Master Thesis
The effect of TMT diversity on firm’s CSR performance in a country-level
sustainability context: Empirical evidence from Europe
Advanced International Business Management and Marketing
Dual Award (17/18)
Supervisors
Dr. Rudi W. de Vries (University of Groningen)
Dr. Iain Munro (Newcastle University)
Pit Leo Rissiek
S2652536
B7080746
1
Abstract
In order to enrich and support previous research on the acceptance of minorities on corporate boards, this thesis aims to illuminate the complexity of top management teams’ (TMT) relationship with corporate performance measures. Following recent trends, this study examines whether and how TMT diversity relates to firms’ industry-specific sustainability performances. Specifically, the effects of three TMT diversity dimensions, namely, gender, age and nationality diversity, are regressed on respective CSR performance scores (ESG data). Moreover, reacting on predictions made based on the upper echelons theory, this research is among the first to inspect the effects of a country-specific sustainability environment on the relationship between TMT diversity and CSR performance.
Data of 117 finance industry related firms from 12 EMU countries is analysed performing multiple regressions and moderation tests. Interestingly, the quantitative analysis provided no statistical evidence indicating a relationship between gender diversity and CSR performance. Contrary to hypothesised effects, TMT age diversity is negatively related to CSR performance, hinting towards intra-TMT generation conflicts. However, confirming diversity synergy forces, the regression analysis revealed that TMT nationality diversity is positively associated with CSR performance. Adding to the upper echelons theory, the results further provide the imperative suggestion that the association between a firm’s TMT diversity and CSR performance is not significantly influenced by the company’s country specific sustainability environment.
Besides discussed managerial implications, this study substantially contributes to equality efforts that demand the appreciation of diverse TMTs and confirms the multilevel intricacy of diversity and performance relationships. This dissertation’s outcomes further academically empower researcher and enterprises to achieve their goal to understand or pursue high CSR scores in order to entice investments. Therewith, substantial contributions and future inspirations are made on a societal as well as firm level.
Acknowledgements
2
List of Contents
1. Introduction ... 3
2. Literature review ... 8
2.1. Firm CSR performance ... 8
2.2. TMT member characteristics and the upper echelons theory ... 9
2.3. TMT diversity and firm CSR performance ... 11
2.4. The effect of a board’s environment: the country’s sustainability environment ... 16
3. Identified research questions ... 18
3.1. How is TMT diversity related to firm’s industry-specific sustainability performance? .... 18
3.2. How is a firm’s home country’s sustainability environment influencing the relationship between TMT diversity and CSR performance? ... 19
4. Hypotheses development ... 20
4.1. TMT diversity and CSR performance ... 20
4.2. Country sustainability environment moderation ... 21
4.3. Research model ... 22
5. Research Methodology ... 23
5.1. Dependent variable: CSR performance... 23
5.2. Independent variables ... 24
5.3. Moderation effect of country sustainability environment ... 25
5.4. Sampling method and sample size ... 26
5.5. Statistical tests ... 27
5.6. Control variables ... 28
6. Analysis ... 30
7. Results ... 35
7.1. Multiple linear regression results ... 35
7.2. Hayes’ moderation test results ... 38
8. Discussion ... 42
8.1. TMT diversity and CSR performance ... 42
8.2. Country sustainability environment moderation ... 46
8.3. Evaluated effects of control variables on given findings ... 51
9. Limitations and conclusions ... 52
9.1. Limitations ... 53
9.2. Conclusions ... 55
References ... 62
3
List of Figures
Figure 1: Research outline ... 7
Figure 2: Research model clarification ... 22
Figure 3: Research level clarification ... 23
Figure 4: Normal P-P Plot of Regression Standardised Residuals – ESG data ... 31
Figure 5: Histogram of ESG performance with perfect normal distribution curve ... 32
Figure 6: Scatterplot of standardised ESG data to test homoscedasticity ... 32
List of Tables
Table 1: Main implications of the research methodology ... 29Table 2: Descriptive Statistics of regression sample ... 30
Table 3: Coefficients and Variance Inflation Factors ... 32
Table 4: Durbin Watson test – Model Summary ... 33
Table 5: Multiple regression Model Summary... 35
Table 6: Anova test – Multiple regression analysis outcome ... 36
Table 7: Coefficients of independent variables ... 37
Table 8: Anova test – Multiple regression analysis with control variables ... 37
Table 9: Coefficients of independent variables and control variables ... 38
Table 10: Model summary – Country sustainability environment & Gender diversity ... 38
Table 11: Interaction Model – Country sustainability environment & Gender diversity ... 39
Table 12: Model summary – Country sustainability environment & Age diversity ... 39
Table 13: Interaction Model – Country sustainability environment & Age diversity ... 40
Table 14: Model summary – Country sustainability environment & Nationality diversity ... 40
Table 15: Interaction Model – Country sustainability environment & Nationality diversity ... 41
4
List of Abbreviations and Acronyms
CSR Corporate Social Responsibility
ESG Environmental, Social and Governance
TMT Top Management Team
EMU European Monetary Union
OLS Ordinary Least Square
WRDS Wharton Research Data Services
ROA Return on Assets
5
1. Introduction
While boards progressively comprise more diverse members, hindrances to accept these minorities on boards remain to exist (Heidenreich, 2012). In response, a range of researchers pursued to describe diverse board’s contributions towards firm performances (Miles et al., 2014). Following main corporate strategy trends, scholars additionally examined Corporate Social Responsibility (CSR) performance outcomes (Melo, 2012).
The necessity to investigate associations between Top Management Team (TMT) and CSR performance is threefold. Whereas enterprises pursue high CSR scores in order to entice investments (Webb, 2004), exploring TMT’s relationship with CSR performance is of high strategic importance. The latter deals with the possible strategic aim of encouraging critical or alternative thinking among board members. Assuming that CSR is an outcome of boards’ decisions, this dissertation additionally proposes that examining the link between TMT diversity and CSR would provide more scientific evidence to accept minorities in TMTs.
6 TMT diversity and CSR performance as well as sustainability related environmental influences, this thesis establishes the following two research questions.
How is TMT diversity related to a firm’s industry-specific sustainability performance and how is the firm’s home country’s sustainability environment influencing this relationship?
While the first research question examines the relationship between TMT diversity and CSR performance in under-researched geographical (Europe) and industrial (Finance Industry) areas, the latter proposes a moderating effect of a board’s home country’s sustainability environment.
7 The following illustration (Figure 1) visualises the logical outline of this research.
1. Introduction
2. Literature Review 3. Research Questions
4. Hypothesis Development 5. Research Methodology
6 - 8. Analysis, Results and Discussion
9. Limitations and Conclusions
8
2. Literature Review
The following literature review firstly touches upon previous research considering firm’s CSR performance. Subsequently, research on TMT diversities is reviewed. Afterwards, findings on relationships between TMT diversities and CSR are examined and finally, the effect of a board’s environment is analysed.
2.1. Firm CSR performance
9
2.2. TMT member characteristics and the upper echelons theory
The vast majority of studies that evaluate board characteristics and diversities focus on their impact on corporate financial performance (Jeong and Harrison, 2017). The main findings indicate a positive relationship of diverse board member’s characteristics on financial firm performances (Thomas et al., 2012).
Additionally, scholars examined decision making behaviour on an individual level. Frequently, researchers base their theoretical models of TMT member’s individual effects on decision making on the upper echelons theory by Hambrick and Mason (1984). Generally, the upper echelons theory states that organizational outcomes can partially be predicted by managerial background characteristics of the top-level management team (Hambrick and Mason, 1984). With regards to TMT members, the upper echelons theory does not dictate specific organisational positions. Nonetheless, scholars generally identify TMT members by their responsibility to make strategic decisions, to set the enterprise’s mission and vision and to establish policy blueprints (Hiebel, 2014; Kock, 2012). Specifically, TMTs commonly consist of top-ranking executives such as the CEO, chairman, managing director, president, executive directors, vice-presidents and others (Talke and Kock, 2011).
10 evaluation processes, which in turn influence their choices. Rather than applying a pure rational model for strategic decisions, the theory assumes that strategic choices result from both, the TMT member’s characteristics and the objective situation (Hambrick and Mason, 1984). Adding to the initial upper echelons model which proposed only direct relationships between TMT characteristics and organizational outcomes, the researchers further developed and updated the theory. Reacting on criticism (i.e. Lawrence, 1997), upper echelons theory-based studies increasingly focus on mechanisms or processes through which TMT characteristics lead to firm strategy and behaviour (Yamak, 2014). Consequently, updated models incorporate TMT processes as crucial parts of the causal chain explaining TMT members’ contributions to firm level outcomes (Finkelstein et al., 2009).
Due to difficulties of capturing directors’ cognitive frames, research on TMTs predominantly utilise observable characteristics such as nationality or gender as proxies for cognitive frames. In other words, the upper echelons research domain applies the above mentioned TMT member characteristics to examine strategic decisions and their processes (Marquis and Lee, 2013). Studies upon the concept have further shown that the TMT’s cognitive basis “[...] denotes a broad range of mental activity, including proprioception, perception, sense making, categorization, inference, value judgments, emotions, and feelings, which all build on each other” (Nooteboom et al., 2007, p. 1017). Hence, boards with more diverse (cognitive) characteristics allow the TMT members to stimulate each other and subsequently, create novelty.
11 decisions and the corporate’s CSR performance outcome. Additionally, the idea of character-related internal TMT differences is utilised for TMT diversity’s association with CSR performance. The theory further proposes that the individual’s characteristics are influenced by their respective environment (Hambrick, 2007). In specific, TMTs’ external environments are recognised as important moderators in the association between TMT composition and organizational outcomes (Finkelstein et al., 2009; Carpenter et al., 2004). Hence, scholars such as Yama et al. (2014) highlight the existence of upper echelons theory related moderation models in which the external environment moderates the relationships between TMT composition, TMT processes and organizational outcomes. This environment is later related to country specific sustainability environments that potentially influence individuals and groups (also see section 2.4.). Henceforth, literature of TMT’s influence on CSR performance is examined.
2.3. TMT diversity and firm CSR performance
Examining TMTs increasingly incorporates placing considerable attention to diversity (Hoffmann and Meusburger, 2018). TMT diversity is most commonly defined as the distribution of personal attributes among the members of the TMT (Nielsen, 2010; Jackson et al., 2003). Overall, the TMT diversity and CSR performance research domain is characterised by conflicting findings.
12 Nonetheless, most research indicated a positive association between TMT diversity and CSR reporting or performance. Among others, scholars utilised diversity factors such as gender, age or nationality in their CSR performance studies (Hagendorff and Keasey, 2012; Post et al., 2011; Bernardi and Threadgill, 2010). Generally, the positive association is supported by stressing diversity related network or synergy effects (Ibrahim, 2016) and improved decision-making processes (Zhang, 2012).
Similarly, the agency theory is frequently utilised to explain the relationship (Francoeur et al., 2008). The agency theory claims that the shareholders' and managers' interests and hence behaviours are likely to differ from each other (Jensen and Meckling, 1997). According to research from for instance Francoeur et al. (2008) an agency-theoretic rationale involves women (or external stakeholders, ethnic minorities and foreigners) often bringing a fresh perspective on complex issues. The researchers additionally claim that this can help correct informational biases in strategy formulation and problem solving (Westphal and Milton, 2000; Dewatripont et al., 1999). Other research supports these claims and adds that board diversity potentially improves monitoring management by increasing board independence (Carter et al., 2003). These effects lower the agency problem and therefore, lead to the aforementioned expected positive association between diversity factors and CSR performance.
13
Gender diversity and CSR performance
A variety of studies found a positive association between gender diversity and CSR performance (Bernardi and Threadgill, 2010; Krüger, 2009; Francoeur et al., 2008; Webb, 2004; Siciliano, 1996).
Research indicates that gender diversity among TMT members is important for effective corporate governance (Bear et al., 2010; Adams and Ferreira, 2009). Elizabeth Webb (2004) analysed board structures in socially responsible firms. Subsequently, the American business researcher found evidence signifying that socially responsible enterprises employ more outsiders and women directors than non-socially responsible firms. Additionally, Webb acknowledges that socially responsible firms score higher on the Governance Index (DSI) than the matched sample. Here, the Governance Index scores refer to rated and published firm specific scores that are published by external and firm-independent parties and comparable to ESG scores. Generally, higher scores are associated with better corporate governance structures (Webb, 2004). These findings are supported by for instance researchers such as Wang and Coffey (1992) who suggest that women and minority directors place more importance on social issues and the welfare of diverse stakeholders due to their social and economic background. In line with these findings, the implementation of CSR-related activities is supported by female board members (Bear et al., 2010). Specifically, a higher proportion of female TMT members increases the engagement of charitable initiatives. Studies indicating a positive influence of gender diversity on corporate giving confirm aforementioned claims (Marquis and Lee, 2013).
14 article analysing women directors’ contribution to board decision-making and strategic involvement, the scholars Nielsen and Huse (2010) highlight that female TMT members can produce various alternatives to organizational strategies and procedures due to their different points of view.
Age diversity and CSR performance
Besides gender diversity’s association with CSR, limited research indicates that age diversity among board members tends to influence CSR (Hafsi and Turgut, 2013). Studies documenting age diversity’s relationship with CSR performance are characterised by conflicting findings.
Some scholars produced evidence indicating that age diversity is negatively associated with CSR performance and methodologically underline the findings with analyses based on aforementioned Social Psychology theories (Liu and Zeng, 2017). Correspondingly, Liu and Zeng (2017) found that more diversely aged TMTs potentially lead to lower CSR performances due to inter-personal tensions. Additionally, established members might shield information and knowledge from young entrants due to trust issues or the intrinsic fear of losing power. These tensions and information withholdings weaken synergy potentials and consequently, diminish CSR performances. Other researchers failed to produce significant findings (Post et al., 2011).
15 greater value on society’s welfare while younger managers tend to recognise and strategically incorporate environmental and ethical issues (Diamantopoulos et al., 2003). Consequently, these diversity related synergies lead to the creation of effective and holistic CSR policies and eventually, a better CSR performance (Hafsi and Turgut, 2013).
Nationality diversity and CSR performance
Likewise, researchers evaluated the relationship between board’s nationality diversity and CSR performance. Whereas some scholars found support for a positive relationship (Ibrahim, 2016), papers from for instance Hafsi and Turgut (2013) fail to produce significant results.
16 These suggestions are supported by research from for instance Hillman et al. (2002). Accordingly, racial minority directors tend to come from diverse backgrounds and hold advanced degrees which enhances the above-mentioned decision-making effectiveness. The necessity for racial minorities to possess a more advanced degree can be traced back to a demanded better performance than their counterparts (Kelly et al., 2015). Similarly, research indicates that racial minorities value social issues and the welfare of diverse stakeholders to a larger extent due to their social and economic background (Wang and Coffey, 1992).
2.4. The effect of a board’s environment: the country’s sustainability environment
In line with the upper echelons theory, scholars highlight the importance to consider the country level context in which the organisation is embedded (Crossland and Hambrick, 2007). Previously, academics found a moderating effect of institutional characteristics on the relationship between board diversity and financial performance (Francoeur et al., 2008). Here, institutional characteristics referred to institutional complexities of regulations and national policies. Correspondingly, researchers disclosed that the relationship between female directors and corporate performance can be moderated by external (institutional) influences (Low et al., 2015). This positive moderation effect is ratified by the country's attitude towards women at work.
17 (Wiersema and Bird, 1993). Following the upper echelons concept, authors further found evidence of moderating effects of national social cultures on the association between TMTs’ strategic leadership behaviours and innovation outcomes (Elenkov et al., 2005).
18
3. Identified Research Questions
In the following, two research questions are raised. While the first research question deals with identified research conflicts in the TMT- CSR performance domain, the latter addresses the potential moderation effect of a country’s sustainability environment.
3.1. How is TMT diversity related to firm’s industry-specific sustainability performance?
In line with previous findings evaluated above, research on TMT’s relationship with CSR performance is characterised by quantitatively scarce as well as conflicting findings (Rao and Tilt, 2016; Krüger, 2009). Moreover, most studies utilise American company data (Melo, 2012; Post et al., 2011; Webb, 2004). In order to identify differences, this research applies data originating from the under-researched European area (also see section 5.3.). In 1984, Hambrick and Mason noted that TMT member characteristics and their effects are likely to vary with different industry contexts. Hence, to enhance internal validity a data set of a single industry, specifically of the finance industry, is utilised. The finance industry sector and its challenging public relations environment (Bourne, 2016), accurately reflects the need for a deep understanding of CSR.
Research Question 1
Sub-research questions
How is TMT diversity related to firm’s finance industry-specific CSR performance?
How is a TMT's gender diversity related to a firm’s finance industry-specific CSR
performance?
How is a TMT's age diversity related to a firm’s finance industry-specific CSR performance?
How is a TMT's nationality diversity related to a finance firm’s industry-specific CSR
19
3.2. How is a firm’s home country’s sustainability environment influencing the relationship between TMT diversity and CSR performance?
According to findings from for instance Low et al. (2015), the correlation between board characteristics and firm performance can be moderated by the board’s environment. Nonetheless, in prior CSR performance studies, this effect has been commonly disregarded. The findings suggest that previous research potentially misinterpreted the strength of TMT diversity’s relationship with CSR. In specific, it is argued that the potential effect of the country’s quality of sustainability environment has been underestimated.
Hereafter, I contend that a country’s sustainability environment moderates the relationship between TMT diversity and the firm’s CSR performance. A promotional environment, psychological forces and stakeholder pressures potentially enable more diverse boards to transform strategic CSR policies into practical strategies (Castelló et al., 2009). Since the upper echelons theory proposes that individual experiences, values and personalities can be driven by the TMT member’s country of origin (Hambrick, 2007), contrasting countries is deemed to be justified. Therefore, the following research question deals with country specific sustainability environments and their relationship with TMT diversity and CSR performance.
Research Question 2
Sub-research question
How is a firm’s home country’s sustainability environment influencing the relationship between TMT diversity and CSR performance?
Is the previously proposed relationship
20
4. Hypotheses Development
The following hypotheses are developed considering the research questions as well as previous findings signifying the type of relationship.
4.1. TMT diversity and CSR performance
The first hypothesis builds upon the research question targeting TMT diversity’s relationship with a firm’s industry-specific CSR performance. Previous research indicated that generally, TMT diversity and CSR performance are positively correlated (Bernardi and Threadgill, 2010).
Specifically, greater TMT diversity encourages closer monitoring of management decision-making pertaining to CSR to guarantee that diverse stakeholder interests are acknowledged in corporate strategy decisions (Kang et al., 2007; Luoma and Goodstein, 1999). These claims are in line with the above evaluated agency theory (Francoeur et al., 2008) and diversity related synergies (Ibrahim, 2016).
Henceforth, the previously recognised three TMT diversity dimensions’ impacts on the firm’s industry-specific CSR performance are hypothesised.
Hypothesis 1
Hypothesis 1a Hypothesis 1b Hypothesis 1c
TMT diversity positively correlates with the finance industry-specific CSR performance
TMT gender diversity positively correlates with the finance industry-specific CSR performance. TMT age diversity positively correlates with the finance industry-specific CSR performance.
TMT nationality diversity positively correlates with the finance industry-specific CSR
21
4.2. Country sustainability environment moderation
Moreover, the literature analysis indicated a potential moderation effect. Subsequently, it is hypothesised that a country’s sustainability environment moderates the relationship between TMT diversity and a firm’s CSR performance (Low et al., 2015). Accordingly, research points towards strengthening effects of a country’s sustainability environment due to social acceptance as well as CSR supportive stakeholder pressures that encourage the voluntary act of CSR. Thus, the following hypotheses are created.
Hypothesis 2
Hypothesis 2a
Hypothesis 2b
Hypothesis 2c
The relationship between board diversity and the industry-specific CSR performance is stronger in a high-ranked sustainability environment than in a low-ranked country sustainability environment.
The relationship between TMT gender diversity and the finance industry-specific CSR
performance is stronger in a high-ranked
sustainability environment than in a low-ranked country sustainability environment.
The relationship between TMT age diversity and the finance industry-specific CSR performance is stronger in a high-ranked
sustainability environment than in a low-ranked country sustainability environment.
The relationship between TMT nationality
diversity and the finance industry-specific CSR
22
4.3. Research model
The following research model (Figure 2) summarises the two developed main hypotheses.
23
5. Research Methodology
The above raised explanatory research questions are addressed applying quantitative research methods while following an deductive approach (Saunders and Lewis, 2012). This study is based on a positivism methodology which incorporates hypothesis testing in order to obtain an “objective” truth (Rossman and Rollis, 1998). The figure below (Figure 3) illustrates the methodological level of research applied in this study.
Figure 3: Research level clarification. Adopted from: (Saunders and Lewis, 2012; Niglas, 1999)
5.1. Dependent variable: CSR performance
Due to an overemphasis on samples stemming from global Fortune 500, S&P 500 or Global 100 databases (Setó‐Pamies, 2013; Bear et al., 2010), industry powers such as varying board objectives including different prioritisations of CSR policies, have been widely disregarded. Moreover, research on TMT diversity drivers is highly US centred which is supported by the overproportioned utilization of databases solely containing American firms (e.g. KLD database) (Melo, 2012; Post et al., 2011; Webb, 2004).
Therefore, the firm related variable ‘CSR performance’ is addressed utilizing ESG data from Arabesque’s S-Ray scores as a proxy (Appendix 1). Arabesque ranks firm’s environmental, social and governance performances based on normative principles of
Positivism • Epistemological Level
Quantitative
Paradigm • Methodological Level Quantitative
Methods • Technical Level Quantitative
24 the United Nations Global Compact such as Human Rights, Labour Rights, the Environment, and Anti-Corruption (Arabesque, 2017). Arabesque S-Ray systematically investigates big data of over 200 ESG metrics and scores firms accordingly on a scale from 0 – 100. Correspondingly, higher Arabesque S-Ray scores reflect better CSR performances while lower scores are associated with poorer CSR performances. The data allows filtering for industries and countries which is essential for the proposed relationships due to the identified under-researched finance industry and European geographical area. The ranking is accessed through Bloomberg Terminal (Newcastle University) and ESG data regarding firms of the European Economic and Monetary Union (EMU) is extracted. Correspondingly, external economic and monetary influences are minimised.
5.2. Independent variables
In the following the three identified diversity dimensions are methodologically addressed. In specific, measurements or scales are developed and ratified.
Gender TMT diversity
25
Age TMT diversity
Following other scholars (i.e. McIntyre et al., 2007), this study addresses age TMT diversity by calculating the standard deviation. The analytical method allows to assess the spread of directors’ ages and therefore, the standard deviation measure is judged to be appropriate with respects to the age diversity research question.
Nationality TMT diversity
In accordance with aforementioned research (i.e. Ibrahim, 2016), the proportion of directors from different countries is computed to examine nationality diversity. Specifically, the TMT’s number of foreign nationals is divided by the TMT’s total number of board members. The resulting percentage is utilised to address the nationality diversity research question.
5.3. Moderation effect of country sustainability environment
Next, the above theorised moderating force of a country’s sustainability environment is considered. In specific, related research methods with regards to RobecoSAM’s Country Sustainability Ranking are identified.
26 environments. Essentially, RobecoSAM’s Country Sustainability Ranking is utilised to rank EMU countries with regards to their ESG environment’s quality (Appendix 3).
5.4. Sampling method and sample size
This study follows a non-probability sampling approach (Greenfield, 2016). The convenience sampling is based on a non-random selection within the finance industry. The sampling technique is chosen due to limited Arabesque’s S-Ray data availability and restricted access to corresponding board data. All data incorporates the calendar year 2017.
Based on a screening of ESG firm data availability by relevant industry sectors (Arabesque, 2018), the finance sector is characterised by the highest percentage of data available with respect to the EMU countries chosen.
27 Board member data for the variables gender, age, and nationality TMT diversity is retrieved from the Wharton Research Data Services (WRDS) (University of Groningen) and annual reports of the year 2017. The firm CSR performance ESG scores are extracted from the Arabesque dataset, accessed through Bloomberg Terminal (Newcastle University). In line with the chosen time frame, ESG data of 12 months is extracted. Due to relatively low deviations within the chosen period, ESG performance averages are computed. Due to no available ESG data for Slovenia and the Slovak Republic (Appendix 4), these countries are excluded from the analysis.
5.5. Statistical tests
The following paragraphs elaborate on statistical methods aiming at examining the relationships between the defined diversity dimensions and firm CSR performance. Furthermore, tests related to the hypothesised moderation effect of a country’s sustainability environment are ratified.
TMT diversity and CSR performance
28 examining board diversities’ relationship with firm performance (i.e. Hafsi and Turgut, 2013).
Country Sustainability Environment’s moderation effect
The moderation related Hypotheses H2a-c are statistically assessed utilising SPSS and the PROCESS plug-in by Andrew F. Hayes. The above explained moderator variable (sustainability environment) has ordinal characteristics since the order of the values is known but their respective differences are undefined. Henceforth, statistical significance is calculated based on assessing the interaction term (Saunders and Lewis, 2012). The analysis allows interpreting differences between the moderator and its relationship with the independent and dependent variable. The applied test is further in line with related research from for instance Low et al. (2015). Hence, it is sub-concluded that the moderation test accurately addresses the stated hypotheses (H2a-c).
5.6. Control variables
The following control variables are based on the developed research questions and are backed with evidence stemming from previous research.
29 On a board control level, TMT size is accounted for. The number of directors on each firm’s board is accessed through the BoardEx database (accessed through University of Groningen). Previous research produced statistically significant support that indicates that board size is positively associated with CSR performance (i.e. Zhuang et al., 2018). Hereafter, TMT size is accounted for by incorporating the number of directors into the analysis.
The following table (Table 1) summarises the main implications of the research methodology.
Research Methodology
Variable type Measure Data source Type of
data Statistical test CSR Performance Dependent variable Industry-specific ESG score (0-100) Arabesque S-Ray Interval ---
Gender Diversity Independent
variable Proportion of male directors on boards (%) Wharton Research Data Services Interval Regression
Age Diversity Independent
variable Standard Deviation of age on boards Wharton Research Data Services
Interval Regression
Nationality Diversity
Independent
variable Proportion of directors from different countries (%) on boards Wharton Research Data Services Interval Regression Country Sustainability Environment
Moderator EMU Country
Sustainability Ranking RobecoSAM Ordinal Hayes’ moderation test
Financial Performance
Control variable
Return on Assets (ROA) Compustat Global
Interval Regression
TMT Size Control
variable
Number of directors on each firm’s board
BoardEx Interval Regression
30
6. Analysis
In the following, OLS analysis steps with regards to the variables CSR performance (ESG), gender diversity (Gender), nationality diversity (Nationality) and age diversity (Age) are provided.
Firstly, the extracted ESG and diversity data is screened for abnormalities, missing values, and distribution patterns in order to verify the validity of performing a multiple linear regression (Saunders and Lewis, 2012). Therefore, the five key assumptions for multiple linear regressions are tested namely, linearity, multivariate normality, no multicollinearity, no auto-correlation and homoscedasticity (Newbold et al., 2013).
Additionally, since OLS regressions are sensitive to outlier effects, abnormalities are identified with boxplot analyses and eliminated from the dataset. Hence, outliers are recognised and excluded from the observation set. After accounting for missing data and outliers, a total sample size of 117 observations is utilised for the regression analysis (Appendix 4). Generally, a confidence interval of 95% (α = 0,05) is utilised throughout all performed statistical tests (Saunders and Lewis, 2012). The following table (Table 2) summarises descriptive statistics related to the main variables.
Descriptive Statistics
N Range Minimum Maximum Mean Std. Deviation
ESG 117 98,217800749 0,770848523 98,9886492725 42,0915120556 23,8228580830
Gender 117 0,545 0,455 1,000 0,72476 0,124341
Age 117 11,0 2,6 13,6 7,928 2,2018
Nationality 117 0,8 0,0 0,8 0,251 0,2459
31
Linearity
The first assumption relates to the dependent variable (CSR performance) which needs to be a linear function of the independent variables (Newbold et al., 2013). Three performed ANOVA linearity tests supports the assumption of linearity (Appendix 5). The deviation from linearity values are statistically insignificant on a 95% confidence interval for all linearity tests between the independent variables and CSR performance. Graphical tests in form of simple scatterplots confirm these tests (Saunders and Lewis, 2012).
Multivariate normality
In this assumption, the regression method demands that the dependent variable follows a normal distribution (Newbold et al., 2013). The condition of multivariate normality is tested with a P-P-Plot (Figure 4). The normal probability plot is a graphical technique for assessing whether or not a data set is approximately normally distributed. Accordingly, the data is plotted against a theoretical normal distribution in such a way that the points should form an approximate straight line (Saunders and Lewis, 2012). The performed analysis thus confirms approximate normality. The histogram for the dependent variable (ESG) further hints that the data is approximately normally distributed (Figure 5).
32
Multicollinearity
Next, similarities between the independent variables are assessed. A multicollinearity test is performed to avoid issues related to interpreting the partial effect of the independent variables on the dependent variable (Saunders and Lewis, 2012). Specifically, the independent variables must not perfectly correlate (Newbold et al., 2013). Since the Variance Inflation Factor (VIF) values observed lie between 1-10, no multicollinearity is assumed (Table 3). The insignificant correlation values between the independent variables support this assumption. Additionally, the correlations provide the first indication for a possible regression significance between nationality and gender diversity with CSR performance which are analysed in the ‘Results’ chapter (section 7.).
Coefficients Model Unstandardised Coefficients Standardised Coefficients t Sig. 95,0% Confidence
Interval for B Collinearity Statistics B Std. Error Beta Lower Bound Upper
Bound Tolerance VIF
1 (Constant) 47,932 14,061 3,409 ,001 20,074 75,791
Gender 8,109 17,613 ,042 ,460 ,646 -26,786 43,003 ,960 1,042
Age -2,069 ,997 -,191 -2,075 ,040 -4,044 -,093 ,955 1,047
Nationality 18,640 8,858 ,192 2,104 ,038 1,092 36,189 ,971 1,030
Table 3: Coefficients and Variance Inflation Factors
33
Autocorrelation
Fourthly, OLS regression analysis demands little or no autocorrelation. Essentially, autocorrelation occurs when the residuals are not independent from each other. In other words, when the value of y(x+1) is not independent from the value of y(x) (Newbold et al., 2013). Henceforth, a Watson test is performed. Since the Durbin-Watson value of 1,986 lies outside the upper and lower levels given the current sample size of 117 observations, no significant autocorrelations take place. Consequently, it is concluded that this assumption is met.
Model Summaryb
Model R R Square Adjusted R Square
Std. Error of the
Estimate Durbin-Watson
1 ,289a ,083 ,059 23,109390478963828 1,986
a. Predictors: (Constant), Gender, Age, Nationality b. Dependent Variable: ESG
Table 4: Durbin Watson test – Model Summary
Homoscedasticity
34 It is sub-concluded that all five key assumptions for a multiple linear regression are met. Finally, the above ratified regression is performed applying the ENTER method. The results are provided in section 7 and discussed in section 8.
Additionally, in order to test the proposed moderation effect, Hayes’ moderation test is performed. The interaction term ‘country qualitative sustainability environment’ (Environment) is added as a moderator in each relationship between the respective diversity dimension and CSR performance proxy ESG.
35
7. Results
Firstly, SPSS outcomes of the OLS regressions are interpreted. Thereafter, findings related to the performed moderation tests are provided.
7.1. Multiple linear regression results
A multiple regression is performed to predict CSR performance (ESG) from gender diversity (Gender), age diversity (Age) and nationality diversity (Nationality). These independent variables statistically significantly predicted CSR performance, F= 3,424, p < 0,02, Adjusted R² = 0,059 (Table 5). Two out of three variables added statistically significantly to the prediction, given p < 0,05. The specific findings are described below.
Firstly, the model summary is screened. R-squared is a statistical measure of how close the data are to the fitted regression line (Newbold et al., 2013). The value of 0,083 illustrates that the multiple linear regression explains 8,3% of the variance in the data.
Model Summaryb
Model R R Square Adjusted R Square Std. Error of the Estimate
1 ,289a ,083 ,059 23,109390478963828
a. Predictors: (Constant), Gender, Age, Nationality b. Dependent Variable: ESG
Table 5: Multiple regression Model Summary
36
ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 5486,350 3 1828,783 3,424 ,020b
Residual 60346,964 113 534,044
Total 65833,314 116
a. Dependent Variable: ESG
b. Predictors: (Constant), Gender, Age, Nationality
Table 6: Anova test – Multiple regression analysis outcome
37 Coefficientsa Model Unstandardised Coefficients Standardised Coefficients t Sig. 95,0% Confidence Interval for B Collinearity Statistics
B Std. Error Beta Lower
Bound Upper Bound Tolerance VIF 1 (Constant) 47,932 14,061 3,409 ,001 20,074 75,791 Gender 8,109 17,613 ,042 ,460 ,646 -26,786 43,003 ,960 1,042 Age -2,069 ,997 -,191 -2,075 ,040 -4,044 -,093 ,955 1,047 Nationality 18,640 8,858 ,192 2,104 ,038 1,092 36,189 ,971 1,030
a. Dependent Variable: ESG
Adding the control variables to the model, led to the following changes in significance. The updated Anova test reveals that the multiple linear regression remained to be overall significant on a confidence level of = 0,05 (Table 8).
ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 6616,065 5 1323,213 2,433 ,040b
Residual 52220,787 96 543,967
Total 58836,852 101
a. Dependent Variable: ESG
b. Predictors: (Constant), Gender, Age, Nationality, BoardSize, ROA
Table 8: Anova test – Multiple regression analysis with control variables
The control variables themselves, are not statistically significant (Table 9). Nonetheless, adding the control variables to the model led to substantial changes in the significance of nationality diversity. Consequently, only the age diversity variable remains significant (p < 0,05). These results indicate an influence of the two added control variables on the association between nationality diversity and CSR performance. The findings and implications stated above are discussed and assessed in section 8.
Table 7: Coefficients of independent variables Table 8: Coefficients of independent variable
Table 9: Coefficients of independent variables Table 10: Coefficients of independent variable
Table 11: Coefficients of independent variables Table 12: Coefficients of independent variable
38 Coefficientsa Model Unstandardised Coefficients Standardised Coefficients t Sig. 95,0% Confidence Interval for B
B Std. Error Beta Lower Bound Upper Bound
1 (Constant) 65,734 18,258 3,600 ,001 29,492 101,976 Gender -3,475 19,976 -,017 -,174 ,862 -43,128 36,177 Age -2,701 1,055 -,251 -2,562 ,012 -4,795 -,608 Nationality 12,243 9,619 ,125 1,273 ,206 -6,850 31,337 BoardSize -,399 ,504 -,084 -,792 ,430 -1,400 ,602 ROA 76,112 59,369 ,135 1,282 ,203 -41,735 193,958
a. Dependent Variable: ESG
Table 9: Coefficients of independent variables and control variables
Performed binary linear regressions between each independent diversity variable and CSR performance confirmed the findings and relationship directions of the multiple linear regression. The tests support multicollinearity results which illustrated no significant inter-correlations among the independent variables.
7.2. Hayes’ moderation test results
In the following, results and findings related to the performed interaction tests are provided. Generally, overall significant models are produced while no individual interaction term is observed to be statistically significant ( = 0,05).
With regards to country qualitative sustainability environment´s moderating forces on the association between gender diversity and CSR performance, the overall statistical model is significant with F(3,113) = 6,9, p < 0,001, R2 = 0,1548 (Table 10).
Model Summary
Model R R Square MSE F Df1 Df2 Sig.
1 ,3935 ,1548 492,3931 6,9002 3 113 ,0003 a. Predictors: Gender
b. Moderator: Environment c. Dependent Variable: ESG
Table 10: Model summary – Country sustainability environment & Gender diversity
39 However, the interaction model details variable specific significance values (Table 11). Observing significance of the individual interaction terms show no statistical evidence supporting a moderation of the country qualitative sustainability environment (p > 0,05). While the overall model is significant, the data fails to reflect a statistically significant moderation on a 95% confidence interval.
Interaction Model
Model Coefficients se t Sig.
1 Constant 75,2599 31,5222 2,3875 ,0186 Environment -3,7161 4,1504 -,8954 ,3725 Gender -20,1979 41,8957 -,4821 ,6307 Interaction 1,6654 5,6655 ,2939 ,7693 a. Predictors: Gender b. Moderator: Environment c. Dependent Variable: ESG
Table 11: Interaction Model – Country sustainability environment & Gender diversity
Similarly, the moderation test related to age diversity and CSR performance produced a significant model with F(3,113) = 7,6, p < 0,001, R2 = 0,1679 (Table 12). Thus,
a significant relationship is observed between age diversity and CSR performance while accounting for country qualitative sustainability environment as a moderation effect.
Model Summary
Model R R Square MSE F Df1 Df2 Sig.
1 , 4098 , 1679 484,7523 7,6027 3 113 ,0001 a. Predictors: Age
b. Moderator: Environment c. Dependent Variable: ESG
Table 12: Model summary – Country sustainability environment & Age diversity
40
Interaction Model
Model Coefficients se t Sig.
1 Constant 63,6631 17,2785 3,6845 ,0004 Environment -1,4812 2,1640 -,6845 ,4951 Age -,6042 2,2012 -,2745 ,7842 Interaction -,0989 ,2623 -,377 ,7067 a. Predictors: Age b. Moderator: Environment c. Dependent Variable: ESG
Table 13: Interaction Model – Country sustainability environment & Age diversity
Finally, an overall significant model with F(3,113) = 7,3201, p < 0,001, R2 = 0,1627
is produced with regards to country qualitative sustainability environment´s moderation in the association between nationality diversity and CSR performance (Table 14).
Model Summary
Model R R Square MSE F Df1 Df2 Sig.
1 , 4034 , 1627 487,7978 7,3201 3 113 ,0002 a. Predictors: Nationality
b. Moderator: Environment c. Dependent Variable: ESG
Table 14: Model summary – Country sustainability environment & Nationality diversity
41
Interaction Model
Model Coefficients se t Sig.
1 Constant 57,5778 7,0401 8,1786 ,0000 Environment -2,4408 ,7721 -3,1610 ,0020 Nationality 4,1650 17,3030 ,2407 ,8102 Interaction ,8750 2,2354 ,3914 ,6962 a. Predictors: Nationality b. Moderator: Environment c. Dependent Variable: ESG
Table 15: Interaction Model – Country sustainability environment & Nationality diversity
42
8. Discussion
The performed statistical tests provide partial support for the previously validated hypotheses. The above analysed findings are discussed next.
8.1. TMT diversity and CSR performance
The first hypotheses dealt with the constructed research question regarding TMT diversity’s association with a firm’s industry-specific CSR performance. The general rationale for the positive correlations were ratified with previous research highlighting diversity’s impact on decision making processes (Kang et al., 2007; Luoma and Goodstein, 1999), synergies (Ibrahim, 2016) and agency theory related informational biases (Francoeur et al., 2008). The overall ANOVA test confirmed that the independent variables in the regression model predict CSR performance significantly well. Nonetheless, the three diversity dimensions’ respective associations with CSR performance are of high importance and henceforth discussed.
Gender diversity and CSR performance
It was hypothesised that TMT gender diversity positively correlates with the firm’s
finance industry-specific CSR performance (H1a). The rationale behind the proposed
relationship was built upon research indicating that gender diversity improves corporate governance structures (Webb, 2004) and encourages the inclusion of social issues and the welfare of diverse stakeholders into TMT decisions (Wang and Coffey, 1992).
43 diversity supports stronger CSR performance (i.e. Bernardi and Threadgill, 2010; Krüger, 2009).
It is worth to acknowledge that potential significant associations might be detected by further increasing the sample size. Nonetheless, these statistically inconclusive results are similar to Pletzer et al.’s research (2015) who investigated the association between percentages of women on the boards and firm performances. Referring to Pletzer et al.’s findings, the management professor Katherine Klein (2017) highlights possible explanations for the insignificant findings.
Accordingly, women appointed to TMTs may in fact not differ substantially from their male colleagues in terms of their values, experiences, and knowledge (Klein, 2017). Since the assumption of improved CSR performance rests on aforementioned synergies of diverse “[...] experiences, values and personalities” (Hambrick, 2007, p. 334), this claim potentially explains statistically insignificant results. Essentially, female TMT members are selected for their professional achievements, knowledge, and skills. If female and male board members are rather similar in their overall experiences, values and personalities, the addition of women to a TMT may not substantially improve the team’s cognitive variety as one might expect based on the upper echelons theory (Hambrick, 2007).
44 effective implementation of such policies. This may explain why no statistically significant relationship between TMT gender diversity and CSR performance can be observed why commonly acknowledged diversity synergies exist.
Age diversity and CSR performance
Secondly, it was hypothesised that TMT age diversity positively correlates with the firm’s
finance industry-specific CSR performance (H1b). Even though previous research is
characterised by conflicting findings (Liu and Zeng, 2017; Post et al., 2011) dominant findings pointed towards a positive association between age diversity and CSR performance (Hafsi and Turgut, 2013).
Interestingly, the above performed multiple linear regression supports findings claiming that age diversity is negatively associated with CSR performance. This study produced statistical evidence (p < 0,05) indicating that a higher standard deviation of age in TMTs can lower ESG performance scores and therewith, CSR performance. Contrary to scholars that highlight positive synergies stemming from experienced older TMT members and flexible younger directors (Hafsi and Turgut, 2013), the outcomes above support analyses based on aforementioned Social Psychology theories (Liu and Zeng, 2017). In specific, the analysed data confirms the potential effect of inter-personal tensions that weaken positive synergies. Consequently, CSR performance outcomes are potentially negatively influenced by a more age-diverse TMT.
decision-45 making processes as disagreement and inability to converge take their toll (Manzoni et al., 2010). Here, inter-personal polarisation issues refer to personal battles and inhibited discussions among TMT members. These tensions can diminish the aforementioned synergy potential and ultimately, lead to a lower quality of CSR related strategies. Nonetheless, it is crucial to acknowledge that the correlation (Pearson correlation: -0,208) and the regression’s standardised coefficient (-0,191) between age diversity and CSR performance are relatively weak. Hence, age diversity’s actual impact on CSR performance needs to be analysed carefully and suggests future research potential.
Nationality diversity and CSR performance
Thirdly, the literature review supported the hypothesis claiming that TMT nationality
diversity positively correlates with the firm’s finance industry-specific CSR performance (H1c). Correspondingly, network effects related to opinions and perspectives, such as life
experiences, culture, behaviour, and norms of the country or region (Ibrahim, 2016), positively contribute to CSR decision-making processes of TMTs (Zhang, 2012).
46 Likewise, the positive association between nationality diversity and CSR performance manifests claims incorporating that most foreign TMT members are usually outsiders and more independent (Masulis et al., 2010). Due to their widely spread physical and psychological international network with stakeholders, a high level of transparency and disclosures are anticipated. Similarly, Ayuso and Argandona (2007) declared that the knowledge incorporated into TMTs by foreign directors, benefits CSR reporting strategies. Additionally, a measurable benefit from other nationalities on boards include increased levels of board activity (Estelyi and Nisar, 2016). Accordingly, foreign nationality directors actively provide their inputs which is reflected in attendance behaviour, committee assignments and corporate governance.
Finally, the results upkeep Wang and Coffey’s claim of racial minorities’ strategy principles or tendencies (1992). Specifically, the scholars hypothesised that racial minorities place more importance on social welfare issues of diverse stakeholders due to their social and economic background. Ceteris paribus and in comparison to a nationality homogeneous TMT’s CSR performance, these strategy principles or tendencies potentially enhance CSR performance.
8.2. Country sustainability environment moderation
47 (Castelló et al., 2009). Respective moderation results are discussed below by referring to the three diversity dimensions individually.
Country qualitative sustainability environment, gender diversity and CSR performance
Firstly, it is hypothesised that the relationship between TMT gender diversity and the
firm’s finance industry-specific CSR performance is stronger in a high-quality sustainability environment than in a low-ranked country sustainability environment (H2a). Accordingly, a promotional environment and cultural acceptance potentially
enables women on boards to influence CSR related decisions and improve the creation of more holistic strategies (Castelló et al., 2009). Specifically, studies disclosed that the association between female directors and corporate performance can be moderated by external (institutional) influences (Low et al., 2015). Moreover, a beneficial environment might encourage women to speak up or challenge strategy norms. Consequently, previous literature hints towards improved CSR performances in high ranked sustainability environments with regards to gender diversity. Contrary, low ranked qualitative sustainability environments potentially hinder gender diverse boards in investing money into CSR strategies.
48 One reasonable explanation for the lack of statistical significance can be related to recent political engagements. A variety of EMU countries started to adopt legislative or voluntary initiatives to encourage gender diversity in TMTs. In a report of the European Commission from January 2018, national policies for gender-balanced boards are outlined. France, for instance, passed a law which requires 40 % gender parity on the board of every publicly listed firm. Additionally, Italy issued a law requiring listed and state-owned companies to employ one-third of their board members to be female and Spain provided numerical targets of 30% by 2020 (Senden and Kruisinga, 2018). Since these adopted legislative or voluntary initiatives differ among EMU countries and are adopted with varying strictness, the regression analysis performed may be influenced. However, the presence of such initiatives indicates the awareness and assumption of effects related to women on board on the governance of companies (Adams and Ferreira, 2009). Another justification of the insignificant moderation effect might be due to an insufficient sample size.
Country qualitative sustainability environment, age diversity and CSR performance
Secondly, the literature review suggested that the relationship between TMT age
diversity and the firm’s finance industry-specific CSR performance is stronger in a high-quality sustainability environment than in a low-ranked country sustainability environment (H2b). The hypothesis is underlined with findings of Low et al. (2015) who
49 Interestingly, the performed moderation test of a TMT’s country qualitative sustainability environment on the correlation between age diversity and CSR performance failed to produce statistical support. Thus, the dataset examined does not statistically upkeep enhancing forces by beneficial sustainability environments on the relationship between TMT age diversity and CSR performance.
The findings contradict scholar’s claims incorporating that a promotional environment enabled more diverse boards to transform strategic CSR policies into practical strategies (Castelló et al., 2009). This might be due to unsubstantial forces of sustainability environments on TMT age diversity’s relationship with CSR performance. In specific, the results show that the sustainability environment defined in this study neither encourages nor diminishes aforementioned possibilities of generation conflicts (Milliken and Martins, 1996). Hence, decision making processes remain to be influenced by inter-personal conflicts within TMTs (Manzoni et al., 2010), indifferent from the TMT’s operational sustainability country environment.
Country qualitative sustainability environment, nationality diversity and CSR performance
Lastly, research supported a claimed relationship between TMT nationality diversity and
the firm’s finance industry-specific CSR performance is stronger in a high-quality sustainability environment than in a low-ranked country sustainability environment (H2c). This exploratory hypothesis’ theoretical rational contained evidence indicating
50 forces are mentioned to potentially influence board’s openness to accept CSR strategy inputs from foreign nationals.
However, the moderation tests performed failed to produce evidence for these claims. The country sustainability environment seems not to significantly influence the relationship between nationality diversity and CSR performance. These findings contradict previous results stemming from the environmental context research which seems to moderate the relationship between racial diversity and firm performance (Richard et al., 2007).
One possible explanation for the statistically insignificant finding is related to the examined finance industry. The international finance industry environment may be insusceptible to some extent with regards to external sustainability environment forces and TMT nationality diversity. Different nationalities on finance industry boards may be influenced by other external forces than the sustainability environment when taking CSR related decisions. The international competitive environment with complex stakeholder networks hence may outweigh the country sustainability effects on the relationship between nationality diversity and CSR performance.
51 the upper echelons theory. Specifically, TMTs’ external environments were initially assumed to be important moderators in the association between TMT composition and organisational outcomes (Finkelstein et al., 2009; Carpenter et al., 2004). The results however, contradict the claims and therewith shed light on the complexity of the relationships. Accompanying limitations to the upper echelon’s theory’s nature are further discussed in section 9.
8.3. Evaluated effects of control variables on given findings
52 The following table (Table 16) summarises the discussed findings of the given hypotheses. Results overview Statistical test Findings Hypothesis 1
a. TMT gender diversity positively correlates with the
finance industry-specific CSR performance. Regression Not significant
b. TMT age diversity positively correlates with the
finance industry-specific CSR performance. Regression (-) negative significant relationship c. TMT nationality diversity positively correlates with
the finance industry-specific CSR performance. Regression (+) positive significant relationship
Hypothesis 2 a. The relationship between TMT gender diversity and
the finance industry-specific CSR performance is stronger in a high-ranked sustainability environment than in a low-ranked country sustainability
environment.
Hayes’ moderation test
Not significant
b. The relationship between TMT age diversity and the finance industry-specific CSR performance is stronger in a high-ranked sustainability environment than in a low-ranked country sustainability environment.
Hayes’ moderation test
Not significant
c. The relationship between TMT nationality diversity and the finance industry-specific CSR performance is stronger in a high-ranked sustainability environment than in a low-ranked country sustainability
environment.
Hayes’ moderation test
Not significant
53
9. Limitations and Conclusions
Finally, limitations are acknowledged and conclusions are drawn. Additionally, managerial implications, the thesis’ contributions and future research suggestions are stated.
9.1. Limitations
As elaborated on above, Hambrick and Mason (1984) suggested in their upper echelons theory that cognitive diversity in TMTs is needed for success and that demography can serve as a proxy for underlying deep-level personal factors such as personality, power, values, and interests. Moreover, Carpenter et al. (2004) labelled these TMT characteristics as observable variables. This thesis utilised observed demographics which is partly decided based on Carpenter et al.’s piece of advice as well as previous research on TMT diversity’s relationship with firm outcomes (i.e. Hagendorff and Keasey, 2012; Post et al., 2011; Bernardi and Threadgill, 2010). Hence, one essential limitation relates to the observable proxies accompanying the upper echelons theory. In specific, one is uncertain about which aspects of the deep-level attribute is being captured by the proxy (Priem et al., 1999). Therefore, limitations occur in demographics–based studies with regards to what the measured demographics actually reflect. Consequently, the mechanisms or processes through which the three diversity dimensions are related to CSR performance have been assumed and remain largely unexplored.
54 might be discriminatory and unfair to the disadvantaged individuals. Therefore, supplementing demographics–based studies, such as this dissertation, with psychological research is essential. Other critiques involve upper echelons theory’s nature which emphasises internal factors of TMT members for strategic choices. Consequently, it is argued that external forces such as successful strategies of other companies, are neglected. Such external success dynamics may influence TMT’s decisions and might be utilised as cases in the decision-making process. Moreover, the upper echelons theory and hence this study, is limited to the decision made by the TMT. In case of more complex and interdependent structures, decisions or tasks may be transferred to lower levels in the firm. These possibilities are disregarded and thus, constitute theoretical limitations to this research.
Generally, the restricted choice of EMU countries and the utilisation of firms operating in the finance industry, limit this study’s research scope. Moreover, one needs to acknowledge the sample’s characteristics of solely public and mostly internationally operating enterprises. Therefore, the proposed moderating environmental effect might have a more substantial scope than the applied country’s sustainability effect. In other words, it is recognised that the assessed qualitative country sustainability environment does not reflect the entire moderating environmental forces.
55 Nationality diversity is assessed by computing the percentage of foreigners on board. Even though the statistical assessment is in line with previous research (Ibrahim, 2016), it neglects the number of different nationalities on board and their magnitude. Specifically, the method is subject to limitations since geographical as well as cultural distances within the diversity dimension are disregarded.
Furthermore, this research potentially is threatened by an omitted variable bias. Such bias involves falsely omitting an independent variable that correlates with both the dependent variable and one or more of the included independent variables (Greenfield, 2016). The extensive literature analysis is performed to minimise this threat.
9.2. Conclusions
This thesis aimed to confront conflicting findings and to follow other study’s indications in the TMT and CSR performance research domain. In the following, relevant hypotheses, findings and implications are briefly highlighted. In conclusion, contributions and future research potential are outlined.
TMT diversity and CSR performance
