The Labor Market Performance of
Individuals with Foreign Backgrounds.
MASTER THESIS WITHIN: Economics NUMBER OF CREDITS: 30 credits
PROGRAM OF STUDY: Economic Analysis AUTHOR: Evelina Ceylan
Master Thesis in Economics
Title: The Labor Market Performance of Individuals with Foreign
Backgrounds.
Authors: Evelina Ceylan
Tutor: Paul Nystedt
Date: 2021-06-15
Key terms: Self-employment, Immigrants, Children of Immigrants, Labor
market
________________________________________________
Abstract
This paper uses individual data from a collected survey, performed in Sweden by the SOM institute, to study individuals with foreign backgrounds in the labor market. We use The Ordinary Least Squares model, where we control for age, education, and gender to explore the difference in incomes between immigrants, children of immigrants, and natives in wage-employment, unemployment, and self-employment. The contribution of this paper is the second generation immigrants, we will assess their performance on the labor market in order to evaluate if self-employment is a profitable alternative. The second generation immigrants act as a benchmark for a functioning integration policy, it is therefore crucial to examine if we can observe any labor market barriers for the second-generation immigrants. The result display that immigrants do perform worse in both wage-employment and self-employment compared to natives.By being self-employed, immigrants earn 25.9 percent less than if they would have been wage-employed. The situation for the children of immigrants is different. Children of immigrants seem to perform better than immigrants on the labor market, especially in employment. One could therefore conclude that since immigrants struggle with finding wage-employment, self-employment may be an alternative. However, self-employment should not be an option for the children of immigrants. It seems that they succeed in finding wage-employment, and they do better in wage-employment compared to employment. So, the promotion of self-employment should be more cautiously made since it may not always have a good economic outcome for the individual.
Table of Content
1. Introduction ... 1
2. Literature review ... 5
2.1 Individuals with foreign background in wage-employment ... 5
2.2 Individuals with foreign background in unemployment ... 8
2.3 Individuals with foreign background in self-employment ... 10
2.4 Difference across immigrant groups ... 15
2.5 Previous finding on the difference between the Natives, first and second-generation immigrants on the labor market. ... 16
3. Data ... 18
3.1 Sample construction ... 18
3.2 Descriptive statistics ... 22
3.2.1 Monthly income distribution for natives, immigrants and children of immigrants ... 22
3.3 Summary statistics ... 24
4. Method ... 26
5. Empirical analysis ... 29
5.1 Result from Descriptive statistics ... 29
5.2 The performance of immigrants, children of immigrants and natives on the labor market ... 30
5.2.1 The performance of Immigrants on the Labor Market ... 31
5.2.2 The performance of Children of Immigrants on the Labor Market ... 34
5.3 The effect of country of origin on labor market performance ... 36
6. Conclusion ... 38
References ... 41 Appendix ... I
Appendix 1: Table of the probability of being in the lowest income group (1), retired (2) and student (3). ... I Appendix 2: Income Distribution with Restrictions Including First Income Category ... II Appendix 3: Income distribution Self-employed Natives ... II Appendix 5: Income distribution Self-employed Immigrants ... III Appendix 7: Income distribution for Wage-employed Immigrants ... III Appendix 8: Distribution of Residuals Model 1 ... IV Appendix 9: Distribution of Residuals Model 2 ... IV Appendix 10: Distribution of Residuals Model 3 ... IV Appendix 14: Income Distribution for Non-European immigrants ... VII Appendix 15: Income Distribution for Other-European immigrants ... VII
1
1. Introduction
As the barriers in the world keep decreasing and people and goods move with more ease, labor force will start moving across borders to get a chance of a better life. The worldwide education expansion (Schofer & Meyer, 2005), have decreased the barriers to the labor market for immigrants. So, globalization and higher education level have somewhat erased the borders; one could therefore imagine that natives and immigrants should be treated more equally on the labor market. However, this is not the case. It is crucial to improve the integration of immigrants into the labor market. Due to the increased inflow of individuals across national borders in the last decade, it has become a more urgent problem. In 2016, the total immigration to Sweden was the highest ever, with over 163,000 immigrants (Statistic Sweden, 2020). In 2019, nearly 120,000 immigrants got residence permission in Sweden, 36% were labor immigrants, 16% were refugees, and 8% were European Union members; these numbers were significantly lower in 2020, most likely due to the pandemic. (Migration Borders, 2020) With this said, one can understand that it is essential to integrate these individuals in the labor market and society; otherwise, it will cause problems. Still, more groups face difficulties in the labor market apart from immigrants, one being the children of immigrants or the second-generation immigrants, which we may call them. Looking at the second-generation immigrants, they as well keep growing. In 2019, there were 615,234 children of immigrants in Sweden. Children of immigrants performance in the labor market is often seen as the benchmark for a successful integration policy because it helps to reduce the unemployment rates in a country (Andersson & Hammarstedt, 2011). Self-employment has been a well-promoted solution to reverse poor labor market outcomes and high unemployment rate for individuals with foreign backgrounds. Studies have shown that immigrants, especially non-western immigrants, are over-represented in self-employment in many countries. They are also more likely than similarly skilled natives to be self-employed. (Borjas 1986; Fairlie and Meyer 1996; Clark and Drinkwater, 2000; Lofstrom, 2002) However, it is not only immigrants who have a higher probability of becoming self-employed. Children of parents that are self-employed have a higher likelihood of being self-employed than children of wage-employed (Andersson & Hammarstedt, 2010). Since immigrants are more likely than natives to be self-employed, so will their children.
2
Many studies focus on the factors that influence immigrants to choose self-employment over wage-employment. (Fairly & Meyer, 1996; Kloosterman et al., 1999; Clark & Drinkwater 2000, 2002; Ekberg & Rooth 2003; Anderson & Wadensjö, 2004; Andersson & Hammarstedt, 2015; Hou & Wang, 2011) Many factors apart from the expected earnings affect individual's choice, such as unemployment risk, language barriers, lack of country-specific skills, lack of networks and discrimination. These difficulties are often called push factors, due to that they push immigrants into self-employment. In addition, one usually talks about pull factors such as specific human capital, inherited knowledge, networks, ethnic enclaves, and home country traditions; these factors significantly affect immigrants' decision to choose self-employment. (Kloosterman et al., 1999; Andersson & Hammarstedt, 2015) Edin et al. (2013) define ethnic enclaves as a municipality where the concentration of an ethnic group is at least twice as large as the share of the ethnic group in the entire population. This concentration may pull immigrant into self-employment since they can satisfy a demand that natives cannot. Also, they may gain both financial and human capital from these ethnic enclaves.
Self-employed immigrants play a significant role in integrating other immigrants into the labor market; immigrants are more willing to employ immigrants than natives (Hammarstedt & Miao, 2019). There is also heterogeneity within the group of immigrants. Some groups of immigrants are more likely to become self-employed. There are many theories as to why this is. Some argue that home country traditions play a significant role in explaining the high self-employment rates for some immigrant groups. Others say that discrimination in the labor market is more prominent for some groups compared to others. (Hout & Rosen 1999; Fairlie & Robb, 2007; Clark & Drinkwater 2000; Andersson & Hammarstedt, 2015; Hammarstedt & Shukur, 2009) Therefore, one may understand why promoting self-employment is a well-used approach to increase participation in the labor market for individuals with a foreign background. While the outcome for the first-generation immigrants in self-employment is well-established, less is known about the second-generation immigrants in self-employment. This paper focus on children of immigrants and their economic performance in the labor market. It is crucial to examine if we can observe any labor market barriers for the
second-3
generation immigrants or if self-employment is an active choice not dependent on struggles obtaining wage employment. If they, as their parents, struggle on the labor market, the assimilation policies that have been established do not operate as we want them to. Segregation between individuals with foreign backgrounds and natives can have significant effects on society. The purpose of this paper is to present the incomes of children of immigrants in Sweden in wage-employment, self-employment, and unemployment and compare them to native incomes. As well as assess if children of immigrants labor market performance have improved compared to immigrants relative to native incomes to evaluate if self-employment is a profitable alternative. Which will allow us to assess if children of immigrants enter self-employment for the same reasons as their parents. This study investigates if the earnings in self-employed differ from the earnings
in wage employed in Sweden. A comparison of the incomes of the first- and second-generation immigrants with natives in Wage-employment, Self-employment, and Unemployment. The contribution of this study is the evaluation of the performance in
regards to the the second generation immigrants in the labor market, particularly in self-employment. The findings are mainly for researchers in their work to determine the actual income difference between children of immigrants and natives. The study will not determine the probability of becoming self-employed, nor will it present the characteristics of self-employed individuals with foreign background. One can predict that in some cases the second-generation immigrants still exhibit disadvantage in the labor market. A possible answer to the research question is that the increased return in wage-employment for the second-generation immigrants due to higher education and higher language proficiency will make self-employment less attractive for the second generation. Therefore, the second-generation immigrants that choose self-employment may be the ones that have a hard time finding wage-employment, which may translate into lower incomes for the second-generation immigrants compared to natives; and the first-generation immigrants. It may also be the case that because the children surpass their parents in many ways, they may do better in self-employment. They may see additional opportunities compared to their parents and will, therefore, have a higher income than their parents.
Four things will be addressed in this study, previous literature 2. on, (2.1) Individuals with foreign background in wage-employment, (2.2) Individuals with foreign background
4
in Unemployment, (2.3) Individuals with foreign background in Self-employment, (2.4) Differences across immigrant groups, (2.5) Previous studies investigating the difference between Natives, first and second-generation immigrants in the labor market. The section following those the (3), (4), (5), (6), will contain, Data, Method, Empirical analysis, and lastly, conclusion.
5
2. Literature review
2.1 Individuals with foreign background in wage-employment
Borjas (1994) argues that the skills across immigrant cohorts that entered the US have decreased. The later arrived cohorts are less skilled; their earnings gap compared to natives will be bigger than for previous cohorts. Over time, the type of immigrants has changed from mostly labor force immigrants to more and more refugees. Theory implies that those who migrate will be the individuals with high skills and high education, positive selected immigrants that will perform well in the labor market. However, since many immigrants seek residency for other reasons than employment, this will not apply; we will therefore have a negative selection of immigrants. (Neuman, 2019) Kogan (2011) study the German labor market and the immigrants who have entered since 1990. The data reveal that the immigrants arriving closer to the year 2010 have, on average, higher education level than earlier cohorts, and some groups are even better educated than the national average. Despite this, immigrants primarily from Eastern Europe, Africa, and Middle East face significant disadvantages on the German labor market. Since the education level has increased across most countries, more recent immigrant cohorts should be more educated than previous cohorts. One can observe a higher education level among immigrants from central and eastern Europe. However, if the education and skills are not transferable to the host country due to low language proficiency and a highly regulated labor market. Then well-educated immigrants may still struggle with finding wage-employment.
Although the education level has increased globally, resulting in better adaption to the labor market, immigrants still face barriers to enter the labor market and obtaining a wage-employment. The low prospects in wage employment for immigrants, caused by lack of language proficiency, low education level, poverty, unemployment, and discrimination, is presented by Fairly and Meyer (1996) as the disadvantage theory. These factors will make it difficult for immigrants to obtain a wage-employment. However, assimilation theory states that second-generation immigrants should experience fewer difficulties finding wage-employment than their parents. The reason being that they resemble natives in knowledge level and language proficiency more than their parents, they are also more
6
familiar with western business and market institutions. (Alba & Nee, 1977) Good education mobility makes them more likely to have higher education and higher knowledge about the host country's education system. This may lead to better economic outcomes for second-generation immigrants than for the first generation. (Kloosterman et al. 1999; Sanders & Nee 1996)
Although one may imagine that the discrimination will decrease due to higher mobility globally, the theory of rising segregation explains a phenomenon of more recently arrived immigrants becoming targets of increasing discrimination. The reason for this may be a change in the immigrant groups. The immigrants arriving in more recent years are, to a greater extent, refugees and not labor immigrants, which was overrepresented in previous cohorts. These immigrants have been observed to have a lower average skill. Some parts may also be attributed to the fact that the most rapidly growing groups of immigrants are the ones that differ the most from natives. The larger the distance is from the origin country to the host country, the more struggles the immigrants will experience in the labor market. This is true for religion and culture as well; when there are significant differences between home and host country, both immigrants and their children will struggle more in order to enter the labor market. (Cutler et al., 2008; Fleischman & Dronkers, 2010) Carlsson and Rooth (2007) investigate ethnic discrimination in the recruitment process in the labor market by sending two similar job applications, except for the name being different, to different advertised job openings. They conclude that the individuals with a foreign-sounding name have a 10-percentage point lower probability of being called for an interview than the applicants with a native-sounding name. Immigrants in Sweden from the Middle East and Africa are more likely than other immigrants to become targets of discrimination on the labor market due to their looks or foreign-sounding names (Åslund & Rooth, 2005). The risk of being discriminated against is still expected to be significant for the second-generation immigrants with a Southern European background and even higher for those with a non-European background. (Ekberg & Rooth, 2003) Individuals with parents from Western and Eastern Europe have a relatively great success in the labor market with incomes in line with natives. However, looking at individuals with parents from Southern Europe andoutside Europe, they struggle in the labor market. (Hammarstedt, 2002)
7
Immigrants face difficulties in the labor market, however the income gap between immigrants and natives in wage-employment is smaller than the income gap in employment. Immigrants in wage-employment earn more than immigrants in self-employment; this could be explained by the fact that there might be a positive selection into wage-employment that decreases the income gap between natives and immigrants in paid work. Andersson (2009) concludes that immigrants would be better off entering wage- employment, at least from an economic perspective. Children of immigrants surpass their parents in many ways; still, they exhibit difficulties in the labor market; children of natives have a higher probability than children of immigrants to become high-income earners (Ekberg & Rooth, 2001). Additionally, as children of immigrants enter the labor market, they have lower labor incomes than individuals with Swedish-born parents and a higher dependency on welfare. Österberg (2000) also concludes that children of immigrants compared to children of natives are more likely to have lower incomes than their parents. Also, they have considerably greater difficulties moving up the earning distribution. However, one must consider that they compare non-European immigrants with natives, the most disadvantaged group.
The type of welfare state or government has a significant effect on immigrants' outcomes. It will either increase immigrants' incentives to work or increase their chances for an income from another source (Kogan, 2011). A more generous welfare state will have adverse effects on the economic outcome of immigrants compared to a less generous welfare state. Kogan (2011) found that immigrant men in countries with a less generous welfare state and a more liberal labor market face fewer disadvantages in the labor market. Although institutions in a county will influence individuals, few results show that policies have a sizeable explanatory power regarding the labor market outcomes of immigrants. (Careja, 2019) One can observe that wages, which could measure productivity, are low in emerging countries, wage-employment is often regulated, and there are barriers to enter the labor market. Self-employment, on the other hand, is less regulated and more easily accessible. Therefore, in these countries, self-employment will be the primary employment type. This is a well-established theory, Poschke (2019) describes the case when a country has a high rate of self-employment as a product of friction in the labor market. Such as large hiring cost and low matching efficiency. Will make it hard to find wage-employment; this will not only cause a high unemployment but it will also increase
8
the attractiveness of self-employment. The paper concludes that the friction in the labor market in emerging countries will drive the high unemployment rate and the high self-employment rate. However, Sweden is far from being an emerging country, nor do they have a high unemployment rate. Still, there are barriers for immigrants when it comes to accessing wage-employment. Hjerm (2004) highlights the impact of the welfare state and concludes that there is greater risk of immigrants being excluded from the labor market in a more regulated labor market such as the Swedish labor market. For example, in the US and, to some extent, the UK, where the labor market is less regulated, immigrants face other risks, such as ending up in low-paid jobs. In Sweden, the risk of ending up in poverty is small since the welfare state will protect the immigrants. However, being outside the labor market will have effects. So, immigrants' success in the labor market is highly dependent on how regulated the labor market is in the country where they arrive.
2.2 Individuals with foreign background in unemployment
Most countries aim is to attract highly educated immigrants; however, the potential of these immigrants is often not utilized (Kogan, 2011). Contrary to what one may think, Fleischmann and Dronkers (2010) found that higher education did not affect either the first or the second generations' unemployment risk. Except for males with only one native-born parent. Looking at immigrants options in the labor market, one must consider the high unemployment rate among immigrants. Since the immigrants have changed from mainly labor force immigrants to more refugees, so has their performance in the labor market. In the 1950 and 1960, the immigrants that arrived performed well in the labor market and had an even lower unemployment rate than natives. However, this changed, and now immigrants face a high risk of being unemployed. Although the second-generation immigrants are more assimilated due to education possibilities and good language proficiency, they still run a higher risk of being unemployed than natives, especially those with non-European backgrounds. (Behrenz et al., 2007) So, individuals with foreign background both first and second-generation immigrants are more likely to be unemployed compared to natives, which makes unemployment the most likely alternative to self-employment for individuals with foreign background. Careja (2019) conclude by using Eurostat data that one can observe differences between natives and immigrants, especially among those that descents from non-EU countries, they have the
9
worst economic outcomes. Non-EU immigrants have higher long term unemployment rate as well as youth unemployment.
In late 2020, the unemployment rate amongst immigrants where 18,9 percent and only 4,7 percent amongst natives in Sweden (Statistic Sweden, 2020). The non-western immigrants are the most discussed since they are the most disadvantaged. Andersson (2009) shows that the difference in employment rate is more substantial than the wage difference, which implies that the biggest issue for immigrants is to enter the labor market. It is often debated that immigrants are forced to become self-employed in order to escape long-term unemployment, especially in markets with minor differences in wages between natives and immigrants due to regulations. (Blume, Kræn, et al., 2009; Klosterman, 2000) Other studies on the Swedish market show that being unemployed and having difficulties finding employment will further contribute to immigrants entering self-employment. Immigrants in unemployment are more likely to turn to self-employment than natives in unemployment. (Dohlmann, 2001). Immigrants are also more likely to exit self-employment into unself-employment compared to natives. (Andersson, 2006) Many enter employment from unemployment, which may push the individuals into self-employment as a last resort for better income. (Fleischmann & Dronkers, 2010; Hou & Wang, 2011; Andersson, 2006; Dohlmsnn, 2011) Hjerm (2004) compare the disposal income between immigrants in entrepreneurism and non-participants in the Swedish labor market and conclude that entrepreneurs have barely 1000 SEK higher disposable income in self-employment than unemployment, when comparing the years 1992 and 1998. Hjerm (2004) suggests that the lack of a significant difference between the two groups could indicate that immigrants entered self-employment in order to escape unemployment.
It has been concluded that children of immigrants as well face a higher risk of unemployment. However, mainly those with parents from non-European counties. Still, children of immigrants may face additional risk since they may have higher requirements and require higher reservation wages and shorter working hours. Abada et al. (2014) examined the group-specific unemployment rates in Canada and observed that it increased from immigrant fathers' to their sons', however not by much. Compared to native fathers and sons unemployment rates that decreased, it shows that children of immigrants still struggle in the labor market. They may still face high unemployment
10
risks since it is more difficult to find a higher job on the occupation hierarchy. Fleischmann and Dronkers (2010) conclude that since the children of immigrants similar to their parents still are affected by their country of origin, one could say that in general, the assimilation process among immigrants in Europe is small or non-existing. The success of immigrants in the labor market depends on their country of origin.
In order to qualify for unemployment benefits in Sweden, have to work at least six months prior to applying for the benefits (The Employment Service, 2021). This becomes problematic for individuals without labor market experience. So, without entering self-employment, many immigrants that did not have the opportunity to join the labor market would have been worse off. Previous studies (Andersson & Hammarstedt, 2015; Ekberg & Rooth 2003) on Sweden showed that years since migration have a significant positive effect on being self-employed. The longer an immigrant has been in the host country, the more likely it is that they become self-employed, which leads one to believe that most people were, in fact, eligible for unemployment benefits at the time they entered self-employment.
2.3 Individuals with foreign background in self-employment
The first and most straightforward explanation of why individuals choose to become employed is because it is a rational choice. Their expected earnings or utility in self-employment is higher than their gains in wage-self-employment. The theory of predicted earnings differential is presented by Clark and Drinkwater (2000). However, it is an extension and development of Lucas (1978) and Evans & Jovanovic (1978), which states that the choice of the sector depends on comparing relative returns. Additional to this, which we will discuss later, is the importance of entrepreneurial ability among individuals and their uneven distribution among groups. Edin & Åslund (2001) use data from Sweden and conclude that immigrants possess lower earnings than natives.
The low income in self-employment results from being unfavored in the labor market with low prospects in wage employment. This will reduce the opportunity cost of self-employment for immigrants. The loss of the return in wage-self-employment when choosing self-employment will be small due to the low wages and the barriers they face. This can be further explained by assuming that we have two groups, natives and immigrants. The theory of predicted earnings differential shows that the wages for natives in
self-11
employment are larger than for immigrants. We know that discrimination lowers immigrants earnings in wage-employment. The theory also implies that since the earnings are larger for natives compared to immigrants in self-employment but mainly wage-employment; the entrepreneurial abilities are more immense for natives than immigrants. A higher proportion of immigrants than natives will choose self-employment since their lower earnings in wage-employment reduce the opportunity cost of stepping into self-employment for immigrants. Immigrants will have lower reservation wage compared to similarly skilled natives and will therefore accept a lower income than natives as self-employed. Immigrants face many difficulties in the labor market, theory therefore suggest that there is a negative selection of individuals that enter self-employment. Self-employment should in theory be an alternative for those who have specific abilities and characteristics that makes them suitable for this kind of employment, however since immigrants face difficulties in the labor market they may choose self-employment even if they may not possess the abilities that are required to run a business successfully. We have now concluded that immigrants have higher self-employment rate than natives, as well as why. (Borjas 1986; Fairlie and Meyer 1996; Clark and Drinkwater, 2000;
Lofstrom, 2002) However, due to this the children of
immigrants will, in turn, have a higher rate of self-employment than those who have native parents which is called intergenerational transmission. It is well known in the literature that the probability of becoming employed is higher among children of self-employed than among children of wage-self-employed (Andersson & Hammarstedt, 2010; Fairlie, 1999; Hout & Rosen, 1999). However, if the main reason for the high self-employment rate among immigrants is difficulties on the labor market. Then one should observe a smaller gap between immigrant's and native's self-employment rate for second-generation immigrants. This is in line with the assimilation theory, which states that the second-generation immigrants should experience fewer difficulties finding wage employment compared to their parents due to that they resemble natives in knowledge level and language proficiency more than their parents. (Alba & Nee, 1977) Sanders & Nee (1996) show that the second generation is more motivated by opportunities in the economy and the benefits that come with self-employment compared to the previous generation. Le (2000) analyze immigrants in Canada, Turkey, and the US and conclude that the higher level of education reduces the probability of being
self-12
employed among second-generation immigrants, although only if they are not located in an ethnic enclave.
Considering the high unemployment rate and the low earnings in wage-employment it implies that immigrants do not only face difficulties entering the labor market, but they also have problems remaining in the labor market. Self-employment becomes a more attractive alternative, which may push minorities into self-employment. Clark and Drinkwater (2000) show that differences between individual predicted earnings in wage-employment and self-wage-employment have a powerful influence in deciding if an individual should enter self-employment or not. This implies that there are differences in wage between immigrants and natives, which suggest that discrimination exist. Abada et al. (2014) states that the attractiveness of self-employment among children of immigrants have decreased since the second-generation immigrants have better prospects in wage-employment compared to their parents. Still, they are expected to perform better in self-employment compared to their parents due to higher financial literacy, which is proven to have a positive impact on becoming self-employed due to better credit from lending institutions (Bobeva 2010).
Hou and Wang (2011) find that immigrants, primarily the newly arrived, were more likely to enter self-employment due to a lack of other options in the labor market compared to natives. This will most likely increase their wage gap compared to natives in employment. However, Hou and Wang (2011) show that the choice of entering self-employment was not due to difficulties in the labor market for either natives or immigrants. It was the ability to be independent, have freedom, escape the hierarchy and further develop and challenge themselves that made them choose self-employment. This highlight the fact that it is not only the income that play a part when deciding to enter self-employment or not. Self-employment can yield a type of utility that wage employment is not able to. It may also be the case that immigrants value this freedom more than natives, which could explain why they choose to stay in self-employment despite having low income. (Andersson, 2009) This suggests that the factors that influence the choice are the same for both natives and immigrants, although we can observe that the effects differ. (Dohlmann, 2001; Andersson, 2006)
13
Andersson and Wadensjö (2004) show through register data that self-employed immigrants in Sweden do in general have lower incomes than their counterparts in wage-employment. They also show that immigrants are overrepresented in self-employment in both Sweden and Denmark. Anderson and Wadensjö (2004) continues to state that immigrants choose self-employment for the reason that they have difficulties getting other types of employment. This indicates that they are becoming self-employed as a last resort. Despite the fact that many studies show that immigrant's economic performance in the labor market is lower compared to natives. Lofstrom (2000) conclude that immigrants that work in high skill occupation earn 12 percent higher earnings in self-employment compared to self-employed natives in 1989. They also find that there is a greater variation in earnings among immigrants in wage-employment compared to self-employment when controlling for country of origin.
It requires great risk to step into self-employment or become an entrepreneur, due to the uncertainties of the economic outcome. However, immigrants take this risk on because compared to the uncertainty of being unemployed, it may seem like a better choice. Comparing the work conditions of immigrants in wage-employment and self-employment, one can see that the work conditions do not change much whether they are in wage-employment or self-employment. They work long and uncomfortable hours with low hourly wage, which is not much different from the conditions in self-employment. (Sanders & Nee, 1996) Also, immigrants are more likely to take on risks since they have already taken risks migrating to a new country. (Andersson & Wadensjö, 2004) Which will reenforce the statements made, that the risk will be smaller for immigrants since their other alternatives are not much more certain when it comes to economic stability.
Lofstrom (2002) use the census population data for year 1980 and 1990 and observe that immigrants in self-employment are on average more educated and have better language proficiency compared to immigrants in wage-employment. This suggest that there is a positive selection of immigrants into self-employment in the US labor market and that immigrants may not choose self-employment as a last resort. The data also show that natives in self-employment do better than their counterpart in wage-employment. This is further verified when they compare the earnings gap between immigrants and natives in
14
and self-employment, which show that the earnings of immigrants in wage-employment do not converge to native income. Immigrants' earnings in self-wage-employment on the other hand seems to converge and surpass the native earnings in both wage and self-employment. Although there is a positive selection Lofstrom (2002) believes that these immigrants would not have reached native incomes in wage-employment.
Promoting self-employment for immigrants solve many problems that immigrants face on the labor market. Such as, decreasing the unemployment rate as well as increasing the assimilation of immigrants into society and the labor market. By having self-employment as an option in the income decision, the probability of immigrants entering the labor market will increase significantly. Therefore, this type of employment is advocated as a good solution and option for immigrants; however, not much concern is taken to the fact that immigrants may not perform that well in self-employment. (Hjerm, 2004) Immigrant's success is an important measure, it shows how well the integration policies function. Therefore, one may understand how vital the share of self-employed immigrants is. Although, self-employment may be a good option for many immigrants it is not for all, therefore one must stress the risk and economic outcome more. Hjerm (2004) summarize Aldrich and Waldinger, 1990; Waldinger et al., 1990 main three factors that drive the high representation of immigrants in self-employment. The first is the market structure and the policies which will affect immigrant's ability to access the labor market and get job opportunities. The second is the individual or the group characteristics that is often mentioned in this context, as well as selective migration and the importance of ethnic network. The third factor that affect the high level of immigrants in self-employment is how the different ethnic groups will adapt to their new environment. This adaption can look different for the same ethnic group in different countries since this is affected by the first factor, the market structure. Hjerm (2004) present an example of this, where Armenians often are entrepreneurs in Los Angles but not in Moscow, where they more often hold higher job positions.
The reason for entering self-employment seem to vary with ethnicity, individual characteristics, welfare type, unemployment risk and the overall economic dynamic and climate in a country and the area of settlement. Ethnic enclaves, will decrease the probability of having a good education and high language proficiency. However, the ethnic enclaves have a more minor effect on the second generation than the
first-15
generation immigrants their ties to their ethnic group are weaker compared to their parents. (Abada et al., 2014)
Andersson and Hammarstedt (2015) show that the unemployment rate is statistically insignificant, which indicates that immigrants are not pushed into self-employment as a last resort for an income. However, another study with data from Sweden and Demark show that non-western immigrants are overrepresented in self-employment in both countries (Anderson & Wadensjö, 2004). Anderson and Wadensjö (2004) also conclude that immigrants choose self-employment for the reason that they have difficulties in getting other types of employment. Still in self-employment, immigrants have lower earnings than immigrants in wage-employment, which indicates that they are becoming self-employed as a last resort. So the literature is divided on whether immigrants become self-employed as a last resort or not, however the outcome depends on the labor market barriers.
2.4 Difference across immigrant groups
The disadvantage theory explains why immigrants are pushed towards employment, but it fails to explain why there are significant differences in the self-employment rate between immigrant groups. Although the choice of the sector will be necessity driven and based on rationality, it will be influenced by individual's abilities and specific characteristics. The individuals with manager abilities will have lower opportunity cost and a higher chance to succeed in self-employment. Therefore, the individuals that choose self-employment will be those that have the highest manager abilities. (Lucas, 1978) These characteristics or manager abilities may be more common among individuals from certain countries, in 1996 the self-employment rate was 58 percent in Turkey, 46 percent in Greece, 11 percent in Sweden, 10 percent in Denmark, 8 percent in the US and 7 percent in Luxembourg (OECD) which indicate that immigrants from Turkey and Greece may be overrepresented in self-employment in Sweden. Therefore, one might imagine that these ethnic groups possess specific characteristics that will pull them into self-employment. Since these abilities will give individuals higher likelihood to succeed in self-employment, we may find that some minority groups may have higher incomes in self-employment compared to other minorities. Fairlie and Meyer (1996) studied the employment rate for immigrants in US and investigated if the self-employment rate in the host country were positively related to the self-self-employment rate
16
in their home country. They estimated the probability of being self-employed or being wage employed based on the potential earnings in each occupation. However, their findings do not support their hypothesis that the most disadvantaged groups have the highest self-employment rates. They find that the more successful an ethnic group is, in terms of earnings in both wage- and self-employed, the higher is the probability of being self-employed. This result show that it is not specifically discrimination and language difficulties that usually leads to self-employment. Instead, immigrants choose to become self-employed because it gives high relative returns, this implies that immigrants are not pushed into self-employment they become self-employed due to that they have good prospects in self-employment. .
Blum & Kræn et al (2009) use danish data and conclude that immigrants from turkey, Vietnam and Iran and immigrants with no citizenship are the groups that are most exposed in the labor market. These groups are the ones that have the most distinct self-employment tradition from their home country, which would make them more successful in self-employment compared to other groups. These groups do indeed use self-self-employment as an escape out of unemployment. The authors suggest that these groups use self-employment as a last resort for an income. In the paper, they use a method where they increase the education level of the groups of immigrants that are most exposed, this did not change the result significantly. Therefore, they suggest that policy makers should focus on decreasing the entry barriers and discrimination in the labor market. Such actions could be to reduce the importance of language proficiency in the labor market or investing in language increasing measures among immigrants. As well as creating employment that requires less formal skills.
2.5 Previous finding on the difference between the Natives, first and second-generation immigrants on the labor market.
Abada et al. (2014) examine the expected earning difference between wage- and self-employment, struggles in the labor market and if ethnic enclaves influence the choice of self-employment in Canada. They study the difference between natives and immigrants also between first- and second generation. The expected earning differentials on the probability of becoming self-employed increased for sons in 2006 compared to their fathers in 1981. The result suggests that there are now greater struggles in obtaining high
17
income in self-employment. When controlling for struggles in the labor market in wage-employment, it had a strong effect only on immigrant fathers. The impact of difficulties in wage-employment decreases over time for young men. This could be a sign of decreasing barriers in the labor market for the second-generation immigrants; they have fewer struggles finding wage-employment.
Borjas (1993) investigates if the earnings differentials observed between immigrants and natives in the firstgeneration increase or decrease in the second generation. The paper investigates immigrants in the US at different points in time, 1940 and 1970. Immigrants observed in 1970 is assumed to be the children of immigrants observed in 1940; comparisons within the age groups are made so that it is more likely that the actual parents and children are captured. Borjas (1993) finds that it is about 7 percent increase in relative earnings across generations for all national groups. The paper concludes that there is some regression towards the mean across generations, but parent's relative earnings are still a strong predictor of children's relative earnings. Hammarstedt and Palme (2012) have data on parents and their actual biological children in Sweden. They compare intergenerational mobility among immigrants and natives for different immigrant groups. They conclude that the intergenerational earnings mobility is lower among immigrants than among natives. Immigrants inherit the position of their parents to a higher extent than natives do Hammarstedt and Palme’s (2012) result show that the second-generation immigrants have turned a 5 percent earnings disadvantage compared to natives for the first generation into a 1.6 percent earnings advantage compared to natives for the second generation. However, second-generation immigrants from Turkey, Greece, the Middle East, and Africa face significantly greater earnings disadvantage than their parents did. They interpret their findings as evidence for the economic status of the family being more important in the intergenerational transmission for immigrants. This could be due to that immigrants are less likely to have access to networks outside the family. The difference between groups may be observed due to that immigrant groups preference may differ when it comes to investing in their children. Therefore, if the institutional setting in the country of origin gives low public support for education and requires sizeable investment from the parents, this might generate large investments from the parents in the host country as well.
18
3. Data
3.1 Sample construction
In the analysis, The SOM Institute Cumulative Dataset is used. The data set is created from the The National SOM survey by the University of Gothenburg. The survey has been performed in Sweden each year since 1986 and the dataset include a selection of questions from the survey that is asked multiple years, since the focus is on the time series. (Swedish National Data Service, 2021) However, since the income variable was restricted from year 2011 to 2018 the study will be conducted on these seven years.
The survey includes information on country of birth, for the individual as well as their parents, which makes it possible to distinguish between natives, first-generation immigrants and second-generation immigrants. The second generation immigrants are individuals born in Sweden, with either one or both parents born outside Sweden. Natives were defined as those born in Sweden with two native parents and immigrants are those who are born outside Sweden. The country of birth is distinguished between those born in Sweden, those born in another country in Scandinavia, those born in Europe and those born outside Europe.
By comparing the income between immigrants, their children and native one can evaluate the outcome of self-employment. The study compares the incomes of self-employed, wage-employed and unemployed from year 2011 to 2018. These variables have been created through the survey questions asking individuals to which occupational group they belong. The employed variable consists of employed blue-collar workers, self-employed with no employees with 1-9 employees and with 10 or more employees. The wage-employed group consist of white-collar workers, white-collar worker with supervisory status, white-collar worker in a senior leadership position, blue-collar worker, blue-collar worker with supervisory status. Farmers is not included in the sample, one could technically include them as self-employed however it is not an option for most people it is rather a passion or inherited tradition etcetera. The unemployed are those that answered that their current labor market situation is unemployed. Since the aim of the
19
paper is to study the labor market outcome of individuals we still include unemployed since it is an option to self-employment they are also still a part of the labor force. The dataset also includes personal characteristics, such as education which is classified in three levels, high, low and medium education level. Low being completed grades until year 9 or below, medium having competed school above year 9 and high being studies or degree at university. We also control for gender, a dummy is included in the regression that is equal to 1 for female and 0 for male. Although information on children and family status do exist it is not included. As Chiswick and Miller (1995) show, both children, family status and ethnic enclaves do not affect the income. Therefore, in this analysis, we will assume that children and family status affect individual's income the same independent of if the individual is wage-employed or self-employed native or have foreign background. The effect of ethnic enclaves has briefly been mentioned; however, this will not be controlled for in the analysis since these effects are minor in Sweden. (Hjerm, 2004)
Since the aim is to study the labor market outcomes of individuals we exclude individuals below 16 and above 75. However, the age is set to those between 25 to 60, in order to capture the group that are most attached to the labor market. (Neuman, 2019) Still, we include six more years than the observed prime age (25-54) in order to give individuals a bit more time to establish in the labor market. After restricting the data, we are left with 11,169 observations. After removing people below 25 and above 60 we will get a data set with individuals that are more attached to the labor market. Still, there are groups that are a part of the data set that do not belong to the labor force, we will therefore exclude students, old age pensioner and early retirement as well as disability pensioners. To determine the characteristics of the removed observations we run a simple Ordinary least square model, this show that students are more likely to be wage-employed and belong to the second generation immigrants (Appendix 1, Column 2). Looking at the retired individuals that are removes, they are more likely to be self-employed and natives (Appendix 1, Column 3 ) We are now left with a data set with 9,945 observations where the individuals included are relatively attached to the labor market. After these restrictions are made, we capture the group that are the most attached to the labor market.
20
Although, the groups that are less attached to the labor market are removed there are still people in the data set that may have less attachment to the labor market since they have very low incomes. However, since we cannot restrict our dataset, so it only includes self-employed or other self-employed that have worked for a sufficiently long time to get a stable income. All self-employed, wage-employed and Unemployed are included as well those who might not stay in that type of employment for an extended period. Consequently, we will remove the first income category of 10,000 SEK or less each month. Since removing the first income group might give us a biased result it is important to determine the characteristics of the excluded observations. It appears that it is unemployed, immigrants and students that are in the first income group which was expected. (Appendix 1, Column 1) Since, self-employed are not most likely to be in the lowest income group we will get a less biased result. (Appendix 1, Column 1) The data is now restricted to age, students, retired, background and employment type as well as the first income category, we end up with a total of 8,474 observations. The income categories that are included are the ones between 10 000 to 60 000 and above. These individuals will most likely be the ones that have the most stable incomes and are the most attached to the labor market. Having less than 10 000 in earning from employment is unlikely if this is their main income source and in the case of a full time employment from either employment type. Although, we may lose some of the self-employed with very low earnings which is important to analyze; The concentration in the left hand side of the income distribution have a negative impact on the result. (Appendix 2)
There are many variable that could have an impact on the income of immigrants, however we only control for education and gender. When controlling for these variables we will be left with, the final amount of 8,454 observations. Unfortunately, the dataset does not include information on skills, talent, abilities, or job experience neither on language proficiency nor years since migration which has been proven to influence the earnings of immigrants. (Christl et al. 2017) As mentioned above one could control for the average hours worked per week, to remove some of the individuals that are less attached to the labor market. This information is included in the survey. Although it will give a good measure for the actual performance it is not included. The reason for this is that it might show a large negative outcome for immigrant entrepreneurs, therefore the difference between native and immigrant earnings might be somewhat underestimated. Including
21
working hours will not give a fair result since self-employed work long hours, but gain other advantages that wage-employed do not. (Hjerm, 2004)
Since the information is drawn from a survey, it may not be as reliable as if it the data set would have been created from registered information. However, the data have information so that many groups in the labor force are included as well as other information to give us the opportunity to distinguish between backgrounds making it a useful dataset. The dependent variable in all regressions is monthly personal income, a drawback is that the income variable is not corrected for inflation. However, for the years that are included in the analysis 2011 to 2018, the inflation was stable with an average of 1.3 (SCB, 2021) with three peaks, 2011, 2017, and 2018. The monthly income may also include payments from social transfers such as sick leave benefit and parental leave benefit etcetera. Since the survey do not state that it is only income from employment. There is also other limitation, one being that the income variable is a category variable which will decrease the accuracy since we do not know the individuals actual income. The income variable will be further discussed when looking at the distribution of incomes as well as the methodology. The paper investigates the earnings of labor market participants over 8 years. The structure of the labor market and opportunities for self-employed might have changed, this is unfortunately difficult to regulate in the regression. (Waldinger et al., 1990) Also, earnings from self-employment may look different compared to earnings from wage-employment they may not take out as large of a salary. However, they have other benefits also at the end of the year they will be able to receive a part of the profit. It is also important to mention that it may exist a heterogenicity among the respondent. Those with high income may be less or more likely to answer the survey truthfully or answer at all which will decrease the accuracy of the study. The two last decade the response rate have decreased significantly for this type of surveys'. Still, the SOM survey have a response rate at 50 percent on average each year which is considerably high for this type of survey. Although, the response rate is comparably high missing out on half of the samples do of course have effects on the representability of the survey result. The survey fails to collect opinions from similar individuals each year, notably people with low political interest, low trust in authorities and researchers, people who are between 20-30 years old, people with less stable life situations and people who were born outside of Europe. Groups that have a high response rate are older people,
22
people who are married, people with higher educations and women. People between 16-39 tend to reply less often than their counterparts, they have an average response rate at 43 percent. (Falk et al., 2021)
3.2 Descriptive statistics
3.2.1 Monthly income distribution for natives, immigrants and children of immigrants
Figure a: Income distribution for Natives Figure b: Income distribution for Immigrants
A useful tool to determine the performance of individuals on the labor market is to analyze their distribution of incomes. The histograms show the concentration in the different income categories of monthly incomes, category 2 being 10 000-14 999, category 3 being
15 000-19 999, category 4 being 20 000-24 999, category 5 being 25 000-29 999,
category 6 being 30 000–34 999, category 7 being 35 000-39 999, category 8 being 40
000-44 999, category 9 being 45 000-49 999, category 10 being 50 000-54 999, category
11 being 55 000-59 999, category 12 being More than 60 000. Observing the distribution (Figure a, b, c) it is clear that immigrants, are the ones that have the worst economic outcome on the labor market. Immigrants distribution are left skewed, they have the highest representation in the lowest income categories. With the highest peak in the category, 15 19 999 SEK followed by a high representation in the category 20
23
Figure c : Income distribution for children of Immigrants
Although, none of the groups have fully normally distributed incomes, native and children of immigrants distribution somewhat match with the one of a normal distribution. Comparing natives and children of immigrants one cannot observe large differences, they both peak in 20 000-24 999. (Figure a, b) Children of immigrants compared to natives have a bit higher representation in the lowest income category. While all groups have a peak in the highest income category, that do not follow the normal distribution, children of immigrants display the highest representation in the last income group being 60 000 or more. Overall, one could conclude that the distribution of the children's income is similar to those of natives.
However, in the case of self-employment the distribution of incomes for children of immigrants differ a bit from natives. Natives incomes is quite normally distributed, and their distribution is quite flat meaning that there are quite large concentration of both high and low incomes (Appendix 3). However, they as well have a high concentration in the low income categories. The distribution of children of immigrants in self-employment is more flat compared to natives. They both have a high concentration in the left hand side of the distribution however children of immigrants have a high and fairly stable concentration in the four middle income categories, which then decrease significantly (Appendix 4). Immigrants income distribution is also concentrated in the left hand side, they have the highest concentration in the low income categories. Their concentration is fairly stable until the fifth income category, 25 000-29 999 where the concentration decrease drastically. (Appendix 5)
24
Comparing the performance in self-employment with the performance in wage-employment, one can observe that all groups perform better in wage-employment natives as well (Appendix 6). The largest difference is for immigrants, their performance increase significant in wage-employment (Appendix 7). The distribution display that the children of immigrants peak in 25 000-29 999 in self-employment, compared to wage-employment where they peak in 20 000-24 999 (Appendix 8). However, in self-employment children of immigrants have a higher concentration in the two lowest income categories. So, there are a much larger deviation in the incomes of self-employed. The similarities of children of immigrants and natives in wage-employment is large both peak in 20 000-24 999, and they have nearly the same pattern in the distribution.
3.3 Summary statistics
Table 2: Summary statistics
Variables Native Immigrant Children of Immigrant
Self-Employed .0838653 (0.28) .1098598 (0.31) .0860158 (0.28) Wage-Employed .9161347 (0.28) .8901402 (0.31) .9139842 (0.28) Unemployed .0439542 (0.20) .1086629 (0.31) .0642945 (0.25) Monthly income 5.974029 (2.38) 5.386068 (2.41) 5.948101 (2.44) Age 25-60 44.2092 (9.98) 44.42055 (9.46) 42.91204 (9.97)
Low Education level .1197496
(0.32) .13559 (0.34) .0823239 (0.27) Medium Education level .4647049 (0.5) .3903232 (0.49) .4687268 (0.50)
High Education level .4155455
(.49) .4740868 (0.5) .4489493 (0.5) Female .5041077 (0.5) .5344015 (0.5) .529094 (0.50) Observations 7115 693 678
25
Table 2 display some descriptive statistics for our sample. Looking at the representation for the different employment types, one could observe that immigrants are the ones that have the highest representation of self-employed in our sample. Natives are the ones that have the lowest representation in self-employment, however only slightly less than children of immigrants. From these statistics, one could also see that Natives are the ones having the highest incomes, however not far more than children of immigrants. Children of immigrants have almost as high monthly income as natives, however with the largest deviation. All background groups have an average monthly income at 25 000 to 29 999 SEK.
It appears that immigrants have slightly higher education attainments, since they have the highest representation in high education level as well as low however not in middle education level. Which is surprising, it is well known in the literature that immigrants struggle due to low education attainments. However, this may be an outcome of low response rate among immigrants with low education attainments (Falk et al., 2021). Looking at the education level for natives compared to children of immigrants they have quite similar representation. Children of immigrants have slightly lower representation in the lowest education level but higher representation in the highest level. Looking at immigrants, it emerges that they are the oldest, they also have significantly higher representation in unemployment compared to the other background groups. This could explain why they have the lowest income. Overall, the representation of men and females in all groups are quite equal, slightly higher female rate among immigrants. Looking at the observations one could note that immigrants and children of immigrants have significantly lower amount of observations compared to natives.
26
4. Method
Our aim is to study the success of self-employed immigrants as well as children of immigrants. This will be achieved by comparing the first generation immigrants and the second generation immigrants' monthly incomes with self-employed natives. We will also perform this comparison for wage employed as well as unemployed to be able to understand what other options immigrants have in the labor market. This will help us evaluate if self-employment is a profitable option for immigrants. As discussed previously, it may be the case that the differences that are observed in monthly income between immigrants and natives is caused by the fact that immigrants and their children might chose self-employment as a last resort to enter the labor market. There are many ways to investigate the incomes of individuals in different type of employment, such as Oaxaca decomposition, however since there are many variables that this dataset cannot control for this approach would have given a biased result. Instead, a simple Ordinary Least Squares (OLS) model is used. Six regressions are performed, this to try to study the relationship between each background group and the employment type. For the three first regressions, natives, immigrants and the second generation immigrants is investigated separately with a regression each. Where monthly income act as the dependent variable, as in all regressions and the three different employment types are included as dummies with wage-employment as a reference. We also control for age education and gender. The following three regressions, the background groups are included as dummies. Each regression is instead performed separately for the different employment types, where wage-employment act as a reference. There are many variables that could explain why specific groups enter a type of employment. However, since the analysis will not study the characteristics of self-employed immigrants, we will only control for the most fundamental variables usually controlled for when studying incomes. Such as age, education level and gender. We estimate the following pooled OLS regression for natives, immigrants and second-generation immigrants.
𝑦!" = 𝛽# + 𝛽%𝐷!,'(+ 𝛽)𝐷!,*(+ 𝛽+𝑋!,,-+ 𝜀!" (1)
Where yit is the economic performance of individual i each month t. As the measure of
27
includes individual variables for the employment type, if the individual is self-employed, unemployed or wage-employed which is included as a reference. 𝑋,- stands for the unique characteristics. ε./, is the error term.
Although equation (1) does give us a good understanding of individuals with different backgrounds performance on the labor market, we perform another regression (2). The same individual characteristics is included, where the pooled OLS regression is estimated individually for each employment type including the different backgrounds as dummies, natives, immigrants (1gen) and second-generation immigrants (2gen). Where natives are included as a reference.
𝑦!" = 𝛽# + 𝛽%𝐷!,#012+ 𝛽)𝐷!,%012+ 𝛽+𝑋!,,-+ 𝜀!" (1)
Additional to this we choose to perform a separate regression for the impact of country of origin, where we look at the self-employment rate and the incomes compared to individuals born in Sweden where the second-generation immigrants are included. A pooled OLS with fixed year effects is also performed. This in order to evaluate if the inflation and real wage increase and structural changes in the labor market have an effect on the result. However, the time span is not that large and the labor market have most likely not changed significantly. Also, observing the inflation it have been quite stable and low during the years examined (SCB, 2021). This implies that adding fixed year effects, should not change the result significantly.
The method of Ordinary Least Squares holds under certain assumptions. The most important to discuss in this case is the minimization of the squared residuals or the error term being normally distributed. This method is believed to function on both continuous but also discreate variables. In our case the lack of a continuous dependent variable will restrict the information. Since, our income variable is discrete it will take on the values 12500, 17500, 22500, 27500, 32500, 37500, 42500, 472500, 52500, 57500 and 62500. Although, it is not a full display of all the possible incomes that one can have and much information is lost it give us a relatively good representations since we have 11 categories. (Horrace & Oaxaca, 2006; Hellevik, 2009; Von Hippel, 2015,
28
Since our income variable will increase continuously, we can fit it and the observations into a line. Had our dependent variable not looked like this, rather being a joint variable on wage-employed, self-employed and unemployed all at once we would not have been able to use OLS. Inspecting the distribution of the error term one could observe that it is fairly close to being normally disturbed, which fulfill the assumption (Appendix 9, 10, 11). Another problem, is that the data is not bound so our observations will be able to take on values over and below the set categories. This could have been solved by creating dummy variables for each income category, however it is not a major problem in this setting if the observations do take on another value. Due to all this, our estimates will not be the Best Linear Unbiased Estimators (BLUE). One can therefore not adapt the result directly to other settings, however it still give us an indication of how children of immigrants perform on the labor market which is our aim. The purpose is to get a general idea of what the relationship looks like, the probabilities are not as important as getting a perception of the relationship. This is the reason for still using the OLS model although it may not be the most optimal in this setting, along with the simplicity of interpretation. The simplicity of being able to interpret our result outweigh the accuracy of using another model. Which may not give us that different results since our dependent variable have 11 categories and the distribution of the error term is quite normal (Appendix 9, 10, 11). (Horrace & Oaxaca, 2006; Hellevik, 2009; Von Hippel, 2015,
