The influence of CEO characteristics on
firm-level innovation in the manufacturing
industry
By
Dirk Jacob van der Zee (S200360555, S3189511) Dissertation supervisors:
Dr. R.W. de Vries Dr. Q. Yu
DDM - Advanced International Business Management and Marketing Rijksuniversiteit Groningen
Newcastle University Business School
ABSTRACT
In the last 2 decades, the relationship between CEOs and innovation has been left under
examined. Many people argue that this implies that the findings of previous research still hold
true today and that it is not worth researching anymore. However, much has changed in this
business world in the last 20 years, especially the manufacturing industry has undergone
significant changes as the result of the technological revolution. Therefore, in this study we will
re-examine and add to the previous models in order to test the relationship between CEO
characteristics and innovation. Using a sample of 1.832 publicly traded companies operating in
all areas of manufacturing, we found that CEO characteristics explain approximately 22% of the
variance in relative R&D expenditures per employee between firms when controlling for firm
and ownership characteristics. In terms of specific CEO characteristics, we found that financial
investment in R&D is greater at firms which have CEOs that are younger, have lower tenure, are female and originate from countries with a high score long-term orientation and a low score on power distance. In addition, the results indicate that the level of education of the CEO does not influence innovation , but the type of education does. Finally, a U-shaped relationship was found
between R&D expenditures and CEO ownership. Based upon the findings in this paper, we
provide suggestions for managers and companies how they can implement the results of this
ACKNOWLEDGEMENTS
I would like to dedicate this thesis to my grandfather, Dirk ‘Dicky’ van der Zee, who sadly
passed away during the process of writing this thesis, may your soul rest in peace.
I also want to offer this endeavour to our God, the Almighty for the wisdom he bestowed upon
me, the strength, the peace of mind and good health in order to finish this research during this pandemic.
I would like to express my deep and sincere gratitude to my research supervisors, Dr. R. W. de
Vries, University of Groningen and Dr. Q. Yu, Newcastle University Business School, for giving
me the opportunity to do research and providing guidance throughout this research.
I am extremely grateful to my parents and grandparents for their love, prayers, caring and
TABLE OF CONTENTS
INTRODUCTION
4
THEORETICAL BACKGROUND AND HYPOTHESIS
8
Existing research 9 Hypothesis 14 METHODOLOGY 25 Dependent variable 25 Independent variable 26 Control variables 27
ANALYSIS
32
DISCUSSION
40
Managerial implications 47CONCLUSION
51
Limitations and future research 53
INTRODUCTION
In today's business landscape, companies need to excel in many different facets in order to gain a competitive advantage, or at least to stay in business. One of these facets, by some authors argued as the most crucial, is innovation (Tohidi & Jabbari 2012; Wellener, Umbenhauer, Dollar, Zale & Ashton 2020; Zamiatina 2019). Innovation is of vital importance because it forces companies to develop new or enhanced products and it gives companies the ability to upgrade
their processes or create completely new, more effective ones and it forces companies to
modernize their technologies. These benefits from innovation are, more often than not, drives for
a company to maintain, or create a competitive advantage (Ettlie 1998; Scherer 1984). One
industry in which innovation is central to a company's success, now more than ever, is
manufacturing (Gurski 2019). The main reason being, the technological revolution, which
brought a world in which technologies are constantly evolving caused by the ‘new’ digital
platforms available to humans. (Gurski 2019; Wellener et al. 2020). Wellener et al. (2020)
believe that the manufacturing industry will be impacted by five disruptive factors. These are,
economic patterns, trade dynamism, digitalization, talent/future work and electrification. These
changes will force companies to constantly innovate, to try and keep with their competition
(Gurski 2019; Tohidi & Jabbari 2012; Wellener et al. 2020). Zamiatina (2019) states:
Research has shown that many variables influence the amount and type of innovation within a firm. These variables include, but are not limited to, the firm’s industry (e.g. Pellegrino and Piva
2020; Scherer 1984), firm’s strategy (e.g. Asensio-López, Cabeza-García & González-Álvarez
2019; Baysinger & Hoskisson 1989; Baysinger et al. 1991; Doğan 2017; Karlsson & Tavassoli
2015) and institutional shareholders (e.g. Asensio-López, Cabeza-García & González-Álvarez
2019; Baysinger et al. 1991; Graves 1988; Hansen & Hill 1991). Nonetheless, despite the
extensive amount of research on the topic of innovation, one factor has been left under examined for the last 20 years. Namely the role of the Chief Executive Officer (CEO). This relationship has been studied, to some extent, between the late 80’s to early 2000’s, by authors such as Chaganti
and Sambharya (1987), Thomas et al. (1991), Rajagopalan and Datta (1996) and Barker and
Mueller (2002).
The reason for this focus by academics on the relationship between CEO and innovation
is rooted in the Upper Echelon theory (UE). The Upper Echelon theory is a management theory
first published by Hambrick and Mason in 1984. It states that CEOs and other top managers, who are responsible for strategic formulation and enactment, when interpreting possible strategic
situations or options are influenced by their own experiences, values, personality and other
the emerging psychological and cognitive moderators of UE variables are presently reinforcing the centrality of dominant coalitions, in that they affect their decision-making processes and strategic choices
As mentioned before there have been studies which tried to determine the influence of
CEO characteristics. However, in the last two decades, no new studies have been written on the
subject. Some people argue that this implies that researchers believe that the findings of these
older studies still hold true today (Whitehead 2018). However, as mentioned before, the
manufacturing industry has undergone significant changes in the last two decades, therefore
assuming that the old findings still hold true today might be wrong to some extent. Moreover,
since the manufacturing industry nowadays is very competitive, small mistakes might force
companies to go out of business. The most respected study on the relationship between CEO
characteristics and innovation is the one carried out by Barker and Mueller (2002). Barker and
Mueller (2002) created a very extensive model to test the relationship. This model included age,
stock ownership, career experience, education and tenure as the independent variables and
included corporate strategy, ownership structure, and other firm-level attributes as control
variables. The authors used a dataset consisting of publicly traded firms spanding many
industries. The main contribution of the paper to the literature is that it is the most
comprehensive study done, at that point in time, on whether CEOs matter in determining
important organizational outcomes. This study, by Barker and Mueller, will be the foundation for
been done by other authors (e.g Barker & Mueller 2002; Chaganti & Sambharya 1987), creating
more accurate results, compared to multi-industry studies. In addition, new variables will be
introduced in the model to make it more complete compared to older studies. These variables are nationality and gender. Furthermore, by re-examining the relationship, this study will provide proof if the researchers were correct or not in assuming that little has changed in the last 20 years regarding the influence of CEO characteristics on innovation.
As developed above, the main proposition of the paper is as follows,
Main Proposition. The relative level of innovation of a firm in the manufacturing industry is significantly influenced by characteristics of the CEO, or top management in the absence of a CEO, controlling for firm-level factors.
Based upon the main proposition the following research question is developed to explore the
relation between CEO characteristics and innovation,
Research Question: What is the influence of CEO characteristics on the level of
innovation in a firm operating in the manufacturing industry?
The remainder of this paper will proceed as follows. First, the relevant literature will be
reviewed. Next the hypotheses and the conceptual framework will be developed. Afterwards we
THEORETICAL BACKGROUND AND HYPOTHESIS
The main proposition of this paper comes with three conditions, which were also proposed by
Barker and Mueller (2002). The first condition is that CEOs have the discretion to determine or
at least influence the financial investment in research and development (R&D), which is the
variable used in this paper to determine innovation. Mansfield (1968) describes the decision on how much money to spend on R&D as a risky decision which often leads to failure. Therefore he
expects that CEOs monitor financial investment very closely. The second condition is that CEOs
have the highest power and the final decision when it comes to decision making. This was
confirmed by Zahra and Pearce (1989) who mention that CEOs more often than not have the
central power within a firm. Moreover, they have the ability to influence the top management
composition which in turn has influence on R&D spending. The third condition revolves around
the accuracy of the Upper Echelon theory. In order for this research to be valid and provide
results which are an accurate depiction of real life business situations, we need the CEO’s
decisions to be influenced by his or her own age, education, tenure, gender, national background
and other factors, which is what the Upper Echelon theory states (Barker & Mueller 2002).
Based upon these conditions being fulfilled, we can confidently assume that the results we find have valuable implications for the ‘real world’.
Existing research
Cabeza-García & González-Álvarez 2019; Choi & Lim 2017; Doğan 2017; Heimonen 2012; Karlsson & Tavassoli 2015; Lööf & Heshmati 2002; Pellegrino and Piva 2020). While there are
some studies which look at the role of the CEO or top level management on the amount of
financial investment into R&D, there have been no new studies in the last two decades.
Nonetheless, in the last 20 years there have been studies on the relationship between CEO
characteristics and firm performance. Diks (2016) who conducted research on 483 CEOs of the
S&P 500 firms, found that CEO age negatively influenced firm performance. In addition, he
found that both CEO tenure and CEO stock compensation positively relates to firm performance.
In contrast to these findings are those of Liu and Jiang (2020), who conducted research on a
sample consisting of 10.446 observations from Chinese companies taken during the period 2008
until 2016. They concluded that CEO age had no apparent impact on firm performance. In their
paper they also stated that their findings contradicted previous research. Moreover, also
contradicting other studies, Liu and Jiang found that CEO tenure had a negative impact on
performance (Liu and Jiang 2020). A similar study to that of Liu and Jiang was carried out by Naseem, Lin, Rehman, Ahmad and Ali (2019) who sampled 179 firms from Pakistan between 2009 and 2015. Their results showed that CEO duality negatively influenced overall firm
performance and that CEO duality increased the likelihood of debt financing. In addition, CEOs
with longer tenure are more likely to act opportunistically, by putting their own interest above
those of the firm, creating agency costs for the company. Furthermore, they found that age,
gender and education significantly influenced the CEO’s decision making and consequently the performance of the firm. While both the study of Liu and Jiang (2020) and that of Naseem et al.
of the samples. Another study on the relationship between firm performance and CEO characteristics by Sani (2019) found that CEO education has a positive relation with profitability. Similarly, if the CEO has prior experience in the firm, the profitability will increase when he or
she becomes CEO. These studies, containing data from the last 20 years, are proof of the
relationship between CEO characteristics and firm performance. However, because of the vast
difference between firm performance and innovation, it is highly unlikely the results will be
transferable, nonetheless, it does provide a starting point for determining which factors could
influence the level of innovation in a firm.
Furthermore, a related area which has seen research in the last two decades is that of
factors influencing innovation. A very recent study performed by Pellegrino an Piva (2020)
concluded that both the characteristics of the industry and the period of time the company has
existed significantly influences the amount and type of innovation within the firm. Their results
suggest that relatively young firms who conduct business in industries which are primarily
classified as ‘entrepreneurial’ are more likely to translate R&D investments into innovation
compared to older firms. However, more mature firms have the upper hand in more ‘routinized
sectores’, such as low-tech manufacturing, in which cost reduction is the main purpose of the innovation (Pellegrino & Piva 2020).
When it comes to financial variables and innovation many relationships appear to be present.
The most prominent ones being market value, sales and capital expenditures (Harmantzis &
Tanguturi 2005). A study conducted on 200 firms, operating in the telecommunications
sales positively influenced the R&D expenditure at the end of the year. Cash reserves was the only financial variable tested that did not significantly influence financial investment into R&D
(Harmantzis & Tanguturi 2005). These results are supported by Bellalah (2003) who found that
financial performance positively influences R&D spending. As previously stated, many studies
have tried to determine the influence of different forms of strategy on innovation (e.g.
Asensio-López, Cabeza-García & González-Álvarez 2019; Baysinger & Hoskisson 1989;
Baysinger et al. 1991; Doğan 2017; Karlsson & Tavassoli 2015). Dogan (2017) argues that firms
who have a strategy which does not purely focus on the product and the process, but goes beyond
by creating a systematic and holistic approach which integrates innovation in the strategy by
aligning the companies missions, objectives and the organizational culture, in order to allow for
optimal implementation of the innovation focused strategy (Dogan 2017). This argumentation is
supported by many other authors, such as Baysinger et al. (1991) and Karlsson and Tavassoli
(2015).
When we try to find studies which closely represent the one proposed in this paper, it
becomes evident that only a few of these studies exist. One such study was performed by
Chaganti and Sambharya (1987). They tried to determine the relationship between the top
management team and the amount of money spent on innovation by examining three firms in the
tobacco sector. They found that managers with lower tenure and backgrounds in production are
more likely to pursue a strategy which focuses on innovation. However, the limited scope of the
study makes the findings not generalizable. In addition, the study was carried out over 35 years
ago, in the time that has passed much has changed in the business world, further reducing the
found that CEOs with higher education are more likely to initiate innovative projects (Thomas et al. 1991). Once again, because of the limited scope of the study, the results can not be applied to the manufacturing industry. Rajagopolan and Datta (1996) argue that CEO characteristics vary
greatly between industries and which characteristic is the most dominant influence when they
make decisions.
One of the most recent studies on CEO characteristics and innovation was conducted
using a sample of French firms listed on the Euronext Paris. The study was conducted by
Mezghanni (2010) and contained 103 firms operating mainly in France. The analysis resulted in
an inverted U-shape relation between R&D expenditure and CEO age as well as tenure. The
author believes this implies, ‘’the existence of a critical CEO age and a critical point in time over CEO tenure at a firm before which CEO increases the amount spent in R&D activities and after which CEO begins to exhibit investment myopia by gradually reducing the amount spent in R&D activities. ‘’ (Mezghanni 2010, p. 1). Moreover, a U-shaped relationship was found with
CEO ownership, meaning that R&D investment is negatively (positively) associated with CEO
ownership at low (high) levels of CEO stock ownership. However, as with other studies
mentioned in this paper, the study by Mezghanni is very limited in its generalizability because of the choice of sample.
The benchmark study for the relation between CEO characteristics and innovation is the
one created by Barker and Mueller (2002), as mentioned before, that study is also used as a
foundation for this paper. Barker and Mueller analyzed the relation between CEO characteristics
and R&D spending. They researched this by using 172 firms from the 1989 and 1990 Business
account for the industry differences they divided the amount of money spent on R&D by the
industry average. Their results showed that some individual CEO characteristics have a
significant influence on R&D spending. First, younger CEOs are more willing to invest money
into R&D. Second, CEOs that have more of their wealth invested into company stock are more
likely to financially invest into R&D. Finally, CEOs with experience in the field of marketing
and or engineering prefer investing into R&D compared to CEOs with alternate experience.
Furthermore, contradicting existing literature, they concluded that the amount of education the
CEO underwent did not significantly affect the financial investment into R&D (Barker and Mueller 2002). Moreover, after they ran their initial analysis, the authors created subgroups for the variable tenure, these groups were 1-3 years, 1-4 years and 8+ years. When analyzing the data using these subgroups they found that the CEO effects on R&D spending increased with
longer tenure. The authors argued that this implied that CEOs may mold R&D spending to their
own preferences as their time with the company increases (Barker & Mueller 2002). One of the strengths of this research is that it can be generalized to an extent, since the authors included firms from all different industries in their sample. However, in spite of the fact that the authors used firms from many different industries, it might be possible that the results do not accurately
represent all industries, there may be some exceptions. The reason being that all the industries
the results might differ for the manufacturing industry. In addition, the data used in the study is over 30 years old (1989-1990), which allows for questioning the accuracy of the findings in this new day and age. Van Paasschen (2017) argues that the business landscape has changed drastically in the last few decades. He states: (p.1)
Virtually everything about the way people live and do business is changing faster than ever before. Digital technology, global development, urbanization, and business disruption represent both a major opportunity and a threat in the global economy
Based upon all the studies and theories mentioned in this section, the following hypotheses were
created.
Hypothesis CEO age
The first, probably most often research characteristic, is the age of the CEO. Many
authors have argued that older CEOs lack the stamina, both physical and mental to adapt to new
ideas and learn how to behave and operate differently, which is vital for innovation (Barker &
Mueller 2002; Diks 2016). Moreover, a study in the field of psychology performed by Frey,
Mata and Hertwig (2015) found that both younger and older people use relatively simple yet
successful strategies to learn, which remain unaffected by the cognitive decline which older
people experience. However, when the situation becomes more complex, older participants were
situations, which occur during the entire process of innovation, older CEOs will have less
cognitive ability to understand the innovation process compared to their younger counterparts.
This gap in abilities between young and old might lead to older CEOs being more hesitant to
engage in innovation.
Furthermore, Hambrick and Mason (1984) argue that the older the CEO becomes, the less likely he or she will be to propose new ideas which go against the status-quo. The reason
being that they want to keep things the way they are, which makes them feel more comfortable.
In addition, MacCrimmon and Wehrung (1986) believe that the older the CEO becomes the more risk averse he or she becomes. Meaning, they will be less inclined to make a decision which contains a significant amount of risk, even though it might become very profitable, which is most often the case for innovation. Moreover, the technological revolution has made the
manufacturing industry even more complex, making it even harder for older CEOs to overcome
their cognitive decline. Therefore, the following hypothesis is constructed:
Hypothesis 1: CEO age is negatively associated with firm-level innovation in the
manufacturing industry
CEO tenure
The second CEO characteristic which will be researched in this paper is CEO tenure.
When it comes to the relation between CEO tenure and the level of innovation the literate is divided. Miller (1991) argues that the longer the tenure of the CEO, the less likely he or she is to
make risky investments or change the environment of the firm. The author believes that long
consistency over constantly evolving the firm, which is what is required for firms to be innovative (Barker & Mueller 2002). In addition, Li and Yang (2019) found that when a CEO’s tenure advances, the amount of money spent on exploitative innovation increases relative to the
amount spent on exploratory innovation. Since exploitative innovation, in most scenarios,
requires less investment and risk compared to exploratory innovation, causing the financial
investment in innovation to decrease if CEO tenure advances. Furthermore, according to
Mezghanni (2010), CEOs that have a short time left at the firm (high tenure) are more risk averse
in their decision making. The reasoning behind this is that they ‘’ prefer investments with shorter
time horizons where cash flows are more predictable in order to enhance their own wealth‘’
(Mezghanni 2010, p. 22).
In contrast, Adams, Almeida, and Ferreira (2005) argue that CEOs with longer tenure
have more power compared to CEOs with lower tenure, causing them to prefer riskier
investments, which could result in higher payoffs. In order to achieve this, they would need to invest more money in R&D to increase the chance of creating a new invention, thus increasing the level of innovation at the firm. Moreover, Hirshleifer (1993) states that CEOs with relatively short tenure are less likely to invest a lot of money in R&D. The reason for this being the fact that they want to build their reputation and therefore do not want to invest a lot of money in risky projects. They prefer short term profits over long term success in the beginning of their career as a CEO, which hampers innovation.
Thus, it is clear that there is no real consensus in the existing literature on the relationship
between CEO tenure and innovation will be tested in the paper, leading to the following hypotheses:
Hypothesis 2a: CEO tenure is negatively associated with firm-level innovation in the
manufacturing industry
Hypothesis 2b: CEO tenure is positively associated with firm-level innovation in the
manufacturing industry
CEO education
The third CEO characteristics discussed in existing literature is CEO education.
Innovations are usually accompanied with newly created products or enhanced technologies,
which can either be completely new or upgraded versions of ones already present within the
firms. Since innovation is a complex process, which requires great cognitive abilities, authors
argue that CEOs which have enjoyed higher education are more likely to see the potential of new
innovations and the mental capacity to understand what the added value of the invention will be.
Meaning they will be more likely to invest into R&D to create such inventions (Barker and
Mueller 2002). Furthermore, a study conducted by researchers from UC Berkeley found that
higher levels of education is associated with a decline in age-related cognitive decline. Meaning that CEOs with higher education will suffer less from cognitive decline when their age increases, allowing them to still see the potential of new innovations and understand the added value as they get older (Maclay 2017). Leading to the following hypothesis:
Furthermore, Barker and Mueller (2002) believe that the type of education also influences the relation between CEO characteristics and innovation. They argue that MBA programs mainly attract students who are more conservative and risk averse (Finkelstein and Hambrick 1996;
Hambrick and Mason 1984) In addition, in the MBA programs students are taught how to avoid
big mistakes and how to reduce the risk of an investment. All these aspects of MBAs and their
students seem to do little for innovation. Moreover, Barker and Mueller (2002) argue that the
same can be said for legal degrees. Since legal degrees place absolutely no emphasis on
innovation, legal students are taught how to apply the law and not think of inventive ways to
create something new or better. In contrast, Tyler and Steensma (1998) believe that CEOs who
have enjoyed a degree in science or engineering are more likely to be in favor of high levels of
innovation. The reason being that CEOs with such a degree have a more complete understanding
of technology and consequently innovation (Barker & Mueller 2002). Based upon these studies
and their findings the following hypothesis were created
Hypothesis 4: The type of education achieved by a CEO is associated with firm-level innovation in the manufacturing industry
To be specific:
Hypothesis 4a: Firm-level innovation in the manufacturing industry is negatively
associated with the number of business degrees earned by its CEO.
Hypothesis 4b: Firm-level innovation in the manufacturing industry is negatively
Hypothesis 4c: Firm level innovation in the manufacturing industry is positively associated with the number of science and engineering degrees earned by its CEO.
CEO ownership
The fourth CEO characteristics researched in this paper is CEO ownership. According to
economic theory, managers will change the financial investment in R&D based upon the amount
of company stocks he or she owns (Beyer, Czarnitzki & Kraft 2011). The rationale behind this theory is the following, innovative projects, in general, have a significant amount of risk attached to them. If a project fails, managers could face detrimental backlash on their career, which could
even lead to the manager being fired. Therefore, managers with low ownership will underinvest
in R&D to avoid the risks associated with it. On the other hand, Beyer et al. (2011) found that
when managers feel entrenched in the firm, meaning they have enough shares to feel that their
job is almost guaranteed, which removes the fear of losing their job if the R&D project fails, they will (over)invest in R&D in order to stimulate growth. These theories suggest that there is a positive linear relation present, however many researchers argue that the relation between these two variables is not linear.
Ghosh et al (2007) argue that there is an inverted U-shape relationship between CEO
ownership and R&D expenditure. To be more specific, they argue that at low stock ownership
(between 0 and 5%) CEOs are more willing to take the risk and invest in R&D projects.
However, when the CEOs ownership increases (between 5 and 25%) they become more hesitant
ownership over 25% the authors argue that no relationship is present anymore between R&D expenditures and CEO ownership (Ghosh et al 2007). In addition, the same type of relation was found by Cho (1998), who conducted research on the influence of insider ownership on financial investment into R&D. While he did also find an inverted U-shape relationship, his cutoff points
were different. He found that R&D expenditure had a positive relation with insider ownership
until 7%. However, between 7,1% and 38% the relation turns into a negative one. Finally, when
the insider ownership exceeds 38% no significant relationship appears to be present anymore. Another paper that confirms a non-linear relation between the two variables is the one by
Abdullah et al (2002). One big difference between this study and the studies mentioned above is
that Abdullah et al (2002) found a W-shaped relationship compared to the inverted U-shaped
relationship of Ghosh et al. (2007) and Cho (1998). To be more precise, R&D investment will
decrease as the ownership of managers increases between 0 and 5% and between 10 and 15%.
However, between 5 to 10 % and beyond 15% the investment into R&D will increase as
managerial ownership increases.
Contradicting all the studies above, Mezghanni (2010) found a significant U-shaped
relation between CEO ownership and R&D expenditures. They found a negative relationship
until the CEO ownership reaches a point of 45,2%. Afterwards an increase in the amount of
stock owned by the CEO positively influences the amount of financial investment into R&D.
According to the authors this means that agency theory is correct and that at a level of ownership
above 45,2 percent an increase will result in stronger incentives for CEOs to increase the
financial investment into R&D to ensure future growth. This usually aligness their interest with
being more hesitant to invest, because it makes managers myopic and more reluctant to invest in R&D projects that can reduce their profits significantly (Mezghanni 2010).
When looking at the research mentioned in this section it is clear that there is no real consensus on the relationship between CEO stock ownership and R&D expenditure. However,
the study of Mezghanni (2010) most accurately represents the study carried out in this paper,
therefore the hypothesis will be based upon the findings of that paper. This leads to the following hypotheses:
Hypothesis 5: CEO stock ownership has a u-shaped association with firm-level
innovation in the manufacturing industry
CEO nationality
The fifth characteristic which will be tested in this study is the nationality of the CEO.
CEO nationality will be measured by using the 6 Hofstede dimension of culture. Douglas and
Wildavsky (1982) discovered that countries who score high on the individualistic dimension of
Hofstede are more likely to take risks, these countries are, for example, the USA, Australia and
the Netherlands. In contrast, countries which score high on the power distance dimension,
meaning they embrace hierarchy instead of egalitarian structures, prefer more cautious operating
systems resulting in less risk taking and consequently less innovation (Kim & Park 2010; Li,
Griffin, Yue & Zhao 2013). In addition, countries that score high on uncertainty avoidance will
prefer more safe investments, which results in less risky, high rewarding innovation. The reason
for this is that these countries are less comfortable with the uncertainty that comes with R&D
R&D, CEOs that originate from countries with a high score on this dimension have a view that is
futuristic and long term oriented, which results in more investment into R&D projects which
usually are long-term investments (Kim & Park 2010; Hofstede, Hofstede & Minkov 2010; Li,
Griffin, Yue & Zhao 2013). Furthermore, the other two Hofstede dimensions, ‘Masculinity’ and
‘Indulgence’ seem to have no real correlation with R&D expenditures (Kim & Park 2010; Li, Griffin, Yue & Zhao 2013). Thus leading to the following hypotheses:
Hypothesis 6: The nationality of a CEO is associated with firm-level innovation in the manufacturing industry
To be specific:
Hypothesis 6a:CEO’s who originate from a country with a high score on individualism have a positive effect on the level of innovation in the manufacturing industry
Hypothesis 6b: CEO’s who originate from a country with a low score on power distance
have a positive effect on the level of innovation in the manufacturing industry
Hypothesis 6c: CEO’s who originate from a country with a low score on uncertainty
avoidance have a negative effect on the level of innovation in the manufacturing industry
Hypothesis 6d: CEO’s who originate from a country with a high score on long-term
orientation have a positive effect on the level of innovation in the manufacturing industry
CEO gender
One of the factors that was left out of the previous studies on the relation between CEO
underrepresentation of females at the top of the hierarchy, meaning boardrooms, top
management and CEOs. This underrepresentation caused the lack of research on how gender
influences variables such as firm performance and innovation (Cooney 2020). However, for the
past two decades, women have made inroads into this traditionally male dominated area. In
2019, over 29% of the senior management roles were occupied by females (Thornton 2020). A
study on female representation in top management found that female representation in top
management increases the performance of a firm, however, this is only true to the extent that a firm is more focused on innovation as a part of its strategy ( Dezső & Ross 2012). Furthermore, a study performed by Han, Cui, Chen & Fu (2019), showed that in Chinese firms, when compared to men, female CEOs have significantly promoted both incremental and radical innovation, both types of innovation are heavily present in the manufacturing industry (Yamamoto & Bellgran
2013). Based upon these studies it is reasonable to assume that female CEOs put more emphasis
on innovation compared to their male counterparts. Leading to the following hypothesis:
Hypothesis 7: Female CEOs have a positive effect on the level of innovation within a firm operating in the manufacturing industry.
METHODOLOGY
The sample which will be used in this paper is drawn from the Wharton Research Data Service
(WRDS) databases. To be sampled, firms need to have data available on all the different
variables (DV, IVs and CVs) and operate in the manufacturing industry. To determine in which
industry the firm operates in, the NAICS ( North American Industry Classification System) codes
are used. Moreover, in order to avoid reverse causality, all the independent and control variables will be lagged on year behind the DV. The data for the dependent variable is taken from 2019,
while the data for the independent and control variables are taken from 2018. The next part of the
paper will explain how each variable of the model is calculated. In addition, at the end of the section a table containing all the variables is displayed (table 1).
Dependent variable
The dependent variable (DV), which is used to measure innovation, is calculated by
dividing the R&D expense of a company by the number of employees the company has. This
method was also used by authors such as Baysinger et al. (1991), Hill and Snell (1989), Scherer
(1984) and Barker and Mueller (2002). These authors believe this method to be a more accurate
representation of firm innovation, compared to the alternative method of dividing R&D
expenditure by firm revenue. In addition, there are differences between areas of manufacturing,
therefore the R&D expense per employee will be divided by the average R&D expense per
employee of the area in which the company operates (Pellegrino and Piva 2020). The areas are
determined by using the companies NAICS codes. The reason for choosing this system over the
the private and public sector, the SIC codes are less standardized and mainly used by the public
sector. However, the NAICS codes are more accurate, used by governments and used for both
national and international purposes (Henneberry 2020). The companies used in this papers
dataset were grouped into areas by their first 4 digits of their code. For example, all companies that have an NAICS code that starts with 3111 are grouped in the area ‘animal food manufacturing’. Based upon this system, different areas in manufacturing were created.
Independent variables
CEO age is measured in years (Daellenbach et al. 1999). CEO tenure is the number of years the person in question is CEO of the given company (Barker & Mueller 2002; Mezghanni 2010). CEO education level is measured on a four-point scale , with 0 = no college degree, 1 = undergraduate degree, 2 = master's degree or JD, 3 = Ph.D. degree. The number of business degrees a CEO has will be used as the actual number in the regression. The same method is used for the science and/or engineering degree. When it comes to the legal degree, a dummy variable will be created with 0= no JD degree, 1=CEO obtained a JD degree (Barker & Mueller 2002). Furthermore, CEO ownership is measured by dividing the shares owned by the CEO by the total
shares outstanding, this method was used by Ghosh et al. (2007) and Mezghanni (2010).
Moreover, in order to test the inverted U-shape relation, both CEO ownership and its square are introduced in the model. CEO nationality is measured by the country in which the CEO is born or spent the majority of his or her childhood. Each of the four Hofstede dimensions
(individualism, power distance, uncertainty avoidance and long-term orientation) is a separate
from the Netherlands, would score 80 on the individualism variable, 38 on power distance, 53 on uncertainty avoidance and 67 on long-term orientation. CEO gender is measured with 0 being a female and 1 being a male.
Control variables
The main argumentation of this paper indicates that R&D spending is influenced by
personal characteristics of the CEO. However, in order to fully test this relationship, control
variables should be included to account for firm and ownership characteristics which influence
R&D spending. This is of vital importance because without the control variables, it is not obvious whether the change in R&D spending is related to the CEO or other factors, such as firm
size (Mezghanni 2010; Scherer 1984). Therefore, any study that wants to accurately examine the
relation between CEO characteristics and innovation should control for variables that previous authors have shown could influence the dependent variable. All the control variables (leverage, firm size, growth opportunities and past performance) are listed below, including the theoretical reason for inclusion and how the variable is measured.
Leverage
Barker and Muller (2002) argue that leverage makes managers more hesitant to invest in
risky long-term projects. The reason being that they prefer increasing current cash flow for debt
purposes. Long and Malitz (1985) mention that firms that want to borrow money will not invest
in R&D projects, because these projects create intangible assets, which are less conducive as
(1989), Bhagat and Welch (1995), Nam et al. (2003) and Chen and Hsu (2009). In this paper, leverage will be calculated by dividing total debt by total assets. This was also done by Mezghanni (2010) and Lee and O'Neill (2003).
Firm Size
Many authors have found a significant relation between R&D spending and firm size
(Baysinger and Hoskisson 1989; Mezghanni 2010; Scherer 1984). The rationale behind the
findings is that large firms have the resources to create and benefit from sustained R&D
programs and the innovation that comes with it ( Emodi, Murthy, Emodi and Emodi 2017;
Scherer 1984).
However, the variable which is used to measure firm size differs between papers.
According to Li and Dang (2015) many different proxies for firm size are used while the
majority of the authors using these measures do not consider how different the result can be
based upon the variable chosen. In their paper, Li and Dang (2015) investigated the influence of
using different variables to measure firm size, these being, total assets, total sales and market
capitalization, they used these 3 proxies because they are the most commonly used in research
papers. The findings of Li and Dang (2015) suggested that when it comes to R&D expenditure
both total assets and market cap have the same significance and R2 , meaning they are the most
Growth opportunities
Firms which are characterized by high growth opportunities earn a larger percent of their
profits from assets not yet owned compared to other firms (Lee and O'Neill 2003). This indicates
that those firms have more incentives to invest in R&D projects (Ghosh et al. 2007). Growth
opportunities will be measured by using the market-to-book ratio, calculating it as the market
value of equity at the end of a year plus the book value of debt divided by the book value of total assets, which was also done by Cho (1998), Lee and O'Neill (2003) and Mezghanni (2010).
Past performance
Another factor which influences the amount of money a company spends on R&D is their past performance. In a study done by Hundley at al. (1996) it was found that US firms were less likely to spend money on R&D if their profits were below zero. Moreover, a study performed by
Osma and Young (2009) found that firms who do not reach their targeted earnings performance,
will significantly cut the amount of financial investment into R&D for the next accounting
period. This suggests that CEOs feel less comfortable spending money on risky R&D projects if
Table 1: Variable overview
Variable Measurement Source of data
Dependent variable R&D expenditure, per employee
(R&D expenditure divided by the number of employees) / average R&D
expenditure per employee of
manufacturing area in which company operates
WRDS, Compustat - Capital IQ
Independent variables
CEO age CEO age in years WRDS, BoardEx
CEO tenure Number of years served as CEO WRDS, BoardEx
CEO education level Education level is measured on a four-point scale, with 0 = no college degree, 1 = undergraduate degree, 2 = master's degree or JD, 3 = Ph.D. degree.
WRDS, BoardEx and Linkedin
CEO type of education Type of education is measured using three different types of degrees. First, business degrees, which is measured as the actual number of degrees the CEO has. Second, science and/or engineering degrees are measured in the same way as business degrees. Third, legal degrees are measured using a dummy variable, with 0= no JD degree and 1=CEO obtained a JD degree.
WRDS, BoardEx and Linkedin
CEO ownership Measured by dividing the shares owned by the CEO by the total shares
outstanding
WRDS, BoardEx and WRDS, Compustat - Capital IQ
CEO nationality Nationality is measured with four variables, one for each Hofstede
distance, uncertainty avoidance and long-term orientation). The value of each variable is determined by the country's score on each dimension. For example, a American CEO would score,
individualism = 91, power distance = 40, uncertainty avoidance = 46 and
long-term orientation = 26
CEO gender Gender of CEO, 0= female, 1= male WRDS, BoardEx
Control variables
Leverage Measured by dividing total debt by total
assets, expressed in percentage
WRDS, Compustat - Capital IQ
Firm size Total assets WRDS, Compustat -
Capital IQ
Growth opportunity Measured using market-to-book ratio. This is calculated by adding the market value of equity at the end of a year and the book value of debt, then dividing it by the book value of total assets, expressed in percentage
WRDS, Compustat - Capital IQ
ANALYSIS
After the data was collected, the Post-Cook’s distance description test was run on all the variables to get a first impression of the data and to detect outliers which were removed from the dataset. The hypotheses were tested using an ordinary least squares regression analysis. The
means, standard deviation, minimum, maximum and median are presented in table 2. This study
covers 1.832 firms operating in the manufacturing industry. In terms of the dependent variable,
the average firm was spending 1,238 times (or 23,8% more) the amount of money on R&D per
employee compared to their area of manufacturing. Nonetheless, the standard deviation of 0,611 indicates that many of the firms in the sample were spending much more or less than their area’s average.
Moreover, with regards to the independent variables, the mean age of the CEOs is
approximately 47 years, with the minimum and maximum age being 27 and 87 respectively. The
CEOs, on average, served 10,5 years, with the longest tenure being 32 years and the shortest
being 1 year. The level of education (1,762) is somewhere between undergraduate (1) and
master's degree (2). The business, legal and science/engineering degrees means were 0,542, 0,09 and 0,611 respectively. The average ownership of the CEOs is 15,62% of their firm's total stock.
When it comes to nationality and its variables we can see that the average scores are,
individualism 61,55, power distance 41,10, uncertainty avoidance 57,90 and long-term
orientation 64,05. Out of the 1,832 CEOs, 1702 were male and 130 were females. Finally, the
An examination of the Pearson correlation matrix, displayed in table 3 and 4, suggests that there is no collinearity present between the independent variables, since all the correlation coefficients are below the typically used cutoff of 0,80 (Berry & Feldman 1985).
Table 2. Descriptive statistics
Note: N= 1.832
Variables Mean Standard Deviation
Minimum Maximum Median 1. R&D expenditure, per
employee (relative to area of manufacturing)
1,238 0,611 0.02 43,56 1,104
2. CEO age 49,65 7,908 27,00 87,00 51,00
3. CEO tenure 10,561 8,405 1,00 32,00 9,00
4. CEO education level 1,762 0.831 0,00 3,00 2,00
5. Number of business degrees 0,542 0,721 0,00 3,00 1,00
6. Legal degree 0,09 0,260 0,00 1,00 0,00
7. Number of science and/or engineering degrees
0,611 0,699 0,00 4,00 1,00
8. CEO ownership 15,62 20,56 0,00 81,1 10,50
9. Individualism 61,55 15,71 8 (Ecuador) 91 (USA) 54,00
10. Power distance 41,10 17,17 18 (Denmark) 100 (Malaysia) 48,00
11. Uncertainty avoidance 57,90 9,83 8 (Singapore) 87 (Peru) 63,00
12. Long-term orientation 64,05 18,76 14 (Morocco) 88 (Japan) 49,00
13. Gender 0,929 0,257 0,00 1,00 1,00
14. Leverage 16,8 10,1 0,1 41,9 28,6
15. Firm size 320.071.983,77 146.267.992,04 285.000 24.733.664.220 87.334.600
16. Growth opportunity 2,27 1,19 -17,00 19,00 1,82
Table 3: Pearson correlation coefficients
Note: Number of observations = 1.832. Correlation significance level: †p ≤ 0,10 ∗p ≤ 0,05 ∗∗p ≤ 0,01
Table 4: Pearson correlation coefficients continued
Note: Number of observations = 1.832. Correlation significance level:†p ≤ 0,10 ∗p ≤ 0,05 ∗∗p ≤ 0,01
To control for firm and ownership effects on R&D expenditure, the regression analysis was performed in a stepwise manner, presented below in table 5. Model 1 looks at the influence of
firm and ownership effects on R&D expenditure per employee and indicates that approximately
22% of the variance in the firm’s relative R&D expenditures per employee can be explained by the control variables used in the first model. To be more specific, firm size (p ≤ 0,001) and past performance (p ≤ 0,05) are associated with an increase in R&D expenditure per employee. This indicates that the more total assets a firm has, the more money they will spend on R&D, which
confirms the rationale that large firms have the resources to create and benefits from sustained
R&D programs (Emodi et al. 2017; Scherer 1984).
Table 5: Regression of firm and CEO characteristics on relative R&D expenditure per employee
Firm and Ownership
controls Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
Intercept 0,702 (0,239) (0,240) 0,783 (0,222) 0,771 (0,217) 0,679 (0,199) 0,656 (0,203) 0,698 (0,233) 0,777 (0,157) 0,566 Leverage -0,054† (0,031) -0,053† (0,036) -0,051† (0,037) (0,029) -0.048 -0,051† (0,033) (0,020) -0.049 (0,021) -0,048 (0,017) -0,038 Firm size 0,00000000332*** (0,00000000118) 0,00000000329*** (0,00000000106) 0,00000000327*** (0,00000000109) 0,00000000299*** (0,00000000103) 0,00000000337*** (0,00000000126) 0,00000000321*** (0,00000000114) 0,00000000311*** (0,00000000111) 0,00000000247*** (0,00000000093) Growth opportunity 0,0314 (0,0196) (0,0185) 0,0291 (0,0203) 0,0288 (0,0188) 0,0307 (0,0177) 0,0329 (0,0192) 0,0262 (0,0188) 0,0261 (0,0177) 0,0302 Past performance 0,0926* (0,0418) (0,0414) 0,0922* (0,0402) 0,0889* (0,0406) 0,0907* 0,0909* (0,0406) (0,0393) 0,0888* 0,0921* (0,0422) 0,0877† (0,502) CEO Characteristics (Hypothesized direction) CEO age (-) -0,0092*** (0,0044) -0,0085*** (0,0037) CEO tenure (-) -0,0085* (0,002) -0,0053† (0,0008)
CEO education level (+) 0,026
(0,005) (0,004) 0,023
Number of business
degrees (-) (0,011) 0,035 (0,011) 0,028
Legal degree (-) 0,043
(0,032) (0,029) 0,042
Note: †p ≤ 0,10 ∗p ≤ 0,05 ∗∗p ≤ 0,01 ∗∗∗p ≤ 0,001, one-tailed coefficient tests. N = 1.832. Coefficients are presented with standard errors in parentheses.
In addition, the positive relation between R&D and past performance might indicate that firms
that reach their targeted earnings performance (or atleast obtain some profits) are more likely to
financially invest into R&D (Hundley at al. 1996; Osma and Young 2009). On the other hand,
leverage (p ≤ 0,10) is associated with a decrease in R&D expenditure per employee, confirming that highly leveraged firms are more hesitant to invest into R&D, in order to increase current cash flow for debt purposes (Barker and Muller 2002; Hansen and Hill 1991).
In model 8 all the CEO characteristics are added in order to test the research question put forward in this paper. As indicated by the change in adjusted R-squared at the bottom of table 5,
CEO characteristics explain 19% of the variance in model 8. This number supports the main
proposition of this paper, namely, that the relative level amount of R&D per employee of a firm
in the manufacturing industry is significantly influenced by characteristics of the CEO
controlling for firm-level factors.
In model 2 until model 7, each hypothesis is tested separately in order to most accurately
determine its influence on the dependent variable. In terms of specific hypotheses, CEO age,
displayed in model 2, was significantly negatively associated with R&D expense (p ≤ 0,001),
which indicates that older CEOs (compared to younger CEOs) are less willing to invest money
into research and development, ceteris paribus. The next hypothesis that was tested, presented in
Adjusted R-Square 0,22 - - - 0,41
Change in adjusted R-Square (over model with no CEO characteristics)
model 3, looks at the influence of CEO tenure. Since there was no real consensus present in the
literature regarding the type of relation (positive or negative) between CEO tenure and R&D
expenditure, both a positive and negative hypothesis was introduced in this paper. After
analyzing the data, a significant negative relationship was found (p ≤ 0,10). Meaning that as
CEO tenure increases, the amount of money spent on research and development decreases. With
regards to the hypotheses about education, which are presented in model 4, we can conclude the
following. First, the level of education does not significantly influence the amount of money
spent on R&D (p>0,10). Second, the type of education does significantly influence the amount of financial investment into R&D. To be more specific, only the number of science and/or engineering degrees seems to significantly influence the amount of money spent on R&D (p ≤
0,05). On the other hand, both the number of business degrees and whether or not the CEO has
obtained a legal degree, seem to have no significant relation with R&D expenditures (p>0,10). Model 5 confirms, in accordance with the hypothesis, that there is a significant U-shaped
relation between CEO ownership and R&D investment per employee. This can be concluded
since both coefficients on CEO ownership and CEO ownership² are respectively negative and positive and both significant (both p ≤ 0,05). The turning point can be determined by calculating the vertex of the quadratic relationship. The vertex is calculated by the following formula: linear
coefficient/(2*quadratic coefficient) (Schechter 2014). Based upon this we can conclude that
R&D spending decreases as CEO ownership rises up to 32,86%, (0,0677/(2*0,00103)) after
The coefficients pertaining to nationality (model 6), show that there is a significant
relation between CEO nationality and R&D expenditure. Nonetheless, only two out of the four
hypotheses/Hofstede dimensions are significant. The first variable tested, individualism, was found to not be significant (p>0,10), therefore disproving the hypothesis. Second, power distance was found to be significant (p ≤ 0,05). The results indicate a negative relationship, which is in
accordance with the hypothesis. Third, once again, no significant relationship was found between
uncertainty avoidance and R&D expense (p>0,10). Finally, the second Hofstede dimension
which was found to have a significant influence on the dependent variables was ‘long term
DISCUSSION
Given the results found in this study we can clearly answer the research question posed in this
paper. Using a dataset consisting of 1832 firms operating in the manufacturing industry we found that many CEO characteristics significantly influence the level of innovation in a firm operating
in the manufacturing industry. This is consistent with the upper echelon theory, which predicts
that individual managers can influence critical decisions at their firms and that these are
influenced by the characteristics of the managers. The results of this study indicate that the
following characteristics significantly influence R&D expenditures: age, tenure, type of education, stock ownership, nationality and gender. Below we will discuss each hypothesis
separately, both the significant and non-significant ones. In order to provide the most accurate
and complete answer to the main proposition and research question posed in this paper. In
addition, at the end of this section, all the hypotheses, their significance level, and whether they are confirmed or not are displayed in table 6, to provide an overview of the results.
The first hypothesis presented and tested in this paper focused on the influence of CEO age on
R&D expenditure. The hypothesis states the following, CEO age is negatively associated with
firm-level innovation in the manufacturing industry. After analyzing the data we concluded that a
significant negative association is present between the two variables. In addition, the analysis
also suggests that this relation is the strongest among all the other variables tested in this paper.
Based upon the results we can conclude that older CEOs are less willing to invest money into
rationale behind the relationship is differing between authors. One of the most common rationale proposed in the existing literature is that older CEOs are less willing to change the status-quo
within the firms, because they do not have the physical and/or mental stamina anymore as they
become older. Therefore, they oppose adapting to new ideas and operating systems, which is a
vital component of innovation (Barker & Mueller 2002; Diks 2016). This also ties in with the fact that older people suffer more from cognitive decline than their younger counterparts making them less aware of the benefits of innovation (Frey et al. 2015). Furthermore, MacCrimmon and
Wehrung (1986) argue that as CEOs become older they become more risk averse. Meaning they
will be less inclined to make a decision which contains a significant amount of risk, which is most often the case for innovation. While the results from this paper do not definitely proof
which of these underlying reasons is true, it is reasonable to assume that all of the
aforementioned theories have some influence on the negative relationship between CEO age and
firm-level innovation.
The second hypothesis focused on association between CEO tenure and R&D expenditures. Since there was no consensus in the literature on the influence of CEO tenure, we
created both a positive and negative hypothesis. Namely, CEO tenure is negatively associated
with firm-level innovation in the manufacturing industry or CEO tenure is positively associated with firm-level innovation in the manufacturing industry. After conducting the analysis we can
conclude that there is a significant negative relationship present between the two variables.
Meaning, that the longer the tenure of the current CEO is, the lower the investment in R&D will be. According to the existing literature, this relationship is explained by the fact that CEOs that
environment, causing them to prefer stability over innovation (Miller 1991). In addition, authors
such as Barker and Mueller (2002) and Mezghanni (2010) argue that CEOs who reach the end of
their career (meaning their tenure is high) prefer less risky investments in order to maximize their own compensation/wealth. However, this assuration cannot be confirmed by this paper because of the lack of knowledge on the retirement plans of the CEOs.
Many authors and theories have often treated age and tenure as one variable, our results
suggest this should not be done in the future. While both influence R&D expenditure in the same direction (negative), the significance level is very different, causing age to be a much stronger predictor of R&D expenditure compared to tenure.
The third hypothesis tried to determine the relation between the level of education of the CEO and R&D expenditure. To be exact , the level of education achieved by a CEO is positively associated with firm-level innovation in the manufacturing industry. Our results indicate that the
level of education does not significantly influence the amount of money spent on R&D. This
finding contradicts other studies (e.g. Bantel and Jackson 1989; Kimberly and Evanisko 1981; Maclay 2017). One possible explanation for this is that the majority of CEOs in the data set have an education that falls somewhere between undergraduate (1) and master's degree (2). Therefore,
the results may focus almost solely on the differences between these two levels of education and
not on the differences between no-degree CEOs and undergraduate degree CEOs, since not many
CEOs in the sample have no degree. Thus, if the education level was regressed using dummy
variables, making the findings in this paper not the only exception (e.g. Barker and Mueller 2002; Daellenbach et al. 1999).
The fourth (set of) hypotheses focussed on the type of education obtained by the CEO.
The hypothesis states the following, the type of education achieved by a CEO is associated with firm-level innovation in the manufacturing industry. The results pertaining to this hypothesis showed that the type of education, contrary to the level of education, does significantly influence
financial investment into R&D. To go more into detail, the first type of education tested was
business degrees. The hypothesis states, firm-level innovation in the manufacturing industry is
negatively associated with the number of business degrees earned by its CEO. The results
however do not support this hypothesis, since no significant relationship was found. Thereby, not confirming the theories posed by authors such as Finkelstein and Hambrick (1996) and Hambrick and Mason (1984), who argue that MBA programs mainly attract students who are more conservative and risk averse. In addition, in the MBA programs, students are taught how to avoid big mistakes and how to reduce the risk of an investment. All these aspects, in general, do
not promote or even discourage innovation. However, as mentioned, the results of this report
indicate that this influence is not significant. The second type of education included in the model
was the legal degree. The hypothesis states that, firm-level innovation in the manufacturing
industry is negatively associated with its CEO having earned a legal degree . Once again, the
results do not support the hypothesis and the rationale behind it. The rationale, argued and
confirmed by Barker and Mueller (2002) states that legal degrees place absolutely no emphasis
on innovation, since legal students are taught how to apply the law and not think of inventive
amounts of money into R&D. However, contradicting the findings of Barker and Meuller (2002),
this analysis found no significant relationship between the two variables. The third and final type
of education included in the model is the number of science and/or engineering degrees. The
hypothesis states, firm level innovation in the manufacturing industry is positively associated
with the number of science and engineering degrees earned by its CEO . The results suggest a
significant positive relationship between the two variables, confirming the theories of Tyler and
Steensma (1998), who argue that CEOs who have enjoyed a degree in science or engineering are
more likely to be in favor of high levels of innovation. The reason being that they have a more complete understanding of technology and consequently innovation, which increases the likelihood of investing more money in R&D. To summarize, firms wanting to increase their
R&D expenditures should focus on the type of education of their CEO, namely increasing the
number of science and/or engineering degrees, and not focus on the level of education of the
CEO.
The fifth hypothesis presented and analyzed in this paper focused on the influence of
CEO stock ownership on R&D expenditure. To be exact, the hypothesis states, CEO stock
ownership has a u-shaped association with firm-level innovation in the manufacturing industry.
After examination, an U-shaped relationship between CEO stock ownership and the dependent
variable was found, this findings is consistent with agency theory. Meaning that when CEO
ownership is below the threshold of 32,86%, an increase in ownership will make the managers more hesitant to invest in R&D as this will reduce current profits and his/her returns on the stock
he or she owns. However, when CEOs pass the ownership threshold, their focus shifts from the
