Measuring Job Satisfaction of European Entrepreneurs with Past Unemployment:

Measuring Job Satisfaction of European Entrepreneurs with Past Unemployment:

Forced or Free, Happy or Not?

Clemens Pegritz Studentnr.: S2529998

23th of June 2014

University of Groningen

First supervisor: Dr. Florian Noseleit

Second supervisor: Dr. John Qi Dong

Word count: 10,840

2

3 Contents

Abstract ... 4

Introduction ... 5

Literature Review ... 8

What does Happiness mean to Economics and Entrepreneurship? ... 8

Job Satisfaction and Self-Employment ... 9

The Stigma of Unemployment ... 11

Necessity and Opportunity Entrepreneurship... 12

Methodology ... 15

Data and Measures ... 15

Control Variables ... 16

Results ... 19

Descriptive Results ... 19

Regression Analysis ... 21

Discussion ... 25

The Benefits of Entrepreneurship ... 25

Past Unemployment and Interaction Effects ... 25

The Differences between Necessity and Opportunity Entrepreneurship ... 27

Implications ... 28

Target Groups and the Dangers of Long-term Unemployment ... 28

Policy Examples and Possible Solution ... 29

Limitations and Future Research ... 30

Limitations ... 30

Future Research ... 30

Conclusion ... 32

References... 33

Appendix ... 36

4 Abstract

Combining data from 26 countries of the European Social Survey (ESS), this study analyzes job satisfaction in the context of employment relation and the influence of past unemployment. The study uses self-employment as a proxy for entrepreneurship and compares the job satisfaction of entrepreneurs to employed people. Adding past unemployment of at least one year to both kinds of employment relation leads to four distinctive groups. The regression analysis of job satisfaction, conducted in two models, tries to answer three main questions: Do entrepreneurs enjoy higher job satisfaction than employees, like established literature suggests? Do entrepreneurs still receive the proclaimed benefits of being “their own boss”, although past unemployment is expected to reduce their job satisfaction? Finally, are entrepreneurs, which have encountered past unemployment, less happy with their jobs than those who did not? The analysis of the last question reveals similarities to the literature of necessity entrepreneurship, but the study tries to shed light on all four introduced groups in terms of job satisfaction. The results indicate entrepreneurs receive higher satisfaction from their jobs than employees. Adding past unemployment to both groups does not remove the benefits of entrepreneurship. It most profoundly affects employees, because employees that faced past unemployment of one year or more are the group with lowest job satisfaction. The differences in job satisfaction between the groups of entrepreneurs with and without past unemployment seem only to be related to an income effect. The findings can be relevant for policy implications.

Keywords: Job satisfaction; Entrepreneurship; Self-employment; Past Unemployment

5 Introduction

The goal of this study is to analyze the question, how past unemployment influences the job satisfaction of entrepreneurs using employees as a comparison group. There are three main reasons why this is a compelling topic to study.

First, entrepreneurship has become an important topic for economic and innovation research in recent years. Different studies show that it might have a positive impact on economic growth (Audretsch and Keilbach 2004, Van Stel et al. 2005, Audretsch 2009, Hartog et al. 2010), job creation (Hartog et al. 2010), technology diffusion (Schmitz 1989) and market competitiveness between firms (Audretsch 2009). The interdependency of entrepreneurs and the economic system can also be found in their indirect relationship with innovation efforts (Audretsch 2008). Already Schumpeter expected entrepreneurs to find new solutions and become innovators to gain a competitive advantage over other market participants (Schumpeter 1934).

Because of their ability to create innovations, studying and understanding entrepreneurial characteristics can therefore also contribute to the field of innovation management.

Second, self-reported job satisfaction is a useful measurement for individual utility gained from work, which is why integrating or applying it can be beneficial for economics and business administration. Especially in early stages of self-employment, entrepreneurs are often not willing or able to report significant information about their salary for surveys, which favors the approach of looking at job satisfaction. It is captured by most of the established socio- economic panels via survey questions, which also cover determinants and influential factors contributing to it. One literature stream uses this information to investigate, if any benefits stem from being in a certain employment relation.

Benz and Frey (2008b), for example, show that self-employed individuals are

significantly more satisfied with their work than employees of a company or organization, which

holds for several countries around the world (Benz and Frey 2008b). They identify self-

determination, which for instance means having an interesting job and being independent at

work, as the most influential factor leading to the increased job satisfaction compared to

employees (Benz and Frey 2008b). In 2004 Frey et al. introduced the concept of procedural

utility in order to explain these non-monetary effects of self-employment (Frey et al. 2004). It

proclaims that people do not only value outcomes, like salary or working hours, but also the

circumstances contributing to them (Frey et al. 2004). In this sense the procedural utility

6

approach differs from traditional economic theories, because there utility only stems from instrumental values, like higher income (Benz and Frey 2008b). The economic concept that comes closest to it is the theory of compensating wage differential, which is an established theory in the field of labor economics (Benz and Frey 2008a). It explains the matching of workers and companies in the way that unpleasant job characteristics have to be compensated through higher wages (Rosen 1974, Smith 1979). Therefore also non-monetary effects are incorporated. They result from positive working conditions, like having an interesting job or autonomy, but can also be negative, like dangerous working conditions (Rosen 1974, Smith 1979). The difference to procedural utility is that the mentioned circumstances are not used to explain work-related satisfaction, but to calculate a fitting wage and hence again only express utility in a monetary fashion.

Building upon the idea to investigate non-pecuniary effects on job satisfaction, Block and Koellinger (2009) compared the start-ups of nascent entrepreneurs and found that necessity entrepreneurs are less happy with their start-ups than opportunity entrepreneurs (Block and Koellinger 2009). They are also accounting procedural utility for the difference in job satisfaction, because the group of nascent necessity entrepreneurs encounters push factors to start a new business and therefore is unable to make an optimal decision for themselves (Block and Koellinger 2009). These push factors can include current unemployment or missing alternatives at the labor market and create the conditions that lead to the negative utility effects on job satisfaction (Block and Koellinger 2009). On the downside, their sample originates from a survey of nascent entrepreneurs, where some respondents might not have started their business yet. Furthermore it does not include employees, making a comparison as carried out in the focal paper of Benz and Frey (2008b) impossible.

As discussed above, previous literature has shown that individual utility can be a useful

instrument for researchers to either explore the influence of necessity and opportunity

entrepreneurship or different employment relations on job satisfaction. This study tries to

synthesize both approaches using the large amount of observations of the European Social

Survey (ESS). To do so, it focuses on past unemployment instead of necessity and opportunity

entrepreneurs. The trait of past unemployment of at least one year has the advantage that it can

also be found with employees, which enables the comparison to the study of Benz and Frey

(2008b). If it is applied to employees and self-employed people, it enables to differentiate

7

between four distinctive groups. The main focus of this study lies on the group of entrepreneurs that were unemployment for one year or more. These entrepreneurs are used as a proxy for necessity entrepreneurship, which enables a comparison to the literature about necessity entrepreneurship, like for example the study of Block and Koellinger (2009). The idea behind the comparison is that longer periods of unemployment will lead to increased pressure to find a job, which matches the definition of push factors. Such factors make an optimal employment decision for the mentioned group more difficult. Especially, if no other options are available, people are pushed towards becoming self-employed, which is why they are called necessity entrepreneurs.

As self-employed people only account for a fraction of the working population, those who reported past long-term unemployment are even fewer. In contemplation of an analysis that considers the employment situation and past unemployment of respondents, the observations of 26 countries of the ESS 2010 data are combined to create a fitting sample. In summary, the influence of past unemployment of one year or more on entrepreneurs constitutes the center of this study, but including employees in the analysis enables the possibility of having a comparison group to them.

Third, the implications of this study might be applicable for policy makers. The aim is to provide results to improve policies in such a way that they will enhance entrepreneurship capital (Audretsch and Keilbach 2004). Different studies have shown that a large amount of latent entrepreneurs exists in developed countries (Blanchflower and Oswald 1998, Blanchflower et al.

2001). One reason why not more people engage in entrepreneurship is that they have to overcome financial entry barriers (Benz and Frey 2008b). Governments can try to identify those people and support them with subsidies or lump sum payments to overcome entry barriers and start an own business (Meager 1996). Usually policy makers intend to support people that are not likely to become entrepreneurs, because otherwise the actions might not be as effective, which is called dead-weight (Meager 1996). Focusing on groups that suffered from past unemployment, the results of this study will give information about individuals that during this time span were not able to escape unemployment by becoming self-employed or just did not especially value it earlier. Therefore, self-selection, like it is the case when people leave their job in order to pursue an opportunity as an entrepreneur, is less feasible.

The results indicate that individuals are better off, if they chose self-employment over an

employed job, having a background of past unemployment. Employees with this trait are the

8

worst off group in terms of job satisfaction and therefore one possible target group for policies.

Although not explicitly the focus of this study, this advice might also be applicable for currently unemployed job seeker.

Literature Review

What does Happiness mean to Economics and Entrepreneurship?

Happiness research can be a valuable addition to economic and entrepreneurship research. There is the argument that economics should take individual happiness into account, because this kind of perspective is missing in economic standard theory (Benz and Frey 2008b).

A more general approach can be used to connect theory and real world scenarios, which is especially important for economists that provide advice to decision makers (Frey and Stutzer 2002, Block and Koellinger 2009). Introducing happiness research in economic policy improves the prospects of cost-benefits analysis of the effect of government expenditures (Frey and Stutzer 2002). Determinates of happiness can also be examined on a macro level. There are studies investigating the relationship between happiness and GDP, inflation and the unemployment rate (Frey and Stutzer 2002). For this study only micro level factors on happiness are relevant, because it focuses on one single year. Therefore external economic factors are less important as when using data including a time dimension. Finally, the field of entrepreneurship research can profit from learning about the inclination of individuals towards starting a business or about those who are already running one, which again might help to improve policy implications (Block and Koellinger 2009).

The scientific term for happiness is subjective well-being, which means that individuals

are directly asked about their overall life satisfaction (Benz and Frey 2004). Capturing well-

being directly has the advantages that respondents also take circumstances like past experience,

expectations and procedural utility into account (Frey and Stutzer 2002). In this way it differs

from decision utility and the rational choices of the homo economicus, which for example cannot

fully explain entrepreneurial behavior (Block and Koellinger 2009). On the downside, measuring

happiness is susceptible to biases, which makes it impossible to approximate an absolute value of

happiness (Frey and Stutzer 2002). Such biases occur because of the formulation and order of

survey questions or the condition of the participant. In most cases the aim of happiness research

9

is to study the composition of happiness, which is possible because the errors of surveyed people are usually not systematic (Frey and Stutzer 2002). This leaves subjective well-being as a satisfying proxy for individual happiness.

From the perspective of happiness research, it is important to consider the difference between job and life satisfaction. The latter is a difficult subject to investigate, because many factors influence a person’s life. For instance, self-employed may enjoy higher job satisfaction because of higher autonomy and job variety, but income variation, long working hours and other common problems within self-employment can reduce life satisfaction under certain circumstances, which makes it often hard to establish causalities (Blanchflower et al. 2001, Benz and Frey 2008b). As this study investigates only work related aspects, like for instance employment relations and past unemployment, the scope can be reduced to job satisfaction. It follows the literature stream about individual job satisfaction, where, identical to the explained case of happiness, a survey question is used to measure self-reported job satisfaction (Blanchflower 2000, Benz and Frey 2004, 2008b, 2008a). Influential factors can be extrinsic, like salary, working hours and job security, or intrinsic satisfaction levels, which are related to the perception of work and procedural utility (Rose 2003, Frey et al. 2004). They include autonomy, the quality and nature of the job and social relations (Rose 2003). Also the employment status of a person influences job satisfaction in a strong manner (Benz and Frey 2008). Unemployment makes an exception, because there is a relationship between it and overall happiness. Researchers found that unemployed people experience high non-monetary costs and a significant decrease in life satisfaction (Clark and Oswald 1994, Winkelmann and Winkelmann 1998, Clark et al. 2001). This effect even occurs, if they receive the same income as prior to their unemployment (Frey and Stutzer 2002). Other important factors are of socio-demographical nature, like country, age, gender, education and relationship status. In order to explain determinants of job satisfaction, it is important to control for these non-work-related aspects and therefore they will be discussed in the methodology section and used as control variables in the analysis.

Job Satisfaction and Self-Employment

Self-employment is the most elementary form of entrepreneurial activity and is often

used as a proxy for entrepreneurship (Blanchflower et al. 2001, Block and Koellinger 2009,

10

Block and Wagner 2010). Studies show that a large amount of latent entrepreneurs exist in developed countries and that the main problem, faced by potential entrepreneurs, is to raise the necessary capital to start a business (Blanchflower and Oswald 1998, Blanchflower et al. 2001).

To answer why these people would prefer to become self-employed, a large body of literature looks into the relationship between job satisfaction and self-employment.

Benz and Frey conducted three studies in which they compared the job satisfaction of employees and self-employed people and consistently found evidence that the latter group is more satisfied with their jobs (Benz and Frey 2004, 2008a, 2008b). Previous studies from different European countries support these findings (Blanchflower and Oswald 1998, Blanchflower 2000). In their focal paper “The value of doing what you like: Evidence from the self-employed in 23 countries”, Benz and Frey (2008b) found that the relationship holds in 23 countries in Europe, North America, Eastern Europe and non-western countries, whereupon only a few non-western countries showed exceptions (Benz and Frey 2008b). The identified main reasons for the higher satisfaction levels of the self-employed are higher autonomy and having an interesting job (Benz and Frey 2004, 2008a, 2008b). There is a strong correlation between these two attributes, which are combined in the term of “self-determination” and together account for 80 percent of the job satisfaction differentials in western countries (Benz and Frey 2008b).

Interestingly, extrinsic work characteristics, like salary, job security and the possibility for advancement only explain a very low percentage of the differentials (Benz and Frey 2008b). A seemingly matching result was found in a study by Hamilton (2000), which concluded that people except lower wages in order to become self-employed (Hamilton 2000). Because of the mentioned findings, Frey et al. (2004) introduced the concept of procedural utility, which stems from three main ideas: First, utility is understood as well-being. Second, procedural utility also includes non-instrumental determinants, like pleasures and displeasures of processes. Finally, the concept acknowledges that humans have a sense of self, which means that people care about how they are seen by others, and how they perceive themselves.

The concept matches with the findings of Benz and Frey (2008b), as the authors think that people seem to not only value the outcomes of their job, but also the process and conditions leading to the outcomes (Frey et al., 2004, Benz and Frey 2008b, Block and Koellinger 2009).

The results point to the idea that individuals receive value from having autonomy or the feeling

of “being your own boss”, which can almost be seen as a consumption good in economic terms

11

(Benz and Frey 2008b). These arguments and findings suggest that self-employed people have higher job satisfaction than employees, which leads to the formulation of the first hypothesis:

Hypothesis 1: Self-employed respondents have higher self-reported job satisfaction than employees.

The Stigma of Unemployment

Employment is thought to be one of the most influential factors considering well-being and happiness research strongly suggests that losing it is an unfortunate event that severely depresses people (Clark and Oswald 1994, Frey and Stutzer 2002). An interesting paradox about work is that many people tend to see their job as a burden, but empirical evidence indicates that having none significantly decreases well-being, even if there is no change in income (Frey and Stutzer 2002). This is supported by many studies, which find that unemployed individuals tend to be significantly less happy than employees (Winkelmann and Winkelmann 1998, Clark et al.

2001).

Examining mental distress, Clark and Oswald (1994) found in their research that unemployed receive an around twice as high mental distress score than employees. The results rise with the level of education, and when only people in working age are considered, the scores are highest for the class of people between 30 and 49, followed by the group above 50 (Clark and Oswald 1994). Even if controlled for specific fixed effects, the non-pecuniary effects of being unemployed on satisfaction seem to be much higher than the one associated with income loss (Winkelmann and Winkelmann 1998). Although this negative relationship is quite established, the causality could be interpreted in the way that unhappy people might perform worse than others and therefore get laid off by their employer (Frey and Stutzer 2002). Winkelmann and Winkelmann (1998) were able to refute this argumentation, because they found strong evidence that the causality runs from unemployment to unhappiness and not the other way round (Winkelmann and Winkelmann 1998), which can be attributed to psychological and social factors (Feather 1990, Frey and Stutzer 2002).

Returning to the comparison of self-employed people and employees in terms of job satisfaction, this study tries to investigate the relevance and influence of past unemployment.

High levels of past unemployment, especially if the duration exceeds one year, have a strong

12

negative effect on life-satisfaction (Clark et al. 2001) and many studies suggest that it is important for policy makers to recognize the dangers of long-term unemployment, as it might have detrimental effects on the future prospect of employment (Winkelmann and Winkelmann 1998, Clark et al. 2001, Frey and Stutzer 2002). Accordingly, the benefits of being self-employed might not be valid for the self-employed people that were confronted with past long-term unemployment. For instance, Darity and Goldsmith (1996) expect that the adverse effects of being unemployed and suffering from low levels of psychological wellbeing will lead to lower productivity and changes in search strategies, which may include the preference of occupational choice (Dartiy and Goldsmith 1996). Therefore the second hypothesis is derived from introducing the characteristic of past unemployment to the relationship of the first hypothesis:

Hypothesis 2: Past unemployment of at least one year offsets the relationship that self- employed respondents have higher self-reported job satisfaction than employees.

Necessity and Opportunity Entrepreneurship

A fundamental decision for job seekers is whether to become independent contractors on

the market or rather work as employees at a firm or institution (Benz and Frey 2008b). It seems

as if the circumstances and the background of an individual have a big influence on the decision

of becoming an entrepreneur and whether a new business is successful or not. To include more of

this information the Global Entrepreneurship Monitor (GEM) started dividing between

opportunity and necessity entrepreneurs in 2001, followed by recent literature about this topic

(Bergmann and Sternberg 2007, Block and Sandner 2009, Block and Wagner 2010). The

division of both groups relates to whether “push” or “pull” factors were predominant in the

decision-making process to become an entrepreneur. For example, the group of necessity

entrepreneurs is often characterized by past or ongoing unemployment and missing opportunities

on the job market (Block and Koellinger 2009). These circumstances function as push factors,

because they increase the pressure on an individual to find and accept a job. The absence of other

options for work may finally lead to entering entrepreneurship. In contrary, opportunity

entrepreneurs are defined as people that engage in entrepreneurship, because they want to pursue

a business opportunity (Block and Sandner 2009, Block and Wagner 2010). They are pulled into

13

starting a new business, because they expect the opportunity to be profitable compared to their previous employed situation. The two classifications are used to further explore socioeconomic characteristics and determinants of success of difficult starting conditions (Block and Wagner 2010). For instance, researchers try to find answers for whether becoming a necessity entrepreneur is desirable from an economic perspective, because opportunity entrepreneurs tend to remain longer in self-employment (Block and Sandner 2009) and are able to exploit more profitable opportunities (Block and Wagner 2010).

For this study, the relevant question is whether necessity entrepreneurs have lower job satisfaction than opportunity entrepreneurs or whether no significant difference exists between both groups. For instance, Block and Koellinger (2009) use a similar approach to the focal study of Benz and Frey (2008b) about job satisfaction of employees and self-employed people, but they focus on nascent necessity and opportunity entrepreneurs. Their findings show that people are significantly less satisfied with their start-ups, if they were unemployed before and lacked other job opportunities (Block and Koellinger 2009). Furthermore, they illustrate that influential factors on start-up satisfaction are high levels of independence and creativity, but the financial success of the start-ups is accountable for the strongest positive effect (Block and Koellinger 2009).

The authors give two possible explanations, why both groups differ in job satisfaction levels (Block and Koellinger 2009): The first reason is related to the aspirations of entrepreneurs, who were unemployed for a period longer than 12 month. This group probably does not have the individual preference of becoming self-employed and their reduced utility might stem from expected disadvantages, like unsecure pay and high working hours. The second reason lies in the decision process itself. As already mentioned it is possible that push factors are limiting the options and might lead to the reduced utility for necessity entrepreneurs. Finally, Block and Sandner (2009) look at reasons that might separate both groups from the perspective of opportunity entrepreneurs. They discuss whether following a business opportunity causes entrepreneurs to especially value entrepreneurship and whether this self-selection includes higher human capital endowments to pursue opportunities.

As the study also tests for interaction effects between the variables of unemployment and

self-employment and the sample includes employees, who can be affected by past unemployment

too, the predictive margins of the job satisfaction will be calculated to oppose the different

14

groups. Here, past unemployment of at least one year will be used to divide entrepreneurs into two distinct groups. As already mentioned, long periods of unemployment create a push factor, which is why the group of self-employed people that encountered them, can be compared to necessity entrepreneurs. Based on the discussed findings and the similarity of past long-term unemployment and necessity entrepreneurship push factors, the third hypothesis is derived as:

Hypothesis 3a: Self-employed respondents, which encountered past unemployment of at least one year, have lower self-reported job satisfaction than self-employed respondents that did not.

However, there are also reasons to expect contradicting findings to the study of Block and Koellinger (2009). First, the distinction between necessity and opportunity entrepreneurs itself might be critically discussed. One argument against the term of necessity entrepreneurs is that starting a new business cannot be forced and is a personal decision, particularly in countries where a basic living is guaranteed by the welfare system (Kautonen and Palmroos 2010).

Second, other studies found less evidence that differences between necessity and opportunity entrepreneurs proceed after they started their own business (Block and Sandner 2009, Block and Wagner 2010, Kautonen and Palmroos 2010).

For instance, it is hard to prove that higher human capital endowment of opportunity entrepreneurs leads to higher job satisfaction for this group. In fact it is quite difficult to measure what excels entrepreneurs, because they have to fulfill different tasks and therefore require a broad skillset (Lazear 2004, Block and Sandner 2009). Researchers in the field of opportunity and necessity entrepreneurs try to solve this problem by focusing on the years of education in the profession used for the start-up or business (Block and Sandner 2009, Block and Wagner 2010).

Controlling for this variable the effect of higher capital endowment for opportunity entrepreneurs is not significant anymore (Block and Sandner 2009). In addition, the education in the conducted profession has a stronger influence on the discovery and exploitation of opportunities for necessity than for opportunity entrepreneurs (Block and Wagner 2010).

It appears that education does not necessarily set the two groups apart, which leaves the

underestimation of income as the strongest argument against the idea that push factors divide

necessity and opportunity entrepreneurs in terms of job satisfaction. It is likely that necessity

15

entrepreneurs for the main part want to earn a livelihood, which puts high emphasis on the variable of income (Kautonen and Palmroos 2010). Block and Koellinger (2009) find that income has the biggest impact on start-up satisfactions, but it seems surprising that although they are controlling for it, the results still show that necessity entrepreneurs gain significantly less utility from their start-ups than opportunity entrepreneurs. As the study relies on nascent entrepreneurs that may still have the start of their business ahead of them, it is possible that the respondents underestimate the importance of income. Kautonen and Palmroos (2010), for example, do not find strong support for differences in job satisfaction of necessity and opportunity entrepreneurs. In their study they define start-up satisfaction as the preference to proceed in self-employment over the possibility to switch back to paid employment and show that the negative effect on satisfaction for necessity entrepreneurs is relatively small and not long-lasting (Kautonen and Palmroos 2010). They expect that a satisfactory level of income and its regularity would offset the differences between opportunity and necessity entrepreneurs in terms of start-up satisfaction (Kautonen and Palmroos 2010). The study will also use a model including compensating control variables for household income and working hours. As differences in job satisfaction might be due to higher income of self-employed respondents that did not face past unemployment, a contradicting hypothesis is formulated as follows:

Hypothesis 3b: Self-employed respondents, which encountered past unemployment of at least one year, do not have lower self-reported job satisfaction than self-employed respondents that did not.

Methodology Data and Measures

The data for the empirical analysis stems from the 5

Round of the European Social

Survey (ESS). The ESS is a cross-national socio-economic survey for academic purposes. Its

advantage, compared to well-established panels like the German Socio-Economic Panel or the

British Household Panel Survey, lies in the multitude of European countries participating. The

2010 survey includes a module about work, family and well-being, which is especially useful for

the research questions at hand. All participating countries asked the same questions to enable

16

consistency across countries. The ESS 2010 survey consists of 46,149 observations from 27 European countries. The sample for the analysis consists of all individuals that are either employed or self-employed and responded to the questions to construct the control variables.

Latvia had to be excluded, because there were no observations of self-employed individuals that reported past unemployment of at least one year. This leaves a sample constructed from data of 26 European countries, which are specified in Table 5 in the Appendix. It consists of 22,686 observations, whereof 19,465 or 85.80 percent are employees and 3,221 or 14.20% self- employed.

Self-reported job satisfaction is used as the main explanatory variable in the empirical analysis. Similar to established literature about job satisfaction, the dependent variable is assessed with the question: “How satisfied are you in your main job?” (G53) (Blanchflower 2000, Benz and Frey 2004, 2008a, 2008b). Individuals are asked to answer the question using a scale from 0 to 10, with 0 being extremely dissatisfied and 10 extremely satisfied. In this study self-employment will be used as a proxy for entrepreneurial activity, which is commonly found in the literature about job satisfaction (Blanchflower et al. 2001, Block and Koellinger 2009, Block and Wagner 2010) and the research area of entrepreneurship and balanced skills (Bublitz and Noseleit 2013). In its simplest configuration it describes a person that has created a job for him/herself, but it also includes people that start their own venture. The self-employed individuals and also employees will be targeted by utilizing the question, “In your main job are/were you...” (F21). The definition of self-employment is quite similar and consistent across countries (Blanchflower et al. 2001), which is an advantage for the approach of this paper that works with countries form different European regions. This dummy variable is the first independent variable of interest. Finally, individuals that suffered an unemployment spell of at least one year were identified in the questionnaire as well. The question used in the survey is

“Any period of unemployment and work seeking lasted 12 months or more?” (F37), which again leads to a dummy variable that is the second independent variable for the regressions.

Control Variables

For the regression analysis, socio-demographic characteristics can explain a part of the

variation of job satisfaction. Therefore control variables will be applied, which are also derived

from survey questions. For Model I, these include country, age, age squared, years of education,

17

gender and partnership status. A second regression Model II additionally includes compensating control variables, which for this study are the average number of hours worked per week including overtime and household income. In the following subsections the control variable will be briefly discussed:

Age

With increasing age, entrepreneurship gets less desirable for the group of latent entrepreneurs, but at the same time the probability of becoming self-employed increases with age, which is most likely due to more experience, self-confidence and available capital (Blanchflower and Oswald 1998, Bergmann and Sternberg 2007). On the other hand, the difficulties of uniting work and family life and considerations about the remainder of the working career and retirement decrease the probability of becoming self-employment in later years, which leads to an inverted U-shaped relationship between age and propensity of entrepreneurs (Bates 1995, Bergmann and Sternberg 2007). Therefore the control variables also incorporate a quadratic term, age squared.

Education

The reasons to become an entrepreneur are versatile and widely debated (Bergmann and Sternberg 2007). Entrepreneurs are facing a great variety of tasks and activities, which require certain capabilities (Lazear 2004, Bergmann and Sternberg 2007, Block and Sandner 2009).

Therefore educations seems to be an important control variable. The years of education are related to the entrepreneur’s human capital endowment, which in turn can influence the decision to become an entrepreneur and also job satisfaction (Block and Wagner 2010)

Gender and Partnership

Gender is used as a control variable, because there are gender-specific differences in

employment relations, which also applies for entrepreneurship (Carter 1997). In industrialized

countries the propensity of male entrepreneurs is higher than of female entrepreneurs (Reynolds

et al. 2004, Xavier et al. 2013). Because of the focus on European countries, this overweight is

18

also expected to be found in this study. Furthermore a possible family life makes it reasonable to include a partnership variable, which shows if the individual lives in a partnership or marriage.

Country

The dummy variable, which is representing the observations of the participating countries, permits to compare them to the first country in the regression. It shows the differences between countries and if working in a particular country has a significant effect, when it comes to job satisfaction. The group of self-employed with prior unemployment spell of at least one year is difficult to target, because both characteristics reduce the number of observations. In order to counteract this problem, it was necessary to combine the data of the all available countries for the sample. As the ESS always surveys around the same amount of people within each country, they do not represent the correct sizes of populations. As no population weights were applied, the sample represents European employees and self-employed, but not the European working force.

Working Hours and Household Income

Working hours and household income are the compensating or extrinsic control variables in this study. Benz and Frey (2008b) found that monetary benefits only account for a small amount of satisfaction differentials between employees and self-employed people. On the other hand, for necessity entrepreneurs wage was identified as the single most important contributor to start-up satisfaction (Block and Koellinger 2009). If income has a strong relation to job satisfaction, it will take out a lot of the variation of the explaining variables of the model.

Therefore it will be applied in a separate Model II. Working hours represent the counterpart to

income in its effect on job satisfaction. In order to improve the proxy of self-employment for

entrepreneurship other studies have excluded part-time self-employed form their study (Block

and Koellinger 2009, Bublitz and Noseleit 2013). One argument to do so is that in their case, the

causality between occupation and job satisfaction is less clear (Block, Koelligner 2009). Apart

from using the income and working hours as control variables in the regression, this study uses a

t-test to compare the means of different groups in the descriptive statistics. If part-time self-

19

employment is stronger in one of the compared groups, it should change the composition and size of the mean, which would be visible through a significant t-test.

Results

Descriptive Results

The two independent variables separate four distinctive groups: Employees

, Employees

, Entrepreneurs

and Entrepreneurs

. Identically with the codification of the data, UE1 (1 = yes) means possessing the trait of past unemployment of at least one year and UE0 (0 = no) means not possessing it. The focus of the study lies on the group of Entrepreneurs

and therefore the descriptive statistics are used to compare the means of different variables to the other groups. Tables 1 – 3 show the comparisons in the order of the size of the mean job satisfaction, starting with the highest of group Entrepreneurs

.

Table 1

Note: The stars show whether the differences of the means ∆ (Mean) are significant. They are based on t-tests for the equality of means and χ

-tests for the quality of proportions. A p-value has to be less than *** p<0.01, ** p<0.05, * p<0.1 in order to rejected H

.

Table 1 illustrates the two groups of entrepreneurs. As already expected after discussing gender in the control variables section, both, Entreprenerus

and Entrepreneurs

, show a higher percentage of male entrepreneurs (66% vs. 60%). The proportion of men is significantly higher in the group of Entrepreneurs

(p<0.1). The descriptive statistics show significant

Entrepreneurs

Entrepreneurs

Variable Mean Std. Dev. Mean Std. Dev. ∆ (Mean)

Job statisfaction 7.69 1.95 7.22 2.3 0.47

Age 46.22 11.81 43.85 10.61 2.37

Education (years) 13.45 4 13.43 4.25 0.02

Working hours (incl. overtime) 47.39 16.75 43.09 17.66 4.3

Gender (dummy, 1 = male) 0.66 0.47 0.6 0.49 0.06

Partner (dummy, 1 = yes) 0.74 0.44 0.66 0.48 0.08

Household income 6.45 2.72 5.44 2.73 1.01

Children (dummy, 1 = yes) 0.53 0.5 0.54 0.5 -0.01

20

differences between both groups of entrepreneurs in terms of job satisfaction, as Entrepreneurs

have a higher mean of job satisfaction (7.69 vs. 7.22). On the other hand this could be explained by their significantly higher household income (6.45 vs. 5.44). Furthermore Entrepreneurs

also significantly excel Entrepreneurs

in mean working hours (47.39 vs.

43.85) and age (46.2 vs. 43.9). Interestingly, it is not possible to reject the H

in case of the difference of educational years, which means that Entrepreneurs

do not significantly differ in the variable of education from the comparison group without past unemployment of one year or more (13.45 and 13.43). Respondents of both groups are very likely to be in a relationship (73.7% and 65.4%) and the probability of having children lies at around 50% (53% vs. 54%), which is similar and not significant in all four groups.

Table 2

Note: As described above (Significance: *** p<0.01, ** p<0.05, * p<0.1).

Table 2 compares the group of employees without the trait of past long-term unemployment of with entrepreneurs possessing it (Employees

vs. Entrepreneurs

).

Probably the most important information from Table 2 is that two groups are on a similar level of mean job satisfaction (7.31 vs 7.22), with Employees

on the higher end, which will be further investigated by the regression analysis and looking at the predictive margins. Employees

have a significantly higher mean household income (6.3 vs. 5.44) and a lower mean working time (40.35 vs. 43.09) than Entrepreneurs

. The same applies to the age of both groups (41.77 vs.

1

If not explicitly mentioned differently, significant always means on a p<0.01 level.

Employees

Entrepreneurs

Variable Mean Std. Dev. Mean Std. Dev. ∆ (Mean)

Job statisfaction 7.31 1.91 7.22 2.3 0.09

Age 41.77 11.85 43.85 10.61 -2.08

Education (years) 13.77 3.57 13.43 4.25 0.34

Working hours (incl. overtime) 40.35 11.24 43.09 17.66 -2.74

Gender (dummy, 1 = male) 0.5 0.5 0.6 0.49 -0.1

Partner (dummy, 1 = yes) 0.67 0.47 0.66 0.48 0.01

Household income 6.3 2.51 5.44 2.73 0.86

Children (dummy, 1 = yes) 0.51 0.5 0.54 0.5 -0.03

21

43.85). Like in the comparison from Table 1, no significant differences in the means of education are found (13.77 vs. 13.43).

Table 3

Note: As described above (Significance: *** p<0.01, ** p<0.05, * p<0.1).

Table 3 matches the weakest group in terms of job satisfaction, Employees

, against the focal group of Entrepreneurs

. The differences in means of job satisfactions are the seconded highest of all three tables (6.77 vs. 7.22). Other significantly lower means are found in the variables of education (12.69 vs. 13.43) and working hours (38.2 vs. 43.09). Although higher for Entrepreneurs

, the difference of mean household income is not significant and there are also no strong differences in mean age of both groups. Employees

is the only group that possesses a higher proportion of female members with women accounting for 56% of the group.

Regression Analysis

The regression analysis illustrates the effects that the independent variables self- employment and unemployment of at least one year and the control variables have on the dependent variable job satisfaction. As both independent variables are a trait that you either possess or not, they are added as dummy variables, “Past unemployment” (0 = no, 1 = yes) and

“Self-employed” (0 = Employee, 1 = Self-employed) to the regression. The analysis also calculates whether there is an interaction effect between these two variables on job satisfaction.

Similar to established literature, one model (Model I) uses basic control variables for age, age Employees

Entrepreneurs

Variable Mean Std. Dev. Mean Std. Dev. ∆ (Mean)

Job statisfaction 6.77 2.33 7.22 2.3 -0.45

Age 43.2 10.62 43.85 10.61 -0.65

Education (years) 12.69 3.63 13.43 4.25 -0.74

Working hours (incl. overtime) 38.2 11.63 43.09 17.66 -4.89

Gender (dummy, 1 = male) 0.44 0.5 0.6 0.49 -0.16

Partner (dummy, 1 = yes) 0.62 0.48 0.66 0.48 -0.04

Household income 5.27 2.66 5.44 2.73 -0.17

Children (dummy, 1 = yes) 0.55 0.5 0.54 0.5 0.01

22

squared, years of education, gender, country and partner, whereas the other model (Model II) also includes compensating control variables for household income and weekly hours worked including overtime (Benz and Frey 2008b, Block and Koellinger 2009). The reasoning behind using two models is that compensating variables might be endogenous in the case of self- employment and therefore Model I will not be affected by an endogeneity bias (Benz and Frey 2008b). Model II has the advantage that controlling for household income and working hours eliminates differences in the material situation that could influence job satisfaction (Benz and Frey 2008b).

Table 4

Note: Design weights applied; standard errors are clustered across countries. (Significance: ***

p<0.01, ** p<0.05, * p<0.1)

Table 4 shows the results of the two regressions. The findings show that the variable past unemployment is significantly negative in Model I and II (-0.460 and -0.345) and self- employment has significant positive coefficients in both models (0.334 and 0.418). It is important to notice that because of the interactions terms, it is not possible to compare the single

Model I Model II

Variables Coef. Std. Dev. Coef. Std. Dev.

Past unemployment -0.460*** (-0.0835) -0.345*** (-0.0891)

Self-employed 0.334*** (-0.0984) 0.418*** (-0.0838)

Interaction effect between

Unemployement & Self-employed 0.123 (-0.158) 0.294** (-0.127)

Age -0.0128 (-0.0117) -0.0111 (-0.0126)

Age squared 0.000232* (-0.000136) 0.000237 (-0.000142)

Education 0.0349*** (-0.00865) 0.00504 (-0.00781)

Working hours -0.00352* (-0.00174)

Household income 0.108*** (-0.00977)

Gender -0.0406 (-0.0387) -0.0401 (-0.0417)

Partner 0.123*** (-0.0388) -0.0348 (-0.0457)

Country (see Appendix)

Constant 7.364*** (-0.258) 7.197*** (-0.298)

Observations 22,686 17,452

R-squared 0.07 0.082

23

coefficients of past unemployment and self-employment independently. Therefore it is more interesting to focus on the interaction effect of the two variables. Model I illustrates a positive interaction effect between past unemployment and self-employment (0.123), but it is not statistically significant. If the standard error is subtracted from the interaction term, it would slightly turn negative (0.123 – 0.158 = -0.035). An important difference is found in Model II, which also displays a positive interactions effect, but it is higher (0.294) and significant on a p<0.05 level. Furthermore the compensating variables of Model II show a significant contribution, working hours has a negative coefficient (p<0.1) and household income a significantly positive coefficient (p<0.01).

Other differences concerning the significance are identified for the variables of age squared (p>0.1, only for Model I), education (p>0.01, only for Model I) and partner (p>0.01, only for Model I). The variable partner shows different algebraic signs in both models, because it has a low, but significant, impact in Model I and is slightly negative in Model II. Table 5 in the Appendix displays the regression coefficient for each country of both models on job satisfaction, using Belgium as a reference country. The calculations for Portugal are missing in Model II, because the data for the compensating control variables is not available for this country. Except for Portugal and also Cyprus in Model II, all country results are significant.

In order to interpret the single coefficients of unemployment and self-employment and compare the four distinctive groups the predictive margins of the regression results are estimated.

Figure 1 and 2 illustrate the linear prediction lines of job satisfaction, connecting

Entrepreneurs

with

and Employees

with

. The values of each group’s job

satisfactions level and 95% confidence intervals, based on Model I and II, are summarized in

Table 6 and 7 in the Appendix.

24 Figure 1

Figure 2

6. 8 7 7. 2 7. 4 7. 6 7. 8

L ine ar P re di c ti on of Job S a ti sfa ct ion

No Yes

Any period of unemployment and work seeking lasted 12 months or more

Employee Self-employed

Predictive Margins of Model I with 95% CIs

7 7. 2 7. 4 7. 6 7. 8 8

L ine ar P re di c ti on of Job S a ti sfa ct ion

No Yes

Any period of unemployment and work seeking lasted 12 months or more

Employee Self-employed

Predictive Margins of Model II with 95% CIs

25 Discussion The Benefits of Entrepreneurship

Under the assumption that self-reported job satisfaction represents the utility gained from work, many studies of European countries found that self-employed individuals are significantly happier with their jobs than employees and respectively gain higher utility from it (Blanchflower and Oswald 1998, Blanchflower 2000, Benz and Frey 2004, 2008a, 2008/23). The predictive margins of both models confirm these findings and Hypothesis 1. The overall differences are best captured through the marginal effects in the two figures. Figure 1 of Model I illustrates that holding unemployment constant on 0 (0 = no), Entrepreneurs

have higher job satisfaction than Employees

(7.64 vs. 7.30). The same applies for the other side where the value of unemployment is 1 (1 = yes) with higher job satisfaction for Entrepreneurs

, but there the 95 percent intervals is slightly overlapping with the one of Employees

. The low number of Entrepreneurs

leads to a high standard error and causes the overlap of job satisfaction intervals, although their estimates strongly differ (7.30 vs. 6.85). In summary, it is visible that the linear prediction line of self-employed people lies above the one of employees, and computing the slope of the linear prediction lines results in it being less steep for self-employed people with no past unemployment (-0.34) than for employees at the same starting point (-0.46). Figure 2 of Model II strengthens the argumentation to accept Hypothesis 1, because it gives a similar impression, but without the overlap on the right side. Again holding past unemployment constant on 1 (1 = yes), the differences of self-employed people and employees in job satisfaction (7.73 vs. 7.02) and slopes (-0.05 vs. -0.34) of the predictive margins are even be bigger as in Figure 1 (also see Table 6 in the Appendix for the computed values of predictive margins).

Past Unemployment and Interaction Effects

As already discussed in the literature review, encountering an unemployment spell of one

year has strong negative effects on life-satisfaction and might even influence future behavior

(Dartiy and Goldsmith 1996, Clark et al. 2001). Compared to the approach of Benz and Frey

(2008b), the introduction of past unemployment as an independent variable for the regression

allows testing its interaction effects with the variable of self-employment. Another advantage of

this approach is that both independent variables create four distinct groups, which are

26

comparable in terms of job satisfaction using predictive margins. The marginal effects of Model I and II were already applied in the last section to show that self-employed people in this study have a higher job satisfaction than employees. In order to decide whether to accept or dismiss the second hypothesis three more findings are important to be discussed. The first is exposed in the comparison of Entrepreneurs

and Employees

in the descriptive statistics. It shows that mean job satisfaction of Entrepreneurs

is higher than of Employees

and that the former group also excels them in education and household income. It seems like adding past unemployment leads to a stronger distinction between both groups, which makes it unlikely that Entrepreneurs

do not benefit from being self-employed in terms of job satisfaction anymore.

This is strengthened by the comparison of marginal effects of Entrepreneurs

and Employees

. The last section already showed that the predictive margins of job satisfaction of Entrepreneurs

are higher than of Emploeyees

for both models. Figure 2 includes the second relevant finding for Hypothesis 2, as it displays that moving from past unemployment being 0 (0 = no) to 1 (1 = yes), self-employed people only lose minimal job satisfaction. To be more precise, the predictive margins of job satisfaction of Entrepreneurs

are nearly equal to the marginal effects computed for Entrepreneurs

(7.78 vs. 7.73). In contrast there is a large gap between Employees

and Employees

in the calculated predictive margins of Model II (7.36 v. 7.02). Even the “advantaged” group of employees that did not face past unemployment, Employees

, is exceeded by the group of Entrepreneurs

in predicted job satisfaction. As entrepreneurs are not strongly affected by the past unemployment in Model II and it there only increases the gap between them and employees in terms of marginal effects, Hypothesis 2 should be rejected. Past unemployment seems to influence employees stronger. The findings could be interpreted in the way that it is difficult to find a satisfying job with a resume of past unemployment, whereas it does not necessarily have to be a stigma for self-employed people.

The descriptive statistics match this idea, because there Employees

are the worst off group in terms of job satisfaction.

Finally, also the interaction effects can be used to illustrate that Entrepreneurs

gain

utility from being self-employed, in spite of their previous unemployment spell. If the interaction

effect would be negative and significant, the characteristic of having encountered past

unemployment would take away the benefits of increased utility of being self-employed. This

can be ruled out, because the regressions of Model I and Model II lead to a positive coefficient in

27

terms of the interaction effect of long-term unemployment on self-employment (0.123 vs. 0.294).

In Model II the value is significant (p<0.05), whereas in Model I it is only positive and subtracting the standard error would put the value slightly below zero (0.123 – 0.158 = -0.035).

The positive coefficients of the interaction effect together with the significantly positive coefficients of being self-employed in both models show that the higher job satisfaction due to self-employment is also applicable for Entrepreneurs

, which again leads to the rejection of Hypothesis.

The Differences between Necessity and Opportunity Entrepreneurship

The study uses previous long-term unemployment to create two groups of entrepreneurs, which show similarities to necessity and opportunity entrepreneurs. It is important to point out that previous long-term unemployed individuals that become entrepreneurs have not necessarily been pushed into doing so. For instance in-between unemployment and self-employment, respondents may have had a stretch of employed work, which makes it a less strict than most definitions of necessity entrepreneurship. It is also unsure, if entrepreneurs that did not encounter past unemployment are pursuing an opportunity or leave their work as an employee, like it should be the case for opportunity entrepreneurship. Nevertheless the author expects that both approaches correlate under certain circumstances. A period of unemployment should positively influence the decision to find a job quickly. Under this assumption, the longer this stretch gets, the more likely it is that becoming self-employed will turn into a necessity or push factor for the unemployed individual.

The descriptive statistics of this study are comparable to the results of other studies about

necessity and opportunity entrepreneurs which show that past unemployment can be used as a

proxy for necessity entrepreneurs (Block and Koellinger 2009, Block and Sandner 2009, Block

and Wagner 2010). Especially the descriptive statistics of necessity and opportunity

entrepreneurs of Bock and Sandner (2009), which are based on the data from the German Socio

Economic Panel Study (GSOEP), are similar to the groups of Entrepreneurs

and

Entrepreneurs

, but they show even lower values of job satisfaction for necessity

entrepreneurs, which could be due to the different sample, but also based on the definition of

necessity entrepreneurs. These findings, which are summarized in Table 1, can be seen as an

indicator to accept Hypothesis 3a, because Entrepreneurs

clearly exceed Entrepreneurs

in

28

mean job satisfaction (7.69 vs. 7.22). Furthermore, Entrepreneurs

work more hours per week and show higher earnings. As income tends to have a big influence on job satisfaction, the regression analysis will be used to identify, whether the salary might explain the differences in job satisfaction. Finally, in terms of education there are no significant differences found in the descriptive statistics of Entrepreneurs

and Entrepreneurs

, which would contradict the idea that higher human capital endowment of Entrepreneurs

sets both groups apart.

Model I is inconclusive about whether to accept Hypothesis 3a or 3b. The interaction effect is positive, but not statistically significant. Looking at the marginal effects seems to favor Hypothesis 3a, because Figure 1 illustrates a strong difference between both groups of entrepreneurs in predictive margins of job satisfaction (7.64 vs. 7.30), but the 95% intervals of the corresponding values are for the most part overlapping, which reduces its validity. Model II has the highest explanatory power for the third hypothesis, because it also uses the compensating control variables. There the predicted satisfactions levels of both groups of entrepreneurs only show a small difference (7.78 vs. 7.73). Furthermore the 95% interval of Entrepreneurs

completely resides in the confidence interval of Entrepreneurs

, which leads to rejecting Hypothesis 3a or the acceptance of the contradicting Hypothesis 3b. The differences of both models and especially Figure 2 suggest that the effect of reduced job satisfaction for Entrepreneurs

is driven by an income effect for the self-employed. The results imply that the aspiration levels and the push factor associated with past unemployment probably do not have a strong influence on entrepreneurs and put more emphasis on the success of the new business being the decisive factor in terms of job satisfaction.

Implications

Target Groups and the Dangers of Long-term Unemployment

There is the chance that self-employed individuals might be more satisfied with their jobs

due to self-selection (Benz, Frey 2004). This bias would occur, if the majority of individuals

seeking self-employment are individuals that especially value its characteristics. Focusing on a

group of entrepreneurs that faced unemployment of one year or more, the author of this thesis

argues that it is less likely that their sample is biased via self-selection. On the one hand the

findings of this study indicate that Entreprenerus

are still able to benefit from the benefits of

29

being “their own boss”. At the same time it highlights the dangers of long-term unemployment for employees, because although happening in the past, it led to much lower job satisfaction for them as for self-employed people. Therefore the author expects that individuals that suffered from past or ongoing long-term unemployment are an interesting target group for policy makers to promote entrepreneurship amongst them. In terms of job satisfaction this two groups might encounter less reduction in utility from having the stigma of former unemployment in self- employment than in a employed relation.

Policy Examples and Possible Solution

Labor market policies can influence the decision-making process of becoming an entrepreneur or starting a new business (Bergmann and Sternberg 2007). In recent years policy makers created a range of programs in order to make self-employment more attractive, because entrepreneurship is attributed to be an important cornerstone of a functioning and innovative economy (Meager 1996, Van Stel et al. 2005, Bergmann and Sternberg 2007, Audretsch 2009).

Relevant examples for this study stem from labor market policies that promote self-employment as an alternative to dependent employment for currently unemployed individuals. These policies can include push factors, like reduced unemployment benefits and the obligation to accept work offers, but also pull factors that lower entry barriers for self-employment (Bergmann and Sternberg 2007). For instance promotion programs, subsides and lump sum payments can reduce the monetary costs to enter self-employment and start an own business. Most prominent is the

“Ich-AG” or “Me-Inc”, introduced by the federal employment agency (“Bundesagentur für Arbeit”) at the beginning of 2003, which enabled an interest-free subsidy for a limited time (Bergmann and Sternberg 2007, Block and Sandner 2009). It was part of the “Harz IV” reform program in Germany, which between 2003 and 2005 led to an absolute and relative increase of necessity self-employment (Bergmann and Sternberg 2007).

In view of this study such a policy could improve job satisfaction, because it is targeting the above described “disadvantaged” groups. But the study also shows that income has a strong effect on how satisfied entrepreneurs are with their work, because after controlling for it the group with and without past unemployment nearly resided on the same level of job satisfaction.

Therefore steps have to be taken to improve the income and its stability for nascent

30

entrepreneurs. For example, Kautonen and Palmroos (2010) expect that the training of entrepreneurial and business skills and the creation of a sustainable business plan would further increase the level and stability of income (Kautonen, Palmroos 2010). In case of the “Ich-AG”

policy makers already applied changes in 2004, which are that nascent entrepreneurs have to hand in a business plan before starting their new business (Bergmann, Sternberg 2007). Finally, the descriptive statistics showed no significant difference in the variable of education between the two groups of entrepreneurs. In line with previous studies that derived similar findings, the author supposes that the best outcomes for nascent entrepreneurs and start-up programs will be reached, if the participants are able to work in the area of their educations (Block and Sandner 2009).

Limitations and Future Research Limitations

This study relies on self-reported and therefore subjective survey data. Thus, it is not possible to rule out biased or satisficing responses. Another possible limitation is that many factors are influencing job satisfaction, which makes it difficult to include control variables for all of them. The R

of both models are low, but similar to other studies of this research area. This might be explained by missing of variables that capture psychological factors in the regressions, like self-determination and having an interesting job. Finally, governments usually do not act as benevolent authorities and cannot focus only on job satisfaction. Frey and Stutzer (2012) propose in their research that trying to only maximize different happiness indicators would not create an optimal welfare policy (Frey and Stutzer 2012). Therefore, additional measures are required to improve policies.

Future Research

Since the Global Entrepreneurship Monitor (GEM) introduced the classifications of

necessity and opportunity entrepreneurship in 2001, they became a popular research topic within

the entrepreneurship literature. Results of different studies have shown that the separation can be

valuable, but the author of this study thinks that is also important to define and further examine

groups of entrepreneurs with different characteristics, e.g. past unemployment. This will lead to