UNIVERSITY OF AMSTERDAM FACULTY OF ECONOMICS AND BUSINESS
Entrepreneurial business performance: The
effects of personality characteristics,
self-other agreement and balance
Abstract
This thesis empirically studies the effect of personality on entrepreneurial business performance. The cross-sectional data-set consists of Dutch entrepreneurs and self-employed, most of them only just started a business. The relationship is assessed in three ways. First the effect of nine personality factors that link to the task of entrepreneurship on business performance is studied, but no significant relationships are found. Second the relationship between business performance and self-other agreement in these person-ality factors is investigated. To link self-other agreement to business performance is novel in entrepre-neurship research. However, the results show no consistent effects. Third a test of the Jack-of-all-trades theory of Lazear (2005) is performed. By using the number of job roles before entering in entrepreneurship as a balance measure, the theory was confirmed. The balance measure that was constructed from the personality characteristics did not turn out to have an effect on business performance.
Liselotte van Thiel Studentnumber: 0522090
Master: Business Economics, specialisation Organisation Economics
14th of August, 2014
Supervisors: Dr. J. Sol
Content
1. Introduction 1
2. Review of the literature and hypothesis development 2
2.1 Personality characteristics and competencies that link to business performance 2
2.2 Self-other agreement 6
2.3 The Jack-of-all-trades theory 8
3. Methodology and data description 11
3.1 Gathering of the data 11
3.2 Data description 12
3.3 Methodology of the analysis 19
4. Results 20 4.1 Main results 20 4.2 Robustness checks 29 5. Conclusion 31 5.1 Main conclusions 31 5.2 Discussion 33 References 35
Appendix 1: Questionnaire of the E-Scan (Driessen, 2005) - English translation 40
Appendix 2: Additional survey - English translation 41
Appendix 3: Initial participation E-mail and reminders 50
Appendix 4: Reliability and factor analyses for the ‘soft’ measure of success 53 Appendix 5: Derivation of the log likelihood function for the ordered logit regression 56
analysis
Appendix 6: Results of model 2(ii), 3(iii) and 3(iv) 58
1. Introduction
Entrepreneurship is a hot topic. As Europe slowly crawls out of years of recession, more and more peo-ple choose to become self-employed. For exampeo-ple, in The Netherlands the percentage of nascent entre-preneurs and new owner-managers has increased from 5.2% of the people aged between 18 and 64 in 2008, to 9.3% in 2013 (GEM, 2014). However, the number of bankruptcies has more than doubled when the figures of 2008 and 2013 are compared (CBS, 2014). Looking at European figures, a less extreme but similar image is sketched. The number of firms with less than 10 employees grew in 2009-2010 with 3.8%. Yet a decrease was found in 2011-2012 of 3.0%. This indicates an increase of both start-ups and business failures (Gagliardi et al., 2013). The increased rate of new business start-ups and failures draws a lot of attention to the subject of entrepreneurship. Entrepreneurship is said to be the motor of the econ-omy, as it stimulates economic growth and innovation (Audretsch, 2004). This illustrates the importance of entrepreneurship research in general and, as is the focus of this thesis, research to the success-factors of entrepreneurs.
Hamilton (2000) found that both an entrepreneur’s initial income and income growth is lower than what he would have gained in paid employment. So why would anyone engage in entrepreneurship? These lower earnings suggest a compensation by non-pecuniary benefits. It seems that these people choose to be an entrepreneur because of other reasons, for example ‘being your own boss’, greater free-dom at work, or because of personality factors (Hamilton, 2000; Walker and Brown, 2004). Moreover, according to Holland (1985), individuals choose those occupations that match their personality. There-fore they will look for an occupation with a high level of fit between work environment and their char-acter. This results in higher job satisfaction and higher performance (Holland, 1985). The influence of personality on business performance is the main topic of this thesis. Therefore, the first research question concerns the effect of personality traits and competencies on business performance.
Business performance is measured in several different ways: by the gross income of the entre-preneur, the operating profit of his firm, and the size of the firm. Additionally a soft measure of success is used because there is debate in the literature whether financial measures are the best way to measure business success (Sternberg, 2004; Walker and Brown, 2004). An entrepreneur is defined in this thesis as someone who has founded and/or owns a business.
The personality traits and competencies of entrepreneurs are assessed by the E-Scan of Driessen (2005). Since most studies only use self-assessed measures, the advantage of this model is that it not only contains the self-assessed personality characteristics and competencies, but also the view of others who knew them well. As Yammarino and Atwater (1993) argue, self-assessments are unreliable and the comparison with the view of others can give a more complete picture.
Several scholars have found that those individuals that have ratings that are ‘in-agreement’ with the ratings of others, are more effective and perform better in the workplace (e.g. Atwater et al., 1998; Bass and Yammarino, 1991; Fleenor et al., 1996). This thesis will extent this reasoning to the business performance of entrepreneurs, and so the second research question concerns the relationship between business performance and self- and other-ratings. To the best of my knowledge, this relationship has not
been studied before.
The above discussed research questions address personality traits and competencies in isolation, but Lazear’s (2005) Jack-of-all-trades theory (JAT) proposes that entrepreneurs have to be multifaceted and should possess a balance in skills. Balance might contribute to entry in entrepreneurship, because if someone is a generalist, his income will be maximised as an entrepreneur. While for a specialist it is more beneficial to be in paid employment (Lazear, 2005). The second implication of the JAT theory - that income increases with the level of balance in skills - is tested. Balance in entrepreneurial compe-tence, as assessed by balance in the personality traits of the E-Scan, is used as a proxy for balanced skills. Skill balance and balance in personality traits and competencies are not the same, but an entrepreneurial personality could correlate to a balanced skill-set (Stuetzer et al., 2013B). Additionally a more traditional proxy for balanced skills: the number of job roles before entering in entrepreneurship, is used to test the income implication of the Jack-of-all-trades theory.
The practical implications of this study lie in several fields. Entrepreneurs can benefit by an assessment that could indicate whether they have an entrepreneurial personality. The effect of feedback might increase self-awareness of the entrepreneur and might improve personal development. Hence, the results could be important for training purposes, entrepreneurial education, career counselling and gov-ernments that wish to stimulate entrepreneurship. Furthermore, the self- and other assessments combined may help investors in risk analyses for (starting) entrepreneurs.
This thesis is outlined as follows: the next section discusses the related literature and the devel-opment of the hypotheses. The third section describes the data-set and methodology. The results are described in the fourth section and the main conclusions and discussion follow in the fifth section.
2. Review of the literature and hypothesis development
This chapter discusses the theory and previous empirical findings on the three subjects outlined above. First, entrepreneurial personality traits and competencies and their relation to business performance are discussed. Second, the subject of self-other agreement is addressed. Third, the Jack-of-all-trades theory is described.
2.1 Personality characteristics and competencies that link to business performance
A lot of research has been done to define the characteristics and competencies that determine whether someone will become a successful entrepreneur. At the end of the 1980s, several scholars concluded that a relationship between successful entrepreneurship and competencies and traits does not exist (for ex-ample Gartner, 1988; Low and MacMillan, 1988; Shaver and Scott, 1991). However, during the last decade many other researchers have found the contrary. Among others, Rauch and Frese (2000; 2007) and Zhao et al. (2010) find significant correlations between success and personality. These contradictory conclusions may be caused by mistakes in study design, methodology or by the use of hypotheses that lacked theoretical foundations in the earlier studies (Hisrich et al., 2007; Zhao et al., 2010). This chapter investigates which personality characteristics lead to success and will describe these characteristics.
Both the decision to become an entrepreneur and entrepreneurial performance are influenced by the personality characteristics and competencies of the individual (Rauch and Frese, 2007; Zhao et al., 2010). Personality traits or characteristics are defined as dispositions to show some kind of response in different circumstances (Caprana and Cervone, 2000; Rauch and Frese, 2007). Even though the terms competency and competence are often interchanged, they are not the same. A competence is the ability and willingness to perform a task, and a competency refers to the dimensions of behaviour that lie behind competent performance. Hence, competencies are needed to acquire competence in an area (Armstrong, 1998; Burgoyne, 1989; Moore et al., 2002).
Several authors have used general personality tests1 to investigate the relationship between busi-ness performance and personality (Ciavarella et al., 2004; Costa and McCrae, 1988; Schmitt-Rosermund, 2004; Zhao and Seibert, 2006; Zhao et al., 2010), however the results are inconsistent. Rauch and Frese (2000; 2007) and Driessen (2005) argue that general personality tests are not related to the content of entrepreneurship, and are therefore not suitable to investigate the relationship between entrepreneurial success and personality. Tests to study this relationship should conceptualize the characteristics and competencies an entrepreneur needs to operate his business successfully. For example, the E-Scan de-veloped by Driessen (2005) has only incorporated characteristics and competencies that are assumed to cohere to the ‘entrepreneurial competence’. In the view of Driessen, a competence is a mixture of knowledge, skills and abilities, motivation and personal characteristics needed to perform a task well. Since the E-Scan is used in this study, a more elaborate description of this model follows in section 3.1.
Several characteristics have been studied repeatedly in relation to entrepreneurship entry and success. These characteristics are - among others - need for achievement, need for autonomy, internal locus of control, risk attitude, creativity, and social competence. They show a positive relation with entrepreneurial business performance (Begley and Boyd, 1987; Crant, 1996; Fairlie and Holleran, 2012; Rauch and Frese, 2000; 2007). Below these five concepts are discussed briefly2. Yet, business perfor-mance might be dependent on other factors as well, for example on human capital attributes such as knowledge, skills, experience and education (Unger et al., 2011). The personality characteristics might be mediated or moderated by other factors, for instance business strategies, planning, and growth orien-tation (Baum et al., 2001; Rauch and Frese, 2007). It is beyond the scope of this thesis to discuss these in detail.
Need for achievement
There are several implications of having a high need for achievement. First, it results in having a prefer-ence for challenging tasks of moderate difficulty instead of preferring very difficult tasks or routine tasks. Consequently, individuals with high need for achievement set challenging goals. Second, they look for feedback on their performance in order to assess their goal accomplishment. Third, they continuously
1 Examples of general personality tests are the ‘Big Five personality test’ (Tupes and Christal, 1961; Digman and Takemoto-Chock, 1981) and the ‘Myers-Briggs Type Indicator’ (MBTI) (Myers-Briggs and McCaully, 1988)
2 The E-Scan consists in total of nine traits and competencies. It is beyond the scope of this thesis to discuss them all in detail. Nev-ertheless, results for all nine E-Scan measures will be reported and definitions of the variables can be found in section 3. An elabo-rate review of all traits and competencies of the E-Scan can be found in Driessen’s dissertation (2005).
search for ways to improve their performance (Begley and Boyd, 1987; McClelland, 1961; Rauch and Frese, 2000). As a consequence, individuals with a high need for achievement are more likely to plan for the future and they feel responsible for their task outcomes. A strong need for achievement impacts the choice of occupation and in particular, it might be an indicator for a preference for entrepreneurship (Collins et al., 2004; McClelland, 1961). Meta-analyses of Collins et al. (2004) and Rauch and Frese (2007) show that entrepreneurs have a higher need for achievement than other groups of people, and that need for achievement is a predictor of entrepreneurial performance.
Need for autonomy
Need for autonomy is often mentioned as one of the main drivers for entrepreneurship. An individual with a strong need for autonomy avoids restrictive environments and prefers to be independent in setting his goals, planning, and decision making. He is more intrinsically motivated by a job that contains these factors (Caliendo and Kritikos, 2011; Rauch and Frese, 2007). Rauch and Frese (2007) find that need for autonomy positively correlates with entrepreneurial performance.
Internal locus of control
Individuals who possess a high internal locus of control (as opposed to an external locus of control), believe they can influence their destiny, or simply put, they feel they can control what happens in their life. They have the belief that they determine their own performance, without being reliant on the actions of others. Consequently, if someone with a high internal locus of control chooses to become an entre-preneur, he will believe that his own actions will determine his business outcomes and therefore he will exert more effort and persistence. Theoretically, this will result in higher entrepreneurial success (Begley and Boyd, 1987; Rauch and Frese, 2000, 2007; Rotter, 1966). This notion is supported by Brockhaus (1980) who made a comparison between successful and unsuccessful entrepreneurs. Nevertheless, Begley and Boyd (1987) investigated the relationship between several traits and business success and differences between founders and non-founders. They show that the relationship between internal locus of control and success is not significant.
Internal locus of control should not be confused with self-efficacy, notwithstanding these con-cepts are closely related. Locus of control measures several areas at once, where (general) self-efficacy3 is task specific self-confidence. So even though someone has a high locus of control, self-efficacy can be low if he does not believe himself to be skilled in the tasks to be performed. However, high internal locus of control induces self-efficacy to evolve easier from past experiences (Bandura, 1977; Chen et al., 1998; Gist, 1987). In addition, many authors show (for example: Baum and Locke, 2004; Evans and Leighton, 1989; Boyd and Vozikis, 1994) that entrepreneurial self-efficacy has a positive effect on en-trepreneurial success and entry.
3General self-efficacy is a task specific knowledge of the own abilities and 'entrepreneurial self-efficacy' narrows this definition by including all roles and tasks that an entrepreneur has to perform (Chen et al., 1998; Zhao et al., 2005).
Creativity
Creativity is a competency that is important for entrepreneurial talent (Hisrich et al, 2007). An implica-tion of being creative is that the individual is able to create high quality, new ideas, it enhances flexibility and induces the individual to solve problems in ways that others do not see (Sternberg, 2004). Starting entrepreneurs often need creativity to come up with a business idea or plan, but it is important for (more) established entrepreneurs as well: to look for new opportunities and to solve problems that might arise. Amabile (1997) argues that creativity is a necessary factor to obtain success as an entrepreneur.
Even though creativity directly relates to opportunity identification, it is not the same as innova-tion: creativity is a way of thinking that forms new ideas, while innovation is implementing that new idea, by developing and exploiting this into an invention (Heunks, 1998; Hills et al., 1997; Ardichvilly et al, 2003). Heunks (1998) shows that innovation correlates to growth and efficiency in small firms and finds an indirect relationship between creativity and business performance in small and medium-sized firms.
Social competence
Social competence is a combination of factors, for example making a good first impression, persuasion and the ability to perceive others accurately. Compared to other groups of people, successful entrepre-neurs are more able to interact effectively with others, that is, their social competence is higher, their social perception is higher, and they are better able to adapt to new social situations. Having a high degree of social competence is beneficial for entrepreneurs because it allows them to better deal with other people - e.g. customers - and to constitute, sustain and exploit their social networks (Baron, 2000; Baron and Markman, 2003; Caliendo and Kritikos, 2011). Accordingly, Baron and Markman (2003) study the relationship between social competence and entrepreneurial financial success (gross income) in two industries, high-tech and cosmetics. They find that social competence, as measured by social perception and social adaptability, positively relates to entrepreneurial gross income in the cosmetics industry4. This holds for social perception and expressiveness in the high-tech industry. Because of the cross-sectional design, this study could be prone to reversed causality because the advantages of entre-preneurial success might improve somebody’s social competence (Baron and Markman, 2003).
Hypothesis 1
Several competencies and traits that match to the tasks of entrepreneurship, and relate to business success have been discussed in this section. The relationship between traits and competencies and business per-formance is tested by the use of the ‘E-Scan’, a test for entrepreneurs with nine different variables that cohere to the entrepreneurial competence. Based on Driessen (2005), the expectation is that all nine variables have a positive influence on business performance5. Hypothesis 1 is therefore:
4 Baron and Markman use a significance level of 5% and find that only social perception is significant. However they report that social adapatibility is significant at the 10% level.
H1: Business performance increases with the score on the nine different competencies and traits meas-ured by the E-Scan; which concern (i) need for achievement, (ii) need for autonomy, (iii) need for power, (iv) social orientation, (v) internal locus of control, (vi) endurance, (vii) market awareness, (viii) crea-tivity, and (ix) flexibility.
2.2 Self-other agreement
As mentioned in the introduction, most studies about entrepreneurial competence use self-assessed measures. However, the disadvantage of self-assessed measures is that many people are unable to accu-rately judge their own traits and abilities. For example, they might be unaware of their incompetence or they might underestimate their own skills. This thesis compares self-assessed measures to feedback given by others. The rationale behind using this feedback is that self-ratings are influenced by personality and certain cognitive biases. In other words, the self-assessed measures might not reflect the ‘true’ traits and competencies. Even though in certain situations the ratings of others (‘other-ratings’) could be more valid, they do not necessarily reflect the ‘true’ values either. They merely provide an extra source of information on the traits and competencies of the focal character (Ackerman et al., 1992; Atwater et al., 1998; Fleenor et al., 1996; Harris and Schaubroeck, 1988; Kruger and Dunning, 1999).
Self-other agreement is defined as the degree of congruence between the ratings of the focal character and other people. These ‘others’ usually are peers, superiors and subordinates; multisource assessment is most commonly used in company settings (e.g. 360 degree feedback) and regard the per-formance of employees. Typically the scores of others are averaged and presented to the employee as feedback instrument to create self-awareness about perceived strengths and weaknesses of the employee and to promote personal development (Atwater et al., 1998; Fleenor et al., 1996; Van Velsor et al., 1993; Yammarino and Atwater, 1993). Multisource rating assessments are less commonly used in entrepre-neurial settings and have been executed only a few times (Baron and Markman, 2003; Hoehn et al., 2002; Lans et al., 2010; Mulder et al., 2007).
Feedback given by others can be used to determine whether the self-rating scores are in-agree-ment, higher, or lower than given by ‘others’. The designation often used is: raters’ or
‘over-estimators’ rate themselves higher than others do. ‘In-agreement raters’ rate themselves more or less the
same as others and ‘under-raters’ or ‘under-estimators’ rate themselves lower than others (Van Velsor et al., 1993). Many scholars have found that self- and other-ratings often do not correlate very well with each other (e.g. Atwater et al., 1998; Fleenor et al., 1996; Harris and Schaubroeck, 1988). This disagree-ment can be twofold; either the focal character is (partly) erroneous in his ratings, or the other-rater is (partly) erroneous in his ratings (Atwater et al., 1998; Van Velsor et al., 1993). These errors can be caused by personality factors, cognitive biases, or both.
Personality factors that influence self-other agreement
The personality factors that are most often mentioned in the literature to influence ratings are self-esteem and self-awareness (Fleenor et al., 2010; Yammarino and Atwater, 1993). Brown et al. (2001)
found in an experimental study that self-esteem directly influences self-evaluations: people with high self-esteem assume that their own abilities are higher, people with low self-esteem regard their own abilities as lower.
Self-awareness is defined as the degree of knowledge of how a person is viewed by others. There-fore it stems from the individual’s ability to observe himself. Even though self-other rating agreement is often used as a proxy for self-awareness, there is some disagreement in the literature about the effect of self-awareness: some found that both under- and over-estimators have low self-awareness, this stems from the fact that the self-ratings differ from those of others (Atwater and Yammarino, 1992; Fleenor et al., 1996). While Van Velsor et al. (1992) found in a study among managers, that over-estimators have the highest values of self-awareness, and under-estimators have the lowest values of self-awareness.
In addition, there seems to be a causality problem concerning the effect of multi-rater feedback on self-awareness: multi-rater feedback is said to enhance self-awareness, yet some researchers found that self-awareness directly influences the focal characters ratings (Atwater and Yammarino, 1992; Fleenor et al., 1996; Van Velsor et al., 1992; Yammarino and Atwater, 1993). Thus, it appears that the relationship between self-other ratings and self-awareness is a vicious circle. Fleenor et al. (1996) argue that rater discrepancies might be caused by some other phenomenon such as arrogance or modesty.
Cognitive biases that influence self-other agreement
Another cause of self-other differences could be cognitive biases. The focal character might fall prey to several egocentric biases which induce inflated ratings. First, he might attribute positive outcomes to himself, while he attributes negative outcomes to others. This is also known as attribution theory or
self-serving bias. Second, the defensiveness effect can inflate the self-rating, because the focal character tries
to enhance his evaluation if he falls prone to this bias. This can be caused by a strong achievement motivation. The third egocentric bias which might influence the focal character is the above average
effect: the self-rater believes he is above average and this may inflate self-ratings (Atwater et al., 1998; Harris and Schaubroeck, 1988; Kruger and Dunning, 1999; Sosik and Megerian, 1999; Yammarino
and Atwater, 1993). Another bias that is often mentioned is overconfidence. Overconfidence means that
a person has an excessive degree of confidence, this causes him to overestimate his own abilities and competencies and therefore inflates self-ratings (Camerer and Lovallo, 1999; Kruger and Dunning, 1999).
A bias that induces the focal character to underrate his competencies is the false consensus effect. The focal character assumes that other individuals are just as competent as he is and therefore scores himself lower than he should (Atwater et al., 1998; Harris and Schaubroeck, 1988; Kruger and Dunning, 1999; Yammarino and Atwater, 1993).
The other-raters might be prone to several biases as well, for example leniency bias, where the evaluator inflates all ratings. Another possibility is halo error, when the ratings of the evaluator are based on his overall impression of the focal character instead of his performance on specific items (Hoyt, 2000;
Holzbach, 1978; Yammarino and Atwater, 1993). In short, many biases could be the cause of self-other differences, and not every individual will fall prone to the same biases.
Self-other agreement in entrepreneurship research
Most studies in large companies found that self-ratings are higher than other ratings, and that in-agree-ment raters are more successful in the terms of performance and effectiveness (Atwater et al., 1998; Bass and Yammarino, 1991; Fleenor et al., 1996; Kruger and Gilovich, 2004; Yammarino and Atwater, 1993). In the studies that used entrepreneurs, the relationship between self-other agreement and entrepreneurial performance have never been tested, and the results on discrepancy between self- and other-ratings are somewhat different than in large companies (however this could be caused by the use of very small datasets6). Baron and Markman (2003) concluded in their study about social competence that the mean ratings of the entrepreneurs do not differ from the ratings of others. Mulder et al. (2007) and Lans et al. (2010) investigate the role of self-other differences in entrepreneurial competence in the Dutch horticul-ture sector. They use a triangular approach: on top of self-assessments, both an expert and someone within the business assessed the entrepreneur. These studies found that entrepreneurs rate their compe-tencies lower than others do. Contrarily, Hoehn et al. (2002) concluded that entrepreneurs tend to over-rate themselves. They studied social competence of entrepreneurs and compared self-assessments to evaluations of social competence by specialised outsiders. They argue that an inflated self-view is inher-ent to inher-entrepreneurship, because it enhances the propensity to undertake economic activities. A field that is more often studied in relation to entrepreneurship is overconfidence. Many studies show that entre-preneurs are overconfident, and as posited before, this could induce inflated self-ratings (for example: Baron, 2000; Forbes, 2005; Koellinger et al., 2007).
Hypothesis 2
The main finding in the literature is that in-agreement raters are found to be more successful in perfor-mance and effectiveness, because in-agreement raters might have higher self-awareness and might be less susceptible for certain cognitive biases. In the case of entrepreneurs, especially overconfidence might be relevant. Yet, to study the underlying constructs is beyond the scope of this thesis and therefore only the relationship between self-other agreement and business performance will be studied. The fol-lowing hypothesis is constructed:
H2: Business performance increases with a higher level of ‘agreement’ between self- and other ratings
2.3 The Jack-of-all trades theory
While section 2.1 discussed the findings of the importance of several traits and competencies for
6The size of the datasets used in these studies are: Baron and Markman (2003): n=15. Lans et al. (2010): n=36. Hoehn et al. (2002): n= 66 and Mulder et al. (2007): n=10.
preneurship in isolation, the Jack-of-all-trades (JAT) theory put forth by Lazear (2005) argues that en-trepreneurs should possess a balance in skills.
The JAT-theory proposes that individuals with balanced skills (generalists) are more likely to become an entrepreneur and will profit more from being an entrepreneur than those with unbalanced skills. Individuals that excel in only one skill (specialists) benefit more from being in paid employment. Lazear (2005) derives his predictions from a model which shows that the specialist’s income is deter-mined by his strongest skill (his specialisation) and that the income of an entrepreneur is dependent on his weakest skill. Consequently, a (would-be) entrepreneur benefits by increasing the level of his weakest skill, in order to attain a more balanced skill set. An employee benefits by investing in his strongest skill, and thereby increasing his level of specialisation (Lazear, 2005). Lazear provides the following intuition: when an entrepreneur lacks some kind of skill or knowledge, he will need at least a basic level to be able to select and hire a skilled applicant. Accordingly, an entrepreneur would not profit from any specialised skill. Employees on the other hand need more specialist skills and do not profit from balanced skills.
Empirical findings of JAT’s implications
The first implication of the Jack-of-all-trades theory, that generalists are more likely to become entre-preneurs, has been confirmed in several studies. For example, Lazear (2004; 2005) found his theory to hold among Stanford MBA graduates. The probability of being an entrepreneur is found to increase with variety of occupational fields, measured by the number of roles performed in their work history, and with the number of educational fields before entering the labour market. Wagner (2003) confirms the JAT hypothesis by using the changes of occupational fields and the amount of professional training received in a probit regression for being self-employed. Other scholars find as well that balanced skills seem to increase the likelihood of becoming an entrepreneur (Åstebro and Thompson, 2007; Hsieh et al., 2011; Lechman and Schnabel, 2014; Oberschachtsiek, 2009; Stuetzer et al., 2012).
However, other studies do no support these findings. Silva (2007) shows that these findings might be driven by individual unobservable variables: when panel techniques are used instead of cross-sectional data, skill variety does not seem to matter for the probability of becoming an entrepreneur. Also Lechman and Schnabel (2014) find limited support for the JAT theory. In a cross-sectional analysis they show that the probability of being self-employed increases with the number of changes of profes-sion, but only for solo entrepreneurs. That is, entrepreneurs without business partner or employees. They found no relationship between the number of professional trainings and the probability of being self-employed. Moreover, Brixy and Hessels (2010) find in a study among German and Dutch nascent entre-preneurs, that being a generalist even reduces the likelihood of a successful start-up. Thus, even though the first implication of the JAT theory has been confirmed in several studies, the overall findings in the literature seem to be mixed.
The second implication of the JAT theory, that income is higher when an entrepreneur has more balanced skills, has not been tested very often and the results are mixed as well. Hartog et al. (2010) use several ability measures and show that entrepreneurs with more balance in these abilities have a higher income. Bublitz and Noseleit (2014) find that income is only higher for those with a high skill level, but
not for those with balanced basic skills. And Åstebro and Thompson (2011) find that the probability of becoming an entrepreneur increases with a varied skill set, but this varied skill set has a negative effect on entrepreneurial income. A problem with the use of income is that the wage premium might be de-pendent on self-selection into entrepreneurship based on low or high ability (Hamilton, 2000). Hartog et al. (2010) use panel data and a difference-in-difference approach, to control for self-selection effects that might arise because of time invariant unobserved ability or other unobserved characteristics. Bublitz and Noseleit (2014) and Åstebro and Thompson (2011) cannot control for this effect or for unobserved var-iables. However, Bublitz and Noseleit use a proxy for unobserved variables in a robustness check which supports their findings. Åstebro and Thompson study the effect of variety on income, arguing that variety might be less susceptible to problems as self-selection, wealth constraints and learning. Altogether, there is some support in the literature of the income implication of the Jacks-of-all-trades theory.
An important critique on the JAT theory follows from these findings: an entrepreneur might benefit from specialists skills and an employee might benefit from balanced skills. Hessels et al. (2014) argue that specialist skills for entrepreneurs might enhance innovation and the entrepreneur might profit from specialisation when the demand for specialists goods is larger. Furthermore, they find that skill variety and not skill balance improves start-up success and that this relationship is moderated by both process- and product innovation. Skill variety is not the same as a balanced skill set per se, because it allows the entrepreneur to be specialised (Hessels et al., 2014). This notion is supported by Lechman and Schnabel (2014) who find that entrepreneurs not only need more basic skills, but they also need a higher number of expert skills than employees. Moreover, Bublitz and Noseleit (2014) find that entre-preneurs benefit the most from skill balance, but employees of small businesses benefit from balanced skills as well.
Hypothesis 3
In this study the second implication - the more balanced an entrepreneurs’ skills, the higher his income - will be examined as a test for the JAT theory. One might argue that if income increases because of balance, this could also account for operating profit, and for other measures of business performance. Therefore, instead of only studying entrepreneurial income, several other business performance measures are used, as is described in the following section. As a proxy for having a balanced skillset two variables are used: (i) balance in entrepreneurial competence, that is, balance in the nine traits and com-petencies of the E-Scan, and (ii) the number of job roles before entering in entrepreneurship. Balance in entrepreneurial competence is not a direct measure of skill balance. However, one could argue that en-trepreneurial competence closely relates to the concept of enen-trepreneurial skills, because a competence is defined as both the ability and willingness to perform a task. Following the JAT theory, the expectation is that both balance measures relate positively to business performance and therefore the hypotheses are:
H3(i): Business performance increases with balance in entrepreneurial competence H3(ii): Business performance increases with the number of job roles
3. Methodology and data description
3.1 Gathering of the data
The E-Scan
The data-set used in this thesis consists of data retrieved from the ‘E-Scan’ and an additional data-set acquired by an extra survey. The E-Scan is developed and provided by the company 'Entrepreneur
Con-sultancy’ and it is a self-assessment test developed to measure 'entrepreneurial competence'. According
to Driessen (2005), a competence is a mixture of knowledge, skills and abilities, motivation, and personal characteristics needed to perform a task well. In his dissertation he has distinguished several character-istics and competencies that determine someone's entrepreneurial competence. Entrepreneurial compe-tence can be seen as an indicator of the future success of an entrepreneur. The E-Scan consists of nine items; three of which are competencies or skills (flexibility, creativity, and market awareness) and six of the items are personality traits (need for achievement, need for autonomy, need for power, social orien-tation, internal locus of control and endurance). All items were initially measured on a scale between 0 and 100, but are divided by 10 to facilitate interpretation (i.e. the items are rescaled to 0-10). Personality traits are assumed to be stable over time, while competencies are subject to change (Driessen and Zwart, 1999; 2006; Driessen, 2005). The E-Scan questionnaire is displayed in appendix 17. An overview of the
definitions of the competencies and traits is given below.
Need for achievement is the individual drive to perform as well as possible, it is the willpower to achieve self-set goals and to keep improving these goals (Driessen, 2005; McClelland, 1961). Or in the words of Armstrong (2009, p. 325) it is 'the need for competitive success measured against a personal standard of excellence'.
Need for autonomy is the need to be able to make decisions independently and to do what the person himself deems valuable (Driessen, 2005).
Need for power is the need for control and power over others. It means wanting to impose your own will and to influence others (Driessen, 2005; McClelland, 1961).
Need for affiliation (social orientation) is the extent to which someone is oriented towards other people. It contains the notion that entrepreneurs need people and networks to realise their ideas and to what extent they are able to make business contacts and to exploit them in their favour, therefore it relates to social competence. Need for affiliation is referred to by Driessen (2005) as social orientation.
Internal locus of control (effectivity) is the perceived ability of an individual of their own capacities and self-esteem in their relationship with others. It is the perceived ability of being able to achieve ones goals without help from others and the perception that rewards are contingent on the indi-vidual's own behaviour (Begley and Boyd, 1987). Driessen (2005) calls this ‘effectivity’. Endurance is being able to continue ones actions on both short and long term, despite any obstacles
and misfortune that can be encountered (Driessen, 2005).
Market awareness is being able to understand and know your customers and potential customers. Fur-thermore it contains the ability to incorporate this information into the execution of business activities (Driessen, 2005).
Creativity is a way of thinking that enables people to form innovative ideas. It enables people to think out of the box and to see and try new opportunities. A sub-factor of creativity is ‘risk taking propensity’, defined by Driessen (2005) as the ability to cope with uncertainty combined with the willingness to take losses. Since risk taking propensity is not measured as a separate factor in the E-Scan, it will only be used in the robustness checks.
Flexibility is being able to react and adapt to changes in the environment. Flexible entrepreneurs can quickly adapt their way of doing business to new situations, for example to a new competitor or a change in market demand (Driessen, 2005).
More than 400,000 entrepreneurs in The Netherlands and abroad have completed the E-Scan survey to form an idea about their strong and weak points in running a business (http://www.entrepreneurconsul-tancy.nl). The data-set contains both starting and more experienced entrepreneurs, see Table 1 for de-scriptive statistics of the sample.
Survey method
The data consists of the already existing dataset of the E-Scan and information gathered by an additional survey send out in September 20128. The survey is displayed in appendix 2.
In total, 550 people have been approached for the additional survey. These people have com-pleted the E-Scan between January 2010 and June 2012. This group has been approached by e-mail, with the request to fill out the additional survey. The initial e-mail and reminders are shown in appendix 3, the e-mails have been personalised by using the recipient’s name in the header and included a personal code to be able to link the data of the E-Scan to the data of the survey. Two reminders have been sent with intermezzos of four weeks. The response rate after the initial e-mail was 10%, the additional re-sponse rates after the first and second reminder are respectively an additional 6.3% and 5.8%. Hence the final response rate is 22.18%, corresponding with a total of 122 respondents. Twenty-one people had to be omitted because of practical reasons9. Consequently, the dataset consists of 101 respondents. The next
section continues by showing some descriptive statistics of the dataset and a description of the main variables.
3.2 Data description
In this section the business performance measures are described, followed by the measures for self-other
8This survey gathered more information than will be used for this thesis, for additional purposes. The survey consists, among oth-ers, of the subjects: general information of the respondent, business performance measures, work experience, role models (based on questions of Bosma et al. (2012), and estimation of future success (based on questions of Cooper et al. (1988))
9 They have been omitted from the data because they either did not give permission to link their results of the E-Scan to the second questionnaire, or they indicated not to be a (starting/ex-) entrepreneur.
agreement and the skill balance measures. Table 1 presents an overview of the variables: means, standard deviations, medians, minima and maxima are displayed. For gross income, operating profit, firm size and age, frequency tables are given and for gender, age, gross income and operating profit a comparison is made with the population of Dutch entrepreneurs. The most striking results of the table are discussed with the corresponding subject.
Business performance
In this research operating profit and gross income of the entrepreneur are used as financial performance measures. The advantage of these measures is that they do not require any calculations by the entrepre-neur because –presumably- the entrepreentrepre-neur knows them by heart. Additionally, as non-financial measures I use firm size and a ‘soft’ measure of success.
Entrepreneurial gross income is measured as the income before taxes and social benefits, but including bonuses, prizes, and other payments. According to the Global Entrepreneurship Monitor (GEM)10, income can be measured by the use of a multiple choice question with classifications based on
the country's average entrepreneurial income in Euro (GEM, 2012). The main advantage of using clas-sifications for income is that an entrepreneur might be more willing to answer this question honestly as opposed to an open-ended question about income. As shown in table 1, the average income of the Dutch entrepreneur was €34,800 (CBS, 2013). Following GEM, bands with a spread of €10,000 are created in the survey.
Operating profit is measured as the return before taxes of the business, minus all costs made to
achieve the return. The same strategy in compiling the classifications for operating profit has been used as for entrepreneurial income. The average operating profit in The Netherlands in 2010 was € 57,460.40 (CBS, 2012). Categories are made with a spread of €10,000, for which the spread increases when oper-ating profit exceeds €100,000. As band spreads are difficult to use in ordinary least squares regressions, extra variables are created for both gross income and operating profit by using the middle of the range of each respondents answer. These are addressed in table 1 as ‘gross income (in Euro)’ and ‘operating profit (in Euro)’. Since several outliers are present, gross income in Euro and operating profit in Euro are transformed to logarithms to be used in the OLS regressions.
Firm size is measured as the number of full time equivalents (FTE) employed, including the
entrepreneur. Firm size is composed of the number of FTE plus the number of hours worked by the entrepreneur transformed to FTE. For example, an entrepreneur that works twenty hours per week be-comes 0.5 FTE, and working forty hours or more bebe-comes 1 FTE. Table 1 shows that 70.1% of the sample falls in the category of 0-1 FTE and 92.2% has less than 5 FTE. This might follow from the fact that the sample consists mostly of starting entrepreneurs. As for operating profit and gross income in Euro, firm size is transformed to logarithms in the OLS regressions to control for outliers.
Additionally, a ‘soft measure of success’ is used as an alternative way to assess business perfor-
10 GEM is a global research project that makes cross-national, annual assessments on entrepreneurs' activity, aspiration and atti-tudes.
mance. Because in the economic literature it is questioned whether financial measures for entrepreneurial performance are truly the best way to measure the success of a business (Hamilton, 2000; Sternberg, 2004; Walker and Brown, 2004). For example, Walker and Brown (2004) argue that financial factors are not the most important drivers to start up and continue a business, especially in the small
business sector. For most entrepreneurs factors as personal satisfaction, a flexible lifestyle, pride in their work and personal achievement are more important than wealth creation. Therefore, a ‘soft measure of success’ is created by the use of exploratory reliability and factor analyses. An elaborate description of these analyses can be found in appendix 4. The soft measure consists of four factors. The two factors
Table 1: Descriptive statistics of the main variables
Traits and competencies Mean SD Median Min Max n Difference scores Mean SD Median Min Max n
Need for achievement 7.97 1.18 8.15 4.45 10.00 101 ΔNeed for achievement 0.17 1.21 0.19 -2.92 4.85 78 Need for autonomy 6.45 1.20 6.67 3.06 9.44 101 ΔNeed for autonomy 0.53 1.57 0.51 -3.88 4.44 78 Need for power 6.02 1.38 6.00 2.54 9.34 101 ΔNeed for power 0.54 1.49 0.62 -4.40 5.00 78 Social orientation 7.65 1.34 7.86 3.57 10.00 101 ΔSocial orientation 0.01 1.28 0.12 -0.38 2.62 78 Internal locus of control 6.99 1.27 6.94 3.06 10.00 101 ΔInternal locus of control 0.40 1.46 0.30 -3.42 3.62 78 Endurance 7.69 1.18 7.71 4.58 10.00 101 ΔEndurance -0.13 1.60 -0.42 -3.61 7.08 78 Market awareness 7.13 1.43 7.50 3.75 10.00 101 ΔMarket awareness 0.28 1.68 0.22 -4.16 5.00 78 Creativity 8.24 1.23 8.41 4.66 10.00 101 ΔCreativity 0.40 1.31 0.38 -3.26 4.51 78 Flexibility 7.54 1.01 7.62 4.29 9.52 101 ΔFlexibility 0.46 1.20 0.24 -3.09 3.69 78 Average of E-Scan 7.30 0.88 7.38 4.93 9.52 101 Average difference scores 0.30 0.98 0.31 -0.21 4.03 78 Average Feedback 7.07 0.80 7.13 4.76 9.28 78
Coefficient of variation 0.16 0.05 0.15 0.07 0.32 101 5.53 4.65 4 1 30 100
Mean SD Median Min Max n Control variables Mean SD Median Min Max n
Gross income 3.93 2.98 3.00 1.00 17.00 71 Age 41.98 9.57 42 23 62 101 Gross income (Euro) 24,859.15 28,820.73 15,000.00 0.00 150,000.00 71 Education (years) 14.40 1.89 15 6 16 101 Operating profit 4.58 4.08 3.00 1.00 18.00 71 Company age (years) 3.21 3.8 1.9 0.25 20 72 Operating profit (euro) 50,070.42 112,489.70 15,000.00 0.00 75,000.00 71
Firm size 2.46 4.55 1.00 1.00 33.00 77
Soft measure of success 0.00 1.00 0.23 -3.24 1.71 70 Hours worked 32.00 12.03 40 0 40 77
Firm size Nr. FTE Frequency: Percentage: Dummies
0-1 54 Male 59.4% 1-2 9 Under-estimators 35.9% 2-3 6 Over-estimators 30.8% 3-4 2 Solo entrepreneur 59.7% 4-5 0 Age <25 5-10 2 25-34 10-20 3 35-44 >20 1 45-54 55-64 >65 Operating profit
In Euro (category) In Euro (category)
<0 (1) 9 9 12.7% 0-10,000 (2) 20 0-10,000 (2) 25 35.2% 10,001-20,000 (3) 9 10,001-20,000 (3) 4 5.6% 20,001-30,000 (4) 13 20,001-30,000 (4) 8 11.3% 30,001-40,000 (5) 6 30,001-40,000 (5) 7 9.9% 40,001-50,000 (6) 4 40,001-50,000 (6) 5 7.0% 50,001-60,000 (7) 4 50,001-60,000 (7) 1 1.4% 60,001-70,000 (8) 2 60,001-70,000 (8) 1 1.4% 70,001-80,000 (9) 0 70,001-80,000 (9) 1 1.4% 80,001-90,000 (10) 0 80,001-90,000 (10) 2 2.8% 90,001-100,000 (11) 1 90,001-100,000 (11) 0 0.0% 100,001-110,000 (12) 1 100,001-150,000 (12) 3 4.2% 110,001-120,000 (13) 1 150,001-200,000 (13) 1 1.4% 120,001-130,000 (14) 0 200,001-250,000 (14) 1 1.4% 130,001-140,000 (15) 0 250,001-300,000 (15) 0 0.0% 140,001-150,000 (16) 0 300,001-500,000 (16) 2 2.8% >150,000 (17) 1 500,001-750,000 (17) 0 0.0% >750,000 (18) 1 1.4%
* These figures are taken from CBS (2013), and concern age and gender of the Dutch entrepreneurs in 2008
0.0% Business performance variables Gross income 70.1% 11.7% Frequency: 3.9% 2.6% 0.0% 7.8% 2.6% 1.0% 23.8% 2.6% 35.4% 2.8% 8.5% 34.7% 29.7% 22.7% 12.7% Percentage: 28.2% 12.7% 18.3% Frequency: Percentage: 33.0% 0.0%
Mean of operating profit of Dutch entrepreneurs in 2010
(CBS, 2013) (in Euro) 57,460.40
Mean of gross income of the Dutch entrepreneurs in 2010
(CBS, 2013) (in Euro) 34,800.00 <0 (1) 0.0% 0.0% 0.0% 1.4% 0.0% 0.0% 1.4% 1.4% 1.4% 5.6% 5.6% 10.9% 5.3% 1.3%
Sample percentage: Population percentage*:
68.6% 0 0 30 101
Job roles
Entrepreneurial work-
with the highest factor loadings are: ‘success’ and ‘content’11. Both items are measured on a 7-point
Likert scale. The item ‘Success’ was queried by the question: ‘Do you feel your business is successful?’ and ‘Content’ by the question: ‘How content are you with your company?’. The other two factors regard ambition and can be found in appendix 4. Based on the factor analysis, regression values for the soft measure of success are calculated and they have a mean of 0 and a standard deviation of 1 as is shown in table 1.
Self-other agreement measures
The E-Scan is a self-assessment test. As argued in section 2.2, self-assessment tests have some disad-vantages, because people's view of themselves does not always reflect their actual behavior (Kruger and Gilovich, 2004). Additionally, the prediction of other people could be used as an extra source of infor-mation (Dunning et al., 2004). In total, acquaintances of 78 respondents have filled in the E-Scan survey concerning the entrepreneur, this is referred to as ‘feedback’. The number of feedback ranges between 1 and 4 per respondent.
To start, the differences between the average feedback and the own results are calculated, the so called ‘self-other-differences’ (Atwater and Yammarino, 1992). Table 1 shows the descriptives of these difference scores. Endurance has a negative mean and median, so apparently most people underrate themselves on this subject compared to others. However, endurance also has the largest maximum dif-ference score (70.8). Most people seem to overrate themselves on the matter of need for autonomy and need for power since the highest values of mean and median are found for these difference scores. Ad-ditionally, by the use of a paired sample t-test it is tested whether the focal character’s results indeed differ from the average feedback. For need for autonomy, need for power, internal locus of control, cre-ativity and flexibility this difference is significant at the 5% level.
To test whether business performance is higher for those entrepreneurs that have a higher degree of agreement with others, the absolute difference scores are used. Simply using the difference scores would result in negative values for under-raters, and higher values for over-raters than for in-agreement raters. Thus, using absolute scores causes those with a high level of agreement to obtain a low value, and those that either over-rate or under-rate themselves to obtain a high value.
The above method does not distinguish between over-raters and under-raters, since they both obtain high values in the absolute difference scores. Therefore agreement categories are created: ‘in-agreement raters’, ‘over-estimators’, and ‘under-estimators’. Following Fleenor et al. (1996), Van Velsor et al. (1993), and Yammarino and Atwater (1993), a comparison is made between the standardized scores of the E-Scan and the standardized average feedback. The advantage of standardizing these scores is that it eliminates possible asymmetrical effects that can be caused by unequal variances and thus corrects for differences in response style between the focal character and the other-rater (Fleenor et al., 1996). The following steps are taken to create the agreement categories: first the scores from the nine E-Scan vari-ables of the focal character are averaged to obtain an ‘average E-Scan score per respondent’. These
scores are standardized as z-values, so the sample mean becomes 0 with a standard deviation of 1. Sec-ond, the average feedback scores are standardized as z-scores in the same way. Third, the standardized self-ratings and standardized other ratings are compared: if the standardized self-rating is more than a half standard deviation (i.e. 0.5) above the standardized other ratings, the individual is categorized as ‘over-estimator’. Correspondingly, if the standardized self-rating is more than a half standard deviation under the standardized other rating, the individual is categorized as ‘under-estimator’. Otherwise, the individual is marked as ‘in-agreement rater’. Hereafter, dummy variables are created for ‘under-estima-tors’ and ‘over-estima‘under-estima-tors’. As is shown in table 1, 35.9% is categorized as under-estimator, 30.8% as over-estimator and 33.3% as in-agreement rater. The over-estimator dummy will be included in the re-gressions for absolute difference scores as an interaction term, to distinguish between over- and under-estimators.
Measures for skill balance
The measures that are used to test the hypotheses regarding the Jack-of-all-trades theory are described below. The number of job roles before becoming an entrepreneur is used as an indicator for balanced skills. As Lazear (2005) argues, acquiring many different job roles is a standard way of acquiring a variety of skills. The mean of the number of job roles is 5.53 and it has a rather large standard deviation of 4.65 and a maximum of 30. Balance in entrepreneurial competence is measured by the coefficient of variation of each respondents E-Scan outcomes. The coefficient of variation can be used as an inverse measure of balance. The advantage of using the coefficient of variation (CV) instead of the standard
deviation, is that it is a normalized value of the standard deviation: it is the standard deviation relative to the mean and therefore it is dimensionless (Abdi, 2010), and controls for skill level. A disadvantage of using the CV is that the figure might become infinite when the mean approaches zero and it can only be
used for measurements with real zero. As is shown in table 1, these disadvantages do not have to be taken into account because the scores of the E-Scan items have a real zero (they range from 0 to 100) and the means are not small (Abdi, 2010; Hartog et al., 2010). To facilitate the interpretation in the regression analyses, the inverse of the CV will be used:
𝐶𝑉𝑖𝑛𝑣 =𝑆𝑡𝑎𝑛𝑑. 𝑑𝑒𝑣. 𝑀𝑒𝑎𝑛1 ⁄ = 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛𝑀𝑒𝑎𝑛
Control variables
Several control variables are used in the regression. Standard controls as gender and age will not be further discussed. The other variables are briefly described.
Education is measured in years of schooling. The respondents indicated their highest type of
education. The nominal years of schooling necessary to finish the indicated type of education is used as education measure. The company’s age is measured in years just as entrepreneurial work experience. I chose to use entrepreneurial instead of general work experience, because entrepreneurial work experi-ence seems to be a better predictor for entrepreneurial success (Bublitz and Noseleit, 2014). Furthermore, general work experience correlates strongly to age (r=0.72), which is already included as control.
T a b le 2 : C o rr el a ti o n s Tr ai ts an d c omp ete n ci es S el f-oth er d iffe re n ce me as u re s Bal an ce me as u re s C on tr ol v ar iab le s (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) (8 ) (9 ) (1 0 ) (1 1 ) (1 2 ) (1 3 ) (1 4 ) (1 5 ) (1 6 ) (1 7 ) (1 8 ) (1 9 ) (2 0 ) (2 1 ) (2 2 ) (2 3 ) (2 4 ) (2 5 ) (2 6 ) (2 7 ) (2 8 ) (2 9 ) (3 0 ) (3 1 ) (3 2 ) (3 3 ) (3 4 ) G ro ss i n co m e (i n E u ro ) (1 ) 1.000 O pe ra ti n g p ro fi t (i n E u ro ) (2 ) 0.451 *** 1.000 F ir m s iz e (3 ) 0.418 *** 0.598 *** 1.000 S of t m ea su re o f su cc es ( 4 ) 0.125 0.162 0.097 1.000 N ee d fo r ac h ie ve m en t (5 ) 0.134 -0.070 -0.113 0.409 *** 1.000 N ee d fo r au to n om y (6 )-0.009 -0.134 -0.045 0.053 0.159 1.000 N ee d fo r po w er ( 7 ) 0.258 ** 0.130 0.105 0.156 0.503 *** 0.129 1.000 S oc ia l or ie n ta ti on ( 8 ) 0.050 -0.148 -0.210 * 0.139 0.449 *** 0.140 0.465 *** 1.000 In te rn al l oc u s of c on tr ol ( 9 ) 0.090 0.000 -0.079 0.288 ** 0.636 *** 0.233 **0.574 *** 0.492 *** 1.000 E n du ra n ce ( 1 0 ) 0.035 -0.045 -0.113 0.364 *** 0.698 *** 0.240 **0.387 *** 0.261 *** 0.540 *** 1.000 M ar k et a w ar en es s ( 1 1 ) 0.005 -0.079 -0.258 ** 0.278 ** 0.653 *** 0.222 **0.507 *** 0.486 *** 0.673 *** 0.613 *** 1.000 C re at iv it y ( 1 2 ) 0.118 -0.101 -0.144 0.102 0.561 *** 0.098 0.409 *** 0.372 *** 0.492 *** 0.375 *** 0.622 *** 1.000 F le xi bi li ty ( 1 3 ) 0.019 -0.061 -0.116 0.290 *** 0.617 *** 0.112 0.272 *** 0.395 *** 0.497 *** 0.488 *** 0.561 *** 0.582 *** 1.000 |Δ N ee d fo r a ch ie ve m en t| (1 4) -0.283 ** -0.142 -0.115 -0.077 -0.068 0.018 -0.147 -0.011 -0.096 0.111 0.002 -0.043 0.093 1.000 |Δ N ee d fo r au to no m y| (1 5) -0.309 **-0.171 -0.193 -0.006 0.035 0.164 -0.081 0.171 0.073 0.173 0.238 ** 0.083 0.203 * 0.325 ***1.000 |Δ N ee d fo r po we r| (1 6) 0.358 *** 0.245 * 0.066 0.004 0.082 -0.044 0.210 * 0.108 -0.038 0.088 0.105 -0.033 -0.050 0.141 0.075 1.000 |Δ So ci al o ri en ta tio n| (1 7) -0.170 -0.031 -0.012 -0.247 *-0.242 ** 0.043 -0.209 *-0.305 ***-0.165 0.044 -0.175 -0.252 **-0.130 0.247 **0.205 * 0.068 1.000 |Δ In te rn al lo cu s o f co n tr ol |( 1 8 ) |Δ En du ra nc e| (1 9) -0.200 -0.183 0.033 -0.145 -0.158 0.073 -0.098 -0.042 -0.076 -0.080 0.012 -0.179 -0.045 0.576 ***0.279 *** 0.071 0.276 *** 0.388 *** 1.000 |Δ m ar ke t a wa re ne ss | (2 0) -0.119 -0.090 0.028 -0.111 -0.043 0.010 0.018 -0.032 -0.034 0.078 -0.032 -0.089 -0.070 0.431 ***0.296 *** 0.155 0.482 *** 0.521 *** 0.526 *** 1.000 |Δ C re at iv ity | ( 21 ) 0.064 0.016 -0.009 -0.096 -0.078 0.053 0.012 -0.005 -0.064 0.105 0.010 -0.068 0.024 0.518 ***0.215 * 0.205 * 0.204 * 0.248 ** 0.360 *** 0.379 *** 1.000 |Δ Fl ex ib ili ty | ( 22 ) 0.053 -0.010 -0.046 0.054 0.225 ** 0.041 0.013 0.166 0.083 0.253 ** 0.257 ** 0.191 * 0.372 *** 0.458 ***0.186 0.128 0.051 0.235 ** 0.230 ** 0.168 0.503 ***1.000 U n de r-es ti m at or ( 2 3 )-0.209 -0.169 -0.100 -0.264 **-0.453 ***-0.194 *-0.285 ***-0.460 ***-0.448 ***-0.425 ***-0.409 ***-0.418 ***-0.362 *** 0.113 -0.238 **-0.138 0.077 0.013 0.143 0.121 -0.109 -0.233 ** 1.000 In -a g re em en t es ti m at or ( 2 4 ) 0.235 * 0.257 * 0.135 0.373 *** 0.262 ** 0.093 0.224 ** 0.175 0.352 *** 0.226 ** 0.245 ** 0.138 0.131 -0.413 ***-0.106 -0.203 *-0.226 **-0.260 **-0.266 **-0.314 ***-0.195 *-0.222 **-0.529 *** 1.000 O ve r-es ti m at or ( 2 5 )-0.013 -0.074 -0.030 -0.080 0.211 * 0.093 0.064 0.301 *** 0.105 0.219 ** 0.181 0.307 *** 0.256 ** 0.305 ***0.356 *** 0.351 *** 0.151 0.251 ** 0.124 0.194 * 0.312 ***0.469 ***-0.499 ***-0.471 *** 1.000 N u m be r of j ob r ol es ( 2 6 ) 0.132 -0.040 -0.131 -0.326 *** 0.107 0.035 0.279 *** 0.281 *** 0.202 **-0.005 0.199 ** 0.120 0.047 -0.156 -0.132 0.011 -0.012 -0.043 -0.226 **-0.144 -0.103 -0.049 0.057 -0.021 -0.036 1.000 In ve rs e co ef fi ci en t o f va ri at io n ( 2 7 ) M al e (2 8 ) 0.250 ** 0.255 ** 0.123 -0.220 *-0.116 -0.159 0.176 * 0.085 -0.092 -0.007 0.020 0.131 0.006 -0.047 -0.194 * 0.259 ** 0.010 0.131 0.012 -0.117 0.034 0.038 -0.082 0.047 0.048 0.160 0.097 1.000 A g e ( 2 9 ) 0.151 -0.032 0.036 -0.201 * 0.033 0.040 0.074 0.070 0.122 0.130 0.108 0.029 -0.032 0.134 0.007 0.086 0.069 0.122 -0.015 0.046 0.265 **0.056 -0.042 -0.011 0.049 0.268 *** 0.112 0.098 1.000 C om pa n y ag e ( 3 1 ) 0.207 * 0.396 *** 0.722 ***-0.010 -0.098 -0.168 -0.065 -0.278 **-0.161 -0.092 -0.227 *-0.074 -0.023 0.090 -0.197 0.176 0.084 0.131 0.211 0.145 0.067 0.019 -0.050 0.003 0.048 -0.174 -0.110 0.062 0.042 1.000 E du ca ti on ( in y ea rs ) (3 0 )-0.044 -0.068 -0.149 0.059 -0.036 0.101 -0.104 -0.001 0.024 -0.113 -0.076 -0.052 0.062 -0.142 -0.014 -0.035 -0.220 **-0.193 *-0.160 -0.132 -0.069 -0.104 0.138 -0.079 -0.061 -0.143 0.075 -0.190 *0.050 -0.169 1.000 E n tr ep re n eu ri al w or k * e xp er ie n ce ( 3 2 ) H ou rs w or k ed p er w ee k ( 3 3 ) 0.372 *** 0.250 ** 0.209 * 0.216 * 0.216 * 0.093 0.062 0.068 0.110 0.130 0.095 -0.072 0.180 0.014 -0.079 0.080 0.088 0.063 0.013 -0.022 -0.075 0.112 -0.238 * 0.120 0.125 -0.149 0.174 0.060 -0.122 0.070 -0.237 ** 0.015 1.000 S ol o en tr ep re n eu r (3 4 ) -0.322 ***-0.289 ***-0.393 ***-0.249 **-0.155 -0.026 -0.108 0.098 -0.139 -0.019 -0.004 0.011 0.037 0.141 0.263 **-0.132 0.029 -0.206 0.080 0.002 0.175 -0.025 0.109 -0.123 0.009 0.060 -0.199 *-0.113 0.008 -0.178 0.219 *-0.220 *-0.412 *** 1.000 *** s ig ni fic an t on th e 1% le ve l ** s ig ni fic an t on th e 5% le ve l * s ig ni fic an t on th e 10% le ve l -0.022 0.133 -0.017 -0.057 0.052 P er fo rm an ce m ea su re s 0.037 -0.203 0.013 -0.172 1.000 0.359 *** 0.136 0.108 0.035 0.132 0.213 * 0.211 * 0.057 0.061 -0.101 -0.007 0.605 *** 0.388 *** 0.507 *** 0.131 0.242 *** 0.416 *** * -0.075 0.233 ** 0.470 *** 0.162 0.307 *** 1.000 -0.037 -0.316 *** 0.237 ** 0.234 0.089 ** 0.117 -0.080 0.111 0.072 -0.087 -0.161 -0.053 0.136 -0.202 * 0.039 0.158 -0.205 0.172 * 0.177 0.124 0.112 -0.055 0.098 0.037 0.031 0.013 -0.009 -0.125 -0.013 -0.211 -0.023 0.011 0.242 ** ** 1.000 * 0.041 0.186 * 0.302 *** -0.239 0.155 -0.037 0.113 0.095 -0.157 0.012
The number of hours worked by the entrepreneur is measured per week and will not be included in the regressions for firm size because firm size already consists of the number of hours worked per week. Being a solo-entrepreneur might matter for the jacks-of-all-trades-theory, because if an entrepreneurs lacks a certain skill, his associate or a specialised employee might supplement it (Lechman and Schnabel, 2014). On the other hand, solo entrepreneurs might build their business around one single specialisation and therefore might be less balanced (Lazear, 2005; Hessels et al., 2014). Either way, both explanations call for including this dummy. The solo-entrepreneur dummy is one for those entrepreneurs that do not have a business partner and do not have any employees.
Correlations
Table 2 shows the correlations between the variables. The business performance measures correlate sig-nificantly with each other, except for the soft measure of success. The soft measure of success does correlate with several of the traits and competencies. The business performance measures in general do not correlate with many of the independent variables, only the absolute difference score of need for power and the in-agreement dummy do. The control variables gender, hours worked per week and solo entrepreneur also correlate significantly with most of the business performance measures.
The traits and competencies of the E-Scan have many significant correlations with each other. This might be caused by the fact that it is supposed to measure ‘entrepreneurial competence’. Addition-ally, several significant correlations are found between the agreement dummies and the balance measures. Also most of the self-other difference measures correlate significantly with each other. The number of job roles correlates significantly with the coefficient of variation. The table of correlations does not give strong support for the expected relationships between business performance and the inde-pendent variables.
Comparison to the population of Dutch entrepreneurs
In table 1 a comparison between the sample data and the population of Dutch entrepreneurs (CBS, 2013) is provided concerning the variables: gender, age, gross income, and operating profit12. Women are
over-represented in the current sample and the respondents seem to be slightly older compared to the popula-tion of Dutch entrepreneurs. The mean of the gross income (€ 24,859.15) and operating profit (€ 50,070.42) of the sample are lower than the figures of CBS as well. However the relative difference for operating profit is not as big as for gross income13 (CBS, 2013). These comparisons lead to the
conclu-sion that the dataset is representable for the population only to a certain extent. The differences might be due to several factors. In the first place, the sample might not be a complete random draw from the population, since the respondents have self-selected into doing the E-Scan. Consequently this research could be prone to sample selection bias. Second, the sample size is small. Third, the sample data consists
12 I have also made a comparison between the education level of the respondents and those of the Dutch labour force. As many other authors found, the respondents are higher educated than the Dutch labour force. (ao. Álvarez-Herranz et al., 2011; Cooper and Dunkelberg, 1987; Lee and Tsang, 2001; Thompson, 1986).
13 Concerning gross income, the respondents earn on average 28,57 percentage point less than the average Dutch entrepreneur, while the average operating profit is 12,87 percentage point lower.
