Political Ideologies in the Boardroom: The Effect of Boards of Directors’ Political Orientations and Opposing Values on Corporate Social Responsibility
Master Thesis
MSc BA: Change Management Faculty of Economics and Business
University of Groningen
Aileen Weiss s2929775
Supervisor: Dr. Björn Mitzinneck Co-assessor: Prof. Dr. Jordi Surroca
Date: 24.06.2019
Title
Political Ideologies in the Boardroom: The Effect of Boards of Directors’ Political Orientations and Opposing Values on Corporate Social Responsibility
Abstract
This paper examines the influence of a board’s political orientation and composition, within the liberalism-conservatism continuum, on organizational corporate social responsibility (CSR) performance. The research proposes an extension of upper echelon theory to boards of directors and the impact of their political ideology on organizational CSR performance. Specifically, it is hypothesized that (1) board liberalism will positively influence organizational CSR performances;
(2) variance between political ideologies within boards will positively moderate this relationship;
(3) a critical mass of three or more individuals of both, republicans and democrats, will positively moderate the relationship. To test the hypotheses, a sample of 629 boards of directors of the S&P 500 organizations between 2000– 2013 were examined. Yielding to 4933 observations, the political orientation of boards of directors on average as well as their composition resulted in mixed findings. Board liberalism positively influenced organizational CSR performances Resulting in the conclusion that upper echelon theory can be extended to boards of directors. Variance between political ideologies within boards had no significant effect. Contrary to expectations only a critical mass of republicans within boards of directors had a positive impact on organizational CSR performances. However, given the complexity the antecedences of organizational CSR performances hold, a holistic focus on not only within-firm factors but also external components needs to be taken.
Keywords: Upper echelon theory, Political ideology, Boards of Directors, Corporate Social
Responsibility, Critical Mass Effect
Introduction
The corporate social responsibility (CSR) performance of organizations carries high value for society, stakeholders and organizations itself and is therefore, demanded as well as scrutinized to a great extend (Gupta, Briscoe, & Hambrick, 2017). Literature indicates that at top management and CEO level, political ideologies impact the selection of CSR practices in firms (Gupta et al., 2017; Christensen, Dhaliwal, Boivie, & Graffin, 2015; Chin, Hambrick, & Trevino, 2013). These findings resonate with upper echelon theory which states that in order to understand an organization’s behavior, one has to consider the dispositions and values of top executives (Hambrick, 2007). Upper-echelon theory however, solely focuses on top executives in relative isolation and has given less consideration to composition and inherent dynamics of groups of top decision makers who impact decision making processes of organizational CSR practices (Harris &
Helfat, 2007). How varying and opposing political orientations, represented in boards of directors, influence an organization’s CSR performance thus remains an unanswered question.
In order to extend upper echelon theory to boards of directors, political ideologies need to be considered. Especially, given that literature claims ideologies to be an underlying mechanism and driver for individuals’ actions, decision-making processes and ultimately outcomes (Gupta, Nadkarni, & Mariam, 2018; Gupta & Wowak, 2017). Political orientations classify different values and attitudes associated with ideas of how society should look like and how to achieve this ideal state (Jost, 2006). Research thus, proposes a continuum to best explain the individual differences between political ideologies (Gupta, Nadkarni, & Mariam, 2018; Jost, 2006; Tetlock, 2000).
Within research on upper echelon theory, it has been asserted that liberal CEO’s are more
inclined to foster an organizations’ CSR performance than CEO’s with a conservative ideology
(Chin et al., 2013). Giving raise to expect that boards of directors with either a liberal orientation
or a conservative orientation differently impact, organizational CSR performances (Gupta et al., 2017).
As boards of directors are first and foremost considered to be groups, research on group dynamics needs to be contemplated by means of competing values, when extending upper echelon theory to boards of directors. Especially, given their influence on organizational outcomes (Christensen et al., 2015; Chin et al., 2013; Torchia, Calabro & Huse, 2011; Konrad, Kramer &
Erkut, 2008; Harrison & Klein, 2007; Miller, Burke & Glick, 1998). Specifically, it is necessary to investigate the impact of variance between the degree of endorsement of political ideologies on organizational outcomes (Chin et al., 2013). Furthermore, borrowing from research on gender diversity, the critical mass effect, which focuses on the impact on organizational performances and outcomes made by oppositions within boards of directors, needs to be considered (Torchia et al., 2011; Konrad et al., 2008; Kanter 1987; Granovetter, 1978; Kanter, 1977). The current research will thus, take into account the variance of political orientation along the political spectrum as a moderator. Additionally, going beyond simple variance research, the impact of opposing political orientations within boards of directors and their impact on organizational CSR performance will be implemented by introducing critical mass as a moderator.
To test the proposed considerations, quantitative research of S&P 500 organizations from
2000 until 2013 was conducted. A panel data regression analysis delivered supporting results on
the positive impact of board liberalism on organizational CSR performances. However, contrary to
previous research a critical mass effect of republicans rather than liberals, present in boards of
directors, had a significant impact on organizational CSR practices. These findings imply that a
within-firm focus can be beneficial when investigating organizational CSR performances and thus
contributing to the extension of upper echelon theory to boards of directors. The focus on whether
political ideologies are associated to specific CSR practices was counterintuitive and contradictory.
Giving raise to believe that CSR practices are in the best interest of board members in general, regardless their political orientation, based on the organizational benefits obtained through CSR practices. By this the current paper contributes not only to research on antecedents of organizational CSR performance but points into research directions that take into account a more holistic picture on organizational CSR performances.
Literature Review
CSR. The importance of corporate social responsibility (CSR) becomes evident when considering the description of its involvement in ‘social and environmental concerns to (firms’) business operations and in their interactions with stakeholders on a voluntary basis’ (Commission of the European Communities 2001, p.6). The relevance for stakeholders as well as society of CSR practices additionally, prevails from its link to the debate about the purpose of organizations itself (Margolis & Walsh, 2003; Friedman, 1970; Kaysen, 1957). By means of its definition as actions that generate greater social good and go beyond an organization’s interest (Gupta et al., 2017), the demand of stakeholders for organisations to endorse more CSR practices has been a factor that many large organisations are faced with in addition to the prevailing external factors such as governmental regulations and competition (Gupta et al., 2017; Kassinis and Vafeas, 2006). To be more specific, CSR practices include factors and actions that will for example increase equality of demographics as well as socioeconomic minorities (Chin et al., 2013). Additionally, research has linked organizational endorsement of CSR practices to firm performance and growth, innovation, financial performance (Bocquet, Les Bas, Mothe, & Poussing, 2015; Margolis & Walsh, 2003) as well as societal well-being (Margolis & Walsh, 2003). Thus, organizational endorsement of CSR practices holds beneficial and valuable outcomes for the organization as well as society.
Associated with top management, literature predicts a managerial over-emphasis on CSR
practices in order to secure own reputation and status among stakeholders (e.g., Barnea and Rubin,
2010; Cespa and Cestone, 2007; Hemingway and Maclagan, 2004). Simultaneously, a literature stream that recognizes endorsements of CSR practices as normative and ceremonial driven, the question about the underlying driving factors of CSR practices has been raised (Surroca, Tribó, &
Waddock, 2010; Surroca and Tribó, 2008) and research, on what factors influence an organisations’
CSR strategy and performance, emerged (Chin et al., 2013).
Boards of Directors. When considering factors that can influence organizational outcomes, the monitoring, evaluating and advising nature of boards of directors needs to be regarded (Veltrop, Molleman, Hooghiemstra, & van Ees, 2018). Boards of directors and their influence on organisational outcomes has become evident (Harris & Helfat, 2007). For instance, boards of directors are deeply involved in determining the directions of an organization and have to regularly approve and vote on major decisions. Meaning that boards of director’s choices strongly shape an organization’s course of action (Veltrop et al., 2018; Harris & Helfat, 2007). Thus, boards of director’s share governance of organizations with the CEO, especially by means of ensuring ethical, social and legal demands such as CSR practices and to attain an organization’s economic potential (Godos-Diez et al., 2018).
Upper Echelon Theory. Another factor that needs to be examined when investigating organizational outcomes are values and beliefs, past experiences as well as personalities of top managers and executives, as proposed by upper echelon theory (Hambrick & Mason, 1984).
According to upper-echelon theory, personality, values and beliefs have an impact on perceiving
and interpreting situations as well as decision making (Hambrick & Mason, 1984). Furthermore,
they not only contribute to the interpretation of situations but also exert influence on decisions
which ultimately impact organizational outcomes and performances (Hambrick, 2007). As
summarized by Finkelstein, Hambrick and Cannella (2009), CEO’s personalities get reflected in
organizational outcomes. However, recent research has moved away from a focus on CEO’s
personalities and instead investigated CEO’s values as well as ideologies instead (Chine et al., 2013). This approach, to a broader conceptualization of preferences held by top executives, focuses on ideologies which can be best conceptualized as an interconnection of values, forming a value system which is held by an individual as proposed by Joost et al., (2009). Specifically, executives’
values and attributes will be portrayed in organizational outcomes. Thus, in order to understand an organization’s performance, one has to consider the dispositions and values of top executives (Hambrick, 2007). This expounds that characteristics in terms of values and morals of leading figures of an organization influence an organization’s outcomes due to their own ideologies (Christensen et al., 2015; Miller et al., 1998). Additionally, it signifies that political ideologies held by top executives impact their decision-making processes.
Political Ideology. The impact of political orientation has been asserted in literature by means of considering political ideologies as a driving factor in decision making processes (Gupta et al., 2018). Research focusing on values and morals, has outlined that, individuals endorse different political ideologies based on their morals, values and ideologies (Graham, Haidt, &
Nosek, 2009). Thus, an individual’s morals and values become prevalent as political ideologies.
Which in turn, are proposed to be the driving factors to evaluate as well as interpret society around
us, based on the established underlying cognitive patterns. Thus, explaining people’s actions as
well as motivations (Gupta et al., 2018; Gupta and Wowak, 2017; Jost et al., 2009). The most
common conceptualization of ideologies is the liberalism-conservatism spectrum (Jost, 2006; Jost
et al, 2003). To be more specific, within this spectrum, also referred to as the right-left continuum
(Gabel & Huber, 2000; Castles & Mair, 1984), it has been proposed that liberals emphasize values
such as social justice, progressivism and control over markets. While conservatives value
traditionalism as well as individualism and free markets (Jost, 2006; Jost et al, 2003). When
defining political orientation along a conservativism-liberalism continuum as proposed by Jost et
al., (2009) it is additionally crucial to assert that compared to conservatives, liberals heavily emphasize on aspects such as fairness, reciprocity as well as harm and care (Graham et al., 2009).
Furthermore, research on political ideologies indicates liberals to believe in values of equality as well as an egalitarian system. By means of dismissing the conservative attitudes that people are unequal and are in need of differential systems with authority and bureaucracy (Gupta et al., 2018).
Political Ideology and CSR. Put into the context of organizational CSR performance, it becomes evident that a top executives’ political ideology impacts the organizational CSR performance (Chin et al, 2013). Research has shown that values as well as political orientations of CEO’s impact CSR initiatives such as egalitarian honorariums (Chin & Semadeni, 2017; Chin et al., 2013). More specifically, the distinction between liberals and conservatives along a continuum has previously shown that liberal CEO’s were more likely to favor organizational CSR practices than CEO’s with a more conservative ideology (Chin et al, 2013). Additionally, research outlines that the underlying morals such as fairness and equality as well as reciprocity of liberals will drive an organization to CSR practices such as increasing equality within the company as well as embracing sustainability by reducing behaviors that will harm the environment (Chin et al, 2013).
Furthermore, research exerts that CEOs who hold liberal ideologies are more likely to endorse
equal compensation practices (Chin & Semadeni, 2017). The proposed conceptualization of
ideologies as well as its impact on organizational CSR performance and organizational outcomes
has not only been tied to the CEO as well as top management but it has also been extended to an
organizational level. Stating that prevailing ideologies held by members of an organization,
significantly impact organizational outcomes and even exceed CEO values (Gupta et al., 2017). It
can thus, be expected that the same effect of political orientation on organizational CSR
performance can be found within the board level of an organization. Thus, by extending these
findings from top executives and CEO’ s as well as the organizational level to a focus on boards of directors, I hypothesize that:
H1: The more liberal the board of directors, the higher an organization’s CSR performance.
Variance of Political Ideology. When determining the impact of political orientations on organizational CSR performance, heterogeneity and variance cannot be ignored (Chin et al., 2013).
However, upper echelon theory in its essence, fails to account for variance along the liberalism-
conservatism spectrum within boards of directors. Thus, when extending upper echelon theory to
boards of directors, the factor of political ideology variance needs to be considered. Specifically,
looking at variance between political orientations it becomes evident that values on the liberalism-
conservativism continuum as proposed by Jost (2006) can be differentiated clearly. Based on
research by Chin et al., (2013) CEO political ideologies were found to be spread along the
conservativism-liberalism continuum. To be more specific, a third of the investigated CEO’s held
highly conservative political ideologies and about fifteen percent could be categorized into the
highly liberal spectrum along the continuum (Chin et al., 2013). However, the remaining CEO’s
were arranged along the more moderate zone of the liberalism- conservativism spectrum (Chin et
al., 2013). Meaning that individuals who are members of a board of directors differ not solely on
their direction towards the opposing continuum, they additionally differ along the intensity and
extremity towards which they endorse either orientation (Chin et al., 2013). In other words,
variance within for example one direction of political orientation can be present due to the
conceptualization of political ideologies as a continuum (Chin et al., 2013; Jost et al., 2009; Jost,
2006). It can thus, be said that individual’s ideologies can be endorsed to a slight, moderate or
fervent extend. Whether the extent of advocacy of political orientation has an impact on organizational CSR performance remains unclear. Research on cognitive diversity asserts a positive effect on organizational outcomes (Harvey 2013; Miller et al., 1998). Furthermore, literature proposes a positive influence on organizational outcomes when a presence of opposing ideologies within a boardroom is accounted for (Torchia et al., 2011; Konrad et al., 2008). Thus, novel to this literature stream, variance of individuals along the liberalism-conservatism spectrum, might have a positively moderating effect on the relationship between political orientation of boards of directors and an organizations CSR performance. Thus, it can be hypothesized that:
H2: The positive effect of board liberalism on organizational CSR performance is stronger when variance between political orientation is greater.
What remains unclear when merely focusing on variance between political orientations however, is how differences in political ideologies impact organizational outcomes. When moving away from variance research (e.g. Chin et al., 2013), towards more recent literature focusing on the underlying mechanisms that oppositions achieve in order to impact organizational outcomes, the impact of diverse political ideologies becomes evident (Torchia et al., 2011; Konrad et al., 2008).
Diverse Political Ideologies. Opposing political orientations within a board of directors can
be conceptualized as cognitive diversity by means of prevailing differences in believes, values,
preferences, ideologies as well as perspectives (Harvey 2013; Miller et al., 1998). Going beyond
the consideration of variance alone this suggests that, competing values in form of holding differing
political ideologies should be taken into account. Thus, when extending upper-echelon theory to
boards of directors with cognitive diversity in terms of opposing political orientations and their
impact on CSR practices, group dynamics need to be investigated further. It can be argued that when considering opposing political ideologies within boards of directors the overarching conservative political ideology held by corporate America cannot be neglected (Bonica, 2016).
Specifically, it is commonly said that corporate America is slightly skewed to the conservative side of the political spectrum based on the existing preference to make donations to presidential candidates from the republican party (Bonica, 2016). Hence, the more common baseline of political orientation among boards of directors is considered to be conservative. However, it has also been argued that even though organizations in America are conservative leaning, especially in industries such as gas, coal as well as oil, a heterogeneity of political orientation in boards of directors persists (Bonica, 2016). Additionally, recent research on opposing views within boards of directors proposes that opposite views and ideologies are able to act as a tipping point of how for example organizations have performed before (Chesterman, & Ross-Smith, 2006).
Critical Mass Effect. Taking inspiration from literature on gender diversity, a numerical
threshold of female directors on boards in order to obtain improved firm performance as well as
innovativeness will be examined (Torchia et al., 2011; Konrad et al., 2008). What is crucial about
research on the critical mass effect, is its occurrence with three or more women on boards of
directors, however, not with less than three (Torchia et al., 2011). Respectively, research shows
that when including a minority of at least three into boards of directors, they will be able to make
an impact (Torchia et al., 2011; Konrad et al., 2008). In other words, the critical mass effect refers
to the idea that by reaching a specific number of diversities, it is able to tip the balance, in order
for a pattern to emerge that goes beyond the expected (Chesterman, & Ross-Smith, 2006). Research
on the critical mass effect has asserted that the threshold which tips the balance of the prevailing
patterns by minorities, do have influence on organizational outcomes and organizational
performance (Torchia et al., 2011; Konrad et al., 2008). Put into the context of diversity in terms
of political ideologies and orientations, it can be argued that a certain threshold of minorities influences organizational outcomes. In line with the current research, it has been established that the underlying values of conservatives and liberals strongly deviate from another (Jost et al, 2009;
Jost, 2006). However, what still remains unclear is how boards of directors, in which members hold differing political ideologies impact organizational CSR performance. As gender as well as cognitive patterns, such as political ideologies, can be diverse within boards of directors, research focusing on boards should include the factor of diversity and contemplate an impact due to a threshold and tipping point triggered by diverse cognitive patterns (Torchia et al., 2011; Konrad et al., 2008; Kanter 1987; Granovetter, 1978; Kanter, 1977). Considering the impact of opposing political orientations within boards of directors on an organisations CSR performance thus, makes the critical mass effect interesting to use as a moderator of the relationship between political orientation of boards of directors and organizations CSR performance. I order to extent upper echelon theory to group diversity in terms of diverse political orientations and to take into account a threshold of opposing political views, the critical mass effect, I therefore, propose that:
H3a: A board with three or more individuals holding a liberal political orientation will increase the impact of a board’s political orientation on an organization’s CSR
performance.
External pressures. Previous literature on political ideologies held by either CEOs, TMTs
or organizations found positive effects of liberalism on organizational CSR performance (Gupta et
al., 2017; Christensen et al., 2015; Chin et al., 2013). While conservatism has not yet been as
positively linked to CSR performances as liberalism has, societal scrutiny on organizations as well
as governmental legislations and market competition, can have an impact on CSR performances
(Gupta et al., 2017; Christensen et al., 2015; Bocquet, et al., 2015; Chin et al., 2013; Margolis &
Walsh, 2003). As society has become more critical with organizations that do not portray the image or values of CSR performances, the demand for equality, sustainability and diversity has increased (Gupta et al., 2018; Gupta and Wowak, 2017; Bocquet, et al., 2015; Margolis & Walsh, 2003;
Meyer & Rowan, 1977). Based on pressure from stakeholders on organizations for more CSR practices, more conservative boards of directors will experience higher pressure to conform to the CSR standards as they will be more critically evaluated increased (Gupta et al., 2018; Gupta and Wowak, 2017; Meyer & Rowan, 1977).Thus, it can be argued that by means of these external pressures on decisions about CSR practices an over-emphasis on CSR practices to secure reputation among stakeholders of conservative leaning individuals can take place (e.g., Barnea and Rubin, 2010; Cespa and Cestone, 2007; Hemingway and Maclagan, 2004). Furthermore, it can be argued that due to the beneficial effects of CSR performance for not only society and the environment but also for organizations (Gupta et al., 2017; Bocquet, et al., 2015; Chin et al., 2013; Margolis &
Walsh, 2003; Meyer & Rowan, 1977), supporting CSR practices is in the interest of many individuals on boards of directors. Thus, when going beyond recent research which solely links liberalism to increased CSR performance (e. g. Gupta et al., 2018; Gupta and Wowak, 2017; Chin et al., 2013) a link between CSR performance to positive organizational outcomes makes supporting CSR practices lucrative for conservatives. Respectively, a critical mass of conservatives would have the same effect than the hypothesized critical mass effect of liberals on organizational CSP performance. It can thus be hypothesized that:
H3b: A board with three or more individuals holding a conservative political orientation
will increase the impact of a board’s political orientation on an organization’s CSR
performance.
Specific types of CSR and Political Ideology.
1When inspecting the effect of political ideologies regarding CSR support, a deeper understanding of which political ideologies associate with which specific sub-type of CSR practices are interesting to examine. Thus, the current paper will additionally take into consideration and analyze the most significant sub-categories of CSR practices to contribute to research on associations between specific CSR practices and political orientations.
Taken together, the hypotheses result in the following conceptual model:
Figure 1. Conceptual Model
1 Correlations with subcategories of the net KLD score, showed diversity and environmentalism to be the most significant ones (Table 1). Additional models were run with the former and latter sub categories
Method Sample and Data Structure.
The sample to test these hypotheses consists of companies listed within the Standard &
Poor’s 500 from 2000-2013 within the U. S. Thus, a focus on organizations with great economic and political influence on the U.S was given. Additionally, this sample provides extensive information about donation behavior of boards of directors to political candidates as well as organizational CSR performance. For this research multiple data sources have been used, in particular the database of Ideology, Money in Politics and Elections (DIME), the Kinder, Lyndenberg, Domini (KLD) measure as well as BoardEx and Compustat. After controlling for missing data and equal sample size, the final sample contained 629 organizations with a total of 4933 observations.
Measures.
All measures have been proposed by previous research and have been validated by literature.
CSR performance (DV). Actions that have been described as corporate social responsibility (CSR) are mainly focused on steps that generate greater social good and go beyond an organization’s interest. Additionally, some of these actions are demanded on a legal basis (Gupta et al, 2017). Thus, CSR can be defined as “actions that appear to further some social good, beyond the interest of the firm and that which is required by law” (McWilliams and Siegel, 2001, p. 117).
To measure organizational CSR performance, this research makes use of data presented by Kinder,
Lyndenberg, Domini (KLD). The KLD is an annual measure of organizations that investigates CSR
profiles. For the analysis the annual net score of an organizations CSR profile based on strengths
and concerns will be calculated for organization’s individually as well as the industry sector in line
with Gupta et al., (2017) and Chin et al., (2013). In order to generate a global KLD score, the
strengths and concerns of the five categories, namely corporate governance, diversity, employee relations, human rights and environmentalism, are focused on. Models will be run with the global KLD score as well as net score of the most significant sub categories (see Table 1).
Political orientation (IV). In line with previous research conducted by Bonica (2016) Christensen et al., (2015) and Chin et al. (2013), political ideologies of board members is measured based on donation behavior towards a specific political candidate and its political positioning. In line with Parsons (1951) political orientation can be conceptualized as: “ideologies are the shared framework of mental models that groups of individuals possess that provide both an interpretation of the environment and a prescription as to how that environment should be structured” (p. 24).
Ideologies can thus be used to best explain the “relational, epistemic, and existential needs or motives” (Napier & Jost, 2008) that ultimately drive an individuals’ behavior (Gupta et al., 2018).
Information on political ideologies were obtained through the DIME which offers information on
circa 100 million individuals as well as organizations in the time span of 1979 and 2012. DIME
records contributions to candidates made by individuals and boards of directors to either
republicans or democrats, political action committees (PAC’s) and political candidates. This data
is publicly accessible and provides personal information of donors such as names, addresses,
employers and occupations. Based on the amount as well as the number of contributions made, to
a political candidate and their standing on the democratic-republican continuum, the political
orientation of the donor was established and depicted in CF-scores (see Bonica, 2016). Ultimately,
resulting in a continuous spectrum of political ideologies between -2 and +2, with -2 indicating
contributions made to the liberal candidates, while +2 displays donations made to the conservative
candidates. For interpretive means, the mean CF Scores of boards were inverted by multiplying
them times -1. To test the hypotheses, a dataset has been constructed through a fuzzy matching
process linking individuals of S&P’s 500 organizations to the provided DIME dataset (see Bonica
2016; Christensen et al., 2015; Chin et al., 2013). This fuzzy matching process was facilitated by and algorithmic pre-cleaning, followed by manual identification of true positives and false positives
2. To minimize human error, the coding processes was done by two independent coders, first individually, followed by resolving disagreements among coders. DIME provides information on political orientation for each individual on boards of directors. The individual CF Scores of board members were summed and divided by the number of serving board members, to obtain the average political orientation for each board of directors within the S&P 500. Following validation by Christensen et al., (2015) as well as Chin et al., (2013), missing information of political ideologies of directors were replaced with 0 providing a neutral political orientation in the average calculation. Figure 2 shows the distribution of political orientation across boards.
2 The identification was further enabled by information available on Bloomberg’s database which is managed by S&P Global Market Intelligence.
Figure 2. Distribution of BOD’s political orientation
Moderator Variables: Variance & CME
3. To test hypotheses 2, 3a and 3b a two-way interaction effect is calculated. In order to create a moderator variable for hypothesis 2, the standard deviation of the CF scores was squared. Based on Torchia at al., (2011) and Konrad et al., (2008) a critical mass is present in boards if 3 or more individuals of a minority group is present. Thus, for hypothesis 3a and 3b a critical mass dummy variable, indicating boards of directors with three or more liberals or conservatives present was created.
3 These moderator variables were measured in t-1
0 .2 .4 .6 .8 1 D e n si ty
-2 -1 0 1 2
BOD CFscore (inverse)
Control Variables
4. Literature on political ideologies as well as CSR performance has asserted that ‘industry average’ is important to control for (Gupta et al., 2017; Chin et al., 2013).
Meaning that the industry in which an organization is set in as it can impact the amount of CSR endorsement, as pressure to adapt can emerge (see Gupta et al., 2017) and differences in political ideologies and organizational CSR performances across industries are present (Gupta et al., 2017;
Bonica, 2016). Thus, industry level effect of political ideology and CSR performance of organizations is controlled for. By using the mean scores of both variables and calculating their average based on the two-digit SIC peer coding system. Additionally, in line with Gupta et al., (2017) a common array of control variables such as performance of organizations, measured in terms of Tobin’s Q and Return on Assets (ROA), as well as firm size, measured in term of total assets and number of employees as a robustness check are controlled for. As it can be argued that large organizations find themselves under more pressure to conform to more CSR practices (Perrault & Quinn, 2018; Gupta et al., 2018; Gupta et al., 2017; Bonica, 2016; Meyer & Rowan, 1977). Board size will thus be controlled for in terms of number of board members on the boards, due to its impact on the critical mass effect (Torchia et al., 2011; Konrad et al., 2008) and gender diversity, a sub-category of the KLD for as suggested by Brammer, Millington & Pavelin (2007).
As proposed by Udayasankar (2007) visibility of industry sector was included in the models based on the reasoning that organizations with low visibility are confronted with less pressure to engage in CSR practices. To measure visibility, the business-to-business orientation, by industry was estimated, also known as an inverse variable of Lev, Petrovits, & Radhakrishnan (2010) customer orientation measure. After a correlation matrix has been established, individual KLD scores will also be controlled for, based on the subcategories that have the most significant effect on the global
4 These control variables were measured in t-1 except ‘Previous year CSR’, which was measured in t-2
net KLD score
5(Gupta et al., 2017). Previous year CSR scores were additionally taken into account (Gupta et al., 2017). Furthermore, the proportion of directors with backgrounds in environmental, social and public relations functions, drawn from BoardEx will be controlled for in order to ensure for confounding variables such as past experience and expertise (Capriotti & Moreno, 2007;
O'Neill, Saunders, & McCarthy, 1989).
Data Analysis. As the dataset holds longitudinal characteristics, the appropriate analysis in form of a panel data regression was conducted. Conceptually, fixed effects models are not in line with controlling for industry level CSR performance as well as industry sector political ideology.
Additionally, a fixed-effect model is not appropriate due to the time invariant variable of political ideology (StataCorp, 2009). Thus, random and mixed effects models are primarily used, following the suggestion made by Bollen and Brand (2010). Specifically, the xtreg command for GLMs in Stata 15.0 was used (Chin et al., 2013). In line with Chin et al., (2013) to establish temporal precedence of predictors, I am using a lag in all models
6. Additionally, the number of observations was controlled for by creating a minimal sample indicator to ensure comparability.
Results
Table 1 and 2 depict correlations of variables measured as well as descriptive statistics.
Various correlations exceed the threshold value of 0.3, thus the variance inflation factor (VIF) was calculated in order to check for collinearity. The VIF value of 1.43 based on a model which includes all measured variables indicates no substantial multicollinearity threats as all VIFs are below 4 (Hair, Black, Babin, & Anderson, 2010). Table 3 summarizes model 1-3b and describes results of random panel data regression models for the effect of political ideologies on organizational CSR
5 See footnote 1
6 See also footnote 3 and 4
performance, taking into account variance and opposing political orientations. Testing for hypothesis 1, model 1 depicts the random-effect baseline model. A significant effect is established (p<0.01). Thus, support for hypothesis 1 was found. All control variables were significant in the random-effects model, except business-to-business orientation, proportion of directors with backgrounds in environmental functions, board size and the industry level of political ideologies.
A robustness check additionally confirmed the results. Moreover, random-effects models, testing the sub-score of environmentalism and diversity of the KLD measure were both significant (p<0.01). For environmentalism all control variables were significant except for industry level of political ideologies and proportion of directors with backgrounds in social functions. Whereas, for diversity all control variables were significant except for business-to-business orientation, proportion of directors with backgrounds in environmental functions, firm performance (Tobin’s Q) and the industry level of political ideologies.
Testing for hypothesis 2, model 2 depicts the interaction effect of variance of political
ideology and board liberalism on organizational CSR performance. The random-effects model
found no significant effect of the interaction. All control variables were significant in the random-
effects model, except business-to-business orientation, proportion of directors with backgrounds
in environmental and social functions, and firm performance (Tobin’s Q). The robustness check
in form of a mixed-effects model support these findings. Hence, support for hypothesis 2 was not
found. When running a random-effect model for the sub-scores of environmentalism and diversity
of the KLD measure, no significant effects were found. Control variables were significant for
environmentalism except for industry level of political ideologies and proportion of directors with
backgrounds in social functions. Whereas, for diversity all control variables were significant except
for business-to-business orientation, proportion of directors with backgrounds in environmental
functions, firm performance (Tobin’s Q) and the industry level of political ideologies.
Testing for hypothesis 3a, model 3a depicts the results of the interaction effect of a critical mass of liberals and a boards political ideology on organizational CSR performance. The interaction effect of a board’s political ideology and a critical mass of liberals however, was insignificant. Thus, hypothesis 3a was not supported. All control variables were significant, except business-to-business orientation, proportion of directors with backgrounds in environmental and social functions, industry level of political ideologies as well as firm performance (Tobin’s Q) and board size. The insignificance was additionally confirmed by a mixed effect model robustness check. When running a random-effect model for the sub-scores of environmentalism and diversity of the KLD measure, no significant effects were found. However, control variables were significant for environmentalism except for industry level of political ideologies and proportion of directors with backgrounds in social functions. Whereas, for diversity all control variables were significant except for business-to-business orientation, proportion of directors with backgrounds in environmental functions, firm performance (Tobin’s Q) and the industry level of political ideologies.
Testing for hypothesis 3b, model 3b depicts the results of the interaction effect of a critical
mass of conservatives and a boards political ideology on organizational CSR performance. The
random-effects model, including control variables found a significant effect (p<0.01). Thus,
hypothesis 3b was supported. All control variables were significant, except business-to-business
orientation, proportion of directors with backgrounds in environmental and social functions,
industry level of political ideologies as well as firm performance (Tobin’s Q) and board size. The
significance was additionally confirmed by a mixed effect model robustness check. When
examining the sub-scores environmentalism and diversity of the KLD measure, a significant effect
was found for both (p<0.05; p<0.01). Again, control variables were significant for
environmentalism except for industry level of political ideologies and proportion of directors with
backgrounds in social functions. Whereas, for diversity all control variables were significant except for business-to-business orientation, proportion of directors with backgrounds in environmental functions, firm performance (Tobin’s Q) and the industry level of political ideologies.
Generally, post hoc robustness tests were conducted. Mixed-effect models were run as
random-effects models to examine whether results remain constant and vice versa. The robustness
tests were conducted by substituting control variables by means of using number of employees as
a firm size measure instead of making use of total assets. Similarly, Tobin’s Q was substituted by
ROA, as previous studies have shown to deliver differing results depending on using market based
or accounting based performance measurement (Adams & Ferreira, 2009). Lastly, the models were
run with the most significant subcategories of the net KLD score, by means of diversity and
environmentalism, to examine the consistency of results. For all hypotheses the robustness check
found consistent results with the global KLD measure.
Variable Model 1 Model 2 Model 3a Model 3b
Industry BOD CF Score -0.728* -0.725* -0.824* -0.833**
(0.327) (0.326) (0.326) (0.322)
Industry KLD Net Score 0.388*** 0.387*** 0.401*** 0.380***
(0.043) (0.043) -0.043 -0.043
Total Assets 0.000** 0.000** 0.000* 0.000**
(0.000) (0.000) (0.000) (0.000)
Tobin's Q -0.005 -0.005 -0.007 -0.000
(0.030) (0.030) (0.029) (0.028)
B2B 0.02 0.02 0.02 0.02
(0.090) (0.090) (0.090) (0.090)
Prior Year KLD Netscore 0.663*** 0.663*** 0.659*** 0.665***
(0.015) (0.015) (0.015) (0.015)
Boardsize 0.049** 0.049** 0.028 0.028
(0.016) (0.016) (0.016) (0.016)
Backround in Environmental
Functions 1.192 1.187 1.085 1.005
(0.685) (0.685) (0.683) (0.679)
Backround in Social Functions -0.406 -0.399 -0.462 -0.421
(0.479) (0.479) (0.479) (0.474)
Backround in Public Relations
Functions 2.469*** 2.469*** 2.445*** 2.368***
(0.333) (0.333) (0.332) (0.330)
BOD CF Score -0.429* 0.183 0.286
(0.154) (0.167) (0.188)
Variance 0.038
(0.125)
CM Dem 0.480***
(0.095)
CM Rep 0.428***
(0.098)
BOD CF Score (H1) 0.505***
(0.107)
BOD CF Score × Variance (H2) 0.216
(0.323)
BOD CF Score × CM Dem (H3a) -0.167
(0.240)
BOD CF Score × CM Rep (H3b) 0.731***
(0.222)
Constant -0.643** -0.669** -0.700*** -0.640**
(0.204) (0.216) (0.212) (0.204)
(***) p < 0.001, (**) p < 0.01, (*) p < 0.05