The impact of role models on female's selection into self-employment, and the moderating effect of education

The impact of role models on female's selection into

self-employment, and the moderating effect of

education

Astrid Dontje

University of Groningen Faculty of Economics and Business

Business Administration Small Business and Entrepreneurship

Supervisor: Dr. F. Noseleit Co-assessor: Dr. O. Belousova

|1

Abstract

Female self-employment rates are consistently lower than those of men, and there appears to be a difference in the rates of female self-employment across countries. Nonetheless, reasons for these country differences are to date still inconclusive. One explanation for differences in female self-employment rates between countries that has received almost no attention, is the influence of role models on the share of the female self-employed in a country. Especially for females, role models are particular important in making nontraditional occupational choices. In the present paper the influence of self-employed parents and female managers, both acting as role models, on the share of the female self-employed in a country is examined. Also attention is paid to whether the effects vary by educational level. Data from 24 countries that participated in the European Social Survey (ESS) are used to test the relationships. The results show that a high share of self-employed parents in a country has a positive impact on the amount of females that enter self-employment. Paternal role models, and not maternal role models, have the strongest influence on the share of the female self-employed. Moreover, the influence of self-employed parents is the strongest for those countries in which females have lower levels of formal education. With regard to the other type of role models used in this study, the female managers, results show that a high share of female managers in a country causes fewer women to enter self-employment. However, this relationship is not moderated by educational level. All in all can be concluded that female self-employment rates are to a large extent influenced by role models.

Key words: role models, female self-employment, female entrepreneurship, female

|2

Table of Content

2.1 Definition and importance of self-employment 8

2.2 Gender gaps in self-employment 9

2.3 Role models and female self-employment 10

2.4 Self-employed parents as role models 11

2.4.1 Social learning through self-employed parents 11

2.4.2 Same sex transmissions 12

2.4.3 Moderating role of education 13

2.5 Female managers as role models 14

2.5.1 Glass ceiling push effects 14

2.5.2 Moderating role of education 16

2.6 Conceptual model 17

3. Research Design 18

3.1 Research method 18

3.2 Data collection and sample 18

3.3 Measurements 19 3.3.1 Dependent variable 19 3.3.2 Independent variables 19 3.3.3 Moderator variable 20 3.3.4 Control variables 20 3.4 Data analysis 21

3.4.1 Assumptions Ordinary Least Square regression 21

3.4.2 Regression model 22

4. Results 23

4.1 Cross-country differences among self-employed 23

4.2 Descriptive statistics 24

4.3 Regression analysis 26

5. Discussion and Implications 29

5.1 Findings 29

5.2 Theoretical implications 31

5.3 Practical implications and policy implications 33

5.4 Limitations and future research 33

|3

References 35

|4

1. Introduction

1.1 Background

Women's labor force participation has grown substantially during the last decades (DiCecio et al., 2008) and this trend also appears in self-employment. There is an increase in women pursuing careers in employment, but still, men are more than twice as likely to be self-employed (Díaz & Welter, 2011; OECD, 2015). The objective of many countries to increase the number of self-employed females remains elusive since, despite focused support, women still constitute a minority of self-employed individuals throughout almost all economies (Baker et al., 2003). Stimulating female self-employment is important since they are argued to be important for economic development. They do not only contribute to employment creation and economic growth, they also make a contribution to the diversity of self-employment in the economic process (Verheul & Thurik, 2001). The rate of female self-employment varies, however, substantially between countries. According to Minniti et al. (2005), who researched female entrepreneurship across 34 countries, the female entrepreneurial activity rates range from 39.1% to 1.2%. Explanations for these country differences in female self-employment rates are up to now very inconclusive, and the justification is argued to be complex and multifaceted (Azmat et al., 2006).

|5

Two types of role models in nontraditional careers for females can be particular relevant when considering women's selection into self-employment: self-employed parents and female managers. First, Hisrich and Bursh (1984) showed that to stimulate women to see self-employment as a potential form of self-employment, they need persons in their environment who already are running their own business. At the individual level, self-employed parents, who serve as entrepreneurial role models, play the most important stimulating role for offspring to set up a business (Schmitt-Rodermund, 2004). Moreover, in order to demonstrate the importance of entrepreneurial parents for women, having entrepreneurial parents has been found to influence girls' entrepreneurial intentions but not boys' (Kickul et al., 2008). Second, besides the importance of role models who are already running their business, another type of role models, stimulating females to start up a business themselves, can be found in the occupation of females within organizations. There are still some formidable barriers that prevent women from making it through top management positions of organizations. One of the main reasons often cited by researchers for the lack of female managers is the absence of female managerial role models (Sealy & Singh, 2006). Korac-Kakabadse and Kouzmin (1997) highlight that this is due to a male dominance in managerial culture, which is historically attributable to the working population consisting only of men. This has become embedded within organizational cultures leading them to be constructed on masculine terms (Due Billing & Alvesson, 2000). As a result, women in organizations struggle to participate legitimately within those masculine ethics, which often precludes women from progressing to management positions (Due Billing & Alvesson, 2000). Females are therefore increasingly leaving their corporations to go it alone because of frustration at hitting the 'glass ceiling', the invisible barrier that block women from management positions in the corporate world (Terjesen, 2005). This glass ceiling can motivate women to start up a business themselves (Familoni, 2007).

However, from extant evidence can be concluded that human capital, and in specific education, plays a crucial role in the decision-making making process of individuals. Reynolds et al. (1999) state the importance of considering human capital when examining selections into self-employment. Human capital refers to the knowledge and skills acquired through formal and informal learning that resides within individuals (Becker, 1964). Especially education is found to be an important factor that needs to be considered in explaining employment decisions (Livanos et al., 2010). The effects of education on self-employment decisions are however contradicting; some argue that education has a positive effect on the selection into self-employment, while others argue for a negative relationship (Parker, 2004). As effects of education on self-employment are mixed, this research expect this relationship to be more complex. Therefore a model is suggested consisting of education in conjunction with self-employed parents and female managers, acting both as role models for females, to explain the share of the female self-employed in a country.

1.2 Research objective

self-|6

employed in a cross-country setting. This is relevant, since role models are particular important in (nontraditional) occupational choices for females. In countries with high levels of self-employed parents, females are exposed to entrepreneurial role models, which results in a start-up motive to follow the example of the self-employed parent(s). In addition, in countries with relatively few females in management positions, and thus a lack of female managerial role models, self-employed females have different start-up motives because of their possibly underlying desire to escape from the so called 'glass ceiling'. Moreover, the effects of role models may vary by educational level, and consequently, also attention is paid to the human capital of the females in a country. Aldrich and Kim (2007) discuss in their paper the need to investigate the extent to which education moderates the link between parental occupation and children's occupational destinations. They argue that it is unclear whether it is the human capital or the parental origins that enables people to enter self-employment, and that it is possible that those two work in conjunction. In addition, findings in the literature are not uniformly in favor of a strong parental role model effect; inconsistent results are reported from some studies (Brenner et al. 1991; Kim et al. 2006). These studies do not explain why some individuals follow parental role models while others do not and it therefore appears that the effect of parental role models on self-employment decisions is not universal, but may be more complex than assumed (Chlosta et al., 2012). Combining this with the arguments of Aldrich & Kim (2007), education is used as a moderator on the self-employed parents - female self-employment relationship. Moreover, the level of education is especially important when considering glass ceiling issues, since it is argued that only those females who are relatively higher educated perceive frustrations at hitting glass ceilings (Weyer, 2007).

self-|7

employment (Bullough & Abdelzaher, 2013). Second, this research also wants to commit to management scholars by providing insights into the influence of female representation in management functions on the share of the female self-employed. Third, this paper contributes to theories on role models and family business scholars by linking the importance of parental entrepreneurial role models and managerial role models to female's selection into self-employment. The final contribution will be to human capital theory. It will be examined whether the level of education of females in a country has a moderating role in the self-employed parents - female self-employment relationship and the female managers - female self-employment relationship.

1.3 Research question

In order to be able to achieve the research objective, the following research question is formulated:

What is the impact of the share of self-employed parents and female managers on the share of the female self-employed in a country, and what is the influence of education on these relationships?

1.4 Chaptering

|8

2. Theoretical Framework

The purpose of this chapter is to provide a theoretical discussion from which the variables used within this study will be derived. This will result in an examination of the influence of self-employed parents and female managers, both acting as role models for females, on the share of the female self-employed in a country. The moderating role of education on these relationships will also be discussed in this chapter. Based on existing evidence, hypotheses are formulated and finally these hypotheses are integrated into a conceptual model. But first, the importance of self-employment and the gender gap in self-employment will be discussed before the factors influencing female self-employment shares will be examined.

2.1 Definition and importance of self-employment

What perhaps has been noticed when reading the paper so far, is that the terms self-employment and entrepreneurship are used interchangeably. Although no unanimity exists, the present paper uses the terms self-employment and entrepreneurship as synonyms of each other. Sometimes 'the entrepreneur' refers simply to someone who is self-employed. At other times it implies specific functions such as an innovator or a risk-bearer (Hamilton, 2000). There is no doubt about the fact that entrepreneurs take on several varying roles, blurring the definition. The multifaceted nature of entrepreneurship makes it therefore a difficult to measure variable. Sanandaji (2011) points out, it is inexact to consider self-employment as the same thing as entrepreneurship because not all entrepreneurs are self-employed and vice versa. But, Faggio and Silva (2012) for example showed certain positive bonds in using self-employment as a proxy of entrepreneurship. Although the findings are mixed, in this study the terms entrepreneurship and self-employment are understood as the same occupational form. They are both defined as the generation of income by an individual’s own business or professional activity (Grilo & Thurik, 2008).

|9

2.2 Gender gaps in self-employment

The level of self-employment activities varies from one social group to the other, from one region to the other as well as from one country to the other. Generally, regardless of country, the studies from the past years concerning the level of entrepreneurship activity show that, although the importance of female entrepreneurship shows an upward tendency in any economy, men still own more businesses than women (Singer et al., 2014). In 2014 for example, there were approximately 24 million self-employed men in the EU and 12 million self-employed women (Singer et al., 2014). The reason for this so called 'gender gap' in entrepreneurship is said to be twofold. According to the Global Entrepreneurship Monitor (GEM, 2014) women start up fewer businesses than men. 11 percent of women around the world start new businesses which is 30 percent less than men do. Second, women have a lower success rate than men when it comes to sustaining the business. This results in another 10 percent gender gap in entrepreneurship, in the rates at which women and men sustain their businesses. Over the past couple of years, the gender gap in the start-up rate has decreased but a similar decrease has not been seen in the success rate gender gap.

So, growth in entrepreneurship among women has occurred, since the start-up rates have increased, despite (or perhaps because of) patterns of discrimination in the workplace and sex role stereotyping (Burgess & Borgida, 1999). An underlying premise of most commentaries on female entrepreneurs is therefore that motivations for women to pursue entrepreneurial careers are vastly different from those afforded by men (Starr & Yudkin, 1996). This is also acknowledged by Delmar and Davidsson (2000) and Shane et al. (1991) who argue that those factors explaining the employment rate of men have limited value in explaining the self-employment rate of women.

|10

Especially those factors explaining the share of the female self-employed are of interest (Azmat et al., 2006), since the participation of females in self-employment still lags behind and this is seen as an untapped resource. So, what conditions impact on women entrepreneurship? What determines the share of women entrepreneurs in a country and what explains differences in those shares between countries? Practices and orientations of entrepreneurs are shaped and determined by the macroeconomic and the socioeconomic environment in which they operate (Jamali, 2009). Differences in environmental factors such as technology, culture, and government regulations are all contextual variables that are yet examined to explain rates of female self-employment within and between countries (Den Dulk, 2001). One macro-level explanation that might offer new insights for differences in female self-employment rates at the country-level may be found in the availability of role models for females. To date, role models have not been researched in a cross-country (macro) setting (Laspita et al., 2012). Especially for females role models are particular important, because a lack of role models for females has been identified as a barrier for women who wish to enter nontraditional careers, such as self-employment (Nauta et al., 1998).

2.3 Role models and female self-employment

An individual’s decision to engage in a particular type of behavior is often influenced by the opinions and actions of others, the way in which others demonstrate their identities and the examples provided by others (Bosma et al., 2012). Those 'others' can also be referred to as role models. Gibson (2003, p. 199) defines a role model as "a person an individual perceives to be similar to some extent, and because of that similarity, the individual desires to emulate aspects of that person’s attributes or behaviors." Actually, the lives and activities of role models influence another person in some way (Gibson, 2003).

Role models are particular critical in the career decision-making process. Research has showed a relationship between the influence of role models on a variety of career-related outcomes, such as career aspirations (Nauta et al., 1998), career choice (DeSantis & Quimby, 2004) and attitudes toward nontraditional careers (Nauta & Kokaly, 2001). Especially the influences of role models on attitudes toward nontraditional careers are relevant in this study. For females, nontraditional careers can for example be found in supervisory/management jobs and engagement in self-employment activities (Simpson, 2005). A lack of role models in nontraditional careers has been identified as a barrier for women who choose to enter these professions (Nauta et al., 1998). Indeed, research of Gilbert (1985) has showed that females perceive role models to be especially important when they want to enter nontraditional careers.

|11

role models on female's selection into self-employment will finally result in a push or a pull motivational factor to start a business. This study uses the influence of two different types of role models, in nontraditional careers, on the selection of females into self-employment: both the impact of self-employed parents and female managers are examined. These role models will provide different startup motives regarding the just mentioned push and pull terminology. The next sections will further elaborate on it.

2.4 Self-employed parents as role models

There is still a myth that entrepreneurs are heroic, risk taking, independent, innovative individuals born with these traits (Bird & Brush, 2002). Bird and Brush (2002) conclude that the entrepreneur is consistently described in exactly the same words as those used to describe the manhood. This suggests that constructions of femininity are not compatible with entrepreneurial stereotypes that are founded upon masculinized attributes. Evidence shows that as women grow up in an entrepreneurial family, they are more likely to see beyond the just mentioned myths and stereotypes in entrepreneurship (Kickul et al., 2008). Self-employed parents serve as role models and can give a realistic job preview of self-employment (Kickul et al., 2008). Aldrich et al. (1998) argue that, since children are especially exposed to their parents’ behaviors, entrepreneurial parents can influence their children's career choices to a great extent. The single strongest predicator of entrepreneurship is also found to be parental entrepreneurship. Arum and Müller (2004) showed for example that having an entrepreneur for a parent increases the probability that a child ends up as an entrepreneur by a factor of 1.3 to 3.0. Moreover, empirical evidence from studies of Lindquist et al. (2012) and Scherer et al. (1989) showed that it does not matter if the entrepreneurial parents were successful or not. Entrepreneurial parents who perform better (or worse) do not have higher (or lower) rates of intergenerational transmissions.

The influence of self-employed parents has, however, ambiguous results regarding the so called underlying 'push' and 'pull' motivation to start a business. Both a push and a pull mechanism can be used to explain the influence of self-employed parents on the selection into self-employment. On the one hand, the family influence dynamic could be interpreted in pull terms: the potential entrepreneur could benefit from the advice and examples provided from the entrepreneurial parents (Wagner, 2005). But on the other hand, Wagner (2005) argues that it could also correspond to a push factor: the individual starts an entrepreneurial career because he or she is pushed by the family to perpetuate the entrepreneurial tradition, although this may not correspond with his or her personal desires.

2.4.1 Social learning through self-employed parents

|12

orientation, such as their potential interest to become entrepreneurs themselves (Davidsson & Honig 2003). Wang and Wong (2004) and Schmitt-Rodermund (2004) provide support for parental role models generally encouraging family offspring to become self-employed. Moreover, it has been found that having entrepreneurial parents influences most girls' entrepreneurial intentions but not boys' (Kickul et al., 2008).

However, studies investigating the role of parental entrepreneurial role models pertain to the individual level. Its validity at the country-level is under researched in the entrepreneurship literature. So, are the relationships found at the individual level also valid at the societal level: countries in which a lot of parents are self-employed, do they have a higher share of female employment than other countries? It is argued that the degree to which women enter self-employment is likely to be dependent on the share of entrepreneurial parents in a country. In countries with relatively more self-employed parents, self-employed females have persons in their lives who can serve as a role model, and by observing their parents, they can see beyond the myths and stereotypes in entrepreneurship (Kickul et al., 2008). An increase in choosing self-employment over dependent employment is expected to be more pronounced in countries with relatively more self-employed parents. The following hypothesis is proposed:

Hypothesis 1a: Countries with a high share of self-employed parents exhibit a higher share of self-employed females.

2.4.2 Same-sex transmissions

But, it appears that paternal role models and maternal role models influence their offsprings' decision to enter the world of self-employment to a different extent. Many researchers argue that especially fathers' participation in entrepreneurship is the most important determinant of entrepreneurship among offspring, and in particular among sons (Fairlie & Robb, 2007a). According to Dunn and Holtz-Eakin (2000), fathers' self employment experience has a stronger effect on the probability that sons become employed than mothers' self-employment experience does. This suggest there exists a difference in parent-child transmissions of entrepreneurship among the same sex.

Based on role model identification theory, role models are more often of the same gender (Ruef et al., 2003). More generally, homophily is prevalent in many relationships; that is, individuals have a tendency to bond easier with similar others (McPherson et al., 2001). Therefore it is argued that the transmission of entrepreneurship from mothers to daughters is significantly stronger than that from fathers to daughters. However, empirical evidence on the maternal role in employment is scarce. Recently, Greene et al. (2013) find that self-employed mothers encourage a greater propensity for self-employment in their daughters. Also Lindquist et al. (2012) show that the transmission of becoming self-employed from mothers to daughters is significantly stronger than that from fathers to daughters.

self-|13

employed. However, this effect may be stronger for maternal than for paternal role models in influencing their daughter's self-employment decision, based upon role model identification theory. This leads to the following hypothesis:

Hypothesis 1b: In a country, the share of self-employed mothers will have a larger impact on the share of self-employed females than the share self-employed fathers.

2.4.3 Moderating role of education

While the above arguments and studies support the notion that, through social learning and role model identification, parental role models influence the share of the female self-employed in a country, there appears to be variance in the extent of this influence. Findings in the literature are not uniformly in favor of a strong parental role model effect and insignificant or inconsistent results are reported from some studies (Brenner et al., 1991; Kim et al., 2006). It appears that the effect of parental role models on decisions to enter self-employment is not universal. One factor that may influence the relationship between parental role models and subsequent shares of female self-employment, is the level of education. To date, little empirical evidence exists on the influence of education on this relationship (Aldrich & Kim, 2007), although it is argued that there is a need to investigate the extent to which education moderates the link between parental occupation and children's occupational destinations (Aldrich & Kim, 2007).

Two opposing views can be developed about the moderating effect of education on the parental self-employment - female self-employment relationship. A first view is that the level of education can serve as a substitute for the influence of a role model. Higher levels of education may result in a lower impact of entrepreneurial role models on the share of the female self-employed in a country, because role models are less used as a source of inspiration and learning. Despite the motivational benefits of an entrepreneurial family background for forming career paths, Zellweger et al. (2011) indicate that higher education is less likely to intensify the selection into self-employment. First, individuals from an entrepreneurial family are more likely than those without a similar background to access the critical resources and social networks (Zellweger et al. 2011). Because this group is able to gain access to these valuable inputs, it reduces their necessities of additional inputs from higher formal education. It is also expected that as the level of education increases, individuals acquire increasing amounts of exposure to decision making models (Heckler, 1989). Heckler (1989) argues that as a result, it is less prone that parental choices will be copied, and instead an independent style of decision making is developed. An independent decision making style can be expressed in the career path that is chosen. Fritsch and Rusakova (2012) indeed find a significant and positive relationship between the presence of a parental role model and the decision to become self-employed for less-educated people.

|14

knowledge (role model influences) requires a certain level of knowledge and skills. In addition, the more highly educated can therefore better be able to recognize the value of their entrepreneurial parents and assimilate it (Cohen & Levinthal, 1990). They can subsequently use this knowledge to start up their own business. However, empirical evidence for this point of view is lacking.

In sum, it can be said that shares of female self-employment depend also upon characteristics other than parent's self-employment status only: education can be an important moderating factor in this relationship. Because the 'absorptive capacity view' lacks empirical evidence, it is hypothesized that in countries where females have higher levels of education, women are less influenced by role models within an entrepreneurial family than in countries in which females have lower levels of education.

Hypothesis 2: The impact of employed parents to push or pull females into self-employment is more pronounced in countries in which females have lower levels of education.

2.5 Female managers as role models

While a relationship between employed parents and the decision to become self-employed is important for females, another link- one between the presence of female managers in a country, acting as role models for other females, and the decision to become self-employed- can also be very crucial. During the last century, the size of the female labor force has rapidly grown in most developed countries (Bennett & Dann, 2000). However, the employment conditions of females have not always matched those of males. Often, female workers have difficulties in accessing higher level employment positions and it is harder for them to get promoted. Researchers have pointed at discrimination against women on the labor market (Burgess & Borgida, 1999). Women at the top of today’s largest companies are for example very scarce. Even after 30 years of equality legislation, only 5.3% of executive directors in the UK’s top 100 companies in 2013 were women (FTSE, 2013).

The organizational reality in many countries is that women struggle with rigid and male-dominated hierarchies, lack of credibility in a masculine culture and promotion based on uninterrupted linear career paths with little flexibility (Rosener, 1990). Just as Due Billing and Alvesson (2000) argue, organizations continue to 'think manager: think male'. One reason cited by researchers for a lack of female managers is the absence of female managerial role models within organizations (Schwartz, 1989). The more female role models there are available in management positions to inspire and motivate women, the more likely it is to see women making decisions to go for the top jobs in organizations (Schwartz, 1989). A theory that can be used to explain a lack of female managerial role models in a country is glass ceiling theory (Terjesen, 2005), which will be discussed next.

2.5.1 Glass ceiling push effects

|15

2005). Jackson (2008) state that the result is that talented productive women are finding that they cannot gain access to the highest levels and highest pay scales of the economy, especially within male dominated organizations. It is therefore not surprising that women are leaving corporate employment in increasing numbers as their need for challenge, achievement and status is left unfilled. Patterson and Mavin (2009) argue that entrepreneurship is a viable and often attractive alternative for women who wish to gain greater control of their working lives and escape career glass ceiling constraints. Increasingly, professional women are therefore leaving waged employment in favor of self-employment (Patterson & Mavin, 2009).

Weiler and Bernasek (2001) state that women's frustration about working in male dominated environments is without question a major driver in women starting their own businesses. One third of the women-initiated enterprises are in fact motivated either by barriers to advancement or lack of challenges in current careers (Patterson & Mavin, 2009). They argue that female's decision to enter self-employment, and leave their former corporations, is dependent on the career progression in their former employment position. Women who reach the management ranks in any organization are unlikely to strike out on their own as they are already running their own businesses inside a larger corporation (Patterson & Mavin, 2009). Devine (1994) also showed that a much larger percentage of female entrepreneurs come from white collar (nonsupervisory) jobs compared with men, which supports the noted frustration with the glass ceiling for female entrepreneurs. This entrepreneurial motivation can be classified as a 'push factor'. Push factors are negative motivations which move individuals towards forms of entrepreneurship (Amit & Muller, 1995). The presence of a glass ceiling within organizations in a country can be argued to result in a 'glass-ceiling push effect' as a motivation for females to become self-employed.

It can be said, based on the just mentioned effect of the 'glass-ceiling push motivation', that there exists a relationship between female managers and female self-employment. In countries with relatively few females in management positions, self-employed females have different start-up motives and aspirations because of the possible underlying motivation to escape from the glass ceiling. In contrast, when there are relatively many females in management positions in a country, and therefore little glass ceiling issues, females can enter management positions more easily and the option to start a business themselves becomes less attractive. They are exposed to female managerial role models and since role models are especially important to women in nontraditional careers, such as in management (Simpson, 2005), the barriers to choose for a management position will be lower. This will reduce the amount of females that opt for self-employment. What is clear from both the theoretical and empirical literature is that women already in management positions are much more likely to prefer waged employment and to stay in waged employment (Hewlett & Luce, 2005). An increase in choosing self-employment over dependent employment is therefore expected to be more pronounced in countries with relatively few females in management positions. The following hypothesis is proposed:

|16

2.5.2 Moderating role of education

The arguments in favor of the 'glass-ceiling push effect' support the notion that through a low share of female managers in a country, females are to a certain level 'pushed' into self-employment. But there appears to be variance in the extent of this influence. The effect may be more complex than assumed, and different factors may influence the relationship between the share of female managers and the share of the female self-employed in a country. One important factor that has the potential to influence this relationship is the level of education of the females in a country (De la Rica et al., 2008). It appears that the effect of the glass ceiling on the self-employment decision varies between high- and low-educated females.

Surprisingly, in many countries women are now outnumbering men in college level education and women are also graduating more successfully (Becker et al., 2010). The reversal of the gender gap in education implies that in many countries, for the first time in history, there are more highly educated women than highly educated men (Esteve et al., 2012). But, quite remarkable: the number of females in higher positions in the labor market show actually no improvement in recent years, despite the increase in education levels. The proportion of women in (senior) management roles in 2014 is stuck at 24%, the same as 2013, 2009 and 2007. The numbers fell in 2011 and 2012, when women were hard hit by the global economic crisis (Grant Thornton, 2014). So, still, the number of females in (senior) management positions have not matched the higher education levels of females.

|17

In contrast, Weyer (2007) argues that lower educated women tend to remain in manual labor positions. With lower expectations in the first place, they do not feel the same frustrations against the glass ceiling (Weyer, 2007). Women with more education are consequently more likely than lower-educated women to feel the effects of a glass ceiling in the workplace. These higher educated females see the glass ceiling as an incentive to enter into self-employment, resulting in the following hypothesis:

Hypothesis 4: The impact of the glass-ceiling to push females into self-employment is more pronounced in countries in which females have higher levels of education.

2.6 Conceptual model

To present the hypotheses proposed in this chapter in a visual way, a conceptual model is developed. Figure 1 shows the conceptual model for hypothesis 1-4.

|18

3. Research Design

In this chapter the methodology of the study will be explained. First an elaboration on the research method is included, followed by the way in which the data is obtained. Also attention is paid to the measures of the variables used in this study, and finally this chapter ends with a discussion of the data analysis.

3.1 Research method

In this study a theory testing approach will be used. According to Van Aken et al. (2012), a theory testing method is the right approach when literature streams are elaborated, but evidence on the theoretical explanations of the phenomenon is still inconclusive. So there is a gap in the literature field that must be closed. Since this is the case for the influence of self-employed parents and female managers on the share of the female self-self-employed, and the role of education in these relationships, a theory testing approach will be conducted. Van Aken et al. (2012) propose a 4-step plan when performing a theory testing approach. These steps will be followed in the current study: 1) definition of the business phenomenon and identification of the literature gap; (2) development of hypotheses and a conceptual model; (3) large scale data collection and statistical data analysis and (4) interpretation of results, comparison with the hypotheses, drawing conclusions, providing implications and setting limitations (van Aken et al., 2012).

Previous studies that examined the determinants of national female self-employment rates have been confined to only a couple of countries (Parker & Robson, 2004). Although these studies have been able to draw conclusions about causes of female self-employment rates within particular countries, they have two major drawbacks. First, they cannot explain differences in self-employment rates between countries. Second, national time-series studies often work with only short spans of data, which results in tests of statistical significance with a lack of power and reliability (Parker & Robson, 2004). These problems can be avoided by exploiting panel data. As Hsiao (2003) argues, panel data contains observations of multiple occurrences obtained over multiple time periods for the same countries. With panel data one is able to examine the data across and within countries over time. Panel data enjoys the advantage over static cross-sections or single-country time-series data of looking at more than just one time period and country (Hsiao, 2003). For these reasons, this study uses panel data with the major advantage that tests on panel data are much more powerful than for the conventional single-country case, making inferences more reliable (Banerjee et al., 1993).

3.2 Data collection and sample

|19

data source. The ESS is a cross-national survey that has been conducted every two years across Europe since 2001. The survey measures the attitudes, beliefs and behavior patterns of diverse populations in more than thirty nations. Rounds 1 to 6 (2002-2012) are used in this study to have a large enough sample for deriving characteristics of the female self-employed at the country-level. Only those countries are used which had a total of at least 100 female employed, 100 female managers and 100 females from which the parents are self-employed, because non-significant results can be found with a sample size below 100 respondents (Pallant, 2005). The final sample represents 24 countries in Europe. In the appendix, table A1 is included to give an overview of the observations across countries and the associated ESS rounds. Because not all countries are included in each wave, the final dataset takes the form of an unbalanced panel dataset.

With large numbers of individuals in a sample within each country but a small number of countries, reliable individual-level effects can be estimated for each country but country effects are likely to be unreliable (Bryan & Jenkins, 2013). The number of countries used in the current study is relatively small for quantitative research (Keller, 2009), so there is a possibility that the results of this study may lead to incorrect conclusions. Reliable results of regression analyses are contingent on sample sizes being large. In particular, a large number of groups (countries) is needed in order to estimate reliable country effects. According to Bryan and Jenkins (2013) at least 25 countries for linear regression models are required. With fewer countries, estimates of country-level parameters are likely to be estimated imprecisely and this will not be adequately reflected in test statistics (Bryan & Jenkins, 2013). This will consequently lead to faulty conclusions about country effects. Although the number of countries in the current study is reasonably well to the lower limit, no true problems of reliability with small sample sizes will be expected since the minimum requirement of the amount of countries is (almost) met. Accordingly, Selig et al. (2008) also mention a sample size of at least 20 countries as the minimum number.

3.3 Measurements

3.3.1 Dependent variable

Share of self-employed females. Self-employment is defined as the generation of income by

an individual’s own business or professional activity (Grilo & Thurik, 2008). This variable is operationalized as the number of self-employed females divided by the total female workforce in a country. To make clear whether a person is self-employed, the following question is included in the ESS: "In your main job are you employee or self-employed?".

3.3.2 Independent variables

Share of self-employed parents. In this study variables are used to indicate whether the

|20

questions are included reflecting the occupation of a person's parents: "When you were 14, did your father work as an employee, was he self-employed, or was he not working then?" and "When you were 14, did your mother work as an employee, was she self-employed, or was she not working then?". The age of 14 is probably chosen in the survey since this represents the age when individuals begin to develop visions of themselves which in turn influence occupational planning (Super, 1990). Although Jodl et al. (2001) argue that identity consolidation is not achieved until late adolescence (age of 17-22), the early adolescence (age of 11-14) is an important period too. There is evidence that points at the critical period of the early adolescence in which parents are particularly salient, as youth start to explore different conceptions of who they are and who they want to be (Eccles et al., 1998). Creamer and Laughlin (2005) found that the younger the daughter, the stronger the influence of significant others (and parents in particular) and Goldstein and Oldham (1979) also argue that especially during the early adoloscence, parents play an important role als role models.

Share of female managers. A manager is defined as a person who is responsible for planning

and directing the work of a group of individuals, monitoring their work, and taking corrective action when necessary (Fayol, 1949). The ESS included a question to determine if someone has a management position within an organization. The following question is considered: "In your main job, do you have any responsibility for supervising the work of other employees?". Because it is important to distinguish between managers and self-employed for the purpose of this study, also a question regarding the employment relationship is considered: "In your main job, are you employee or self-employed ?". Only those who are considered as an employee are included in the share of female managers. To measure the share of female managers in a country, the number of female managers is divided by the total female workforce.

3.3.3 Moderator variable

Level of education. The level is education is operationalized as the highest level of formal

education in terms of how much and what kind of education an individual has completed, as manifested by the educational certificates awarded (Braun & Müller, 1997). It can be measured by the highest educational qualification achieved or by the number of years of education completed. This latter one assumes that the longer the individual stayed in education, the higher the level. Especially in cross-national comparative studies it is recommended to use 'years of schooling' as a measurement of educational level. Average years of schooling turns out to be a statistically significant measure to provide useful information about a country's educational capital stock (Barro & Lee, 2015). Therefore the average years of schooling of the females in a country are used as a measure of the level of education.

3.3.4 Control variables

|21

per capita. There is a strong negative correlation between the level of GDP per capita and the

share of self-employment (Acs et al., 1994). Besides GDP per capita, also female

unemployment rates basically can act as a factor that influences female self-employment rates

(Audretsch & Thurik, 2000). Results are however mixed. On the one hand the opportunity costs for unemployed individuals to become self-employed are relatively low, and this will favor their choice for self-employment (Staber & Bogenhold, 1993). On the other hand, high unemployment may be related to an economic depression, which makes prospects for starting a business bleak (Storey, 1991). There will also be controlled for female labor force

participation rates. A growing participation of women in the labor market implies a

decreasing share of self-employment in the labor force (Acs et al., 1994). Finally, population

density may also have an influence on the share of female self-employment. Urban areas often

result in economies of scale, through which small-sized entrepreneurship comes under pressure (Bais et al., 1995). On the contrary, thinly populated areas will often have many small businesses (Bais et al., 1995). Consequently, population density may be expected to have a negative effect on the rate of female self-employment in a country.

The ESS data is merged with country-level data on GDP per capita and population density for the years 2002 till 2012 provided by Eurostat and the Worldbank.

3.4 Data analysis

Following the literature (for example Acs et al. 1994; Staber & Bogenhold, 1993) the ordinary least-squares regression method (OLS) will be used to test the hypotheses developed in chapter 2. OLS regression examines the independent effects of each variable on the share of female self-employment while controlling for all of the other variables in the equation. The share of the self-employed females is regressed on the share of self-employed parents and female managers in a country. The analysis will be performed by first having control variables, then adding the main effect variables and finally testing the moderating variable.

3.4.1 Assumptions Ordinary Least Square regression

|22

calculating an alternative (robust) standard error that allows for the possibility of heteroskedasticity.

3.4.2 Regression model

To test the hypothesis whether countries with a high share of self-employed parents exhibit a higher share of female self-employed, and whether the effect varies by educational level, the equation below will be estimated. To examine whether the share of female managers is negatively related to the share of the female self-employed in a country, and whether the effect varies by educational level, these factors are also added to the equation:

SEFct = αc + β1SEPct + β2SEPct * EDU + β3FMct + β4FMct * EDU + β5GDPct + β6UNRct +

β7LFPRct + β8PDct + t (1)

The dependent variable SEFct represents the total share of the female self-employed in a

country c in period t. The explanatory variables are the total share of self-employed parents (SEP) and female managers (FM) in a country. The interaction term between the share of self-employed parents and the level of education (EDU) is also included in the model, as well as the interaction term between the share of female mangers and the level of education. In addition to the explanatory variables, the following control variables are added: GDP per capita (GDP), female unemployment rates (UNR), female labor force participation rates (LFPR) and population density (PD). Finally, ε represents the error term.

To test whether the share of self-employed mothers will have a larger impact on the share of self-employed females than the share self-employed fathers, the following regression will be estimated:

SEFct = αc + β1SEMct + β2SEFct + β3GDPct + β4UNRct + β5LFPRct + β6PDct + t (2)

|23

4. Results

This chapter presents the results of the statistical analysis performed in order to draw conclusions about the hypotheses. First some cross-country differences among the self-employed are displayed followed by the descriptive findings. The outcomes of the correlation analysis and finally the ordinary least squares regression analysis are discussed in this chapter.

4.1 Cross-country differences among self-employed

Although the participation of females in self-employment increased considerably during the last decades, self-employment is less common among women than among men. Figure 2 displays female self-employment as a share of all females in the workforce and the share of male self-employed over all males in the workforce. The 45°-line shows the deviation from gender-parity. As the graph in figure 2 shows, the share of self-employed among men is higher than for women. Especially in Ireland, there is a relatively large gender gap, compared to other European countries. The share of self-employed males exceeds .22, while the share of employed females is barely .06. This also applies to Italy, where the share of self-employed males is around .3 and the share of self-self-employed females is far below .2. The gender gap is the smallest in Luxembourg, Poland and Bulgaria. The differences between male and female shares of self-employment are close to .04 for these countries. The total average proportion of self-employed females is .08 and .15 for males.

|24

4.2 Descriptive statistics

Table 1, on the next page, shows (a) the means and standard deviations and (b) the Pearson correlations between the variables used in this study. The correlation matrix reveals that three variables are highly inter-correlated (above .8). A high correlation is detected between the share of employed parents, the share of employed mothers and the share of self-employed fathers. These correlations range between .739 and .992, because the paternal and maternal self-employed are included in the parental one. Hence, separate analyses are used for the just mentioned three independent variables because highly inter-correlated variables inflate the size of the error terms and weaken the analysis (Tabachnick & Fidell, 1996). Moreover, almost all the variables of interest in this study, the share of female self-employed, the share of self-employed parents, the share of female managers and the educational level, correlate significantly to one another as shown in table 1. Since the variance inflation factor (VIF) of the different variables are well below 10, there are however no multicollinearity problems

|25

|26

4.3 Regression analysis

|27

After controlling for the female labor force participation rate, female unemployment rate, GDP per capita and population density, the results in table 2 show that the share of employed parents in a country has a significant positive effect on the share of the female self-employed in a country (model II, β = .671, p < .01). This provides support for hypothesis 1: Countries with a high share of self-employed parents exhibit a higher share of self-employed females. Moreover, a significant positive relationship is also found between the share of self-employed mothers (model III, β =.655, p < .01), the share of self-self-employed fathers (model IV, β = .787, p < .01) and the share of the female self-employed in a country. Both parents have a significant positive effect on the share of the female self-employed in a country, but the share of self-employed fathers has a stronger effect (model V, β = .509, p < .01) on the share of self-employed females than the share of self employed mothers (model V, β = .305, p < .01). Hence, hypothesis 1a: The share of self-employed mothers will have a larger impact on the share of self-employed females than the share self-employed fathers, is not supported.

|28

Turning back to model II from table 2, the results show that the share of female managers in a country has a significant negative influence on the share of the female self-employed in a country (model II, β = -.324, p < .01). These results support the negative relationship expected in hypothesis 3: Countries with a low share of female managers exhibit a higher share of female self-employed. To explore whether the effect of self-employed parents and female managers on female self-employment shares differs by educational level, years of education is added as a moderator variable in model VII. The moderator (years of education) has a significant negative effect on the share of the female self-employed in a country (Model VI, β = -.143; p < .05). Model VII includes the interaction variables. Education is found to be a significant negative moderator of the positive relationship between the share of self-employed parents and the share of female self-employment in a country (model VII, β = -.331, p < .01). This provides support for hypothesis 2: The impact of self-employed parents to push or pull females into self-employment is more pronounced in countries in which females have lower levels of education. This implies that when the years of education for females increases in a country, the positive relation between the share of employed parents and the share of self-employed females, will become weaker. With regard to the moderating effect of education on the female managers - female self-employment relationship, it seems to be that the negative relationship becomes weaker when the level of education of females is higher in a country. However, education does not significantly moderate the relationship between the share of female managers and the share of the female self-employed in a country. Therefore the results do not provide support for hypothesis 4. The impact of the glass-ceiling to push females into self-employment is not more pronounced in countries in which females have higher levels education.

To facilitate interpretation, the significant moderating effect is plotted in figure 3. This figure shows the moderating effect of educational level on the self-employed parents - female self-employed relationship. When the level of education of females is high in a country, the relationship between the share of self-employed parents and the share of self-employed females becomes weaker; the share of self-employed females stays almost steady when the share of self-employed parents increases. To contrast this, when the level of education is lower, the share of self-employed females becomes higher, and the positive relationship between the share of self-employed parents and the share of self-employed females becomes stronger. The share of self-employed females increases when the share of self-employed parents increases at lower levels of education.

|29

5. Discussion and Implications

The main purpose of this study is to gain insight into the relationship between role models (self-employed parents and female managers) and shares of female self-employment. Additionally, the effect of education on these relationships is also examined. This chapter starts with the most important findings of the study and these findings will be related to existing literature. The implications of the findings for theory, practice and policy makers will also be discussed. Subsequently, the limitations of this study and suggestions for future research are presented and this chapter will end with the main conclusions.

5.1 Findings

Recently, female self-employment has become an important topic within the entrepreneurship literature, but it is still a very much under researched literature field (Bullough & Abdelzaher, 2013). Female self-employment is not only relevant because it results in the creation of employment and economic growth, it also contributes to the diversity in self-employment (Verheul & Thurik, 2001) and it creates a work environment in which job and family could be combined (Achtenhagen & Welter, 2003). However, rates of female self-employment still lag behind that of men and there are substantial differences in the rates of female employment across countries. Micro-level explanations for varying rates in female self-employment are best documented in the literature, but they do not account for the variety across and within countries, and they also do not demonstrate how much variation is explained at the level of countries. Therefore this study uses a macro-level approach to account for differences in female self-employment rates across countries. One macro-level explanation that might offer new insights into diverging female self-employment rates, is the availability of role models within countries for females. Based on a sample of 24 European countries, this study investigated how country differences in the presence of role models are related to the selection of females into self-employment. The influence of self-employed parents and female managers, both acting as role models for the females in a country, on the share of the female self-employed is examined. Because education can be especially important in the just mentioned relationships, the level of education is considered as a moderating variable influencing the self-employed parents - female self-employment relationship and female managers - female self-employment relationship

|30

can give a realistic job preview of self-employment and females can see beyond the myths and stereotypes that surrounds entrepreneurship to date (Kickul et al., 2008).

However, self-employed fathers and self-employed mothers influence their offspring decision to enter the world of self-employment to a different extent. Based on role model identification theory, it is expected that role models are more often of the same gender, and therefore the transmission of entrepreneurship from mothers to daughters should be significantly stronger than that from fathers to daughters (Greene et al., 2013; Lindquist et al., 2012). Findings of this study do, however, not provide support for these 'same sex transmissions'. Although the results show that both self-employed mothers and self-employed fathers have a significant positive effect on the share of the employed females in a country, the influence of self-employed fathers is stronger than the influence of self-self-employed mothers. Prior research also shows mixed results regarding the influence of the self-employed father and mother. Russel et al. (2003) for example found that paternal role models play a more significant role in their children’s occupational choice than do maternal role models. This argument is supported by Dunn and Holtz-Eakin (2000), who also found that children of self-employed fathers enter self-employment to a higher extent than those of self-employed mothers. This suggests that females might experience their father’s self-employment as more intense than their mother’s and therefore rates of female self-employment are more influenced by self-employed fathers.

With regard to the other type of role models used in this study, the female managers, the results show that there is a negative relationship between the share of female managers and the share of self-employed females in a country. This is in line with the arguments concerning the 'glass ceiling push effect'. In countries with few females in management positions, self-employed females have different start-up motives and aspirations because of the possible underlying motivation to escape from the glass ceiling. In these countries the share of employed females is relatively higher as compared to countries in which the share of self-employed managers is high. This is in line with the arguments of Weiler and Bernasek (2001) and Patterson and Mavin (2009), who state that women's frustration about working in male dominated environments is without question a major driver in women starting their own businesses. Females will leave their corporations to 'go it alone' because of frustration at hitting the glass ceiling (Familoni, 2007). The current research has indeed found that the possible presence of a glass ceiling, so relatively few females in management positions, motivates women to start up a business themselves. The findings are also in line with Simpson (2005), who argues that when there are already a lot of females in management positions, other females are exposed to these managerial role models and this will reduce the amount of females that opt for self-employment.

|31

parents, but the average educational level of the females is also high, those countries have a lower share of self-employed females. Valuable inputs from an entrepreneurial family reduces the necessities of additional inputs from higher formal education (Zellweger et al., 2011) and lower levels of education are associated with less decision making models whereby it is more likely that parental choices will be copied (Heckler, 1989). This is in line with findings from Fritsch and Rusakova (2012), who found a more positive relationship between the presence of a parental role model and the decision to become self-employed for less-educated people. However, education does not moderate the relationship between female managers and the share of the female self-employed in a country. This finding is surprising, since it was hypothesized that especially those higher educated females experience glass ceiling issues in their environments and be therefore more pushed into self-employment. But it appears that the effect does not vary significantly by educational level. Prior research has, however, concluded that education plays an important role in the glass ceiling - self-employment transition for females (De la Rica et al. 2008; Baraka, 1999). The missing significant evidence in this study may be due to the way in which the glass ceiling is measured. There is to date a scarcity of measures of the glass ceiling (Smith et al., 2012). Besides using the share of female managers as a proxy for the glass ceiling, as is done in the current study, a potential more accurate way to measure glass ceiling issues in a country is by looking at women's beliefs about the glass ceiling. For example the Career Pathways Survey (CPS) measures beliefs about glass ceilings which are congruent with the latest theoretical understandings of the glass ceiling (Smith et al., 2012). This instrument measures different aspects of the glass ceiling such as acceptance, resignation, and promotion scores. There is a possibility that by using a more subjective measure of the glass ceiling, education has a significant impact on the glass ceiling - female self-employment relationship.

5.2 Theoretical implications

The findings of this study will be of value for different literature streams, because it increases knowledge about the drivers of female self-employment, the importance of role models, and the influencing role of human capital. To date, little research exists on the factors explaining the self-employment rate of females from a macro-perspective (Bullough an Abdelzaher, 2013). The findings of this study contribute to macro-level studies on female entrepreneurship by exploring the relationship between the availability of role models for females in a country in nontraditional careers and consequently female's selection into self-employment.

|32

females play an important role in explaining country differences in female self-employment rates and the results show that variation in female self-employment rates can be explained from a role model perspective.

On the one hand, the findings show that in countries with relatively a lot of self-employed parents, more females are exposed to entrepreneurial role models and consequently those countries exhibit higher female self-employment shares. This confirms research from Kickul et al. (2008) who argue that especially for females, parental entrepreneurial role models are important because it allows them to look beyond the stereotypes in entrepreneurship, and start up a business themselves. Research on the influence of parental entrepreneurial role models on females is, however, scarce. The focus is most often on the influence of parents on both children, without distinguishing in gender, such as research from Arum & Müller, (2004), Wang and Wong (2004) and Schmitt-Rodermund (2004). The current research expands our knowledge about the positive relationship between employed parents and female self-employment rates at the macro-level. Interesting to theories on 'same-sex transmissions' is also the surprising result of this study that self-employed fathers have a larger influence on female's selection into self-employment than self-employed mothers.

On the other hand, a lack of female managerial role models in the corporate world, a sign of potential glass ceiling issues in a country, serves as a motivation for females to start up a business themselves. Empirical evidence of the push effect of the glass ceiling on female self-employment rates is very scarce, and evidence in a cross-country setting is to date nonexistent. Terjesen (2005), Familoni (2007) and Patterson & Mavin (2009), who all performed micro-level research, argue that professional women are increasingly leaving waged employment in favor of self-employment because of glass ceiling issues. Findings of the current study show that in countries with relatively few female managers, female self-employment rates are higher. Few female managers in a country can be a sign of glass ceiling issues, and by starting up a business, females can escape from this glass ceiling. Country differences in female self-employment rates can therefore be explained from a glass ceiling perspective.