Master Thesis
Does International Experience Matter?
The Influence of CEOs’ International Experience on Firm Innovativeness
Tanja Meier s3352625
t.s.meier@student.rug.nl
Supervisor: Dr Esha Mendiratta M. Sc. International Business & Management
University of Groningen
b7080749
t.s.meier2@newcastle.ac.uk
Supervisor: Dr Iain Munro
M. Sc. Advanced International Business Management & Marketing Newcastle University Business School
Date of Submission: 17th November 2018
Abstract
This thesis aims to advance the current understanding in research of which Chief Executive Officer (CEO) characteristics drive firm innovativeness. Based on the upper echelons theory and concepts of social psychology, this thesis argues that CEOs with more international experience are more open-minded and risk-tolerant and therefore spur firm innovativeness, measured through Research & Development (R&D) intensity. This thesis answers the question whether a high level of international experience of CEOs is related to a high level of R&D intensity of firms and examines how CEO power moderates this relationship. This thesis focuses on the chemical industry in Germany and develops a novel index for international experience (IE Index) that considers all three sources of international experience: education, work and nationality. Using a four-level scale, the international experience of CEOs is evaluated based on three different aspects. First, the depth of the CEOs’ international experience measured through the number of years, second, the breadth of international experience measured through the number of foreign countries, and third, the degree of foreignness measured through the cultural distance between countries. Panel data for a five years period (2012 to 2016) is gathered from several sources such as BoardEx, Compustat and Bloomberg. To test the relationship between the international experience of CEOs and the R&D intensity of firms this thesis runs a Generalised Least Squares test using a Random Effect Model and Maximum Likelihood Estimates. The results suggest that more international experience of CEOs is not inevitably related to higher levels of R&D intensity. Furthermore, the results do not corroborate that CEO power moderates this relationship. The main contribution of this thesis is its novel approach of measuring the international experience of CEOs in a more comprehensive way. Existing approaches either do not consider nationality as a source of international experience, or do not distinguish international experience by its depth, breadth and degree of foreignness. This new approach of measurement provides research with new insights regarding the relationship between CEO characteristics and firm innovativeness. Furthermore, it contributes to closing the aforementioned gap in innovation literature and encourages future research to further examine which determinants explain the variance in firm innovativeness and firm performance.
Acknowledgements
Table of Contents
Abstract ... 2 List of Figures ... 5 List of Tables ... 5 List of Abbreviations ... 6 1. Introduction ... 72. Literature Review and Hypotheses Development ... 9
2.1. Firm Innovation ... 9
2.2. The Influence of CEOs on Organisational Outcomes ... 10
2.3. The Influence of CEOs’ International Experience ... 13
2.4. The Moderating Effect of CEO Power ... 17
3. Methodology ... 19 3.1. Sample ... 19 3.2. Data Collection ... 20 3.3. Measurement ... 21 3.3.1. Dependent Variable ... 21 3.3.2. Independent Variables ... 22 3.3.3. Control Variables ... 28 3.3.4. Moderator Variable ... 30 3.4. Analytical Methods ... 31 4. Results ... 32 4.1. Descriptive Statistics ... 32 4.2. Correlations ... 33
4.3. Regression Results and Hypotheses Testing ... 34
5. Discussion ... 38
6. Conclusion ... 42
Appendix 1 – SIC Codes ... i
Appendix 2 – List of Firms and CEOs ... ii
Appendix 3 – Variances of Hofstede’s Cultural Dimensions ... v
Appendix 4 – CEO Nationalities ... vi
Appendix 5 – Hausman Test ... vii
Appendix 6 – Breusch-Pegan Lagrangian Multiplier Test ... viii
Appendix 7 – VIF Values ... ix
Appendix 8 – Firm R&D Intensity and CEO Nationalities ... x
List of Figures
Figure 1 – Conceptual Research Model ... 18Figure 2 – R&D Intensity Calculation ... 22
Figure 3 – KS Index to Measure Cultural Distance ... 23
Figure 4 – KS Value Frequencies ... 26
Figure 5 – IE Index Calculation ... 27
Figure 6 – ROA Calculation ... 29
Figure 7 – R&D Average and CEO Nationality ... x
List of Tables
Table 1 – Means, Standard Deviations, Minimum/Maximum, Median, Standard Errors ... 32Table 2 – Correlation Matrix ... 34
Table 3 – Regression Results and Hypotheses Testing ... 37
Table 4 – Description SIC Codes ... i
Table 5 – Firms and their CEOs ... iv
Table 6 – Variance of Hofstede's Cultural Dimensions Calculated by Kandogan (2012) ... v
Table 7 – Dummy Coding CEO Nationalities ... vi
Table 8 – Results Hausman Test ... vii
Table 9 – Breusch-Pegan Lagrangian Multiplier Test ... viii
List of Abbreviations
al. aliud
AG Aktiengesellschaft (public limited company) CEO Chief Executive Officer
Dr Doctor
e. g. exempli gratia
FE Fixed Effect
GLS Generalised Least Squares
H Hypothesis
i. e. id est
IE International Experience
IND Indulgence
Int. International
KS Kogut and Singh
LM Lagrangian Multiplier
MBA Master of Business Administration M. Sc. Master of Science
OECD Organisation for Economic Co-Operation and Development OLS Ordinary Least Squares
PhD Philosophiae Doctor
R&D Research and Development
RE Random Effect
ROA Return on Assets
SIC Standard Industrial Classification TMTs Top Management Teams
UAI Uncertainty Avoidance
UK United Kingdom
UNESCO United Nations Educational, Scientific and Cultural Organization
1.
Introduction
In research and practice innovation is considered to be a key driver of firm top-line revenue and firm performance (Oke et al., 2009; Torchia et al., 2011). Especially with regard to growing competition in global markets, firms increasingly consider innovativeness as a fundamental factor that provides firms with a competitive advantage, expanded market shares, enlarged productivity and positive reputation (Barker and Mueller, 2002; Gunday et al., 2011; Hill and Snell, 1989).
Within the last two decades, research showed increasing interest in factors that drive innovativeness of firms and therefore their competitiveness and firm performance (Gunday et al., 2011). Prior research, for example, examined the influence of firms’ geographic location (Allred and Steensma, 2005), board composition (Allemand et al., 2017) and national regulations (Blind, 2012) on the innovativeness of firms. Due to the vital importance of innovation for competitiveness and firm performance, strategic innovation decisions are usually made by Chief Executive Officers (CEOs) and Top Management Teams (TMTs) (Barker and Mueller, 2002). Although the field of research on innovation is growing, only few studies examined the influence of CEOs on the innovation activities of firms so far (Barker and Mueller, 2002; Sariol and Abebe, 2017). This is surprising in view of the influence researchers often attribute to CEOs in strategic decision-making (e. g. Chin et al., 2013; Maitland and Sammartino, 2015). The few studies that focus on the influence of CEOs on innovation decisions consider the role of the CEOs’ age, level of education, tenure, functional background and prior work experience (Chaganti and Sambharya, 1987; Daellenbach et al., 1999; Thomas et al., 1991).
The majority of research that examines the reasons for firm innovativeness concentrates on firm-level factors, e. g. firm size, corporate strategy and ownership structures (e. g. Baysinger et al., 1991; Hong et al., 2016). Recent research increasingly focuses on the influence of externalities, e. g. industry, innovation clusters, networks and other external sources, on firm innovation (e. g. Allred and Steensma, 2005; Azadegan et al., 2013; Bellamy et al., 2014; Blind, 2012; Mazur, 2016; Nomaler and Verspagen, 2016). Von Hippel et al. (2011), for example, stresses the importance of consumers as an external source for the development of new products and services. Najafi-Tavani et al. (2018), to name another example, encourage firms to cooperate with external partners in innovation networks to facilitate the development of new ideas and the exchange of information. Especially with a glance to modern communication technologies, such as the internet and social media, the establishment of corporate networks and close relationships between firms and consumers seem to be more feasible than ever. Some scholars such as Benkler (2007) and Lessig (2002) claim that through these technologies the role of firms as the main provider of information for consumers could be sustainably changed as consumers are enabled to generate and freely share information without the influence of firms (Benkler, 2007; Lessig, 2002).
Although firm-level factors and externalities certainly influence firm innovativeness to a certain extent, this thesis only focuses on the influence of CEOs, as they are considered to be the main drivers of innovation. They are the ones who at last decide whether or not, and to which extent, firms invest in R&D activities.
This thesis is structured as follows. First, it reviews and discusses relevant literature and develops appropriate hypotheses to answer the research question. In the methodology part, the data collection process, the measurements of variables and the statistical tests are described. In the following part, the test results are presented and discussed. Finally, in the conclusion part, this thesis illustrates the implications of the results, points out the limitations of this thesis and gives an outlook on future research.
2.
Literature Review and Hypotheses Development
2.1. Firm Innovation
In literature a plurality of perceptions and definitions of the term innovation exist (Atalay et al., 2013; OECD, 2015). Generally speaking, innovation is a complex process based on the concept of novelty (Gupta et al., 2007). On the firm-level, research distinguishes several types of innovation, i. e. (1) product, (2) process, (3) marketing, (4) organisation (OECD, 2005, 2015), (5) managerial (Damanpour, 1991) and (6) business model innovation (Kırım, 2007). This thesis follows the definition of innovation by Gunday (2011) and Drucker (1985). The authors define innovation as the process of equipping in improved, new capabilities or improved utility (Drucker, 1985; Gunday et al., 2011). This definition is chosen, first, because it is general enough to account for all types and sorts of firm innovation regardless of its kind, extent or origin. Such an approach allows for a more holistic view on the predictors of firm innovation. Second, this definition was chosen because it considers innovation activities as a time-consuming process rather than a one-time event. Therefore, it better reflects reality with regard to the complex procedures related to the development and implementation of new products, services, processes, management and business model practices.
Several scholars examined which institutional and firm-level factors actually cause a firm to be innovative. Based on the resource-based-view and the organisational learning theory, some scholars argue that partnerships, alliances and certain network structures can positively influence firm innovativeness (Azadegan et al., 2013; Bellamy et al., 2014; Elenkov and Manev, 2005, sa; Sampson, 2007). Other scholars showed that the industry structure, national context, and regulatory frameworks have a direct impact on firm innovation (Allred and Steensma, 2005; Baysinger et al., 1991; Blind, 2012).
Another field of research particularly examines the influence of the composition and ownership structures of management boards and the characteristics of TMTs on the innovativeness of firms. Baysinger et al. (1991) and Kor (2006) showed that a high insider representation on the board, a concentration of equity among institutional investors and knowledge regarding the skills and idiosyncratic habits of fellow TMT members positively influence investments in innovation. TMT members with long tenures, on the other hand, cause lower levels of innovation investments (Kor, 2006).
Within this particular field of research only few scholars have focused on the influence of CEOs on the innovativeness of firms. Based on the upper echelons theory, the next chapter reviews existing literature in this field and argues for the important role of CEOs in corporate decision-making processes.
2.2. The Influence of CEOs on Organisational Outcomes
The Upper Echelons Theory
Murphy, 2012). Hambrick and Mason (1984) argue that each person has an individual perspective on situations due to different cognitive bases, values and experiences. These perspectives act as a filter which narrows the options people consider as available (Yunlu and Murphy, 2012). As a consequence CEOs and TMT members make different decisions that can ultimately explain the variance in strategic choices and organisational outcomes (Buyl et al., 2011; Hambrick and Mason, 1984).
With the upper echelons theory the authors provided novel insights that encouraged researchers to further examine the role of CEOs and TMTs on organisational outcomes (Cannella and Holcomb, 2009). The vast majority of this research examines the influence of demographic factors, e. g. gender, age, tenure and functional background, on strategic choices and organisational outcomes. Peni (2014) argues that there is a positive relationship between the presence of women in the TMT and firm performance. Auden et al. (2006) found a positive relationship between the age, functional background and tenure of TMTs and firm performance. The authors showed that TMTs that are homogeneous in terms of age, and diversified in terms of functional background and team tenure, positively influence firm performance measured through return on assets (ROA) (Auden et al., 2006). Díaz-Fernández et al. (2014) revealed that there is a significant and negative relationship between the diversity of TMT members regarding their education-level and firm performance measured through ROA and return on equity. In other words, TMTs with different levels of education seem to diminish firm performance (Díaz-Fernández et al., 2014). Although findings regarding the influence of demographic factors such as age and gender are mixed in literature, it is widely accepted that the specific characteristics of TMT members can, to a certain degree, explain the variance in strategic choices and organisational outcomes.
The Influence of CEOs on Firm Innovation
Research about the influence of CEOs on firm innovation in particular is rare. This will probably change as some researchers argue that individual differences between CEOs, e. g. regarding their personal attitude towards innovation, have more influence on innovation outcomes than other institutional factors such as urbanisation and community wealth (Damanpour, 1991).
tenures. The authors did not find evidence that a higher education of a CEO results in higher levels of innovation (Barker and Mueller, 2002). The positive relationship between the work experience of CEOs and firm innovation was confirmed by other studies. Datta and Guthrie (1994) revealed that CEOs with a technical functional background are more likely to advance firm innovation. In contrast to Barker and Mueller, the authors further argue that higher levels of education are positively related to innovation investments. According to Custódio et al. (2017), general managerial skills of CEOs are more likely to result in a higher degree of firm innovation compared to specialised CEOs who gained in depth experience in only one organisational area, industry or firm (Custódio et al., 2017). The authors claim that generalist CEOs have knowledge that goes beyond one specific technological domain, and that, in turn, promotes innovation (Custódio et al., 2017). Young et al. (2001) showed that hospitals with older, better educated, and innovation experienced executives are more likely to adopt innovative business practices. Their findings regarding the influence of CEO age are in contrast to Barker and Mueller (2002), whereas their findings regarding the influence of the level of education are in line with Datta and Guthrie (1994). Yunlu (2012), and Cho and Kim (2017) showed that CEOs with shorter career horizons are related to less corporate breakthrough innovations and less investments in innovation activities. They argue that CEOs who are about to retire want to protect their own success and therefore reject investments in risky innovation strategies (Cho and Kim, 2017; Yunlu and Murphy, 2012).
Next to demographic CEO characteristics, some scholars identified CEOs’ personality traits to be a reason for variance in organisational outcomes. Galasso and Simcoe (2011) showed that firms with overconfident CEOs, i. e. CEOs who underestimate the probability of failure, drive innovation investments. Hirshleifer et al. (2012) also found a positive relationship between the overconfidence of CEOs and the level of firm innovation. Internal locus of control and self-directive values of executives were identified as other factors that can drive innovation in firms (Berson et al., 2008; Miller and Toulouse, 1986). Abdel-Khalik (2014) in his recent study showed that risk-tolerant CEOs lead to higher levels of investments in innovation. His findings are plausible as investments in innovation activities require executives to take higher risks (Kor, 2006).
As the aim of this thesis is to examine the influence of international experience of CEOs in particular, the next section discusses previous findings in literature regarding the influence of international experience of CEOs on organisational outcomes in more depth and develops hypotheses based on the upper echelons theory and concepts of social psychology.
2.3. The Influence of CEOs’ International Experience
Due to the growing number of international acquisitions and foreign alliances, international experience of CEOs and executives becomes increasingly important in business (Bartlett and Ghoshal, 1989). As mentioned earlier, the influence of the experience of CEOs on corporate outcomes and strategies is already discussed in literature (e. g. Barker and Mueller, 2002; Datta and Guthrie, 1994; Wang et al., 2016). However, not many papers consider the role of
international experience in particular (e. g. Rodenbach and Brettel, 2012; Sambharya, 1996).
In these papers, international experience is described as a valuable source of knowledge and competitive advantage (e. g. Carpenter et al., 2000; Daily et al., 2000; Reuber and Fischer, 1997). Knowledge based on international experience decreases the uncertainty that is attached to foreign environments, provides access to international networks and can thereby increase firm performance (Athanassiou and Nigh, 1999). Following the argumentation of the upper echelons theory, international experience shapes the CEOs’ perspectives, skills, values and attitudes. This, in turn, influences how CEOs notice, interpret and process information and consequently determines which strategic decisions they favour (Hambrick, 1994; Wang et al., 2016).
levels of firm innovation. As it is discussed later, firm innovation can be measured by the R&D intensity of a firm. Therefore, hypothesis H1 is formulated accordingly:
H1: The international experience of CEOs is positively associated with the R&D intensity of their firms.
Research distinguishes three sources of international experience: international work experience, international education experience and experience through foreign nationality. However, these three sources of international experience have not been integrated into one index so far. Each source of international experience is discussed in the following sections.
International Work Experience
Based on the arguments of Wang (2016), this thesis defines international work experience as prior international work experience, which is the amount of time a CEO spent abroad working in various positions. CEOs can gather international work experience either through assigned or self-initiated expatriation (Schmid et al., 2018). Increasing international work experience enlarges the knowledge base of CEOs and thereby augments their self-confidence to competently handle situations they are confronted with (Wang et al., 2016). Managers with this kind of experience are considered to have the knowledge and understanding that is necessary to successfully manage complex and dynamic operations (Perlmutter and Heenan, 1974). Hermann and Datta (2006) show in their study that CEOs with international work experience show a greater propensity to choose risky greenfield investments and acquisitions over joint ventures. With their study the authors confirm that CEOs with international work experience are more risk-tolerant and willing to engage in activities with unpredictable outcomes. Based on the findings in literature, I argue that CEOs with international work experience are more risk-tolerant and confident to cope with unpredictable situations. That is why I assume that CEOs with international work experience tend to advocate R&D investments. Therefore, hypothesis H1a is formulated accordingly:
International Education Experience
Therefore, hypothesis H1b is formulated accordingly:
H1b: The international education experience of CEOs is positively associated with the R&D intensity of their firms.
Foreign Nationality
In this thesis foreign nationality is understood as the discordance between the nationality and hence cultural background of the focal CEO and the country of his or her employment. Existing business research examined the influence of cultural diversity on the board level. In this area of research, scholars found mixed results regarding the relationship between culturally diverse boards and firm performance (Delis et al., 2017; Earley and Mosakowski, 2000; Larkey, 1996; Milliken and Martins, 1996). Some scholars argue that cultural diversity on the board level can result in conflicts, dissatisfaction and social barriers between board members, which can decrease the effectiveness of decision-making, and subsequently firm performance (e. g. Dumas et al., 2013; Frijns et al., 2016; García-Meca et al., 2015; Verkuyten et al., 1993). In contrast to these findings, other researchers argue for a positive relationship between culturally diverse boards and firm performance (e. g. Delis et al., 2017; Estélyi and Nisar, 2016). This group of researchers argues that foreign nationalities can be a valuable source of knowledge regarding foreign markets, business practices and regulations, different skills and abilities. Thereby, cultural diversity on the board level allows the board members to make better decisions based on a more diverse knowledge base and skill portfolio (Delis et al., 2017; Muttakin et al., 2015; Nielsen and Nielsen, 2008; Phillips-Wren, 2018). Decisions become more creative (Adams et al., 2015; Swann et al., 2004) and informed as different views, opinions and perspectives can be compared and discussed (Adams et al., 2015; Nielsen and Nielsen, 2008; Phillips-Wren, 2018). More informed decisions, in turn, can increase firm performance (Estélyi and Nisar, 2016). In addition, a more diverse knowledge base and skill portfolio can decrease uncertainty of board members regarding strategic and far-reaching decisions and can thereby increase risk-tolerance regarding the adoption of novel business practices (Bang and Frith, 2017).
risk-aversion of board members regarding risky strategic decisions such as R&D investments. Therefore, it is assumed that firms with CEOs with a foreign nationality are more likely to invest in the development of novel products and services. Therefore, hypothesis H1c is formulated accordingly:
H1c: The foreign nationality of CEOs is positively associated with the R&D intensity of their firms.
2.4. The Moderating Effect of CEO Power
The upper echelons theory argues that CEOs as the firm leaders play a significant role in the decision-making process of firms (Hambrick and Mason, 1984). However, their influence on organisational outcomes is limited. Research about managerial discretion examines whether and in particular when CEOs have an impact on organisational outcomes and firm performance. Managerial discretion can be defined as “the degree of influence that executives might exert” (Sirén et al., 2018, p. 956). The higher the discretion of CEOs, the higher is their influence and effect on organisational outcomes. According to Sirén (2018), the discretion of CEOs is composed of three types of factors: (1) organisational factors, i. e. the degree to which a firm itself is able to exert possible actions and to empower the CEO to execute them, (2) environmental factors, i. e. the degree to which the environment and context allows actions and change, and (3) individual factors, i. e. the degree to which a CEO is personally able to envision and exert possible actions. As mentioned earlier, the aim of this study is to examine the influence of CEOs’ international experience on R&D intensity. Therefore, this thesis concentrates on the individual factors of managerial discretion as they are most likely to influence the relationship between the international experience of CEOs and the R&D intensity of firms. Organisational and environmental factors are not considered in this thesis.
possesses, the higher is the probability that his or her international experience influences the R&D intensity of firms. Therefore, hypothesis H2, H2a and H2b are formulated accordingly:
H2: The influence of a CEO’s international experience on R&D intensity is moderated by the power the CEO has relative to the board.
H2a: The influence of a CEO’s international experience on R&D intensity is higher for a CEO with a high level of power relative to the board.
H2b: The influence of a CEO’s international experience on R&D intensity is lower for a CEO with a low level of power relative to the board.
The literature review shows, first, that innovation can be a driver for firm performance and productivity. These findings stress the importance and relevance of innovation research. Second, in line with the upper echelons theory, CEOs and TMTs play a crucial role in the decision-making process of firms. They are to a great extent responsible for the variance of firms regarding R&D spending and innovation investments. Third, foreign international work and education experience, as well as the foreign nationality of CEOs probably have an impact on firms’ R&D intensity. The hypotheses that are developed based on these findings are summarised and visually presented in the following conceptual model.
The following chapter explains the methodology used to test the proposed hypotheses. First, the sample and industry that has been selected to test the hypotheses is described. Afterwards, the data collection process and the measurement of all variables is explained in detail. Finally, adequate statistical tests are proposed and applied to test the hypotheses.
CEO’s International Experience
- Int. Work (H1a+) - Int. Education (H1b+) - Foreign Nationality (H1c+) R&D Intensity of Firm CEO Power relative to the Board H1+ H2, H2a, H2b Control Variables: Revenue Revenue t-1 Firm Performance CEO Age CEO Nationality
3.
Methodology
3.1. Sample
As suggested by Dess et al. (1990), the proposed thesis focuses on one single industry and one single country in order to control for country effects (e. g. governmental subventions) and industry effects (e. g. high levels of competition). For this reason, the sample is drawn from the German chemical industry only. This industry is classified as group 28 – “Chemicals and Allied Products” – in the United States (US) Standard Industrial Classification (SIC). An overview of all sub-groups within this industry class is presented in the appendix (see Appendix 1).
The chemical industry has been chosen as the empirical setting of this study because innovation and R&D plays a crucial role in this industry. Due to increased global competition, the development and implementation of innovations are one essential key for success in this industry (Marcinowski, 2006). That is why firms in the chemical industry on average show relatively high investments in R&D compared to other industries.
The chemical industry in Germany is characterised by a high degree of variation regarding R&D expenditures in absolute numbers. Some firms strongly invest in R&D activities, others barely do. The reason for that is that the German chemical industry includes small, highly specialised laboratories which are restricted in budget and man power as well as global corporations, e. g. Bayer and Beiersdorf, which are able to fund cost-intensive R&D projects on a regular basis. Another reason for this high variation next to firm size, is the difference of operational focus. Some firms focus on the cost-intensive development and commercialisation of new chemical products, e. g. new pharmaceuticals, whereas other firms are specialised in the production of off-patent chemicals and generic medicinal products that do not necessarily require prior investments in R&D (German Economic Institute, 2018). The wide range of R&D expenditures within the chemical industry ensures sufficient variation of the dependent variable and therefore a higher generalisability of the results.
Information about German firms operating in the chemical industry was collected for a five years period (2012 to 2016) for the end of each financial year (31st December, 31st March, 30th September, or 30th November respectively). Panel data allows to eliminate the possibility of extraordinary R&D expenditures in the focal year, accounts for individual heterogeneity, and improves the reliability of statistical results (Baltagi, 2005; Hsiao, 2007). The collected panel data was strongly balanced, which means that data was available for all companies for all years.
3.2. Data Collection
Firms were selected from the Orbis and Compustat database, the register of membership of the Association for the German Pharmaceutical Industry and the Association of Research-Based Pharmaceutical Companies. All companies were considered, first, which operated in the chemical industry in Germany (SIC 28) between 2012 and 2016, and second, which had on average at least ten employees during that time.
Information about the R&D intensity of firms, the control variables, and the moderator variable was primarily gathered from annual reports, financial statements, corporate websites, LinkedIn profiles, vitas published on Bloomberg, and the online database Federal Gazette, a central platform for legal firm announcements and news published by the German Federal Ministry of Justice and Consumer Protection.
All firms and CEOs with missing or ambiguous information were excluded from the sample. Complete information was available for 60 firms and their CEOs for the period 2012 to 2016. Three outliners in terms of R&D expenditures were excluded from the sample. One third of these firms (n=19) appointed a new CEO between 2012 and 2016. Therefore, the number of observed CEOs exceeds the number of observed firms. The final sample includes 57 firms and 76 CEOs. The final sample amounts to 285 observations (57 firms x 5 years = 285 observations) in total. According to Roscoe (1975) and Harrell (2015), 285 observations are sufficient for meaningful statistical analyses. A list of all firms and CEOs is depicted in the appendix (see Appendix 2).
The following chapter introduces all variables that are necessary to test the hypotheses and describes how every variable is measured in detail.
3.3. Measurement
3.3.1. Dependent Variable
R&D Intensity
Innovativeness of firms can be measured in several ways. Deeds and Hill (1996), for example, measure innovativeness by the rate of new product developments. Others assess the innovativeness of firms through industry surveys (Oerlemans et al., 2013), or by the number of granted patents (Lahiri and Narayanan, 2013; Lin et al., 2012). The measurement of innovativeness by the number of granted patents is not sufficient with regard to the chemical industry. First, firms within this industry often spend several years and make high investments in R&D until they can apply for a patent. As a consequence, the number of granted patents per firm within the 2012 to 2016 period is assumed to be low for the specified sample. Second, as discussed earlier, some firms in this industry only focus on the production of chemicals and not the development of new products and patents. Hence, it is expected that a substantial number of firms within the specified sample has never received a granted patent between 2012 and 2016. Another way to assess the innovativeness of firms is the measurement of the R&D
et al., 1991; Hill and Snell, 1989; Leeuw et al., 2014; Scherer, 1984). This thesis follows the latter approach and calculates R&D Intensity using the following formula:
The number of employees is measured as the average number of employees in the respective years 2012 to 2016. As mentioned earlier, the size of firms within the chemical industry varies strongly. Some firms have only few, other firms thousands of employees. As it is expected that bigger firms can afford higher investments in R&D, the measurement of R&D Intensity seems to be advantageous because it additionally allows to control for firm size. Furthermore, this measurement is assumed to be more stable than other common divisions related to R&D, e. g. the quotient of R&D expenditures and sales (Hill and Snell, 1989; Scherer, 1984). The dependent variable R&D Intensity is a continuous numerical variable.
3.3.2. Independent Variables
International Experience
The international experience (IE) of CEOs represents the independent variable in this study. An index was developed to measure the overall IE of CEOs. This index considers, first, the CEOs’ international work experience (IE1 Int. Work), second, the CEOs’ international education experience (IE2 Int. Education), and third, the CEOs’ IE through foreign nationality (IE3
Foreign Nationality). While the international work and education experience is included in
several other studies that examine the influence of CEO characteristics on diverse organisational outcomes (e. g. Herrmann and Datta, 2005, 2006; Laufs et al., 2016; McDougall et al., 2003; Musteen et al., 2010; Reuber and Fischer, 1997), the influence of foreign nationality of CEOs is mostly neglected with just few exceptions (e. g. Nielsen and Nielsen, 2008; Simmonds and Smith, 1968). This thesis integrates the foreign nationality of CEOs and thereby develops a novel and more comprehensive index measuring the overall IE of CEOs.
Following the terminology of other scholars such as Magnusson and Boggs (2006), and Maitland and Sammartino (2015), the newly developed IE index measures the (a) depth, (b) breadth and (c) degree of the CEOs’ IE. International work and education experience are measured by its (a) depth, (b) breadth and (c) degree, whereas IE based on foreign nationality is only measured by its (c) degree.
The (a) depth of international work and education experience is measured by the number of years a CEO worked or studied abroad. The number of years is rounded off to whole numbers. This measurement is important to acknowledge that CEOs who worked or studied abroad for many years gained more IE and therefore are more likely to be open-minded and less risk-averse than CEOs who worked and studied abroad for less than one year, or who did not work and study abroad at all.
The (b) breadth of international work and education experience is measured by the number of foreign countries in which the CEOs worked or studied in. A country is considered as foreign when it does not match the nationality of the CEO. United Kingdom (UK), for example, is considered as a foreign country, when the focal CEO has a German nationality. In contrast, UK is not considered as a foreign country, when the focal CEO has a British nationality. This measurement is important to acknowledge that CEOs who worked or studied in many distinct countries gained more IE and therefore are more likely to be open-minded and less risk-averse than CEOs who worked and studied abroad in only one country, or who did not work nor study abroad at all.
The (c) degree of international work experience, education experience and foreign nationality is measured by the cultural distance between the CEOs’ countries of origin and the countries the CEOs worked or studied in. This study uses the index of Kogut and Singh (KS) (1988) to measure the cultural distance between countries. Although this index is criticised to have several limitations, the KS index is still one of the most popular indices for cultural distance used in cultural and business studies (Kandogan, 2012; Shenkar, 2012). The KS index is based on Hofstede’s (2001) original four dimensions of culture, that are Power Distance, Individualism, Masculinity and Uncertainty Avoidance. The cultural dimensions Long Term Orientation (LTO) and Indulgence (IND) are excluded from the study, because values for these dimensions are not available for all countries (Kandogan, 2012). The calculation of the KS index is depicted below.
The values of Iij and Iiu are taken from the internet website of Hofstede (Hofstede, 2018). They can range from 0 to 100 (Hofstede, 2001). The variance Vi of each dimension of Hofstede’s four
cultural dimensions are taken from Kandogan’s (2012) study. The calculated variances Vi of each dimension are presented in the appendix (see Appendix 3).
The measurement of (c) degree is important to acknowledge the role of cultural distance. CEOs who are born or have worked and studied in culturally very diverse countries gained more IE, and therefore are more likely to be open-minded and less risk-averse than CEOs who worked and studied in countries which are culturally similar to their countries of origin. A Chinese CEO, for example, presumably gains more IE when working and studying in Germany, than an Austrian working and studying in Germany.
The (a) depth, (b) breadth and (c) degree of international work and education experience and foreign nationality are all measured by using an ordinal four level scale. Thereby, the lowest level of IE, that is the absence of IE, is coded as 0, and the highest level of IE is coded as 3. The following section explains how the international work and education experience, and foreign nationality is coded in detail.
Coding of CEOs’ International Experience
The (a) depth of international work experience is measured in years worked abroad. The CEOs in the defined sample have worked abroad for seven years on average (median: 4 years). This high average can be explained by considering that some CEOs in the sample started their careers abroad and never moved back to their country of origin. Some of these CEOs worked abroad for more than 25 years (n=5). Based on these considerations, the (a) depth of international work experience is coded as 0 when the CEO has not worked abroad, as 1 when the CEO has worked abroad for one to four years, as 2 when the CEO has worked abroad for five to twelve years, and as 3 when the CEO has worked abroad for at least thirteen years.
Therefore, two different coding scales are used for international work and international education experience.
The (b) breadth of international work experience is measured by the number of foreign countries a CEO has worked in. The CEOs in the defined sample have worked in 1.3 foreign countries on average (median: 1 country). Therefore, the (b) breadth of international work experience is coded as 0 when the CEO has not worked in a foreign country, as 1 when the CEO has worked in one foreign country, as 2 when the CEO has worked in two foreign countries, and as 3 when the CEO has worked at least in three foreign countries.
The (b) breadth of international education experience is measured by the number of foreign countries a CEO has studied in. The CEOs in the defined sample have studied in 0.33 foreign countries on average (median: 0 countries). Therefore, the (b) breadth of international education experience is coded as 0 when the CEO has not studied in a foreign country, as 1 when the CEO has studied in one foreign country, as 2 when the CEO has studied in two foreign countries, and as 3 when the CEO has studied at least in three foreign countries.
A student has the possibility to study abroad by choosing a degree programme in a foreign country or by spending one or more semesters abroad. Hence, it is possible for students to study in many countries during the course of their studies. The same is applicable for employees. Especially with a glance to our globalised world it gets more feasible for employees to work abroad in several, different countries. Within the sample the CEO with the highest (b) breadth of international work experience worked in six foreign countries. As it is assumed to be possible to study in six foreign countries as well, the coding scale for the (b) breadth of international education and international work experience is set up identically.
As mentioned earlier, the (c) degree of international work experience, education experience and foreign nationality is assessed by the KS index measuring cultural distances. The values of this index can range from 0 to maximum 11.9. Figure 4 shows the frequencies of all rounded KS values calculated for all 5,184 possible pairs of countries that are considered in Hofstede’s analysis.
When measuring the (c) degree of international work and education experience, only the most foreign country is considered for coding. A German, for example, who worked or studied in Singapore and in Spain, is assigned to a coding group by calculating the KS value between Germany and Singapore, and not Germany and Spain, as only the most foreign country, which in this case is Singapore, is considered.
After coding all three elements of IE, i. e. (a) depth, (b) breadth and (c) degree, the average coding of the CEOs’ overall international work experience (IE1 Int. Work) and international education experience (IE2 Int. Education) is added to the degree of foreign nationality (IE3
Foreign Nationality). In theory the values of the so calculated IE index (IE Total) can range
from 0 (= CEO has no IE) to 9 (= CEO has high IE).
In a final step, the reliability and consistency of the developed IE index is estimated by calculating the Cronbach’s Alpha. Cronbach’s Alpha allows to assess whether international work experience (IE1 Int. Work), international education experience (IE2 Int. Education) and foreign nationality (IE3 Foreign Nationality) can be combined into one single, meaningful variable (IE Total). The calculated Cronbach’s Alpha is α=0.6042 which is slightly below the common threshold of α=0.7 (Nunnally, 1978). The coding criteria were changed several times to increase the Cronbach’s Alpha. Changing the coding criteria did not lead to an improvement of Cronbach’s Alpha. Schmitt (1996) argues that values below α=0.7 are not necessarily an impediment to use a developed scale. Therefore, the initial coding criteria as described earlier was assessed as the one that provides the highest level of reliability and consistency. Figure 5 summarises the calculation of the IE index and gives an overview of the final coding criteria.
27
3.3.3. Control Variables
To mitigate omitted-variable bias several control variables, both on the firm and the CEO-level, are considered. The next section explains the relevance and measurement of these control variables in detail.
Revenue
Revenue is added to the model to control for firm size and to take the availability of financial
means to fund R&D investments into account. Research shows that larger firms are more committed to R&D and are better able to innovate (Abdel-Khalik, 2014; Cohen and Klepper, 1996). In research, firm size is measured in multiple ways. Some studies assess firm size by the number of employees (e. g. Datta and Guthrie, 1994; Gunday et al., 2011; Herrmann and Datta, 2006), others by firm revenues (e. g. Cohen and Klepper, 1996; Custódio et al., 2017; Daily et al., 2000; Finkelstein, 1992; Gómez-Mejia and Palich, 1997). This thesis follows the latter, first, because the number of employees is already incorporated in the measurement of the dependent variable R&D Intensity, and second, because a high number of employees does not necessarily imply a high level of financial means to fund R&D investments. Information about Revenue is gathered from the financial annual statements published in the Federal Gazette. Revenue is a continuous numerical variable.
Revenue t-1
This thesis also controls for revenue of the previous year (t-1). It is assumed that not only the revenues of the focal year, but also the revenues of the previous year have an impact on decisions regarding R&D expenditures. Firms with a low revenue in the previous financial year are assumed to invest less in R&D to compensate the revenue deficit of the previous year. In contrast, firms with a high revenue in the previous financial year are more likely to increase their R&D and other corporate expenditures. Information about previous revenues is gathered from the financial annual statements published in the Federal Gazette. Revenue t-1 is a continuous numerical variable.
Firm Performance
Previous studies showed that Firm Performance has a great influence on strategic decisions made by CEOs and TMTs (Crossland et al., 2014). Firms tend to cut their R&D investments when their Firm Performance is low, and, in contrast, tend to increase their investments in R&D programmes when their Firm Performance is high (Barker and Mueller, 2002; Cho and Kim, 2017; Cyert and March, 1992; Daellenbach et al., 1999). Therefore, this thesis controls for Firm
which was calculated using the formula depicted in Figure 6 (Daellenbach et al., 1999; Díaz-Fernández et al., 2014).
Information about the net income, interest expenses and total assets is extracted from annual financial statements. ROA is chosen as a measurement for Firm Performance, first, because it is difficult for CEOs and TMTs to manipulate this measure, and second, because it is one of the most common measures of Firm Performance in business research (Daily et al., 2000; Gómez-Mejia and Palich, 1997). Firm Performance is a continuous numerical variable.
CEO Age
This study not only considers firm-level variables to influence the R&D Intensity of firms, but also CEO characteristics. One CEO characteristic that is often considered in research is the age of CEOs. Most research shows that older CEOs are less open-minded to novel strategies and processes, and are less willing to take risks (Serfling, 2014; Wang et al., 2016). Younger CEOs, in contrast, are more risk-tolerant (Hambrick and Mason, 1984). They tend to pursue more aggressive strategies compared to older CEOs (Yim, 2013). The reason for that is that younger CEOs have limited experiences and a less developed cognitive schema that would allow them to fully understand the consequences of new strategies and processes (Wang et al., 2016). Based on these findings, I posit that the age of CEOs does influence their decision-making regarding R&D expenditures and therefore R&D Intensity of firms. It is assumed that younger CEOs are related to higher levels of R&D Intensity. The CEOs’ age is primarily gathered from corporate websites and interviews published online. CEO Age is a continuous numerical variable.
CEO Nationality
Research shows that the nationality of a CEO influences corporate outcomes and strategic decision-making (Delis et al., 2017). The nationality of board members, for example, was identified to influence entry mode choices (Kogut and Singh, 1988; Nielsen and Nielsen, 2011) and Corporate Social Responsibility disclosure practices (Muttakin et al., 2015). Nationality can be described as a “superordinate determinant of a person's self-identity, derived through a meaning system shared with others” (Earley and Mosakowski, 2000, p. 26). To a certain degree nationality determines how risk-tolerant and open-minded a CEO is (Geletkanycz, 1997; Kogut and Singh, 1988). CEOs who have a nationality which is characterised by high levels of
tolerance, e. g. Singapore and Denmark, are assumed to favour uncertain innovation investments more than CEOs with nationalities which are characterised by high levels of risk-aversion, e. g. Greece and South Korea (Hofstede Insights, 2018). CEOs with a nationality that is characterised as risk-tolerant are assumed to make decisions that favour unpredictable actions and R&D investments. Information about the CEO Nationality is collected from corporate websites, interviews and vitas published online. The nationalities of CEOs are then assigned to an integer number to compute a dummy variable for nationality. An overview of all nationalities that occur in the sample and their assigned numbers are presented in the appendix (see Appendix 4). CEO Nationality is a nominal categorical variable.
3.3.4. Moderator Variable
CEO Power
CEO Power relative to the board represents the moderator variable in this thesis. Several studies
argue for a multidimensional construct to measure CEO Power in a comprehensive way (e. g. Chin et al., 2013; Finkelstein, 1992; van Essen et al., 2015). This thesis follows the most recent approach of Schmid et al. (2018) and constructs an index of CEO Power by using three proxy indicators, i. e. (1) Prestige Power, (2) Expert Power, and (3) Structural Power. These three proxy indicators are standardised and summed to compute an index for CEO Power.
(1) Prestige Power is measured through the number of board mandates a CEO currently holds in other organisations besides the one in the focal firm. Such mandates are, for example, positions in the councils of industry associations and employers’ federations. A high number of mandates increases the status and legitimacy of the CEO, and hence the power of the CEO relative to the board (Schmid et al., 2018).
(2) Expert Power is measured through the number of years since the CEO was appointed. The longer a CEO is in place, the higher is his or her expertise (Wang et al., 2016). The higher the CEO’s expertise, the higher is the probability that the CEO’s perceptions are considered in board decisions (Bhagat and Bolton, 2008; Ryan and Wiggins, 2001; Schmid et al., 2018). Furthermore, CEOs can establish valuable relationships with stakeholders over time that, in turn, strengthen their position (Combs et al., 2007).
more supervisory directors were appointed after the CEO the higher is the number of loyal board members and the higher is the power of the CEO (Pollock et al., 2002; Schmid et al., 2018).
The study of Schmid et al. (2018) additionally uses a fourth proxy which is Ownership Power. In their study the authors measure Ownership Power through the percentage of shares a CEO holds. They argue that the more shares a CEO owns the higher is the CEO’s influence on the board agenda and firm decisions such as R&D expenditures. This thesis does not consider Ownership Power, because data on ownership was unavailable for the majority of the observations. The moderating variable CEO Power is a continuous numerical variable.
3.4. Analytical Methods
4.
Results
4.1. Descriptive Statistics
The descriptive statistics in Table 1 give an overview of all 285 observations of all 57 companies and their 76 CEOs for the years 2012 to 2016. This table show that the dependent variable R&D Intensity varies strongly between the firms. The R&D Intensity of the observed firms ranges from 0€ minimum (n=29) to 288,135.60€ (Mologen AG in 2016) maximum. This can be explained, first, by the fact that some firms do not invest in R&D as they are specialised in the production and not the development of chemicals, second, because some firms outsource their R&D activities to external laboratories or affiliated companies, and third, because of the wide difference in the number of employees which is used for the calculation of R&D Intensity. The number of employees in the sample ranges from 14 employees (Nanorepro AG in 2012 to 2016) to 117,400 employees (Bayer AG in 2014). The relatively low median (14,626.82€) and mean value (36,095.02€) shows that most firms exhibit a R&D Intensity in the lower spectrum. Similar to the dependent variable R&D Intensity, the control variables Revenue and Revenue
t-1 show a high degree of variance. The Revenue of firms varies from minimum 1,332€ (B.M.P.
Pharma Trading AG in 2014) to 46,324,000,000€ maximum (Bayer AG in 2016). This high variance is not surprising as the development of new chemicals often requires firms to make high investments over years until a product is fully developed and can be placed on the market to yield a return. Therefore, it is possible that revenues are small over the years and R&D expenses high. The positive median value of Firm Performance (0.013) measured through ROA shows that more than 50% of the companies in the sample are profitable.
N Mean Std. Dev. Min Max Median Std. Error
R&D Intensity 285 36,095.02 5,3667.2 0 288,135.6 14,626.82 3,178.971
Revenue 285 3.43e+09 8.57e+09 1,332 4.68e+10 2.78e+08 5.07e+08
Revenue t-1 285 3.16e+09 8.14e+09 500 4.63e+10 2.43e+08 4.82e+08
Firm Performance 285 -.0577168 .5249394 -7.036518 1.632294 .0133786 .0310947
CEO Nationality 285 8.975439 2.499879 1 13 10 .14808
CEO Age 285 54.04912 6.849826 31 76 54 .4057487
CEO Power 285 1.98e-16 2.172719 -3.022421 7.93524 -.4792027 .1287008
IE Total 285 2.003509 1.893586 0 8.333333 2 .1121664
IE1 Int. Work 285 1.225731 .987857 0 3 1.333333 .0585156
IE2 Int. Education 285 .3847953 .7145939 0 3 0 .0423289
IE3 Foreign Nationality 285 .3929825 .8092467 0 3 0 .2986282
The independent variable IE Total ranges from 0 minimum to 8.33 maximum. This shows that some CEOs have gained no IE at all (n=22), whereas few others gained very high levels of IE (n=4 for IE Total>6) during their life so far, either through prior employments abroad, prior studies abroad, or through their foreign nationality. All three elements of IE, i. e. IE1 Int. Work,
IE2 Int. Education and IE3 Foreign Nationality, range from 0 minimum to 3 maximum. Most
CEOs gained IE through employments abroad (1.23), followed by their foreign nationality (0.39) and international education (0.38).
The CEOs in the observed sample have 13 different nationalities. The average CEO is German (n=59) and 54 years old. In 2012, the youngest CEO was 31 and the oldest CEO 72 years old. All together the CEOs worked in 25 and studied in 15 different foreign countries. Most CEOs hold a PhD degree (n=37).
4.2. Correlations
The correlations of the variables are depicted in Table 2. Values above 0.8 are an indicator for multicollinearity between variables. Values above 0.9 are considered as a problematic level of correlation (Tabachnick and Fidell, 1996). Table 2 shows a very high correlation of 0.9784 between the control variables Revenue and Revenue t-1. This is not surprising as the revenue of the previous year is obviously related to the revenue of the following year. Due to the very high correlation it was considered to exclude the variable Revenue t-1 from the analysis. When included into the models, the variable Revenue t-1 showed basically no effect (β=3.16e-07; p=0.61). Therefore, and because of the possible threat of multicollinearity the variable Revenue
t-1 was excluded from all further analyses. Another high correlation of 0.8188 exists between IE Total and IE1 Int. Work. That is also not surprising as the variable IE Total is a composite
of IE1 Int. Work and two other variables. This correlation is not problematic because IE Total and IE Int. Work do not enter the same model.
Table 2 – Correlation Matrix
4.3. Regression Results and Hypotheses Testing
The GLS analysis is performed in a step-wise manner to test the hypotheses. In total 13 models are used. The Log Likelihood (LL) ratio of each model is compared with each other to assess improvements between the models. The closer this ratio converges to zero, the higher is the goodness of fit of the model. The results of all models are presented in Table 3.
In Model 1 (LL=-3271.5404) the effect of the control variables on R&D Intensity is tested. As discussed in the previous section, the control variable Revenue t-1 is excluded due to high levels of correlation. The control variable Revenue (β=-694e-07, p=0.260) shows basically no effect and therefore is dropped from further models. The control variable Firm Performance (β=-1536.099; p=0.474) shows a negative, but not significant effect. The coefficient of the control variable Firm Performance remains negative and insignificant in all further models. CEO
Nationality (β=1713.68; p=0.091) and CEO Age (β=1404.63; p<0.01) both show positive and
highly significant results.
In Model 2 (LL=-3272.1368) the independent variable IE Total is added to the three remaining control variables to test hypothesis H1. Hypothesis H1 predicts a positive relationship between the IE of CEOs and the R&D Intensity of their firms. In Model 2 the control variable CEO
Nationality (β=1679.771) remains significant at the 10% level and the control variable CEO Age (β=1341.111) at the 1% level. The control variable Firm Performance is still not significant
(β=-1523.908; p=0.481). The independent variable IE Total shows positive, but insignificant results (β=337.2098; p=0.766). Therefore, hypothesis H1 is not supported.
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (1) R&D Intensity 1 (2) Revenue .0265 1 (3) Revenue t-1 .0304 .9784** 1 (4) Firm Performance -.1672** .1949** .1950** 1 (5) CEO Nationality .0608 -.0695 -.0627 -.0150 1 (6) CEO Age -.0646 .1267* .1207* .0841 -.2136** 1 (7) CEO Power .0460 .2542** .2467** .1420* .0793 .3722** 1 (8) IE Total .1121 .2543** .2287** .0439 -.4193** .0662 -.0899 1
(9) IE1 Int. Work .0870 .3066** .2885** 1296* -.2815** .1028 -.0593 .8188** 1
(10) IE2 Int. Education .0355 -.0324 -.0572 -.0421 -.0611 -.0331 -.1252* .6351** .2584** 1
In Model 3, Model 4 and Model 5 the independent variables IE1 Int. Work, IE2 Int. Education, and IE3 Foreign Nationality are added separately to test hypotheses H1a, H1b and H1c respectively. In Model 3 (LL=-3272.1806), the independent variable IE1 Int. Work (β=-78.39011; p=0.977) is added. Only the control variable CEO Age (β=1330.82; p<0.01) shows significant results. The same results appear for Model 4 (LL=-3272.0805) which includes the independent variable IE2 Int. Education (β=1270.695; p=0.654). The only significant variable is CEO Age (β=1358.416; p<0.01). Model 5 (LL=-3272.1303) includes the independent variable IE3 Foreign Nationality (β=880.8727; p=0.32). In this model CEO Age (β=1330.878; p<0.01) as well as CEO Nationality (β=1740.597; p=0.09) show significant results. All separately tested, independent variables in Model 3, Model 4 and Model 5 show insignificant negative (IE1 Int. Work) or insignificant positive (IE2 Int. Education, IE3 Foreign Nationality) results. Therefore, hypotheses H1a, H1b and H1c are not supported.
In Model 6 (LL=-3272.0247) the three aforementioned independent variables are added to the model in combination. Model 6 confirms the insignificant results of Model 3, Model 4 and
Model 5. When tested in combination, IE1 Int. Work (β=-1029.793; p=0.746), IE2 Int. Education (β=1606.097; p=0.667), and IE3 Foreign Nationality (β=410.2968; p=0.903) all
show insignificant results.
Model 7 (LL=-3264.8037) includes the control variables and adds the moderator variable CEO Power. The results of Model 7 show that CEO Age (β=2491.944), CEO Nationality (β=3180.75)
as well as the moderator variable CEO Power (β=-5263.539) are highly significant (p<0.01). Surprisingly, Model 7 reveals that CEO Power has a negative impact on R&D Intensity. In Model 8 (LL=-3264.7672) the control variables, the moderator variable CEO Power and the independent variable IE Total are tested. The results are similar to the results of previous models. CEO Age (β=2495.505), CEO Nationality (β=3183.079) and CEO Power 5317.477) are highly significant (p<0.01). Firm Performance 1448.159) and IE Total (β=-298.1482) still show insignificant results. In contrast to Model 2, IE Total now appears to have a negative impact on the R&D Intensity of firms.
Model 9 (LL=-3263.9513) considers the interaction effects between CEO Power and IE Total
Power seems to not moderate the relationship between the IE of CEOs and the innovativeness
of their firms. Therefore, hypotheses H2, H2a and H2b are not supported.
In Model 10, Model 11 and Model 12 the independent variables IE1 Int. Work, IE2 Int.
Education, and IE3 Foreign Nationality and their interaction effects are added separately. In Model 10 (LL=-3263.468), the independent variable IE1 Int. Work (β=-1060.212; p=0.722) is
included. In line with most other models, the control variables CEO Age (β=2333.863; p<0.01),
CEO Nationality (β=3131.171; p<0.01) and the moderator variable CEO Power (β=-4924.471;
p<0.01) show highly significant results. The same results occur for Model 11 (LL=-3263.7765) which includes the independent variable IE2 Int. Education (β=3573.481; p=0.334). In this model CEO Age (β=2405.808), CEO Nationality (β=3230.112) and CEO Power (β=-4815.052) are highly significant at the 1% level. Model 12 (LL=-3264.5817) adds the independent variable
IE3 Foreign Nationality (β=2214.672; p=0.574) and also shows high significance for CEO Age
(β=2435.087; p<0.01), CEO Nationality (β=3251.945; p<0.01) and CEO Power (β=-5081.239; p<0.01). All separately tested independent variables in Model 10, Model 11 and Model 12 show insignificant negative (IE1 Int. Work) or insignificant positive (IE2 Int. Education, IE3 Foreign
Nationality) results. The respective interaction effects with CEO Power show insignificant
results as well. Model 10, Model 11 and Model 12 thereby support the findings of Model 3,
Model 4, Model 5 and Model 9: IE of CEOs is not related to R&D Intensity and CEO Power
does not have a moderating effect.
Model 13 finally includes all variables and interaction terms. The LL value of Model 13
(LL=-3262.4778) is closest to zero. Therefore, Model 13 is the primary model with the highest goodness of fit to test the hypotheses. This model includes all control variables and all independent variables and their interaction effects with the moderator variable separately. All prior outcomes are confirmed in Model 13. The control variables CEO Age (β=2336.036; p<0.01) and CEO Nationality (β=3451.615; p<0.01) show positive and significant, and the moderator variable CEO Power (β=-4827.946; p<0.01) negative and significant results. The independent variables IE2 Int. Education (β=3204.25; p=0.439), IE3 Foreign Nationality (β=2382.444; p=0.545) and the interactions effects between CEO Power and IE1 Int. Work (β=1416.592; p=0.327) and IE2 Int. Education (β=1645.427; p=0.373) show positive and insignificant results. The control variable Firm Performance (β=-1135.102; p=0.585), IE1 Int.
Table 3 – Regression Results and Hypotheses Testing
Model (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
Dependent Variable R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity R&D Intensity
Control Variables
Revenue -6.94e-07
(6.16e-07) excl. excl. excl. excl. excl. excl. excl. excl. excl. excl. excl. excl.
Revenue t-1 excl. excl. excl. excl. excl. excl. excl. excl. excl. excl. excl. excl. excl.
Firm Performance -1536.099 (2144.743) -1523.908 (2163.587) -1577.054 (2156.04) -1496.749 (2162.286) -1498.251 (2170.68) -1443.768 (2171.75) -1402.143 (2063.016) -1448.159 (2068.826) -1439.572 (2077.318) -1419.802 (2062.338) -1406.016 (2070.583) -1303.552 (2080.571) -1135.102 (2078.2) CEO Nationality 1713.68 * (1015.011) 1679.771* (1018.84) 1663.207 (1017.696) 1628.184 (1019.716) 1740.597* (1046.436) 1666.286 (1077.339) 3180.75*** (1058.629) 3183.079*** (1058.583) 3075.225*** (1059.38) 3131.171*** (1073.42) 3230.112*** (1062.436) 3251.945*** (1083.46) 3451.615*** (1142.664) CEO Age 1404.63 *** (421.4575) 1341.111*** (418.8754) 1330.82*** (418.1988) 1358.416*** (421.8015) 1330.878*** (417.5985) 1356.994*** (423.8907) 2491.944*** (502.3472) 2495.505*** (502.3972) 2299.1*** (523.4711) 2333.863*** (530.1799) 2405.808*** (504.5224) 2435.087*** (511.8869) 2336.036 *** (529.6831) Independent Variables IE Total 337.2098 (1133.549) -298.1482 (1102.449) 932.5158 (1462.22)
IE1 Int. Work -78.39011
(2683.292) -1029.793 (3176.898) -1060.212 (2977.27) -2195.815 (3595.888)
IE2 Int. Education 1270.695
(2832.687) 1606.097 (3738.172) 3573.481 (3695.42) 3204.25 (4141.793)
IE3 Foreign Nationality 880.8727
(2766.488) 410.2968 (3363.202) 2214.672 (3483.075) 2382.444 (3936.835) Moderator Variable CEO Power -5263.539 *** (1342.838) -5317.477*** (1357.113) -4594.193*** (1467.986) -4924.471*** (1482.768) -4815.052*** (1385.078) -5081.239*** (1305.691) -4827.946*** (1495.295) Interactions
IE Total x CEO Power 766.4752
(600.7855) IE1 Int. Work x
CEO Power
1550.535 (1220.846)
1416.592 (1446.715) IE2 Int. Education x
CEO Power
2322.448 (1621.709)
1645.427 (1848.471) IE3 Foreign Nationality x
Based on these results all hypotheses are rejected. IE of CEOs does not drive innovativeness of firms (H1, H1a, H1b, H1c). Furthermore, CEO Power does not have a moderating effect on the relationship between the IE of CEOs and the R&D Intensity of their firms (H2, H2a, H2b). These results and possible explanations are discussed in the following chapter.
5.
Discussion
The review of existing research about firm innovation and the influence of CEOs on organisational outcomes suggested that IE of CEOs are likely to influence the R&D intensity of firms. Based on literature, I argued that IE enhances the knowledge base and skill portfolio and increases the risk-tolerance and openness of CEOs. Both an enhanced knowledge base and skill portfolio as well as an increased risk-tolerance and openness was assumed to facilitate risky R&D investments and thereby the R&D intensity of firms. The empirical results presented in the previous chapter do not support this assumption.
Independent Variable: IE of CEOs
The empirical results suggest that a higher level of IE of CEOs does not necessarily lead to higher levels of R&D intensity of firms. Even when testing each source of IE, i. e. international work experience, international education experience, and foreign nationality separately, the empirical results rebut the key assumption of this thesis.
There are several possible explanations for these opposing results. The first reason relates to the novel measurement of IE. This thesis developed a new, more comprehensive approach that integrates international work experience, international education experience and the foreign nationality of CEOs into one index to measure and assess their overall IE. Prior studies that examined the influence of CEOs’ IE on organisational outcomes primarily focused on the (a) depth and/or (b) breadth of CEOs’ IE (e. g. Herrmann and Datta, 2005, 2006; Laufs et al., 2016; McDougall et al., 2003; Musteen et al., 2010; Reuber and Fischer, 1997), but often neglected the (c) degree of IE and the foreign nationality of CEOs. Thus, the empirical results that suggest that IE of CEOs do not influence the R&D intensity of firms might be caused by the novel approach of this thesis of measuring the IE of CEOs.
