Exploring the effect of CEO background characteristics on the new ventures
performance and the mediating effect of innovation
Master Thesis – Business studies: International Management University of Amsterdam
Date: 29-06-2015 Version: final thesis
Student: Elke Nijhoff [10607579]
MSc. in Business Administration – International Management Track Supervisor: Dr. Ilir Haxhi
Statement of originality
This document is written by student Elke Nijhoff who declares to take full
responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and
that no sources other than those mentioned in the text and its references have
been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision
of completion of the work, not for the contents.
ABSTRACT
New ventures are dynamic and innovative firms promoting economic growth and development; however, their survival rates are relatively low. Previous research on the upper echelon perspective indicates a positive relationship between managerial background characteristic and firm performance. However little is known about the nature of this relationship. In the context of new ventures we argue that innovation may play an important role in explaining this relationship, as innovation is the key source of long-term entrepreneurial success. Thus, in this study, we explore the mediating effect of innovation (i.e., the number of patents) on the relationship between the CEO background characteristics (i.e., age, gender, education and nationality) and new venture performance. For a sample of 200 new ventures in 12 OECD countries our findings show: first, a weak effect of CEO age, gender, education and nationality on new venture performance. Second, a positive effect of male CEOs on new venture innovation outcome. Finally, we found support for innovation as a partial mediator between managerial background characteristics and new venture performance when the characteristics (i.e., age, gender, education, and nationality) are tested together. We contribute to existing literature by gaining a better understanding on the CEO background characteristics - new venture performance relationship and start the debate on innovation as a mediator. As the managerial background characteristics combined appear to be useful for influencing innovation and success of new ventures, this research may guide entrepreneurs in their evaluation of new ventures and help them in investment decisions and innovation strategies.
ACKNOWLEDGEMENTS I would like to express my sincere gratitude to Dr. Ilir Haxhi - my thesis supervisor, for his guidance and supervision during the entire research process. His enthusiasm about the topic certainly inspired me to challenge myself and to achieve the best possible result. Furthermore, I want to thank my fellow team members Sanne Duursma and Natalia Kafkova for their joyful companionship in this project. Lastly I would like to thank my current employer, Estelle Roux, who gave me the time and space I needed to complete this study.
TABLE OF CONTENTS
1. INTRODUCTION ... 6
2. LITERATURE REVIEW ... 10
2.1. New ventures and their definition ... 10
2.2 Innovation, definition and measurement ... 11
2.3 Theoretical perspectives related to innovation and new ventures ... 13
2.4 The mediation ... 15
2.4.1 Managerial background characteristics and innovation ... 15
2.4.2 Innovation and new venture performance ... 15
3. THEORETICAL FRAMEWORK ... 17 3.1 Age ... 19 3.2 Gender ... 20 3.3 Nationality ... 21 3.4 Education ... 22 4. METHOD ... 24 4.1 Data Sources ... 24 4.2 Sample ... 24 4.3 Variables ... 25 4.3.1 Dependent variables ... 25 4.3.2 Independent variables ... 26 4.3.3 Mediating variable ... 27 4.3.4 Control variables ... 27 4.3.5 Method of Analysis ... 28 5. RESULTS ... 32 5.1 Descriptive analysis ... 32 5.2 Correlation analysis ... 33 5.3 Regression analysis ... 35
5.3.1 Managerial background characteristics – ROA regression (path c) ... 35
5.3.2 Managerial background characteristics – patents regression (path a) ... 36
5.3.3 Patents – ROA Regression (path b) ... 39
5.3.4 Mediation – indirect effect ... 40
5.4 Sample firms based in Nordic countries ... 41
5.4.1 managerial background characteristics – patents regression (path a) ... 41
5.4.2 Mediation – indirect effect ... 43
6. DISCUSSION ... 43
6.1 Managerial background characteristics – new venture performance relations ... 43
6.2 Managerial backrgound characteristcs – innovation relationship ... 46
6.3 Innovation – new venture performance relationship ... 45
6.4 Mediation ... 46
6.5 Nordics Sample ... 47
6.6 Theoretical contributions and managerial implications ... 48
7. LIMITATIONS ... 49
8. FUTURE RESEARCH ... 50
9. CONCLUSION ... 51
10. REFERENCES ... 54
1. INTRODUCTION
New ventures have been the driving force in the modern economy and will play an important role in the future as entrepreneurs work to meet the economic needs of people through the creation of new businesses (Mazzarol, Volery, Doss, & Thein, 2014). However, although new firms are a vital part of the economy, the future for all new firms is uncertain and the failure rate is high. A recent project coauthored by Berkeley & Stanford faculty members and ten new venture accelerators as contributors analyzed 3,200 high growth new ventures. The results show that within three years, 92% of startups failed (Marmer, Herrmann, Dogrultan, Berman, Eesley, & Blank, 2011). To better understand the factors relating to survival and growth many scholars have researched performance determinants that are present before large financial or other investments have been made (e.g., Hambrick and Mason, 1984; Cooper, Gimeno-Gascon, Woo, 1994; Okamuro, Kato, Honjo, 2009).
As new ventures generally have scarce initial resources, the background characteristics of their founders and managers are one of their main performance determinants (Criaco et al., 2014). The upper echelons theory, introduced by Hambrick and Mason, argues that organizational outcomes can be viewed as reflections of the values and cognitive bases of powerful actors in the organization (Hambrick and Mason, 1984). In line with the upper echelons view, a number of small business studies investigate the effect of background characteristics on firm success in general (e.g., Davidsson and Honig 2003; Wilson, Wright and Altanlar, 2014). These studies differ however in their measurement of background characteristics and firm performance. Unger, Rauch, Frese and Rosenbusch (2011) show, by conducting a meta-analysis, that the founder’s background characteristics are indeed a predictor of firm success. Initial resources such as the general background of the entrepreneur, management know-how and industry know-how, can be a buffer against the liabilities of newness and smallness (Cooper at al., 1994). In summary, much research has looked at the relation between managerial
background characteristics on firm performance. However, as of today little is known about the concept through which managerial background characteristics influence new venture performance.
Innovation has been widely recognized as an essential driver of economic growth. This research examines if innovation explains the relation between managerial background characteristics and new venture performance. Innovation in the context of smaller firms has received a lot of interest in recent literature, due to the importance of small and medium sized enterprises (SMEs) for economic and technological development (Ács and Audretsch, 2006). It has become an increasingly common subject of political action and scientific studies over the last two decades, as interest occurred from the widespread understanding that development and dissemination of innovations transform the economy into a more dynamic and knowledge-based economy (Dubina, 2013). Many existing studies argue that the emergence of new ventures and their post-entry performance are a source of regional innovation, economic growth and productivity and that small businesses play a significant role in a large fraction of innovations (e.g., Kato et al., 2015; Ács and Audretsch, 2006). Schumpeter (1934) introduced one of the first definitions for innovation and argued that continuous innovation activity is the key source of long- term entrepreneurial success. Substantial recent literature still suggests that in order to survive and thrive in increasingly hyper competitive markets, innovation is the only solution (e.g., Kim and Maubourgne, 2005; Bausch et al., 2008).
Innovation may have a mediation effect in explaining the relationship between managerial background characteristics and new venture performance, as research shows that managerial background characteristics predict new venture performance and innovation is the key source of long-term entrepreneurial success. To date, there is a whole body of literature on the relationship between managerial background characteristics and firm performance, based on the upper echelon theory. It is found, and therefore assumed in this study, that managerial background characteristics are a predictor of firm success (Unger et al., 2011). Also, research indicates that
there is a definite a link between innovation and new venture performance (Bausch et al., 2009). However we identified two gaps in the literature. First, existing innovation and upper echelon studies seem to neglect the possible mediating effect of innovation on the relationship between background characteristics and new venture performance. Second, research on the direct effect between managerial background characteristics and innovation is scarce and mixed. Evidence has been found that certain managerial background characteristics affect innovation activity in new ventures (e.g., Lynsky, 2004; Chaganti et al., 2008.) However, other studies have not discovered a connection between managerial background characteristics and innovation outcome (e.g., Davidsson and Honig, 2013).
To fill these gaps in the literature, we propose the following research question: What is the mediating effect of innovation on the relationship between managerial background characteristics and new venture performance? We suggest that managers are a crucial source of organizational recourses in new ventures. These organizational recourses provide communication and resource flows necessary for innovation to occur (Russel, 1990). Innovation in turn, helps new ventures in adapting to the changing environment and often is critical for firm survival.
We answer the research question by first looking at i) the direct effect between managerial background characteristics and innovation outcome and second ii) the possible mediation effect of innovation on the relationship between managerial background characteristics and new venture performance. We test the hypotheses by comparing patent, return on asset (ROA) and managerial background data from 200 new ventures from 12 OECD countries. OECD stands for Organization for Economic Co-operation and Development (OECD, 2011).
This study is unique by focusing on innovation as a possible cause of the relationship between organizational outcome and managerial background characteristic. We integrate the upper echelons theory with innovation activity. Also, so far only a handful of studies have tried to determine the role of managers on the performance of small and medium sized enterprises
(SMEs), particularly new ventures (e.g., Wilson et al.,2014). Directors in new and small firms play the critical role in providing recourses in the early stages (Zahra et al., 2009). Especially, given that new ventures have no track record. Thus, this study makes a number of contributions. First, it is one of few studies to examine the upper echelon theory in new ventures as opposed to a more mature firm. Second, it is one of the first studies to look at the nature of the managerial background – new venture performance relationship by integrating innovation activity. Hereby we start the debate on whether innovation is a mediator in explaining the relationship between managerial background characteristics and new venture performance.
This study is also relevant for practice. Our data suggest that, in the context of new ventures, the managerial background characteristics have a strong influence on the firm’s innovation activity. In turn the innovation activity has a strong influence on the firm performance. These findings could be very valuable for venture capitalists, private equity investors or other actors that plan to financially invest in new ventures. This research may guide entrepreneurs in their evaluation of new ventures and help new ventures in investment decisions and innovation strategy. In addition our contributions could help new ventures in their decisions regarding managers’ background characteristics criteria, by becoming more aware of the effect of the characteristics.
The remainder of this research is organized as follows. In section two we discuss previous literature related to the relationships mentioned in the research question, followed by hypotheses we developed based on the literature review. In the third section we describe the research methods, after which we report the research results. Following, there is a discussion based on the results. The final section entails acknowledgements of the limitations of the research, suggestions for future research and a clear and concise answer to the research question.
2. LITERATURE REVIEW 2.1 New ventures and their definition
In general new ventures are seen as an important source of economic development. Scholars believe new ventures will play an important role in the future as entrepreneurs work to meet the economic needs of people through the creation of new businesses (e.g., Fritsch, 2013; Mazzarol et al., 2014). However, scholars have been challenged to find one unifying general definition. For example, Bartlett defines a new venture as “a company, newly organized to exploit an idea […], founded by an individual sometimes referred to as the “entrepreneur” or the “founder”” (1999, p.6). Low and McMillan relate new ventures to the “creation of new enterprises” (1988, p.141). Bygrave refers to a “process of becoming rather than state of being” (1989, p.21). Blank and Dorf (2012) define a new venture as an organization established to search for a repeatable and scalable business model. The word ‘search’ differentiates large new ventures from small businesses. The latter implements a well-known, already existing business model, whereas a new venture enters an unknown or innovative business model. Nowadays examples are Amazon, Uber and Google.
As the subject new ventures has many different relevant aspects it has a wide range of perspectives and views, which results in different definitions. The definitions do not contradict each other, but they certainly address the topic from different perspectives. Academics do however agree upon the fact that a new venture is a relatively young, growing company, approximately six years or younger (Blank and Dorf, 2012).
Gartner (1985) states that the creation of a new venture is influenced by four factors: the environment, the individuals, the process and the organization itself. He states that a new venture is an independent entity that competes in the marketplace, which could be at a national or international level. He also stresses the fact that individuals with expertise are of great importance, if not crucial for the new venture.
a business is to conduct research in order to assess, validate and develop the business concepts. In addition is it difficult to establish further and deeper understanding on the business concepts and their commercial potential (Blank and Dorf, 2012). Fritsch (2013) states that investors are attracted to these new ventures for their risk/reward profile and scalability. They have lower bootstrapping costs, higher risk, and higher potential return on investment. Successful new ventures have the potential to grow rapidly with limited investment of capital, labor or land and are therefore more scalable then established businesses.
In the literature both terms start-up and new venture are used to define the above-described concept. For consistency reasons the term new venture is used throughout this study.
2.2 Innovation, definition and measurement
It is argued that innovation is the main driver behind economic growth and the key distinguishing element of entrepreneurship (e.g., Crossan and Apaydin, 2010; Davidsson, 2004). Schumpeter (1934) was the first to state that continuous innovation activity is the key source of long-term entrepreneurial success. Substantial literature nowadays even suggests that in order to survive and thrive in increasingly hyper competitive markets, innovation is the only solution. (e.g., Kim and Maubourgne, 2005; Bausch et al., 2008). Innovation is however far from a clearly defined concept.
Much research has been done on the subject innovation, a good working definition is however still missing. Schumpeter introduced one of the first definitions for innovation that emphasizes the novel aspect, summarized as ‘doing things differently’ (Crossan and Apaydin, 2010). Consequently the definition of innovation is supported by the concept of newness. However, innovation is not solely something new, it also needs to add value for the firm, the costumer and the stakeholders (Schramm et al., 2008). Not every change is considered to be innovation. The diverse characteristics of both concepts separates change from innovation, but the interrelatedness should
not be denied. Innovation presupposes changes, but not all change presupposes innovation (Crossan and Apaydin, 2010).
Innovation can be produced, internally through research and development, or adopted, externally bought (Crossan and Apaydin, 2010). Either way, after acquisition it must be exploited and add value to the firm. Baregheh, Rowley and Sambrook (2009) state that innovation can entail a process, a discrete item (product, program and/or service) and an attribute of firms. According to them, innovation is mostly defined by the type and nature of the innovation. In order words, innovation is mostly defined by something new or improved (nature) or by the type of output or result (e.g., product or service). Baregheh et al. (2009) provide the following definition: ‘Innovation is the multi-stage process whereby organizations transform ideas into new / improved products, services or processes, in order to advance, compete and differentiate themselves successfully in their marketplace.’ (p.1334).
Many scholars faced problems concerning the measurement of innovation (Smith at al., 2005). Trying to measure something that is in itself a novelty requires specification of innovation indicators that are dependent on the situation (Smith et al., 2005). Researchers are still looking for generally acceptable indicators of innovation. Current research suggests choosing innovation indicators by looking at what is being investigated (Kleinknecht, Van Montfort, & Brouwer, 2002). Two common measurements of innovation are the input, R&D expenditure, and the output, number of patents (Hausman, Hall and Griliches, 1984).
R&D expenditure as an indicator has several pros and cons. Data has been gathered regularly and is easy comparable across different firms and industries (Kleinknecht et al., 2002). On the downside, R&D expenditure says nothing about the output, real introduction of new products, and is only one of several inputs. It is not about the amount of money, or capital, spend on something that will make it successful. In other words, the biggest spenders do not have to be the most innovative firms (Kleinknecht et al., 2002).
Patents show how initial ideas for a new product, process, service or technical development are conversed into manageable and tangible products or services. Some scholars are skeptic about using patents as innovation indicator, as patents may not completely reflect the quality of innovation outcomes. Granted patents do however reveal the degree of the firm’s innovative activities. Even scholars who are critical towards the use of patents as innovation indicator, admit that patents can be an appropriate indicator in the context of high-tech sectors (Hausman, Hall and Griliches, 1984; Arundel and Kabla, 1998). Arundel and Kabla (1998) note that especially the choice of the studied industry is important when using patents as indicator, because the level of competitiveness forces firms to build protective walls of patents. The chosen industries for this study are the top five intensive innovative industries measured by the organization for economic cooperation and development EOCD (EOCD, 2011).
2.3 Theoretical perspectives related to innovation and new ventures
This research builds on the exciting theories of innovation and performance determinants. Crossan and Apaydin (2010) developed a multi-dimensional framework of organization innovation. In this framework both the recourse based view and the upper echelon perspective are mentioned as innovation determinants.
We use resource-based literature to develop hypotheses on the relationship between managerial background characteristics and new venture performance. Intending to posit that the managerial background characteristics of new ventures can bring resources, which enhances innovation and new venture performance. Moreover, managerial background characteristics create crucial new venture resources with significant effects on firm performance (e.g., Wilson et al., 2014; Davidsson and Honig, 2003). In the literature both managerial background characteristics and human capital are used to define similar concepts. For consistency reasons the term managerial background characteristics is used throughout this study.
The upper echelon perspective, introduced by Hambrick and Mason, emphasizes the role of top management team (TMT) skills and knowledge for firms’ innovative performance. This literature stream argues that organizational outcomes, both strategies and effectiveness, can be viewed as reflections of the values and cognitive bases of powerful actors in the organization (Hambrick and Mason, 1984). The definition of the TMT is an often-discussed topic in TMT research. Some researchers exclusively focus on the top-level executives or even the chief executive officer (CEO), others include vice presidents, directors and non-executive board members (Carpenter, Geletkanycz and Sanders, 2004). In this study we focus on the CEO, as most strategic decisions in entrepreneurial firms are made by only one or a few individuals (Gartner, 1985).
As new ventures generally have scare initial resources, the background characteristics of their founders are one of their main business assets (Criaco et al., 2014). Organizational outcomes both strategies and effectiveness are viewed as reflections of the values and cognitive base of powerful actors in the organization. It is expected that to some extent, such linkages, can be detected empirically (Hambrick and Mason, 1984). Complex decisions are largely the outcome of behavioral factors rather than a mechanical quest for economic optimizing (Coad, 2013). The cognitive structures of TMT members determine how they gather and filter information, interpret this filtered information and decide to act based on their interpretation of the information. Cognitive structures are a function of the TMT’s demographic characteristics (such as education and age) (Carpenter et al., 2004).
It has often been argued that founders’ and managerial background characteristics play a significant role in determining firm performance (e.g., Kato et al., 2015; Coad, 2013; Unger et al., 2011). For example, TMT members with higher levels of education are able to interpret more complex information and make faster strategic decisions (Wally and Baum, 1994). Also, older TMT members are often more resistant to change and will therefore prefer less risky strategic actions (Carpenter et al., 2004). Researchers have analyzed a variety of managerial background
characteristics such as director heterogeneity (Wilson, et al., 2014), entrepreneurial experience (Kato, et al., 2015), education (Lynskey, 2004) and age (Carpenter et al., 2004). The analysis of all these characteristics is beyond this research. Therefore it is decided to focus on four variables; age, education, nationality and gender.
2.4 The mediation
2.4.1 Managerial background characteristics and innovation
The previous paragraphs show that managerial background characteristics can have a positive influence on new venture performance. However, the effect of the managerial background characteristics on specifically small firms’ and new ventures’ innovative performance is mixed. Kato et al., (2015) provide evidence that certain types of founders’ human capital are directly associated with innovation outcome. Chaganti et al. (2008) find a relationship between the founding team members’ background and new ventures’ propensity to seek and pursue entrepreneurial opportunities. Whereas Lysnskey (2004) does not discover a connection between the CEOs human capital and new product development in new ventures. Davidsson and Honig (2013) also do not discover an effect of the entrepreneur’s human capital on first product sales of profitability. De Winne and Sels (2010) suggest that new ventures’ owners only contribute indirectly by hiring more highly educated employees and using more human resource practices, and not contribute directly to the firms innovative outcome by generating ideas and recognizing opportunities.
2.4.2 Innovation and new venture performance
Recently not only established firms but also new ventures have been given attention as the sources of regional innovation and productivity (Audretsch, Keilbach and Lehmann, 2006). New ventures have limited resources and experience; this negatively influences them in being successful in innovation. Although new ventures often face considerable resource constraints, they can innovate
successfully (Audretsch, Keilbach and Lehmann 2006). The relationship between new ventures and innovation is often described as a new venture advantage by international business literature. Due to several reasons innovation is known to have an enhancing effect on the survival of new ventures. Introducing innovative products, services, processes, or business models tailored to attract niches are additional opportunities for new ventures to stand out from competitors (Porter, 1980). New ventures can benefit from brand loyalty of consumers and reduces price sensitivity of demand, because costumers value the uniqueness of the innovation (Lieberman and Montgomery, 1988). Crossan and Apaydin (2010) state that innovation is an opportunity for entrepreneurial firms to gain rents through the temporary establishment of a monopoly and considers continuous innovation activity as the key element of long-term entrepreneurial success. As new ventures can act faster than their larger and settled counterparts, they can obtain these monopoly rents for a longer period of time. Musteen and Ahsan (2013) argue that new ventures have less rigid routines, lower levels of formalization and an entrepreneurial culture. Due to a less rigid routine, new ventures are able to adapt quicker to rapidly changing environments. Lumpkin and Dess (1996) state that new ventures are more likely to be alert and oriented towards entrepreneurship. In addition, Schumpeter (1934) states that innovation has a positive impact on market power of small-medium enterprises (SMEs). A disadvantage that new ventures may experience is the liability of newness.
Liability of newness arises when firms have less experience and less legitimacy in the eyes of potential investors and other stakeholders (Baum and Oliver, 1996; Stinchcombe, 1965). Newly established firms have to learn new roles, determine mutual relations and structure the fields of benefits and sanctions, which is one of the reasons why these new corporations experience high costs (Stinchcombe, 1965).
However, Bausch et al. (2008) applied meta-analyses techniques to aggregate prior empirical research on the innovation-performance relationship. Their findings indicate that innovation has a positive effect on the performance of small and medium sized firms (SME’s).
They also find that innovation has a stronger effect on younger firms than on more established firms. This outcome suggests that the liability of newness of younger firms can in contrary also be an asset for new firms. Bausch et al. (2008) indicate that new ventures own unique capabilities to create and appropriate value through innovations.
3. THEORETICAL FRAMEWORK
There is a broad range of literature, based on the upper echelon theory, on the relationship between managerial background characteristics and firm performance. It is found that managerial background characteristics are a predictor of new venture success (Unger et al., 2011). However little is known about the nature of this relationship and the mediating variables through which managerial background characteristics influence new venture performance. One of the key success factors for new ventures is innovation. The importance of innovation and new ventures has been the topic of a considerable amount of research. Various scholars state that innovation is the key distinguishing element of entrepreneurship. (e.g., Ács and Audretsch, 2006; Kato et al., 2015, Davidsson, 2004). We suggest that innovation has a mediation effect in explaining the relationship between managerial background characteristics and new venture performance, as research shows that managerial background characteristics predict new venture performance and innovation is the key source of long-term entrepreneurial success.
A B
C
As shown in the literature review a positive relationship between managerial backgrounds characteristics and new venture performance, path C in figure 1, and a positive relationship between innovation and new venture performance, path B in figure 1, has been found and is therefore assumed in this study. However, research on the relationship between managerial background characteristics and innovation, path A in figure 1, is mixed. Research on innovation as a possible mediator, figure 1 as a whole, has not been conducted.
We first look at the direct effect between managerial background characteristics and innovation. This relationship has not received extensive attention in recent research and the results are mixed. However, it is found that generic managerial background characteristics, such as educational background, effects innovation outcome (Lynsky, 20114; Kato et al., 2015). The managers’ background characteristics add resources, such as knowledge and experience, to the new venture. These resources drive managers decisions towards innovation strategies and therefore can enhance innovation. In a study on strategy and performance of new ventures a similar result was found. There is a relationship between the founding team members’ background and new ventures’ pursue of entrepreneurial opportunities (Chaganti et al., 2008).
Secondly, we look at the indirect mediation effect. This study suggests that innovation serves to clarify the nature of the relationship between managerial background characteristics and new venture performance. We believe that innovation plays this important role in governing the relationship as innovation has been recognized widely as an essential driver of economic growth especially through new ventures (e.g., Ács and Audretsch, 2006; Kato et al., 2015). Knowing that managerial background characteristics affect new venture performance and that innovation activity is the key source of long-term entrepreneurial success, we assume that i) managerial background characteristics influence the new ventures’ propensity to seek and pursue innovation activities and ii) that innovation in turn positively influences new venture performance. Recent research seems to neglect innovation as a mediator, however several recent studies have introduced the concept and
mentioned the relevance of researching innovation as a possible mediator on the managerial background characteristics – new venture performance relationship (Wilson et al., 2014; Kato et al., 2015). Recent research does argument that innovation serves as a key mediator between environmental uncertainty and performance (Han et al., 1998).
In the next sections we discuss the specific managerial background characteristics used in this study. First by looking at the direct effect between the characteristic and new venture performance and secondly by looking at the mediation effect of innovation on the characteristic – new ventures performance relationship. Like mentioned before we focus on the managerial background characteristics of the CEO. Previous studies have also used the CEO as a representative of the TMT (e.g. Barker and Mueller, 2002), as most strategic decisions in entrepreneurial firms are made by only one or a few individuals (Gartner, 1985). Finally, figure 2 shows a schematic representation of the hypotheses.
3.1 Age
There has not been extensive research about the association between the age of managers and new venture innovation outcome. But the ones that exist yield consistent results. It is believed that young managers attempt the novel, the unprecedented, taking risks. Young managers appear to be associated with corporate growth (e.g., Carpenter, 2004; Audretsch et al., 2006). A negative association has been found between managerial age and the ability to integrate information in making decisions. However there appears to be a positive association with tendencies to seek more information, to evaluate information accurately, and to take longer to make decisions (Hambrick and Mason, 1984). We suggest that the CEOs’ age negatively influences innovation outcome as older CEOs tend to stick to routines and innovation entails breaking routines. In a study on the characteristics of innovation a similar result was found. Older managers will be less willing to commit to changing organizational conditions and routines (Huber et al. 1993). Therefore we
hypothesize the following.
H1a: CEO age is negatively related to new venture innovation outcome.
We suggest that younger CEOs enhance innovation. In turn, innovation positively influences new venture performance (Bausch et al., 2008). Young managers tend to make faster decisions with confidence. Fast decision-making decreases the companies’ rigid routines and levels of formalization (Crossan and Apaydin, 2010), which results in faster responses when needed and better performance following from innovation. Therefore, innovation should mediate the relationship between CEO age and new venture performance.
H1b: The negative relationship between CEO age and new venture performance is mediated by innovation.
3.2 Gender
Research on the effect of gender on innovation is mixed. Stelter (2002) states that women tend to have a more transformational leadership style than their male counterparts, suggesting that female managers will positively influence innovation adoption. While DiTamoso and Farris (1992) found that woman R&D engineers tend to rate themselves lower than man do on innovativeness. Nevertheless, it is known that men and women differ in communication style, willingness to take risks, values and socialization (Nieboer, 2015). We suggest that gender may affect the chosen venture innovation strategies, as men are more willing to take risks compared to their female counterparts. A similar study on leadership and gender found that male top managers would initiate more innovation because they are more willing to change the status quo, and would more easily decide to adopt the innovation and allocate resources to it (Hooijberg and DiTomaso 1996).
Therefore, we propose the following hypothesis.
H2a: Male CEOs are positively related to new venture innovation outcome.
We suggest that male CEOs enhance innovation. In turn, innovation enhances new venture performance (Bausch et al., 2008). Innovation activities essentially goes from failure to failure in an effort to learn from each failure and discover what does not work in the process (Blank and Dorf, 2012). Men are more willing to take these risks and learn from the innovation failures, as women are more risk averse than men (Nieboer, 2015). This in turn will lead to better new venture performance. Therefore we hypothesize that innovation mediates the positive relationship between
male CEO and new venture performance.
H2b: The positive relationship between male CEOs and new venture performance is mediated by innovation.
3.3 Nationality
Local knowledge is important for new venture survival (Wilson et al., 2014). Johanson and Vahlne (2009) state that local knowledge may help enhance knowledge about the local customer base, suppliers’ relationships and regulations. However foreign managers can bring heterogeneity that enhances social capital from wider networks (Love et al., 2010). Boschma, Eriksson, and Lindgren (2009) find that inflows of unrelated skills from other regions are more likely to create communication problems. Overall, we suggest that local managers will bring advantages to new venture innovation, due to familiarity with the local context and a therefore bigger resource and capability base. This is similar to recent research findings that indicate that local managers, known with the company local environment can utilize local knowledge and professional networks to
support early stage new venture innovation and development (Wilson et al., 2014). Therefore we
hypothesize the following.
H3a: New ventures with local CEOs are more likely to engage in innovation than new ventures with foreign CEOs.
We suggest that local CEOs are positively related to innovation. In turn, innovation is positively related to new venture performance (Bausch et al., 2008). Innovation through a broader resource and capability base enhances the opportunity for entrepreneurial firms to gain rents through the temporary establishment of a monopoly and possible continuous innovation activity (Crossan and Apaydin, 2010). Therefore we hypothesize that innovation mediates the positive relationship between local CEOs and new venture performance.
H3b: The positive relationship between local CEOs and new venture performance is mediated by innovation
3.4 Education
Highly educated and more experienced managers are expected to be more successful in opportunity recognition (Hambrick and Mason 1984; Shane 2000; Ucbasaran et al. 2009). Bates (1990) found that human capital measured by years of education is strongly linked to business viability. According to Smith et al. (2005), education helps people to improve their understanding of what they know, to better manage time and recourses, to monitor results and to predict outcomes. We suggest that new ventures with high educated CEOs (i.e., greater knowledge base) to be more innovative than new ventures with low educated CEOs. High-educated CEOs will add more knowledge and resources to the new venture. This is in line with research that states that innovation
is considered a function of knowledge. It takes some level of existing knowledge to develop new knowledge (Smith et al. 2005). Therefore we propose the following hypothesis.
H4a: CEO education level positively influences new venture innovation outcome.
We suggest that higher CEO education increases new venture innovation. In turn innovation increases new venture performance (Bausch et al., 2008). Highly educated and more experienced managers are expected to be more successful than lower educated and inexperienced managers in opportunity recognition (Hambrick and Mason 1984; Shane 2000; Ucbasaran et al. 2009). This can lead to better new venture performance following from innovation. Therefore we hypothesize the following.
H4b: The positive relationship between the amount of education and new venture performance is mediated by innovation.
4. METHOD 4.1 Data Sources
To the best of our knowledge, there exists no publicly available data source for new ventures and innovation activities. In order to construct a data set of new ventures we conducted secondary sources. We used the data from Orbis. Orbis contains comprehensive information on listed and unlisted companies worldwide. In this research we define a new venture as a growing firm that is six years or younger, founded by an individual. The inclusion criteria we used to retrieve information about these new ventures are the following: all companies with an active status (active or default of payment); all companies that are incorporated in 2009 or later (after crisis); all companies are categorized by a small status, meaning that they do not have more than twenty-five employees. Their operating revenue has a maximum of 1 million euros and the total assets of these new ventures have a maximum value of 2 million euros. Orbis provided the country in which the new venture is located, the number of employees, the number of patents, the ROA, the CEO gender and the CEO age. We obtained the CEO education from LinkedIn, using a recruiter LinkedIn account. This account gave access to a wider range of profiles and revealed more background information than a regular LinkedIn profile.
4.2 Sample
The hypotheses are tested on a sample of 200 small and medium sized new ventures incorporated in 2009 or later and located in twelve OECD countries. Choosing firms incorporated in 2009 or later means that all firms where incorporated post-crisis and are six years or younger.
There were some restricting variables, which reduced the sample size. The number of patents, which is used as a proxy for innovation (Hausman, Hall and Griliches 1984), was not reported for all companies. Companies, of which no data was reported, were excluded from the
were also excluded from the sample. Information considering the education of managers was not provided on Orbis and therefore searched for on LinkedIn. We randomly selected companies with one or more patents to search their managers’ educational information until a database of 100 companies with patents was obtained. Subsequently, we randomly selected companies with zero patents to search their managers’ educational information until another database of 100 companies without patents was obtained. This resulted in a sample of 200 firms, of which 50% were firms with one or more patents and 50% were firms with no patents at all. As companies for which no data was reported were excluded from the sample and it was decided to pick 100 companies with at least one patent and 100 companies without patents, this resulted in an unbalanced sample of 200 companies from twelve different OECD countries with different numbers of patents from zero to Thirteen. Furthermore, we planned to include research and development expenses as an extra proxy for innovation. Unfortunately the data did not contain enough research and development expenses information, therefore this criteria is excluded from the sample.
4.3 Variables
4.3.1 Dependent variables
The dependent variable in this research is new venture performance. Referring to Chandler and Jansen (1992), two separate dimensions are particularly relevant for new ventures’ financial performance. The first is profitability, which relates to earnings, while the second is growth, referring not only to growth in revenue but also the change in market share. Taking into account that profitability is often hard to measure in the case of young new ventures, the measurement of performance in this research will focus on the growth of revenues. The measurement that will be used for this variable is return of assets (ROA), retrieved from Orbis. The results of patents do not have to occur within one year, which is why this study takes the average ROA over a four year time
period (last available year plus the three years before).
4.3.2 Independent variables
The independent variable of this study is managerial background characteristics. This variable is measured by managerial-specific characteristics, which can be retrieved from the data source. Due to data restrictions we only use CEO background characteristics. Previous studies have also used the CEO as a representative of the TMT (e.g. Barker and Mueller, 2002), as most strategic decisions in entrepreneurial firms are made by only a few individuals (Gartner, 1985). In the literature different terms are used to describe the CEO, such as Manager director or President. For consistency reasons the term CEO is used throughout this study.
In this research we use four types of characteristics. These characteristics (i) are likely to have an impact on de firms innovation capabilities, (ii) describe the main characteristics of new venture CEO and (iii) are accessible, so a database could be built. The first characteristic is age defined as the actual age of the CEO, measured as a continuous variable. Age is an often used characteristic in upper echelon studies (Carpenter et al., 2004). The second characteristic is gender, defined as the sex of the CEO. Gender is measured as a dichotomous variable (0 = male, 1 = female). This is an often-used measurement for gender (e.g. Damanpour and Schneider, 2009). The third characteristic is nationality, defined as the CEO being local or not. Managers born and raised in the same country as where the company is located are considered local managers. The nationality of the CEO will influence the knowledge and recourse base the CEO brings to the firm (Wilson et al., 2014). The last characteristic is education, defined by the level of education the CEO obtained. As education statistics around the world do not always use the same indicators, we used the International Standard Classification of Education (ISCED) developed by UNESCO as a guideline for education levels. ISCED facilitates comparisons of education statistics around the world on the basis of uniform and internationally agreed definitions (UNESCO, 2011). The level of education in this
research is measured by a 5-point scale (1 = no higher education; 2 =post-secondary education; 3 = bachelor; 4 = master; 5 = PhD or equivalent). This scale is similar to the one used in the study of Damanpour and Schneider (2009).
4.3.3 Mediating variable
The effects of managerial background characteristics may differ according to the types of innovation outcomes, but it is difficult to determine appropriate measures of innovation outcomes. An indicator for measuring innovation that is frequently used, despite its limitations, is the number of patents granted, also called intellectual property rights (Hausman, Hall and Griliches 1984). This information will be gathered from the Orbis database. Although granted patents may not completely reflect the quality of innovation outcomes, it does reveal the degree of the firm’s innovative activities. Acknowledging the limitations of granted patents we decided to also capture the R&D expenses, unfortunately the R&D data was not valid and had to be excluded.
4.3.4 Control variables
Industry was used as a control variable. Controlling for industry type is important as certain types of industries make more use of patents. Controlling for industry is also important as ROA is used as a performance measure in this research, and some industries require a larger amount of assets for a similar amount of profits compared to other industries (Lieberman and Montgomery, 1988). The industries included in this research are the five major most intensive innovation industries measured by EOCD (2011) based on Eurostat CIS-2006 and CIS-2004.
1. Manufacturing
2. Wholesale and retail trade, repair of motor vehicles 3. Information and Communication
5. Professional, scientific and technical activities
We dummy coded the companies as 1 if they belonged to a sector and 0 if they did not. Professional, scientific and technical activities industry was used as the reference group.
Company size was also used as a control variable and measured as the natural logarithm (In) of the number of employees, as found in Orbis. We did this because research suggests firm size influences firm performance (Kogut and Singh, 1988).
Another control variable we used is the country of origin (country in which the firm is located). Like the previous control variables also this variable is provided by Orbis. Country of origin is believed to be related to the ease of obtaining patents (Rosenbusch, Brinckmann and Bausch, 2011). The companies have been dummy coded as 1 if they originated in a specific country and 0 if they did not originate in that specific country. The Nordics was used as the reference group as most firms originated in the Nordics.
4.3.5 Method of Analysis
Following Baron and Kenny (1986) hierarchical multiple regressions were used to test the direct effects. Regression analysis is used when one or several independent variables are hypothesized to affect one dependent variable. Thus, regression analysis finds the best fitting straight line through a set of points. This relationship can be expressed in the following equation:
Where Y is the dependent variable β0 is the intercept term
βn are the n coefficients for independent variables ε is the error term
Here the parameter β0 and β1 are called the regression coefficients. They are the intercept (β0) and the slope (β1) of the straight line relating Y to X1 and X2. The slope can be interpreted as the number of units by which the independent variable would increase if the dependent variable would be increased by one unit. The error term denotes the difference between the estimated X1 and the actual X1 (Sekaran and Bougie, 2009).
In order to test the model four steps were conducted. Table 1 shows the variables included in the different steps of the hierarchical regression analyses. First a multiple regression is used to test that the causal variable managerial background characteristics; age, gender, education and nationality, are correlated with new venture financial outcome. This is path C in figure 3. It demonstrates the effect of the independent variables on ROA, while controlling for the control variables as specified above. Control variables, company industry, country and number of employees, were entered first and then the independent variables were added. This step established that there is an effect that may be mediated.
Hierarchical multiple regressions were used in the second step to show that the causal variables are correlated with the mediator innovation, measured by the number of patents. Therefore the mediator was treated as if it were an outcome variable. This is path A in figure 3.
A similar analysis was conducted in order to measure if innovation affects the outcome variable, ROA. This is path B in figure 3. Both managerial background characteristics and number of patents were treated as independent variables. It was not sufficient to just correlate the mediator with the outcome because the mediator and the outcome may be correlated because they are both caused by the causal variables, managerial background characteristics. Thus the managerial background characteristics variables were controlled in establishing the effect of the number of patents on ROA.
Finally, to establish that innovation, measured by the number of patents, completely mediates the managerial background characteristics – ROA relationship, the effect of managerial
background characteristics on ROA controlling for number of patents should be zero. When these numbers are zero we can speak of a full mediation effect. If this is not the case, there is still the possibility of partial mediation. We performed mediation analysis using PROCESS (Hayes, 2012). PROCESS has the ability to test if there is a relationship between the independent and dependent variable through a mediator. Modern thinking about mediation and one of the assumptions of PROCESS is that mediation analysis does not require evidence of a simple association between X and Y in order to estimate and test hypotheses about indirect effects (Hayes, 2012). This means that even if previous regressions did not show direct effects an indirect, mediation, effect was still possible.
Figure 3. Method of analysis
5. RESULTS 5.1 Descriptive analysis
To start, we discuss the descriptive statistics of the firms included in the sample. The mean of each variable and the standard deviation can be found in table 2.
All firms in the sample originate from twelve OECD countries. Most firms in the sample were located in Nordic countries, namely 47.5% were located in Sweden, 16.5% in Norway and 8.5% in Finland. Besides the Nordics quite some firms were located in Italy (12%) and France (5.5%). Finally only a couple of firms from the sample were located in Czech Republic, Germany, Spain, Japan, Korea, Slovakia or United Kingdom. Most firms in the sample (44%) operated in industries with professional, scientific and technical activities. 23.5% operated in information and communication industries, 16.5% in manufacturing industries, 14% in wholesale en retail trade industries and only 2% operated in industries with financial and insurance activities. The number of employees varied from one to twenty-five. The average number of employees of the firms in the sample was 4,58.
In this study we focus on the CEO of the new venture. 85% of the firms CEOs in the sample were male. CEO age varied over a large range, which can be concluded from the large standard deviation, relative to the mean. The mean age was 43,94 but the sample included ages from 24 up to 74. More than half of the firms CEOs in the sample (54%) was granted a Master degree. 31.5% was granted a Bachelor degree and only 9% a PhD degree. Less than 6% of the sample received a post-secondary degree or attended no higher education at all.
In this study we measure innovation by the number of patents owned by the firm. The largest number of patents was thirteen, but the average was 1,33. ROA per company is measured as the average ROA over the last available year, which for all companies is 2014 or 2013, and the
three years before that. ROA mean of all companies is 116.51% with a standard deviation of 2.39.
5.2 Correlation analysis
In order to assess the level of correlation between the different variables, we performed a correlation analysis. The results can be found in table 2. We discuss the correlations of 0.4 and higher here.
First, the correlation between number of patents and number of employees is 0.511. This might be because more employees increase the amount of resources within a firm and more resources increase the ease of obtaining patents. Second, the correlation between ROA and number of patents is -0.409. This is a negative correlation meaning that when the number of patents increases, the ROA decreases. This is a remarkable finding, as one would expect that a higher amount of patents would result in a higher ROA (Bausch et al., 2008). We will explain a possible reason for this later. Variables other than the ones discussed above have lower correlations, which are not likely to influence the analysis. The control variables country and industry cannot be included in the correlations, as these variables are categorical and correlations are meant to measure a linear relationship (Field 2013).
Before conducting a multiple regression analysis the data is tested for multicollinearity. To test multicollinearity we evaluate the tolerance and variance inflation factor (VIF). When the VIF is above 20 and the tolerance below 0.20 it means that several independent variables are correlating too high with each other (Field, 2009). As shown in table 3 the data experienced no problems with multicollinearity. VIF scores show no values above 20.
Table 2. Mean, Standard deviation and correlation matrix of the variables.
Table 3. Collinearity diagnostics
Dependent Variable: ROA *Significance on a 0.10 level **Significance on a 0.05 level
5.3 Regression analysis
In the following sections we report the results from the regression analysis. The control variables country and industry have been dummy coded. Professional scientific industry is the reference group with respect to industry type since this industry is represented the largest in the sample. The Nordics is the reference group with respect to the country in which the firm is located since the Nordics is represented the largest in the sample.
5.3.1 Managerial background characteristics – ROA regression (path c)
We performed hierarchical multiple regressions to investigate the ability of managerial background characteristics (age, gender, education and nationality) to predict new venture financial outcome,
Standardized Coefficients
t Sig.
Collinearity Statistics
Beta Tolerance VIF
(Constant) 3.655** .000 # Employees -.193 -2.388** .018 .601 1.664 Industry 1 -.264 -3.434** .001 .660 1.516 Industry 2 -.217 -2.803** .006 .653 1.532 Industry 3 -.049 -.668 .505 .731 1.368 Industry 4 -.057 -.886 .377 .951 1.051 CR .072 1.065 .288 .848 1.179 DE .013 .195 .845 .906 1.103 ES .046 .726 .469 .963 1.038 FR -.097 -1.495 .137 .931 1.074 IT .139 1.915* .057 .745 1.341 JP -.069 -1.051 .294 .915 1.093 KR .115 1.623 .106 .779 1.284 SVK -.033 -.515 .607 .967 1.035 UK .068 1.031 .304 .910 1.099 Gender .038 .583 .561 .911 1.097 Age of CEO -.066 -.989 .324 .871 1.148 Local or non-local .039 .606 .545 .969 1.032 Education -.077 -1.143 .254 .873 1.146 Patents -.282 -3.571** .000 .626 1.599
between X and Y is tested. The results can be found in appendix A. This table was put in the appendix as this regression tests the assumed (and proven) relationship between managerial background characteristics and firms performance. It helps explain the model but does not answer any of the hypotheses.
In the first step of the hierarchical multiple regression we entered the control variables industry, firm country and number of employees. This model (1.1) was statistically significant F (4,185) = 3.907; p < 0.001 and explained approximately 22.8% of variance in ROA. In the following models (1.2 to 1.5) the different managerial background characteristics are tested individually. All models were significant with p < 0.001 and F (5,185) = between 3.200 and 3.700. The introduction of the individual managerial background characteristics after controlling for industry, country and number of employees added very little explained variance to the model, all R2 changes were below 0.01. In the final model (1.6) all managerial background characteristics were tested simultaneously after controlling for industry, country and number of employees. This model explained 24.6% (R2=0.246) of the variance in new venture performance. The introduction of age, education level, nationality and gender simultaneously explained an additional 1.8% (R2 change= 0.018).
In sum, when controlling for industry, country and nationality the managerial background characteristics individually add only marginal to the explained variance in ROA. Even when combining all predictor variables in one model it only adds 1.8% to the explained variance in ROA. All the models as a whole have significant F values P < 0.001, but none of the F changes were significant and none of the managerial background characteristic predictor variable are statistical significant.
5.3.2 Managerial background characteristics – patents regression (path a)
and nationality, to predict innovation (number of patents), after controlling for country, industry and number of employees. The results can be found in table 4. In the first step of the hierarchical multiple regression we entered three predictors; industry, firm country and number of employees. This model (2.1) was statistically significant F (14,185) = 7.069; p < 0.001 and explained approximately 34.9% of variance in number of patents. In the following models (2.2 to 2.5) the different managerial background characteristics are tested individually. All models were significant with P < 0.001 and F = (15,185) between 6.00 en 7.10. In model 2.2, after entry of CEO age, the total variance explained by the model was 34.9% (R2=0.349). The introduction of CEO age after controlling for industry, country and number of employees did not add any explanatory power (R2 change = 0.000). In model 2.3, after entry of CEO education level, the total variance explain by the model was 35.6% (R2=0.356). The introduction of CEO education level explained an additional 0.8% (R2 change = 0.008). In model 2.4, after entry of CEO nationality, measured by being local or not, the total variance explained by the model was 35.2% (R2=0.352). The introduction of CEO nationality explained an additional 0.4% (R2 change = 0.004). In model 2.5, after entry of CEO gender, the total variance explained by the model was 36.4% (R2=0.365). The introduction of CEO gender after controlling for industry, country and number of employees explained an additional 1.5% (R2 change = 0.015). In the final model all managerial background characteristics were tested simultaneously after controlling for industry, country and number of employees. This model explained 37.4% (R2=0.374) of the variance in new venture performance. The introduction of age, education level, nationality and gender simultaneously explained an additional 2.6% (R2 change= 0.026).
In sum, when controlling for industry, country and nationality, the variables education and nationality increase the explained variance in patents by less then 1%. The age of the CEO does not increase the explained variance in patents at all. The CEOs gender explains 1.5% of the variance. All the managerial background characteristics combined increases the explained variance of the
model by 2.6%. However, the predictor variable age, education and nationality were not statistically significant. Also the models F changes were not significant. Only the gender of the CEO shows a significant result (p < 0.05). The B coefficient for gender is -0.788. We found support for hypothesis 2A. New ventures with male CEOs are more likely to engage in innovation as apposed to new ventures with female CEOs. The hypotheses 1B, 3B and 4B are not supported.
Table 4. Regression analyses with Patents as dependent variable.
2.1 2.2 2.3 2.4 2.5 2.6 Variables Constant 0.481* (0.231) 0.454 (0.494) -0.484 (0.685) 0.464* (0.232) 0.666** (0.246) -0.398 (0.938) Control variables Manufacturing industry .591 (0.430) .592 (0.432) .647 (0.430) .606 (0.430) .592 (0.426) .665 (0.429) Wholesale industry -1.129* (0.452) -1.127* (0.457) -1.025* (0.456) -1.132* (0.452) -1.142* (0.448) -1.031* (0.458) Information industry -.700* (0.340) -.697* (0.351) -.711* (0.338) -.730* (0.341) -.828* (0.342) -.853* (0.353) Financial industry -1.035 (0.957) -1.037 (0.961) -.982 (0.954) -1.020 (0.957) -.996 (0.948) -.938 (0.951) Professional, scientific industry Croatia -.669 (1.422) -.664 (1.430) -.808 (1.420) -.634 (1.422) -.803 (1.411) -.904 (1.418) Germany -.680 (1.131) -.680 (1.134) -.631 (1.127) -.650 (1.131 -.830 (1.123) -.757 (1.123) Spain .017 (0.849) .017 (0.851) -.108 (0.850) .043 (0.849) -.127 (0.844) -.234 (0.849) France .002 (0.588) .000 (0.591) -.108 (0.591) .026 (0.589) -.079 (0.584) -.177 (0.591) Italy .630 (0.456) .627 (0.462) .541 (0.458) .645 (0.456) .580 (0.452) .493 (0.462) Japan 2.886* (1.343) 2.879* (1.356) 2.618+ (1.351) 2.926* (1.344) 2.789* (1.332) 2.531+ (1.354) Korea -.749 (0.938) -.754 (0.949) -.808 (0.936) -.735 (0.938) -.906 (0.933) -.970 (0.942) Slovakia -1.029 -1.027 -1.386 -1.004 -1.196 -1.534
UK 1.257 (1.378) 1.258 (1.382) 1.336 (1.375) 1.297 (1.379) 1.174 (1.366) 1.288 (1.368) Nordics Number of employees 0.221*** (0.031) 0.221*** (0.031) 0.215*** (0.182) 0.220*** (0.197) 0.219** (0.030) 0.211*** (0.014) Independent Variables Age 0.001 (0.014) 0.002 (0.014) Education 0.273 (0.182) 0.278 (0.182) Nationality 0.201 (0.197) 0.155 (0.196) Gender -0.785* -0.788* (0.377) R Square 0.349 0.349 0.356 0.352 0.364 0.374 R Square change Compared to model 2.1 0.00 0.00 0.008 0.004 0.015 0.026 Adjusted R Square 0.299 0.299 0.304 0.299 0.312 0.312 F statistic 7.069*** 6.563*** 6.791*** 6.669*** 7.007*** 6.019*** *** indicates P < 0,001, **indicate P < 0,01 *Indicates P < 0,05, + indicates P < 0,10.
Standard error in parentheses.
5.3.3 Patents – ROA Regression (path B)
We conducted a similar analysis in order to measure if innovation affects the outcome variable, ROA. The results can be found in appendix B. This table was put in the appendix as the regression helps in explaining the model, but does not answer any of the hypotheses. Both managerial background characteristics and number of patents were treated as independent variables. It was not sufficient to just correlate the mediator with the outcome because the mediator and the outcome may be correlated because they are both caused by the causal variables, managerial background characteristics. Thus, the managerial background characteristics variables were controlled in
establishing the effect of the number of patents on ROA. This model (4.2) was statistically significant F (1,180) = 3.980; p < 0.001 and explained approximately 29.6% of variance in ROA. The model showed a significant F change of 3.459 P < 0.005. The predictor variable, patents, recorded a significant Beta value (β = -3.046). This indicates that there is a significant negative relationship between the number of patents and ROA.
5.3.4 Mediation – indirect effect
To test mediating effects in hypotheses 1B, 2B, 3B and 4B we used the PROCESS function of SPSS. The results can be found in table 7. During these tests there is again controlled for country, industry and number of employees. The models (4.1 to 4.5) were significant with a P value of <0.001. The SPSS output showed that there is no direct effect between the predictor variables and ROA. This is congruent with the first regression analysis that has been conducted (model 1.1 to 1.6). The total effect (direct effect plus indirect effect) is also not significant for all predictor variables. When looking at the mediation effect in table 6, indicated as indirect effect, the relationship between gender and ROA seems to be significantly mediated by the numbers of patents, LLCI=0.7282 and ULCI 4.9151***, P < 0.001. This said, we found support for hypothesis 2B. No support has been found for hypotheses 1B, 3B and 4B. However, when taking all predictor variables together, innovation significantly mediates the effect of age, gender, nationality and education on ROA.
Table 5. PROCESS coefficients.
4.1 4.2 4.3 4.4 4.5 Independent Variables Age -0.1463 -0.1545 Education -2.1682 -2.408 Nationality 1.1355 1.3645 Gender 2.1399 2.5546
