC
Moskowitz, M. (1972). Choosing socially responsible stocks. Business and Society, 1, 71-75.
Peabody, D. (1985). National characteristics. Cambridge, UK, Cambridge University Press.
Porter, M. E., Van der Linde, C. (1995). Green and competitive: ending the stalemate.

NU
Harvard Business Review, 73(5), 120-134.
Roulet, T. J., Touboul, S. (2015). The Intentions with Which the Road is Paved: Attitudes to
Liberalism as Determinants of Greenwashing. Journal of Business Ethics, 128(2), 305-
MA
320.
Russo, M. V., Fouts, P. A. (1997). A resource-based perspective on corporate environmental
performance and profitability. Academy of management Journal, 40(3), 534-559.
Schwartz, S. H. (1994). Beyond individualism/collectivism: New cultural dimensions of
D

values. Sage Publications, Inc.

TE

Servaes, H., Tamayo, A. (2013). The impact of corporate social responsibility on firm value:
The role of customer awareness. Management Science, 59(5), 1045-1061.
Singelis, T. M., Triandis, H. C., Bhawuk, D. P. S., Gelfand, M. J. (1995). Horizontal and
P

vertical dimensions of individualism and collectivism: A theoretical and measurement

CE

refinement. Cross-Cultural Research, 29(3), 240-275.

Solomon, R. C., Hanson, K. R. (1985). It's good business. New York, Atheneum.
Stulz, R. M., & Williamson, R. (2003). Culture, openness, and finance. Journal of Financial
AC

Economics, 70(3), 313-3

Turban, D. B., Greening, D. W. (1997). Corporate social performance and organizational
attractiveness to prospective employees. Academy of Management Journal, 40(3), 658-
672.
Vitell, S. Festervand, T. (1987). Business ethics: Conflicts, practices and beliefs of industrial
executives. Journal of Business Ethics, 6(2), 111-122.
Vitell, S., Nwachukwu, S., Barnes, J. (1993). The effects of culture on ethical decision-
making: An application of Hofstede's typology. Journal of Business Ethics, 12(10), 753-
760.
Vitell, S. J., Paolillo, J. G. P., Thomas, J. L. (2003). The perceived role of ethics and social
responsibility: A study of marketing professionals. Business Ethics Quarterly, 13(1), 63-
86.
Waddock, S. A., Graves, S. B. (1997). The corporate social performance-financial

28
 ACCEPTED MANUSCRIPT

performance link. Strategic Management Journal, 18(4), 303-319.

Waldman, David A., et al. (2006). Cultural and leadership predictors of corporate social
responsibility values of top management: A GLOBE study of 15 countries. Journal of
International Business Studies, 37(6), 823-837.
White, H., (1980). A heteroscedasticity-consistent covariance matrix estimator and a direct

T
test for heteroscedasticity. Econometrica, 48, 817-838.

IP
Williamson, D., (2002), Forward from a critique of Hofstede’s model of national culture.
Human Relations, 55(11), 1373–1395.

R
Wintoki, M., Linck, J., Netter, J., (2012). Endogeneity and the dynamics of internal corporate

SC
governance. Journal of Financial Economics, 105(3), 581-606.

NU
MA
D
P TE
CE
AC

29
 ACCEPTED MANUSCRIPT

Appendix A
Variable description
Variable Description Source
Employee relation The social index, among ESG indexes (i.e., environmental, Thomson Reuters
social, and corporate governance), is used as the measure of ASSET4

T
employee relation, which is aimed to measure the quality of

IP
relation between a firm and its employees.
EW employee Equally weighted employee relation index is defined as the equal Authors’
relation weighting of the mostly employee-related (five) sub-social calculations based

R
indexes: (a) employment quality index, (b) training and on Thomson
development index, (c) health and safety index, (d) diversity Reuters ASSET4

SC
index, and (e) human rights index.
PCA employee The first principal component of employee relation index is Authors’
relation defined as the principal components of five categories: calculations based

NU
employment quality index, training and development index, on Thomson
health and safety index, diversity index, and human rights index. Reuters ASSET4
Power distance Power distance is defined as “the extent to which the members of Hofstede Centre
institutions and organizations within a country expect and accept
MA
that power is distributed unequally” (Hofstede, 1997). A perfect
score on power distance is 100 points. In high power-distance
societies, individuals admit the inequality of power and
differences between superiors and subordinates, are unwilling to
D

disagree with superiors when they are asked to follow some

specific rules and believe that superiors are eligible for special
TE

privileges (Hofstede, 1983). In contrast, in low power-distance

societies, individuals are less likely to tolerate the class
distinction and are not afraid to speak up when they disagree with
P

superiors.
Collectivism Collectivism/individualism “describes the relationship between Hofstede Centre
CE

/individualism the collectivity and the individual that prevails in a given society”
(Hofstede, 1980). A perfect score on collectivism/individualism
is 100 points. “Collectivism places more importance on interests
AC

of the group over those of individuals, emphasizing harmony and

closeness among the group members, while on the other hand,
individualism pertains to societies in which the ties between
individuals are loose; everyone is expected to look after himself
or herself and his or her immediate family” (Hofstede, 1997). In
collectivistic societies, a person considers her/himself as a part of
a group and values reciprocation of favors, a sense of belonging,
and respect for tradition (Schwartz, 1994). In individualistic
societies, people tend to value personal time and freedom
(Hofstede, 1984), pursue independence and pleasure (Schwartz,
1994), and seek to accomplish their personal goals.
Masculinity Masculinity/femininity dimension refers to clearly differentiated Hofstede Centre
/femininity social gender roles. A perfect score on masculinity/femininity is
100 points. Masculinity is defined as “a preference for
achievement, heroism, assertiveness, and material success”
(Hofstede, 1984). Individuals in masculine societies are supposed

30
 ACCEPTED MANUSCRIPT

to be assertive, aggressive, ambitious, competitive, and success-

oriented, while those in feminine societies are characterized as
tender, humble, caring, benevolent, and relationship-oriented
(Hofstede, 1984).
Uncertainty Uncertainty avoidance pertains to “the extent to which the Hofstede Centre

T
avoidance members of a culture feel threatened by uncertain and unknown
situation” (Hofstede, 1997). A perfect score on uncertainty

IP
avoidance is 100 points. People deal with uncertainty through
different means by resorting to technology, laws, and religions,

R
and the tolerance for uncertainty can vary culture by culture.
Individuals in the culture with high uncertainty are risk-averse,

SC
resist changes, place a premium on job security, favor written
rules, and are intolerant of deviations from standard practices in
contrast to those in the culture with low uncertainty avoidance

NU
(Hofstede, 1984).
Tobin’s q It is widely used as a measure of firm value and estimated as S&P Capital IQ
(book value of assets + market value of equity - book value of
equity) / (book value of assets).
MA
Leverage The ratio of book value of debt divided by total assets. S&P Capital IQ
Log (total assets) Defined as the logarithm of total assets. S&P Capital IQ
ROE Estimated as net income divided by the equity. S&P Capital IQ
Capital expenditure The ratio of capital expenditure expense to total sales. S&P Capital IQ
D

R&D/total assets Defined as the ratio of R&D expense divided by the total assets. S&P Capital IQ
Log (GDP per capita) Used as a measure of economic development and estimated as World Bank
TE

the logarithm of average of the yearly GDP per capita for each
country.
GDP growth Used as a measure of GDP growth volatility and estimated as the World Bank
P

volatility standard deviation of growth in GDP per capita for five years
(Eun et al., 2015)
CE

Governance Used to measure a country’s governance system similar to that of World Bank
indicators Morck et al. (2000) and estimated as the sum of six indexes from
Kaufmann et al. (2009). These indexes measure (i) voice and
AC

accountability, (ii) political stability and absence of

violence/terrorism, (iii) government effectiveness, (iv) regulatory
quality, (v) rule of law, and (vi) control of corruption.

31
 ACCEPTED MANUSCRIPT

Table 1 Descriptive statistics

Standard
Variable Obs. Mean Median P(0.25) P(0.75)
deviation
Employee relation 16,693 54.243 56.600 30.374 24.970 83.910

T
EW employee relation 16,693 49.192 48.807 10.962 41.831 56.809

IP
Power distance 16,693 44.650 40.000 12.881 35.000 54.000
Collectivism 16,693 26.886 11.000 21.856 9.000 49.000

R
Masculinity 16,693 62.247 62.000 19.994 62.000 66.000
Uncertainty 16,693 57.570 46.000 22.100 46.000 85.000

SC
Tobin’s q 16,693 1.263 0.808 1.967 0.566 1.355
Leverage 16,693 0.246 0.234 0.169 0.118 0.352

NU
Log (total assets) 16,693 8.705 8.596 1.350 7.768 9.579
ROE 16,693 0.123 0.120 0.279 0.057 0.199
Capital expenditure 16,693 0.090 0.045 0.157 0.025 0.089
MA
R&D/total assets 16,693 0.020 0.001 0.037 0.000 0.024
Log (GDP per capita) 16,693 10.585 10.672 0.415 10.545 10.787
GDP growth volatility 16,693 31.172 27.181 18.702 17.040 43.875
Governance indicators 16,693 7.770 8.140 2.145 7.300 8.870
D

This table reports the mean, median, standard deviation, first quartile, and third quartile for each variable
used in our empirical tests.
P TE
CE
AC

32
 ACCEPTED MANUSCRIPT

Table 2 Correlation matrix

Panel A. Correlations between Hofstede variables and Log (GDP per capita)
Log (GDP per capita)
2012 2011 2010

T
IP
Power distance -0.747 -0.709 -0.703
(0.000) (0.000) (0.000)

R
Collectivism -0.703 -0.585 -0.608
(0.000) (0.000) (0.000)

SC
Masculinity 0.147 0.172 0.187
(0.000) (0.000) (0.000)
Uncertainty avoidance -0.334 -0.186 -0.196
(0.000) (0.000) (0.000)

Panel B. Correlations between Hofstede variables

NU
MA
Power distance Collectivism Masculinity
0.751
Collectivism
(0.000)
-0.110 -0.404
D

Masculinity
(0.001) (0.000)
TE

0.320 0.516 -0.072

Uncertainty avoidance
(0.000) (0.000) (0.008)
This table presents the Pearson (Spearman) correlations between Hofstede cultural dimension and log
P

(GDP per capita), and between Hofstede cultural dimensions. Correlations in bold are significant at < 1%
level. p-value is in parentheses.
CE
AC

33
 ACCEPTED MANUSCRIPT

Table 3 Employee relation, firm value, and cultural difference: Fixed effect analysis

Tobin’s q
(1) (2) (3) (4) (5)
Employee relation: a 0.365*** 0.355*** 0.349*** 0.331*** 0.342***

T
(5.39) (5.30) (5.08) (4.74) (4.76)
axb 2.781***

IP
(3.72)
axc 0.717***

R
(2.85)
axd 1.215***

SC
(3.83)
axe 0.687**
(2.27)

NU
Power distance residual: b -2.397
(-1.21)
Collectivism residual: c -1.384
(-1.38)
MA
Masculinity residual: d -1.943*
(-1.75)
Uncertainty avoidance residual: e -0.938
(-1.01)
D

Leverage 0.184* 0.187* 0.178* 0.159 0.178*

(1.87) (1.90) (1.81) (1.62) (1.80)
TE

Log (total assets) -0.196*** -0.194*** -0.195*** -0.194*** -0.196***

(-10.69) (-10.72) (-10.59) (-10.56) (-10.66)
ROE 0.559*** 0.563*** 0.565*** 0.560*** 0.565***
P

(7.86) (7.87) (7.92) (7.89) (7.92)

Capital expenditure 0.054 0.064 0.043 0.043 0.049
CE

(0.46) (0.54) (0.36) (0.37) (0.42)

R&D/total assets 0.039*** 0.038*** 0.038*** 0.038*** 0.038***
(4.89) (4.85) (4.84) (4.82) (4.81)
Log (GDP per capita)
AC

-1.063*** -0.918* -0.715 -0.979*** -0.914***

(-3.60) (-1.74) (-1.57) (-3.23) (-2.64)
GDP growth volatility -0.003* -0.003 -0.003 -0.003* -0.003
(-1.71) (-1.59) (-1.49) (-1.78) (-1.48)
Governance indicators -0.092 -0.114 -0.128* -0.120* -0.119
(-1.40) (-1.56) (-1.71) (-1.85) (-1.46)
Intercept term Yes Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes Yes
Industry fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Robust standard errors Yes Yes Yes Yes Yes
R-squared 0.335 0.336 0.335 0.336 0.335
Number of observations 16,693 16,693 16,693 16,693 16,693
This table presents the estimation results from regressing Tobin’s q on employee relation (ER) and its
interaction terms with the residuals of Hofstede’s cultural dimensions over the period of 2002-2012. t-
statistics appear in brackets and are based on robust standard errors used in a study by White (1980).
***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.

34
 ACCEPTED MANUSCRIPT

Table 4 Adjusting sample composition bias I: Weighted least square (WLS)

Tobin’s q
(1) (2) (3) (4) (5)
Employee relation: a 0.349*** 0.349*** 0.358*** 0.312*** 0.341***
(6.76) (6.79) (6.99) (5.78) (6.49)

T
axb 2.689***

IP
(3.92)
axc 0.722***

R
(3.56)
axd 1.147***

SC
(4.08)
axe 0.684***
(2.87)
Power distance residual: b -3.297*

NU
(-1.78)
Collectivism residual: c -1.938**
(-2.06)
MA
Masculinity residual: d -2.081**
(-2.02)
Uncertainty avoidance residual: e -1.319
(-1.51)
Leverage 0.122 0.121 0.114 0.104 0.115
D

(1.48) (1.48) (1.39) (1.27) (1.40)

TE

Log (total assets) -0.200*** -0.197*** -0.198*** -0.197*** -0.199***

(-13.61) (-13.64) (-13.50) (-13.47) (-13.53)
ROE 0.480*** 0.484*** 0.486*** 0.482*** 0.486***
(8.78) (8.81) (8.87) (8.81) (8.87)
P

Capital expenditure 0.08 0.082 0.070 0.071 0.075

CE

(0.82) (0.84) (0.71) (0.73) (0.76)

R&D/total assets 0.041*** 0.041*** 0.041*** 0.040*** 0.040***
(6.03) (6.00) (5.98) (5.96) (5.96)
Log (GDP per capita) -0.972*** -0.610 -0.408 -0.895*** -0.713**
AC

(-3.55) (-1.23) (-0.95) (-3.19) (-2.18)

GDP growth volatility -0.001 -0.001 -0.001 -0.001 -0.001
(-0.89) (-0.55) (-0.44) (-0.71) (-0.46)
Governance indicators -0.010 -0.055 -0.077 -0.053 -0.064
(-0.18) (-0.82) (-1.12) (-0.93) (-0.83)
Intercept term Yes Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes Yes
Industry fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Robust standard errors Yes Yes Yes Yes Yes
R-squared 0.345 0.346 0.346 0.346 0.346
Number of observations 16,693 16,693 16,693 16,693 16,693
This table presents weighted least square (WLS) results for 16,693 firm-year observations over the period
of 2002-2012. t-statistics appear in brackets and are based on robust standard errors used in a study by
White (1980).
***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.

35
 ACCEPTED MANUSCRIPT

Table 5 Adjusting sample composition bias II

Panel A. Except U.S. sample
Tobin’s q
(1) (2) (3) (4) (5)
Employee relation: a 0.313*** 0.284*** 0.223* 0.263** 0.248*

T
(2.96) (2.72) (1.65) (2.40) (1.95)

IP
axb 2.604***
(3.43)
axc 0.813*

R
(1.87)

SC
axd 1.152***
(3.56)
axe 0.665*
(1.71)

NU
Power distance residual: b -0.798
(-0.36)
Collectivism residual: c -0.376
MA
(-0.34)
Masculinity residual: d -1.928
(-1.44)
Uncertainty avoidance residual: e 0.028
(0.03)
D

Leverage 0.478*** 0.480*** 0.473*** 0.441*** 0.466***

(3.12) (3.11) (3.07) (2.87) (3.01)
TE

Log (total assets) -0.180*** -0.176*** -0.177*** -0.176*** -0.178***

(-6.18) (-6.15) (-6.03) (-6.05) (-6.14)
ROE 0.679*** 0.681*** 0.680*** 0.681*** 0.681***
P

(5.13) (5.13) (5.13) (5.16) (5.13)

CE

Capital expenditure -0.123 -0.093 -0.138 -0.142 -0.131

(-0.53) (-0.41) (-0.60) (-0.61) (-0.57)
R&D/total assets 0.019 0.018 0.019 0.018 0.018
(1.49) (1.42) (1.45) (1.42) (1.41)
AC

Log (GDP per capita) -0.712*** -0.906 -0.764 -0.569* -0.823**

(-2.69) (-1.59) (-1.63) (-1.86) (-2.45)
GDP growth volatility -0.001 -0.001 -0.001 -0.001 -0.001
(-0.35) (-0.41) (-0.35) (-0.51) (-0.35)
Governance indicators -0.217** -0.221** -0.224** -0.204** -0.219**
(-2.28) (-2.32) (-2.33) (-1.97) (-2.26)
Intercept term Yes Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes Yes
Industry fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Robust standard errors Yes Yes Yes Yes Yes
R-squared 0.331 0.333 0.332 0.333 0.332
Number of observations 10,046 10,046 10,046 10,046 10,046

36
 ACCEPTED MANUSCRIPT

Panel B. Except Belgium, France, Japan, Poland, and Spain samples

Tobin’s q
(1) (2) (3) (4) (5)
Employee relation: a 0.424*** 0.537*** 0.449*** 0.508*** 0.494***

T
(4.88) (5.33) (5.08) (5.71) (5.52)
axb 4.884***

IP
(3.69)
axc 0.778**

R
(2.08)
axd 2.238***

SC
(3.61)
axe 1.249*
(1.90)

NU
Power distance residual: b -2.779
(-1.09)
Collectivism residual: c -0.923
(-0.73)
MA
Masculinity residual: d -2.282
(-1.53)
Uncertainty avoidance residual: e -1.252
(-1.04)
D

Leverage -0.088 -0.072 -0.084 -0.095 -0.083

(-0.70) (-0.58) (-0.67) (-0.76) (-0.66)
TE

Log (total assets) -0.238*** -0.231*** -0.237*** -0.238*** -0.237***

(-10.79) (-10.85) (-10.79) (-10.84) (-10.87)
ROE 0.572*** 0.569*** 0.576*** 0.568*** 0.577***
P

(7.50) (7.53) (7.52) (7.49) (7.52)

Capital expenditure 0.128 0.143 0.125 0.134 0.129
CE

(0.93) (1.03) (0.90) (0.97) (0.93)

R&D/total assets 0.042*** 0.043*** 0.042*** 0.042*** 0.042***
(4.42) (4.53) (4.44) (4.47) (4.41)
AC

Log (GDP per capita) -1.332*** -1.360** -1.190** -1.240*** -1.205***

(-3.87) (-2.16) (-2.24) (-3.60) (-3.23)
GDP growth volatility -0.013*** -0.013*** -0.013*** -0.013*** -0.012***
(-4.39) (-4.39) (-4.35) (-4.52) (-4.22)
Governance indicators -0.417*** -0.444*** -0.428*** -0.437*** -0.437***
(-4.26) (-4.35) (-4.16) (-4.49) (-4.02)
Intercept term Yes Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes Yes
Industry fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Robust standard errors Yes Yes Yes Yes Yes
R-squared 0.341 0.343 0.341 0.343 0.341
Number of observations 12,926 12,926 12,926 12,926 12,926
Panels A and B present robustness tests for alleviating the concern of sample composition bias from
regressing Tobin’s q on employee relation (ER) and its interaction terms with the residuals of Hofstede’s
cultural dimensions. t-statistics appear in brackets and are based on robust standard errors used in a study
by White (1980). ***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.

37
 ACCEPTED MANUSCRIPT

Table 6 Alternative independent variable I: Equally weighted employee relation

Tobin’s q
(1) (2) (3) (4) (5)
EW employee relation: a* 0.747*** 0.787*** 0.779*** 0.666*** 0.732***
(4.64) (4.75) (4.85) (4.10) (4.54)

T
a*xb 7.851***

IP
(4.03)
a*xc 1.933***
(2.93)

R
a*xd 4.612***
(5.10)

SC
a*xe 2.532***
(3.25)
Power distance residual: b -4.704**

NU
(-2.10)
Collectivism residual: c -1.925*
(-1.85)
Masculinity residual: d -3.537***
MA
(-2.80)
Uncertainty avoidance residual: e -1.852*
(-1.72)
Leverage 0.187* 0.177* 0.178* 0.158 0.170*
D

(1.90) (1.80) (1.80) (1.60) (1.71)

Log (total assets) -0.181*** -0.180*** -0.182*** -0.179*** -0.182***
TE

(-10.44) (-10.46) (-10.53) (-10.36) (-10.65)

ROE 0.565*** 0.570*** 0.570*** 0.565*** 0.572***
(7.95) (7.96) (7.99) (7.96) (8.01)
P

Capital expenditure 0.037 0.050 0.032 0.027 0.035

(0.31) (0.43) (0.27) (0.23) (0.30)
CE

R&D/total assets 0.040*** 0.040*** 0.040*** 0.039*** 0.039***

(5.13) (5.06) (5.04) (5.02) (4.98)
Log (GDP per capita) -1.115*** -0.982* -0.779* -0.956*** -0.978***
AC

(-3.79) (-1.86) (-1.71) (-3.16) (-2.82)

GDP growth volatility (5yr) -0.003 -0.003 -0.002 -0.003* -0.002
(-1.63) (-1.48) (-1.41) (-1.95) (-1.34)
Governance indicators -0.088 -0.110 -0.122 -0.124* -0.115
(-1.32) (-1.51) (-1.62) (-1.89) (-1.41)
Intercept term Yes Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes Yes
Industry fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Robust standard errors Yes Yes Yes Yes Yes
R-squared 0.334 0.335 0.335 0.336 0.335
Number of observations 16,693 16,693 16,693 16,693 16,693
This table reports robustness tests by applying alternative measure of employee relation, EW employee
relation, for 16,693 firm-year observations over the period of 2002-2012. t-statistics appear in brackets and
are based on robust standard errors used in a study by White (1980).
***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.

38
 ACCEPTED MANUSCRIPT

Table 7 Alternative independent variable II: Principal components analysis

Tobin’s q
(1) (2) (3) (4) (5)
PCA employee relation: a** 0.454*** 0.483*** 0.478*** 0.411*** 0.451***
(4.83) (4.94) (5.08) (4.34) (4.82)

T
a**xb 4.518***

IP
(4.00)
a**xc 1.155***

R
(2.93)
a**xd 2.758***

SC
(5.14)
a**xe 1.512***
(3.29)
Power distance residual: b -0.836

NU
(-0.44)
Collectivism residual: c -0.976
(-1.00)
MA
Masculinity residual: d -1.269
(-1.21)
Uncertainty avoidance residual: e -0.610
(-0.70)
Leverage 0.188* 0.179* 0.180* 0.160 0.172*
D

(1.91) (1.82) (1.81) (1.62) (1.73)

TE

Log (total assets) -0.181*** -0.181*** -0.183*** -0.179*** -0.183***

(-10.48) (-10.50) (-10.56) (-10.41) (-10.70)
ROE 0.565*** 0.569*** 0.569*** 0.564*** 0.571***
(7.95) (7.96) (7.99) (7.96) (8.01)
P

Capital expenditure 0.036 0.049 0.031 0.026 0.034

CE

(0.30) (0.42) (0.27) (0.22) (0.29)

R&D/total assets 0.040*** 0.040*** 0.040*** 0.039*** 0.039***
(5.13) (5.06) (5.04) (5.02) (4.98)
Log (GDP per capita) -1.116*** -0.979* -0.778* -0.960*** -0.978***
AC

(-3.79) (-1.85) (-1.70) (-3.17) (-2.82)

GDP growth volatility -0.003 -0.003 -0.002 -0.003* -0.002
(-1.62) (-1.48) (-1.41) (-1.94) (-1.33)
Governance indicators -0.088 -0.110 -0.122 -0.125* -0.116
(-1.33) (-1.51) (-1.63) (-1.91) (-1.42)
Intercept term Yes Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes Yes
Industry fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Robust standard errors Yes Yes Yes Yes Yes
R-squared 0.334 0.336 0.335 0.336 0.335
Number of observations 16,693 16,693 16,693 16,693 16,693
This table reports principal components analysis (PCA) results for 16,693 firm-year observations over the
period of 2002-2012. t-statistics appear in brackets and are based on robust standard errors used in a study
by White (1980). ***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.

39
 ACCEPTED MANUSCRIPT

Table 8 Mitigating endogeneity problem: Two-stage least squares (2SLS)

Tobin’s q
(1) (2) (3) (4) (5)
Employee relation: a 0.345*** 0.327*** 0.322*** 0.324*** 0.316***
(3.54) (3.36) (3.27) (3.30) (3.18)

T
axb 2.861***

IP
(5.63)
axc -2.544***

R
(-3.20)
axd 0.753***

SC
(2.92)
axe -1.456***
(-3.64)
Power distance residual: b 1.208***

NU
(5.11)
Collectivism residual: c -1.994***
(-4.85)
MA
Masculinity residual: d 0.715***
(3.13)
Uncertainty avoidance residual: e -0.994***
(-3.22)
Leverage 0.198** 0.201** 0.192** 0.174** 0.191**
D

(2.30) (2.35) (2.23) (2.02) (2.23)

TE

Log (total assets) -0.194*** -0.191*** -0.192*** -0.193*** -0.193***

(-11.93) (-11.73) (-11.71) (-11.83) (-11.81)
ROE 0.559*** 0.564*** 0.565*** 0.560*** 0.566***
(11.94) (12.06) (12.08) (11.98) (12.08)
P

Capital expenditure 0.052 0.062 0.04 0.043 0.047

CE

(0.52) (0.63) (0.40) (0.43) (0.47)

R&D/total assets 0.039*** 0.038*** 0.038*** 0.038*** 0.038***
(7.03) (6.96) (6.97) (6.88) (6.88)
Log (GDP per capita) -1.069*** -0.904*** -0.704*** -0.982*** -0.908***
AC

(-4.72) (-3.25) (-2.65) (-4.33) (-3.79)

GDP growth volatility -0.003* -0.003* -0.003 -0.003* -0.003*
(-1.88) (-1.69) (-1.63) (-1.81) (-1.69)
Governance indicators -0.090 -0.113** -0.128** -0.118** -0.118**
(-1.62) (-1.99) (-2.24) (-2.12) (-2.07)
Intercept term Yes Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes Yes
Industry fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
F-test of instruments (P-Value) 0.000 0.000 0.000 0.000 0.000
Sargan overidentification test 3.660 3.805 3.607 3.903 3.672
(P-Value) 0.160 0.149 0.165 0.142 0.159
R-squared 0.326 0.328 0.327 0.328 0.327
Number of observations 16,592 16,592 16,592 16,592 16,592
This table shows the robustness results for mitigating endogeneity problem using two-stage least squares
(2SLS). t-statistics appear in brackets.
***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.
40
 ACCEPTED MANUSCRIPT

Table 9 Mitigating endogeneity problem: Dynamic system generalized method of moments

Tobin’s q
(1) (2) (3) (4) (5)
Employee relation: a 0.863** 0.775** 0.731** 0.814** 0.814**

T
(2.31) (2.24) (2.33) (2.26) (2.38)
axb 0.696*

IP
(1.89)
axc 0.448**

R
(2.04)
axd 0.376**

SC
(2.23)
axe 0.311*
(1.88)

NU
Power distance residual: b -2.009
(-1.48)
Collectivism residual: c -0.552*
(-1.66)
MA
Masculinity residual: d -0.899***
(-3.64)
Uncertainty avoidance residual: e -0.412**
(-2.24)
Tobin’s Qt-1
D

-0.015 -0.013 -0.015 -0.015 -0.014

(-1.43) (-1.13) (-1.35) (-1.32) (-1.28)
TE

Leverage 0.223 0.089 0.180 0.199 0.191

(1.61) (1.02) (1.37) (1.50) (1.44)
Log (total assets) -0.593** -0.605** -0.521** -0.564** -0.546**
P

(-2.20) (-2.21) (-1.99) (-2.08) (-2.04)

ROE 0.348 0.448** 0.406** 0.431** 0.422**
CE

(1.59) (2.04) (2.09) (2.10) (2.05)

Capital expenditure -0.065* -0.056 -0.053 -0.055 -0.057*
(-1.81) (-1.62) (-1.61) (-1.62) (-1.65)
AC

R&D/total assets -0.095*** -0.075* -0.080** -0.076** -0.083**

(-2.65) (-1.87) (-2.15) (-2.05) (-2.22)
Log (GDP per capita) -0.269 0.806 0.195 0.144 0.139
(-0.57) (1.33) (0.98) (1.37) (1.17)
GDP growth volatility 0.002 0.001 0.001** 0.001*** 0.001***
(1.53) (0.69) (2.54) (3.01) (2.80)
Governance indicators 0.072 0.067 0.086 0.080 0.086
(1.03) (0.90) (0.87) (0.92) (0.91)
AR(1) test 0.000 0.000 0.000 0.000 0.000
AR(2) test 0.121 0.107 0.105 0.113 0.106
Hansen test of Overidentification 0.130 0.106 0.111 0.103 0.118
Diff-in-Hansen test of Exogeneity 0.527 0.435 0.601 0.471 0.656
Number of Observations 9,586 9,586 9,586 9,586 9,586
This table shows the robustness results for mitigating endogeneity problem using dynamic system generalized method of
moments (GMM). The AR(1) and AR(2) tests are tests for first-order and second-order serial correlation in the first differenced residuals.
The null hypothesis is no serial correlation. The Hansen test of over-identifying restrictions is a test with the joint null hypothesis that
instrumental variables are valid, i.e., uncorrelated with error terms. The difference-in-Hansen test of erogeneity is a test with the null
hypothesis that the subsets of instruments that we use in the levels equation are exogenous. Robust z-statistics are presented.
***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.
41
 ACCEPTED MANUSCRIPT

Table 10 Granger causality tests

Employee EW employee Tobin’s PCA employee

Tobin’s qt Tobin’s qt
relationt relationt qt relationt
(1) (2) (3) (4)
Tobin’s qt-1

T
-0.007 0.001 -0.024** 0.001 -0.024** 0.001
(-0.07) (0.28) (-2.06) (0.32) (-2.07) (0.26)

IP
Tobin’s qt-2 0.099 0.001 0.067*** 0.001 0.067*** 0.001
(1.01) (0.71) (5.84) (0.87) (5.83) (0.43)

R
Tobin’s qt-3 0.187** 0.001 0.258*** 0.001 0.258*** 0.001
(2.28) (0.55) (26.97) (0.95) (26.97) (0.93)

SC
Employee relationt-
0.005 0.676***
1
(0.06) (58.50)

NU
Employee relationt-
0.193** 0.133***
2
(2.09) (10.30)
Employee relationt-
MA
0.016 0.016
3
(0.15) (1.61)
EW employee
-0.003 1.591***
relationt-1
D

(-0.01) (72.33)
EW employee
0.900** 0.179***
TE

relationt-2
(2.01) (6.32)
EW employee
-0.507 0.166***
P

relationt-3
(-1.45) (7.14)
CE

PCA employee
0.036 0.937***
relationt-1
(0.17) (73.25)
AC

PCA employee
0.520** 0.109***
relationt-2
(1.96) (6.59)
PCA employee
-0.331 0.086***
relationt-3
(-1.59) (6.36)
Leverage 0.197** -0.009 0.167* -0.009* 0.167* -0.009
(2.32) (-0.92) (1.93) (-1.70) (1.92) (-1.64)
-
Log (total assets) -0.064*** 0.015*** -0.053*** -0.001 0.001
0.053***
(-3.93) (10.60) (-4.16) (-1.42) (-4.16) (0.52)
ROE 0.194*** 0.015*** 0.192*** 0.009*** 0.192*** 0.010***
(3.09) (2.95) (4.15) (3.04) (4.15) (3.41)
Capital expenditure -0.217*** -0.015 -0.252** -0.004 -0.254** -0.009
(-3.12) (-1.26) (-2.19) (-0.59) (-2.21) (-1.41)
R&D/total assets 0.011** 0.002*** 0.011* 0.001* 0.011* 0.001**
(1.97) (3.16) (1.92) (1.74) (1.92) (2.34)
42
 ACCEPTED MANUSCRIPT

Log (GDP per

-0.142 -0.046* -0.009 0.013 -0.010 0.011
capita)
(-0.74) (-1.66) (-0.04) (0.83) (-0.04) (0.72)
GDP growth
0.008*** -0.001 0.007*** 0.001 0.007*** 0.001
volatility (5yr)

T
(5.16) (-0.48) (4.56) (0.59) (4.56) (0.50)
Governance

IP
0.165** -0.043*** 0.237*** -0.008 0.236*** -0.011**
indicators
(1.99) (-4.67) (3.36) (-1.57) (3.34) (-2.27)

R
Intercept term Yes Yes Yes Yes Yes Yes
Country fixed
Yes Yes Yes Yes Yes Yes

SC
effects
Industry fixed
Yes Yes Yes Yes Yes Yes
effects
Year fixed effects Yes Yes Yes Yes Yes Yes

NU
Robust standard
Yes Yes Yes Yes Yes Yes
errors
R-squared 0.436 0.809 0.401 0.890 0.401 0.890
Number of
9,586 9,586 9,586 9,586 9,586 9,586
MA
observations
This table reports the results of Granger causality tests for 16,693 firm-year observations over the period of
2002-2012. The 1-, 2-, and 3-year lagged Tobin’s q, employee relation, EW employee relation, PCA
employee relation variables are used to estimate the model. The figures are in bold type if Granger
causality test is statistically significant at the 5% level or lower. t-statistics appear in brackets and are
D

based on robust standard errors used in a study by White (1980).

TE

***, **, and * denote statistical significance at the 1, 5 and 10 % levels, respectively.
P
CE
AC

43
 ACCEPTED MANUSCRIPT

Highlights

 Employee relationship responsibility(ERR) is a sub-sector in CSR worthy of an analysis.

 ERR covers diverse areas (ex. human rights) and builds up the reputation of a firm.

T
 Firms with good ERR demonstrate significantly higher levels of firm value.

IP
 Positive impact of ERR on firm value is strengthened with national culture.

R
 This seems to be caused by greater marginal effect of ERR in some cultures.

SC
NU
MA
D
P TE
CE
AC

44