Entrepreneurship after displacement

Empirical research regarding the individual characteristics that influence the decision to become an entrepreneur is now extensive. Parker (2009) summarizes the empirical evidence and concludes that the probability of becoming an entrepreneur increases with age because the potential entrepreneur, for example, acquires more experience and develops his or her social network. Furthermore, the empirical evidence shows that entrepreneurial activities tend to decrease when a person approaches retirement age. This possible non-linear relationship implies that an inverted U-shaped relationship between age and entrepreneurship is expected. To empirically test this relationship, a squared age variable can be introduced in the empirical analysis. Regarding the performance of the entrepreneurial venture, Parker (2009) concludes that most empirical studies find the age of the entrepreneur to be positively related to survival.

Related to the discussion on the relationship between entrepreneurship and age is the issue of experience and the propensity for an employee to become an entrepreneur and for his or her subsequent performance as an entrepreneur. Shane (2003) distinguishes between general business experience, industry experience, functional experience (e.g., marketing, management), start-up experience, and vicarious experience (observing others being entrepreneurs). On this matter, previous empirical evidence shows that individuals with entrepreneurial experience and individuals with entrepreneurial aspirations are more likely to become entrepreneurs in the future (Hyytinen and Ilmakunnas 2007). According to Bhide (2000) and Dunkelberg et al. (1987), successful entrepreneurs and entrepreneurs who show higher than average growth rates commonly identify their business idea while still at their previous place of employment. Parker (2009) also concludes that the empirical evidence suggests that entrepreneurial experience and industry experience are positively related to the survival of the venture. However, most other variables measuring experience (including managerial experience) are not found to enhance survival chances.

In this paper’s empirical analysis, there is also information on whether an individual has participated in any labor market policy measures, which indicates whether they have a history of unemployment. Regarding the transition from unemployment to entrepreneurship, Carrasco (1999) found that unemployed individuals have a higher probability of becoming self-employed compared to employees. It is also interesting to note that longer durations of unemployment are associated with a higher probability of transition into entrepreneurship (see, e.g., More and Mueller 2002). However, the performance of ventures started by the unemployed is less successful compared to employees who become entrepreneurs (Andersson and Wadensjö 2007). According to a literature review by Parker (2009), previous experience of unemployment does not promote survival and instead has a negative relationship with firm survival.

Furthermore, the empirical evidence shows that women are less likely than men to become entrepreneurs (Parker 2009). According to the Global Entrepreneurship Monitor, the entrepreneurial activity among the Swedish male population is approximately 9%, while the corresponding figure for women is approximately 4% (Singer et al. 2015). The differences in goals and types of entrepreneurial ventures started by women and men also result in performance differences. Parker (2009) concludes that there is a gender entrepreneurship earnings gap and that female entrepreneurs underperform relative to men with regard to turnover, employment creation, and venture survival. For instance, Lohmann and Luber (2004) find that approximately 40% of women who are self-employed remain in business after 5 years, while the corresponding figure for men is approximately 60%.

Education and income can be expected to be positively related to entrepreneurial activity. It can be argued that a certain amount of human capital is required to be involved in innovative and entrepreneurial activities. Furthermore, well-educated employees generally have jobs with more independence, which may imply that they have access to social networks required for entrepreneurial activities. Income is also an indicator of the individual ability of a potential entrepreneur. On the other hand, education and income are usually correlated, and thus the opportunity cost of entrepreneurship may be high. Even if the evidence is far from conclusive, according to Parker (2009), most empirical studies find a positive relationship between education level and transition to entrepreneurship. Regarding the performance of the entrepreneurial venture, survival rates are found to be positively related to an entrepreneur’s education level.

Furthermore, it is suggested that immigrants have a higher propensity to become entrepreneurs. Numerous potential reasons for such an overrepresentation have been proposed (see Parker 2009). Some suggested reasons are that immigrants have access to ethnic capital that may be utilized in an entrepreneurial setting, that their mobility to paid employment is blocked due to language barriers, non-validated foreign qualifications, or even discrimination, or that immigrants are self-selected risk takers. Summarizing the empirical evidence, Parker (2009) concludes that there is a positive relationship between immigration as a determinant of entrepreneurship. Regarding the relationship between immigrant status and survival rates of entrepreneurial ventures, Parker (2009) concludes that there is either no relationship or a negative relationship, although the conclusion is derived based on a small number of studies. Finally, opportunities for entrepreneurial activities vary substantial across regions with entry rates being the highest in metropolitan regions (see Barreneche García 2014; Nyström 2007, 2009). Hence, the probability of transition into entrepreneurship can be expected to be higher in metropolitan regions.

Hoetker and Agarwal (2007) argue that for a successful knowledge diffusion to take place, it is important for an entrepreneur to use an existing firm as a template to observe and perhaps interact with their rules and routines to fully benefit from knowledge spillovers. Hence, it can be argued that previous employment experience in the same industry should have a positive effect on the performance of the entrepreneurial venture created after displacement even if, in the case of business closure, the entrepreneur do not have possibilities to interact with the previous employer.

Regarding the characteristics of the former workplace, the extent to which size and age of the establishment influence the transition to entrepreneurship has been researched. It has empirically been shown that employees in small firms are more prone to become entrepreneurs (e.g., Hyytinen and Maliranta, 2008). What explains this small firm effect on entrepreneurial spawning? According to Elfenbein et al. (2010), a selection effect may be prevalent because employment in small firms is similar to entrepreneurship. Hence, individuals who prefer independence and less bureaucracy and individuals who are less risk averse choose to work in small firms. Furthermore, the lower wages paid by small firms (see, e.g., Wagner 1997 and Waddoups 2007) imply that the employees are more likely to transit to entrepreneurship in order to increase their income. However, even if an alternative to staying in the firm does not exist in cases of business closure, it can be argued that the wage received at the closed business influences the income-related benefits received through unemployment insurance (see section 3.2). Hence, the wage at the closed business can still be argued to influence the opportunity cost of entrepreneurship. A final argument is that employment in small firms develops entrepreneurial learning and skills that could be valuable in pursuing an entrepreneurial career.

As for the size of the establishment, it can also be hypothesized that a number of institutional factors may be of importance in the transition process following a business closure. These arguments are mainly valid when transitions to entrepreneurship are not primarily driven by the identification of opportunities that an employee would like to explore. For several reasons, it can be argued that employees at larger firms may have better chances of finding new employment. First, larger firms may be able to provide more support in the transition process in connection with a closure. Second, there may also be a positive signaling effect associated with previous employment at a large company. Small and new companies face a competitive disadvantage when recruiting relative to large and established firms because they are unable to rely on their reputation and name to attract applicants (Aldrich 1999). Attracting more applicants for a position implies that a company can select higher quality employees from the larger applicant pool (Turban and Cable 2003). Having this background in mind when recruiting displaced employees from a larger company implies the possibility that the employee selected for a large firm came from a larger pool of applicants, which should therefore signal high quality. Finally, employees in large firms are more likely to be covered by a collective agreement that ensures additional support in the transition process (see the discussion below on institutional support in case of firm closure).

Even though new firms are often small, small firms are not always young. Nevertheless, some of the reasons discussed can be argued to be valid for young firms as well. There may be selection mechanisms at work, implying that less risk-averse individuals select into employment in new firms, as well as entrepreneurial learning and lower wages in new firms (Nyström and Elvung 2014) affecting the opportunity cost of entrepreneurship. It can also be argued that available institutional support increases with the age of the firm. However, labor mobility after working with a new firm is relatively less researched in the empirical literature (Nyström 2015). The empirical evidence on the new firm effect on entrepreneurial spawning is less extensive than on the small firm effect on entrepreneurial spawning. Nevertheless, Wagner (2004), Sørensen (2007), and Elvung (2016) find a firm age effect on entrepreneurial spawning.

Regarding profitability, it can be argued that there is a higher probability of identifying profitable ideas worth exploring as an entrepreneur after a closure if the closed establishment was profitable. Many entrepreneurs discover the idea behind their entrepreneurial venture at their previous employer (Bhide 2000). In particular, the empirical evidence suggests that entrepreneurs who got their business ideas at their previous place of employment tend to perform better in terms of higher growth rates (Dunkelberg et al. 1987). In the case of business closure, it should be stressed that entrepreneurs who close a business still describe it as successful (Bates 2005); hence, there are possibilities to identify viable business opportunities, especially in businesses showing positive returns. In summary, firm size and age are hypothesized to be negatively related to transition to entrepreneurship while performance in terms of profitability is argued to be positively related to the probability of becoming an entrepreneur and to the performance of created ventures.

Swedish unemployment insurance has two parts; a basic amount and an income-related benefit that can be received for a maximum of 300 days.⁴ The basic amount applies to individuals who are 20 years or older and fulfill a working requirement.⁵ This basic amount is currently set at approximately €32⁶ per day.⁷ To qualify for income-related benefits, the working requirement mentioned above must be fulfilled. In addition, the individual must have been a member of an unemployment insurance fund for at least 12 months. The income-related benefit compensates 80% of the previous income up to a maximum of €68 per day⁸ (Swedish Public Employment Service, 2015). This corresponds to a monthly salary of €1870, which corresponds to a salary lower than what 90% of what Swedish employees earn. Hence, most Swedish employees would in fact not receive 80% of their salary in case of unemployment, and this has resulted in the development of a complementary insurance market where, for example, labor unions offer insurance plans that cover income losses due to unemployment above the mentioned compensation limit for income-related benefits. At the same time, it should be mentioned that approximately 90% of employees are members of an unemployment fund. However, it is not known how large the share of these employees is who would qualify for income-related unemployment benefits.⁹ In the beginning of the unemployment period, there are no explicit requests from the Swedish public employment service to apply for jobs outside one’s current field or in other regions. However, as the unemployment period increases, there is greater pressure to apply for jobs in different sectors, fields and regions.¹⁰

Job security councils have developed as a part of the Swedish model, which implies that the parties on the labor market take responsibility for the readjustment process with respect to displacement. The support provided by job security councils is considered unique in an international context (Walter 2015). Job security councils are a complement to the unemployment insurance, and the support they offer is available for employees where firms have signed a collective agreement with a union. The funds were established in the period 1972–2012, and the number of employees covered by the funds has accordingly been increasing. Today, more than 3 million employees, i.e., most Swedish employees, are covered by this potential additional support in the case of displacement. There are currently ten organizations connected to the job security councils, and the goal, organization, and extent of the support offered differ (Walter 2015). Some of the job security councils offer additional economic compensation, but for the most part, they provide, for instance, information and coaching on how to find a new job. Annually, approximately 40,000 employees receive some type of support from job security councils (Walter 2015). The largest job security council in the private sector reports that approximately 2% of the employees involved in their support activities end up as self-employed. This is similar to the figure reported by Magnergård (2013). Interestingly, they also find that almost 80% of them perceive that their new job is either more or equally qualifying compared to the job they lost (TSL 2015). However, it should be mentioned that the access to the support offered by the job security councils are connected to the employer having a collective bargaining agreement. According to Företagarna (2011), 60% of the Swedish firms do not have such an agreement and small firms and firms in new and fast-growing industries are overrepresented among firms without collective agreement.

What are the implications of the prevalence of the abovementioned benefits and institutions with regard to the incentives to become an entrepreneur? A reasonable conclusion is to assume that before the 300 days covered by the income-related unemployment benefits, the incentives to embark on an insecure path such as entrepreneurship would be relatively low (given that one qualifies for the income-related unemployment benefit). Hence, necessity-based entrepreneurial activities can be assumed to be rather low during this period. However, as the end of the unemployment insurance period approaches, the opportunity cost decreases and it is reasonable to assume that necessity-based entrepreneurial activities increase. However, it should be noted that before 2000, it was possible to qualify for an additional period of unemployment insurance. In 2000, a labor market policy aimed at stimulating unemployed workers to intensify their job-seeking activities was introduced, thereby removing the possibility of qualifying for such additional periods of unemployment insurance and reducing compensation levels.

To identify displaced employees, the same methodology to define displacement as applied in Nyström and Viklund Ros (2014) is used. The methodology implies imposing a flexible time window to identify displacement and has been developed by Eliason and Storrie (2004) and used by von Greiff (2009) and Nyström (2018). The argument for imposing a flexible time window is that the closure process may start several years before an establishment disappears in the statistics. This is particularly true for closures of large establishments. Using this flexible time window methodology to identify firm closures implies that the closure process is defined to be 1, 2, or 3 years long, depending upon the size of the establishment and the pace of job reductions.¹¹

The individuals affected by a displacement and the rest of the variables used in the empirical analysis are identified through a matched employee-employer database covering all employees and establishments in Sweden and are provided by Statistics Sweden. The coverage of the database enables us to identify displacements and transitions to entrepreneurship during the period 2001–2010.¹² This paper uses only the establishments that have closed due to exits and hence establishments closed due to mergers or splits are excluded. One of the disadvantages with the dataset is that we are able to follow the employees only on a year-to-year basis and that we have limited information on what they do if they exit the labor market. There is information available on whether they have received some income from receiving unemployment benefits or either education related to the unemployment insurance system. However, there is no information on whether, for instance, they enter other educational activities or decide to retire. Thus, following von Greiff (2009), employees older than 55 and younger than 25 are excluded the year prior to displacement because individuals in these age groups have a higher probability of exiting the labor market for education or retirement. For further details about the definition of displacements, see Nyström and Viklund Ros (2014). Table 1 provides a description of the variables included in the empirical analysis. For further details of the variable, see also SCB (2010). Table 2 presents the descriptive statistics for the data included in the empirical analysis.¹³ It should be noted that the number of firm closures is substantially larger than the number of files for bankruptcies. According to Tillväxtanalys (2009), about 15% of all Swedish firm closures are associated with a bankruptcy procedure.¹⁴

The empirical analysis consists of two parts. In the first part, the determinants of the choice to become an entrepreneur after displacement are investigated. Because the dependent variable in this analysis has a binary outcome (becoming an entrepreneur or not), a logit-model is estimated. The estimation is corrected for heteroscedasticity by using robust standard errors. Odd-ratios are calculated to enable the interpretation of the size of the effects.¹⁵ An odds ratio exceeding one should be interpreted as the variable having a positive effect on the probability of becoming an entrepreneur, while an odds ratio below one indicates a negative relationship. For the analysis of performance of the entrepreneurial ventures, the development of turnover, employment, and the survival is analyzed. In line with previous research on firm survival (see, e.g., Mata and Portugal 1994; Esteve-Pérez and Mañez-Castillejo 2008; and Box 2008), the Cox proportional hazards model is applied. The Cox (1972) proportional hazards model is a semi-parametric estimation method where the covariates are multiplicatively related to the hazard, in this case, the closure of the firm. An estimated hazard ratio below one should be interpreted as the covariates having a negative effect on the probability of hazard (closure of the firm), while a hazard ratio above one implies a positive relationship.

Table 3 presents the results from the logit-model estimating the probability of a displaced employee becoming an entrepreneur. In the analysis, the probability of entering entrepreneurship 1, 2, or 3 years after displacement is reported. The purpose of this distinction is to explore whether the transition patterns differ over time assuming that the prevalence of necessity-driven entrepreneurship increases the longer the time passes since the business closure.

As expected and consistent with previous empirical research (e.g., Hyytinen and Maliranta 2008), working for a large establishment decreases the probability of transition into entrepreneurship. In line with our previous expectations, the performance of the prior establishment turns out to be statistically significant, with odds ratios being slightly above one during the first year. However, the odds ratio is below one during the second and third year after closure. Hence, if the transition to entrepreneurship is immediate, it is more likely that the transition is based on identifying a profitable opportunity at the closing business. Regarding the age of the closure, the probability of transition to entrepreneurship has an odds ratio that exceeds one during the first year. Hence, the idea that employees in younger firms closing down are more likely to become entrepreneurs is not supported. For the second and third year, the odds ratio is below one, but the size of the firm age-effect is very small during all 3 years.

Regarding individual characteristics, similar to the general evidence on the decisions to become entrepreneurs, males have a higher probability of becoming entrepreneurs. This finding is valid for all 3 years after displacement. Also in line with previous empirical research, the transition into entrepreneurship increases with age but with a decreasing rate (squared age). Individuals who have entrepreneurial experience have a higher probability of transition into entrepreneurship. However, it should be noted that for the first year of observation, the size of the effect is overestimated due to a methodological problem related to the fact that it is only possible to observe the transition into entrepreneurship on a yearly basis.¹⁶Furthermore, as expected, employees in metropolitan regions are more likely to become entrepreneurs after displacement. Individuals with qualified, specialist and in particular, managerial occupations have a higher probability of becoming entrepreneurs. Finally, an interesting finding is that individuals who have participated in labor market policy measures (either education or income-related unemployment benefits) also have a higher probability of transition into entrepreneurship after displacement.

Figures 3, 4, 5 and show the average development of employees, turnover, and operating profit for the firms started the year after displacement and that survive at least 5 years. The employment development of these firms seems rather moderate in that they start by employing approximately 2.5 employees on average, which increases to approximately 3.2 employees after 5 years. The average turnover at the beginning of the period is 12,800 SEK, increasing to 15,800 SEK after 5 years. For operating profit, the average profit is 800 SEK in the beginning of the period and increases to 1100 SEK after 5 years. These are all quite moderate performance figures. However, looking at the distribution of performance (provided in Appendix 2) some of these entrepreneurial ventures turn out to be successful in terms of performance. The maximum values for these firms are 7.5 million SEK in turnover, 725,000 SEK in terms of operating profit and 313 employees.

Table 4 provides the results of the Cox proportional hazard model studying the survival of the firms generated by displaced employees. Regarding the characteristics of the closing establishment, the results show that coming from a large closure increases hazard rates, i.e., entrepreneurial ventures started by displaced workers in small firms have higher chances of survival. Thus, this result supports the idea that firms generated by displaced employees in small firms are more viable because the employees of these firms may have acquired valuable entrepreneurial knowledge during their time as employees in small and new firms. Entrepreneurship spurred from older firms has lower hazard ratios, but the size of the effect is rather small. Furthermore, entrepreneurial activity spurred from profitable firms reduces hazard ratios, which supports the hypothesis that these displaced employees may have identified a profitable business idea at their previous place of employment. As expected, entrepreneurial ventures started in the same industry as the entrepreneur was working before reduces the hazard rate indicating that starting a business in the same industry generates business ideas with higher chances of survival.

Regarding the individual characteristics, the empirical findings are in line with previous empirical research (Parker 2009) and suggest that the hazard rate is lower for male and older entrepreneurs. A higher wage at the closed establishment also decreases hazard rates. Thus, it is argued that a higher wage at the prior place of employment implies that the potential benefits from entrepreneurship are estimated to be rather high (more likely opportunity-based entrepreneurship), which should enhance the probability of survival. Furthermore, consistent with previous empirical findings on venture survival, entrepreneurs with foreign backgrounds have a higher hazard ratio. However, firms started by individuals with higher education surprisingly show lower survival. Being involved in labor market policies in immediate relation to the firm closure increases hazard rates of the entrepreneurial venture started. This is consistent with the findings by Andersson and Wadensjö (2007) who find that unemployed individuals perform worse than employees starting an entrepreneurial venture. Somewhat surprisingly the effect of entrepreneurial experience does not turn out statistically significant in the empirical analysis. Among the occupational categories, having an office occupation, managerial experience and qualified occupation increase hazard ratios. This is also somewhat surprising because previous research according to Parker (2009) indicates that most types of experience are weakly related to firm survival. Hence, it is possible that these findings imply that experience plays a different role for survival of the ventures when the venture is created by displaced workers. For instance, when individuals with a more qualified occupational background engage in entrepreneurial activities in the aftermath of the displacement, it is potentially a temporary solution to joblessness waiting for a more attractive opportunity to appear. Finally, entrepreneurial activities initiated by entrepreneurs in metropolitan regions have hazard ratios exceeding one reflecting the higher turbulence in terms of entry and exit rates in these regions (Nyström 2007, 2009). Appendix 3 provides the Kaplan-Meier survival estimates which indicate that approximately 80% of the ventures survive their first year while approximately 40% survive 4 years.