(1)Master Thesis Analysis of the regional distribution of the entrepreneurship-prone personality profile in the Netherlands

(1)

Master Thesis

Analysis of the regional distribution of the entrepreneurship-prone personality profile in the Netherlands.

Does it affect entrepreneurial activity and what is the impact of regional knowledge base?

by

Harmen Anema S3165027

University of Groningen Faculty of Economics and Business MSc BA - Small Business and Entrepreneurship

Supervisor: Michael Wyrwich Co-assessor: Erzsi Meerstra-De Haan

Date of submission: 13-01-2021

Word count: 10.765

(2)

Abstract

Multiple studies have investigated the regional distribution of the entrepreneurship-prone personality profile for multiple countries, including the US, the UK and Germany. Additionally, these studies report that this regional distribution influences the level of regional entrepreneurial activity in these countries.

As this relationship has not yet been investigated in the Netherlands, this paper aims to fill that gap by investigating this relationship and subsequently extending it by analysing if and how it is affected by the regional knowledge base. The analysis is conducted using data from the Gosling-Potter projects, Eurostat, CBS and VSNU. The results of the analysis show that the entrepreneurship-prone personality profile is regionally clustered in the Netherlands. Also, a positive relationship is found between the profile and the regional start-up rate. Regarding the regional self-employment rate, the results do not provide evidence for such a relationship. The results also show that this relationship is positively affected by the presence of an older classical university in a region and opposed to expectations, negatively affected by the presence of a technical university in a (neighbouring) region. To conclude, this paper found that the regional start-up rate is positively affected by the entrepreneurship-prone personality profile of a regional population and that the regional knowledge base, in some cases, plays an important role in this effect.

Keywords: entrepreneurship-prone personality profile, Big Five personality traits, regional start-up rate, regional self-employment rate, regional knowledge base.

(3)

Table of contents

Introduction 4

Literature review 6

Regional differences in personality traits 6

Big Five traits and entrepreneurship-prone personality profile 8

Regional knowledge base 10

Hypotheses 11

Methodology 15

Data collection 15

Measurements 16

Analysis 18

Results 19

Descriptive- and correlation analysis 19

Hypothesis testing 21

Discussion and conclusion 28

Discussion 28

Theoretical and practical implications 29

Research limitations and further research 30

References 32

Appendices 39

(4)

Introduction

Over the past few years, the topic entrepreneurship experienced a rise in popularity in the field of social science. Especially the relationship between an individual’s personality and entrepreneurship is increasingly becoming a topic of interest. (Obschonka, Schmitt-Rodermund, Silbereisen, Gosling & Potter, 2013). The literature on this relationship focuses primarily on two topics: the prediction of entrepreneurial outcomes (e.g. the likelihood of the entrepreneur achieving success) and entrepreneurial behaviour (e.g. the likelihood of an individual to become an entrepreneur) (Kerr, Kerr & Xu, 2017). The present study focuses on the latter. With regard to entrepreneurial behaviour, multiple studies have shown that individuals differ in their propensity to engage in self-employment (Fritsch & Rusakova, 2010; Blanchflower, Oswald, &

Stutzer, 2001) and that certain personality traits can play an important role in this (Rauch & Frese, 2007;

Brandstätter, 1997). Although entrepreneurial behaviour is often studied on the individual level, it can also be investigated on the regional level. For instance, from a cultural perspective, multiple studies have found that regional culture is an important factor in the emergence of new business formation (Fritsch & Wyrwich, 2019; Kibler, Kautonen & Fink, 2014; Mueller, 2005, Davidsson, 1995). Another perspective to investigate entrepreneurial behaviour on a regional level is by looking into the prevalence of certain personality traits in regions. For example, Rentflow (2010; 2008) found compelling evidence for the existence of regional differences concerning the prevalence of the Big Five personality domains of extraversion, agreeableness, conscientiousness, neuroticism, and openness (Goldberg, 1992). Furthermore, there is an increasing number of studies that provide evidence for the existence of an entrepreneurship-prone personality profile that functions as a predictor of entrepreneurial characteristics within individuals. This profile ‘reflects a characteristic constellation of traits that make entrepreneurial behaviour more likely’ (Obschonka, et al., 2013) and is based on the Big Five personality traits model (Fritsch, Obschonka, & Wyrwich, 2019;

Obschonka, Silbereisen, & Schmitt-Rodermund, 2010, Obschonka, Silbereisen, & Schmitt-Rodermund 2011, Obschonka, Silbereisen, & Schmitt-Rodermund 2012; Schmitt-Rodermund, 2004). Elaborating on the articles of Rentfrow (2010; 2008), Obschonka, et al. (2013) and Fritsch et al. (2019) found that the presence of the entrepreneurship-prone personality profile among populations is also not equally distributed, as their results show significant deviations in the distribution of this profile across regions within the UK, the US and Germany and thus indicate that it is regionally clustered. These regional differences are proven to affect the entrepreneurial activity of the region (Obschonka et al., 2013; Fritsch et al. 2019), referring to the regional start-up rate and the self-employment rate (Obschonka et al. 2013;

Blanchflower et al, 2001). In addition, the knowledge base of a region might also play an important role in the occurrence of these effects, as the regional knowledge base is proven to affect entrepreneurial activity (Acs, Braunerhjelm, Audretsch and Carlsson, 2009; Audretsch and Lehmann, 2005). However, as Acs et al. (2009) argued in their article, a strong regional knowledge base alone is not sufficient. A regional

(5)

population that has a certain propensity to engage in entrepreneurial activity is also required (Fritsch &

Wyrwich, 2018).

The present study aims to contribute to the existing literature on entrepreneurship-prone personality profile of a population and its relationship with entrepreneurial activity (Fritsch et al. 2019; Obschonka, 2013) by empirically examining this relationship in the Netherlands and subsequently extending it by incorporating the moderating variable of regional knowledge base. As of today, there is no empirical evidence for this relationship in the Netherlands. This study aims to fill that gap. Hence, in order to conduct this study, two research questions have been formulated. The first research question addresses the regional distribution of the profile and its effects on entrepreneurial activity: Is there a regional clustering of the entrepreneurship- prone personality profile in the Netherlands and does this clustering affect regional entrepreneurial activity?. The second research question relates to the moderating role of regional knowledge base on the relationship mentioned in the previous question: Does a strong regional knowledge base enhance the positive relationship between the regional population’s entrepreneurship-prone personality profile and regional entrepreneurial activity? With regard to policy-making, this study provides insights into the regional distribution of entrepreneurship-prone people. This could help policy makers, for instance local or provincial governments, to effectively direct their entrepreneurship stimulating measures to the regions that are actually prone to this, as directing these measures to less entrepreneurship-prone regions with a weaker knowledge base would be less effective and would probably yield lower results.

This paper will continue as follows: in the second chapter, a literature review will be given that provides a theoretical background including all topics that are relevant to this study. At the end of this chapter, testable hypotheses are formulated. The third chapter will address the methodological aspects of this study.

Subsequently the results will be discussed in the fourth chapter. This paper will conclude with a fifth chapter in which relevant conclusions will be drawn, limitations and implications of this study will be discussed and possible directions for future research will be presented.

(6)

Literature review

This literature review is divided into multiple sections. The first section will discuss the phenomenon of regional differences in personality traits. Subsequently, an elaboration will be provided on how an entrepreneurship-prone personality profile of a region can be derived based on the Big Five personality dimensions. After that, the role of a region’s knowledge base will be elaborated. This chapter will end with a section in which hypotheses are formulated.

Regional differences in personality traits

Research on geographical differences in personality suggest the existence of a personality trait variation across countries, but also across regions within those countries (Rentfrow, Gosling & Potter, 2008). A theory for explaining how personality operates at the geographical level is developed by Rentfrow, et al.

(2008). In their article they developed a theory that elaborates the processes of how ‘personality could emerge, persist and become expressed geographically’ (Rentfrow et al., 2008). Their theory is based on the assumption that: “If a geographical region scores higher on a particular personality dimension, then the population of that region will generally score higher than other regional populations on personality traits associated with that dimension” (Rentfrow et al., 2008). Consequently, behavioural tendencies linked to these personality traits will also be more pervasive in these regions than tendencies linked to less common personality traits (Rentfrow et al., 2008).

The emergence of these regional differences can be explained by looking at historical migration patterns.

Rentfrow et al. (2008) indicate two plausible explanations in their article. The first explanation relates to genetic founder effects which, in short, entails the migration of people with common genetic characteristics to certain geographical regions. As a consequence of this, certain ‘‘restricted gene pools of non-random samples of personality traits’ are likely to have emerged, which in turn caused a disproportionately large number of individuals in this region to develop certain similar personality traits. A second plausible explanation relates to social founder effects. The lifestyles, habits and histories of the early settlers could have caused the establishment of social norms in certain regions. Also, as a consequence of socialisation processes, regional populations acquired personality traits consistent with the behavioural tendencies that were common and admired in those particular regions (Rentfrow et al, 2008; Kitayama, Ishii, Imada, Takemura, & Ramaswamy, 2006; Nisbett, 2003; Hofstede, 2001).

However, as these explanations might account for the emergence of regional personality differences, they are not explanatory for the persistence of these differences. According to Rentfrow et al. (2008) the mechanisms of selective migration, social influence and environmental influence are important contributors to the persistence of regional personality differences. Selective migration refers to the tendency of certain

(7)

individuals to live near similar individuals, since these similar individuals are expected to understand and share the same languages, cultures and lifestyles (Rentfrow et al., 2008). Social influence has to do with the clustering of attitudes and beliefs is likely to occur when individuals engage in frequent social interaction with each other (Rentfrow et al, 2008; Bourgeois & Bowen, 2001; Latané, 1981). As a consequence of this social interaction, certain attitudes become more prevalent in a region not due to people choosing to live with other individuals that share the same interests and beliefs, but rather as a consequence of individuals influencing each other (Rentfrow et al, 2008; Bourgeois & Bowen, 2001). Environmental influence relates to the effect of the physical environment on the personality of individuals in certain regions. For instance, regional climate, level of urbanisation and neighbourhood characteristics all influence the commonality of certain psychological and behavioural tendencies among regional populations (Rentfrow et al., 2008). Each of the above described mechanisms contribute to the persistence of regional personality differences over time. However, the expression of personality at a geographical level is caused by other factors. In their article, Rentfrow et al. (2008) introduced a model that helps to explain the ways in which certain personality traits could become geographically expressed. As stated in their article, the

“model depicts a series of processes that could each cause personality to be represented geographically and, in turn, affect the prevalence of certain personality traits within a region”. For the sake of simplicity, the present paper will not go into detail with regard to the model’s processes. However, a clarifying overview of these processes and their interrelatedness is given below. Overall, the article of Rentfrow et al.

(2008) provides a logical explanation of how geographical variation in personality traits emerges, persists and becomes expressed. The theory described in their paper forms the basis of the variation in geographical distribution of the entrepreneurship-prone personality profile that is investigated in this study.

(8)

Big Five traits and entrepreneurship-prone personality profile

A personality trait can be described as ‘a habitual pattern of behaviour, thought or emotion (Guy, Kim, Lin

& Monocha, 2011). Even though humans possess and exhibit numerous different traits, the basic personality of an individual can be identified by just a few primary traits (Guy et al., 2011). Personality trait theories help identify these traits and provide an adequate means for indicating variations in an individual’s personality; scoring how weakly or strongly these primary traits are exhibited by individuals (Guy et al., 2011). The Big Five approach on which the entrepreneurship-prone personality profile is based is such a theory and is regarded as ‘the best established and most cross-culturally valid model of personality structure’ (Obschonka et al., 2013; Benet-Martinez & John, 1998; Digman, 1990; Schmitt, Alik, McCrae

& Benet-Martinez, 2007). This Big Five approach consists of five personality dimensions: extraversion, agreeableness, conscientiousness, neuroticism, and openness (Goldberg, 1992). These dimensions are widely used in prior research in the field of personality traits (Kerr et al., 2017; Rentfrow et al., 2008) and more recently also entrepreneurship (Zhao & Seibert, 2006). The studies from Fritsch et al. (2019) and Obschonka et al. (2013) use this Big Five approach as the foundation of their entrepreneurship-prone personality profile framework, as they describe the profile as ‘an entrepreneurial constellation of Big Five traits within the person’. Since the present paper aims to build on these studies, the same framework will be adopted. Before elaborating the relationship between entrepreneurship-prone personality profile and entrepreneurial activity an overview will be given on each of the Big Five dimensions and its link with entrepreneurship.

According to Zhao & Seibert (2006), ‘extraversion describes the extent to which people are assertive, dominant, energetic, active, talkative and enthusiastic’ (Antoncic, Bratkovic Kregar, Singh & DeNoble, 2015; Costa & McCrae, 1992; John 1990; Goldberg, 1990). Individuals that are more extroverted are found to be more cheerful, jovial, merry, optimistic and prone to seek leadership roles (Antoncic, 2015; Goldberg, 1990), while individuals that are more introverted tend to be more reserved, quiet and independent (Zhao

& Seibert, 2006). Research from Costa, McCrae & Holland (1984) found that persons with a higher score regarding extraversion also have greater interest in entrepreneurial activities and occupations. In addition, Caliendo, Fossen & Kritikos (2014) state that ‘extroverted individuals tend to be more sociable, enabling them to develop social networks more easily, which may result in stronger partnerships with clients and suppliers’. Ciavarella, Buchholtz, Riordan, Gatewood & Stokes (2004) found that ‘being assertive, seeking leadership roles and developing networks are positively related to entrepreneurial development in terms of entry decision and entrepreneurial survival’ (Caliendo et al., 2014). Overall, entrepreneurs are found to score high on the dimension of extraversion (Fritsch et al., 2019; Caliendo et al., 2014).

Regarding agreeableness, individuals that score high on this dimension are found to have a more forgiving and trusting nature and are regarded as more altruistic and flexible (Caliendo et al., 2014; Zhao & Seibert,

(9)

2006). These traits suggest that these high-scoring individuals are more cooperative, while lower scoring individuals are more likely to be self-centred, suspicious, ruthless and hard-bargaining (Caliendo et al., 2014; Zhao & Seibert, 2006; Costa & McCrae; 1992; Digman, 1990). Zhao & Seibert (2006) argue that the negative effects of agreeableness are ‘detrimental for those in an entrepreneurial role’, as entrepreneurs operate with limited access to resources and legal protections. As a consequence, they are more likely to suffer from bargaining advantages (Zhao & Seibert, 2006). Overall, entrepreneurs are found to score low on agreeableness (Zhao & Seibert, 2006).

Conscientiousness refers to ‘an individual's degree of organisation, persistence, hard work and motivation in the pursuit of accomplishment’ (Zhao & Seibert, 2006). Individuals that score high on this dimension are regarded as achievement-driven and are characterised by their drive to search for innovative solutions (Antoncic et al., 2015; McClelland, 1961). Also, McCelland (1961) found that entrepreneurs scored high on the need for achievement as ‘they take responsibility for their decisions, prefer decisions involving a moderate degree of risk, dislike repetitive, routine work and are interested in concrete knowledge of the results of the decisions’ (Antoncic et al., 2015). Rickman (2000) pointed out that this need for achievement is closely linked to conscientiousness and can therefore also be regarded as a substitute for this trait. Zhao

& Seibert (2006), in line with Howard & Howard (1995), conclude that a high level of conscientiousness is often seen in entrepreneurs. They even make the claim that among the Big Five personality traits, conscientiousness is the closest related to the entrepreneurship status (Antoncic et al., 2015). Overall, entrepreneurs are found to score high on conscientiousness (Caliendo et al., 2014; Zhao, Seibert, Lumpkin, 2010; Zhao & Seibert, 2006).

Zhao & Seibert (2006) describe neuroticism as the ‘individual differences in adjustment and emotional stability’. Individuals that score high on this dimension are more susceptible to experiencing feelings of

‘anxiety, hostility, depression, self-consciousness, impulsiveness and vulnerability’ (Zhao & Seibert, 2006;

Costa & McCrae, 1992). On the other hand, individuals that score lower on this dimension can be regarded as self-confident, calm, even tempered, relaxed and able to tolerate stressful situations (Caliendo et al., 2014; Zhao & Seibert, 2006). Once individuals engage in entrepreneurial activities, they inevitably have to deal with stress and uncertainty, as they work in uncertain environments with uncertain outcomes (Caliendo et al., 2014). Being resistant is therefore a desirable trait in entrepreneurs, because it helps them to effectively manage and control their business. Additionally, entrepreneurs can be described as highly self- confident (Zhao & Seibert, 2006; Chen, Greene & Crick, 1998; Crant, 1996) and are often found to have an internal locus of control (Caliendo, et al. 2014; Simon, Houghton, & Aquino, 1999), which indicates that entrepreneurs are more emotionally-stable. Overall, entrepreneurs are found to score low on neuroticism (Fritsch et al., 2019; Caliendo et al., 2014; Zhao & Seibert, 2006).

(10)

The last dimension, openness, refers to the extent of intellectual curiosity of an individual and its tendency to explore new ideas and seek new experiences (Zhao & Seibert, 2006). Individuals that score high on this dimension are more ‘creative, innovative, imaginative, reflective and untraditional’, while lower scoring individuals are described as ‘conventional, narrow in interests and unanalytical’ (Zhao & Seibert, 2006).

According to Hisrich, Peters & Shepherd (2005) and Churchill (1992) the creation of value through innovation and seizing of opportunities, together with the creation of something new, are essential elements of entrepreneurship (Antoncic et al., 2015). As already mentioned earlier, McClelland (1961) argued that entrepreneurs dislike repetitive and routine work, indicating a strong preference for creativity and, in a more broader sense, openness (Antoncic et al., 2015). Also, openness is considered to be crucial for recognising business opportunities, as it represents the starting point of the entrepreneurial process (Antoncic et al., 2015; Baron, 2007). Overall, entrepreneurs are found to score high on openness (Fritsch et al., 2019;

Caliendo et al., 2014; Zhao et al., 2006).

Based on the described relationships between each of the Big Five traits and entrepreneurship, we can conclude that entrepreneurs score relatively high on the traits of extraversion, conscientiousness and openness, but relatively low on the traits of agreeableness and neuroticism (Fritsch et al, 2019; Caliendo et al., 2014; Zhao & Seibert, 2006). Whereas the relationships between the Big Five traits and entrepreneurship were all conducted on the individual level, the present study, in line with Fritsch et al.

(2019) and Obschonka et al. (2013), aims to do this on an intra-individual level by combining all of these traits into an entrepreneurial profile index. This index will eventually lead to an entrepreneurial Big Five profile, also known as the entrepreneurship-prone personality profile. By doing this, the share of people with an entrepreneurship-prone personality profile in the regional population can be measured (Fritsch et al., 2019). The outcome of this measure can be compared to the ideal entrepreneurial personality structure (high on the traits of extraversion, conscientiousness and openness, low on the traits of agreeableness and neuroticism).

Regional knowledge base

Knowledge plays an important role in the existence and emergence of entrepreneurship. For instance, Acs, Braunerhjelm, Audretsch and Carlsson (2009) argue that business foundation, and particularly the foundation of innovative start-ups, can be considered as a manifestation of knowledge spillovers from existing knowledge sources (Fritsch & Wyrwich, 2018). They based their argument on the Knowledge Spillover Theory of the Firm of Acs (2008) and Audretsch, Keilbach and Lehman (2006). This theory suggests that generation of knowledge and ideas results from investment of incumbent organisations. Based on the expected economic value of this knowledge and these ideas, the incumbent organisation assesses the profitability of further investments in their development and commercialisation (Knoben, Ponds & van

(11)

Oort, 2011). As a consequence of information asymmetry and its uncertain character, the assessment of the economic value of new knowledge and ideas differs across organisations and individuals (Acs et al., 2009).

If the incumbent organisation, for example a research department of a university, eventually decides not to commercially exploit certain knowledge, as it assesses its economic value to be insufficient, the opportunity arises for an individual, that assesses the knowledge to be more valuable, to commercialise that knowledge and attempt to appropriate its economic value by means of starting a new business (Audretsch & Keilbach, 2007). Overall, it is suggested that knowledge spillovers should be considered as a source of entrepreneurial opportunities.

With regard to the empirical evidence supporting this theory, several studies have investigated the relationship between knowledge spillovers and entrepreneurship. For instance, the article of Acs et al.

(2009) establishing the knowledge spillover theory of entrepreneurship, provided evidence for their theory by investigating differences in start-up rates across multiple industries that have different underlying knowledge contexts (Audretsch & Lehmann, 2005). The results indicated that industries with higher levels of investment in new knowledge also exhibited higher start-up rates, while industries with lower levels of new knowledge investment also reported lower start-up rates. Accordingly, the study concluded that entrepreneurship and in particular the start-up rate should be seen as a mechanism through which knowledge spillovers are transmitted. Furthermore, Jaffe, Trajtenberg and Henderson (1993) and Audretsch and Feldman (1996) investigated the spatial characteristics of knowledge spillovers, concluding that knowledge spillovers tend to be geographically bounded and within the spatial proximity to the source (Audretsch &

Lehmann, 2005). These results are in line with the reasoning of Fritsch and Wyrwich (2018), who state that geographic proximity is likely to play a crucial role in the emergence of entrepreneurial knowledge spillovers. In their article, they give two reasons for this. First, new knowledge has the tendency to remain located within the proximity of where it was initially created (Anselin, Varga & Acs, 1997; Boschma, 2005;

Asheim and Gertler, 2006). Second, businesses tend to be founded and located within either the area of residence of the entrepreneur or the area of their former workplace (Figueiredo, Guimaraes & Woodward, 2002; Dahl & Sorenson, 2009).

In conclusion, the Knowledge Spillover Theory of Entrepreneurship suggests that regions with more knowledge generating organisations and thus a larger knowledge base, will accommodate more entrepreneurial opportunities and consequently exhibit higher start-up rates.

Hypotheses

As already discussed in the literature review, the study of Rentfrow et al. (2008) suggests the existence of regional personality differences. They measured this in terms of the Big Five personality traits model.

Fritsch et al. (2019) and Obschonka et al. (2013) elaborate on this in the field of entrepreneurship by

(12)

investigating whether these differences are also present in Germany, the United Kingdom and the United States. Their results indicate that significant differences exist across regions within these countries and that regions with a higher share of individuals with an entrepreneurship-prone personality profile report greater entrepreneurial activity, referring to higher regional start-up rates and self-employment rates (Obschonka et al., 2013). An explanation for this can be found in the article of Rentfrow et al. (2008) discussed in the literature review. In their paper, they make the following assumption: ”if a disproportionately large number of individuals within a region possess certain personality traits, then there should be more psychological and behavioural manifestations of those traits in that region than in other regions where the personality traits are less common”. With regard to entrepreneurial activity, this would imply that regions with a higher share of individuals with an entrepreneurship-prone personality profile would also exhibit greater entrepreneurial activity (Obschonka et al., 2013). In addition, Obschonka et al. (2013) state that with the prevalence of the entrepreneurship-prone personality profile in a region, it is also more likely that entrepreneurial norms and values are prevalent in that particular region, which could encourage entrepreneurial behaviour even among the individuals that are less prone to engaging in entrepreneurship.

Based on the above and in line with the studies of Obschonka et al. (2013) and Fritsch et al. (2019), the following hypothesis can be formulated:

H1: The regional share of people with an entrepreneurship-prone personality profile has a positive relationship with the regional entrepreneurial activity.

As discussed in the literature review, the knowledge base of a region is found to affect the emergence of entrepreneurial opportunities and consequently the number of start-ups in a region (Acs et al., 2009).

Accordingly, regions with a stronger knowledge base exhibit higher start-up rate than regions with weaker knowledge base. Therefore, it is argued by Acs et al. (2009) that generation of new knowledge via research and development is crucial for the emergence of start-ups and in particular innovative start-ups. The present paper focuses on the role of universities in this process for the following reasons: First, universities play an important role in the collection, generation and diffusion of knowledge (Fritsch & Wyrwich, 2018). Also, universities provide relevant inputs with regard to innovation and contribute to the regional presence of human capital that is crucial for identification of entrepreneurial opportunities (Fritsch & Wyrwich, 2018;

Schubert & Kroll, 2016). Lastly, universities function as ‘brokers’ and ‘gatekeepers’ in local innovation systems (Kauffeld-Monz & Fritsch, 2013). A broker should be seen as a network actor that is responsible for the transfer of knowledge between organisations that are not directly linked, while gatekeepers should be seen as actors that are able to provide third parties access to global knowledge sources. Universities should therefore be regarded as central actors in regional innovation networks (Kauffeld-Monz & Fritsch,

(13)

2013). Based on the above, this paper assumes that the presence of a university in a region is an adequate indicator for the strength of the regional knowledge base.

It is expected that the relationship mentioned in the first hypothesis on entrepreneurship-prone personality profile and regional entrepreneurial activity is moderated by the presence of a university. The Knowledge Spillover Theory argues that a strong regional knowledge base and thus also university presence does not naturally lead to new business formation, but that it also requires entrepreneurial individuals that recognise and seize opportunities (Fritsch & Wyrwich, 2018; Acs et al., 2009). Therefore, according to Fritsch &

Wyrwich (2018) the likelihood of the regional population to engage in entrepreneurial activities plays an important role in the occurrence of entrepreneurial spillovers. Reasoning from the expected positive relationship between entrepreneurship-prone personality profile and regional entrepreneurial activity, it can therefore be expected that in regions with a higher share of people with an entrepreneurship-prone personality profile and thus regional populations with a higher propensity to engage in entrepreneurship also exhibit higher levels of entrepreneurial activity (Obschonka et al., 2013) and that in regions in which a university is present this relationship will even be stronger, as those regions have a stronger knowledge base and thus accommodate more entrepreneurial opportunities. Hence, the following hypothesis can be formulated:

H2: The positive relationship between the regional share of people with an entrepreneurship-prone personality profile and regional entrepreneurial activity (H1) will be enhanced in regions that accommodate a university.

Elaborating on the second hypothesis, a distinction can be made between the type of university that is located in a certain region. This distinction is relevant because not all knowledge spilling over from universities is similar, as the nature of certain universities might differ. (Audretsch, Lehmann & Warning, 2005). As a consequence, this also affects the nature of the knowledge base of a region. In line with Fritsch and Wyrwich (2018), the present paper distinguishes between classical universities (CUs) and technical universities (TUs), with the first referring to general universities that offer their students a large scope of academic programmes and the latter referring to technology-orientated universities who solely offer their students academic programmes in the field of natural sciences and engineering. Traditionally, many of the research and teaching areas of technical universities can be linked to industry activities (Goethner &

Wyrwich, 2019). Consequently, technical universities are posited to be more proficient in creating knowledge spillovers and commercialising knowledge and ideas than classical universities (Goethner &

Wyrwich, 2019; Audretsch & Lehmann, 2005; Audretsch & Lehmann, 2005). This position is in line with the findings of Goethner & Wyrwich (2019). In their article on the effect of spatial proximity on

(14)

entrepreneurship at cross-faculty level, they show that the faculties associated with natural sciences and engineering have a greater likelihood of having at least a single grant successfully awarded to a start-up, when compared to the other faculties of social sciences, humanities, arts & culture and sport sciences. This indicates that start-ups spinning off from these faculties are more often valued as promising than those from other faculties. Furthermore, it is proven that knowledge-based, technology-intensive firms can benefit from settling near universities (Bathelt, Kogler & Munro, 2010; Arrow, 1962; Nelson. 1959). For instance, Audretsch and Lehmann (2005) found that especially high-tech ventures benefit from localised knowledge spillovers created by universities (Bathelt, Kogler & Munro, 2010). More specifically, they found that research intensive universities in the field of natural science should be considered to be more attractive to technology-based firms, as their analysis showed that university research in particular fields of natural sciences is positively related to the number of firms clustering around a university, while they did not find such an effect for university research in social sciences (Audretsch & Lehmann, 2005). Overall, these findings suggest that technical universities, which focus on natural sciences and engineering indeed are more adept at creating knowledge spillovers and commercialising knowledge and ideas, as spinoffs rooted in these sciences are more often considered as promising and research in these fields is shown to positively affect the number of firms settled around universities. Therefore, the following hypothesis can be formulated:

H3: The positive relationship between the regional share of people with an entrepreneurship-prone personality profile and regional entrepreneurial activity (H1) will be stronger in regions that accommodate a technical university, than in regions that accommodate a classical university.

(15)

Methodology

In the present study, the theory testing approach will be followed as Van Aken, Berends & Van der Bij (2012) state that this approach should be followed in cases where literature streams are already elaborated.

The present study builds on the trait theory, knowledge spillover theory and, more specifically, the entrepreneurship-prone personality profile and identifies a literature gap that has not been investigated yet.

Therefore, taking the theory testing approach for this paper seems to be the most suitable option. In accordance with the theory testing approach the following steps have been taken: (1) definition of the phenomenon and identification of the gap in the literature, (2) development of conceptual model and hypotheses, (3) data analysis, (4) interpretation of results with the comparison to initial hypotheses, drawing of conclusions and provision of implications, limitations and directions for future research (Van Aken et al., 2012). The step of data collection has been left out, as the present study uses existing, secondary databases.

The remainder of this chapter will continue as follows: first, the collection of the necessary data will be addressed. After that, all measurements used in this study will be discussed. This chapter will conclude with a section describing the approach that was taken to conduct the analysis.

Data collection

As already mentioned before, this study will be conducted in the Netherlands. Previous research has shown that the results found in the US are also generalisable to other highly developed, innovation-driven countries like Germany and the UK (Obschonka et al., 2013, Fritsch et al., 2019; Audretsch, 2007). The Netherlands can also be defined as a highly developed and innovative country, as it is ranked among the ten most developed countries of the world (United Nations Human Development Index, 2020) and ranked fifth in the Global Innovation Index (World Intellectual Property Organisation, 2020). It is therefore interesting to investigate whether the Netherlands, being a comparable country with regard to development and innovation, but not with regard to size, shows similar results. Another interesting reason to investigate the Netherlands is related to the population density of the country. The Netherlands is nearly, and more than twice as densely populated as respectively the UK and Germany (Our World in Data, 2017). Overall, the Netherlands is a suitable country for investigating the entrepreneurship-prone personality profile, as the results will provide insight into how this profile behaves in smaller and more densely populated countries.

Regarding the collection of all the necessary data, the present study uses multiple available databases that will be combined into a single database. The data with regard to personality and the entrepreneurship-prone

(16)

personality profile will be gathered from the Global Gosling-Potter Internet projects (Fritsch et al., 2019;

Rentfrow et al., 2008), which collected data in multiple countries, including the Netherlands. This data has been collected over the years of 1998 to 2015 by means of an online survey (https://outofservice.com), that contained multiple statements on the Big Five personality traits. A total of 111,224 Dutch individuals participated in the survey. After filtering out unuseful cases, a total of 108,403 cases were included in the analysis. Only individuals within the age range of 15-75 years were included in the sample, as individuals outside this range are not considered to be part of the working population (CBS, 2020). In order to assess the representativeness of this sample, a correlation analysis is conducted (Fritsch et al., 2019). A correlation analysis between the population share of the regions in the total population of the Netherlands and the population share of the regions in the total sample population showed a correlation of r = 0.955, indicating that the sample is highly representative for the Dutch situation. The data with regard to regional entrepreneurial activity (regional start-up rate and self-employment rate) will be gathered from the Central Statistical Office of the Netherlands (CBS) and will be collected over the years of 2015 to 2018. Regarding the moderating variable of regional knowledge base (university presence), the data is collected from Studielink (http://info.studielink.nl), which is the official authority with regard to higher education enrolment in the Netherlands. Studielink provides an overview of all the official universities in the Netherlands. Data regarding the age will be collected from the websites of each individual university. Data on the size of the universities will be collected over the years 2015 to 2018 and will be gathered from the Association of Dutch Universities (VSNU). The data on the control variables of regional GDP, regional population density and regional employment share of the manufacturing industry will be gathered from the EUROSTAT database of the European Commission and is, similar to the data on entrepreneurial activity, collected over the years 2015 to 2018. All data with regard to the dependent, moderating and control variables will be collected on NUTS 3 level, which is equivalent to the Dutch COROP level. In order to get the data of the Gosling-Potter project on the same level of aggregation as the data from the CBS and EUROSTAT, respondents of the online survey were asked for their zip-codes. Based on their zip-codes, they can be divided across the 40 COROP regions in the Netherlands. Lastly, averages have been calculated for all the data that has been collected over the course of multiple years (2015-2018). By doing this, the risk of potential biases is mitigated.

Measurements

Entrepreneurship-prone personality profile of a region

Initially, the entrepreneurship-prone personality fit will be measured at the individual level (Fritsch et al.

2019; Obschonka & Stuetzer, 2017). The individuals that participated in the survey were asked to indicate

(17)

to what extent they agreed or disagreed with multiple statements on a five-point likert scale (1= Strongly Disagree, 5= Strongly Agree). The participants were allocated to a Dutch COROP region based on their zip codes. Also, a statistical reference profile of an entrepreneurial personality profile is constructed (high on extraversion, conscientiousness and openness, low on agreeableness and neuroticism) (Fritsch et al., 2019;

Obschonka & Stuetzer, 2017). In order to measure the entrepreneurship-prone personality fit of an individual, the sum of the squared deviations of the individual’s Big Five scores from the reference profile is calculated (Fritsch et al., 2019; Cronbach & Gleser, 1953,measure). Afterwards, these individual values are aggregated to regional level by calculating averages scores per COROP region (based on the participants’ zip codes)(Fritsch et al., 2019).

Regional entrepreneurial activity

Regional entrepreneurial activity will be measured using the regional start-up rate and the regional self- employment rate (Fritsch et al., 2019; Obschonka et al., 2013). The former will be measured by calculating the average annual number of start-ups formed per population in the working age (15-75 years old and expressed in 10000’s) (Fritsch et al., 2019). The latter will be measured by dividing the number of self- employed persons in the private sector by the total regional labour force (Fritsch et al., 2019). This data will all be collected on COROP level.

Regional knowledge base

In line with the article of Fritsch and Wyrwich (2018), the knowledge base of a region will be measured by indicating whether there is a university present in the particular COROP region. This study focuses solely on Dutch universities; other Dutch higher education institutions like Hogescholen (HBOs) are excluded from the analysis. A distinction will be made between classical universities (CUs) and technical universities (TUs). Two binary variables will be formed for the presence of each (Fritsch & Wyrwich, 2018). The universities will be allocated to their specific COROP region based on the zip codes of their headquarters.

Also, as the spillover effect initiated by universities might not be restricted to man-made, administrative borders, this study also investigates whether certain regions are affected by university presence in a neighbouring region. In addition, the age of the universities will also be considered. Following the article of Fritsch et al. (2019) and Fritsch and Wyrwich (2018), a distinction will be drawn between universities founded before the year 1900 and those founded after this year. Lastly, the size of the universities will also be taken into account (Fritsch et al., 2019). The average number of total enrolments per university will be calculated over the course of multiple years. Based on the average number of enrolments of all universities,

(18)

a university will be defined as either ‘large’, when exceeding this average or ‘small’ when falling short (table 12, in the appendix). These steps will be carried out separately for both CUs and TUs.

Control variables

The control variable of population density is incorporated into the analysis, as it ‘can be regarded as a

‘catch-all’ variable for diverse characteristics of the regional environment, because it is correlated with several other metrics that might have an effect on the level of entrepreneurship and regional economic development’ (Fritsch & Wyrwich, 2017). In addition, GDP (gross domestic product) has to be incorporated as well, since it is proven that economic conditions affect entrepreneurial activity, in a sense that individuals prefer to start new firms in more prosperous regions rather than in less prosperous ones (Obschonka et al., 2013; Armington & Acs, 2002; Sternberg, 2009, Fritsch & Mueller, 2007). Province dummies will also be included in this analysis, because they are a relevant actor in entrepreneurial policy making. Their influence should thus be controlled for (Fritsch et al., 2019; Fritsch & Wyrwich, 2017). Lastly, the employment share of the manufacturing industry will also be incorporated as it controls for sectoral structure differences among regions (Fritsch et al., 2019)

Analysis

Before testing on the hypotheses, a pre-analysis will be conducted. This pre-analysis incorporates descriptives on the distribution of the entrepreneurship-prone personality profile in the Netherlands, the start-up rate and the self-employment rate. Thereafter, a correlation matrix that exhibits the correlation between all variables. Testing for construct validity is in this case not necessary, since this study uses secondary data extracted from the CBS and the Gosling-Potter database. After the pre-analysis, the hypothesis will be tested. In line with Fritsch et al. (2019), an Ordinary Least Squares (OLS) regression will be conducted. Multiple regression models will be run for both the start-up rate and the self-employment rate. The effect of the regional knowledge base variables will be measured using the measurements discussed above. Lastly, the control variables will be incorporated in a stepwise manner.

(19)

Results

In this chapter, the results of the data-analysis will be presented. First, results from the descriptive analysis and correlation analysis will be discussed. After that, the results of the Ordinary Least Squares (OLS) regressions will be presented which, subsequently, will be linked to the hypotheses developed in the hypothesis section.

Descriptive- and correlation analysis

The results of the descriptive analysis show that the mean value of the entrepreneurship-prone personality profile in the Netherlands is equal to -19.39 and has a standard deviation of 0.3017. The maximum value of the profile is -18.67, whilst the minimum value in the sample is -19.85. The COROP regions of Groot- Amsterdam, ‘s-Gravenhage and Haarlem show the highest scores: -18.67, -18,70 and -18.93, respectively.

The regions Veluwe, Zuidoost-Drenthe and Oost-Groningen show the lowest scores, scoring: -19.95, -19.85 and -19.81, respectively. This implies that the populations of the former regions can be considered to be the most entrepreneurial, whilst those of the latter regions can be considered to be the least. With regard to entrepreneurial activity, the results show that the mean value of the number of start-ups per 10,000 working age individuals is approximately 99.90 with a standard deviation of 25.5392. The maximum value in the sample is 193.32 and the minimum value is 60,64. The regions scoring the highest on the number of start- ups are respectively: Groot-Amsterdam (193.32) Het Gooi en Vechtstreek (151,26) and ‘s-Gravenhage (147,89), whilst the lowest scoring regions are respectively Delfzijl en omgeving (60,64), Oost-Groningen (67,11) and Zuid-Limburg (74,57). The analysis on self-employment rate shows that the mean value is approximately 0.1580 (15,80) with a standard deviation of 0.0156. The maximum value of the sample is equal to 0.1962 (19.62%), whilst the minimum is 0.1364 (13.64%). When looking at the regional level, Groot-Amsterdam (19.62%), Zuidwest-Friesland (19.52%) and Het Gooi en Vechtstreek (19.44%) exhibit the highest levels of self-employment. In contrast, Noord-Limburg (13.64%), Flevoland (13.81%) and Groot-Rijnmond (13.98%) exhibit the lowest levels of self-employment.

The results of the correlation analysis (table 1) show that some variables are strongly correlated (r=>0.8) to others. For instance, the regional presence of a small technical university is strongly correlated (r=0.854, p=0.01) to the overall presence of a technical university in a region. Furthermore, the analysis also showed that the regional presence of a technical university before the year 1900 is strongly correlated to the presence of a large technical university (r=1.000, p=0.01), indicating that all large technical universities were founded before 1900. Lastly, the regional presence of a classical university before 1900 is strongly correlated to the regional presence of a large classical university (r=0.882, p=0.01). The results also indicate multiple variables that are moderately correlated. For instance, regional technical university presence is

(20)

moderately correlated to classical university presence (r=0.509, p=0.01). Also, both regional presence of a large classical university and regional presence of a classical university before the year 1900, are moderately correlated to regional classical university presence, (r=0.577, p=0.01) and (r=0.509, p=0.01). Based on these results, these previously discussed variables cannot be used in conjunction with each other. However, since these variables are all moderating variables, they are not used simultaneously in one model, as each moderating effect is tested separately. Accordingly, the problem of potential multicollinearity among these variables is mitigated. Another moderately strong correlation is the one between entrepreneurship-prone personality profile and the start-up rate (r=0.721, p=0.01). As these two variables are at the core of this research, they are also tested for multicollinearity by analysing the VIF value. Other variables that are investigated further are the three control variables of population density, gross domestic product and employment share of manufacturing firms, as these show correlations above r=0,5 with each other as well as other variables. However, in all regression models that are run, none of the included variables show VIF values that exceed the critical threshold of 10 (Mason, Gunst, & Hess, 1989). Therefore, all variables can be kept into the analysis.

Table 1: Pearson Correlation Matrix

Notes: * = significant at 10% level, ** = significant at 5% level, *** = significant at 1% level.

(21)

Hypothesis testing

In this section, an overview of the most important and relevant results will be presented. The hypotheses formulated in the theory chapter will be accepted or rejected based on these results. Multiple models have been composed to test for the hypotheses. All models have been tested by means of a OLS regression.

Tables 2,3,7 and 9 show the results of the analysis on the effect of the entrepreneurship-prone personality profile (also referred to as ‘the profile’) on start-up rate and the moderating role of knowledge base. Most models indicate that the profile is positively related to start-up rate. For instance, the results in model 1 indicate, at a significance level of 5%, that for every unit increase in entrepreneurship-prone personality profile, the start-up rate per 10,000 working population increases with 20.124. The second model does not show such an effect, as the effect of the profile is insignificant in this model. Overall, the positive effect of the entrepreneurship-prone personality profile is seen to be significant, at different significance levels, in the majority of the models (1,4,5,7,8,10-13,16-18,37,38,41,49-54). When taking a closer look at the results, a certain pattern can be seen recognised in the significance of the results, as the significance lowers with every control variable added to the model. For instance, model 10 with solely the population density control included, shows a positive relationship at the significance level of 1%, while model 11, with the control variable of GDP added, shows a significance at the 5% level and subsequently, model 12, which also includes the control variable of employment share shows a significance at the 10% level. This pattern is seen among many other models as well.

The results with regard to self-employment rate are less convincing, as only a few models show a significant effect of the entrepreneurship-prone personality profile on self-employment (models 28, 29,31,34-36). For instance, model 28 indicates, at a 10% significance level, that for every unit increase in the profile, the self- employment rate increases with 0.021. Also, similar to the models concerning start-up rate, the significance of the results tends to decrease with every control variable that gets added to the model. The overall conclusion that can be derived from these results is that hypothesis 1 (H1: The regional share of people with an entrepreneurship-prone personality profile has a positive relationship with the regional entrepreneurial activity) can be supported only partially. With regard to start-up rate, the results are quite substantial, as the vast majority of regressions provide significant positive evidence for the existence of this relationship. The results regarding self-employment rate are less substantial, with only a few models reporting significant relationships.

(22)

Table 2: OLS regression results start-up rate

(23)

Table 4: OLS regression results self-employment rate

(24)

With regard to the moderating variable regional knowledge base, multiple measurements have been used to investigate this effect. The first measurement used was the regional presence of a university. Table 2 exhibits the results for the interaction effects of regional university presence. The results indicate that the presence of a university does not affect the strength of the relationship between the entrepreneurship-prone personality profile and the regional start-up-rate. In addition, table 4 shows the results for the same interaction effect, but then regarding the self-employment rate. These results also indicate that regional university presence has no significant influence on the relationship between the profile and the self- employment rate of a region. Besides university presence, the age of a university has also been considered.

Table 7 shows that the presence of a university founded before the year 1900 in a region strengthens the relationship between the profile and the start-up rate in that same region. This indicates that the presence of an older university, in conjunction with a higher share of entrepreneurial individuals, positively affects the start-up rate. For instance, model 40 indicates, at a significance level of 1%, that the presence of such a university in a region, in conjunction with a single unit increase in profile score, increases the start-up rate with 62.424 start-ups per 10.000 working population. In corroboration of this, a similar effect is found when explaining the self-employment rate. Table 8 shows that, albeit the insignificance of the profile, the moderating variable of regional university presence before 1900 is positively significant at respectively the

(25)

5% and the 1% level. The models indicate that in regions where an older university is located, the effect of the entrepreneurship-prone personality profile on self-employment rate increases with respectively, 4,6%, 8,6% and 9,2%. Lastly, the effect of the size of a university has also been investigated. Table 10 (in the appendix) exhibits that the presence of a large university does not significantly affect the start-up rate.

Although model 58 shows a strong positive significance, model 59 does not show such an effect, when controlled for the economic conditions of the region. Additionally, the presence of a small university did not seem to affect the relationship between the profile and the start-up rate at all (table 11, in the appendix).

With regard to the self-employment rate, the regression models turned out to be insignificant. Consequently, these results cannot be interpreted and have therefore been left out of this paper. Overall, the second hypothesis (H2: The positive relationship between the regional share of people with an entrepreneurship- prone personality profile and regional entrepreneurial activity (H1) will be enhanced in regions that accommodate a university.) must be rejected, as the models do not provide enough evidence to substantiate the hypothesis.

(26)

In addition to this, the effect of technical university presence in a region has also been researched. Table 3 shows that only a few models report a significant moderating effect when investigating its effect on the relationship between the profile and start-up rate. For instance, model 10 reports a significant negative effect at the 10% significance level, while models 11 and 12 do not report this significant result. The results in model 10 indicate that the positive relationship between the profile and the regional start-up rate is weakened in the presence of a technical university with an effect of -45.036 start-ups per 10000 working population. These results contradict the expected strengthening effect of regional technical university presence. Regarding the self-employment rate, none of the regressions that have been run yielded significant results, as can be seen in table 5. Furthermore, the effect of the presence of a technical university in a neighbouring region on entrepreneurial activity has also been investigated. These results are more substantial, as all three models in table 3 report a significant weakening effect at a 5% significance level.

This indicates that in regions that accommodate a technical university, and/or regions that border to such a region, a single unit increase in the entrepreneurship-prone personality profile leads to a less positive effect on the start-up rate, than in regions that do not accommodate a technical university nor border to a region that does. In model 16 for instance, the effect of a single unit increase in the profile on the regional start-up rate is weakened by 41.663 start-ups per 10.000 working population. A similar effect can be seen in models

(27)

17 and 18. When investigating the self-employment rate, table 6 shows that technical university presence within, or in a neighbouring region has a significant effect at the 10% significance level. For instance, model 34 reports a negative effect of -2.7% on the relationship between the profile and the regional self- employment rate. The last measurement used to measure the moderating effect of technical university presence is the age of the university. However, the results in table 9 indicate that the age of a technical university does not significantly influence the regional start-up rate. Regarding the self-employment rate, the models turned out to be insignificant. Therefore, these results cannot be interpreted and are thus not presented. The results show, opposed to the expectations, that the relationship between the entrepreneurship-prone personality profile and entrepreneurial activity is affected negatively by the presence of a technical university. However, this effect is less geographically bounded, as it is almost exclusively significant in case the neighbouring regions are incorporated in the regression. Overall, the results do not provide evidence to support the third hypothesis (H3: The positive relationship between the regional share of people with an entrepreneurship-prone personality profile and regional entrepreneurial activity (H1) will be stronger in regions that accommodate a technical university, than in regions that accommodate a classical university.), as technical university presence is shown to weaken the relationship between the entrepreneurship-prone personality profile and entrepreneurial activity.

(28)

Discussion and conclusion

In this chapter, the results of this paper will be discussed and conclusions will be drawn. First, the research questions of this study will be answered and a brief summary of the results will be given. After that, the theoretical and practical implications of the findings of this paper will be addressed. This chapter will end with a section on the limitations of this paper. This section will also include directions for further research.

Discussion

In line with the aim of this study, the present study has investigated the relationship between the entrepreneurship-prone personality profile and entrepreneurial activity. Additionally, the moderating role of the regional knowledge base in this relationship has also been examined. In order to conduct this research, two research questions have been formulated. The first research question is aimed at investigating the distribution of the entrepreneurship-prone personality profile in the Netherlands and its effect on entrepreneurial activity: “Is there a regional clustering of the entrepreneurship-prone personality profile in the Netherlands and does this clustering affect regional entrepreneurial activity?”. The second research question is formulated to investigate the moderating effect of regional knowledge base in the relationship pointed out in the previous question: “Does a strong regional knowledge base enhance the positive relationship between the regional population’s entrepreneurship-prone personality profile and regional entrepreneurial activity?”. To answer the first research question; the results of analysis clearly show that the entrepreneurship-prone personality profile is regionally clustered in the Netherlands. For example, COROP region Groot-Amsterdam (-18,61) scores significantly higher than COROP region Veluwe (- 19,95). This finding is in line with the existing literature on the entrepreneurship-prone personality profile, as Obschonka et al. (2013) and Fritsch et al. (2019) report similar findings for the distribution of this profile in other countries. Furthermore, the entrepreneurship-prone personality profile is found to positively affect the regional start-up rate, as most models that have been run show significant results that support this relationship. These results are in line with the findings of Obschonka et al. (2013) and Fritsch et al. (2019).

However, when investigating the profile’s effect on the self-employment rate, the results from the present study deviate from those found in similar articles like Obschonka et al. (2013) and Fritsch et al. (2019). The vast majority of the regression results indicate that the profile does not significantly affect the self- employment rate. A reason for this could be the lack of distinction between opportunity and necessity entrepreneurship. The choice of becoming self-employed might not always be a voluntary one, as individuals might be forced into entrepreneurship as a consequence of absent or unsatisfying employment alternatives (Angulo-Guerro, Pérez-Moreno & Abad-Guerrero, 2017). Therefore, it is not unthinkable that

(29)

in regions with less employment opportunities and/or -prospects, a relatively high number of less entrepreneurial people is forced into entrepreneurship. This could be a plausible reason why the entrepreneurship-prone personality profile in a certain region might be lower, while at the same time exhibiting relatively high levels of regional self-employment. Another reason for this unexpected result could be related to the share of self-employeds that is active in the agricultural sector. For example, Obschonka et al. (2013) stumbled upon a discrepancy in their data, as the start-up rate in Northern Ireland was relatively low compared to the self-employment rate. The explanation that they provide for this phenomenon is that this might be due to the importance of agriculture in this region. According to Obschonka et al. (2013) the agricultural sector often exhibits a high self-employment rate, even though the average agrarian is not prototypical for entrepreneurship. Since the self-employment rate used in this study is based on all industries as indicated by the CBS, and thus includes the agricultural industry, this might have affected the results of this paper.

To answer the second research question; the results of this paper do not provide enough evidence to support the positive moderating effects of regional knowledge base on the relationship between the entrepreneurship-prone personality profile and entrepreneurial activity. Regarding the start-up rate, the presence of a classical university founded before 1900 is, in line with Fritsch et al. (2019) found to positively affect the start-up rate, while technical university presence within a (neighbouring) region, is found to negatively affect this relationship. The results concerning the self-employment rate show a less significant, but similar effect. A possible reason for the lack of significance regarding the other variables of the regional knowledge base could be due to the high level of population density in the Netherlands. As a small, but densely populated and prosperous country, the Netherlands has one of the most developed economies of the world. Almost every COROP region in the Netherlands accommodates a university or borders to a region that accommodates one. Consequently, and albeit the boundedness of knowledge, it is easy for individuals to travel across COROP regions and still stay within the proximity of the knowledge generating institution. This could have inhibited an adequate measurement of the effect of regional knowledge base on the aforementioned relationship.

Theoretical and practical implications

This study contributes to the existing literature on the entrepreneurship-prone personality profile, by investigating whether the Netherlands sees a regional clustering of this profile like found in other countries that have been investigated before (Obschonka et al., 2013, Fritsch et al., 2019). The results of the analysis clearly show that the entrepreneurship-prone personality profile is regionally clustered in the Netherlands, with multiple regions showing deviating profile scores compared to others. In addition, this study provided evidence for the positive relationship between the entrepreneurship-prone personality profile and