THE ECONOMIC PERFORMANCE OF IMMIGRANTS
The Role of Human and Social Capital

AGNIESZKA KANAS

Kanas, A.M. The Economic Performance of Immigrants. The Role of Human and Social Capital Dissertation, Utrecht University, The Netherlands Cover illustration: Krzysztof Wodiczko, Goscie/Guests, 2009, instalacja wideo/video installation, 17,17 min./minutes. Dzieki uprzejmosci artysty i Fundacji Profile/courtesy of the artist and Profile Foundation, Warsaw. Cover design: Agnieszka Kanas & Sebastian Gryglewicz Printed by: Wöhrmann Print Service ISBN: 978-90-393-5550-3 © Agnieszka Kanas, 2011 All Rights reserved. No part of this publication may be reproduced, translated, stored in a retrival system of any nature, or transmitted, in any form or by any means, electrnically, mechanically, by photocopying, microfilming, recording, or otherwise, without the prior written permission from the author.

THE ECONOMIC PERFORMANCE OF IMMIGRANTS The Role of Human and Social Capital

DE ECONOMISCHE POSITIE VAN IMMIGRANTEN De rol van menselijk en sociaal kapitaal

(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof.dr. G.J. van der Zwaan, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op dinsdag 28 juni 2011 des middags te 2.30 uur door Agnieszka Małgorzata Kanas geboren op 3 februari 1980 te Trzcianka, Polen

Promotoren:

Prof. dr. F.A. van Tubergen Prof. dr. ir. A.G. van der Lippe

Manuscript committee:

Prof. dr. H. Flap (Utrecht University) Prof. dr. F. Kalter (Mannheim University) Prof. dr. K. Phalet (Catholic University of Leuven) Prof. dr. M. Verkuyten (Utrecht University) Prof. dr. H.G. van de Werfhorst (University of Amsterdam)

To my parents

Table of Contents

1

Introduction ...........................................................................................................................................1

1.1 1.2 1.3 1.4 1.5 1.5.1 1.5.2 1.6 2

Background ............................................................................................................................. 1 Research questions and contributions ..................................................................................... 4 Economic performance............................................................................................................ 8 The Netherlands and Germany .............................................................................................. 10 Data sources .......................................................................................................................... 11 Dutch data...................................................................................................................... 11 German data .................................................................................................................. 12 Outline of the study ............................................................................................................... 12

The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants ...15

2.1 2.2 2.3 2.4 2.4.1 2.4.2 2.5 2.5.1 2.5.2 2.6

Introduction ........................................................................................................................... 15 Dutch setting.......................................................................................................................... 18 Theory and hypotheses .......................................................................................................... 18 Data and methods .................................................................................................................. 22 Dependent and independent variables ........................................................................... 23 Methods ......................................................................................................................... 26 Results ................................................................................................................................... 27 Human capital................................................................................................................ 27 Host-country schooling and social contacts with natives .............................................. 32 Conclusions and discussion ................................................................................................... 33

3 Immigrant Self-Employment: Testing Hypotheses about the Role of Origin and Host Country Human Capital and Bonding and Bridging Social Capital ..........................................................................35

3.1 3.2 3.2.1 3.2.2 3.2.3 3.3 3.3.1 3.3.2 3.3.3 3.4 3.4.1 3.4.2 3.4.3 3.5

Introduction ........................................................................................................................... 35 Theories and hypotheses ....................................................................................................... 38 Human capital theory .................................................................................................... 38 Social capital theory ...................................................................................................... 40 Social contacts with natives and host-country human capital ....................................... 41 Data and methods .................................................................................................................. 42 Dependent and independent variables ........................................................................... 44 Job-skill level ................................................................................................................ 45 Method........................................................................................................................... 47 Results ................................................................................................................................... 47 Human capital................................................................................................................ 50 Social capital ................................................................................................................. 50 Model comparison ......................................................................................................... 51 Conclusions and discussion ................................................................................................... 52

4 The Role of Social Contacts in the Employment Status of Immigrants: A Panel Study of Immigrants in Germany...............................................................................................................................55

4.1 4.2 4.2.1 4.2.2 4.3 4.3.1 4.3.2 4.4 4.4.1 4.4.2

Introduction ........................................................................................................................... 55 Theory and hypotheses .......................................................................................................... 57 Social contacts ............................................................................................................... 57 Social contacts with natives and host-country human capital ....................................... 58 Data and methods .................................................................................................................. 61 Dependent and independent variables ........................................................................... 62 Methods ......................................................................................................................... 65 Results ................................................................................................................................... 67 Social contacts ............................................................................................................... 67 Social contacts with natives and host-country human capital ....................................... 70

x

4.5

Conclusions and discussion ................................................................................................... 72

5 Social Contacts and the Economic Performance of Immigrants: A Panel Study of Immigrants in Germany ......................................................................................................................................................75

5.1 5.2 5.2.1 5.2.2 5.3 5.3.1 5.3.2 5.4 5.4.1 5.4.2 5.5 6

Introduction ........................................................................................................................... 75 Theory and hypotheses .......................................................................................................... 78 Social contacts ............................................................................................................... 78 Co-ethnic concentration and origin- and host-country human capital........................... 79 Data and methods .................................................................................................................. 80 Dependent and independent variables ........................................................................... 81 Method........................................................................................................................... 84 Results ................................................................................................................................... 85 Social contacts ............................................................................................................... 85 Social contacts and origin- and host-country human capital ......................................... 89 Conclusions and discussion ................................................................................................... 91

Conclusions and Discussion ................................................................................................................94

6.1 6.2 6.2.1 6.2.2 6.2.3 6.3 6.3.1 6.3.2 6.3.3 6.3.4

Research questions and contributions ................................................................................... 94 Answers to the research questions......................................................................................... 96 Origin- and host-country human capital ........................................................................ 96 Origin- and host-country social capital ......................................................................... 98 Interplay between origin- and host-country human and social capital ........................ 100 Suggestions for future research ........................................................................................... 102 The role of quality, transferability and employer’s uncertainty .................................. 102 The measures of social capital..................................................................................... 103 Longitudinal data on recent immigrants ...................................................................... 103 Immigrant women ....................................................................................................... 104

xi

6.4

The role of human and social capital................................................................................... 104

Appendix ...................................................................................................................................................107

Samenvatting (summary in Dutch)............................................................................................................109

References .................................................................................................................................................117

Acknowledgments .....................................................................................................................................131

Curriculum Vitae .......................................................................................................................................133

ICS Dissertation series ..............................................................................................................................135

xii

1 Introduction

1.1 Background
In the context of increasing migration to Western countries, immigrants’ economic disadvantage has become an important issue in policymaking and scientific inquiry. Sociologists, economists and social demographers have studied the economic difficulties the foreign-born population experiences in great detail (e.g., Borjas 1994; Portes and Rumbaut 1996; Van Tubergen, Maas and Flap 2004). Empirical evidence overwhelmingly shows that immigrants are more likely to be unemployed than are natives; if they have jobs, they often have jobs with lower prestige and lower earnings (Alba and Nee 2003; Borjas 1994; Hall and Farkas 2008; Kogan 2006). In this study, I focus on two important explanations for immigrants’ economic disadvantage provided by human capital theory and social capital theory. According to human capital theory, immigrants’ disadvantage can be explained by their skills and knowledge. It is argued that, on average, immigrants have fewer skills and are less productive in the labor market than the native-born population. In particular, immigrants are less educated than natives, making them less productive and less attractive to employers. Indeed, empirical studies show that immigrants are generally less educated, and that this difference partly explains their disadvantaged position in the labor market (e.g., Borjas 1994; Portes and Rumbaut 1996). An intriguing question remains, however, in studies relying on the human capital approach: why do immigrants still perform worse in the labor market –compared to natives–, while taking education into account? The ‘ethnic residual’ remains even after including other determinants of economic attainment that are informed by the human capital theory (e.g., labor market experience, gender, age, health). Why are immigrants more often unemployed and occupy lower quality jobs than natives, even when they have the same or similar levels of education, work experience, gender, age, and health? Researchers have extended the human capital theory to further explain the ethnic disadvantage (Bratsberg and Ragan 2002; Duleep and Regets 1999; Friedberg 2000; Li 2001; Zeng and Xie 2004). It is argued that immigrants face particular problems in the labor market, because they lack so-called host-country specific skills that natives possess and that are required in the host-country labor market (Duleep and Regets 1999; Van Tubergen and Van

Chapter 1

de Werfhorst 2007). This is most evident for language. In many Western countries, a large part of the immigrant population has a mother tongue different from the host country’s official language, and many immigrants do not speak the host-country language well (Espenshade and Fu 1997; Van Tubergen and Kalmijn 2005). Although immigrants are skilled in their mother tongue, such origin-specific skills are clearly less valued in the hostcountry labor market. By contrast, immigrants who acquire the host-country language significantly improve their economic opportunities (e.g., Chiswick and Miller 1995, 2002). For example, Chiswick and Miller (2002), using the 1990 Census data on immigrants in the United States, showed that keeping other characteristics constant, immigrants who speak English fluently have 14 percent higher earnings than those who are not fluent in English. Besides language, researchers have argued that other forms of host-country specific skills are important as well (Chiswick 1978; Bratsberg and Ragan 2002; Friedberg 2000; Li 2001; Zeng and Xie 2004). More specifically, it is suggested that the educational qualifications and work experience immigrants have acquired in their country of origin are less valued in the host-country labor market than schooling and experience obtained in the host country. It is furthermore assumed that many immigrants do obtain some education and work experience after migration, thereby improving their host-country skills (Van Tubergen and Van de Werfhorst 2007). However, besides the empirical evidence supporting the importance of hostcountry language skills, there is little empirical research on the role of origin- and hostcountry schooling and work experience in immigrant economic performance. A second explanation for the “ethnic residual” is provided by the social capital theory. A longstanding and influential line of research in sociology and economics has been developed that considers the impact of social capital on people’s economic attainment (Coleman 1990). The theoretical importance of social capital has long been recognized in the literature on the native-born population (Burt 1992; Coleman 1990; Granovetter 1973; Ioannides and Loury 2004; Lin 1999; Mouw 2002). The major idea stipulates that having more (resourceful) contacts generally increases people’s economic opportunities. A number of studies have examined the importance of co-ethnic contacts in immigrant economic outcomes (e.g., Aguilera 2005; Aguilera and Massey 2003; Sanders, Nee and Sernau 2002). Immigrants could benefit from social contact with co-ethnics as it provides immigrants with host-country specific information and knowledge, for example, about where to look for a job, what the available jobs are, and how to behave on the job interview (Aguilera and Massey 2003; Fernandez-Kelly 1995). In addition, social contacts can assist immigrants in the job-search process by providing recommendations to prospective employers or entry into work positions (Aguilera and Massey 2003; Hagan 1998; Sanders, Nee and Sernau 2002). There have been also a number of studies that considered the role of ethnic

2

Introduction

enclaves in immigrant economic outcomes (Hagan 1998; Portes and Jensen 1989; Sanders, Nee and Sernau 2002; Waldinger, Aldrich and Ward 2006). These studies have consistently shown that immigrants tend to settle in areas with large concentrations of immigrants (Bauer, Epstein, and Gang 2005; Chavez, Mouw and Hagan 2009), and that these ethnic concentrations can be an important predictor of immigrants’ entry and advancement in the host-country labor market (Hagan 1998; Portes and Jensen 1989; Waldinger, Aldrich and Ward 2006). Specifically, some authors showed that ethnic concentration is associated with higher employment chances and better jobs for immigrants (e.g., Hagan 1998; Portes and Jensen 1989); others showed that economic opportunities provided by ethnic concentrations are merely limited to low-skilled and low-paid jobs (Catanzarite and Aguilera 2002; LewinEpstein and Semyonov 1992). In the research of Putnam (2000) a distinction is made between bonding and bridging social capital. According to this research, connections within one’s own ethnic group are considered a form of bonding social capital, whereas connections with natives are a form of bridging social capital. It is also argued that whereas bonding social capital is crucial for mobilizing in-group solidarity and reciprocal relations, bridging social capital is critical for linkages to external resources and information diffusion. Following Putnam (2000), I distinguish between bonding social capital (which I refer to as origin-country specific social capital) and bridging social capital (which I refer to as host-country specific social capital). Although co-ethnic contacts generally facilitate immigrants’ economic performance, I expect that interacting with natives may be particularly important for immigrant economic outcomes. One reason for this expectation is that natives have had more exposure to the host-country labor market than have immigrants. Another reason is that natives are generally more educated, more often employed, they have higher status jobs and more earnings than immigrants. For these reasons, one may expect natives to possess more information about available jobs and to be better informed on how to apply for jobs. However, previous studies have almost exclusively focused on origin-country specific social capital, and little empirical evidence exists for the role of host-country social capital in immigrant economic outcomes.1

Although I mainly focused on human and social capital, other explanations can be offered to explain the economic performance of immigrants, one of them having to do with racial discrimination (National Research Council 2004; Quillian 2006). There is ample evidence that discrimination hampers the economic mobility of immigrants in the labor market, particularly those from culturally different backgrounds (Van Tubergen 2006). In this study, discrimination could explain why I consistently find that Turkish immigrants in the Netherlands and Germany, the populations most socially and culturally distinct from the native-born population, are more often unemployed, have lower occupational statuses and higher rates of self-employment than other immigrants, even after accounting for origin- and host-country specific human capital, social contacts with natives, and after controlling for multiple demographic variables.

1

3

Chapter 1

1.2 Research questions and contributions
In this study, I assess the role of origin- and host-country human and social capital in the economic performance of immigrants in two Western countries: the Netherlands and Germany. I attempt to answer three general research questions. In answering these questions, I make several contributions to the literature. The first research question is the following: 1. To what extent and why do origin-country specific human capital and host-country specific human capital affect the economic performance of immigrants? In answering this question, I contribute to the literature by studying the role of place of education and work experience in the economic incorporation of immigrants in the Netherlands and Germany. Despite much theorizing about the presumed impact of hostcountry specific skills, little is known empirically. There are few studies that have examined the importance of origin- and host-country schooling, and even less is known about the returns to origin- and host-country work experience. What is even more important is that these studies were exclusively conducted among immigrants in the United States (Akresh 2007; Bratsberg and Ragan 2002; Bratsberg and Terrell 2002; Zeng and Xie 2004), Canada (Li 2001), and Israel (Friedberg 2000), thus, little is known whether the same findings would hold equally in new receiving Western European countries. Second, I assess how the economic returns to origin-country human capital differ between immigrant groups in the Netherlands. Specifically, I examine whether education and work experience obtained in Turkey and Morocco have a smaller effect on immigrant economic outcomes than do education and work experience received in Suriname and Dutch Antilles. Most previous research has not considered immigrant group differences in the returns to origin-country human capital, and therefore, provide little insights into why such differences could occur. However, a few studies have examined the returns to origin-country schooling among different immigrants in the United States (Bratsberg and Ragan 2002; Bratsberg and Terrell 2002; Zeng and Xie 2004) and Israel (Friedberg 2000). Third, I advance previous research methodologically by using better measures of originand host-country human capital. Most of the previous studies examining the returns to originand host-country human capital relied on general population surveys (e.g., census data) that do not directly measure pre- and post-migration schooling (and work experience), but instead use information on age at the time of migration and total years of schooling (Bratsberg and Ragan 2002; Friedberg 2000; Zeng and Xie 2004). Such measures may lead to systematic measurement error of both pre-migration schooling and post-migration schooling (Chiswick

4

Introduction

and Miller 1994).2 In this study, I rely on immigrant specific surveys, which contain direct measures of pre- and post-migration schooling, language proficiency, and more direct measures of work experience. The second research question addressed in this study is the following: 2. To what extent and why do origin-country specific social capital and host-country specific social capital affect the economic performance of immigrants? In answering this question, I improve on previous research by focusing on not only social contacts with co-ethnics, but also social contacts with natives. Previous research almost exclusively focused on origin-specific social capital. Specifically, a number of studies have considered the importance of having co-ethnic family and friends for immigrant economic integration into the host-country (Aguilera 2005; Aguilera and Massey 2003; Sanders, Nee and Sernau 2002). These studies have shown, by providing immigrants with information and social and financial assistance, that co-ethnic contacts promote the economic performance of immigrants (e.g., Aguilera 2005; Aguilera and Massey 2003; Sanders, Nee and Sernau 2002). Although contacts with co-ethnics generally foster economic mobility of immigrants, it can be argued that contacts with natives, in particular, improve immigrants’ economic performance. It can be assumed that in the host-country labor market natives have access to more and better resources than immigrants (e.g., have more and better information about available jobs and application procedures). Moreover, several studies throughout the literature examining the native-born population have suggested that for acquiring novel information about labor market opportunities and influence, ties outside of one’s own social network are crucial (Burt 1992; Granovetter 1973; Putnam 2000). Following the insights from these studies, it can be argued that for immigrants, contact with natives is especially important. In this study, I improve upon earlier research by relying on surveys, which also contain information on social contacts with natives. The second contribution regarding the social capital theory is the use of panel data. Previous empirical studies on immigrants have exclusively relied on cross-sectional data. The use of cross-sectional data is problematic for assessing the effect of social capital and testing the hypotheses rigorously. That is, cross-sectional data do not allow examining whether the presumed positive effect of social contacts on immigrant economic outcomes is causal, reflects reversed causality, or is spurious due to unobserved characteristics of immigrants.

2

I discuss the importance of the possible measurement error in detail in chapter 2 on pp. 22-23.

5

Chapter 1

Figure 1.1 outlines the possible causal, reverse causal, mediated and spurious relations between social contacts and immigrants’ economic outcomes. Although researchers have argued that (resourceful) social contacts help immigrants get (better) jobs (as depicted by arrow A in Figure 1.1), it could be that immigrants who have a better job have more opportunities to make (resourceful) social connections (arrow B). Because having a job (or higher status job) provides opportunities to meet and socialize with resourceful people, a positive correlation between immigrant social contacts and economic outcomes could reflect the tendency that (better) employed immigrants acquire more resourceful social contacts. By using longitudinal data and measuring immigrant social contacts earlier than their economic outcomes, this study provides a better test of a causal relationship between social contacts and immigrants’ economic performance. Furthermore, this study examines whether a positive correlation between social contacts and immigrant economic outcomes could be spurious due to unmeasured time-constant immigrant characteristics. Because people tend to associate with people that are similar to
Figure 1.1 Causal, reverse causal, mediated and spurious effects of host-country human capital and origin- and host-country social capital on immigrant economic outcomes.

Host-country human capital

E F G

A
Origin- and host-country social capital Economic outcomes

B

C D

Unobserved, time-constant characteristics of immigrants e.g., ambition, talents

6

Introduction

themselves (McPherson, Smith-Lovin, and Cook 2001), a positive correlation between social contacts and immigrant economic outcomes could simply reflect the tendency of ambitious and talented immigrants to befriends other ambitious and talented immigrants (or natives) (arrow C) and have better economic outcomes (arrow D) even in the absence of a causal effect of social contacts. By making use of longitudinal data and by following the same individuals over time, this study provides a more rigorous test for a causal effect of social capital. The third research question follows: 3. How do origin- and host-country specific human capital and origin- and host-country specific social capital interplay in the economic performance of immigrants? In answering this question, I make two theoretical contributions. First, I study whether the positive effect of host-country specific human capital on immigrant economic outcomes is mediated or spurious due to host-country specific social capital. Figure 1.1 describes possible scenarios in the relationship between host-country specific human capital, social contacts with natives and immigrant economic outcomes. According to the standard interpretation, hostcountry specific human capital is more valued by employers because skills acquired in the host-country are of higher quality and are more transferable than skills acquired in the country of origin (e.g., Friedberg 2000), (arrow E, in Figure 1.1). In this study, I also examine an alternative interpretation of the positive effect of host-country human capital. One could argue that immigrants with more host-country specific skills benefit from such skills because they are associated with increasing connections with the native-born population. People who learn the second language, and who attend school and work in the host country are more likely to develop relationships with natives (e.g., at school, at work) (arrow F) that promote their economic opportunities (arrow A). In this scenario, higher returns to host-country specific human capital are mediated: investments in such skills lead to better economic outcomes because of their association with increased connections with natives. In a more extreme scenario, the higher returns to host-country human capital are spurious due to host-country specific social capital. Specifically, social contacts with natives can lead to both post-migration investments in human capital (arrow G) and better economic opportunities (arrow A), not that post-migration investments in human capital have an effect on immigrants’ economic outcomes. For example, one could argue that immigrants who marry a native spouse and develop relationships with native friends are more likely to learn the official language, to attend school and to get a (better) job.

7

Chapter 1

Similarly, one could argue that a presumed positive correlation between contacts with natives and immigrant economic outcomes is mediated or spurious due to post-migration investments in human capital. For example, in an extreme scenario, immigrants who learn the language of the host country and attend school in the host country are more likely to find a job (arrow E), but, at the same time, these post-migration investments in human capital could help immigrants to develop social connections with natives (arrow F). By studying the effects of host-country specific human capital and host-country specific social capital at the same time, this study deepens the understanding of the relationship between human and social capital in explaining the economic performance of immigrants. Second, beyond studying mediation and spuriousness, I contribute to earlier research by theorizing how the returns to origin- and host-country specific human capital are moderated by origin-country specific social capital. Specifically, I argue that because co-ethnic concentration facilitates the transferability and reduces employers’ uncertainty towards origincountry skills, immigrants living among many co-ethnics receive larger returns to their origincountry human capital than those living among few co-ethnics. By providing immigrants with jobs where host-country credentials and skills are not important or not necessary, co-ethnic concentration can also reduce the economic returns to host-country human capital. Although several studies have examined the direct impact of ethnic concentration in the region of living on immigrant economic outcomes (e.g., Chiswick and Miller 2002, 2005; Kogan and Kalter 2006; Lewin-Epstein and Semyonov 1992; Tienda and Lii 1987; Tolnay 2001), little is known about the interaction between ethnic concentration and origin- and hostcountry specific human capital. The exception is a study by Chiswick and Miller (2002) that examined the effect of linguistic concentration on the investments in host-country language skills and the economic returns to such skills among immigrants in the United States. In this study, I provide further insights into the relationship between co-ethnic concentration in the region of living and immigrants’ economic outcomes, by comparing the economic returns to origin- and host-country schooling and work experience in regions with different co-ethnic concentrations.

1.3 Economic performance
In this study I focus on four different aspects of immigrants’ economic performance: employment, self-employment, occupational status and income. Previous research on the role of origin- and host-country specific human capital has focused exclusively on immigrant earnings, and little is known empirically about other economic outcomes, that is, employment and occupational status. It could be, for example, that post-migration investments in human

8

Introduction

capital are less important for finding a job, than for having a higher status job or higher income. Likewise, most studies on immigrant social contacts have focused on a single economic outcome. Several studies showed, for example, that living among many co-ethnics facilitates immigrant employment (e.g., Hagan 1998; Portes and Jensen 1989; Waldinger et al. 2006), but it is less beneficial, or even detrimental to having a higher-status or better-paid job (e.g., Catanzarite and Aguilera 2002; Tienda and Lii 1987). The results from these studies suggest that the effect of ethnic concentration may depend on the economic outcome under study. It could also be that contradictory findings for the role of ethnic concentration originate from differences in immigrant populations, discrepancies in the measurement of social capital, or unobserved contextual factors. By replicating the analysis for different types of outcomes, I not only assess whether and how human and social capital resources affect immigrants’ employment and job type, but also examine the extent to which the effects of these resources are generalizable across outcomes. In addition, by examining occupational status and income, this study sheds light on aspects of immigrants’ economic integration that cannot be clearly observed when focusing only on employment status. Specifically, studies on occupational status (or income) provide an indication of the returns to each year of schooling or work experience, and thereby, of the incentives to invest in these forms of human capital. More generally, the expected occupational status and income provide incentives for immigrants to participate in the hostcountry labor market (OECD 2008). Previous studies on the role of self-employment in the economic integration of immigrants have been mixed. On the one hand, researchers have argued, and indeed showed, that problems with transferability and employers’ uncertainty towards credentials and skills acquired in the country of origin push immigrants into self-employment (Bates 1997). In that sense, self-employment was considered as the second-best option for those immigrants who are at risk of unemployment or poverty. On the other hand, however, researchers showed that host-country human capital increases the managerial knowledge and skills needed to deal with host-country institutions, and thus, pulls immigrants towards self-employment (Constant and Zimmerman 2006; Le 2000). According to this argument, self-employment provides a route to upward mobility, facilitating the opportunities for higher-status and better-paying jobs. By studying how human and social capital acquired in the country of origin and destination influence self-employment, this study provides further insights into the discussion about the role of self-employment in the economic integration of immigrants.

9

Chapter 1

1.4 The Netherlands and Germany
I examine the economic performance of immigrants in two Western countries: the Netherlands and Germany. There are several similarities between the two countries with respect to migration history, organization of the labor market, and the economic standing of immigrants. The Netherlands and Germany are recent immigrant countries with a substantial foreign-born population, 10 and 8.8 percent, respectively (Federal Statistics Office 2007; Statistics Netherlands 2010). In the 1960s, the two countries admitted large numbers of lowskilled guest worker immigrants. The guest workers in Germany came from Italy, Spain, Greece, Turkey, Morocco, Portugal, Tunisia, and Yugoslavia (Kogan 2007), whereas in the Netherlands they mainly came from Spain, Yugoslavia, Morocco, and Turkey (Hagendoorn, Veenman and Vollebergh 2003). Although in both countries immigrant guest workers were expected to stay temporarily, in most cases, they decided to stay and brought their families. In addition to immigrant guest workers, starting in the late 1980s, Germany accepted a number of refugees and asylum seekers, mainly from Yugoslavia, and ethnic Germans from Eastern Europe (Kogan 2007). Beginning in the 1950s many immigrants to the Netherlands also came from former colonies of the Netherlands, Suriname and the Dutch Antilles, thus migration is common from these countries (Hagendoorn, Veenman and Vollebergh 2003). The two countries are also similar regarding the organization of the labor market, which is characterized by high segmentation (i.e., low permeability of the labor market between unemployment and employment as well as different types of employment), expensive lowskilled labor and a relatively long maximum duration of unemployment benefits (Empter 2002). In both countries guest worker immigrants, in particular those with Turkish backgrounds, tend to have a higher unemployment rate than natives, and they are overrepresented in low-status, and low-paid jobs (Hagendoorn, Veenman and Vollebergh 2003; Kalter and Granato 2002; Kogan 2007). An important difference between the two countries is that they adapted different integration policies. Specifically, whereas German integration policies were rather minimal focusing on return migration and restricting access to German nationality, the Netherlands promoted immigrants access to the host-country institutions (e.g., housing, education, labor market), offered easy access to Dutch nationality, and at least until 2001, encouraged immigrants to preserve their own cultural identity (Euwals, Dagevos, Gijsberts, Roodenburg 2007). By focusing on the two countries I show whether the results for human and social capital can be replicated across two national contexts, and thus, provide a stronger evidence for the importance of human and social capital for immigrant economic performance. As argued by

10

Introduction

Firebaugh (2008), repeating the analysis for different contexts and new populations allows for testing of the robustness of findings, and should become a common strategy in future research. I also focus on the two countries because there are large-scale, high-quality data available on the immigrant population; these data contain relevant information on the research questions studied here, i.e., direct measures of origin- and host-country specific schooling and immigrant contacts with natives. Both datasets also complement each other in an important way. Specifically, whereas the Dutch survey provides a great opportunity for testing the effect of origin- and host-country human capital on the economic performance, its cross-sectional design makes it difficult to assess the causal effect of social capital. On the other hand, the German survey includes longitudinal information on immigrant social contacts, including connections with natives, which provides a better possibility for testing the effect of social capital.

1.5 Data sources
1.5.1 Dutch data

In chapters 2 and 3, I examine the economic performance of Turkish, Moroccan, Surinamese and Dutch Antillean immigrants in the Netherlands. I make use of a large-scale, crosssectional survey: the Social Position and Use of Welfare Facilities by Immigrants survey (SPVA 1991, 1994, 1998 and 2002). The SPVA survey was administrated by the Institute for Socio-Economic Research (ISEO) and from 1998 was conducted in collaboration with the Netherlands Institute for Social Research (SCP) (De Koning and Gijsberts 2002; Martens 1994; Martens and Tesser 1998; Martens and Veenman 1991; Veenman 1988). The survey was designed to monitor the socio-economic and socio-cultural position of the previously described four largest, non-Western immigrant groups in the Netherlands. The first survey was conducted in 1988 and repeated in 1991, 1994, 1998 and 2002. The survey includes a random sample of Dutch natives and immigrants with Turkish, Moroccan, Surinamese and Dutch Antillean backgrounds, who were the heads of their households at the time of the survey. The sample was drawn from 10 to 13 cities (depending on the survey year), which cover about 50 percent of the four immigrant groups within the Dutch population. The sample size varies between 5,445 persons in 1988 and 8,321 persons in 2002. The response rate is between 44 and 79 percent and varies across immigrant groups and years. What is particularly important for this study is that the SPVA survey includes detailed information about originand host-country specific schooling and contacts with co-ethnics and natives. Additionally, in both chapters the SPVA survey is merged with the neighborhood level data on the percentage

11

Chapter 1

of immigrants with a non-Western background from the Statistics Netherlands (Statistics Netherlands 1998a, 1998b). 1.5.2 German data

In chapters 4 and 5 I examine the economic outcomes of immigrants in Germany. I use the German Socio-Economic Panel (GSOEP 1984-2004). The GSOEP is representative panel survey of the population in Germany administrated annually by the German Institute for Economic Research (DIW Berlin) (Haisken-DeNew and Frick 2005). The survey started in 1984 and provides information on the socio-economic position of all household members. The sample includes Germans living in the old and new German states, immigrant guest workers and more recent immigrants to Germany. In 1984, the survey included 12,000 respondents from the Federal Republic of Germany (West Germany), including 3,000 legal immigrants. The response rate varied from 25 to 70 percent between 1984 and 2004 year, with unsuccessful interviewing and tracking of individuals throughout the surveys as the main causes of sample attrition (Kroh and Spieß 2008). The main immigrant groups participating in the survey are from Italy, Greece, Spain, Turkey and former Yugoslavia, the main sources of guest worker migration; ethnic Germans mainly from Eastern Europe; immigrants from Western Europe, the United States, Canada, Australia, Japan; and other immigrants from developing countries. The advantage of the GSOEP is that it provides longitudinal information over a 20 year period on immigrants’ social contacts, including contacts with German natives and economic outcomes. In chapter 5, I merge the GSOEP survey with data on the percentage of co-ethnics per region (i.e., Bundesländer) obtained from the German Microcensus (Microcensus Scientific Use File 2009).

1.6 Outline of the study
Table 1.1 provides an outline of this study. Chapter 2 examines the employment and occupational status of immigrants in the Netherlands using the SPVA data (SPVA 1998, 2002). In this chapter, I focus mainly on answering the first research question. Specifically, chapter 2 addresses the impact of origin- and host-country schooling on the employment and occupational status of immigrants in the Netherlands. It also examines whether the economic returns to origin-country schooling vary between Mediterranean (i.e., Turkish, Moroccan) and Caribbean (i.e., Surinamese and Dutch Antillean) immigrants, and depends on ethnic concentration in the region of living. In chapter 2, I also answer the third research question about the interplay between origin- and host-country specific human and social capital, by

12

Introduction

studying whether the positive effect of host-country schooling is mediated or spurious due to increased social contacts with natives. Chapter 3 studies self-employment of immigrants in the Netherlands using the SPVA survey (SPVA 1991, 1994, 1998, 2002). In chapter 3 all three research questions are addressed. Specifically, this chapter examines the roles of origin- and host-country specific human and social capital in immigrant self-employment. This chapter also studies the mediation and spuriousness between host-country human capital and social contacts with natives. Chapter 4 examines the employment status of immigrants in Germany using the GSOEP data (1984-2004). It mainly addresses the second and third research questions by examining the impact of social contacts on immigrant employment. Specifically, it studies whether a positive correlation between social contacts and immigrant employment is causal, reflects reverse causality, or is spurious due to time-variant host-country human capital or timeconstant unobserved characteristics of immigrants. Chapter 5 also relies on the GSOEP data (1984-2004), and it examines the occupational status and income of immigrants in Germany. It mainly answers the second and third research questions. Specifically, it studies the impact of social contacts on immigrant occupational status and income. It also examines whether the economic returns to origin- and host-country specific schooling and language skills depend on co-ethnic concentration in the region of living. The study ends with a summary of the most important findings and contributions presented in chapter 6. Additionally, in chapter 6 several points for discussion and implications for future research are raised.

13

Table 1.1 Outline of empirical chapters
Chapter 2 Economic performance Employment Occupational status Human capital • Education abroad and in the Netherlands • Work experience abroad and in the Netherlands • Dutch language skills • Health Social capital • Partner • Dutch contacts • Organization membership: Dutch/ethnic/none • Ethnic composition of contacts • Ethnic concentration • Partner: Dutch/ethnic/none • Dutch contacts • Organization membership: Dutch/ethnic/none • Ethnic concentration • • • • Partner Frequency of contacts Volunteering German contacts Data source SPVA 1998, 2002 in the Netherlands Population Turks, Moroccans, Surinamese, Dutch Antilleans Method of analysis Logistic regression with cluster correction Heckman’s selection model with cluster correction SPVA 1991; 1994, Turks, Moroccans, 1998, 2002 in the Surinamese, Dutch Netherlands Antilleans Logistic regression with cluster correction

3

Self-employment

• Education abroad and in the Netherlands • Work experience abroad and in the Netherlands • Dutch language skills • Education abroad and in Germany • German language skills • Work experience

4

Employment

GSOEP 1984-2004 Turks, Greeks, in Germany Yugoslavians, Italians, Spaniards, Eastern Europeans, Immigrants from developing countries GSOEP 1984-2004 Turks, Greeks, in Germany Yugoslavians, Italians, Spaniards, Eastern Europeans, Immigrants from developing countries

Random effects logistic regression Fixed effects logistic regression

5

Occupational status Annual income

• Education abroad and in Germany • German language skills • Work experience

• • • • •

Partner: German/ethnic/none Frequency of contacts Volunteering German contacts Ethnic concentration

Heckman’s selection model with cluster correction

2 The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants∗

2.1 Introduction
It is widely known in the literature that many immigrants in Western countries are at a disadvantage in the labor market (Borjas 1994; Chiswick 1978; Portes and Rumbaut 1996). Immigrants have more difficulties finding a job, they have longer periods of unemployment and, if they are employed, they often have less prestigious jobs and lower earnings compared to natives (e.g., Alba and Nee 1999; Borjas 1994). A well-known explanation of ethnic inequalities is that immigrants are less skilled and less productive than natives. Because many immigrants come from developing countries, they are often lower educated than natives in Western countries. Furthermore, several authors have argued that the skills immigrants acquired in the country of origin (origin-country human capital) are less valued than skills obtained in the country of destination (Borjas 1994; Duleep and Regets 1999; Friedberg 2000), as these origin-specific skills are of lower quality, difficult to transfer, or employers are more uncertain about these skills. Hence, it is argued that immigrants’ proficiency in the home language is of little use when the official language in the host country is different. Similarly, authors have maintained that educational qualifications and work experience obtained in the country of origin are not equally valued as qualifications and experience acquired in the host country (Friedberg 2000). It is generally assumed that immigrants are particularly disadvantaged in the labor market upon arrival in the host country, but as they acquire host-country human capital they improve their economic position (Borjas 1994). Although many researchers have theorized about the importance of host-country skills, there is only sound empirical evidence for the role of language. There is ample support in the literature that immigrants who acquire the host-country language significantly improve their economic opportunities (e.g., Chiswick and Miller 1995, 2002). Much less is known about the impact of returns to origin-country vis-à-vis destinationcountry education. Only a few studies have addressed this question. These studies were

A slightly different version of this chapter is published in Social Forces (Kanas and Van Tubergen 2009). An earlier version of this chapter was presented at the 8th Annual European Sociological Conference in Glasgow, United Kingdom, September 2007.

∗

Chapter 2

conducted among immigrants in Israel (Friedberg 2000), Canada (Li 2001), Sweden (Duvander 2001) and the United States (Bratsberg and Ragan 2002; Bratsberg and Terrell 2002; Zeng and Xie 2004). Although these studies generally support the presumed higher returns to host-country education, little is known whether these patterns equally hold in other countries – e.g., in the new immigrant countries in Western Europe. What is more problematic, however, is that these studies rely on general population surveys (e.g., census data) that do not directly measure (in years) pre-migration and post-migration schooling. Instead, researchers have used information on people’s age of migration and total years of education to construct measures of (years of) education before and after migration. As argued by Chiswick and Miller (1994), such indirect measures may lead to substantial measurement error and erroneous conclusions. To see how important this measurement error is, consider the following example (cf. Van Tubergen and Van de Werfhorst 2007). An immigrant who has attended five years of education in his country of origin (i.e., from age 6 to 10), who migrated at 25 years of age and then attended school for five more years in the country of destination (i.e., from age 25 to 30) is estimated to have attended 10 years of education in the country of origin and not to have obtained any education after migration. Nevertheless, the majority of studies on the returns to origin- and destination-schooling rely on this indirect measure (Bratsberg and Ragan 2002; Bratsberg and Terrell 2002; Friedberg 2000; Zeng and Xie 2004). Furthermore, some studies (Duvander 2001; Li 2001) only include the level of education and a dummy variable indicating whether the highest education was (probably) obtained in the country of destination. In this way, however, those who obtained their highest education in the receiving nation also include people who were educated in their origin nation. We make three contributions to the literature. First, we examine the returns to origin- and host-country schooling using direct measures. Only two studies have used direct measures of pre- and post-migration schooling (Constant and Massey 2003; 2005). However, these studies relied on different measures for pre-and post-migration schooling (in years and levels of education, respectively), thereby hampering comparisons between the returns to pre- and postmigration schooling. Second, we examine whether the returns to pre-migration schooling differ between two contexts: the immigrant group and the region of living. It is argued in the literature that, because of transferability, quality, and uncertainty, education acquired in some origin countries is valued more than education acquired in other origin countries. This chapter contributes to the growing evidence on this issue (Bratsberg and Ragan 2002; Bratsberg and Terrell 2002; Friedberg 2000; Zeng and Xie 2004). The returns to origin-country schooling may also depend on immigrants’ region of living in the host country. It can be argued that

16

The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants

ethnic concentration increases the transferability and reduces the uncertainty of origin-country schooling, and immigrants living in ethnic concentrations may therefore receive higher returns to their origin-country schooling than those living in regions with few immigrants. Although studies have been done on the impact of ethnic concentration on economic outcomes (e.g. Chiswick and Miller 2002, 2005; Kogan and Kalter 2006; Lewin-Epstein and Semyonov 1992; Tienda and Lii 1987; Tolnay 2001), few studies have specifically examined the cross-level interactions between ethnic concentration and the returns to origin-country schooling among immigrants. Third, we provide an alternative explanation for the presumed (positive) effect of hostcountry schooling on immigrant economic outcomes. Rather than focusing on the common interpretation that emphasizes the lower uncertainty of employers, or the higher quality and transferability of skills (Bratsberg and Ragan 2002; Bratsberg and Terrell 2002; Friedberg 2000; Zeng and Xie 2004), it can be argued that social contacts play a major role. A longstanding and influential line of research in sociology, as well as in economics, has been developed that considers the impact of social capital on people’s economic attainment (Coleman 1990). The major insight is that having more (resourceful) contacts generally increases people’s economic opportunities (Bourdieu 1986; Boxman, De Graaf and Flap 1991; Coleman 1990; Ioannides and Loury 2004; Lin 1999; Mouw 2002; Portes and Sensenbrenner 1993). Specific studies in the field of migration have equally shown that contacts with family and friends promote the economic performance of immigrants (Aguilera 2003; Aguilera and Massey 2003; Nee, Sanders and Sernau 1994; Sanders and Nee 1996; Sanders, Nee and Sernau 2002). However, these earlier studies have focused predominantly on contacts within the own ethnic group. We label this origin-country social capital, as those contacts – maintained either in the country of origin or country of destination – remain within the own ethnic community. Although contacts with co-ethnics generally foster economic mobility, it could be argued that contacts with natives (destination-country social capital) in particular may improve immigrants’ economic performance. Immigrants have predominantly contacts with members of their own ethnic group, who know the host country labor market less well and who have less information on job opportunities than natives. Making this distinction between origin-country vis-à-vis destination-country social capital sheds new light on the presumed impact of origin-country vis-à-vis destination-country schooling on the economic performance of immigrants. From the perspective of social capital theory, one could argue that immigrants’ who have more destination-country schooling benefit from such skills, since acquiring them is associated with increasing contacts with the native population. For example, people who enroll in education in the host country are more

17

Chapter 2

likely to develop contacts with natives, which promote their economic opportunities. Thus, we test an alternative mechanism for the presumed positive effect of host-country schooling on immigrant economic outcomes. We make use of an immigrant survey that has been collected in 1998 and 2002 among four large immigrant groups in the Netherlands: Turks, Moroccans, Surinamese and Dutch Antilleans. The survey has been specifically designed to study these four ethnic minority groups. Sample sizes are large, bilingual interviewers are used, and extensive information on migration history, human capital, social capital and labor market outcomes is included.

2.2 Dutch setting
Before formulating the hypotheses, we briefly discuss the four groups studied here. In 2000, immigrants from Turkey, Morocco, Suriname and Dutch Antilles represented about 66 percent of the non-Western, foreign-born population and about 41 percent of the total immigrant population in the Netherlands (Statistics Netherlands 2008). Turks and Moroccans mainly came to the Netherlands as part of the ‘guest workers’ program in the 1960s and 1970s. Suriname and the Dutch Antilles were former colonies of the Netherlands and migration is common from these countries. The four groups have a higher unemployment rate than Dutch natives and they are overrepresented in lower-paid jobs. This is especially true for Turks and Moroccans, who have the highest rates of benefit dependence (The Netherlands Institute for Social Research (SCP) 2005). The groups also differ with respect to their socio-cultural integration. The longstanding connection between Surinam, Dutch Antilles and the Netherlands has resulted in several advantages for immigrants from these countries, including knowledge of the Dutch language, familiarity with the Dutch educational system and a long tradition of cultural exchange. Surinamese and Dutch Antilleans are rather well-integrated socially. For example, about 25 percent of the Surinamese and 48 percent of Antilleans are married with natives, compared to less than 5 percent among the Turks and Moroccans (Kalmijn and Van Tubergen 2006). Immigrants from Turkey and Morocco were not exposed to the Dutch language before immigration, and virtually all Turks and Moroccans are Muslims. By contrast, Dutch natives and immigrants from Surinam and Dutch Antilles are predominantly Christian or not affiliated with a religion.

2.3 Theory and hypotheses
Human capital theory has been used to explain immigrants’ labor force participation (e.g., Bevelander and Veenman 2004; Sanders and Nee 1996), income (e.g., Chiswick 1978; Zeng

18

The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants

and Xie 2004), occupational status (e.g., Raijman and Semyonov 1995), and job tenure (e.g., Aguilera 2003). Basically, human capital refers to the capability to work productively. According to the human capital theory, the more talented, skilled and capable people have a better position in the labor market. In empirical research, human capital is often measured in terms of education, labor market experience and health (e.g., Chiswick 1978). There is ample evidence that people with higher levels of education, work experience and who have better health have a better position in the labor market. In the field of immigration, however, an important distinction is made between origin and destination human capital. It is argued that human capital acquired in the country of origin is less valued by employers (Bratsberg and Ragan 2002; Chiswick 1978; Friedberg 2000; Zeng and Xie 2004). Although talents, motivation and health seem to be rather context independent (what one could label general human capital), knowledge and skills might be more or less specific to a certain context. This is certainly the case for language skills. Knowledge of the official language of the country of origin (origin-country human capital) is of little usage when the official language of the host country is different. With the exception of a few jobs provided by co-ethnics, most occupations require knowledge of the native language (hostcountry human capital). There is ample empirical evidence that destination-language proficiency has a strong positive effect on labor market outcomes. Immigrants who speak the official language of the host country better are more likely to be employed and have higher earnings than those with fewer command of the host language (Chiswick and Miller 1995, 2002). Likewise, the returns to labor market experience obtained in the country of origin might be less strong than the returns to experience from the host country. Employers are less wellinformed about the occupational career of immigrants before migration than about the experience immigrants obtained in the host country. Furthermore, the knowledge and skills immigrants acquire on the job in the country of origin are presumably less valuable for the labor market in the receiving country. In this chapter, we focus on the returns to origin- and destination-schooling. It can be argued that education obtained in the country of origin is difficult to transfer to the host country and that it is generally of lower quality, since many immigrants come from less developed nations (Friedberg 2000; Zeng and Xie 2004). Furthermore, employers may be reluctant to grant full recognition to foreign credentials as they are simply uncertain about the knowledge and skills that these credentials provide. By contrast, education obtained in the host country provides immigrants with credentials that are fully recognized in the hostcountry labor market. Employers are familiar with those diplomas and the education received in the host country more strongly matches the needs of the labor market. Hence, we

19

Chapter 2

hypothesize that the returns to host-country schooling are higher than the returns to origincountry schooling (H1). An important issue is the possible interplay between national origin and the value of origin-country schooling (Bratsberg and Terrell 2002; Friedberg 2000). Several studies have shown that the returns to pre-migration schooling vary between countries of origin. Bratsberg and Terrell (2002) studied the effect of educational quality on the returns to pre-migration schooling. They showed that the effect of origin-country education increases with the quality of education in the country of origin, as measured by lower pupil-teacher ratios and greater expenditures per pupil. Friedberg (2000) studied the transferability mechanism and showed that in Israel, immigrants from Western countries receive higher returns to pre-migration schooling than immigrants from Asia and Africa. However, both studies relied on indirect measures of pre-migration schooling. We extend this line of research by comparing groups in the Netherlands. We assume that educational qualifications obtained in Suriname and the Dutch Antilles are valued more than qualifications obtained in Turkey and Morocco (Cf. Van Tubergen and Van de Werfhorst 2007). After all, Suriname and the Dutch Antilles are more economically developed than Morocco and regions in Turkey the immigrants come from (i.e., difference in quality) and Suriname and the Dutch Antilles were former colonies of the Netherlands, making the educational system and the labor market more similar to that of the Netherlands (i.e., differences in transferability and uncertainty). It is therefore hypothesize that the returns to origin-country schooling are higher among Surinamese and Antillean immigrants than among Turks and Moroccans (H2). The effect of origin-specific capital may also depend on ethnic concentration. Ethnic concentration is expected to increase the value and transferability of pre-migration knowledge and skills. Immigrants living in ethnically concentrated areas can rely on origin-specific knowledge about ethnic goods, consumer preferences and norms (Chiswick and Miller 2005). Living in ethnic concentration area increases also chances of working for co-ethnic employers who can better recognize and value origin-country schooling. Finally, ethnic concentration may also increase the knowledge and experience of native employers who may be better informed about the value and portability of origin-country schooling than native employers living in areas with few immigrants. Hence, we hypothesize that ethnic concentration increases the returns to origin-country schooling among immigrants (H3). Previous studies have argued that the higher returns to host-country education might be explained by its better quality, transferability and employers’ uncertainty towards origincountry credentials. In this chapter we provide an alternative explanation for the differential returns to origin- and host-country schooling. It can be argued that higher returns to host-

20

The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants

country schooling are due to its relationship with contacts with natives. The idea is that immigrants benefit from host-country schooling as it is associated with increasing contacts with the natives. Emerson, Kimbro and Yancey (2002) showed that the school context might be crucial for the formation of racially diverse relationships. Those who attended racially diverse schools were more likely to attend inter-ethnic as opposed to co-ethnic religious congregations, and to have higher rates of inter-ethnic marriage. Thus, it can be argued that immigrants who are enrolled in education in the host country are more likely to have native friends, acquaintances or a partner. These contacts with natives can provide valuable social capital promoting the chances of immigrants in the labor market. The idea that social capital facilitates immigrant economic integration is not new in the migration literature (Aguilera 2005; Aguilera and Massey 2003; Nee et al. 1994; Sanders et al. 2002). It is argued in the literature that immigrants profit from the resources of others, most notably information and influence. Within their social network, people may provide immigrants directly with information on a job that is available, and they can also inform about where to look for jobs generally, how to present themselves for employers, and how to behave on the job (Aguilera and Massey 2003; Fernandez-Kelly 1995). Furthermore, the contact person can influence the job-matching process by providing entry into desirable occupations (Lin 1999; Mouw 2003). Several studies conducted among immigrants in the United States indeed show the importance of social contacts for immigrants’ economic integration (Aguilera 2002, 2003, 2005; Aguilera and Massey 2003; Nee et al. 1994; Sanders et al. 2002). Based on the foregoing, we hypothesize that social contacts have a positive effect on the economic performance of immigrants (H4). Moreover, several researchers have suggested that contacts with natives may be particularly important for information diffusion and influence (Drever and Hoffmeister 2008; Kahanec and Mendola 2007; Kazemipur 2006). It is argued that natives are better informed about specific job openings, they generally know better how to find jobs, and they know more how to present themselves to employers than immigrants. One reason for this difference in resources is that natives naturally have been exposed for a longer time to the host-country labor market than immigrants, and for that reason they have superior information. Another reason is that natives are less often unemployed, higher educated and have more prestigious jobs than immigrants. Hence, contacts with natives may be helpful in finding a job and improving job quality. Empirically, however, less is known about the presumed positive impact of contacts with natives. Kahanec and Mendola (2007), in their study on immigrants in Great Britain, found that participation in mixed or non-ethnic clubs and voluntary organizations is positively associated with salary employment. However, Drever and Hoffmeister (2008) found that having close German friends provided little advantage in the

21

Chapter 2

job search process, and that having such cross-ethnic contacts was not associated with an improved employment position of immigrants. We examine the influence of contacts with natives for the labor market position of immigrants in the Netherlands. In addition, despite important insights gained from previous studies, little attention was paid to the possible interplay between host-country human capital and contacts with natives. It is hypothesized that contacts with natives explain part of the positive effect of host-country schooling (H5).

2.4 Data and methods
The data come from the Social Position and Use of Welfare Facilities by Immigrants survey (SPVA 1998, 2002). Two waves were combined in order to increase the number of cases. SPVA is a large-scale, cross-sectional, immigrant-specific survey (Van Ours and Veenman 2003). The data are unique in the sense that they contain information on pre- and postmigration human capital. They provide a wide range of information on the socio-economic and the socio-cultural position of four large ethnic minority groups in the Netherlands: Turks, Moroccans, Surinamese and Antilleans. People in cities were overrepresented in the sample frame since most members of ethnic minorities live in cities. The sample frame consists of ten to thirteen cities (depending on the survey year), covering about 50 percent of the four minority groups of the Dutch population. This overrepresentation of immigrants in urban areas might bias descriptive figures on employment and occupational status (i.e., unemployment rates tend to be higher in the cities), but it is less likely to affect our multivariate results. The data have some limitations too. One issue is the cross-sectional design. This makes it impossible to examine the causality between some –though not all– variables. For example, even if we hypothesize that social contacts with Dutch increase the odds of employment, it may also be the opposite, namely, that having a job increases connections to natives. The issue of reversed causality is less problematic for the presumed effects of schooling, which is the main focus of this chapter. With respect to health and social contacts, we will keep the cautionary note in mind and talk about empirical associations. Another issue is non-response. The non-response rate for the 1998 and 2002 waves was the lowest among the Turks (39 percent), and the highest among the Surinamese (56 percent) (Groeneveld and WeijersMartens 2003). These numbers are rather high when compared to surveys in other countries, but they are typical when compared to other surveys within the Netherlands (Van Ours and Veenman 2003). There are several reasons to believe that the low response rate is not of major concern to our conclusions. The non-response rates have been investigated and there is no evidence for systematic non-response in our survey with regard to core indicates such as

22

The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants

gender and education (Groeneveld and Weijers-Martens 2003; Martens 1999). Moreover, special measures were taken to include respondents that are less well-integrated culturally and economically. This means that interviewers were from the same ethnic minority group as the respondent and were matched on gender and interviews could be carried out in the ethnic language (Groeneveld and Weijers-Martens 2003; Martens 1999). The analysis is restricted to male immigrants between the ages of 25 and 60. The age category was chosen based on the presumption that individuals older than 24 years have finished schooling and that individuals older than 60 have left the labor market as a consequence of (early) retirement (Bevelander and Veenman 2004). Immigrants are defined as individuals born outside the Netherlands. Because information was only available for the heads of households our analysis is restricted to these members of the family.1 The focus is on males, because mostly men and only a few women are heads of households among the Turks and Moroccans. All in all, our analysis includes 4,410 respondents. 2.4.1 Dependent and independent variables

We analyze the employment and the occupational status of immigrants. The dependent variables are measured as follows: Employment: Respondents were asked about their employment status. Those who are employed, including self-employed, are contrasted with those who are without work (unemployed, currently available and seeking work and inactive). By combining the labor force participation rates and unemployment rates among the active labor force we avoid the complicated boundary between inactivity and unemployment. Occupational status: Employed respondents were asked about the status of the current job. Occupational status is measured in terms of the International Socio-Economic Index. The ISEI scale measures the hierarchical position of the occupation and is based on a weighted sum of the average education and the average income of occupational groups. To obtain ISEI scores for the occupations we use tools that convert the ISCO-92 classification into ISEI (Ganzeboom, De Graaf and Treiman 1992).

The overrepresentation of immigrants in urban areas and focus on household-heads only might bias our results. To see whether this is the case we performed an additional analysis using a nationally-representative survey of (all) respondents (i.e., Leefsituatie Allochtone Stedelingen 2004/2005). The LAS data include a random sample of immigrants from Turkey, Morocco, Suriname and Dutch Antilles. Because social capital variables present in LAS cannot be compared with those from the SPVA survey, we replicated the models for human capital and control variables only. Our results, not presented here, show that the returns to host-country schooling are higher than the returns to origin-country schooling on both employment and occupational status. Thus, our findings do not change qualitatively and our conclusions remain the same even when we use a nationally-representative sample.

1

23

Chapter 2

We include measures of (origin and destination) human capital, (origin and destination) social capital, and (additional) controls. In order to obtain a parsimonious model, we tested each variable measured on an ordinal level whether they may be entered as a continuous variable. When likelihood ratio tests show that dummy specification does not significantly improve the model, we chose the linear specification. Human capital is measured by four indicators. Education: Respondents were asked about the highest level of completed education in their country of origin and in the Netherlands. In order to facilitate comparisons between education obtained in the country of origin and destination, we constructed five categories: (1. no education, (2. primary, (3. lower secondary, (4. higher secondary, and (5. tertiary. We include both education abroad and education in the Netherlands as continuous variables. Work experience: The survey provides a direct measure of work experience in the Netherlands and a more indirect measure of experience abroad. A separate question asks respondents to report the number of years of work experience in the Netherlands. No such question is included for experience abroad. We therefore used information on age at immigration and the total years of schooling in the country of origin. Experience abroad is measured as: age at immigration – years of schooling abroad – 6. Thus, the survey contains information on actual work experience in the Netherlands and potential work experience in the country of origin. Health: Respondents were asked what their health condition was. The possible answers were: (1. very bad, (2. bad, (3. neutral, (4. good and (5. excellent. Because very few people indicated that their health was very bad (3.1 percent) or bad (14.6 percent), we grouped categories 1 and 2 together. We used 5 as the reference category and included three dummy variables. Dutch language skills: Respondents were asked whether they experience difficulties with speaking the Dutch language. We created a dummy variable contrasting those who speak Dutch fluently with those who experience problems with speaking Dutch.2 We included several measures of social capital. Ethnic concentration at the neighborhood level (four-digit zip codes) was calculated as the population percentage of first- or secondgeneration immigrants with a non-Western background. Non-Western minorities predominantly include immigrants from Turkey, Morocco, and Suriname and Dutch Antilles (Statistics Netherlands 1998a, 1998b). We use figures for the year 1998. Information on group-specific measures at the neighborhood level is unavailable. Dutch contacts: We combined two questions that measure immigrants’ contacts with natives. Respondents were

2

Objective assessment of language skills would be more desirable than self-reported measure of language skills used in this chapter. There could also be a difference between self-reported and interviewer-reported measures of language skills. Research shows, however, that different measures of language proficiency highly correlate (Van Tubergen and Kalmijn 2005).

24

The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants

asked whether they ever received Dutch friends or neighbors as visitors and whether they sometimes associated with the Dutch in their free time. For both questions, respondents could choose between (1. never, (2. sometimes, and (3. often. Answers to these questions are highly correlated (Spearman correlation .69; Cronbach’s alpha .82), and we therefore combined them by adding up the scores on the two items and dividing them by two. Ethnic composition: Next to an absolute measure of contacts with natives, we included a variable that measures the number of contacts with the Dutch in relation to that of co-ethnics. Respondents were asked about the ethnic composition of their social contacts, and we constructed a variable with three categories: (1. most contacts with ethnics, (2. equal contacts with Dutch and ethnics, (3. most contacts with Dutch. We used 1 as the reference category and included two dummy variables. Membership organization: Respondents were asked whether they were a member of an organization and whether the organization was predominantly ethnic or Dutch. We constructed a variable with three categories: (1. no membership, (2. member of a predominantly ethnic organization and (3. member of a predominantly Dutch organization. We used 2 as the reference category and included two dummy variables. Preferably, we would like to have additional measures of social capital, including information on network size, diversity, and resources. Unfortunately, the SPVA survey does not include this information. At the same time, however, it should be noted that none of the previous studies in the literature on the impact of post-migration schooling include measures such as inter-ethnic contacts and ethnic composition of organizations. Married: We constructed a dummy variable indicating those who are cohabitating/married as compared to single people. Caribbean: We contrasted immigrants from Turkey and Morocco (“Mediterranean”) with immigrants from Suriname and the Dutch Antilles (“Caribbean”). We combined immigrant groups to get a sufficiently large number of respondents (especially for the interactions), and because the groups are very homogeneous (e.g., language, religion, economic development). Migration motive: Respondents were asked about their reasons for immigrating. We constructed three categories of the main reasons: (1. work, (2. family, (3. other. SPVA 2002: To control for survey effects, we included a dummy variable indicating the 2002 wave. Table 2.1 presents descriptive statistics for the independent and dependent variables. We checked for high multicollinearity among the independent variables, but correlations did not exceed critical levels. Note, however, that for precisely this reason we did not include additional controls such as age, age at immigration, or length of stay.

25

Chapter 2

Table 2.1 Descriptive Statistics of Independent and Dependent Variables, SPVA 1998 and 2002.
Range Dependent variables Employed Occupational status (ISEI) Independent variables Human capital Education abroad Education in the Netherlands Work experience abroad Work experience in the Netherlands Dutch language skills Health Bad or very bad Neutral Good Excellent Social capital Ethnic concentration (%) Contacts with Dutch Ethnic composition network More with ethnics Equal More with Dutch Membership organization Ethnic Dutch No membership Control variables Married Caribbean Migration motive Work Family Other Survey 2002 0/1 16-88 Mean .67 37.70 S.D.

15.21

1-5 1-5 0-47 0-44 0/1 0/1 0/1 0/1 0/1 .84-79.94 0-2 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1

2.26 1.95 9.99 12.65 .46 .18 .18 .44 .20 32.29 .91 .60 .26 .14 .14 .13 .73 .82 .33 .34 .40 .26 .38

1.15 1.36 8.24 8.58

20.92 .67

2.4.2

Methods

We used logistic regression for the analysis of employment, and linear regression for the analysis of occupational status. To adjust for the fact that respondent’s answers are correlated within neighborhoods, we used cluster correction within Stata 9. Because immigrants’ occupational status was estimated for only those who were employed, we corrected for possible sample selection bias in our sample.3

To compare the individual determinants of occupational status versus labor market participation of immigrants, this study uses a Heckman model (Lee 1983). There are two identifying variables. The region of living as represented by six dummy variables for region of living. We control for region of living because it is likely to influence the likelihood of immigrants’ labor force participation but has no effect on occupational status. The

3

26

The Impact of Origin and Host Country Schooling on the Economic Performance of Immigrants

2.5 Results
We will first discuss the results of the multivariate analyses of employment (Table 2.2) and occupational status (Table 2.3). Model 1 includes measures of human capital; Model 2 adds interactions between certain human capital variables and national origin and ethnic concentration; Model 3 includes only social capital variables; Model 4 includes human and social capital variables simultaneously.4 We will compare the coefficients of Model 2 to that of Model 4, in order to see whether the role of host-country schooling persists when social capital is taken into account. We will compare the coefficients of these different models by a method proposed by Clogg, Petkova and Haritou (1995). 2.5.1 Human capital

It was hypothesized that the returns to host-country schooling are higher than to origincountry schooling (H1). Both education abroad and education in the Netherlands are measured on a five- point scale, ranging from no education to tertiary education. Table 2.2, Model 1 shows that higher diplomas obtained in the country of destination (b = .23) more strongly increase the odds of employment than higher qualifications obtained abroad (b = .16), the difference being not statistically significant, however. With respect to occupational status, we find that the returns to education obtained in the Netherlands are significantly higher (chi2 = 34.60; p = .00). For each unit of increase in education, those who obtained their education in the Netherlands score (5.29 – 2.71 =) 2.58 status points higher than those who obtained a similar education abroad (Table 2.3, Model 1). The results presented here refer to a linear specification of ordered categories of educational levels (ranging from no education to tertiary education). To examine whether our results are sensitive to this specification, we examined alternative measures of education. Using dummy variables for each educational level confirms our conclusions: the returns to education obtained in the host country are higher than education obtained in the country of

second identifying variable is the number of persons in the country of origin the respondent is taking care of. Again, we expect that having dependents in the country of origin influences the likelihood of immigrants’ labor force participation but is not related to occupational status. Both variables are indeed significantly and quite strongly correlated with employment status, but there is no (region of living) or only a weak (number of dependents in country of origin) correlation with occupational status. The rho is significant suggesting possible selectivity in our sample; therefore we report findings from the selection model. 4 Although it is beyond the scope of our study, it is important to mention that even when we control for human and social capital characteristics (Model 4) Turkish and Moroccan immigrants are more often unemployed and have lower status jobs than the Caribbean groups, possibly suggesting stronger discrimination against these Mediterranean groups.

27

Chapter 2

Table 2.2 Logistic Regression of Immigrants’ Employment, SPVA 1998 and 2002.
Model 1 B t-ratios Independent variables Human capital Education abroad (centered) Education in the Netherlands (centered) Work experience abroad (centered) Work experience in the Netherlands (centered) Dutch language skills (Ref. Bad) Health (ref. Excellent) Bad or very bad Neutral Good Social capital Ethnic concentration (%) Contacts with Dutch Ethnic composition network (ref. More with ethnics) Equal More with Dutch Membership organization (ref. Ethnic) Dutch No membership Control variables Married (ref. Single) Caribbean (ref. Mediterranean) Migration motive (ref. Family) Work Other Survey 2002 (ref. 1998) Interactions Caribbean*Educ abroad Caribbean*Exp abroad Ethnic concentration (%) *Educ abroad Ethnic concentration (%) *Exp abroad Constant Number of clusters Number of individuals Nagelkerke R2 Unstandardized coefficients; **p