Immigrant-native differences in long-term self-employment

The register data that we use in the paper consist of Swedish linked employer-employee data combined with administrative data from the Swedish tax authority. The data cover the period 2002 to 2016 and is longitudinal, enabling us to follow individuals over time. The data include information on sector of employment, labor, and capital income, as well as socio-economic and demographic information, such as educational attainment and immigration status.⁶

Throughout the paper, we define natives as those born in Sweden and immigrants as those who are born outside of Sweden. Hence, in our paper, immigrants correspond to first-generation immigrants. Using information on birth region, we further classify immigrants into European and non-European immigrants. The motivation behind this classification is that previous research has shown that non-European immigrants typically are considered to have a disadvantage in the Swedish labor market and are over-represented in the self-employment sector. These patterns are not unique to Sweden, and can be found in many other European countries.⁷

We define a person as self-employed if his/her main source of income are self-employment activities.⁸ To analyze long-term self-employment outcomes, our register analysis focuses on self-employment spells that began between 2002 and 2006.⁹ This allows us to follow individuals for at least 10 years. In principle, we could include earlier self-employment experiences into our analysis, but we choose the relatively narrow interval 2002–2006 in order to focus on businesses that develop during roughly the same point in time and hence face roughly the same business climate and macroeconomic conditions. Furthermore, this ensures that the sample used in the analysis based on register data matches the survey data. The data allow us to distinguish between incorporated and unincorporated business owners, as well as to identify people who are wage-employed. In our analysis, we restrict our attention to individuals aged 20 to 64 and exclude self-employed individuals in the agricultural sector.

To compare self-employment performance between immigrants and natives, we focus on annual taxable labor and capital income from the tax administration. These are the two main sources of economic compensation to individuals and are tightly connected to individual well-being. Labor income represents the sum of the employment income from wage and business activities, minus a general deduction. The capital income variable includes interest income from savings, and dividend income from stocks and ownership in closely held corporations.¹⁰

Figure 1 shows the fraction of people in our sample who stay self-employed for at least T years, where T = 0,...,10 and T = 0 corresponds to the year of self-employment entry. From the figure, we observe that the share who remain in self-employment is higher among natives than among European and non-European immigrants. Ten years after self-employment entry, around 40% of natives are still self-employed, whereas for European immigrants, the corresponding share is somewhat lower. Among non-European immigrants, on the other hand, the corresponding share is 30%.¹¹

Our focus is on immigrant-native differences in long-term self-employment. We therefore focus on individuals who became self-employed in 2002–2006 and remained in self-employment for at least ten consecutive years. These individuals are defined as long-term self-employed in our paper and correspond to those who are still in the sample at T = 10 in Fig. 1. In total, the sample includes 54,486, 4089, and 3095 self-employed natives, European immigrants, and non-European immigrants, respectively.

Our register-based analysis will be divided into two parts. We will first graphically analyze how taxable labor and capital income evolve over the first 10 years following self-employment entry. We will then estimate, in a regression framework, the effect of immigrant background on these outcome measures, averaged over these 10 years. The purpose of analyzing the effect of immigrant background on these ten year averages is to approximate “permanent” income measures for the long-term self-employed.

Table 1 below shows summary statistics for our final sample.¹² The first thing to notice is that, on average, self-employed natives have both higher labor and capital income than self-employed European and non-European immigrants. In terms of individual characteristics at the year of self-employment entry, we find that on average, European immigrants are older, and non-European immigrants are younger than their native counterparts. For all groups, the majority of the self-employed are male, with the fraction being the largest among non-European immigrants and smallest among European immigrants. Non-European immigrants are on average the least educated and European immigrants are the most educated. In addition, as compared to natives and European immigrants, non-European immigrants are more likely to be married and more likely to have children under the age of 18 living at home.

In terms of business characteristics, around half of native self-employed individuals start an incorporated business, whereas only around 14% of self-employed non-European immigrants do so.¹³ Furthermore, we find that about 44% of non-European immigrants start businesses in industries with low barriers to entry while the corresponding share among natives and European immigrants is only around 20%.¹⁴

A key aspect of our contribution is that we combine a register analysis for the total population, with the results from a tailor-made, register-linked survey that enables us to learn about the factors that native and immigrant entrepreneurs themselves consider to be important for their long-term self-employment experience.

The survey was designed uniquely for this paper and was conducted in collaboration with Statistics Sweden between September 2018 and January 2019. It targeted a random sample of the total population of self-employed in Sweden who had been self-employed for ten consecutive years between 2007 and 2016 (we exclude the self-employed in the agricultural sector).¹⁵ A total of around 17,500 survey questionnaires were sent out by regular mail and the response rate was around 40%, corresponding to around 7000 respondents. Among the respondents, the shares of natives, and European and non-European immigrants were about 41, 34, and 25%, respectively.¹⁶

We selected the survey questions based on results from previous research regarding self-employment and immigrant self-employment.¹⁷ To understand the nature of the businesses of the self-employed in our sample, we asked different questions about their self-employment experience, and the characteristics of their firms. It is well known that social and entrepreneurial networks, educational attainment, and access to financial capital are of central importance for the possibility to become and succeed as self-employed, and that the importance of these factors may vary between immigrants and natives (see, for example, Blanchflower & Oswald 1998; Blanchard et al., 2008). We therefore asked the respondents how important they perceived that these factors have been for their self-employment activities. Furthermore, it is well known that immigrants may encounter other problems than natives in their firms (see, for example, Constant & Zimmermann 2006; Aldén & Hammarstedt2016). We therefore asked the respondents about perceived success factors and perceived obstacles in their self-employment activities. Table 18 in Appendix B presents the full set of survey questions.

To assess the representativeness of our survey data, Table 2 presents descriptive statistics where we compare the survey respondents and the population data from which the survey was drawn.¹⁸ The table shows that the survey sample and the corresponding population data from which the survey was sampled are overall quite similar in terms of average characteristics. The only clear exceptions are that non-European immigrants in the survey data are more likely to be college educated than non-European immigrants in the population data (40% versus 26%) and are also more likely to be female (27% versus 18%).¹⁹

Comparing the characteristics of natives and immigrants within the group of survey respondents, we find that the share of women is about 30% for both natives and immigrants. The average age in the year 2016 is about 56, 57, and 54 for natives, and European and non-European immigrants, respectively. Among the survey respondents, both natives and immigrants seem to have similar education levels. Non-European immigrants are more likely to be married and more likely to have children under the age of 18 at home than the other groups. Furthermore, among self-employed non-European immigrants, the share of individuals with an incorporated business is much smaller and the share of individuals working in industries with low barriers to entry is much higher, relative to natives and European immigrants. In terms of income, we find that self-employed natives have on average a much higher disposable income than both European and non-European immigrants.²⁰

We begin with a graphical analysis where we, for different immigrant groups, explore how measures of self-employment performance evolve during the first 10 years following self-employment entry.

In Fig. 2, we analyze the evolution of labor income, focusing on yearly population averages as well as the 25th, 50th, and 75th percentiles. Similar to Fig. 1, T = 0 corresponds to the year of self-employment entry (which is potentially different for each individual).

The figure shows that there is a sizable gap between natives and immigrants, especially in the case of non-European immigrants. The gap decreases with self-employment experience, but a noticeable difference is evident even after 10 years in self-employment. The fact that all the graphs are upwards sloping likely reflects the fact that many businesses grow over time, and therefore generate an increasing stream of income to their owners. We can also see that the gap between natives and immigrants diminishes over time. As time goes by, immigrants become more and more integrated into society (e.g., by learning language skills) and are therefore able to achieve business incomes that are more similar to those of natives.

Figure 3 shows the trajectories for capital income. The figure in the left panel shows capital income trajectories in a graph similar to the first panel of Fig. 2. The perhaps most interesting observation is that, in contrast to the evolution of labor income, we find that the gap in terms of capital income between self-employed natives and non-European immigrants, widens over time. Thus, when it comes to capital income, non-European immigrants do not seem to catch up in the same way as in Fig. 2. The gap between natives and European immigrants is on the other hand quite small, at least when looking at average outcomes. Since a sizable share of individuals earn zero capital income, the right panel of Fig. 3 shows the share of positive (non-zero) capital income among natives and immigrants. The figure shows that the share of individuals with positive capital income increases for both natives and immigrants over the course of self-employment. However, even after a long time in self-employment, e.g., at T = 10, less than 50% of non-European immigrants earn a positive capital income.

In Fig. 4, we analyze the evolution of the distribution of capital income among the self-employed with positive capital income by inspecting the evolution of different percentiles. We observe a clear, widening gap between natives and immigrants in the top quartile of the distribution, in particular between natives and non-European immigrants.

In the analysis so far, we have not considered that natives and immigrants run different types of firms with respect to corporate form and type of industry. Figure 5 therefore shows the share of self-employed who have an incorporated business (left panel) and the share who work in industries with low barriers to entry (right panel), and how these variables evolve over time. We see that the fraction of non-European immigrants who have an incorporated business is very low. The share at T = 0 is about 15% among non-European immigrants, while the corresponding shares for natives and European immigrants are about 35% and 50%, respectively. However, the extent of incorporation increases over time in all immigrant groups, but there is no pattern of convergence or divergence across immigrant groups.²¹ About half of non-European immigrants work in industries with low barriers to entry, whereas the share is only about 20% for natives and European immigrants. Furthermore, it does not seem that long-term entrepreneurs change industries during the first 10 years following self-employment entry. There is a small tendency for European immigrants to leave industries with low barriers of entry, but, overall, the likelihood of moving between low- and high-entry barrier industries is low for both natives and immigrants during the first 10 years.

In Appendix A, motivated by the immigrant-native differences documented in Fig. 5, we re-do Figs. 2 and 4 based on whether an individual starts an incorporated business or not at the time of self-employment entry (see Appendix Figs. 6, 7, 8, 9, 10, 11) and based on whether an individual starts a firm in a low barrier-to-entry industry or not (see Appendix Figs. 12, 13, 14, 15, 16, 17). These results show that capital income is generally much higher among those who start an incorporated business and those who start a business in an industry that does not classify as an industry with low barriers to entry. However, independently of the corporate form and type of industry at the time of self-employment entry, the immigrant-native gap in labor income seems to converge over time, whereas the immigrant-native gap in capital income seems to widen over time.²²

One reason for the stronger growth of capital income among natives could be that natives run businesses that are more likely to generate capital income to their owners as the businesses grow. Another explanation could be that self-employed natives are better at obtaining a larger share of their self-employment compensation in the form of capital income, which is beneficial due to the typically lower marginal tax rate on capital income in the context of the Swedish dual-income tax system (where the taxation of capital income is separated from the taxation of labor income).²³

Finally, we have repeated Figs. 2, 3, 4, and 5 where we compare natives with immigrants, dividing up the analysis depending on if these immigrants arrived to Sweden before or after (or in) 1990. These figures can be found in Figs. 25, 26, 27, 28, 29, 30, 31, and 32 in the Appendix and show that cohorts that arrived to Sweden before 1990 generally have a more positive long-term development of labor and capital income relative to more recent cohorts, especially among European immigrants.

We now turn to a regression approach to examine immigrant-native differences in average labor income and capital income over the course of self-employment. The benefit of the regression approach is that we can control for a set of individual characteristics at the year of self-employment entry. We focus on the following specification: The outcome variable \(\log Y_{i}\) is the logarithm of individual average income over the period T = 0 to T = 10. The purpose of focusing on 10-year averages is to approximate a “permanent” income measure for the long-term self-employed. X_i represents a vector of control variables, including age, gender, marital status, education, and the number of children in the household under age 18, and 𝜖_i is an error term. The variables of interest are the dummies European_i and NonEuropean_i that indicate whether a self-employed person is a European or a non-European immigrant, with the reference group being self-employed natives. The estimates of the coefficients β and γ capture immigrant-native difference in average earnings.

Table 3 presents the estimated coefficients of interest for each outcome variable, with and without controls.²⁴ The results in specification (2), with controls, show that, relative to natives, average labor income is about 24% lower for European immigrants and about 41% lower for non-European immigrants. Mirroring the graphical evidence, the largest immigrant-native differences are found for capital income, which is around 60% lower for European immigrants and around 90% lower for non-European immigrants.²⁵ The average income gap between immigrants and natives may not fully reflect how the gap looks like in different parts of the outcome distribution. Therefore, in Table 12 in Appendix B, we re-run the above specification using quantile regression. For both labor and capital income, the results show that the earnings gap is largest in the bottom of the distribution (at the 25th percentile) and smallest at the top of the distribution (at the 75th percentile).

Notice that the longitudinal nature of the data also allows us to perform a traditional panel data analysis. However, the major advantage of using 10-year averages is that this approximates a measure of permanent long-term income for the long-term self-employed in our sample. This cross-sectional approach also allows us to smooth out year-specific fluctuations in labor and capital income, which is especially important in the case of capital income, due to the volatile nature of capital income flows. However, as a robustness check, we have also performed a traditional panel data analysis in Table 16 in Appendix B. This analysis confirms the large immigrant-native differences in both labor and capital incomes.²⁶

Figure 5 showed that there are large immigrant-native differences in terms of the likelihood to have an incorporated firm and the likelihood to have a firm in an industry with low barriers to entry. It is therefore interesting to estimate the earnings gap when splitting the sample along these dimensions. Table 4 shows that for both labor income and capital income, the earnings gap between self-employed natives and immigrants remains large when restricting the analysis to either individuals with the same corporate form or to individuals who operate in industries with similar barriers to entry. However, the immigrant-native gap is larger among self-employed with unincorporated firms and firms in industries with low barriers to entry.

One key element for the success of self-employment activities is country-specific human capital which is accumulated by immigrants by spending time in the host country. In our sample, at the time of self-employment entry, European immigrants have stayed in Sweden for about 20 years and non-European immigrants have stayed in Sweden for about 14 years. Thus, the two immigrant groups differ not only in terms of their country of birth, but also in terms of their length of exposure to the host country.²⁷ Table 5 repeats the analysis in Table 3 dividing up the analysis depending on the length of stay in Sweden at the time of self-employment entry. The results show that for both European and non-European immigrants, the estimated income differences relative to natives decrease with respect to the duration of stay in the host country. The results suggest that country-specific human capital at the time of self-employment entry can have a long-run impact on business performance and earnings.²⁸ However, it is important to note that there are substantial differences among natives and immigrants, even among those who have stayed a very long time in Sweden before they start their business, especially in the case of non-European immigrants.

Of course, those who have stayed in Sweden for a long time naturally also belong to earlier cohorts of immigrant to Sweden. To shed further light on the role of immigration cohort, Table 6 presents a table similar to Table 5 that divides up the results depending on if the self-employed individual migrated to Sweden before or after 1990.²⁹ The results confirm the pattern observed in the graphical evidence, namely, that cohorts that arrived to Sweden before 1990 generally have better long-term outcomes. In fact, European immigrants who arrived before 1990 have long-term outcomes that are very similar to natives.

The previous literature has found that native self-employed in general have better economic outcomes than immigrant self-employed along a number of economic dimensions. This literature has mainly focused on the choice to become self-employed and short-run self-employment outcomes. Our contribution is to take a long-term perspective, and also analyze important immigrant-native differences in the evolution of capital income along the course of self-employment.

Our graphical analysis showed that labor earnings for long-term self-employed natives and immigrants appear to converge over time, while capital income appears to be widening over time. Moreover, a regression analysis of long-run averages revealed that a substantial immigrant-native gap persists, even after controlling for different factors at the time of self-employment entry. The gap is more pronounced among those with unincorporated businesses, and the widest gap is found when comparing natives and non-European immigrants, and when considering immigrants with a shorter duration of stay in Sweden at the time of self-employment entry or immigrants who arrived more recently to Sweden.

It is well-known that non-European immigrants have difficulties in the Swedish and many European labor markets. Hence, non-European immigrants are more likely than natives to be pushed into the self-employment sector. This is also partly reflected in our results that show that a large number of non-European self-employed immigrants are working in low entry barrier sectors and choose to pursue their business activity in unincorporated firms. However, substantial immigrant-native differences remain even when we restrict attention to those who start the same type of business in the same type of industry. To obtain further insights into the determinants of these remaining differences, we turn to analyze our survey targeting the long-term self-employed.

Table 7 describes the background characteristics of our survey respondents, focusing on individual and family characteristics, business characteristics, language skills, working hours, and perspectives on self-employment. The first three columns show the mean value for each variable for natives, and European and non-European immigrants. The subsequent two columns test for statistically significant differences between each of the two immigrant groups and the native reference group.

We begin by noticing that the average age at self- employment entry is significantly higher among immigrants than that among natives. This holds true especially for non-European immigrants. An interesting question is whether there are immigrant-native differences in the inter-generational transmission of the choice to become self-employed and what role the family plays for the long-term self-employed.³² We find that immigrants are in general less likely to have parents who are self-employed.³³ This could reflect that immigrants are pushed into self-employment due to lack of better alternatives rather than being pulled into self-employment, by, for instance, family traditions. We also find that non-European immigrants are less likely than natives to have family members or relatives work in their business.

Turning to business characteristics, we find that immigrant-owned businesses are on all accounts much more likely to interact with people with a foreign background, in the form of either employee relationships, or relationship with suppliers or customers.³⁴ The difference relative to natives is strongest for non-European immigrants. Furthermore, compared with non-European immigrants, natives report to be more likely to have a personal relationship with their customers.

Language skills are obviously very important for immigrants to succeed in the labor market. However, it is unclear to which extent language skills play a role for the long-term self-employed and whether there are differences between immigrant groups. We find that compared to natives, non-European immigrants are about 40 percentage points less likely to be proficient in Swedish while the corresponding difference between European immigrants and natives is only about 17 percentage points. We also note that non-European immigrants are also less likely to be highly proficient in English compared to natives and European immigrants.

Regarding hours of work, we find no differences in terms of hours of work between natives and European immigrants. However, non-European immigrants work much more, on average almost five more hours per week. This is despite the fact that non-European immigrants, on average, generally earn much less than natives. The proportion of individuals among the long-term self-employed who have another job is however small among both natives and immigrants.³⁵ We also see that among those who report that their partner works in their business, immigrants report that their partner works longer hours than natives do, especially in the case of non-European immigrants. Thus, a clear message is that non-European long-term self-employed immigrants, and their partners, work much more than their native and European counterparts.

A specific purpose of the survey was to go beyond the traditional monetary measures of self-employment success to obtain a broader view of immigrant-native differences in self-employment success. In particular, we asked respondents whether they enjoy being self-employed, whether they instead would have preferred to be employed as a regular employee, whether they consider themselves to have achieved their goals as self-employed and whether they think that they will be self-employed five years into the future. What we find is that the attitude towards self-employment differs dramatically among self-employed natives and immigrants, even though we focus on individuals who all share a long history of self-employment. Compared to natives, immigrants consider self-employment to be less enjoyable, would rather be wage employed, and to a lesser extent feel that they have achieved their goals. This is line with the common explanation that many immigrants are pushed into self-employment due to the lack of better labor market opportunities. Interestingly, immigrants also feel, to a much greater extent than natives, that luck is more important than hard work for economic success.

To sum up, we find large immigrant-native differences in several important dimensions. The most striking ones are those that relate to working hours and perspectives on self-employment.

Table 8 analyzes how the respondents view different factors that could be important in explaining their long-term success in self-employment. The questions focus on the role of the family, social networks, work experience, and access to capital.

Between 25 and 30% of respondents perceive their partner to be important for the success of their business. Non-Europeans consider their partners to be the most important and the difference relative to natives is statistically significant. Children are considered to be less important in general, with only about 11 to 13% of respondents considering them to be important for the success of their business. There is a small tendency for non-European immigrants to consider children to be more important, but the difference relative to the other groups is not statistically significant. Relatives are considered to be less important than both partners and children, but here there is still a clear immigrant-native difference, with around 10% of non-European immigrants considering children to be important for the success of their business, whereas the corresponding share for natives is only around 6%.

Previous literature has found that social networks are important not only in the startup phase of a business, but also for its long-term performance. Our survey shows that former employers and colleagues are important for both self-employed natives and immigrants, but natives value them much more, especially compared to non-European immigrants. Previous business partners appear however to be roughly equally important to immigrants and natives, with around 20% of respondents expressing that such connections are important for their business. Interestingly, a significantly higher share of self-employed non-European immigrants consider previous classmates, neighbors, and friends to be important. The answers to the questions of the importance of the social network suggest that there are potentially large immigrant-native differences in terms of how important networks outside the workplace are considered for the success in self-employment.³⁶ The fact that primarily non-European immigrants seem to, relative to natives and European immigrants, lack social networks with a strong connection to the labor market, could be one explanation for their worse long-term self-employment outcomes.

An important question is to which extent human capital acquired in the home country is transferable to the new host country as individuals migrate and to which extent education acquired in the host country is perceived as valuable for the success in self-employment. Both European and non-European immigrants consider their education in their home country to be valuable, with no clear differences between the two groups of immigrants. Most respondents also consider having an education in the host country (Sweden) to be an important advantage. The shares are 55 and 63%, respectively for European and non-European immigrants, which should be compared to the share among natives, which is 68%. Regarding past work experience in the Swedish labor market, about 60% of natives, 59% of European immigrants, and 55% of non-European immigrants find this factor to be important. Overall, and quite naturally, we see that natives value education and working experience more than immigrants. This is one explanation behind the the fact that a significant proportion of self-employed non-European immigrants work in industries with low barriers to entry, where the human capital requirements are smaller.

Previous research has found that natives and immigrants differ in terms of their ability to obtain capital to fund the startup and growth of their businesses.³⁷ From Table 8, we see that both self-employed natives and immigrants consider bank loans to be an important source of capital, although European immigrants weight the importance less than natives. Furthermore, we find that immigrants, particularly non-European immigrants, are more likely to consider other sources of finance, to be important.

Our survey contained several questions with the purpose of identifying immigrant-native differences in terms of how respondents perceive and rank the importance of different obstacles to self-employment success. The results are shown in Table 9. Overall, we find that non-European immigrants experience more obstacles than natives do. In particular, they perceive high taxes, high salaries, access to capital, tax complexity, and finding appropriate employees to be more of a problem relative to natives, and often also in comparison to European immigrants.

Our results show that high taxes and tax complexity are perceived to be, relative to natives, more problematic among non-European immigrants but not among European immigrants. One natural potential explanation for this is that non-European immigrants are less familiar with European tax systems that share many common features. The immigrant-native divergence in the perceptions about taxes can have important implications since governments often use tax policy to stimulate entrepreneurship. If there are immigrant-native differences in terms of the understanding of the host country’s tax system (which some research seems to indicate is the case; see e.g., Bastani et al., 2020), such measures can potentially exacerbate the immigrant-native differences in long-term self-employment outcomes that we have documented in this paper.³⁸ We also see that that self-employed non-European immigrants have more difficulties in finding employees. This can be due to several factors, such as ethnically segregated hiring networks or ethnic preferences among employees (see, e.g., Giuliano et al., 2009).