University Education as a Compensation Strategy Among Second-Generation Immigrants Susanne Urban Linköping University The aim of this study is to determine whether immigrants and their children use a compensation strategy that involves achieving higher than expected education, given their parents’ level of education and income. The study uses data for all individuals in Sweden who fin- ished elementary school from 1990 to 1992. Parents’ level of educa- tion and income is in general positively associated with higher odds of having university education. However, some immigrant groups show the reverse pattern of the impact of parents’ income. The results support a compensation strategy developed in groups and families with low level of integration in the labor market. INTRODUCTION It is well-known that a parent’s educational level has a large impact on their offspring’s educational career. This impact is not straightforward but can be influenced by a number of factors at societal, group, familial, and individual levels. A number of different theories have been proposed to explain the differences in intergenerational educational careers among immigrant groups. The focus here is to determine whether differences between immigrant groups can be explained based on the level of integra- tion in the group and by parents’ level of integration. Integration is mea- sured as income per level and type of education. Compensation theory, as developed in this article, proposes that a first-generation immigrant group with high levels of education but low-income levels will aim for higher levels of education for their children – “the second generation” – to com- pensate for perceived discrimination in the labor market. This study will test the significance of the compensational strategy at the group and individual levels with the use of large scale quantitative data. The study also uses register and census data for all individuals and their parents who live in Sweden and finish ninth-grade schooling from © 2012 by the Center for Migration Studies of New York. All rights reserved. DOI: 10.1111/imre.12005 IMR Volume 46 Number 4 (Winter 2012):919–940 919 920 INTERNATIONAL MIGRATION REVIEW 1990 to 1992. The output is measured as having a university degree dur- ing 2004–2006. THEORY AND PREVIOUS STUDIES Economic and educational progress among immigrant groups not only differs from the majority population, but it is also unevenly distributed among different groups. Socioeconomic factors such as family income, parental occupation, and family structure have been used to explain inter-group differences (Blau and Duncan, 1967). From an economic per- spective, it seems reasonable that having parents with low payoffs from education would discourage their offspring to invest in higher education. However, a large number of studies have shown that social class can only explain a small variance in academic performance (White, 1982; Schmid, 2001). George J. Borjas (1992) assumes that ethnicity acts as an external- ity in the human capital accumulation process. The skills of the next gen- eration depend on parental inputs and on the quality of the ethnic environment in which parents make their investments or their “ethnic capital.” He concludes that the skills of the next generation depend not only on the skills of their parents but also on the average skills of the ethnic group in the parents’ generation. Explanations based on class have also been complemented with sociocultural perspectives. Here, it is argued that school performance can be more fully explained if factors related to the culture of ethnic groups and the context of the groups′ reception in the receiving country are taken into account. Ogby (1987, 1991) argues that voluntary minorities succeed to a greater degree compared with involuntary minorities because they want to conform to the “rules and mores” of the public schools and develop a community that is more supportive toward higher education. Involuntary minorities, on the other hand, develop an oppositional collec- tive identity that results in an ambivalent view toward the educational sys- tem. Ogby concludes that voluntary immigrant groups in the U.S. (such as Asians) more often promote upward mobility and place a high value on education (as compared with most Latino groups). The importance of maintaining the original culture has been proposed by other researchers in the field (Zhou, 1997; Zhou and Bankston, 1998). Portes (1999) stresses the importance of access to communities with great social and cognitive support. Fejgin (1995) argues that different socialization patterns in different ethnic groups can encourage as well as discourage academic EDUCATION AS A COMPENSATION STRATEGY 921 performance. High educational achievements among South and South- East Asian immigrant children are often explained by the high value their communities attach to education (which, in turn, may be related to Con- fucianism, which emphasizes the importance of study). The socioeconomic and sociocultural perspectives have been further developed by examining the importance of social incorporation and the context of reception in the country (Portes and MacLeod, 1996; Portes and Rumbaut, 1996). Portes and his colleagues argue that political refu- gees, from countries such as Cuba and Vietnam, are treated more sym- pathetically by the receiving society and are able to use governmental and private recourses to create solidarity and entrepreneurial communi- ties. They tend to live in close-knit communities, in which academic achievement is encouraged. This argument is contrary to the argument proposed by Ogby who had difficulties explaining why Mexicans, who may be categorized as voluntary migrants, have such a disadvantaged position in the U.S. labor market. Portes places more emphasis on the importance of group reception and argues that Mexicans and Haitians, who are primarily “economic” migrants, have experienced more discrimi- nation. Therefore, they tend to hold marginal jobs and do not succeed in developing cohesive communities. They are also deprived of the eco- nomic subsidies granted to legal refugees. Waters (1999) argues that negative reception or overt racism might cause a negative cultural devel- opment in a group, which subsequently results in disinvestment in the future and the formation of oppositional identities in the second generation. Immigrants and their children may also be able to use a common ethnicity to overcome structural advantages. The ethnic community can work as a form of social capital that provides support as well as control. The prerequisite for receiving ethnic group support for upward mobility and assimilation is having the opportunity to lean on the material or moral resources available in the family and the immigrant community (Zhou, 1997). Zhou argues that when there is a generational conflict (“dissonant acculturation”), the children will be deprived of family or community resources and be led away from parental expectations. I argue that migrants with a weak position in the labor market will develop strategies to deal with the situation. Furthermore, a negative reception of highly educated groups can cause counteractive agency that includes developing a compensation strategy within the educational sys- tem. Portes and Zhou (1993) and Zhou (1997) discuss three forms of 922 INTERNATIONAL MIGRATION REVIEW adaptation: (1) integration into the middle class, (2) assimilation into the underclass and permanent poverty, and (3) rapid economic advancement within a tight ethnic community. The mode of adaptation considered in this article is integration into the middle class, whereby migrants compen- sate for low prospects in the labor market with high education. A classic study by Duncan and Hoffman (1981) argues that overed- ucated workers who are overqualified for the position they hold earn more than correctly matched workers in the same position. Conversely, under- educated workers earn less than correctly matched workers in the same position. A large number of studies have confirmed this pattern (Rubb, 2003). However, other studies have shown that immigrants get lower pay- off from the level of schooling than the majority population (see, for example, Chiswick and Miller, 2008 and for studies with Swedish data: Österberg, 2000; Behtoui, 2006; Nekby, Vilhelmsson, and Özcan, 2008). Research from Sweden confirms that sex, ethnicity, and class affect marks and choices among post-compulsory school alternatives (Broady et al., 2000; Erikson and Jonsson, 1994; Lindblad, 1994; Dryler, 2001; Jonsson and Rudolphi, 2011; for an international overview see Heath, Rothon, and Kilpi, 2008). Arai, Schröder, and Vilhelmsson (2000) use longitudinal interview material from a representative sample of young people who completed 9 years of compulsory school education in 1998. On the one hand, low compulsory school results and a tendency to drop out of higher education mean that youth with foreign-born parents and are born in Sweden or abroad are more likely to have compulsory school as their highest educational level. On the other hand, Arai, Schröder, and Vilhelmsson (2000) demon- strate that the chance of receiving high marks in school are a third higher for those born outside of Europe than for those born within Europe. They report that the difference in schooling results between youths born abroad and those born in Sweden is statistically significant. Alternatively, the difference between those born in Sweden with one or both parents born abroad and those with both parents who were born in Sweden is sta- tistically insignificant. Differences in marks between youths who were born in Sweden (and their parents) and those born abroad who have com- pleted all their schooling in Sweden are due solely to the background fac- tors used in the study. The most important explanatory variable is whether any household member held a job. Similä (1994) analyzes the educational careers of 1.9 million youths born in 1953–1970 with focus on the odds of having a theoretical EDUCATION AS A COMPENSATION STRATEGY 923 gymnasium qualification1 and 1.2 million youths born in 1953–1964 with focus on the odds of having a university degree. The samples include all youths whose parents answered the Swedish census in 1990. Children to parents born in Sweden have higher odds of having a theoretical gymnasium degree and of having a university degree. However, after controlling for birth year, sex, father’s education, and household socioeconomic position, it is clear that children of for- eign-born parents are more likely to attend a theoretical gymnasium. This result is supported by a later study conducted by Jonsson and Ru- dolphi (2011) who show that students from most ethnic minority groups exhibit a higher propensity of choosing academic over vocational study programs as compared with the majority population in Sweden. As shown in the study by Similä (1994), differences in the odds of having a university degree among children of parents born in Sweden and those born in other countries are negligible when controlling for the back- ground variables (but is still negative). The impact of different birth coun- tries is heterogeneous: having one or two parents born in Poland, Greece, Yugoslavia, or the Baltic states has a positive effect on attaining a degree from a theoretical gymnasium as well as on attaining a university degree. Children from all birth countries/regions, with the exception of Denmark, Finland, Norway, and Latin America, have higher odds of receiving a uni- versity degree. Having parents born in the Nordic countries had a nega- tive effect on having a university degree. Similä suggests that compensatory behavior is the main explanation for differences among immigrant groups. He argues that as immigrants from more distant countries are subjected to greater discrimination, their children receive fewer options, lower access to networks and fewer social contacts that provide access to the labor market. Obstacles in the labor market may lead to compensatory behavior in the educational system – specifically, seeking advantage by taking the route to higher education. Jonsson and Rudolphi (2011) argue that the most plausible explanation to their results is that ethnic minority groups expect higher benefits from academic education than from vocational schooling and that this, in turn, may be due to anticipated discrimination in the (manual) labor market. 1 A theoretical program in the Swedish “gymnasium” (gymnasium follows after the com- pulsory 9 years elementary school) is mainly a university preparation program, as opposed to a vocational program that is directed to the labor market. 924 INTERNATIONAL MIGRATION REVIEW Plans for return migration can also have an impact on their choice of edu- cational paths (Dustmann, 2008). A number of interview studies indicate that young people born abroad or with foreign-born parents are highly encouraged by their parents to climb the intergenerational educational career (Knocke and Hertzberg, 2000; Ljung, 2000; Nordenstam and Wallin, 2002; Lundqvist, 2010). Young people with migrant backgrounds seem to have higher edu- cational ambitions as compared with young people born in Sweden with Swedish-born parents (Arai, Schröder, and Vilhelmsson, 2000). Several other interview studies report that young individuals with foreign-born parents want to compensate for their disadvantage in the labor market with higher education. Ljung (2000) writes of a “driving inner force” among young people in the second year of the gymnasium. This driving force pushes them to advance from their parents’ marginalized position and into the Swedish society. Similar to Similä (1994), Lundqvist (2010) concludes that the availability of social ties of value in the labor market is lower for young immigrants. Subsequently, they have to place greater trust in the educational system. The theory on compensation strategies is in stark contrast to cultural and assimilationist theories. These theories posit that low internal resources (measured as ethnic capital, social capital, or supportive culture) coupled with negative reception by majority society only produce negative outcomes and lead to downward acculturation. Similä (1994) presents a number of other issues that may con- tribute to the explanation. The first is the problem of measurement. Indeed, measuring parental educational level and socioeconomic posi- tion is too imprecise, for it does not capture all relevant details. The type of education, for instance, is not measured in his data. The sec- ond issue is selective migration. Refugees have higher education than compared with labor migrants. Migrants voluntarily choose to migrate and therefore have higher ambitions for the future. The third explana- tion is selective return migration. Migrants from countries closer to Sweden (the Nordic countries) who aim for higher education for their children are more likely to move back to their home countries. Parents from Nordic countries who do not aspire for higher education for their children remain in Sweden. The final issue is school treatment. Language problems for Nordic migrants may not be taken seriously in schools, thereby resulting in insufficient school language support for Nordic migrants. EDUCATION AS A COMPENSATION STRATEGY 925 THE HYPOTHESIS This study will test whether or not immigrants invest in higher educa- tion. We do not put into question whether the level of education is affected by parental educational level as a result of socialization and expected educational payoffs. Rather, we will investigate other inciting mechanisms that operate to influence educational levels. Is it possible that perceived discrimination may motivate immigrants to pursue higher education? The main argument advanced here is that children of immi- grants display a compensation strategy. We argue that foreign-born par- ents motivate their children to attain higher levels of education to counteract perceived labor market discrimination. Higher parental involvement in children’s education may result in both higher school performance and in the preference for university education (i.e., have both primary and secondary effects, in the terminology of Boudon, 1974; Eriksson and Jonsson, 1996) A number of studies have observed that particular immigrant groups in Sweden tend to receive higher education compared with others with the same background (Similä, 1994; Jonsson and Rudolphi, 2011). How- ever, the possibility of a compensation strategy has not previously been developed in detail. According to the compensation hypothesis, ethnic segmentation in the labor market can serve as an incentive for immigrants to develop com- pensation strategies. Perceived discrimination may stimulate parents’ and children’s ambitions for university education because they can interpret the low payoff from education as a result of discriminatory practices in the labor market. This might stimulate immigrants to respond with indi- vidual and group agentic actions to counteract what is perceived as unfair ethnic segmentation. This study tests for the presence of a compensational strategy that can be measured at the individual level (1) and at the group level (2). The hypothesis is as follows: (1) Foreign-born parents who have a low payoff from education will compensate with strategies that encourage their offspring to enroll in university education. (2) Belonging to an ethnic group that has a large proportion of members with low payoffs from edu- cation will increase the compensation strategy to a far greater extent than the impact from parents’ socioeconomic status. This effect will be more likely in groups with high levels of education, where the attitude toward academic education is expected to be more supportive. 926 INTERNATIONAL MIGRATION REVIEW THE SETTING – A POLARIZING LABOR MARKET One of the main characteristics of the Swedish social welfare state has been an educational policy, aiming at leveling out the importance of par- ents’ socioeconomic position for their children’s access to education. In spite of economic incentives (no fees, general access to student allowances, and student loans), active policies for equal school quality and for wid- ened participation, parental educational level still remains an important signifier of their children’s educational careers. Previous research (Shierup, Hansen, and Castles, 2006, chapter 8; Urban, 2008a,b; Rauhut, 2007; Nordström Skans and Åslund, 2009; Socialstyrelsen, 2010) has shown that polarization in the capital city of Stockholm and in Sweden at large is increasing. Urban (2008a) shows that both the proportion of the overall population with low income and the proportion with high income have grown during the period between 1990 and 2004. Differences between men and women have been reduced, but differences between the native and foreign-born populations have become larger. This pattern cannot be explained by differences in educa- tional level, because Stockholm (as well as the country in general) has received a large amount of highly educated refugees and migrants. Eco- nomic polarization drastically increased during the economic crisis in the first half of the 1990s, but decreased in the second half of the 1990s. The economic crisis gives a specific context to the study. The immigrant popu- lation in Sweden has a diverse ethno-national background. Labor migra- tion from other Nordic countries increased by the late 1970s, but the early 1980s saw the migration of individuals, mostly refugees, from Asia, Africa, and Eastern Europe. 9.2 percent of the total population was born abroad in 1990. DATA AND METHODOLOGY The dataset used in this study is retrieved from the LISA database (1990– 2006), which is administered by Statistics Sweden. LISA is a longitudinal database containing annual information compiled from a wide range of registers on personal and demographic variables, namely the education, income, and employment status of all individuals registered as living in Sweden from the age of 16 and over. We have complemented the data with geographic information and marks attained in ninth (final) year of compulsory school. EDUCATION AS A COMPENSATION STRATEGY 927 For this study, we selected a sample of all individuals who had marks from the ninth year of school during 1990, 1991, and 1992 (N: 311,247). A total of 294,606 individuals were still registered in Sweden fourteen years later (in 2004, 2005, and 2006, respectively). Indi- viduals who did not have any parents residing in Sweden, whose parents were both older than 65 or did not have any recorded education were excluded from the sample. Parents without any kind of registered income at the time of their children’s completion of compulsory school and indi- viduals with no income during the measurement year were omitted from the analyses to avoid problems with unregistered emigration. A total of 289,012 individuals remained in the sample. As the data problem is mainly concerned with unregistered foreign education, the registered highest educational level was never overestimated. Table 1 displays more details on the differences between the original pop- ulation and the sample. The results given here are only valid for those with at least one parent who lived in Sweden during the year the child finished their ninth year of compulsory school. Despite the presence of problems concerning information on foreign-born parents’ level of educa- tion, the dataset provides us with the opportunity to analyze the intergen- erational educational career of a large group of individuals entering adulthood between 1990 and 2006. The outcome is measured as the odds ratio of having a university degree (post-high school) 14 years after finishing compulsory school (in years 2004, 2005, and 2006, respectively). The analytical strategy is to investigate differences among individuals from different national/regional backgrounds and to determine what extent the variation among children with parents born in different coun- tries can be explained by a number of background characteristics, namely parental background and average income in the group of origin. The first step identifies some regions of origin with positive effects on odds ratio to have university education, after controlling a wide range of background TABLE 1 POPULATION VERSUS SAMPLE Population 90/91/92 Remain 2004/05/06 Sample Number of individuals 311,247 294,606 289,012 Year of birth, mean 1974.9 1974.9 1974.9 Female, % 49.0 48.5 48.5 Born in Sweden, % 94.1 95.8 96.3 University education, % 44.2 44.3 44.5 928 INTERNATIONAL MIGRATION REVIEW characteristics. We developed separate models for individuals with parents born in Sweden, for those from “Low-odds regions” and for those from “High-odds regions” to study the differential impacts of relative income per level of education in more detail. The parent with the highest level of education is chosen as the rep- resentative parent. The highest completed level of education is measured as follows: (1) elementary school 9 years or less, (2) post-elementary school (high school) up to 2 years, (3) post-elementary school (high school) maximum 3 years, (4) post-high school (university) less than 3 years, (5) post-high school (university) 3 years or longer, and (6) PhD degree. The representative par- ent must have income, and if both parents have the same level of educa- tion, the one with highest income was chosen. The data on highest level of education in LISA are derived from the Register of Education. This register has, in general, good coverage of formal education acquired in Sweden. A number of measures have been developed by Statistics Sweden to improve the data and to collect information on edu- cation obtained abroad. Information on parents’ education in 1990, 1991, or 1992 is based on three sources: register data, the 1990 census (the last one conducted in Sweden), and a survey aimed at immigrants in 1990. Birth country of the representative parent is used to investigate the effects of different ethnic backgrounds. Countries with large proportions in the dataset are not aggregated, but countries with smaller proportions or nearby countries with similar effects have been aggregated in the mod- els. A total of 18 different background regions are used in the models. The latest year of registered immigration is used to construct four addi- tional categories: (1) born in Sweden or immigrated their birth year, (2) immigrated from 1 to 6 years of age, (3) immigrated from 7 to 10 years of age, and (4) 11 years or older when immigrated to Sweden. Information on the level of education and earnings (disposable income) of the representative parent is measured the same year as their child finished school, that is, 1990, 1991, and 1992, respectively. The earnings are divided from the median earnings of the total population (all individuals living in Sweden with income > 0 between the ages of 16–65) by sex, age group (5 year span), and educational level and type.2 This 2 Education is measured according to the International Standard Classification of Education (ISCED). Both length of education and type of education are included in the classification scheme. EDUCATION AS A COMPENSATION STRATEGY 929 measure is called Relative Income Education in the models. The models also test for interaction effects of Relative Income Education by parent′s level of education. The mean of Relative Income Education per country (73 birth countries are registered in the dataset, and smaller countries are aggregated) for the whole population (16–64 years old with income > 0) is used as a measure of group-level integration in the labor market. This variable is called Relative Income Country. Less than one means that indi- viduals born in this country on average have a lower income as compared with the average income for all individuals of the same sex, age group, and level of education (the same year as when the child finished ninth grade). Information on sex and year of ninth-grade completion is also used at the individual level. Type of family, measured the same year as year of ninth-year completion, is used in the models to control for the impact of growing up with one or two parents in the household. A dummy for liv- ing in a metropolitan area (greater Stockholm, Gothenburg, or Malmoe) is also included. Finally, individual marks attained in elementary school are also included in the model. The models have been tested separately for those having a father or a mother as the representative parent. Additional descriptive statistics of variables used in the models are shown in Table 2. FINDINGS According to the compensation hypothesis, individuals who belong to groups who experience discrimination in the labor market will develop a compensation strategy where academic education constitutes an important part. A total of 35 percent of parents born in Sweden have university edu- cation, and 46 percent of their offspring also have university education 14 years after finishing ninth grade (Table 3). The percentages among parents born in other countries are in some cases higher: parents born in East Europe, Iran, South America (exclusive Chile) and in particular par- ents from Other North West (North America, Japan, Australia, and New Zealand) have higher percentages. Parents born in other regions have a relatively smaller percentage of university education. The percentage among children with university education is higher compared with the percentage among their parents for all regions (except from Other North West). Foreign-born parental relative income is generally lower compared with that of parents born in Sweden, but higher than parents from the 930 INTERNATIONAL MIGRATION REVIEW TABLE 2 DESCRIPTIVE STATISTICS Total Number of individuals 289,012 University education, % 44.5 Female, % 48.5 Parent level of education Elementary school less than 9 years, % 8.4 Elementary school 9 years, % 9.1 Post-elementary school (high school) up to 2 years, % 32.8 Post-elementary school (high school) maximum 3 years, % 15.7 Post-high school (university) less than 3 years, % 15.7 Post-high school (university) 3 years or longer, % 16.9 PhD. studies, % 1.3 Parent relative earnings Mean 1.16 Age of migration Born in Sweden or immigrated same year as birth, % 96.5 1–6 year, % 1.5 7–10 year, % 0.7 11 or older, % 1.3 Metropolitan residential, % 36.5 other Nordic countries and Chile. Given the fact that the level of education among immigrants might be underestimated, there is a risk that relative earnings are overestimated. Choosing only the parent who has the highest level of education reduced this problem. Table 4 shows results for logistic regression estimations of the odds of having university education 14 years after finishing ninth grade. The baseline is a male individual born in Sweden, who finished ninth grade in 1990, lives outside the largest metropolitan areas, and has married parents with the representative parent having 9 years of elementary schooling or less, both the parent and the individual are born in Sweden. Table 4 shows models with a stepwise inclusion of the explanatory variables, starting with a model with only the 18 regions of origin (model 1). Most of the regions display lower odds of having university education compared with Sweden. Individuals with parents born in East Europe, Iran, or Other North West have significantly higher odds of having uni- versity education when compared with individuals with Swedish origin. After controlling for sex, year of finishing ninth grade, family type, and age of immigration (ref born in Sweden), the same regions show positive and significant results. South America changes from insignificant to signif- icant positive effects when controlling for these background variables. EDUCATION AS A COMPENSATION STRATEGY 931 TABLE 3 NUMBER OF INDIVIDUALS, PERCENTAGE INDIVIDUALS WITH UNIVERSITY EDUCATION, PERCENTAGE PARENTS WITH UNIVERSITY EDUCATION, AND MEAN RELATIVE PARENTAL INCOME, BY REGION OF ORIGIN University Parent university Parent relative Number education, % education, % income, mean Sweden 255,603 45.61 35.24 1.17 Other Nordic 14,152 32.85 21.31 1.18 Greece 705 39.15 11.63 0.99 Spain 226 40.27 25.22 1.11 East Europe 3622 48.92 40.03 1.08 Other Europe 3113 45.74 35.69 1.09 Other North West 354 56.5 60.73 1.07 Turkey 1946 20.91 6.12 0.98 Iran 886 55.3 38.71 0.98 Yugoslavia 3090 31.68 9.61 1.07 Iraq 219 36.53 31.05 1.04 Arabian Peninsula 325 44 36 1.02 Asia 1144 39.77 19.76 1.11 Chile 1246 29.53 25.12 1.23 Rest South America 728 43.82 41.76 1.16 Africa 1234 30.63 21.07 1.15 Syria 411 23.36 10.95 1.00 Unknown 8 50 50 0.95 The third model also includes parental level of education. It is not surprising to see that the higher the level of parents’ education, the higher the odds ratio for their offspring to have university education. Parents’ relative income education (by sex, age group, and detailed educational level) is also included in this model. The income measure also has, as expected, a positive effect. The effect of having a parent born in East Europe or Iran remains positive when the models include parents’ educational level and relative income. Odds ratios for Greece, Spain, and Asia turn significantly positive. The forth model includes interaction effects between level of edu- cation and relative income, which show that parents’ relative income is somewhat more important for those with a parent with post-elementary or university education, as compared with the relative income of parents with only nine or less years of education (which is the baseline). The fifth model includes the variable Relative Income Country. The measure is used here as an indication of the level of integration of immi- grants from the parent’s country of origin in the Swedish labor market. The hypothesis states that a low level of inclusion in the labor market at the group level will result in higher educational aspirations and thereby in a higher odds ratio for high education. The odds ratio for Relative Income Country is lower than one but is not significant. However, when running this model separately for those with father (n = 181082) and mother 932 INTERNATIONAL MIGRATION REVIEW TABLE 4 LOGISTIC REGRESSION. ODDS RATIO FOR UNIVERSITY EDUCATION, ROBUST STANDARD ERROR BY PARENTS’ COUNTRY OF BIRTH AND BY REGIONAL BACKGROUND. STEPWISE INCLUSION OF EXPLAINING VARIABLES 1 2 3 4 5 Parent’s region of birth OR SE OR SE OR SE OR SE OR SE Nordic 0.58 0.01 0.62 0.01 0.79 0.02 0.79 0.02 0.79 0.02 Other Europe 1.01 0.07 1.04 0.08 0.97 0.04 0.97 0.04 0.93 0.05 East Europe 1.14 0.07 1.32 0.09 1.15 0.04 1.15 0.04 1.09 0.07 Greece 0.77 0.00 0.75 0.00 1.35 0.01 1.34 0.01 1.21 0.10 Spain 0.8 0.00 0.86 0.01 1.15 0.00 1.16 0.00 1.12 0.03 Turkey 0.32 0.00 0.32 0.02 0.64 0.01 0.63 0.01 0.57 0.05 Iran 1.48 0.00 2.36 0.22 2.03 0.13 2.04 0.13 1.82 0.21 Yugoslavia 0.55 0.04 0.56 0.03 0.89 0.05 0.89 0.05 0.84 0.06 Iraq 0.69 0.00 0.93 0.08 0.98 0.05 0.98 0.05 0.90 0.07 Arabian Peninsula 0.94 0.10 1.08 0.13 1.15 0.10 1.15 0.10 1.06 0.12 Africa 0.53 0.10 0.65 0.13 0.82 0.07 0.82 0.07 0.81 0.08 Chile 0.5 0.00 0.7 0.06 0.76 0.03 0.76 0.03 0.79 0.04 Asia 0.79 0.14 0.94 0.16 1.33 0.15 1.33 0.15 1.28 0.13 Other South America 0.93 0.13 1.22 0.19 1.09 0.13 1.09 0.13 1.08 0.13 Syria 0.36 0.00 0.45 0.03 0.76 0.03 0.75 0.03 0.68 0.06 Other North West 1.55 0.06 1.6 0.07 1.02 0.07 1.03 0.07 0.97 0.07 Unknown 1.19 0.00 1.79 0.05 1.85 0.03 1.86 0.03 1.64 0.17 Backgrounda No Yes Yes Yes Yes Parents education Post-elementary 1.42 0.01 1.25 0.03 1.25 0.03 school 2 years Post-elementary 2.64 0.04 2.33 0.04 2.33 0.04 school 3 years Post-high school 4.03 0.05 3.47 0.06 3.48 0.06 (university) <3 years Post-high school 9.46 0.24 8.15 0.19 8.15 0.19 (university) >3 years PhD 24.38 1.21 21.04 0.97 21.04 0.98 Parents relative 1.07 0.00 0.98 0.01 0.98 0.01 income education Relative income/ 1.11 0.02 1.10 0.02 post-elementary education Relative income/ 1.13 0.01 1.13 0.01 university education Relative income country 0.56 0.28 Pseudo R2 0.0051 0.0265 0.1223 0.1223 0.1224 Note: Odds ratios in bold: p > |z| < 0.05. Omitted categories for the independent variables are as follows: male, year of finishing ninth grade 1990, parents married, parent born in Sweden, outside metropolitan area residential, parents education: elementary level or less. a Female, year of finishing ninth grade 1991, year of finishing ninth grade 1992, cohabiting, single father, single mother, other family categories, Metropolitan residential, age when immigrated 1–6 years, 7–10 years, 11 years or older. (n = 107930) as the representative parent, it becomes clear that Relative Income Country has a significant negative effect only if the parent with the highest education is the mother. In a model with only those with mothers EDUCATION AS A COMPENSATION STRATEGY 933 as a representative parent, the odds ratio for Relative Income Country becomes 0.18 (0.10) and significant. This means that if the mother has the highest education in the family (or if it is the same as the father or if she is single) and is from an immigrant group with low level of integra- tion, the individual will have higher odds of having a university degree. For individuals whose father has the highest education, the odds ratio is positive, but not significant (1.26 [0.68]). It should be noted here that if both parents have the same level of education, the parent with the highest income was chosen. The inclusion of Relative Income Country reduces the odds ratios for almost all regions. The odds ratio of having a parent born in East Europe then becomes insignificant. This leads us to conclude that Relative Income Country explains some of the positive odds ratios, but only a small portion. Pseudo R2 only increases from 0.1223 to 0.1224 when Relative Income Country is included in the model, which means that it only helps to modify the model and not to add more explanatory power. In the separate models for individuals with mothers as the representative parent, only having a par- ent born in Iran has positive effects. In the model for those with the father as the representative parent, the effect of having a parent born in East Eur- ope stays positive. This also holds for the other regions of origin that were positive in the total model (Greece, Spain, Iran, and Asia). Finally, a sixth model (not shown) shows the individuals’ marks (varies between 1 and 5) from the ninth grade as a control variable. As marks can also be a part of a compensation strategy, this may overcontrol the data. Marks have, as expected, a positive effect on the odds of having a university degree. Marks do change some of the results: odds ratios for those who have a parent born in Greece, Spain, or Iran rise, and previ- ously negative or insignificant odds ratios for those who have a parent born in Iraq, Chile, or South America become positive. The effect of hav- ing a parent born in Asia is insignificant. The previous negative impact of having a parent born in Africa also becomes insignificant, and the effect of having a parent born in Other North West becomes significantly nega- tive. In general, when taking marks into account, the odds ratios of having a university education becomes higher for most groups compared with those of individuals with parents born in Sweden. The odds ratio of Relative Income Country does not change much as compared with the previous model (rises from 0.56 to 0.58). The general conclusion from this first step of the analyses is that individuals with parents who are born in certain regions have higher odds 934 INTERNATIONAL MIGRATION REVIEW of having a university degree compared with individuals with parents born in other regions or in Sweden. Relative Income Country seems to explain a small proportion of the positive effect from some regions of origin – this supports the hypothesis of a compensation strategy based on higher aspira- tions for high education if belonging to a group with low integration in the labor market. Relative Income Country also has a negative impact on the odds ratio of having university education. However, this seems true only for those with the mother as a representative parent. In the next step of the analyses, we look further at the different effect of parents’ level of education and relative income in different groups of origin. First (Table 5), we compare individuals with a parent born in Sweden with those with parents born in regions that had a neg- ative effect on the odds of having university education in Table 4. The model that only includes individuals with parents born in these Low-odds regions (Nordic countries, Turkey, former Yugoslavia, Chile, and Syria) is quite similar to the model with individuals with parents born in Sweden. However, the effect of having a parent with a PhD degree is smaller as compared with individuals with parents born in Sweden.3 The effect from parents’ relative income is positive in the model for only those individuals born in Low-odds regions, but not significant. When running the same model separately for those with fathers as the represen- tative parent, the positive odds ratio from relative income and the inter- action effect from relative income by university education become significant. It stays insignificant for those with mothers as the representa- tive parent. The effect of having a parent with a PhD degree is also lower in the model for children with parents born in High-odds regions (Spain, Greece, Iran, and Asia) as compared with those with parents born in Sweden, but higher compared with individuals with parents born in Low-odds regions. The effect of relative income is negative but insignificant, but the effect of Relative Income Country is significantly negative. The same pattern stays when running models with individuals with both fathers and mothers as the representative parents. 3 It should be noted that the lowest level of education, which is the baseline, might not be comparable between the separate models. The lowest level of education can vary from no schooling at all and illiteracy, to a full 9 years of schooling. It is therefore not appropriate to compare the precise values of the estimates between models with different groups. The focus is on the pattern and how it differs between groups. EDUCATION AS A COMPENSATION STRATEGY 935 TABLE 5 LOGISTIC REGRESSION. ODDS RATIO FOR UNIVERSITY EDUCATION, ROBUST STANDARD ERROR BY PARENTS’ COUNTRY OF BIRTH, SEPARATE MODELS FOR SWEDEN AND LOW-ODDS COUNTRIES (NORWAY, FINLAND, DENMARK, TURKEY, YUGOSLAVIA, CHILE, AND SYRIA) Low-odds High-odds Sweden, country, country, N = 255,603 N = 22,079 N = 2961 Backgrounda Yes Yes Yes Parents education OR SE OR OR SE Post-elementary school 2 years 1.23 0.04 1.59 0.23 1.62 0.34 Post-elementary school 3 years 2.32 0.07 2.69 0.35 2.08 0.39 Post-high school (university) <3 years 3.48 0.11 3.51 0.47 4.37 1.16 Post-high school (university) >3 years 8.28 0.27 7.06 0.81 8.43 2.51 PhD 21.87 1.35 15.05 3.05 18.38 11.93 Parents income Parents relative income education 0.97 0.03 1.10 0.09 0.99 0.06 Relative income/post-elementary education 1.12 0.02 0.92 0.07 1.05 0.17 Relative income/university education 1.14 0.04 1.10 0.07 0.74 0.18 Relative income country – 1.30 0.45 0.07 0.03 Pseudo R2 0.1222 0.0865 0.0972 Note: Odds ratios in bold: p > |z| < 0.05. Omitted categories for the independent variables are as follows: male, year of finishing ninth grade 1990, parents married, parent born in Sweden, outside metropolitan area residential, parents education: elementary level or less. a Female, year of finishing ninth grade 1991, year of finishing ninth grade 1992, cohabiting, single father, single mother, other family categories, Metropolitan residential, age when immigrated 1–6 years, 7–10 years, 11 years or older. When running the model separately for individuals with parents born in the High-odds countries – Spain, Greece, Iran, and Asia (Table 6) – it seems clear that having a parent who is born in Asia and holds a PhD degree gives much higher odds for the child to have a uni- versity degree as compared with children with parents born in other countries. Among parents born in Spain and Greece, there were too few parents with PhD degrees in the data. The effect of university-educated parents’ relative income is negative for individuals with parents born in Iran and Asia, but insignificant. When running the model for only those with mothers as the representative parent, the negative odds ratio for relative income of the parent having a post-elementary education becomes significant for those with parents born in Asia (the number of mothers born in the other three High-odds countries is too small to be run in separate models). The impact of university-educated parents’ rela- tive income is significantly positive for individuals with parents born in Spain. However, it seems clear that the impact of parents’ income differ among those with parents born in High-odds countries as compared with the general pattern. 936 INTERNATIONAL MIGRATION REVIEW TABLE 6 LOGISTIC REGRESSION. ODDS RATIO FOR UNIVERSITY EDUCATION, ROBUST STANDARD ERROR BY PARENTS’ COUNTRY OF BIRTH, SEPARATE MODELS FOR IRAN, GREECE, ASIA, AND SPAIN 1 2 3 4 Iran, N = 886 Greece, N = 643 Asia, N = 1144 Spain, N = 225 Backgrounda Parents education OR SE OR SE OR SE OR SE Post-elementary 1.82 0.69 0.82 0.34 2.14 0.83 0.56 0.50 school 2 years Post-elementary 2.03 0.75 1.25 0.55 2.71 1.13 0.62 0.59 school 3 years Post-high school 3.79 1.51 2.28 1.68 9.26 6.00 0.36 0.41 (university) <3 years Post-high school 5.67 2.17 7.11 5.28 18.51 13.92 0.72 0.81 (university) >3 years PhD 8.16 6.05 (omitted) 32.33 (omitted) Parents income Parents relative income 1.03 0.18 0.91 0.21 0.92 0.14 0.80 0.30 education Relative income/ 0.89 0.24 1.69 0.62 0.81 0.34 1.65 1.25 post-elementary education Relative income/ 0.64 0.21 1.10 0.77 0.43 0.29 6.22 6.09 university education Pseudo R2 0.0588 0.0667 0.1379 0.1118 Note: Odds ratios in bold: p > |z| < 0.05. Omitted categories for the independent variables are as follows: male, year of finishing ninth grade 1990, parents married, parent born in Sweden, outside metropolitan area residential, parents education: elementary level or less. a Female, year of finishing ninth grade 1991, year of finishing ninth grade 1992, cohabiting, single father, single mother, Other family categories, Metropolitan residential, age when immigrated 1–6 years, 7–10 years, 11 years or older. CONCLUDING DISCUSSION The results reported in this study confirm previous studies on existing inequalities in the labor market. Parental educational level still has a large impact on their children’s educational paths. Most groups of immigrants have lower mean earnings compared with means among natives with the same sex, age, level, and type of education. On the positive side, this study provides some empirical support for the exis- tence of a compensation strategy; children of certain groups of migrants pursue intergenerational educational careers to a larger degree, compared with children with parents born in Sweden. Parents’ level of education and relative income is in general positively associ- ated with higher odds of having university education; however, some immigrant groups show the reverse pattern of the impact of parents’ EDUCATION AS A COMPENSATION STRATEGY 937 income. We also found that average relative income among the parents’ immigrant group is negatively associated with having a uni- versity education. This results support the hypothesis on a compensa- tion strategy developed in groups with low level of integration in the labor market. In particular, after controlling for a range of background variables, individuals with parents born in Iran, Spain, Greece, and Asia have higher odds ratios of obtaining a university education 14 years after finishing ele- mentary school. It is especially the income from mothers and parents with high education that has a negative impact on the odds ratio for university education. This supports the hypothesis for a compensation strategy also on an individual level. With support from qualitative studies conducted in this field, the findings support the existence of group as well as individual compensation strategy. The alternative explanations proposed by Similä, namely the imprecise measures of educational level, selective migration, selective return migration among Nordic migrants, and diverse school treatment, are also relevant in this study. However, the measurement problem is reduced because parent’s earnings are related to the level as well as the type of education in a very detailed manner. The selective migration and return migration expla- nations might be relevant, but they do not reduce the importance of a compensation strategy among the migrants that remain in the country. The result reminds us that “societies are not made up of passive sub- jects resigned to structural domination” (Castells, 1996:495). Individuals and groups develop strategies and agentic practices to counteract discrimi- natory structures in society. REFERENCES Arai, M., L. Schröder, and R. Vilhelmsson. 2000 En Svartvit Arbetsmarknad? – En ESO-Rapport om Vägen Från Skola Till Arbete. Stockholm: Fritzes offentliga publikationer. Behtoui, A. 2006 Unequal Opportunities. The impact of social capital and recruitment methods on immigrants and their children in the Swedish labour market. Linköping: Linköpings universitet. Blau, P., and O. D. Duncan. 1967 The American Occupational Structure. New York: John Wiley & Sons. 938 INTERNATIONAL MIGRATION REVIEW Borjas, G. J. 1992 “Ethnic Capital and Intergenerational Mobility.” The Quarterly Journal of Economics 107:123–150. Boudon, R. 1974 Education, Opportunity & Social Inequality. New York: John Wiley & Sons. Broady, D. et al. 2000 Välfärd och Skola. Stockholm: Socialdepartementet. Castells, M. 1996 The Rise of the Network Society. Oxford: Blackwell. Chiswick, B. R. and P. W. Miller 2008 “Why is the Payoff to Schooling Smaller for Immigrants?” Labour Economics 15:1317–1340. Dryler, H. 2001 “Etnisk segregation i skolan: effekter på ungdomars betyg och övergång till gymna- sieskolan.” In Välfärdens finansiering och fördelning: antologin från kommitténs välfärdsbokslut. Ed. J. Fritzell and J. Palme. Stockholm: Fritzes. Pp. 319–355. Duncan, J. G. J., and S. D. Hoffman 1981 “The Incidence and Wage Effects of Overeducation.” Economics of Education Review 1:75–86. Dustmann, C. 2008 “Return Migration, Investment in Children, and Intergenerational Mobility: Com- paring Sons of Foreign- and Native-Born Fathers.” Journal of Human Resources 43:299–324. Erikson, R., and J. O. Jonsson 1994 Sorteringen i Skolan. Studier av snedrekrytering och utbildningens konsekvenser. Stockholm: Carlssons. Eriksson, R., and J. O. Jonsson ed. 1996 Can Education be Equalized? Boulder: Westview Press. Fejgin, N. 1995 “Factors Contributing to the Academic Excellence of American Jewish and Asian Students.” Sociology of Education 68:18–30. Heath, A. F., C. Rothon, and E. Kilpi 2008 “The Second Generation in Western Europe: Education, Unemployment, and Occupational Attainment.” Annual Review of Sociology 34:211–235. Jonsson, J. O., and F. Rudolphi. 2011 “Weak Performance—Strong Determination: School Achievement and Educational Choice Among Children of Immigrants in Sweden.” European Sociological Review 27:487–508. Knocke, W., and F. Hertzberg 2000 Mångfaldens Barn Söker sin Plats. En studie om arbetsmarknadschanser för ungdo- mar med invandrarbakgrund. Stockholm: Svartvitts förlag. Lindblad, S. 1994 “Skolkarriär och levnadsbana. Om elevers erfarenheter av ungdomsskola och rekry- tering till högre studier.” In Sorteringen i Skolan. Ed. R. Erikson and J. O. Jonsson. Stockholm: Carlssons. Pp. 172–225. Ljung, M. 2000 “Utbildning i broschyrspråkets värld.” In Mångfald i Högskolan. Utredningen om Social och Etnisk Mångfald i Högskolan. Ed. Utredningen om social och etnisk mån- gfald i högskolan. Stockholm: Fritzes offentliga publikationer. Pp. 211–230. EDUCATION AS A COMPENSATION STRATEGY 939 Lundqvist, C. 2010 Möjligheternas Horisont. Linköping: Linköping university. Nekby, L., R. Vilhelmsson, and G. Özcan 2008 “Do Host Country Educations Even Out the Playing Field? Immigrant-Native Labor Market Gaps in Sweden.” Journal of Immigrant and Refugee Studies 6:168–196. Nordenstam, K., and I. Wallin 2002 Osynliga Flickor – Synliga Pojkar. Om ungdomar med svenska som andra språk. Lund: Studentlitteratur. Nordström Skans, O., and O. Åslund 2009 Segregationen i Storstäderna. Stockholm: Studieförbundet Näringsliv och Samhälle. Ogby, J. U. 1987 “Variability in Minority School Performance.” Anthtropology and Education Quar- terly 25:312–334. ——— 1991 “Immigrant and Involuntary Minorities in Comparative perspective.” In Minority Status and Schooling: A Comparatative Perspective. Ed. M. A. Gibson and J. U. Ogby. New York: New York Press. Pp. 3–33. Österberg, T. 2000 Economic Perspectives on Immigrants and Intergenerational Transmissions. Göteborg: Handelshögskolan. Portes, P. 1999 “Social and Psychological Factors in the Academic Achievements of Children of Immigrants: A Cultural Puzzle.” American Research Educational Journal 36:489–507. Portes, A., and D. MacLeod 1996 “Educational Progress of Children of Immigrants: The Roles of Class, Ethnicity, and School Context.” Sociology of Education 69:255–275. ———, and R. Rumbaut 1996 Immigrant America. Berkely: University of California Press. ———, and M. Zhou 1993 “The New Second Generation: Segmented Assimilation and its Variants Among Post-1965 Immigrant Youth.” Annals of the American Academy of Political and Social Science 530:74–96. Rauhut, D. 2007 Vägen Till Sverige: Om Integrationsarbete i Stockholm. Stockholm: Forsknings- och utvecklingsenheten. Rubb, S. 2003 “Overeducation in the Labour Market: A Comment and re-Analyses of a Meta- Analysis.” Economics of Education Review 22:621–629. Schmid, C. L. 2001 “Educational Achievement, Language-Minority Students, and the New Second Gen- eration.” Sociology of Education 74:71–87. Shierup, C., P. Hansen, and S. Castles 2006 Migration, Citizenship and the European Welfare State. An European Dilemma. Oxford: Oxford University Press. Similä, M. 1994 “Andra generationens invandrare i den svenska skolan.” In Sorteringen i Skolan. Studier av Snedrekrytering och Utbildningens Konsekvenser. Ed. R. Eriksson and J. O. Jonsson. Stockholm: Carlssons. Pp. 226–263. Socialstyrelsen. 2010 Social Rapport 2010. Stockholm: Socialstyrelsen. 940 INTERNATIONAL MIGRATION REVIEW Urban, S. 2008a “Stockholms etniska delning. Polariseringen på arbets- och bostadsmarknaden.” In Den Kalla och Varma Staden. Migration och Stadsförändring i Stockholm Efter 1970. Ed. H. Forsell. Stockholm: Stockholmia förlag. Pp. 28–56. ——— 2008b “Polarisering i Stockholm – på gränsen till postfordism.” In Sveriges Sociologförbunds Årsmöteskonferens 2008. Östersund. Waters, M. C. 1999 Black Identities: West Indian Immigrant Dreams and American Realities. MA New York: Harvard University Press Russel Sage Foundation. White, K. R. 1982 “The Relationship Between Socioeconomic Status and Academic Success.” Psychological Bulletin 91:461–481. Zhou, M. 1997 “Segmented Assimilation: Issues, Controversies, And Recent Research on the new Second Generation.” International Migration Review 31:975–1008. ———, and C. L. Bankston 1998 Growing up American: How Vietnamese Children Adapt to Life in the United States. New York: Russel Sage Foundation.