Is There a Dark Side to Philanthropy?
Evaluating the Relationship Between Corporate Giving, Corporate Volunteering and Corporate Social Irresponsibility
Liset van Eibergen Santhagens S3179389
University of Groningen Faculty of Economics and Business
Master Business Administration – Change Management 01 / 03 / 2021
Supervisor: dr. Björn Mitzinneck Co-Assessor: dr. I. Maris-de Bresser
Words: 10130
Abstract
Prior research has established the notion that corporate social responsibility (CSR) and corporate social irresponsibility (CSiR) can occur concurrently or subsequent to one another.
Drawing on moral licensing theory, this study examines whether a similar relationship exists between corporate charitable donations and employee volunteering and subsequent corporate social irresponsibility. In addition, this study tests whether the number of liberal directors moderate the moral licensing effect. The empirical analysis is based on a 14-year panel dataset covering 149 firms from the S&P500, specifically from highly polluting industries. The analysis shows correlation between charitable donations and subsequent CSiR, but not for employee volunteering and subsequent CSiR. Evidence was also found for a moderating effect of number of liberal directors on the licensing effect with regards to charitable donations. Thus, corporate philanthropy can be used to acquire a moral license which can accommodate later corporate misconduct. The number of liberal directors can weaken the licensing effect in terms of charitable donations.
Key words: Corporate Social Irresponsibility, Corporate Philanthropy, Moral Licensing
Theory, Political Ideology, Liberalism
Introduction
Wells Fargo, one the largest and powerful banks of the US, states in its annual report that the bank’s philanthropic activities in 2019 equal $455 million. It also announces a philanthropic commitment of $1 billion over the coming six years to help address the growing housing affordability crisis in the United States (Wells Fargo, 2019, pp. 18 - 19). The same bank recently faced criminal charges for mistreatment of customers over a 14-year period, leading to a settlement of $3 billion. Unrealistic sales goals together with badgering and intimidation of employees resulted in the commitment of widespread fraud for several years (Washington Post, 2020; The New York Times, 2020). While the bank claims to drive positive impact on society (Wells Fargo, 2019) their imposed fines for corporate misconduct from the financial crisis onwards, amount to more than $18 billion (The New York Times, 2020).
Moral licensing theory may offer an insight into why organizations have the tendency
to show both caring and careless behavior towards stakeholders, as the theory states that one
may have the tendency to reward oneself with a moral license that allows for later unethical
behavior (Ormiston & Wong, 2013; Monin & Miller, 2001). This is reflected in keeping a moral
balance towards themselves as well as towards others. In Wells Fargo’s case, the recent
donation of $175 million for the public health crisis may be rewarded with a moral license that
helps to maintain a positive moral image towards themselves and their surroundings, and
this license may subsequently be used to permit misconduct. Corporate misconduct by
organizations is generally referred to as corporate social irresponsibility (CSiR) and in contrast
to corporate social responsibility (CSR), CSiR has a negative impact on a firm’s stakeholders
and their interests. Over the years, scholars have increasingly studied the relationship
between CSR and CSiR of which the different findings suggest that the phenomena can occur
simultaneously and/or subsequent to one another (Ormiston & Wong, 2013; Strike et al., 2006;
Kotchen & Moon, 2011). These studies however, have not considered corporate philanthropy as a distinct activity separate from CSR initiatives. Whereas CSR has become more regulated and more connected to sustainability, corporate philanthropy is more ad hoc in nature and not bound to certain domains (von Schnurbein et al., 2016). This is illustrated by Wells Fargo, as its philanthropy ranges from housing affordability to financial health, as well as a $175 million donation to public needs caused by COVID-19. Especially the voluntariness of corporate philanthropy is what makes the activity of interest with regards to corporate misconduct, where it may be used to address the negative impact of CSiR. In a recent study, Wu et al. (2020) shed light on philanthropy with regards to misconduct and found a positive relationship between charitable donations and pollutant emissions. Rather than moral licensing theory, this study draws on legitimacy theory to develop a theoretical framework that explains how donations are used to offset the harm to a firm’s image caused by environmental misconduct.
Based on moral licensing theory, this study will aim to create an understanding in how
organizations use corporate philanthropy to offset corporate misconduct. The relationship
between philanthropy and CSiR will be examined first. The distinction will be made between
two kinds of philanthropy, namely charitable donations and employee volunteering, for
which separate models will be used. By building on upper echelon theory, it is assumed that
directors can influence the overall strategy of an organization through the use one’s personal
experiences, values and beliefs (Hambrick, 2018). Political ideology, and specifically the
spectrum of liberalism – conservatism, is commonly used to measure one’s values and beliefs
(Chin et al., 2013; Christensen et al., 2015). Liberals are usually more responsive towards
societal issues and are proponents of driving social change (Chin, Hambrick, & Trevino,
Political Ideologies of CEOs: The Influence of Executives' Values on Corporate Social Responsibili, 2013), which underlines liberals preference to pursuit a stakeholder approach over a shareholder approach. Since corporate misconduct can be the result of the neglectance of the firm’s stakeholders and liberals are more likely to incorporate these stakeholders, board liberalism is expected to have a moderating effect on the relationship between philanthropy and irresponsibility.
In this study, I will thus explore the relationship between corporate philanthropy and CSiR, by addressing the following research question:
“What is the effect of corporate philanthropy on corporate misconduct?”
With the application of moral licensing theory on an organizational level, this study aims to clarify how organizations may make grand philanthropic contributions while subsequently causing corporate misconduct. The sample of organizations is extracted from the ASSET4 ESG database and consists of large publicly traded US firms. The dataset includes longitudinal data that enables to see changes throughout time using a panel data analysis.
This study contributes to the theoretical understanding of moral licensing theory in terms of
the applicability of the theory on an organizational level on the concept of corporate
philanthropy. Moreover, the influence of political ideology on moral licensing behavior is
examined – which helps to understand the occurrence of philanthropy and subsequent
corporate misconduct.
Literature Review
Corporate Philanthropy
Nowadays with increased globalization and more attention for pressing social and environment issues, organizations are expected to do more for society than merely adhere to their legal responsibilities (de Paiva Duarte, 2013). Subsequently, performance is not only considered in terms of reaching financials goals (e.g. profit) but also with regards to environmental and societal goals (Caligiuri et al., 2013). A way to comply to this expectation for organizations is through the use of philanthropy. Corporate philanthropy is generally also referred to as corporate giving, which is seen as a form of corporate social responsibility (CSR).
Philanthropy means doing good to others, so that philanthropy on a firm-level translates to using resources, such as financials or employees, to support a social cause (Kabongo, 2013).
The three most common forms of corporate philanthropy are monetary donations,
employee volunteering and the establishment of a corporate foundation (von Schnurbein, et
al., 2016). Monetary donations are for example contributions made to direct causes as disaster
relief, community investments through grants, or contributions to the firm’s own foundation
(von Schnurbein et al., 2016). The use of corporate philanthropy in terms of disaster relief has
over the years become part of doing business (Muller & Kraüssl, 2011), which is shown after
hurricane Katrina when businesses were responsible for a third of the total funds received of
relief efforts. Similarly, the 2004 tsunami with a death toll above 150,000 triggered a large
group of US corporations as contributors to disaster relief (Muller & Whiteman, 2009). For
larger corporations, the monetary donations are usually processed through such foundations,
whereas smaller firms may be more inclined to donate directly to causes or nonprofits
(Kabongo, 2013). Employee volunteerism is a form of employee engagement, where high
employee engagement has found to positively affect matters as job satisfaction and involvement and intristic motivation (Opoku-Dakwa et al.,2018). Aside from monetary donations and volunteering, corporate philanthropy can take many other forms such as the formation of a foundation, support for housing and education, or the providing of expertise, facilities or products (de Paiva Duarte, 2013; Kabongo, 2013). According to a survey by Giving in Numbers, a benchmark of corporate social investment, a total amount of $24.8 billion was totally donated by 250 multi-billion businesses in 2019, of which 27% of the total was given to health and social services, followed by community and economic development, and education (CECP, 2020).
In his study on market reactions towards press releases about disaster relief, Patten (2008)
found that corporations announcing the donation of large amounts lead to more positive
market reactions in comparisons to the announcement of smaller amounts or the exact amount
of $1 million. The fact that most corporations decided to publicly announce their donation,
underlines a probable strategic gain from performing such philanthropic action. The
achievement of economic results or financial gains through the use of philanthropy is referred
to as strategic philanthropy (Kabongo, 2013). This approach to corporate giving indicates that
the engagement in philanthropic actions will eventually and in the long run benefit the
business itself in some way, with the rationale that corporations and society are
interdependent of each other (Kabongo, 2013). Philanthropy can thus benefit businesses in
multiple ways, such as the improvement of employee engagement through the
accommodation of volunteering programs, or the increase in customer loyalty as studies have
found that young people are likely to switch brands if associated with a social cause (Patten,
2008; de Paiva Duarte, 2013). Philanthropy can also play a major role in creating reputational
capital and thereby in terms of managing risks to the corporation’s reputation (Patten, 2008).
Reputation is an intangible asset and has further effect on the attracting of resources and subsequently on a business’ competitive advantage. De Paiva Duarte (2013) further emphasizes that a long-accumulated goodwill obtained through philanthropy can moderate possible negative effects of a firm’s mistakes. With philanthropic activity in mind, one is more likely to see an error as a singular mistake instead of the failure of the organization’s system.
In this way, the investment in corporate philanthropy either through systematic donations to communities or singular donations as disaster relief, creates financial benefits for corporations themselves together with the creation of social welfare.
Where philanthropy is a more responsive and an ad hoc action, CSR entails the incorporation of strategies into an organization that aim to fulfill social, environmental and economic goals that go beyond their legal requirements. The implementation of CSR initiatives ensures that other stakeholders than only the shareholders’ interests are looked after, which is in contrast to Milton Friedman’s (1970) thought that companies’ sole obligation to the public is to make profit. The incorporation of all stakeholders’ interests instead of solely the shareholders’, is also referred to as the stakeholder approach. Since stakeholders and firms are interdependent and thus have effect on one another, a well-built relationship with its stakeholders is of importance. Freeman (2010), founder and proponent of the stakeholder approach, argues that a strong relationship with stakeholders offers an organization insight into all possible effects that can influence the organization. Derived from stakeholder theory is the belief that the maximalization of benefits to stakeholders will benefit the firm itself (Caligiuri et al., 2013).
The implementations of CSR initiatives are diverse and differ per industry. Industrial
business can for example adjust their production processes so that their emissions are
reduced, whereas a coffee brand ensures fair trade throughout its supply chain or real estate’s
investment in sustainable buildings. CSR and philanthropy are both ways to conform to social demands and can therefore be used simultaneously or complementary to one another. Zollo (2004) illustrates that the difference between the two initiatives becomes clear in both the reasons for investing as the way of investing in a certain societal issue. The initiatives may both be tackling the same issue, though philanthropy is more ad hoc with social effect on the short term and a financial benefit in the long run. CSR in return integrates the societal goal in its operations thereby accounts for a more continuous impact on the issue.
Corporate Social Irresponsibility (CSiR)
Corporate scandals create harm to a great number of people every year, of which Wells Fargo’s recent fraud, the Volkswagen emission scandal and the 2008 financial crisis are well- known and relevant examples. Aside from harming people, another stakeholder that is often the victim of irresponsible corporate behavior is the environment, as shown in the oil spill in the Gulf of Mexico by British Petroleum (BP) in 2010 (Ormiston & Wong, 2013). These different scandals across different industries are all examples of corporate social irresponsibility, which shows that CSiR can take many forms and is not bound by industry. However, some might argue that some industries are more open to scrutiny than others as some industries themselves are controversial in nature, such as the oil or tobacco industry (Kotchen & Moon, 2011). Until the last decade, research had mainly focused on the effects of responsible behavior by organizations (CSR) rather than irresponsible behavior (CSiR) (Riera & Iborra, 2017).
Scholars still differ in their way of defining and conceptualizing CSiR hence no consensus yet
has been established on the concept. For instance, Kotchen & Moon (2011, p. 2) refer to CSiR
as “a set of actions that increases externalized costs and/or promotes distributional conflicts”,
thereby focusing on the effects of CSiR on the company itself. Aside from probable penalties
with financial loss, media coverage of irresponsible behavior exposes companies to financial risk (Kölbel et al., 2017) together with the loss of legitimacy (Wu et al., 2020). In turn, a damaged reputation can lead to difficulties with regards to attracting employees, customers and investors (Alcadipani & de Oliveira Medeiros, 2020). In defining CSiR, Riera & Iborra (2017) place more emphasis on the negative effects on the stakeholders and argue the intentionality of CSiR where scandals are the result of strategy rather than isolated events.
Here, CSiR is the result of an organization’s primary focus on self-interest, profit and shareholders’ interests thereby affecting the interests of other stakeholders, such as customers, society and the environment (Jones, 2013).
With interest in CSiR growing over the past years, scholars have started to explore the concept in relation to CSR, rather than seeing the concepts as conceptually the same. The findings of these studies suggest that CSR and CSiR can occur simultaneously, as well as that CSiR offsets CSR and vice versa (Kotchen & Moon, 2011; Ormiston & Wong, 2013; Strike et al., 2006). These results indicate that corporations can behave ethically in one domain of the organization and act irresponsible in another, or that CSR campaigns are started as a response to CSiR in order to limit the negative effects of the irresponsible behavior (Jones, 2013). The findings in these studies are crucial in dismissing the notion that firms are either “good” or
“bad” and show instead that firms can do good and bad deeds both simultaneously (Muller
& Kraüssl, 2011).
Corporate Philanthropy and Corporate Social Irresponsibility
The occurrence of both behaviors raises the question whether and why socially desired behavior may drive irresponsible behavior, and whether this is the case with corporate philanthropy and corporate social irresponsibility. Whereas the aforementioned studies considered CSR, Williams & Barrett (2000) found that philanthropy has a significant moderating effect on the relationship between a firm’s violations and its reputation. The association with irresponsible behavior can harm the reputation of an organization, whereupon the use of philanthropy as a kind of tool can reduce the negative impacts of misconduct on the firm’s reputation (Williams & Barrett, 2000). Muller and Kraüssl (2011) indicate a more direct link between irresponsibility and philanthropy in their study on disaster donations, firm value and reputation. Their findings show that firms that have a reputation associated with irresponsibility, are more likely to make charitable donations as a means of disaster relief. Similarly, Wu et al., (2020) found a positive relation between charitable donations and pollutant emissions. The study differentiates between two type of strategies that small Chinese companies use towards donations: a fire-suppressing approach and the proactive approach, of which the first approach entails making donations after exposure of irresponsible behavior, and the latter comprises long-term donations that build reputational capital as well as moral capital (Wu et al., 2020).
The emphasis of these studies is mainly on the acquiring of reputational capital whereas
Ormiston & Wong (2013) introduce the idea of moral capital through the use of moral licensing
theory on a firm level. This theory states that individuals have the tendency to keep a kind of
moral balance towards themselves aswell as towards their surroundings (Monin & Miller,
2001; Ormiston & Wong, 2013; Blanken, Zeelenberg, & van de Ven, 2015). A requirement of
moral licensing theory to hold is that the morally right act is performed before the morally
wrong act. In this way, an individual may engage in an extent of good deeds after which this person is under the impression that enough moral credits have been acquired to subsequently behave socially indesirable (Monin & Miller, 2001). Thus, moral credits can be acquired through engaging in socially desirable behavior which establishes a moral balance (Godfrey et al., 2009). The possession of this moral balance then enables people to feel licensed to behave irresponsibly (Ormiston & Wong, 2013). Where Ormiston & Wong (2013) applied this theory to firms engaging in both CSR and CSiR, the theory can also be applied to the relationship between corporate philanthropy and CSiR which are also deemed as good and bad deeds.
Here, firms would engage in corporate philanthropy through either monetary donations or employee volunteering, and engage in irresponsible behavior afterwards. The sequence of this behavior is opposed to the more logical or natural sequence where one tries to compensate for an immoral act and does a good deed after, which can be explained through the feeling of guilt. Aside from guilt, the use of philanthropy after the exposure of corporate social irresponsible behavior is also a way to make up for a damaged reputation. When a corporation publicly announcemences charitable giving, the underlying thought may be that the corporation is in fact licensing itself.
As Ormiston & Wong (2013) explain, moral licencing theory was initially found to be
domain-specific, which means that the both moral as immoral behavior occur in the same
domain. Mazar and Zhong (2010) however found that both acts do not necessarily have to be
in the same domain, and that for example environmentalism and subsequent giving less to
charity may also be explained through moral licensing theory. This finding allows moral
licensing to be studied in a more broader sense when for instance the precise domains of the
moral and immoral acts are unknown.
This study builds on the prior finding of Ormiston & Wong (2013) that prior CSR offsets subsequent CSIR, confirming the applicability of moral licensing theory on a firm level.
Whereas Wu et al. (2020) and Muller & Kraüssl (2011) uses legitimacy theory and reputation to explain the positive relationship that was found between charitable giving and irresponsibility, this study will build on moral licensing theory to explain the relationship between corporate philanthropy and corporate social irresponsibility. This contributes to the little understanding there is regarding the application of moral licensing theory on a firm level, rather than the individual or group level. This study also extends the current understanding there is regarding the relationship between philanthropy and CSiR by focusing on moral credits instead of reputational credits. In order to gain a valuable insight into this relationship, the different forms of corporate philanthropy are considered. The most common forms of employee volunteering and charitable giving are examined, of which donations to, or through foundations are included in charitable giving. Finally, the following hypotheses are stated with regards to the relationship between corporate philanthropy and corporate social irresponsibility:
H1: Firms that do charitable donations are more likely to be involved in corporate social irresponsibility
H2: Firms that engage in employee volunteering are more likely to be involved in corporate
social irresponsibility
Political Ideology and its moderating effects
The board of directors together with the CEO form the highest layer of decision-making and accordingly set the strategy and make day-to-day decisions, which can include the decision to engage in philanthropy to glean moral credits. Some theories, such as institutional theory, argue that directors are limited in influencing the decision-making process through existing constraints that are the result of internal and external forces (Marquis & Tilcsik, 2016).
In that view, individual directors as well as the entire board, has limited influence on the actual strategy, performance and policies that an organization pursues. Advocates of the upper echelon theory consider these constraints but argue that directors’ attributes are the starting point of understanding an organization’s behavior (Hambrick, 2018). Upper echelon theory assumes that directors interpret situations in which decisions are to be made through personalized lenses. The directors are then likely to draw upon prior experiences and their characteristics to both interpreted and react to the situations and strategic decisions that they are facing (Hambrick, 2018). In this way, directors are likely to influence the decision-making process through the use of these personalized lenses (Hambrick, 2018). Subsequently, an organization’s behavior may mirror the characteristics, experiences and values of its board of directors. In turn, these attributes may influence the firm’s strategy to engage in moral licensing behavior
Political orientation, especially on the liberalism – conservatism dimension, is seen as an important and stable reflection of someone’s underlying values, that influence their attitude towards matters such as risk and uncertainty (Christensen et al., 2015; Chin et al., 2013).
Schwartz (1996) found that liberals are more likely to be concerned with social issues such as
human rights or the environment. Subsequently, liberals are often proponents of social change
(Chin et al., 2013). In contrast, conservatives beliefs are against change as well as uncertainty,
ambiguity, risk and threats (Peterson et al., 2009). Moreover, Schwartz (1996) found that conservatives are more individualistic and attach more value to a stable and free market, and are therefore generally more concerned with maintaining the status quo.
CSiR can harm an organization’s legitimacy (Wu et al., 2020), reputation and ability to attract resources (Alcadipani & de Oliveira Medeiros, 2020), and exposes firms to financial risks (Kölbel et al., 2017). Taking these consequences together, the exposure of corporate misconduct can form a threat to the stability of an organization. Considering conservatives’
aversion for instability, risk and uncertainty, conservatives may be more inclined to establish a strategy that battles the possible negative consequences of misconduct that threathen the status quo. Wu et al., (2020) underline that the proactive approach towards donations yield reputational and moral credits, that help protect the firm’s legitimacy and thus helps preserving the status quo. Moral licensing too highlights the yielding of moral credits, that in turn help maintain a moral balance towards others. Where the liberal ideology is less concerned with maintaining the status quo, conservatives may be more likely than liberals to perform a good deed, of which the purpose is to protect the organization from any later negative consequences of probable misconduct. Therefore, liberals may be less inclined to engage in corporate philanthropy with the sole purpose to acquire a moral license for later misconduct. Accordingly, I hypothesize the following moderating effect on both the relationship between donations and CSiR, as the relationship between employee volunteering and CSiR:
H3a: A higher number of liberal directors in a board of directors weakens the positive relationship between charitable giving and corporate social irresponsibility.
H3b: A higher number of liberal directors in a board of directors weakens the positive
relationship between employee volunteering and corporate social irresponsibility.
Figure 1: Conceptual Model
Methods
Data Collection and Sample Selection
The goal of this study is to gain an insight into the relationship between corporate philanthropy and corporate social irresponsibility with a possible moderating effect of political ideology of directors. In order to test my hypotheses, the dataset has to meet the requirements of including a measurement of irresponsible behavior, philanthropic activity and director’s political ideology. The suitable sample of firms has been taken from Standard
& Poor’s 500 (S&P 500), which is an index that follows the stock performance of the 500 largest
corporations listed on the United States’ stock exchange. The S&P is commonly used as a
sample source for studies in this academic field. The final sample consist of 149 firms with a
total of 1,393 observations between 2002 and 2014 and includes industries such as oil & gas,
mining, chemicals, construction and food & beverages. The selection of industries is based on
EPA’s report stating that these industries account for 84% of the total industrial emissions in
the United States (EPA, 2008). Companies in these industries have high levels of emissions
and may therefore be more exposed to causing environmental misconduct. Sampling based
on this assumption follows Wu et al., (2020). I expect that firms in these industries with high
emissions are likely to be penalized for exceeding the emission limit for example, which falls
under corporate social irresponsibility. Since this study considers moral licensing to be both
domain-specific as non-domain-specific, all kinds of corporate social irresponsibility across
different domains will be considered and not solely violations with regards to the
environment.
Variables on corporate social responsibility and philanthropy are extracted from Thomson Reuters’ Asset4, whereas variables on corporate social irresponsibility are sourced from the Kinder, Lydenberg and Domini (KLD) database. KLD includes different indicators of social performance in terms of strengths and concerns, of which the latter will be used to indicate corporate misconduct.
With regards to the moderator, the variables on boards’ and directors’ political ideology stem from the DIME database that tracks political contributions to the democrats or republicans of firms and board members. General information of companies such as total assets, number of employees and turnover have been extracted from CompuStat, as well as Asset4. Prior research on directors’ political ideology by Christensen et al., (2015) and Chin &
Semadeni (2017) used similar measurements that use individual political contributions.
Measurement of Variables Dependent Variable
The dependent variable in this study is corporate social irresponsible behavior, which has previously been defined as harm that is being done to organizations’ stakeholders. CSiR will be measured using the KLD measurements. KLD gathers the social performance concerns over the following different issue areas: corporate governance, community, diversity, employee relations, environment, human rights and product. Similar to prior studies on CSiR, the KLD concerns are used as a measurement of CSiR (Kotchen & Moon, 2011; Ormiston &
Wong, 2013; Strike et al., 2006; Muller & Kraüssl, 2011). The issue areas are further categorized
into sub-indicators, such as has union relations in the area of employee relations. The sub-
indicators are assessed and given the binary value of 0 or 1, where no or poor union relations
are given the value of 1. The sum of these sub-indicators account for the total number of concerns in an area. In this study, the sum of all issue areas and thus the sum of all sub- indictors will be used as a measurement of corporate misconduct. By doing so, I follow Mazar
& Zhong’s (2010) finding that the licensing effect can be domain specific as well as non- domain specific and focus on the non-domain specification by taking the sum of all issue areas. In this way, a more global approach is taken with regards to identifying the moral licensing effect, as prior research often considers the different domains.
Independent Variables
As illustrated in the literature review, corporate philanthropy is used in many different ways of which the two most common forms are through monetary donations and employee volunteering. For both variables has Thomson Reuters’ Asset4 been used as a source. With regards to the quality of this dataset, it came to my attention that during a random manual check of correctness for 22 observations, 4 observations turned out to be incorrect. The variables that were checked in this manual inspection were donations (total charitable contributions) and volunteering (employee volunteering). With the assumption that similar errors would be present in the sample that was taken from the original dataset, nearly 20 percent of the sample has been manually checked. The detailed methodology regarding this coding can be found in Appendix 1.
First, based on prior studies on corporate philanthropy (Jia & Zhang, 2014; Brown,
Helland, & Smith, 2006), donations is measured as relative donations. This variable is calculated
by dividing total donations by total revenues. Total donations include all monetary
contributions that a firm has done during a year including direct donations to causes (e.g.,
disaster relief) as well as donations to charitable foundations of firms themselves. The amounts are recorded in US dollars.
Second, the variable volunteering is measured and used as a dummy variable, following Hall & Rieck (1998), where 0 is recorded for no employee volunteering, and 1 is recorded if employees of a firm have participated in volunteering work during that year.
Moderator Variable
In this study, political ideology of the board of directors is expected to have a moderating effect on the relationship between CSiR and philanthropy. In prior studies that include political ideology, the American republican party and the democratic party have generally been translated to the continuum of conservatism and liberalism respectively.
Following Chin et al. (2013) I use political ideology as a reflection of one’s personal ideology.
Directors’ beliefs are expected to influence the organization’s way of serving stakeholders’
interest and thereby have effect on the level of CSiR. Building on the finding that liberalists are more likely to avoid doing harm than conservatives, the moderator variable is on board’s liberalism. The DIME database collects information on donation behavior on political parties and has been used as the source for this variable.
Board Liberalism is measured through the number of liberal directors in the board,
where naturally a higher number translates to a more liberal board, and a lower number
means a more conservative board.
Control Variables
Tobin’s Q is an indicator of market performance calculated by dividing equity market value by equity book value. Firm market performance has been found to be related to corporate social irresponsibility (Jain & Zaman, 2020) as well as corporate social responsibility (Ormiston & Wong, 2013) which can thus have an effect on the dependent variable of corporate social irresponsibility.
Firm size is controlled for through the logarithms of total assets and number of employees.
Larger companies may show better social performance and have higher ethical standards (Wu et al., 2020), which influences corporate misconduct. The size of a firm may be determinant for the level of corporate philanthropy. The use of logarithms controls for the risk of high skewness for both total assets and number of employees (Strike et al., 2006).
Gender ratio is measured through the share of males and females in a board of directors, where the values range between 0 and 1. A gender ratio close to 0 means more female directors and a ratio closer to 1 means more male directors. It was found that the number of women in a board was positively associated with CSR (Bear et al., 2010), and similarly a high proportion of women in boards are effective in reducing CSiR (Jain & Zaman, When Boards Matter: The Case of Corporate Social Irresponsibility, 2020). Therefore, the gender ratio of a board of directors is controlled for.
Governance is controlled for through the KLD indicator for corporate governance strengths, building on Jain & Zaman (2020) who found that certain board-level governance bundles are effective in reducing corporate social irresponsibility.
Mean Prior Corporate Social Irresponsibility is also controlled for its possible effect on
reputation and in turn on charitable donations, as argued by Muller & Kraüssl (2011).
Moreover, Williams & Barrett (2000) found a strong link between misconduct and a negative reputation. Based on this strong link, it is assumed that past misconduct greatly affects reputation regardless of the number of years ago the misconduct took place. Therefore, the mean of each firm’s corporate misconduct has been taken to account for differences in misconduct over the years.
Industry is controlled for, following Ormiston & Wong (2013) and Wu et al., (2020), because of the possible effect of industry on corporate social responsibility. A similar effect may be found for corporate social irresponsibility. Industry is controlled for by using a dummy variable based on the first two digits of the North American Industry Classification Code (NAICS).
Year fixed effects are controlled in order to rule out time trends that possibly influence the relationship between the dependent and independent variable. A dummy variable for year is included in the regression to control for this inference.
Data Analysis
The dataset covers 149 firms from the S&P 500 and includes variables that have been
collected in the time period of 2002 – 2014, making the dataset longitudinal. This means that
for each firm, multiple years of data has been gathered and included in this panel dataset. For
analyzing datasets with longitudinal characteristics, the panel data regression analysis is
generally used. The use of panel data regression enables the testing of hypotheses cross-
sectionally and across years. Especially the latter is of interest, because changes in behavior
over the years can be interpreted with the use of this analysis method. For this analysis, I have
used the statistical program StataSE 16.0. Before running the appropriate panel analysis, the
dataset must be declared to be a panel dataset first. In this step, every firm is assigned an id variable “t_id”, together with setting a time variable “year” – this is done by using the xtset command in Stata. The dataset that I am using is unbalanced according to Stata, which means that some observations are missing for some firms. A balanced dataset has for each year an equal number of variables observed. It is possible to correct an unbalanced dataset if the missing of data is caused by a certain kind of bias. In this dataset, it is expected that missing observations are the result of the data sources not being able to collect the data from, for example, firms’ reporting. Therefore, I will run the panel data regression analysis with an unbalanced panel dataset. The final dataset includes 149 firms with 1,393 observations. In this study, the sequence of events is of importance to test for moral licensing theory. In order to include such sequence, it was chosen to include a lag of 1 year in the independent, moderator variable and all control variables.
The three common methods for analyzing panel data are the fixed effect model, the
random effects mode and population-averaged mode. In deciding what model to use, I have
compared all models in terms of what they explore and assume, and what is most relevant
and suitable with regards to this particular study. The fixed effect model is relevant if the
assumption is that factors within a firm can bias the outcome variables which subsequently
needs to be controlled for. The use of the fixed model leads to the exclusion of time invariant
variables in order to measure the real effect of the predictors (Frees, 2004). The moderation
effect in this study is political ideology, which is seen as consistent over time and thus time
invariant (Chin & Semadeni, 2017; Christensen et al., 2015). Another time invariant variable
used in this study is the control variable for industry. The random effects model builds on the
assumptions that variations across firms are random and that the dependent variable is
influenced by the differences that exist between firms (Frees, 2004; Greene, 2003). However,
the random effects model may cause bias in the estimation of the coefficient, as the random effects model does not estimate separate unit effects (Clark & Linzer, 2012). The alternative analysis model is the population-averaged model, which generally uses the generalized estimation equation. Following Chin & Semadeni (2017), it was decided to use this alternative model to account for possible bias in the random effects model, and the exclusion of time- invariant variables in the fixed effect model. The population averaged model integrates non- independent observations and is generally appropriate for cross-sectional and time series panel data (Chin & Semadeni, 2017). Three separate models will be used for the two different kinds of philanthropy, and the third model will only include the control variables. The first model will include donations as the independent variable, and volunteering will be the independent variable in the second model.
Results
Diagnostics
Panel data regression analysis can be subject to modeling issues, and for that reason it is appropriate to run additional tests in order to check the reliability of the analysis and its results (Frees, 2004; Greene, 2003). These additional checks test for heteroscedasticity, collinearity and autocorrelation.
Heteroscedasticity occurs when the variance of the error term is found to be constant. The opposite, homoscedasticity means no consistency is present in the variance of the error term.
Heteroscedasticity can bias the standard errors, which in turn affects the test statistics and the
confident intervals as well. The null hypothesis stating that there is no homoscedasticity was
rejected for the model with relative donations (P > chi2 = 0.000) and supported for the model with volunteering (P > chi2 = 1.000), indicating the presence of homoscedasticity in the model with relative donations.
Autocorrelation, also referred to as serial correlation, occurs when there is a correlation between values for a certain variable across firms throughout the years in the panel data. The Woolridge test is often used to detect autocorrelation, where the null hypothesis is that there is no first-order autocorrelation. The Woolridge test was used for both the model where relative donations is the independent variable, and the model in which volunteering is the independent variable. For both tests, Prob > F < 0.05 which means the null hypotheses are rejected and it can be assumed that there is autocorrelation present in the analysis. The presence of autocorrelation is in principle not problematic for the analysis, though the standard errors must be interpreted with caution, since these may vary from the true standard errors. The use of robust standard errors in the regression is commonly used to overcome the reliability issues of the standard errors caused by autocorrelation and homoscedasticity. For that reason, for all models the robust standard errors have been added to the regressions.
Collinearity occurs when the independent variables are themselves related. In order to test for this, a Variation Inflation Factors (VIF) test was used. The test indicated that all variables scored between 1.014 and 3.315. Based on this result, no action was taken towards collinearity.
The table stating the results of the VIF test can be found in Appendix 2. Additionally, simple slopes analyses were performed to double check the moderating effect. The graphs are illustrated in Appendix 3.
For each model, robustness was checked through the substitution of variables that
measure the same concept, such as total assets instead of number of employees, and return on
assets for financial performance instead of Tobin’s q. The results of the robustness checks indicate that the results of using different measurements generate consistent outcomes.
Descriptive Statistics
In the table below, a summary of all the variables included in the analyses are listed. For relative donations and number of employees the natural log has been taken in order to correct for skewness. The total number of observations is 1,393 for 149 firms. The mean for corporate social irresponsibility is 3.358 which indicates an average of roughly 3 KLD concerns per firm.
Slightly more than half of the firms have participated in employee volunteering as the mean for volunteering is 0.621.
Obs. Mean Variance Min Max
Corporate Social Irresponsibility 1186 3.943 9.072 0 18
Relative Donations* 1393 .205 1.008 -11.765 5.805
Volunteering 1393 .621 .236 0 1
Number of Liberal Directors 1393 2.04 4.544 0 18
Tobin’s Q 1391 4.236 505.552 -189.297 713.515
Number of Employees* 1297 2.805 1.285 -.48 5.694
Gender Ratio 1375 .845 .01 .4 1
Governance Strengths 1319 .265 .246 0 3
Mean Prior Corporate Social Irresponsibility
1307 9.147 461.433 0 153.846
Figure 2:Descriptive Statistics. * = logged variable
In the figure 3, the covariates and correlations among all variables included in the regression are shown. The table shows that the vast majority of the variables are correlated.
Both number of liberal directors, Tobin’s Q and gender ratio are not correlated with CSiR, comprising the only two values that are not correlated.
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9)
(1) Corporate Social Irresponsibility
1.000
(2) Relative Donations (log)
0.1205* 1.000
(3) Volunteering 0.1156* 0.1163* 1.000 (4) Number of Liberal
Directors
0.0456 0.1793* 0.1796* 1.000
(5) Tobin’s Q -0.0438 0.0047 0.0216 0.000 1.000 (6) Number of
Employees (log)
0.4292* 0.1856* 0.2207* 0.189* -0.014 1.000
(7) Gender Ratio 0.0078 -0.147* -0.262* -0.246* -0.040 -0.370* 1.000 (8) Governance
Strengths
0.3175* 0.1844* 0.223* 0.163* -0.015 0.365* -0.198* 1.000
(9) Mean Prior Corporate Social Irresponsibility
0.6390* 0.0619* 0.1238* 0.122* -0.005 0.383* -0.037 0.246* 1.000
*** p<0.01, ** p<0.05, * p<0.1
Figure 3:Covariates and correlations amongst all variables
Hypotheses Testing
In Figure 4 on the next page, an overview of the results of the random effects regression
analyses is illustrated. In the first model, relative charitable donations are used as the
independent variable. The number of liberal directors is introduced as a moderator on the
relationship between CSiR and charitable donations. The following control variables are
consistent for both models: Tobin’s Q, Number of Employees, Gender Ratio, Governance Strengths,
Mean Prior CSiR and industry and year fixed effects.
Figure 4: Results from the random effects regression analyses for Model 1 (IV: donations), Model 2 (IV: volunteering) and Model 3 (control variables)