An institutional perspective on the self-employment gender gap: What factors should policy focus on?

(1)

An institutional perspective on the self-employment gender gap:

What factors should policy focus on?

21 October 2015

A master thesis written by Lisanne J.A.A. Huijs

Supervisor: F. Noseleit

Second assessor: D.J. Langley

Wordcount: 12.500

(2)

[1]

Preface

With this thesis is will finalize my studies at the University of Groningen. During this time I was able to develop and grow into the person I am today. Whilst studying at the faculty of business and economics, I soon came to the understanding that my interests are so much broader than business. This resulted, amongst other things, in an exchange to Australia. Here I was able to take some sociology courses. These subjects gave me a broader view, and were exactly the for me the enrichment I was looking for in my studies.

In this thesis I am able to combine these two fields. The literature really got me interested for the theme, and I hope, being it only a tiny bit, can add some new insights to this stream.

I would like to thank Florian Noseleit for all the good advice and effort he made for me. Always being just an e-mail away. Thank you!

(3)

[2]

Abstract

Whilst gender inequality is still existing in all countries in the world, the perception of men being superior to women no longer is. In developed countries it is believed that women should have the same opportunities as men. Despite this belief, women are still underrepresented in the labour market and especially in self-employment. Governments try to decrease this gender gap, since women hold great potential for economic development, but their policies mostly have a counterproductive effect.

With this research it is tried to make a first attempt with filling the a gap in academic literature. Till now there is no extensive work on the relation between formal and informal institutions in country level and the gender gap in self-employment available.

The results indicate that in fact the institutions gaining most attention by academics, namely institutions related to family are indeed playing in large role in explaining the gender gap in self-employment. Furthermore, also GDP per capita and the extent to which industries are gendered are important.

This thesis concludes that especially women who already should be able to become self-employed, do nog perceive this as an attractive option. Therefore, practical implications are made, focussing on making self-employment a more desirable employment option for women.

(4)

[3]

1. Introduction

Aletta Jacobs was the first female student in the Netherlands. She started her studies in medicine at the University of Groningen in 1878, whilst this university was already founded in 1614. So it took 264 years for women to be allowed at this university in the Netherlands. In 1878 she graduated and became the first Dutch female doctor. She introduced birth control and she was an advocate of the female suffrage. In 1922 it was the first time that women were allowed to vote (Feinberg, 1996). Now, 137 years later I am going to graduate from the same university as Aletta Jacobs. Still an achievement, but no longer something extraordinary. Women attending university is no exception anymore in most western countries, it has become the norm. Now, in the Netherlands even more women are enrolled at universities than men (Centraal Bureau voor de Statistiek, 2014).

Emancipation has come a long way. Especially in the so called global North, women and men are becoming more equal. Kan, Sullivan and Gershuny (2011) explain it as gender convergence. The shares of men and women are approximating each other in amongst other things education, informal labour and formal labour (p. 234). The education gap is almost closed in most countries and even reversed in some. This is to a large extend due to the commitment of 189 countries to the Millennium Development Goals. Off which a primary goal was to increase the number of girl with a proper education (Ganguli, Hausmann, & Viarengo, 2014, p. 173). The closing gender gap in education is a major step in making girls and women more independent and letting them contribute to economic growth. However, unfortunately, educating more girls does not automatically result in a more equal labour market (OECD, 2012, p.13).

(7)

[6]

The share of business owned by women is significantly smaller than the share owned by men. This gap increases if we only take into account businesses which generate employment to others (OECD, 2012, p. 274-276). Overall, men are starting and exploiting businesses which generate more employment, are more profitable and are more innovative (Thébaud, 2015, p. 2). Whilst the gender gap in employment is there in all countries, the size of the gap and also the total number of self-employed shows great cross-country differences (Pathak, Goltz, & Buche, 2013, p. 478; Thébaud, 2015, p. 4; Torrini, 2005, p. 661). The average share of women being self-employed is 13,0% amongst OECD countries, varying between 4,35% in Norway and 43,44% in Turkey (Kodama, & Odaki, 2011, p. 1237; OECD, 2015a).

This issue, the gender gap in self-employment, is on the political agenda of most western countries. These countries have introduced all kinds of policy initiatives, to close the overall gender gap in the labour market, and sometimes the gender gap in self-employment specifically. These governments do not only want to enhance the position of women in their country, but also see that these women hold great economic potential. Not only do these governments have this ambition, it also turns out that they have a significant influence on the employment status of their citizens (Torrini, 2005, p. 666). As Baumol (1990) explained in his highly cited work, there are always people with entrepreneurial talent and ambition present in society. It is the society in which they live, what the rules of the games in that particular environment are, which makes them to decide to exploit these gifts in a productive or unproductive manner. Some societies even make it most attractive to apply these characteristics in a destructive way, which has a negative impact on society (p. 898-899)(Estrin & Mickiewicz, 2011, p. 399; Kasper & Streit, 1998, p. 242).

What are these rules of the game? Which elements constituting a society influence the employment choice of an individual, and to what extent? It’s still an unanswered question in academics. A relevant questions since great expenditures are made by countries to close the gender gap in self-employment, since they believe there is great economic potential in this. Though, it is argued that these efforts do not always have the desired outcomes.

(8)

[7]

p. 3). This neglects the fact that the birth rate of female owned businesses exceeds the birth rate of male owned businesses, but the self-employment gender gap is still significant and is hardly closing (OECD, 2012, p. 276)

Previous research by Pathak et al. (2013) made an attempt on including a more comprehensive set of institutional variable. However, they solely focused on the relationship with female self-employment and did not look at the gender differences(p. 480). Besides, also this research only takes into account nascent entrepreneurs (p. 486).

There is a need for more extensive work on the relation between formal and informal institutions in country level and the gender gap in self-employment. Research, not only taking into account one type of country level factors like the economic conditions, legislation, or gendered institutions, demographics etc., but in the direction of getting a more comprehensive view (Ahl, 2006, p. 611-613; Ganguli et al., 2014, p. 201; Klyver, Nielsen, & Evald, 2013, p. 486; Pathak et al., 2013, p. 498; Thébaud, 2015, p. 31; Verheul et al., 2006, p. 176-177). Which factors then, explain the greatest part of the self-employment gender gap?

Answer to this question could give a better direction for policy development (Ahl, 2006, p. 613). What should be focused on? What kind of policies indeed foster women’s self-employment in a way it strengthens the position of females and provides economic progression? Now policies, developed to increase gender equality, sometimes actually have the opposite effect (Thébaud, 2015, p. 3).

1.1 Gender inequality

It is a fact that in general men always had and still have a superior position in relation to women. Because of the natural difference in physical strength and the fact that women by nature are the ones who give birth and feed their children. The easy to recognize human characteristic, gender, became a distinction mechanism. In biological terms, men and women do have different characterises. However, the weight and the paradigms we relate to this difference is socially formed (Smith, Noll, & Bryant, 1999, p. 499). We can thus ask ourselves the question, since we are genetically different, why gender inequality is an issue?

(9)

neo-[8]

institutionalist theory (Gerhards, Schäfer, & Kämpfer, 2009, p. 516; Sikora & Saha, 2009, p. 387). The perception of the man being superior to the woman was no longer self-evident. Now, it is perceived that women should have the same opportunities as men. In modern society governments and overarching bodies like the United Nations believe, document and propagate that all people should have equal rights and opportunities. A perception which now, to a large extend, is shared by the citizens of these countries (Gerhards, Schäfer, & Kämpfer, 2009, p. 517).

The United Nations defined equality in the Millenium Declaration as follows:

“Equality: No individual and no nation must be denied the opportunity to benefit from development. The equal rights and opportunities of women and men must be assured.” (United Nations General Assembly, 2000, p. 2).

Gender equality, despite of being goal number three on the United Nations millennium Development Goals, committed to by 189 member states, is still a goal which has not been achieved yet (OECD, 2012, p. 13). Previous research shows that there are barriers prohibiting total equalization. One of the most pronounced inequalities in western countries is the division of domestic work. Women still do the lion share of all the work at home. Especially caring for children and elderly and all the routine chores like cooking, cleaning and laundry are still the responsibility of women. Men are doing more non-routine work now, than four decades ago, which results in a higher share of domestic work performed by men, which implies convergence of the genders. Striking though, is that the inflexible routine jobs are still performed by women (Kan et al., 2011, p. 238; OECD, 2012, p. 13).

Thus, women are the ones who have to manage the work-family conflict (Thébaud, 2011). Research tells us that women found a way to combine their responsibilities at home with formal employment. They become self-employed, most of the times working home-based, giving them more flexibility. Despite the benefit of flexibility, a gender gap in self-employment is still present. Women do not only face gender disadvantages in waged employment, but also in self-employment (Marlow, 2006, p. 593).

(10)

[9]

measurements on human development, namely: reproductive health, empowerment and economic status.1_{Table 1 gives an overview of the GII and its measures.}

Table 1 – The Gender Inequality Index (GII)

GII MATERNAL MORTALITY (DEATHS PER 100.000 LIVE BIRTHS) ADOLESCENT BIRTH RATE (BIRTHS PER 1,000 WOMEN AGED 15-19) SHARE OF SEATS IN PARLIAM ENT (% HELD BY WOMEN) GENDER GAP OF POPULATION WITH AT LEAST SOME SECONDARY EDUCATION (PEOPLE ABOVE 25) GENDER GAP IN LABOUR FORCE PARTICIPATION RATE (PEOPLE ABOVE 15) TOP 3 (SLOVENIA, SWITZERLAND AND GERMANY) 0,033 9 2 28,07 1,52 12,73 BOTTOM 3 OF VERY HIGH DEVELOPED COUNTRIES (CHILE, ARGENTINA AND QATAR) 0,420 36 40 17,24 -2,21 32,70 BOTTOM 3 (AFGHANISTAN, CHAD AND NIGER) 0,707 717 148 18,60 13,82 43,03

Source: United Nations Development Programme (UNDP, 2015)

It shows that women still have a weaker position that men in all countries. For example, with exception of Rwanda with 52% of seat in the parliament held by women in 2013, there are no other countries in the world where women have a majority of seats in parliament. Furthermore, notable are the relatively low gender gaps in the population with at least some secondary education in comparison to the high gender gap in labour force participation (UNDP, 2015).

1_{Human development includes three aspects.}

1. Reproductive health measured by: maternal mortality and adolescent birth rates.

2. Empowerment measured by: Proportion of parliaments seats occupied by women and share of women above 25 with secondary education in relation to share of men.

(11)

[10]

1.2 History of women in the labour force

Table 1 shows that the percentage of women participating in the labour force is substantially lower than the percentage of men. This difference has been even greater in history. In most countries, the gender gap in labour force participation decreased over the last few decades (Euwals, Knoef, & van Vuuren, 2011, p. 729).

The historical development of women’s labour force participation, unfortunately, is hard to analyse. Each country has its own historical path in the development of women’s labour force participation. Besides, there is no consistent cross-country historical data available on this topic. Especially a few decades ago, when women did contribute to production, this was not registered well since most of them weren’t formally waged employed (Pathak et al., 2013, p. 483; Stanfors, 2014, p. 518). There are, however, some generalizable trends which can be identified. In most countries, the relation between economic development and women’s labour force participation is U-shaped (Stanfors, 2014, p. 514).

Before industrialization, women contributed to production mostly on farms or in family firms. For these jobs, they most of the time did not directly get paid, and this type of work was not formally registered. With industrialization, women’s contribution actually declined. First, because of the social norms in that time, not approving women to do industrial labour. Second, because of the differences in physical strength between men and women. Third, because women did not need to contribute to the household income, since most of men’s income raised with industrialization (Fernández, 2007, p2-4; Simonton, 1998, p. 261-268; Stanfors, 2014, p. 520).

This downturn in labour force participation is followed by an increase when development continues (Stanfors, 2014, p. 514). More jobs becoming available, not requiring heavy manual labour, made it more socially accepted for women to work. Besides, with more girls getting a secondary and even tertiary education, increased the economic potential of women (Stanfors, 2014, p. 521). The availability of birth control, decreased the average number of children per family, which increased the opportunity cost of women, making it more remunerative to participate in formal employment (Euwals et al., 2011, p. 734; Ganguli et al., 2014, p. 176; Stanfors, 2014, p. 521).

(12)

[11]

In recent history, with the increasing possibility to work part-time and with new communication possibilities allowing to work from home, it is now also possible to combine wage employment with the household responsibilities (McKie, Biese, & Jyrkinen, 2013, p. 187). Therefore, now women of all ages are working, with or without having children (Balleer, Gomez-Salvador, & Turunen, 2014, p. 1386; Stanfors, 2014, p. 523). Besides, social norms changed in a way that it is now widely accepted that women both have formal work and a family. This is, amongst other things, because of new generations taken over the example, set by the older generation, and let these precedents evolve (Euwals et al., 2011, p. 734).

In spite of the closing gender gap in labour force participation, the gender gap in self-employment did not decrease with the same levels. Besides, female owned businesses usually are smaller in size, and are concentrated in certain sectors (OECD, 2012, p. 16). Mostly in the service sector, where then the entry barriers become so low, that the competition increases. This in turn, decreases the chance for survival and economic growth (Marlow, 2006, p. 595).

It is not in line with expectations for the female labour force participation to increase significantly in the near future in developed countries. Social norms seem to be rather stable, and also many supporting mechanisms to allow women to take part in formal labour are not likely to change a lot in developed countries (Euwals et al., 2011, p. 747). Notwithstanding, chances are high there is still a lot to gain in female self-employment. This does not necessarily implies that more women need to become self-employment, but also that the motives why women become self-employed need to change.

1.3 Women’s potential for economy

Whilst the amount of women with a tertiary education degree has increased significantly, this is not reflected to the same extend in the labour market. Most countries are still neglecting the potential of women’s contributions to the labour market. Not only there are less women than men having a top position at corporations, also the value women-owned businesses could have to a country’s economic development, growth and innovation seems to be highly underestimated (Welter, 2004). Empirical evidence tells us the metaphorical glass ceiling does exist (Kreide, 2003, p. 208; OECD, 2012, p. 15).

(13)

[12]

themselves but also generate employment to others (Euwals et al., 2011, p. 748; Pathak et al., 2013, p. 496; Verheul et al., 2006, p. 151). Additionally, they could contribute largely to the diversity of businesses. Female owned enterprises, in relation to male, usually operate in other sectors and are run in a different fashion with other goals and other structures. This could generate new impulses in the market and could enhance creativity and innovation (Pathak et al., 2013, p. 496; Verheul et al., 2006, p. 151).

Female owned enterprises, unfortunately, are more likely to fail than their male counterparts (Kodama & Odaki, 2011, p. 1237). Women are more likely to start a business out of necessity, to combine their responsibilities at home with generating an extra income for the household, and sometimes because of their inferior position on the labour market (Pathak et al., 2013, p. 484; Thébaud, 2015, p. 5). These women often start enterprises which will remain small. They enter a market with low entry barriers, which therefore is highly competitive. Next to that these women often do not have any managerial education or experience (Marlow, 2006, p. 592; OECD, 2012, p. 16; Thébaud, 2015, p. 5).

Notwithstanding, women with managerial experience do not fall short to men. Research, when controlling for managerial experience, tell us women are as likely to succeed as men (Kodama & Odaki, 2011, p1237). Analysis by the OECD (2012) indicates that female owned enterprises create the same amount of jobs as their male equivalent in the first years of business (p. 286). When taking into account all existing firms, including incorporated enterprises, a significant performance gender gap is apparent.

(14)

[13]

factors could be best focused on with policy measures. In the next section the role of institutions will be elaborated on.

1.4 Formal and informal institutions

Both in social and economic life people need to have a common understanding about how to interact with each other. This common understanding will only work if there are going to be consequences if a person does not comply to these shared agreements (Kasper & Streit, 1998, p. 2). These ‘rules’, also called institutions, have made it possible for humans to interact in a society and to build an economic life (Kasper & Streit, 1998, p1). Kasper & Streit (1998) also refer to the institutions as the: ‘the ‘software’ that channels the interaction of people and the development of society (p. 6).

Institutions are man-made. For a new institution to develop, a rule needs to be adopted by a group of significant size, for a long enough period of time (Kasper & Streit, 1998, p. 30). Some are even valued universally (Kasper & Streit, 1998, p. 71), namely:

- Individual freedom - Justice

- Security - Peace

- Economic welfare

In institutional theory a distinction is made between formal and informal institutions. Formal institutions are hard rules and easy to quantify factors, in contrast, informal institutions are soft not easy to quantify, socio-cultural factors. Both formal and informal institutions differ between societal groups, they can for example be evaluated on country level (Elam & Terjesen, 2010, p. 332; Pathak et al., 2013, p. 480). Institutions are framing, amongst other things, our gender perception (Elveren, 2014, p. 862; Pathak et al., 2013, p. 479).

(15)

[14]

In academic literature the employment choice model only taking into account a person’s personal characteristics and benefit maximization are prevalent (Klyver et al., 2013, p. 475). A minor share of academic work has taken an institutional approach, demonstrating a relation between several institutional factors and gender inequality in labour (Elam & Terjesen, 2010, p. 332).

Governments try to influence, but also make, these institutions in order to achieve certain goals they have set. When the market fails to distribute income and wealth equally, for example due to discrimination, politicians will try to ‘correct this market’ (Kasper & Streit, 1998, p. 304). In most Western societies, these kind of inequalities are not socially accepted, but did not yet have become obsolete. Therefore, politicians feel the need to intervene (Kasper & Streit, 1998, p. 80).

They, for instance, can implement policies to stimulate or discourage certain behaviours. They can also propagate certain ideas and behaviour. For example, in the Netherlands, the government broadcasts television commercials against discrimination. If implemented well, governments hold a powerful tool. Unfortunately, when it comes to the self-employment gender gap, this tool does not always has the desirable effect (Thébaud, 2015, p. 3). To a large extent this is because there is not yet a comprehensive image of what institutions really affect a woman’s choice to become self-employed. When there is a better understanding of these powers, it is likely that governments will be better able to implement measures to decrease this gender gap (Elam & Terjesen, 2010, p. 346).

For each country the current state is different. Most rules, social norms and culture have developed on country level. Each country has its own historical path, which results in different outcomes (Kasper & Streit, 1998, p. 35). In this research it is tried to find which institutions are most related to the self-employment gender gap. Data from multiple countries should result in a general image on which institutions governments could best focus their policies to achieve gender equality in self-employment levels.

1.5 Reading guide

(16)

[15]

2. Literature review

People in a context free of war and coercion, are able to make own individual choices. These choices, although made on an individual level, are likely to be influenced by the context in which these choices are made.

When it comes to employment choices, people are likely to choose for the opportunity which is most likely to give them the most beneficial outcome. Which opportunity is most desirable, as indicated by traditional economic literature, depends on both the monetary and the nonmonetary earnings. This is in literature also referred to as the occupational choice model or the utility function (Leung, 2006, p. 762; Lombard, 2001, p. 216; Malchow-Møller, Markussen, & Skaksen, 2010, p. 39; Torrini, 2005, p. 669). It is assumed that women put more emphasis on the nonmonetary benefits, leading to different outcomes for the model between men and women (Saridakis et al., 2014, p. 357).

Another stream of literature, balancing between the field of sociology and business, emphasises the interaction between the individual and its context. This stream gained less attention by academics, and is still not fully understood (Pathak et al., 2013, 478). From common sense, it seems logic that the individual choice is highly influenced by its context. For example, in an environment where women are still the one who have to take care of children, they need to take into account the cost of childcare into their utility function. The benefits generated by any form of employment, should at least outweigh the cost of child care. Otherwise a woman will not participate in the labour force, because this option is the least attractive outcome of her utility function (Connelly, 1992, p. 17; OECD, 2012, p. 15).

(17)

[16]

2.1 The economy

As explained in the introduction, the role of women evolved over time. These changes were highly related with to economic development in Western countries. We can see, that in less developed countries, the role of women in economy is somewhat comparable to our gender roles, decades ago(Donahoe, 1999, p. 543). The time when women also contributed by informal labour, something which was not formally administered (Pathak et al., 2013, p. 483; Stanfors, 2014, p. 518) .

Multiple researches have indicated the positive relationship between the participation rate of women in the formal labour force and economic development (Stanfors, 2014, p. 455; Elveren, 2014, p. 862; Klyver et al., 2013, p. 474). Despite, none of the OECD countries can be qualified as a third world developing county, there are still disparities between the economic situations in these countries.

2.1.1 GDP per capita

Whilst all OECD countries are industrialized and politically stable countries, there are still some major economic differences. In for example Turkey, the GDP per capita is $ 10.530,- and in Luxembourg this is $ 110.665,- (The World Bank, 2015). Per capita GDP levels are assumed to have a U-shaped relation with the gender gap in wage employment (Elveren, 2014, p. 863; Ganguli et al., 2014, p. 187). In early industrialization there is less need for women to generate income, plus there are less jobs female appropriate jobs available. When economies pass this stage, the gender employment gap tends do decline. Women in this stage, usually are stimulated to join the labour force by policies like parental leave and child care. This is likely to make female self-employment in richer countries even less likely, since the benefits of being employed increase (Klyver et al., 2013, p. 484).

The economic development state is likely to be related to the self-employment levels for both genders, nations with higher GDP per capita levels are having less business owners (Estrin & Mickiewicz, 2011, p. 407; Torrini, 2005, p. 662). Due to increasing GDP levels, wage employment becomes more attractive. Besides, it is likely that the need to flexibly combine generating additional income with caring for children decreases. Mainly because, the men are presumably better able to provide for the family on their own. Therefore, it is proposed that women’s self-employment levels are more affected than those levels for men. This results in the following hypothesis:

H1a2_{: The GDP per capita level positively affects the total self-employment gender gap.}

H1b: The GDP per capita level positively affects the opportunity self-employment gender gap.

(18)

[17]

In countries with a lower GDP per capita, the unemployment levels are higher, and thus the need for alternative income is higher (Thébaud, p. 2015, p. 13). Following, it can be assumed that people in poorer countries are thus more likely to become self-employed out of necessity. For women, wage employment, presumably, is not an appropriate alternative since the facilities to unburden their household responsibilities are less developed. Thus, when there is a high need for additional income, their only option will be self-employment. This reasoning leads to the following hypothesis:

H1c: The GDP per capita level positively affects the necessity self-employment gender gap.

2.2 The market

With economic development, the composition of the labour market changes. The labour market in the agricultural and manufacturing sector become smaller and the service sector usually grows (Rau & Wazienski, 1999, p. 505). An increase in the service sector is associated with more female friendly jobs, requiring less physical power (Ganguli et al., 2014, p. 178). To stimulate further economic development, female labour force participation is needed. This resulted in labour markets developments which promote female participation. For example, in some countries it became more common to hire part-time employees, which made is more easy for women to combine their family responsibilities with formal wage employment (Salladarré & Hlaimi, 2014, p. 296). These developments, in turn, presumably influence the levels of self-employment.

2.2.1 Part-time work

All OECD countries allow companies to hire part-time employees, to what extent the labour markets in these countries have part-time jobs available differs greatly. In the Netherlands, for example, 60,6% of women have a part-time jobs, and 19,6% of men. In contrast, in Hungary only 5,8% of women and 2,9% of men are part-time employed (OECD, 2015b). To give women the opportunity to work part-time, tends to result in higher female labour force participation rates (Balleer et al., 2014, p. 1411; Kreide, 2003, p. 213). The downside of this increasing participation levels, is that women who otherwise would work full-time or start their own firm, are now more drawn to part-time wage employment (Buchmann, Kriesi, & Sacchi, 2010, p. 284).

(19)

[18]

2012, p. 15; Thébaud, 2015, p. 28). Therefore, in compliance with the line of reasoning by Joona (2014), it is likely that ambitious women choose to be self-employment with opportunity motives, to achieve a fruitful career.

H2a: The availability of part-time wage employment positively affects the total self-employment gender gap.

H2b: The availability of part-time wage employment positively affects the necessity self-employment gender gap.

H2c: The availability of part-time wage employment negatively affects the opportunity self-employment gender gap.

2.3 The welfare state

In developed countries, the citizens relative to their needs, might be eligible to receive support from the state. People who are relatively wealthy are contributing to the state in the form of taxes, where the needy are helped by the state in the form of benefits (Kasper & Streit, 1998, 320).

Each country has their own type of welfare state. Some countries only provide the basic benefits to all, and provide additional benefits to the participants in the labour force. Whilst other countries provide welfare to all. In the Scandinavian countries, for instance, all people have equal rights to all benefits. For example, every child is eligible for childcare, whether the parents are employed or not (Welter, 2004, p. 215). Switzerland, in contrast, has only moderate set of social benefits, where for example a father has no right on paternity leave at all (OECD, 2015c).

These social expenditures are extremely costly for states. Each year they have to spend a great share of the gross domestic product on these kind of benefits. So why are governments investing so much? This is, first, because of the idea of social justice (Kasper & Streit, 1998, 320), and second, because these expenditures hopefully will result in economic development in return. Providing paid maternity leave and provision of child care, for example, should encourage women to stay active in the labour force.

2.3.1 Childcare arrangements

(20)

[19]

working hours for participating women (Euwals et al., 2011, p. 736; Ganguli et al., 2014, p. 192). Research tells us that these women become self-employed, because this type of employment gives them the flexibility to combine formal labour with their caring responsibilities (Connelly, 1992, p. 17; Marlow, 2006, p. 593; McKie et al., 2013, p. 185; Thébaud, 2015, p. 2; Wellington, 2006, p. 360). It is a choice made individually. Though, a choice likely to be influenced by context (Elam & Terjesen, 2010, p. 346). The way child care is arranged, is one key element of this context (OECD, 2012, p. 15; Verheul et al., 2006, p. 162).

Each country has its own policies and customs with respect to care for children. The Nordic European countries are now known for their extensive provision of child care by the state. Only, also in Sweden, only two decades ago, the male breadwinner model was prevalent (Stanfors, 2014, p. 517). The state made child care available for every child. Not only with the goal that each woman then would be able to participate in the labour market, but also to improve the well-being and education of all children (Eydal & Rostgaard, 2011, p. 162). In Greece and Turkey, in contrast, it is still common to take care for the children at home. This does not necessarily mean that women cannot participate in the labour force. Due to the importance of family in these countries, tot a large extent, grandparents are the ones taking care of the children (Aycan & Eskin, 2005, p. 458; Svensson-Dianellou, Smith, & Mestheneos, 2010, p. 250).

Euwals et al. (2011) show that in the Netherlands, the negative effect of the presence of children in the household on labour force participation, has decreased significantly between 1992 and 2004. This decline is not explained by declining fertility (p. 744). One of the main explanations for this change was the increasing availability and decreasing cost of child care. In addition, it also became more common to make use of these facilities (p. 749).

Thus, when families have greater access to child care outside the household, women are less likely to become self-employed for flexibility motives. Probably they are more drawn to wage employment, when the context allows them to combine this with having children (Elam & Terjesen, 2010, p. 336; Thébaud, 2015, p. 27). This results in the following hypothesis:

H3a: The availability of child care outside the household positively affects the total self-employment gender gap.

(21)

[20]

their business. When women do not need to choose for self-employment out of necessity, the women who still do start their own firm, are likely to do this with opportunity motives, with growth orientation and ambition (Estrin & Mickiewicz, 2011, p. 412; Thébaud, 2015, p. 28).

H3b: The availability of child care outside the household positively affects the necessity self-employment gender gap.

H3c: The availability of child care outside the household negatively affects the opportunity self-employment gender gap.

2.3.2 Parental leave system

By nature, we need women to give birth. Simply put, without babies our species would extinct. Though, before extinction, our societies and economies in the whole world would be out of balance. These systems build on the principle that each new generation has to fulfil its role in society. The elderly, for example, are depending on the youth to care for them. In most Western societies, this is not only done directly, but also by paying taxes, so the state can provide care. Therefore, it stays important, that women are able to give birth to enough children, to keep the system intact (Johnson, 1997, p.1897). Thus, economies not only benefit from women participating in the labour force, these economies also benefit eventually from the children they give birth to, who later in their turn join the labour force.

Though, this is contradictory. Giving birth and taking care of children, inhibits women to participate in the labour force. At least for the time their physically not able to work. Therefore, Western governments have made rules and arrangements, to facilitate and promote women, to combine childbearing and contributing to economy. These rules and arrangements differ widely across nations (Bruning & Plantenga, 1999, p. 197).

(22)

[21]

According to previous research, the duration of maternity and parental leave are associated with the self-employment level of women (Estrin & Mickiewicz, 2011, p. 412; Joona, 2014, p. 2; Thébaud, 2015, p. 29). Though, there is no unambiguous direction of the relation between these variables. Following, Estrin & Mickiewicz (2011), a positive relation between the provision of maternity leave and the self-employment gender gap is expected (p. 412). However, this is not in conformity with the indications of Joona (2014). She proposes, that extreme durations of maternity leave are positively related with the female opportunity self-employment (p. 4). The research by Thébaud (2015), shows that women are more likely to opt for wage employment if the state provides them with medium length of maternity leave (p. 7). Therefore the following hypotheses are proposed, where the leave for mothers is the sum of maternity and parental leave:

H4a: The duration of leave for mothers negatively affects the total self-employment gender gap.

H4b: The duration of leave for mothers positively affects the necessity self-employment gender gap.

H4c: The duration of leave for mothers negatively affects the opportunity self-employment gender gap.

2.3.3 Unemployment benefits

It is argued that the provision of unemployment benefits has a negative effect on self-employment levels. Wage self-employment is more attractive to both men and women, since they are entitled to receive unemployment benefits. Besides, for the unemployed there is less necessity to start their own business, since they receive their benefits (Estrin & Mickiewicz, 2011, p. 400; Malchow-Møller, Markussen, & Skaksen, 2010, p. 47; Torrini, 2005, p. 667). Notwithstanding, women who voluntarily left the labour force to become a mother, may later encounter troubles to re-enter the labour market. For them, self-employment stays an attractive alternative, since they are usually not eligible for these kind of benefits (Kreide, 2003, p. 211).

(23)

[22]

negatively affect the choice for these women to become self-employed from opportunity motives, since the unemployed have an advantage relative to them. This results in the following hypotheses:

H5a: The level of unemployment benefits negatively affects the total self-employment gender gap.

H5b: The level of unemployment benefits negatively affects the necessity self-employment gender gap.

H5c: The level of unemployment benefits positively affects the opportunity self-employment gender gap.

2.4 Legislation

The employment choice is also likely to be influenced by the rules and regulations of the context in which the decision is made. When it comes to labour, the rights and obligations you have as an employee or employer, are highly dependent on the country in which you are working.

Highly regulated environments on the upside, can create a safe environment for employees and entrepreneurs. For example, making sure that employees cannot be laid off without sufficient grounds, or inhibit large firms to overpower small firms. High administrative load, on the downside, could discourage small firms (Torrini, 2005, p. 667).

According to Estrin and Mickiewicz (2011), the downsides of a weak rule of law, are likely to outweigh these administrative burdens, due to rising transaction costs and increasing risk for small firms (p. 399). Two types of legislation are assumed to be related to the self-employment gender gap, and will be elaborated on.

2.4.1 Minimum wage legislation

Multiple governments are obliging the employers in their countries to pay their employees a minimum wage. This regulations make sure employees will get decent remuneration for their labour. However, not all countries have these sort of rules. The level of this minimum wage Is also different in each country (OECD, 2015d).

(24)

[23]

employees. This only leaves them the option to become self-employed out of necessity (Malchow-Møller et al., 2010, p. 47).

Establishing minimum wages also has an impact on a phenomenon referred to as wage compression. When minimum wages are set relatively high, or are increasing relatively fast, this will result in very small differences in pay between incumbents and new entry employees. Therefore, the incumbent, more productive, employees will feel like they are underpaid. These employees are likely to explore their options to become self-employed, to get higher returns for their efforts (Malchow-Møller et al., 2010, p. 47). This however, is issue is likely to be relative for both genders. Joona (2014), for instance, indicates that ambitious women in Sweden become self-employed, because they cannot achieve a satisfactory job level in wage employment (p. 24).

On the other hand, minimum wages also increase the opportunity cost of not working and of having a low income small business. Therefore, people are more likely to be drawn to waged employment, instead of starting a small firm with necessity motives. Research also indicates that in particular women at the lower ability end, are likely to earn more in wage employment (Ganguli et al., 2014, p. 176).

H6a: Minimum wage levels are positively affecting the total self-employment gender gap. H6b: Minimum wage levels are negatively affecting the necessity self-employment gender gap.

2.4.2 Tax regulations

Previous research proposes that highly regulated environments are associated with high bureaucracy and heavy loads of administration. In these environments, it is likely that tax evasion will be more difficult, and thus on average firms need to spend more time on registering their profits and incomes, in combination to the taxes, and cannot evade this burden.

Especially for smaller firms, this administrative obligations can have a great impact on their productivity, and therefore also on their profits. Therefore, the income these firms generate should not only outweigh the opportunity cost of wage employment, but these opportunity cost plus the loss in profits due to bureaucracy (Torrini, 2005, p. 667).

(25)

[24]

out of necessity, it is not likely there will be a sufficient alternative. Thus despite, of this great negative effect, it is likely these women still will become self-employed, but their profit margin will decline, which also might endanger their survival chances.

When it comes to opportunity driven self-employment, previous research shows that especially this type of self-employment is affected. It is proposed that high administrative responsibilities and taxes are having a discouraging effect, and alternative employment options are becoming more attractive (Estrin & Mickiewicz, 2011, p. 400). This relation however, is not likely to have a gender component.

The research by Torrini (2005) shows that indeed, countries with higher tax evasion possibilities, are having on average higher self-employment rates (p. 681). Building on this finding, it might be the case that this is especially true for smaller firms, where large companies a more interesting for governments to detect. These companies need to pay higher taxes, and are thus more remunerative for the state. Small firms, only generating income for the individual owning the business, might not be worth the effort to detect them. Since especially female owned necessity driven firms are characterised by being and staying small, these companies are likely to be most able to evade taxes. This results in the following hypotheses:

H7a: The opportunity to evade taxes negatively affects the total self-employment gender gap. H7b:The opportunity to evade taxes negatively affects the necessity self-employment gender gap.

2.5 The gendered society & culture

(26)

[25]

Previous literature indicates three specific institutions which could influence the gender gap in self-employment. These will be explained in the remainder of this section.

2.5.1 Gendered industries

The natural characteristics, physical strength, which is highly related to gender, caused men to become the ones who performed the heavy manual labour in history. Thus automatically, manufacturing and agriculture were sectors with large shares of male employees. Nowadays, after industrialization, there are only a few jobs which demand heavy manual labour. Most heavy tasks are now performed by machinery. Therefore, it would be expected that men and women should be able to fill in the same type of jobs. Nevertheless, industries are still highly gendered, with women mainly being active in the service sector. According to previous research, these kind of jobs were perceived as being more congruent to the image society had formed of femininity (Elam & Terjesen, 2010 , p334; Stanfors, 2014, p. 521).

The extent to which industries are gendered in wage employment, is likely to result in gender differences in the division of men and women across industries in self-employment. Besides, gendered industries are also likely to influence the self-employment gender gap. It can be assumed that this is due to the mechanism of statistical discrimination. If by gradual development, women are becoming overrepresented in services. This is likely to evolve in the general assumption that every woman will perform best in the service sector. This prohibits women in general to become self-employed in other industries, since by the stereotype, they are not suited to participate in these sectors. Therefore, they are less likely to get finance for these kind of firms, and are less likely to be get a sufficient client base (Budig, 2006, p. 727; Marlow, 2006, p. 594; Verheul et al., 2006, p. 161).

This leads to overcrowding by female businesses in the service sector, which lowers the entry barriers, but also decrease the changes on survival due to high a highly competitive environment. If these, generally small firm are able to survive, they are not likely to generate growth, simply because the supply will be too large. Thus, since this sector is probably will be overrepresented, women starting a firm with opportunity motives, do not have a lot of growth potential in this sector. Unfortunately, since the entry barriers to other sectors will be higher for women than for men due to this statistical discrimination, their possibilities in these sectors will be relative low as well. This results in following hypotheses:

(27)

[26]

H8b: Gender segregation in the service sector negatively affects the necessity self-employment gender gap.

H8c: Gender segregation in the service sector positively affects the opportunity self-employment gender gap.

2.5.2 Gender education gap

One of the main gender inequality measures used by intergovernmental organizations such as the World Bank, the United Nations, etc. is the gender education gap. It is assumed that women, if they have had the same educational opportunities as men, can reinforce their position in society relative to men (OECD, 2012, p. 79). Besides. it is universally believed that economic, political and societal development is positively influenced by educating children (Ganguli et al., 2014, p. 174).

If they have the opportunity to attend school, girls on average tend to perform better than boys, and are less likely to drop out (OECD, 2012, p. 13). These numbers, unfortunately are not reflected, to the same extend, in the labour force and in self-employment levels (Ganguli et al., 2014, p. 200). Notwithstanding, multiple scholars have demonstrated a positive relation between women’s education levels and female participation levels in both wage employment and self-employment (Balleer et al., 2014, p. 1388; Estrin & Mickiewicz, 2011, p. 405; Euwals et al., 2011, p. 741; Pathak et al., 2013, p. 496; Stanfors, 2014, p. 521).

With more girls getting proper education, the average opportunity cost of women not working increases. Women, through education, invest in their human capital. In turn, higher human capital should result in higher earnings. Therefore, in general, women are likely to be more eager to work when they are educated (Ganguli et al., 2014, p. 175). In addition, having an equal or even a higher educational background relative to men, enhances the position of women on the labour market. Giving them the opportunity to attain better paid and more challenging jobs in wage employment (Stanfors, 2014, p. 521).

Women’s motivations to become self-employment are also likely to be influenced by their educational background. When women have more human capital, are also more likely to have the confidence to start a fruitful business (Estrin & Mickiewicz, 2011, p. 405). These indication lead to the following hypotheses:

(28)

[27]

H9b: : The educational gender gap positively affects the necessity self-employment gender gap.

(29)

[28]

3. Methodology

Based on the literature, hypotheses are drafted. Next to testing the hypotheses, also the weight of each theme in explaining the self-employment gender gap will be analysed. This will result in an analysis giving information about which institution relates to the self-employment gender gap, the directions of these relationships, and the themes on which governments should focus their policies.

The proposed hypotheses will be tested with a multivariate regression analysis using secondary data from multiple sources.

3.1 Gendered institutions

To come to the selection of institutions included in the regression, a systematic literature search is performed. Using the WorldCat search engine, a search was performed with the following keywords: women, female, gender, labour force, self-employment, entrepreneurship, labour market institutions, and institutional economics. Only peer reviewed articles were taken into account.

After scanning each title and reading abstracts of possible relevant articles, a selection of 67 articles was made. Each article included one or a couple of institutions in the analysis. These were filtered into an overview after which a selection was made, based on whether the institution has a proposed gendered aspect and the frequency of using the institution in previous research. The selected institutions were arranged by theme.

(30)

[29]

3.2 Data

Data is gathered from multiple secondary sources, namely: The World Bank, The OECD, Eurostat, GEM, United Nations Development Programme and Transparency International. Data is gathered for the years 2002 up to 2012, since older and more recent data is not yet made publically available for all measures.

Only developed countries are included in the analysis. First and foremost, because, previous research indicates great differences in the self-employment gender gap between developed and developing countries (Kasper & Streit, 1998, p. 28). Second, due to the lack of availability of data for these countries.

Developing countries, to some extent, are now facing the same road ahead, as developed countries faced decades ago. As explained in the introduction, the contribution of women to the labour force changed a lot due to industrialization. The type of labour changes for women, and also the extent to which their contributions are registered and are remunerated. Therefore, it is likely that the challenges these countries face to close the gender gap are different from the challenges of developed countries. Also, it is likely that the measures to help these women are from a completely different nature. In these countries the focus lies at providing in the minimum living standards, instead of improving the economic state and decreasing the social inequalities. To be able to draw conclusions and to provide implications for governments, this article focusses on developed countries.

3.2.1 Sources

The dataset is build up from multiple sources. Each organisation focusses on gathering data concerning specific themes. Therefore, there is nog single database available which contains all themes used in this analysis.

Using multiple sources increases the possibility to find the measure which represents the variables in the regression. Downside, on the other hand, is that not every organisation gathers information annually. Besides, there is no standard for which countries to include in the database, it highly depends on which sources are used, for which countries data will be available.

(31)

[30]

Cooper, & Schindler, 2008, p. 318). This resulted in a dataset containing 328 cases covering 30 countries.

For each institution, one indicator is included in the dataset. Some indicators are indirect measures, for example, the Corruption Perception Index as a measure for the possibility to evade taxes (Torrini, 2005, p. 681). An overview of which measures are used for each institution is given in table 2. Appendix A states an overview of all measures with their sources.

Appendix C presents an overview of the descriptive statistics before standardization, but after data preparation. These values gives a clear and easy to interpret summary of the data used for this research.

Table 2 – Institutions included in regression by theme

THEME INSTITUTION MEASURE(S)

ECONOMY GDP per capita - GDP per capita

MARKET Part-time employment - Incidence of female part-time employment WELFARE Child care - Social expenditure on Early Childhood,

Education, & Care (ECEC) Parental leave - Total leave for mothers

Unemployment benefits - Public expenditure on unemployment compensation

LEGISLATION Minimum wage -Minimum wage level

Tax regulations -Corruption perception index (CPI) SOCIAL & CULTURE Gendered industries -Service sector gender gap

Gender education gap -Gender parity index (GPI)

3.3.2 Reliability and validity

Since this research only uses secondary data, there are a couple of issues which need attention. As explained only renowned sources are used, which should avoid measurement issues and biased data. For each individual variable only one source is used. Using multiple sources could decrease the reliability, since each organization could define a construct in its own terms and could measure it slightly different.

(32)

[31]

To avoid validity issues, only measures, used by previous research to measure the same construct are included. The corruption perception index, for example, is amongst others already used in the research of Torrini (2005), to measure the opportunity to evade taxes.

3.3 Data preparation

Prior to performing the regression analysis, the dataset was prepared. In Appendix B, each step is recorded.

3.3.1 Missing values

First, it was checked whether the values were missing completely at random, or not. Little’s MCAR test was highly significant (χ2_{(1, N=402) = 923,80, p= 0,000), which meant the values were not} missing completely at random. Therefore, simply replacing the missing values by the series means, would possibly result in biases in the test outcomes.

To avoid these biases, subjects, variables or cases, with little or no data entries, were deleted from the dataset. Thus, first countries which had no complete cases were deleted from the dataset. Then, specific cases, too short on data, were deleted. This step was followed by replacing all other missing values either by imputed values or the series mean. The imputation method was preferred, since using panel data, this method is likely to proxy the real values best. Already having only a couple entries for a country for a specific measure, could provide the base to estimate the other years (Huizingh, 2007, p. 179-181). The series mean method was only used when imputation resulted In highly unlikely values, as a result of having only two data entries for a specific measure for a subject.

3.3.2 Data check

To be able to perform a regression analysis, all measures should be approximately be normally distributed. Therefore, for all variables histograms were plotted. These plots showed that the data fulfil this condition. Transformation of all measures using the natural logarithm, resulted in somewhat better distributed data, though this also results in data loss for all negative values. Therefore, this method is not applied.

(33)

[32]

disproportionate effect in relation to the other, simply because a different measurement scale was used.

Applying a univariate analysis indicates that there are no fixed time effects. Therefore it can be assumed that the dependent variables are not interacting with the years the of data cases. Figure 1 shows the profile plot of the analysis between the total self-employment gender gap and time. All subject lines being parallel to each other, indicates that these two variables do not interact.

Figure 1: Profile plot of analysis between total self-employment gender gap and time

All independent variables are standardized before correlation analysis and the regression analysis. Multicollinearity is accounted for whilst performing the analysis. These results indicate that there is some multicollinearity, but none of the outcomes are above the rule of thumb of 10 for the VIF value (the maximum VIF value being 3,531 for the share of women in parliament). Thus, there is no reason to believe the regression results are biased by this phenomenon (Huizingh, 2007, p. 309).

3.4 Analysis

The analysis will include both include a Pearson correlations test and a regression analysis. Executing both will give more insight on both the strength and the direction of the relations between the independent and the dependent variables.

Es ti mate d m ar gi n al m ea n s

Estimated marginal means of self-employment gender gap

(34)

[33]

Since there are three different dependent variables, which are evaluated separately, there are performed three separate regressions. Each regression only having a different dependent variable, but including all independent variables.

For total self-employment gender inequality following equations is estimated:

𝑌

_{𝑇𝑜𝑡𝐺𝐺=𝛽0+𝛽}_𝐺𝐷𝑃_𝑋+𝛽_𝑃𝑇_𝑋+𝛽_𝐶𝐶_𝑋+𝛽_𝑃𝐿_𝑋+𝛽_𝑈𝐵_𝑋+𝛽_𝑀𝑊_𝑋+𝛽_𝑇𝑅_𝑋+𝛽_𝐺𝐼_𝑋+𝛽_𝐺𝐸_𝑋+𝜀

For necessity self-employment gender inequality following equations is estimated:

𝑌

_{𝑁𝑒𝑠𝐺𝐺=𝛽0+𝛽}_𝐺𝐷𝑃_𝑋+𝛽_𝑃𝑇_𝑋+𝛽_𝐶𝐶_𝑋+𝛽_𝑃𝐿_𝑋+𝛽_𝑈𝐵_𝑋+𝛽_𝑀𝑊_𝑋+𝛽_𝑇𝑅_𝑋+𝛽_𝐺𝐼_𝑋+𝛽_𝐺𝐸_𝑋+𝜀

For opportunity self-employment gender inequality following equations is estimated:

𝑌

_{𝑂𝑝𝑝𝐺𝐺=𝛽0+𝛽}_𝐺𝐷𝑃_𝑋+𝛽_𝑃𝑇_𝑋+𝛽_𝐶𝐶_𝑋+𝛽_𝑃𝐿_𝑋+𝛽_𝑈𝐵_𝑋+𝛽_𝐺𝐼_𝑋+𝛽_𝐺𝐸_𝑋+𝜀

3.5 Results

The Pearson correlation is executed for all dependent and independent variables at once. Since there are hypotheses drafted including an assumption about the direction of the relation, the correlation analysis is performed one tailed. Table 3 presents the correlation results.

(35)

[34]

Table 3 – Pearson correlations

Dep1 Dep2 Dep3 Ind1 Ind2 Ind3 INd4 Ind5 Ind6 Ind7 Ind8 Ind9 Dep1: Self-employment

gender gap 1

Dep 2: Necessity motivated self-employment gender gap

,115* ₁

Dep3: Opportunity motivated self-employment gender gap

-,007 ,182** ₁

Ind1: GDP per capita _,467** _,044 _-,034 ₁

Ind2: Incidence of female part-time employment

,096* _,024 _-,052 _,561** ₁

Ind3: Social expenditure on ECEC ,602

** _,054 _-,051 _,415** _,050 ₁

Ind4: Total leave for mothers ,256 ** ,165** -,057 -,181** -,356** ,064 1

Ind5: Public expenditure on unemployment compensation ,302** -,044 -,064 ,238** ,231** ,191** ,014 1

Ind6: Minimum wage

levels ,084 ,077 -,050 ,058 ,367

** _-,164** _-,221** _,245** ₁

Ind7: Corruption perception index (Tax regulations)

,508** _,073 _-,049 _,719** _,510** _,466** _-,076 _,243** _,110* ₁

Ind8: Gender gap in service sector -,548

** _,110* _,093* _-,068 _,055 _-,157** _-,341** _-,267** _,110* _-,174** ₁

Ind9: Gender parity index (education) ,549** -,005 -,050 ,148** -,063 ,388** ,040 ,263** ,037 ,260** -,634** 1

*. Correlation is significant at the 0.05 level (1-tailed). **. Correlation is significant at the 0.01 level (1-tailed). (Listwise N=328)

Table 4 – Regression analysis explaining the Total self-employment gender gap

β t-value VIF

(Constant) ,319 47,660

Ind1: GDP per capita ,069 6,520 2,526

Ind2: Incidence of female part-time employment -,032 -3,231 2,184

Ind3: Social expenditure on ECEC ,087 9,895 1,705

Ind4: Total leave for mothers ,038 4,644 1,460

Ind5: Public expenditure on unemployment compensation

-,004 -,482 1,266

Ind6: Minimum wage levels ,056 7,109 1,399

Ind7: Corruption perception index (Tax regulations) ,018 1,670 2,535

Ind8: Gender gap in service sector -,082 -8,147 2,283

Ind9: Gender parity index (education) ,017 1,616 2,339

(36)

[35]

Table 5 – Regression analysis explaining the Necessity self-employment gender gap

(Constant) ,339 15,278

Ind1: GDP per capita ,011 ,321 2,526

Ind2: Incidence of female part-time employment ,031 ,929 2,184

Ind3: Social expenditure on ECEC -,001 -,020 1,705

Ind4: Total leave for mothers ,134 5,001 1,460

-,030 -1,196 1,266

Ind6: Minimum wage levels ,037 1,415 1,399

Ind7: Corruption perception index (Tax regulations) ,023 ,663 2,535

Ind8: Gender gap in service sector ,132 3,934 2,283

Ind9: Gender parity index (education) ,077 2,278 2,339

R-Square 0,104 Adjusted R-square 0,078 Significance 0,000 N 328 Degrees of freedom 318

Table 6 – Regression analysis explaining the Opportunity self-employment gender gap

(Constant) ,476 24,536,

Ind1: GDP per capita ,019 ,680 1,988

Ind2: Incidence of female part-time employment -,027 -1,030 1,764

Ind3: Social expenditure on ECEC -,016 -,732 1,329

Ind4: Total leave for mothers -,022 -,996 1,315

-,013 -,590 1,208

Ind8: Gender gap in service sector ,025 1,169 1,221

Ind9: Gender parity index (education) ,025 1,152 1,244

R-Square 0,021 Adjusted R-square -,001

(37)

[36]

5. Conclusion

The regression analysis and correlation analysis do not support any of the proposed hypotheses with the opportunity self-employment gender gap as the dependent variable. In the discussion, there will be further elaborated on this matter.

Strong support however is found for hypothesis H1a, H3a and H4b. Thus, based on this research, it can be concluded that higher levels of GDP per capita are increasing the total self-employment gender gap. A strong correlation is found for this relation (r (326)=0,467 , p<0,01) and also the regression analysis supports that GDP per capita explains a significantly predicted the total self-employment gender gap (β=0,069 , t(318)=6,520 , p<0,01).

The proposed positive relation between child care availability (measured by social expenditure on ECEC) and the total self-employment gender gap is also found to be highly significant. The two variables were not only highly correlated ( r (326)=0,602 , p<0,01), but also the regression was significant in the proposed direction (β=0,087 , t(318)=5,001 , p<0,01).

The third hypothesis supported by the correlation analysis ( r (326)=0,165 , p<0,01)) and regression analysis is H4b (β=0,134 , t(318)=9,895 , p<0,01). Therefore it is implied that a longer period of leave for women around and after giving birth, is positively related to the necessity self-employment gender gap. The analyses indicates, in contrast to the direction proposed in the hypothesis H4a, the same relation for leave for mothers and the total self-employment gender gap. Both the correlation ( r (326)=0,256 , p<0,01) and the regression are highly significant (β=0,038 , t(318)=4,644 , p<0,01).

Strong support, but in the opposite direction was also found for H8a and H8b. Thus, in contrast to the propositions, the gender gap in the service sector is negatively related to the total self-employment gender gap ( r (326)=-,548 , p<0,01) (β=-0,082 , t(318)=-8,147 , p<0,01), and the gender gap in the service sector is positively related to the necessity self-employment gender gap ( r (326)= 0,110 , p<0,01) (β=0,132 , t(318)=3,934 , p<0,01).

In table 7 a complete overview is presented of which proposed relations are supported. A shown, H6a is also supported, but only with weak evidence. H2a and H9b are both supported but in opposite direction. Also for these hypotheses the evidence was very weak.

An institutional perspective on the self-employment gender gap: What factors should policy focus on?