A societal perspective on self-employment – Sweden as an example

ABSTRACT

European Union policies emphasise the importance of entrepreneurship and self-employment to maintain economic growth, a solution influenced by the USA and research from schools of business. There are expectations of higher education graduates that they are innovative, will start businesses and will employ others. However, transferring a solution from one country to another may not be as simple as that. This study questions the recommendations made in EU policies which mainly focus on how to support individuals to become self-employed. Extensive data from the Swedish population registry of approximately 90,000 tertiary-educated people, all born in the mid-1970s, is used to analyse the extent to which this group became self-employed. Results show that societal and cultural differences need to be considered when creating such policies.

KEYWORDS:

• Higher education
• tertiary educated
• self-employment
• entrepreneur
• gender

Point of departure: an individual perspective

Creating new businesses and services is seen as a way to keep unemployment levels low and thus preserve or even improve living standards in European countries, as in several other countries. Earning a livelihood has increasingly become an individualised responsibility and it is assumed that individuals with the ‘right’ characteristics will take the lead in starting new businesses (Simoes, Crespo, and Moreira 2016; Yendell 2001). There is a belief that university graduates will be at the frontline of creating these new employment possibilities (Gibb 2011; Roman and Maxim 2017; Teixeira and & Davey 2010), an objective that the European Union’s policy strongly promotes (e.g. European Commission 2012a; European Commission 2013). Moreover, since fewer women in most European countries are self-employed compared to men, they are seen as an ‘under-exploited source’ (European Commission 2017a), a source to be subjected to policies about increased self-employment.¹ European governments, including that of Sweden, have adapted similar policies to come to terms with the perceived low rate of self-employment (e.g. Ahl and Nelson 2015; Europeiska kommissionen 2008; Pettersson et al. 2017).

Business schools have long held a leading position in research about entrepreneurship (e.g. Babson College 2018). Conventional economics has been guided by the theory of ‘constrained optimisation in equilibrium’ a theory that ignores the context in which the individual makes decisions (Foster 2017). Economics generally (Gruszka, Scharbert, and Soder 2017; Foster 2017) have been influenced by physics and mathematics and explanations based on historical development or social and cultural structures have been overlooked. The focus on financial returns, the labour market and on individual traits strongly influences the field of entrepreneurship (Zachary & Mishra, 2011); the variety within education and its diverse fields of knowledge is often neglected when calculations of financial return from education are based on years of schooling,

Moreover, results from research conducted in one type of society or welfare state (liberal) may not easily be implemented in other kinds of societal organisation (e.g. social democratic). Push or pull factors into self-employment such as income or unemployment do not have the same influence on people’s decisions depending on how the welfare state and labour market is organised and how traditions and culture affect people’s perceptions (GEM 2018; Hjorth 2008). Also within the field of economics itself, the assumption of a historically, socially and culturally isolated individual who makes decisions in an abstract world is being criticised (e.g. Dopfer, Foster, and Potts 2004; Foster 2017; Foster and Metcalfe 2012; Gruszka, Scharbert, and Soder 2017). This study contributes with empirical facts showing that ‘one rule’ (Foster and Metcalfe 2012, 431) does not fit all.

Previous large-scale research on self-employment

The extent to which higher-education graduates start businesses (not their intentions) is rarely studied. However, there are some examples with a different focus; the question of whether graduates from internationally ranked universities were more likely to become self-employed than graduates from non-ranked universities was the focus of a Swedish study by Daghbashyan and Hårsman (2014). Using Swedish registry data, they analysed outcomes of three million higher education graduates from the social sciences, education, the natural sciences, technology and medicine. Their results showed that graduates from ranked universities were more likely to become self-employed, except for technology graduates for whom the type of institution did not seem to be of importance. In a Danish study by Buenstorf, Nielsen, and Timmermans (2017) the focus is on those who did not graduate (drop out), if they were more likely to become self-employed than those who graduated. They also analysed register data but reduced the student population to those who had studied the natural sciences, engineering, the social sciences, business, humanities, education and those who had studied mixed fields. Nearly 108,000 individuals were analysed of whom 28% did not complete their studies were educational drop-outs. Drop outs were more likely to register a firm, but they earned less than those who were employed. The authors criticise the idealisation of a few successful drop-outs and stress that studies conducted in the USA are not comparable to the Scandinavian context with its different welfare systems and a different perception of self-employment. Register data with information on the entire population is only available in the Nordic countries, and researchers in other countries need to rely on other sample-based sources of information, often questionnaires. In Austria (Falk and Leoni 2009; Leoni and Falk 2010), census data (probably questionnaires) including 366,000 higher-education graduates in a range of age groups were analysed, and their likelihood of self-employment was estimated. Many influencing factors were included, and the analyses were divided by gender. The results showed that men were more likely to become self-employed than women and that the prevalence of self-employment differed depending on the field of knowledge. Gendered educational choices were an important explanation of the different outcomes in the labour market. The authors also stressed that international comparisons are crucial to become aware of institutional and other country-specific factors. A German study (Beiler 2017), also analysed survey data from questionnaires and administrative data. The author focused on how the economic conditions in a country influence the likelihood of graduates becoming self-employed. He compared graduates aged 23–32 years, from eight fields of study over an eight-year period, but did not consider gender. He found that in prosperous times, this group of graduates was more likely to start a business shortly after graduation and that the longer the period since graduation, trade conditions seemed not to influence the decision. Livanos, Yalkin, and Nuñez (2009) compared Greece and the UK and one reason for this comparison was the large difference in unemployment rates. The researchers utilized data from nationally conducted surveys that included all age groups and all levels of education. Their focus was on women and the proportions of self-employed, employed, and unemployed people in relation to civil status and length of education. In both countries women with less schooling were more likely to be self-employed. Self-employment was more common among married women in Greece, while single women in UK seemed to be slightly more likely to be self-employed. The authors discuss these differences in terms of economic structure and institutional characteristics, such as high unemployment in Greece and its weak social benefit system. Another example in which unemployment rates are high in Italy (OECD 2015, 112). Rosti and Chelli (2009) analysed data from a national labour force survey based on recurring interviews with the purpose of mapping the flow of people in and out of self-employment, employment and unemployment. This study shows a substantial flow of graduates from employment to self-employment and vice versa. Self-employed women were particularly likely to leave self-employment for employment. The explanations were individual, such as women were seen as less able, or because of gender discrimination. The field of knowledge was not considered in this study.

The Swedish case

Since the university sector was established, the aim has been to educate people for positions in administration (Collins 2006). Even today this is valid; tertiary studies are designed for positions in the public sector, and for employment in the private sector. Unemployment among tertiary educated is generally low (OECD 2012; 2015, 112). Also, after the 2008 recession, tertiary-educated people largely remained in the labour force; the highest unemployment level in Sweden was 4.5 per cent in 2010 (people aged 25–64 years) (Eurostat 2018). The availability of basic insurance and unemployment benefits (Hjorth 2008) is also likely to influence peoples’ decisions about self-employment, implying that people are less forced to move, or to become self-employed to financially survive. Swedish unemployment benefits start from approx. 11,000 SEK per month, around 1000 EUR, but are usually augmented by insurance connected to previous employment providing 80 per cent of former salary (European Commission 2017b). Moreover, it is possible that tertiary-educated people take jobs for which they are over-qualified (Archer 2003). To a large extent, women and men make traditional gendered educational choices and this gender division is further upheld by the organisation of the educational system where traditionally male-dominated trades are offered in upper secondary schools, while most of the traditionally female-dominated trades and semi-professions are housed in higher education. This organisation more or less forces most women to go into higher education. The result is a horizontal gender division within higher education by which men make up the majority in engineering programmes and women are to be found studying most other programmes (Lindberg, Riis, and Silander 2011; SCB 2018, 40–41). Although Sweden is considered to be a gender-equal country (EIGE 2018), with extensive paid parental leave and state-sponsored availability of childcare (European Commission 2017b; Hjorth 2008), compared to men, women are more likely to work part-time and take greater responsibility for the family (Delamont 2001; Pettit and Hook 2005; SCB 2018).

In this study, the extent to which tertiary-educated people from various fields of study become self-employed is analysed, with a particular interest in gender differences. The results are discussed in relation to contemporary research on self-employment among men and women; social structures such as parents’ resources; organisational structures such as the educational system and the labour market structure; and the influence of the type of welfare state.

Method: measuring self-employment

Data and population

The data analysed in this study came from Swedish registers (Gothenburg Educational Longitudinal Database) and include information about the students’ education history, work history and income, as well as a variety of background variables (gender, parental education, national origin, etc.). The point of departure for this study is the total population of individuals born in 1974–1976 (N = 321,612) residing in Sweden during the year in which they were 16 years old. The homogenous age group reduces influences from work experience and trade conditions. Policies about entrepreneurship education created by European Union started to occur in mid-1990s when a majority of this population was in education and was the target for these policies. Data from two years were analysed: one measure looked at 2006, when these tertiary-educated people were aged 30–32 years, which is shortly after most of them had finished higher education and the influence of their education could be expected to be quite strong; the second measure looked six years later in 2012 when they were aged 36–39 years. In 2006, the labour market was booming, while in 2012 it may still have been feeling the effects of the recession in 2008. From the total population, those who had completed compulsory schooling and had a grade point average were selected (N = 320,135). Moreover, the individuals also needed to have completed at least two years of higher education; that is, both those who had completed their studies with a degree, and those who had not, were included. For obvious reasons, more people have had time to reach the two-years-study limit, when analysed in 2012 (N = 106,554). ‘Completed’ is defined as not studying during the years prior to and between 2006 and 2012. In addition, they needed to be active in the labour market (N = 91,245). The number of non-active individuals differs because of reasons such as studying, or being unemployed, retired or missing. The population available for analysis in 2006 numbered 71,753 tertiary-educated people, and in by 2012 their number had increased to 85,228 people.

Variables

Dependent variables

Income from employment is divided into two: main income from salaried employment or main income from self-employment. This division is used in both the descriptive and multivariate statistics. To give a more detailed description of what underlies this rough division, the population’s occupational status is divided into four categories: employed; hybrids (one category of hybrids has a main occupation which is employment and another has a main occupation which is self-employment); and solely self-employed. (These categories are used in Table 1).

Table 1. Descriptive statistics occupation in 2006 and 2012.

Variables	All HE students
	n	%	n	%
Gender	Men		Women
Occupation 2006
Employed	32,656	84	47,492	90
Hybrid, mostly employed	3.276	9	2.232	4
Hybrid, mostly self-empl.	471	1	318	1
Self-employed	926	2	399	1
No info/unknown	1393	4	2082	4
Occupation 2012
Employed	30,151	78	45,496	87
Hybrid, mostly employed	4502	12	3921	8
Hybrid, mostly self-empl.	642	2	475	1
Self-employed	1983	5	1024	2
No info/unknown	1444	4	1607	3
Unemployed 2005
Unemployed	845	2	1110	2
Employed/other/unknown	37,877	98	51,413	98
Unemployed 2011
Unemployed	156	0	335	1
Employed/other/unknown	38,566	100	52,188	99
Total	38,722	100	52,523	100

Tertiary educated people totalled 91,245 before reducing the population for those undertaking ongoing studies in 2006 and 2012.

Independent variables

The independent variables are gender (men and women) and parents’ education. The latter is divided into two categories: students whose parents had received at least two years of higher academic education, and others, including those parents whose education is unknown. Parents’ self-employment is a dichotomous variable indicating whether either of the student’s parents were self-employed as a main source of income. Field of study is divided into nine broad fields in which the students had completed most of the credit points. They are (1) humanities and theology; (2) law and social sciences; (3) education; (4) natural sciences; (5) technology; (6) agriculture and forestry; (7) medicine, odontology, and veterinary science; (8) health care; and (9) fine arts. There is also an ‘others’ category, that comprises multidisciplinary studies; such as ‘law and technology’, but also comprises courses that seem to have been incorrectly categorised. This ‘other’ category is included in the analysis, but the results are not shown in the tables with the multivariate analyses. Unemployed 2005 and Unemployed 2011 indicates the number of tertiary-educated people who had no statement of earnings and tax deductions the year that preceded the years assessed in this study. The variable degree indicates if the student finished their tertiary studies with or without a degree. Many students may not have written their final essay, they lack some credit points from a course or they simply did not apply for a degree certificate. People studying at a rate comparable to half-time or more in 2005 and 2011 were deleted, since studying is perceived as their main occupation. Grade point average (GPA) from compulsory school is used as an indicator of ability. It is a five-grade norm-referenced scale, normally distributed, where five is the best. Finally, median monthly income is used in the descriptive analysis, but not included in the statistical analysis. It is problematic to estimate a monthly income among the self-employed; they often do not have a regular income and have the opportunity to influence the finances of their business, meaning that some of the turnovers could be used for reinvestment before being estimated for the tax authorities.

Analysis method

Binary logistic regression (SPSS) and Average Marginal Effects (STATA) were used to study and compare the influence on self-employment of the various factors, and how these influence women’s and men’s tendency to become self-employed (Leeper 2018; Mood 2010). ‘Employed’ is the reference category in the dependent variable, to which ‘self-employed’ relate.

Building models for understanding factors that influence human behaviour in educational science is based on theory and previous research. This means the model has not been built exploratively, step-by-step, introducing the variable that best explains the residual until the model fit no longer improves. Tests of significance were not performed. There is no ‘super-population’ against which the results could be generalised (White and Gorard 2017).

Results: self-employment among tertiary educated

Table 1 shows the distribution concerning occupational status among those who had studied at least two years at the tertiary level. As previously known, more women than men pursue post-secondary studies. Women comprise 58 per cent of these cohorts; this corresponds with general Swedish statistics (UKÄ 2018) and to the mean proportion in the OECD countries (OECD 2018). A large majority were employed or mostly employed. It seems as though the most common type of self-employment is to have a business on the side. Also, it is well-known, and now confirmed when looking at this group that more men than women were self-employed.

Several pieces of research have shown that the number of solo entrepreneurs without any employees are increasing (Fritsch, Kritikos, and Rusakova 2012; Holmquist and Sundin 2017; Yuen et al. 2018). It is mainly middle-aged (or older) professionals with previous work experience in paid employment who transfer to self-employment. The economic conditions when the students finish their education are likely to influence their decision whether to become self-employed. Beiler’s (2017) results suggest that ‘recent’ graduates are more likely to start a business shortly after graduation if the trade conditions are good. In this study, the proportion of self-employed in 2006 when the market was booming, was lower than anticipated. A possible explanation is that students from all fields of study were included in the present study which makes this total population less subject to influence by financial fluctuations. An explanation that seems to fit better with Swedish conditions is that professionals transfer from paid employment to self-employment when they have gained some experience. Looking at the continuity in business ownership, 34 per cent (1021 individuals) were self-employed in both 2006 and 2012. Perhaps the reasons for being self-employed vary during different life stages. A very low proportion was registered as unemployed in the year before the analyses. Generally, tertiary educated people are less exposed to unemployment than other groups (OECD 2012, 119; Åberg 2003). To conclude, whether unemployment is a factor that pushes the tertiary educated to start businesses cannot be drawn because of the low numbers unemployed in this study.

Table 2 shows some individual characteristics and how these were related to the prevalence of self-employment. From the top, those tertiary-educated students who originated from an academically educated family were slightly more likely to become self-employed compared to those who originated from a less-schooled family background. To have resources, such as financial resources, education and/or social networks is likely to influence a transfer to self-employment (Tilly 2005). Since self-employment is comparatively uncommon in Sweden, a minority of the tertiary-educated students have self-employed parents with experience to share. In the descriptive statistics, a tendency to have a positive relationship between self-employed parents and self-employed tertiary students can be seen. (These and other associations will be further explored in Table 3). Turning to field of study, slightly more than half of the student population were educated within the fields of social sciences and technology. In numbers, these were also the fields from which most of the self-employed originated.

Table 2. Descriptive statistics of the whole population tertiary educated and self-employed in 2006 and 2012.

Variables	All HE students				Self-employed 06		Self-employed 12		Self-employed 06		Self-employed 12
	n	%	n	%	n	%	n	%	n	%	n	%
Gender	Men		Women				Men				Women
Parents’ education
Parents’ HE	18,341	47	24,693	47	822	59	1448	55	418	58	847	57
Parents’ less schooled	20,381	53	27,830	53	575	41	1177	45	299	42	652	43
Parents’ employment
Parents’ employed	36,042	93	48,590	93	1285	92	2399	91	656	92	1360	91
Parents’ self-employed	2680	7	3933	8	112	8	226	9	61	9	139	9
Field of study
Humanities/Theology	2062	5	4817	10	118	8	154	6	174	24	284	19
Social sci/Law	10,565	27	16,158	31	385	27	766	29	211	29	474	32
Education	3320	9	13,137	25	33	2	106	4	72	10	211	14
Natural science	2504	7	3271	6	80	6	142	5	37	5	82	5
Technology	17,290	45	6041	12	659	47	1236	47	101	14	202	13
Agriculture/Forestry	167	-	315	1	15	1	24	1	5	0	26	2
Medicine/Odontology	1075	3	1523	3	21	2	57	2	19	3	48	3
Health care	1076	3	6438	12	23	2	48	2	53	7	103	7
Fine arts	383	1	414	1	43	3	67	3	31	4	46	3
Other/unknown	206	1	337	1	20	2	25	1	14	2	22	1
Degree
Degree 2004	25,922	67	39,157	75	806	58			470	66
No degree 2004	12,800	33	13,366	25	591	42			247	34
Degree 2010	30,120	78	44,808	85			1874	71			1165	78
No degree 2010	8602	22	13,366	15			751	29			334	22
Mean GPA	3.8	0.5	3.9	0.5	3.9	0.5	3.9	0.5	4.0	0.5	4.0	0.5
Median income 2006*	26,083		18,191		8325				7575
Median income 2012*	36,517		24,908				12,192				10,608
Total	38,722	100	52,523	100	1397	3.6%	2625	6.8%	717	1.4%	1499	2.9%

Notes: Per cent is calculated vertically and gender divided in relation to the total population within each category. *In 2006, 4157 individuals did not have a salaried monthly income, in 2012, 4125 individuals did not have a salaried monthly income; these were excluded when calculating median income and dispersion.

Percentages have been rounded to the nearest integer.

Table 3. Probability in percent to become self-employed the year 2006 and 2012, gender divided analyses.

	Self-empl 2006	Self-empl 2012	Self-empl 2006	Self-empl 2012
Gender	Men		Women
Parents’ education
Parents’ HE	0.6	0.5	0.4	0.8
Parents’ less schooled	ref	ref	ref	ref
Parents’ employment
Parents’ self-employed	0.3	1.7	0.4	0.9
Parents’ employed/other	ref	ref	ref	ref
Field of study
Human/Theology	13.5	8.2	4.2	5.3
Social sci/Law	6.9	6.6	1.2	1.5
Education	ref	ref	ref	ref
Natural science	7.1	4.5	1.0	0.9
Technology	5.9	5.7	1.9	2.0
Agriculture/Forestry	24.0	18.2	too few	8.9
Medicine/Odontology	3.5	2.9	1.3	1.3
Health care	8.8	4.0	0.7	0.1
Fine arts	27.6	25.4	13.4	12.8
Unemployed year before
Unemployed	1.9	0.6	0.6	3.0
Employed/other/unknown	ref	ref	ref	ref
Degree
Degree	ref	ref	ref	ref
No degree	2.5	1.3	0.5	0.5
Continuous var
GPA	1.5	1.7	0.2	0.9

Of the student population, 18 per cent never completed their studies with a degree (in 2010). It seems as those who became self-employed were slightly less likely to have graduated. A common explanation is that these students had started their business during their studies and did not have time to sit their final exams (Buenstorf, Nielsen, and Timmermans 2017).

Both self-employed men and women showed higher school ability in early schooling than their counterparts who were employed. This can be seen as a confirmation of the perception that the self-employed are able and enterprising (From 2010). Finally, income differences between men and women were large among both the employed and the self-employed. On average, self-employment does not seem to contribute to high-income levels (see for example Buenstorf, Nielsen, and Timmermans 2017), particularly not within the arts (Hårsman 2012). Moreover, in 2012, Table 2 shows that self-employed women face a difficult financial situation. Employed women earned around 25,000 SEK/month (∼2,300 EUR), while self-employed women earned around 10,000 SEK/month (∼870 EUR). A common explanation of the low-income level among self-employed women is that they combine self-employment with unpaid home duties (Bourne 2010; Weber and Geneste 2014). In these analyses, those whose main source of income was from self-employment were defined as ‘self-employed’; meaning that even though the income is low, it may still be their main income. Income from self-employment is difficult to estimate; self-employed people do not have as regular an income as employed people have. Another issue is the apparent reluctance of self-employed people to pay taxes (Alalehto and Larsson 2012; Benedek and Lelkes 2011; Åstebro and Chen 2014). A Swedish study shows that as much as 30 per cent of the total income is underreported (Engström and Holmlund 2009). In the descriptive analyses, median income was reported, since the income distribution was skewed. Among the self-employed, a majority had a low income (two per cent earned around 100,000 SEK/month ∼ 8700 EUR). The income variable was not included in the multivariate analyses.

In Table 3, the influence of single factors (characteristics) on self-employment, were estimated when other factors included in the analysis were the same. The importance of the various characteristics or group belongings becomes more easily discernible.

Starting with a comparison of men and women (in a total analysis not shown in Table 3), it appeared that among the self-employed in 2006 the gender difference was (0.22), meaning that men had an increased likelihood of 22 per cent of becoming self-employed compared to women. This was also the case when parents’ education and self-employment, students’ field of study, occurrence of unemployment and graduation were the same. In 2012, that difference had increased, and men were 37 per cent more likely to become self-employed compared to women.

As can be seen from Table 3, the choice of field of study is the most decisive factor for self-employment. From Table 2, students educated within the fields of social sciences and technology were more likely to become self-employed; however, in terms of proportions, another picture emerges. Somewhat surprisingly, the technology field does not stand out as a field leading to new businesses. On the basis of earlier research (van Rooij 2014), it might have been expected that the investments made to promote self-employment in these fields (European Commission 2012b; HSV 2009), and from technological developments accomplished during studies that could be turned into business ideas and self-employment. Maybe it is the large proportion of ‘ordinary’ engineers who preferred to be employed (by for example Volvo, Ericsson) who counteracted these expectations. Similarly, the fields of social sciences and law, including business, did not appear as obvious starting points for self-employment. Many civil servants are educated within these fields and they outnumber those who start businesses as lawyers or accountants (Ahlin, Gabrielsson, and Wennberg 2013). Students from the field of education (reference group) were quite unlikely to become self-employed, as were students from the fields of health care and medicine. Education is a field in which 80 per cent of the students are female; despite this, only double the number of women as men were self-employed in 2012. Previous research has shown that men educated in low-income occupations were more likely than their female counterparts to leave their careers for another that paid better (Acker 1990; Berggren and Lauster 2014; Kauppinen-Toropainen and Lammi 1993). Teaching professions are comparatively low-paid in Sweden. Recipients of education paths leading to professions circumscribed by public authorisation have difficulties competing within the private market. The deregulation of the state monopoly in health care and education has not facilitated self-employment within these sectors. Small businesses in health care are bought up or driven out of competition by large companies (e.g. Hård, Sundin, and Tillmar 2007; Sundin and Tillmar 2010). Moreover, an increasing number of independent schools are run by large, international school companies (Nilsson and Johansson 2011). Most likely to become self-employed were male students educated within agriculture/forestry, followed by those who had studied fine arts. Among women, fine arts stood out as the most likely departure for self-employment followed by agriculture/forestry and humanities/theology. Swedish forests are mostly privately owned (80%) (Skogsstyrelsen 2013) and often handed down through inheritance from generation to generation. The fine arts fields include music, dance, photography, design, crafts, and literary composition. Among students educated within these fields, 14 per cent were self-employed (Table 2, in 2012). They are often hired for time-limited projects (e.g. actors) or, for crafts persons, as self-employed full time or part time in a workshop of their own or with others (Hårsman 2012). The fields of humanities and religion also seem to be a departure for self-employment. Unfortunately, these analyses do not include information on the labour market sector the students establish themselves in.

Concerning degrees, those who became self-employed, were slightly less likely to have graduated, it applies particularly to men. Six years later, the difference in start-up propensity between graduates and non-graduates had declined (among men). A possible explanation of this equalisation is that professionals start a business after getting experience and building their reputation (Fritsch, Kritikos, and Rusakova 2012; Holmquist and Sundin 2017; Yuen et al. 2018).

Men with higher grades from school are more likely to start a business. Technically, the results should be interpreted such as each grade increase on the five-grade scale increased the likelihood of self-employment by 1.5–1.7 times (150–170%) among men, and by 20–90 per cent among women.

As mentioned previously, because few tertiary-educated people were unemployed the year before the analysis year, the results only lend themselves to a starting point for further research.

Discussion

Neither tertiary-educated Swedish men or women were likely to become self-employed; less than two per cent were registered as having their main income from self-employment in 2006, and 4.5 per cent in 2012 (Table 2). The low prevalence of self-employment in Swedish society is likely to be a result of influences from different societal structures acting both separately and together.

The welfare regime in Sweden is universal (Esping-Andersen 1990), meaning that the state ensures a certain living standard for its population including in the case of unemployment and ill health. In this system, incentives for self-employment (Buenstorf, Nielsen, and Timmermans 2017; GEM 2011; Ulmestig 2013) decrease; people do not immediately have to move or become self-employed to avoid poverty.

A higher education degree is designed for employment within large institutions which further explains some of the extremely low prevalence of self-employment. A large proportion of students educated in technological fields (a large majority of men) find employment within Swedish multi-national companies, graduates from social science and law fields are directed towards positions in governmental institutions, and graduates from the fields of health care and education are directed towards positions in the public sector which correspond precisely or well with their degree.

People start businesses in fields in which they are knowledgeable (Hård, Sundin, and Tillmar 2007). Despite the image of Sweden as a gender-equal country, men and women choose different fields of study, in line with men and women in most western European countries (UNESCO 2012, 81). This individual ‘choice’ about education places people in different positions in various sectors of the labour market and influences their likelihood of starting businesses. Hård, Sundin, and Tillmar (2007) suggest that the profiles of the small business owners are even more gender-segregated than the overall labour market. Women were highly unlikely to become self-employed; partly because of the indirect effect, the gendered educational choices, and partly because of the direct effect of gender discrimination.

Starting a business needs resources, capital and a network (Tillväxtverket 2007; Tilly 2005). When resources are lacking, the dearth of self-employment may be a result of structural obstacles rather than unwillingness. In Sweden, the self-employed may face difficulties in getting credit, and they also may have problems buying an apartment or a house because mortgages will not usually be given to people who cannot prove they have a reliable and steady income.

The contribution of this study

To understand the ‘reluctance’ to become self-employed, the individual and the individual characteristics need to be complemented by studies that consider contextual and historical influences (Ahl 2006, 2008; Calás, Smircich, and Bourne 2009; Dopfer, Foster, and Potts 2004; Foster 2017; Holmquist and Sundin 2002).

Individuals can be inspired by entrepreneurship education, but societal organisation and expectations of family and friends set frames for what is or feels possible to choose and do (Dopfer, Foster, and Potts 2004; Foster 2017). There are cultural differences between Europe and North America (Welter and Lasch 2008), but there are also variations within the Nordic countries (Hjorth 2008), thus, international comparisons are needed (Leoni and Falk 2010) to become aware of how various societal organisations limit or facilitate self-employment.

The data in this study were based on the total population of students born in 1974, 1975 and 1976 who pursued higher education studies during the 1990s and early 2000s. This means that the students went through the same educational system and made their choice of higher education and labour market careers during the same period of life. The large data set without selection bias means the results are reliable and that they provide an overview. Such studies are needed to aid comparisons between the employed and self-employed, between men and women, between the various study fields and labour market sectors, to yield insight into which groups of people who actually start businesses and to serve as a base for improving policies.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Swedish Research Council [grant number 2008-4760].

Notes

1 In this article, we use self-employment, owning a business and entrepreneurship as exchangeable concepts.