migration effect of education

HOWIMMIGRANTCHILDRENAFFECTTHEACADEMICACHIEVEMENTOFNATIVEDUTCHCHILDREN.pdf

Home >Business & Finance homework help >Economics homework help >migration effect of education

HOW IMMIGRANT CHILDREN AFFECT THE ACADEMIC ACHIEVEMENT OF NATIVE DUTCH CHILDREN*

Asako Ohinata and Jan C. van Ours

In this article, we analyse how the share of immigrant children in the classroom affects the educational attainment of native Dutch children. Our analysis uses data from various sources, which allow us to characterise educational attainment in terms of reading literacy, mathematical skills and science skills. Our results suggest that Dutch students face a worse learning environment when they are studying with more immigrant students in the classroom. However, we do not find strong evidence of negative spill-over effects from the presence of immigrant children on the academic performance of the native Dutch students.

The large inflow of immigrants to Europe in past decades has drawn considerable attention to the issue of the impact of immigrants on the labour market outcomes of both the immigrants as well as the natives. Now that immigrants are a substantial part of the population, research is shifting towards assessing educational performance of immigrant children sometimes in comparison to the native children. However, the question of whether immigrant children affect native children’s educational outcomes remains largely unanswered. Our article aims to fill this gap in the literature by analysing whether the presence of immigrant children in the classroom affects the educational attainment of native Dutch children in that classroom.

The Dutch experience presents an interesting case study, since the immigrant students in the Netherlands generally come from families with lower education. This is a feature shared by most European countries and as a result, this article presents relevant findings to a wider European audience. Studying immigrant spill-over effects is helpful when exploring policy implications on how to allocate immigrant students to minimise negative impacts or maximise positive impacts of immigrant children on the educational attainment of native children. Results may also highlight the potential importance of providing additional resources to schools or classes with large numbers of immigrant children.

In contrast to the existing literature that focuses on students in high-schools and above, our article studies the spill-over effects among primary school children, that is, mostly nine and ten-year-old children. Investigating the impact among younger students is of interest, since it allows us to evaluate how native students respond to an exposure to immigrants at a younger age. Moreover, immigrant students may find it easier to assimilate when faced with a new environment at a younger age. If this is the

* Corresponding author: Jan C. van Ours, Department of Economics, Tilburg University, P.O. Box 90153, 5000 LE Tilburg, The Netherlands. Email: [email protected].

The authors thank anonymous referees and the seminar and conference participants in Tilburg University, Australian National University, 2nd TEMPO a Conference in Vienna, 3rd Norface Conference in Mannheim, the CHOICES Workshop in the Hague, the European Society of Population Economics in Bern and the European Association of Labour Economists in Bonn. The authors gratefully acknowledge financial support from the NORFACE research programme on Migration in Europe – Social, Economic, Cultural and Policy Dynamics.

[ F308 ]

Published by John Wiley & Sons, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

case, we may find a smaller effect compared to those reported in the other studies. Moreover, the existing evidence is likely to reflect the accumulated impact from the exposure to the immigrant students during many years. Studying young students allows us to reduce the extent of such an effect.

Our article analyses the spill-over impacts of the share of immigrant students in a classroom on the reading, science and mathematics test scores of native Dutch students in that classroom. The reading test involves reading a long text and answering questions in writing. If the linguistic difficulties experienced by immigrant students interfere with the Dutch students’ learning process, we may observe negative effects of the Dutch students’ reading test scores. In the science and mathematics tests, students are asked to read much shorter phrases and the questions are typically accompanied by visual aids. Nevertheless, successful performance in these subjects requires compre- hension of the subject matter in Dutch. Studying the impacts on multiple subjects separately, therefore, allows us to measure potentially differential impacts of immigrant students across these three subjects. Note that, we do not study the educational attainment of immigrant children; Ohinata and van Ours (2012) does on this issue. In that study, potential spillover effects of the presence of immigrant children on their own educational attainment are not taken into account.

Finally, we also study whether students’ learning environment worsened due to the existence of immigrant students in the same classroom. As Card (2013) argues, potential spillover effects on academic achievement may not be the main concern of native parents. To study potential spillover effects on the learning environment, we investigate whether students experienced for example, more or less bullying or stealing. Dutch students may perform worse if they experience emotional difficulties at school. The results on the learning environment, therefore, present us with complementary evidence to that of the academic performances.

There are several channels through which the presence of immigrant students affect the academic performance of native students. According to Manski (1993), association in behaviour between the two groups of students may come from exogenous, endogenous and correlated effects. When we apply this to immigrant children and native Dutch students, we get the following possible associations. First, immigrant students in the Netherlands typically come from families with lower socioeconomic and educational backgrounds (van Ours and Veenman, 2003). Dutch students, who are studying with immigrant students, may be negatively affected by the presence of such peers in the same classroom. Second, coming from such a disadvantaged background, immigrant students may struggle to perform academically, which may also lead to worse academic achievements of native students. Lastly, parents of both immigrant and native students are likely to select themselves into particular schools. This is sometimes due to immigrants living in segregated areas. It is also possible that parents avoid sending their children to schools where many of the enrolled students have very different backgrounds from themselves. Our article attempts to isolate the correlated effect from the exogenous and endogenous effects but does not attempt to establish the contribution of latter effects separately. Instead, it presents evidence of the combined effects.

Our empirical analysis is based on two waves from two data sets. We use the 2001 and 2006 Progress in International Reading Literacy Study (PIRLS) for information on the

[A U G U S T 2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F309

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

reading abilities of children in the Netherlands. And we use the 1995 and 2007 Trends in International Mathematics and Science Study (TIMSS), which provide information on the mathematics and science abilities of students in the Netherlands.

If immigrant children are randomly allocated to schools across the country, the peer effects can be identified by exploiting the variation in the proportion of immigrant students across schools and classes. However, immigrant families are likely to settle in areas with more immigrants. In fact, Ladd and Fiske (2009) report a high concentration of immigrant students in the four large cities (i.e. Amsterdam, The Hague, Rotterdam and Utrecht). Furthermore, parents of the native Dutch students are reported to choose schools with limited numbers of immigrant students (Karsten et al., 2003; Ladd et al., 2010). In fact, interviews with 931 Dutch parents conducted by Karsten et al. (2003) reveals that avoiding schools with high shares of immigrants was the most frequently cited reason for the choice of their children’s schools. Since immigrant households in the Netherlands typically suffer from lower socioeconomic and educational backgrounds, estimates of negative impacts of immigrants on native students may simply reflect selective school enrolment of both immigrant and native students. To avoid this selectivity problem, our article identifies the peer effects by controlling for the unobserved school characteristics by estimating a school fixed effects model. This identification strategy assumes that once school specific characteristics are controlled for, students are randomly allocated to a particular class within a school. As Ammermueller and Pischke (2009) indicate, primary school students are not generally grouped into classes on the basis of ability or family background. However, we conduct tests to ensure the validity of this assumption.

The remainder of this article is organised as follows. Section 1 gives an overview of previous studies. Section 2 gives an overview of immigration into the Netherlands and discusses characteristics of the Dutch educational system in relation to immigrant children. Section 3 presents the data used for the analysis and discusses the set-up of the analysis. Section 4 describes the parameter estimates. Section 5 concludes.

1. Previous Studies

There are several articles on the topic of educational spill-over effects between immigrant children and native children. The majority of the evidence comes from the US, where the immigrant crowding out effects on the native students are investigated by looking at students’ college and graduate school enrolment rates (Borjas, 2004; Hoxby, 1998), the probability of high school graduation (Betts, 1998; Hunt, 2012), or the number of years of schooling (Betts and Lofstrom, 2000).

An article that is more closely related to ours is Bui (2012), which uses data on fifth grade students from a large urban school district in the Southwest US and studies how limited English proficient students affect each other’s educational outcomes in terms of mathematics, reading and language. Taking school fixed effects into account, she finds that a higher share of limited English proficient students in a cohort leads to an improvement of student achievement, in particular on the mathematics scores.

There are also other studies from outside the US. Gould et al. (2009) use the large influx of Jewish immigrants from the former Soviet Union to investigate peer effects on

F310 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

native Israeli students. They evaluate the impact of exposure to immigrants in the fifth grade on the final matriculation exam pass rate, which is a prerequisite for proceeding to university. They, therefore, study the long-term peer effects of immigrants on the native students. Although these immigrant students come from relatively highly educated families, they faced economic difficulties. As a result, they may have chosen a particular region for settling down where the living expenses were low. To account for the potential selection bias stemming from such self-selection of immigrants to particular regions of Israel, they exploit the random allocation of students to the fifth grade once the numbers of immigrants in the fourth and sixth grades are controlled for. They find that the strong negative impact of the exposure to the immigrant students diminishes once they address the selection bias, although the results remain marginally significant.

Brunello and Rocco (2013) present cross-country evidence from 27 European and Anglo-Saxon countries by using the 2000, 2003, 2006 and 2009 Program for International Student Assessment (PISA). PISA assesses the cognitive abilities (reading, mathematics and science) of 15-year-old students in OECD member countries. They aggregate the micro-level data to the country level in order to avoid selection bias. They also include country fixed effect to control for the selection bias at the country level. They find a significant but small negative impact of immigrant students on native students. The main problem they face is the small sample size as a result of the data aggregation. Therefore, they pool test results from all three subjects. This requires an assumption that the test scores from all three subjects are comparable. However, such an assumption is unlikely to hold due to differential skills required in solving questions from the three subjects.

Jensen and Rasmussen (2011) study the immigrant peer effects in Denmark using the 2000 and 2005 PISA and Danish administrative register data. They address the non-random allocation of immigrant families to certain regions by using the population size of the residence of children as an instrumental variable. They find that a high concentration of immigrant students in school negatively affect the reading and mathematics test scores of native Danish students even after controlling for the potential selection of immigrants to certain regions.

Friesen and Krauth (2011) investigate classroom spillover effects using data from the Canadian province of British Columbia defining immigrant peers as peers that speak non-English languages at home. They investigate numeracy and reading scores of students in the fourth and seventh grade. They also control for endogenous selection of immigrants across schools by including school fixed effects. They find that the effect of immigrant peers on educational attainment varies substantially by the immigrant students’ language at home.

Finally, Geay et al. (2013) use data from the British National Pupil Database between 2003 and 2009 to relate the percentage of non-English speaking children aged 12 in England to the educational performance (reading, writing, mathematics) of native children within the same school. A raw correlation suggests that there is a negative spillover effects but after accounting for differences in school characteristics, these negative spillover effects disappear.

Our article contributes to the existing literature on immigrant spill-over effects on the educational attainment of native children in three ways. First, we contribute to the

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F311

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

relatively scarce European evidence on the issue. Immigrant students in the Netherlands generally come from families with lower education. This is a feature shared by most European countries and as a result, may provide relevant findings for the European situation. Second, our article uses two micro-level data sets, the PIRLS and TIMSS in order to study the peer impacts on various subjects. Our article addresses the endogenous selection problem of immigrant students by exploiting the data availability of multiple classes within the same school. In doing so, our article avoids the small sample problem faced by Brunello and Rocco (2013). Third, in contrast to many of the existing studies, our article investigates the peer effects among young students (i.e. aged 9 and 10). Studying the impact among younger students is of an interest, since it allows us to evaluate how the native students respond to exposures to immigrants early on. Moreover, immigrant students may find it easier to assimilate when exposed to the native environment at a younger age. If this were the case, we may find a smaller effect compared to those reported in the existing studies. Furthermore, estimates in many other articles are likely to reflect the accumulated impact from the exposure to immigrant students over many years. Studying young students allow us to reduce the extent of such effects.

2. Background Information

2.1. Immigrants in the Netherlands

After the Second World War, migrants to the Netherlands moved broadly for the following three reasons. First, large groups of immigrants came from the former Dutch colonies between the middle of the 1940s and 1970s. These include migrants from Indonesia and Moluccas, Surinam and Antilles. Second, foreign workers were recruited in the 1960s and 1970s as guest workers to combat the shortages of labour in the Netherlands. Last, in recent years, some entered as asylum seekers.

The independence of Indonesia in 1949 led to large influxes of Dutch-Indonesian repatriates and Moluccans to the Netherlands. Approximately 300,000 repatriates, half of which were Eurasians, and 12,500 Moluccans migrated to the Netherlands during the two decades. Moreover, approximately 40,000 Surinamese moved to the Netherlands in 1975 when Surinam was decolonised and became independent. Another large flow of migration occurred around 1979 and 1980 when a mandatory entry visa for Surinamese was introduced, since many feared that entry to the Netherlands would become more restricted (Lucassen and Penninx, 1997; Ersanilli, 2007). Finally, there has been a continuous flow of immigrants from the Netherlands Antilles over the past years.

The major hiring of guest workers from Southern Europe, Yugoslavia and particularly from Morocco and Turkey started as a result of the boom of the Dutch economy in the 1960s and its subsequent shortages of unskilled workers in the labour market (van Ours and Veenman, 2005). The number of these immigrants reached approximately 235,000 in 1970 (Penninx et al., 1994). The recruitment stopped in 1973, but further migration from Morocco and Turkey continued even in the 1980s, which were mainly for the purpose of family formation or unification (Ersanilli, 2007).

Political refugees and asylum seekers are another group of immigrants. After the fall of the Soviet Union, many immigrants from eastern Europe moved to the Netherlands.

F312 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

In more recent years, economic and political crises have increased immigrants from diverse backgrounds such as Iraq, Iran, Afghanistan and Somalia.

The number of immigrants in the Netherlands by country of origin in 1996 and 2011 is given in Table 1. As shown, the 1996 Dutch population of 15.5 million consisted of 13 million native Dutch and 2.5 million immigrants while in 2011, the population of 16.7 million consisted of 13.3 million native Dutch and 3.4 million immigrants. Among these immigrants, western immigrants came from countries in Europe (excluding Turkey), North-America, Oceania, Indonesia or Japan. Non-western immigrants came from countries in Africa, Latin-America and Asia (excluding Indonesia and Japan) or Turkey. Over the period of 1996–2011, especially the number of non-western immigrants increased substantially.

An important distinction in the literature on immigrant children is between first and second-generation immigrants. First-generation immigrants include those who were born outside of the Netherlands with at least one parent also born abroad. Second- generation immigrants are those who were born in the Netherlands with at least one of the parents born outside the Netherlands. In the empirical analysis below, we define immigrant children as first-generation immigrant children. Ohinata and van Ours (2011) show parameter estimates with the sum of first and second-generation immigrant children as relevant group of immigrant children. Among non-western immigrants, both the number of first-generation immigrants as well as the number of second-generation immigrants increased substantially. Although Indonesians are one of the major groups of immigrants in the Netherlands, they are unlikely to represent a significant portion of the students at primary schools as they entered mainly in the 1950s and 1960s.

Table 1

Dutch Population by Immigrant Status; 1 January, 1996 and 2011 (000s)

1996 2011

Total First-

generation Second-

generation Total First-

generation Second-

generation

Total 15,493 16,656

of which: Native Dutch 12,955 13,229 Immigrants 2,498 1,283 1,215 3,427 1,735 1,692

of which: Western immigrants 1,327 522 805 1,528 666 862 Non-western immigrants 1,171 761 410 1,899 1,069 830

of which: Indonesians 412 149 263 380 117 263 Moroccans 225 141 84 356 168 188 Antilleans 87 56 31 141 82 59 Surinamese 281 179 102 345 185 160 Turkish 271 167 104 389 197 192

Note. A first-generation immigrant is born outside the Netherlands with at least one parent born outside the Netherlands; a second-generation immigrant is born in the Netherlands with at least one parent born outside the Netherlands. Source. Statistics Netherlands.

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F313

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

2.2. The Dutch Educational System

The highest concentration of immigrant households are found in the four largest cities, that is, Amsterdam, the Hague, Rotterdam and Utrecht. This has two important implications for the purpose of the analysis in this article. First, Dutch parents in large cities have more schools from which to choose. This implies that there is a higher probability of ethnic and socioeconomic segregations in these cities. Second, schools located in the four large cities were likely to have received more funding from the Dutch government, at least until 2006. The Weighted Student Funding (WSF) was in operation between 1985 until 2006 in order to promote equal educational quality among schools and also to assist schools with a larger number of disadvantaged students (Ladd and Fiske, 2009). The scheme calculates a weighting index for each school by taking account of the number of immigrant students as well as disadvantaged Dutch students. This index ranges between 1 and 1.9, where schools with an index of 1 until 1.09 were not given any extra funding. Schools with the index above 1.09 were offered the extra funding, whose amount was reflected by the index. For example, those with 1.9 received 90% more funding per student. The system is made slightly more complex by the fact that money was not directly paid to each school but rather was given to school boards that had the control over the distribution of the allocated funding across the schools. Nonetheless, Ladd and Fiske (2009) show evidence that extra funding was allocated mainly to schools in the four largest cities. The implication of such a treatment is that school principals may have allocated additional resources towards classes with larger numbers of immigrant students. If this is the case, and classes with a high share of immigrant children were being taught by more able teachers or if these classes had more teaching resources, the size of potentially negative peer effects of immigrant students may have been reduced. Below, we investigate to what extent the allocation of educational resources is correlated with the share of immigrant children in a classroom.

3. Data and Set-up of the Analysis

3.1. Data

The data sets employed in this article are collected in the Netherlands through the 2001 and 2006 PIRLS and the 1995 and 2007 TIMSS (see Appendix A for details; the 2003 TIMSS data contain too few observations of schools for the purpose of our analysis). They share similar characteristics, since both types of surveys were designed and conducted by the International Association for the Evaluation of Educational Achievement (IEA). PIRLS assesses the reading abilities of nine and ten-year-olds in 35 countries. Similarly, TIMSS collects information on the mathematics and science abilities in approximately 40 countries. TIMSS also collects information from the eighth graders. However, for the purpose of comparison with the results from PIRLS, only the data on the fourth graders are used for the analysis. Unfortunately, both data sets do not provide information on the ethnic background of the students.

The samples of students in both surveys were selected using a two-stage sampling design. In the first stage, schools were selected using a probability-proportional-to-size

F314 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

sampling scheme. In the second stage, one or multiple fourth grade classes were randomly sampled from each of the selected schools.

The dependent variable is the measure of students’ abilities to read or solve mathematical and scientific questions. To ensure that the tests assess the students’ full abilities without overburdening the participating students, IEA designed the test materials for all three subjects in the following manner. First, the material was divided into eight blocks. Each of these blocks lasted 40 min and they were distributed across ten test booklets. Every student was asked to complete one of these test booklets within 80 min. Due to the design of these tests, each student only completes a fraction of the assessment materials. As a result, the raw scores do not measure the full ability of the students. Instead, plausible values are reported in PIRLS and TIMSS, which use multiple imputation to reveal how the students would have performed should they complete the entire tests (Gonzalez and Kennedy (2003) for more details on the calculation of the plausible values). The reported Cronbach’s alpha reliability coefficient for these data sets range between 0.73 and 0.83. Both in PIRLS and TIMSS, the test scores have been standardised to an international mean of 500 and a standard deviation of 100.

A similar set of covariates are available from both PIRLS and TIMSS. For example, available school characteristics from these two data sets include class sizes and the number of days spent for instructions of each subject and the sizes of population of the regions in which schools are located. Teachers’ years of experience and their age and gender are also available from both data sets. At the individual student level, PIRLS and TIMSS also hold information on age and gender of students and the number of books at home.

There are some differences in the information available from these data sets. For example, the highest educational qualifications of parents are only available in PIRLS. However, both data sets contain information about the number of books at home. In PIRLS, the number of books at home is reported by parents whereas this variable is reported by students in TIMSS. As a result, the information available in TIMSS may be more prone to measurement errors (Ammermueller and Pischke, 2009). Although the number of books at home is no proxy for the parental educational background, there is correlation between the two. Furthermore, the number of books at home is likely to be correlated to income and wealth of the parents.

Table 2 presents the percentage of the first-generation immigrant students as well as the sum of first and second-generation immigrant students in classrooms by the population size of the geographical areas where schools are located. The statistics are calculated using the 2001 and 2006 PIRLS. Table 2 highlights very high concentrations of both first and second-generation immigrant students in schools located in large cities (column (4)).

3.2. Set-up of the Analysis

We are interested in the spillover effects from the classroom share of (first-generation) immigrant students to the education attainment of native Dutch students in that classroom. If students are randomly allocated to schools, we can identify spill-over

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F315

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

effects of immigrant students by estimating the following equation using ordinary least squares (OLS):

yics ¼ bxics þ cCcs þ dMcs þ �ics; ð1Þ where yics denotes the test score for the ith Dutch student in cth class and sth school, xics captures the Dutch students’ individual and family characteristics and Ccs represent classroom characteristics. The key variable for the identification of the spill-over effects is the Mcs, which is the percentage of first-generation immigrant students in each class. The coefficient d would capture both the immigrant students’ influence on the native students due to the immigrant students’ potentially unfavourable family background as well as their better/worse effort to learn during the class. �ics ¼ as þ lcs þ uics, the error term is composed of three terms: as reflects the school specific effect, lcs captures any classroom specific characteristics that are not controlled in the model and uics is the random error term.

Investigation of the immigrant spill-over effects, however, is more complex in the absence of the random student allocation assumption. As indicated before, immigrants in the Netherlands settle mainly in the four large cities. Moreover, native Dutch parents are also observed to place their children in schools with low concentration of immigrant children (Karsten et al., 2003; Ladd et al., 2010). These pieces of information suggest that the students are selectively allocated to schools. As a result, both as and lcs are likely to be correlated with the immigrant proportion variable Mcs, leading the OLS estimator to be biased and inconsistent.

In order to overcome this endogeneity problem, the present article controls for the school fixed effect. The main identification assumption requires that once school specific characteristics are controlled for, students are allocated to each class randomly within a school (i.e. lcs ¼ 0). Therefore, we exploit the variation in the proportion of immigrant students across classes within the same school. This identification strategy excludes schools with just one class. Comparisons of the summary statistics reveal no differences in the average test scores of Dutch students regardless of whether they

Table 2

Percentage of First-generation Immigrant Students and the Sum of First and Second-generation Immigrant Students in Class by the Population Size of the School Location

Population size (000s)

<3 3–100 100–500 >500

First-generation (%) 0–5 0.88 0.76 0.71 0.49 5–10 0.10 0.15 0.07 0.00 10–20 0.00 0.07 0.03 0.51 ≥ 20 0.02 0.02 0.19 0.00

First and second-generation immigrants (%) 0–5 0.53 0.21 0.07 0.00 5–10 0.18 0.24 0.11 0.00 10–20 0.22 0.26 0.35 0.18 ≥ 20 0.07 0.29 0.47 0.82

Source. PIRLS 2001 and 2006.

F316 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

study in schools with one or multiple classes. The schools with just one class are generally smaller and slightly more likely to be located in smaller towns. The identification strategy used in this article closely follows that of Ammermueller and Pischke (2009), who argue that non-random allocation of students does not generally occur in primary schools. Although the focus of Ammermueller and Pischke (2009) is not on the impact of immigrant students, they investigate the effects of student on the peers’ reading plausible scores by using the same identification strategy. Nevertheless, the identifying assumption may be violated if students are non-randomly allocated to classes on the basis of their abilities or school principals allocate more funding and resources to classes with larger numbers of immigrant students. Before we move on to the main analysis, Figure 1 and Table 3 investigate to what extent these two concerns matter.

In order to explore whether immigrant students are allocated randomly to classes, we test whether the observed distribution of the differences in the number of

0 0 1 2 3 4 5 6

0 1 2 3 4 5 6

Predicted

Actual

Predicted

Actual

(a) PIRLS

(b) TIMSS

Fig. 1. Predicted and Actual Differences in Number of Immigrant Students Between Two Classes Within the Same School

Note. Details of how these distributions are calculated are presented in Appendix B.

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F317

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

immigrant students between the two classes within the same school differ from the simulated distribution when we assume random allocation to each class. To do this, we first simulate the distribution of the expected numbers of schools for each difference in the number of immigrant students between the two classes within the same school, assuming that students are randomly allocated to each class. We then compare the predicted distribution of these differences to the actual distribution from our data. If students were indeed randomly allocated to classes, we should observe the two distributions to be similar to each other. Figure 1 presents the results for PIRLS and TIMSS separately. Comparing the two distributions for both data sets and by conducting Fisher’s exact test, we conclude that the actual distributions do not differ significantly from the predicted distributions in either samples, thus, presenting support for the assumption that the allocation of immigrant students across classes within the same school occurred randomly (see Appendix B for more details). Subsection 4.3 conducts further sensitivity analysis that addresses the issue of the ability based class formation of students.

Even if the students were allocated to classes randomly, our identification strategy would fail if teaching resources were non-randomly allocated within schools. Table 3, therefore, tests the assumption of random allocation of teaching resources across classes. The dependent variable in these regressions is the percentage of immigrants in classes and is regressed against class and teacher-level characteristics. If resources are non-randomly allocated, we should observe strong correlations between the

Table 3

Relating the Percentage of Immigrant Students in the Classroom to Classroom Characteristics

Variables (1) (2)

Reading Science/Mathematics

Teaching years 0.21 0.06 (0.13) (0.09)

Teacher is female 2.39* �0.15 (1.42) (1.41)

Age teacher: 30–39 �4.43* �2.63 (2.24) (1.70)

Age teacher: 40–49 �7.07** �1.98 (3.09) (1.81)

Age teacher: 50 or above �8.69** �4.58 (4.18) (3.06)

Class size: 1–19 �2.17 (1.69)

Class size: 20–26 �1.93 (1.16)

Number of classrooms 119 301 Number of schools 55 129 F-test statistics 1.668 0.927

Notes. This Table presents results that tests the random allocation of teaching resources across classes within the same school. The dependent variable is the percentage of first-generation immigrant students in each class. Data employed for column (1) is the 2001 and 2006 PIRLS for reading test scores. The 1995 and 2007 TIMSS for mathematics and science test scores are used to calculate results presented in column (2). The last row of this Table presents the results from joint significance tests. The class size variable is missing in column (2) due to data limitations in TIMSS. Robust standard errors are presented in parentheses. ***p < 0.01, **p < 0.05, *p < 0.10.

F318 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

share of immigrants in classes and these class-level characteristics. Some of the parameter estimates are significantly different from zero. Classes with a high share of first-generation immigrant students are more likely to have a young female teacher. Nevertheless, the last row of Table 3 which reports F-statistics from the joint significance tests shows that we do not find any evidence that teach- ing resources are allocated favourably to classes with high shares of immigrant students.

4. Parameter Estimates

4.1. Preliminary Analysis

To investigate whether there is a relationship between the share of immigrant children in a classroom and the educational attainment of Dutch children, the graphs on the left-hand side in Figure 2 each present a scatter plot of class-level average reading, mathematics and science test scores against the percentages of immigrant students in classes. Clearly, there is a negative correlation between the average test scores and the share of immigrant children. However, this negative correlation may be driven by selective choice of schools both by the immigrant and the native families. For example, parents of children with higher educational skills may have encouraged their children to go to schools with a low percentage of immigrants.

The identification strategy employed in our article assumes that the allocation of students is random once school specific characteristics are controlled for. The graphs on the right-hand side of Figure 2 plot the within transformation of the average test scores and the percentage of immigrant students across classes within the same school. Comparing the left-hand side plots to the corresponding plots on the right-hand side of Figure 2 reveals that the extent of the negative spill-over effects of the immigrant students is reduced once percentages of immigrant students are differenced across classes, except for science. This suggests that once school differences are taken into account, the sizes of raw correlations between educational skills and the share of immigrant children in the classroom is smaller, at least for reading and mathematics test scores. It is also clear from the right-hand side graphs of Figure 2 that there is substantial within school variation of the classroom shares of immigrant students so that potential spillover effects can be identified.

4.2. Baseline Estimates

Table 4 presents the estimates from the school fixed effect model of the immigrant spill-over effects on Dutch students’ various plausible test scores. In each column, the immigrant students peer effects are captured by the variable ‘the percentage of immigrant children in each class’.

The parameter estimates suggest that the presence of immigrant students in the same learning environment has very limited and insignificant impacts on the Dutch students’ academic achievements. The specification assumes a linear spillover effects of immigrant students. When we estimate the model with non-linear spill-over effects, our conclusions remain unchanged. For example, a 1 percentage point increase in the

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F319

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

0 10 20 30 40 50 60 350 400 450 500 550 600 650 700 750

C la

ss -l

ev el

A ve

ra ge

T es

t S

co re

% First Generation Immigrant Students in Class

–200 –150 –100 –50

0 50

100 150 200

–30 –20 –10 0 10 20 30W it

hi n

T ra

ns fo

rm at

io n

of t

he C

la ss

L ev

el A

ve ra

ge T

es t

S co

re s

Within Transformation of the % First Generation Immigrant Students in Class

350 400 450 500 550 600 650 700 750

0 10 20 30 40 50 60

C la

ss -l

ev el

A ve

ra ge

T es

t S

co re

% First Generation Immigrant Students in Class

–30 –20 –10 0 10 20 30 –200 –150 –100 –50

0 50

100 150 200

W it

hi n

T ra

ns fo

rm at

io n

of t

he C

la ss

L ev

el A

ve ra

ge T

es t

S co

re s

Within Transformation of the % First Generation Immigrant Students in Class

350 400 450 500 550 600 650 700 750

0 10 20 30 40 50 60

C la

ss -l

ev el

A ve

ra ge

T es

t S

co re

% First Generation Immigrant Students in Class

–30 –20 –10 0 10 20 30 –200 –150 –100 –50

50 100 150 200

W it

hi n

T ra

ns fo

rm at

io n

of t

he C

la ss

L ev

el A

ve ra

ge T

es t

S co

re s

Within Transformation of the % First Generation Immigrant Students in Class

(a) Reading Scores

(b) Science Scores

Fig. 2. Graphical Evaluation of the Potential Usefulness of Controlling for School Fixed Effect Notes. The data sets employed are the 2001 and 2006 PIRLS (reading) and the 1995 and 2007 TIMSS (science and mathematics). The left-hand side Figures provide scatter plots of the class- level average test scores against the share of first-generation immigrant students for each subject. The right-hand side Figures, on the other hand, present scatter plots after conducting a within transformation to eliminate the school level averages.

F320 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

proportion of immigrant students in class reduces the average Dutch students’ reading score by 0.21 points (column (1)). Similarly, a 1 percentage point increase in the share of immigrant students reduces the science score by 0.40 points but increases the mathematics score by 0.74 points.

Turning to the rest of the estimates in Table 4, age has a negative effect on the reading scores but positive effects on science and mathematics scores. Furthermore, female students perform better in reading tests and worse in mathematics and science. The more books children have at home, the better they perform in their tests. Teachers’ teaching experiences seem to matter little. As there is some evidence of an increase in teacher quality in the first years of experience but little evidence of a continued improvement after the first three years (Rivkin et al., 2005), we also conduct an additional investigation by including a quadratic term for teacher experience. This,

Table 4

Determinants of Educational Attainment – Baseline Results

Variables (1) (2) (3)

Reading Science Mathematics

% of immigrants in class �0.21 �0.40 0.74 (0.50) (0.43) (0.73)

Age of student �7.62*** 19.71*** 27.03*** (2.57) (1.69) (2.30)

Gender of student: 1 if female 6.85* �19.49*** �10.89*** (3.47) (1.96) (2.18)

Number of books at home: 11–25 2.47 17.25*** 17.67*** (5.13) (5.37) (6.27)

Number of books at home: 26–100 13.93** 33.00*** 37.56*** (5.54) (5.46) (6.23)

Number of books at home: 101–200 15.51*** 47.25*** 50.77*** (5.27) (5.16) (6.58)

Number of books at home: more than 200 27.51*** 51.30*** 50.97*** (5.85) (5.22) (6.59)

Teaching experience (in years) �0.53 0.99 1.32* (0.53) (0.60) (0.79)

Teacher is female 7.16 �10.13 �25.24** (5.51) (6.48) (10.04)

Age teacher: 30–39 9.19 �6.52 �20.59 (9.87) (8.68) (12.96)

Age teacher: 40–49 14.93 �24.71** �38.54** (10.85) (10.97) (15.94)

Age teacher: 50 or above 19.78 �31.68** �45.85** (14.35) (15.21) (21.02)

Class-size: 1–19 students 4.03 (12.52)

Class-size: 20–26 students �13.64 (9.34)

Observations 1,320 3,664 3,664 R2 0.063 0.151 0.155 Number of schools 55 129 129

Notes. Immigrant children are first-generation immigrants. The Table presents baseline regressions with school fixed effects. The dependent variables are the test scores on reading/science/mathematics. The data sets employed are the 2001 and 2006 PIRLS for reading test scores and 1995 and 2007 TIMSS for mathematics and science test scores. See main text for more details. Robust standard errors are presented in parentheses. ***p < 0.01, **p < 0.05, *p <0.10.

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F321

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

however, does not affect the magnitude of the spillover effects. Older teachers enhance students’ reading scores, but younger teachers seem to be better at teaching mathematics and science classes.

4.3. Sensitivity Analysis

Table 5 reports results from various sensitivity analyses. Panel (a) presents the baseline estimates, which are taken from the first row of Table 4 for the purpose of comparison with the rest of the estimates.

The results in panels (a) to (c) help us illustrate the importance of controlling for the observed family, classroom and the unobserved school characteristics. Row (b) indicates that not controlling for any covariates or school differences leads all estimates to be negative and significant. For science, the estimate is similar to the baseline result although the estimate is now significant. Therefore, omitting all characteristics from the regression reduces the standard errors for the share of immigrants but hardly affects the magnitude of the estimated effect. However, for reading scores, ignoring the school fixed effects leads to a substantially larger and more negative effect of the share of immigrant children in the classroom while for mathematics, there is a change in sign from positive to negative. In panel (c), family characteristics and classroom characteristics are included but school fixed effects are ignored. This results in the parameter estimates to be close to those in panel (b). In other words, the magnitude of the spillover effects from the share of immigrant children to the educational attainment of native Dutch children is not very much affected by family characteristics and observed school characteristics but it is the unobserved school characteristics that eliminates the observed negative effects.

Although second-generation immigrant students are less likely to suffer linguisti- cally, the number of second-generation immigrant students in the Netherlands is larger than that of the first-generation immigrant students. We, therefore, conduct a separate sensitivity analysis to investigate how the combined share of both first and second-generation immigrant students in classroom affects our results. Here, we define immigrant students as having at least one immigrant parent regardless of where they were born. Panel (d) shows that including second-generation immigrants in the immigrant share calculation hardly makes a difference for spillover effects on reading test scores. The effects of the share of immigrant children on science and mathematics test scores, on the other hand, are more positive when second- generation immigrant students are also included in the calculation. More specifically, the estimated effect on the science test scores becomes positive and closer to zero while the effect on mathematics becomes positive and even significantly different from zero.

In addition, if immigrant students are born from one as opposed to two immigrant parents, this may affect the severity of the spillover effects. This is likely to be the case if having at least one Dutch parent improves the child’s linguistic ability or helps the child to assimilate in the Dutch culture better. Therefore, an additional sensitivity check is conducted in panel (e). In this panel, we define immigrant children as children who have two immigrant parents irrespective of where they themselves were born. The magnitude of the parameter estimates is very much the same as those in

F322 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

T a b le

R eg re ss io n E st im a te s fr om

V a ri ou s S en si ti v it y A n a ly se s

V a ri ab

le s

R e a d in g

S ci e n ce / M at h e m a ti cs

R e a d in g

S ci e n ce

M a th e m a ti cs

E x p la n a to ry

va ri a b le s

S ch

o o l

F E

O b se rv a ti o n s

S ch

o o ls

O b se rv a ti o n s

S ch

o o ls

(a ) B a se li n e e st im

a te s

�0 .2 1

�0 .4 0

0 .7 4

Y e s

1 ,3 2 0

5 5

3 ,6 6 4

1 2 9

(0 .5 0 )

(0 .4 3 )

(0 .7 3 )

(b ) U n co

n d it io n a l (O

L S )

�0 .8 0 * * *

�0 .4 1 * *

�0 .4 5 * *

N o

1 ,3 2 0

5 5

3 ,6 6 4

1 2 9

(0 .2 6 )

(0 .1 8 )

(0 .2 2 )

(c ) W it h sc h o o l ch

a ra ct e ri st ic s

�0 .7 3 * * *

�0 .3 2 * *

�0 .3 9 *

Y e s

N o

1 ,3 2 0

5 5

3 ,6 6 4

1 2 9

(0 .2 5 )

(0 .1 6 )

(0 .2 0 )

(d ) %

F ir st

a n d se co

n d -g e n e ra ti o n

im m ig ra n ts

�0 .2 2

0 .0 5

1 .0 2 * *

Y e s

1 ,3 2 0

5 5

3 ,6 6 4

1 2 9

(0 .2 8 )

(0 .3 1 )

(0 .4 6 )

(e ) %

F ir st

a n d se co

n d -g e n e ra ti o n

im m ig ra n ts

(t w o im

m ig ra n t p a re n ts )

�0 .3 0

�0 .1 3

1 .0 4

Y e s

1 ,3 2 0

5 5

3 ,6 6 4

1 2 9

(0 .4 5 )

(0 .5 5 )

(0 .7 6 )

(f ) A rr iv in g a ft e r a g e fi ve

�0 .3 2

1 .3 1

Y e s

– –

3 ,6 6 4

1 2 9

(0 .7 8 )

(1 .2 9 )

(g ) In cl u d in g p a re n ta l e d u ca ti o n

�0 .4 9

– –

Y e s

1 ,2 5 6

5 5

– –

(0 .5 4 )

(h ) E x cl u d e sc h o o ls th a t g ro u p cl a ss e s

a cc o rd in g to

a b il it y

0 .8 2

�0 .0 2

1 .6 7

Y e s

3 4 1

1 6

– –

(0 .8 3 )

(0 .3 7 )

(1 .2 8 )

(i ) Im

m ig ra n t p a re n ts

w it h lo w

e d u ca ti o n

0 .3 2

– –

Y e s

1 ,2 5 6

5 5

– –

(0 .9 4 )

(j ) Im

m ig ra n ts

w h o d o n o t sp e a k

D u tc h a t h o m e

�0 .2 1

0 .1 3

1 .3 2 *

Y e s

1 ,3 2 0

5 5

3 ,6 6 4

1 2 9

(0 .5 0 )

(0 .4 5 )

(0 .7 6 )

(k ) W it h o u t sc h o o ls si tu a te d in

la rg e

ci ti e s

�0 .3 2

�0 .1 8

0 .1 4

Y e s

1 ,2 3 1

5 2

5 2 1

2 3

(0 .5 1 )

(0 .3 9 )

(0 .4 8 )

N ot es . T h is T a b le

p re se n ts e st im

a te s fr o m

va ri o u s se n si ti vi ty a n a ly se s. T h e d a ta

se ts e m p lo ye d a re

th e 2 0 0 1 a n d 2 0 0 6 P IR

L S fo r re ad

in g te st sc o re s a n d 1 9 9 5 a n d 2 0 0 7

T IM

S S fo r m a th e m a ti cs

a n d sc ie n ce

te st sc o re s. T h e d e p e n d e n t va ri a b le s fo r e a ch

o f th e co

lu m n s a re

th e st a n d ar d is e d te st sc o re s o n re a d in g / sc ie n ce / m a th e m a ti cs .

A si d e fr o m

ro w s (c ) a n d (g ),

th e e x p la n a to ry

va ri a b le s a re

th e sa m e a s u se d in

T a b le

4 . R o w

(c ) in cl u d e s a d d it io n a l sc h o o l- le ve l ch

a ra ct e ri st ic s w h e re a s ro w (g )

in cl u d e s p a re n ta l e d u ca ti o n . S e e th e m a in

te x t fo r m o re

in fo rm

a ti o n . R e su lt s fo r so m e ce ll s a re

m is si n g d u e to

la ck

o f in fo rm

a ti o n in

th o se

p a rt ic u la r ye a rs . S in ce

d a ta

o n a b il it y g ro u p in g o f cl a ss e s is

o n ly

a va il a b le

in 2 0 0 1 P IR

L S a n d 2 0 0 7 T IM

S S , th e sa m p le

si ze s a re

sm a ll e r fo r ro w

(h ).

T h e n u m b e r o f o b se rv a ti o n s fo r

th e sc ie n ce

a n d m a th e m a ti cs

re g re ss io n s in

ro w (h ) a re

5 1 6 (2 3 sc h o o ls ) a n d 3 5 7 (1 5 sc h o o ls ),

re sp e ct iv e ly . R o b u st

st a n d ar d e rr o rs

a re

p re se n te d in

p a re n th e se s.

* * * p < 0 .0 1 , * * p < 0 .0 5 , * p < 0 .1 0 .

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F323

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

panel (d), although the spillover effect on mathematics test scores now becomes insignificantly different from zero.

Our definition of the first-generation immigrants does not differentiate those who arrived in the Netherlands at a very early age from those who came later on. However, a very early exposure to Dutch culture and language may help immigrant children to assimilate better. In such a case, those who arrived early would not face the same difficulties as other first-generation immigrants. In order to see the peer effects of immigrants that are least likely to have been assimilated, the definition of the first-generation immigrants in panel ( f ) only includes those who moved to the Netherlands after the age of five. This information is available only in TIMSS and not in PIRLS. The estimates reported in panel ( f ) show no significant negative impact of the share of immigrant children who came to the Netherlands after the age of five.

In the baseline regressions, the number of books at home dummy variables are included. Although these are not perfect proxies for the parental educational background, they are likely to be highly correlated. Since parental education variables are available in PIRLS, they are included in the regression and the estimate is presented in panel (g). The estimated effects of the share of immigrant children becomes slightly more negative compared to those shown by the baseline estimates, but the effects are still insignificantly different from zero.

The potential grouping of classes by ability is another concern. This is likely to undermine the identification strategy employed in this article, which requires that the students are randomly allocated to classes once the school fixed effects are controlled for. The 2001 PIRLS and the 2007 TIMSS both report whether the classes are grouped according to ability. The estimates reported in panel (h), therefore, only use the sample of schools that report random allocation of students across classes. The reported estimate on the reading test scores is more positive and statistically significant compared to the baseline estimate. The results for the science and mathematics test scores, on the other hand, are similar to the baseline estimates, but the magnitude of the effects are smaller.

The nature of the potential negative spillover effects from the presence of immigrant students may not have to do with their immigrant status but with their low educational background. Therefore, in the sensitivity analysis shown in panel (i), we calculated the proportion of immigrant children in class whose parents have secondary school education or less. As indicated before, we can only do this using the PIRLS data. As shown, the estimated effect on the reading score in panel (i), however, remains insignificant.

A potential reason for the negative spillover effects of studying with immigrant students is their lack of Dutch language proficiency. In panel ( j), we explore this issue by calculating the share of first-generation immigrant students who speak Dutch at home occasionally or not at all. Surprisingly, our conclusions from previous rows still holds here. In particular, we do not observe evidence of negative spillover effects for reading and science. For mathematics test scores, we even observe positive and significant effects.

The discussion in subsection 2.2 highlights more severe segregations of Dutch students from immigrants in larger cities. Panel (k), therefore, presents estimates without schools located in cities with more than 500,000 residents to limit the size of the bias stemming from the segregation of students. The results presented, however, do not alter our conclusions from earlier estimates. Cities with a population over 500,000

F324 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

residents include Amsterdam, Rotterdam and the Hague. We also conducted a separate analyses by excluding Dutch students residing in cities with more than 100,000 residents. Our estimates indicate the same conclusions regardless of the population cut-off points.

So far, we have focused on the spillover impacts of studying with immigrant students in the same classroom and find no significant effects. Our data also allow us to investigate whether immigrant students affect school learning environment of native students. Table 6 presents evidence on whether students encounter difficulties in learning places by looking at how the incidence of bullying is influenced by the share of immigrant students in a classroom. The dependent variables are dummy variables that denote negative experiences at schools reported by students. These experiences include whether students were ever bullied at school, if their possessions were ever stolen, if the students ever felt left out or made fun of and if they were ever hit by other students. Columns (1) and (2) were estimated using the 2001 and 2006 PIRLS, whilst columns (3)–(6) are estimated using the 2007 TIMSS. The estimates from linear probability models with school fixed effects show that native Dutch students experience increased incidents of bullying when they are studying with more immigrant students in the same classroom.

5. Conclusions

Many immigrants have entered European countries in recent decades. Initially, most immigrants were workers attracted by the favourable economic circumstances, entering the labour markets to fill vacancies that were difficult to fill by native workers. Later on, immigrants entered Europe because of family reunion, family formation or because they were seeking asylum. Nowadays, immigration is on top of the political agenda in many European countries. From an economic point of view, an important question is how immigrants affect the economy, in particular the functioning of the labour

Table 6

The Peer Effect of Immigrant Students on the School Learning Environment of Dutch Students

PIRLS TIMSS

(1) (2) (3) (4) (5) (6) Variables Stolen Bullied Stolen Feel left out Made fun of Hit by others

Proportion of immigrants in class

0.61*** 0.57** �0.20 1.15*** 0.46 0.75** (0.20) (0.26) (0.23) (0.40) (0.37) (0.33)

Observations 1,320 1,296 548 546 548 546 Number of schools 55 55 26 26 26 26

Notes. This Table presents the estimated peer effects of first-generation immigrant students on the school learning environment using linear probability models with school fixed effects. Columns (1) and (2) show the impacts on the Dutch students observed in PIRLS whilst columns (3)–(6) illustrate the effect on the immigrant students observed in TIMSS. The dependent variables are the variables that capture negative school experiences. Data employed for this Table is the 2001 and 2006 PIRLS and 2007 TIMSS. For TIMSS, the dependent variables are only available in 2007. To make the interpretation easier, the key independent variable is divided by 100. The variable in this Table, therefore, is the ‘proportion of immigrants in class’ as opposed to the previous ‘% of immigrants in class’ in Table 4. The other independent variables included are identical to those in Table 4. Robust standard errors are presented in parentheses. ***p < 0.01, **p < 0.05, *p < 0.10.

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F325

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

market. However, now that the immigrants have become a substantial part of the population, the focus of research shifts to questions like the integration of immigrants in the society. Research in the area has been done on educational attainment of children showing, for example, that second-generation immigrants do much better than first-generation immigrants thus closing the educational gap between immigrant children and native children.

A relatively new area of research concerns the presence of educational spill-over effects from immigrant children to native children. This is the main topic of the current article. We analyse how the share of first-generation immigrant children in the classroom affects the educational attainment of native Dutch children in the same classroom. In our analysis, we use data from various sources which allow us to characterise educational attainment in terms of reading literacy, mathematical skills and science skills.

A major identification problem when establishing potential educational spill-over effects is related to student selection into schools. If schools with a relatively high share of immigrant children attract Dutch children whose educational skills are different from those in schools with a relatively low share of immigrant children, we might erroneously conclude that the presence of immigrant children has negative spill-over effects on Dutch children. We solve this potential selectivity by investigating within school variation. We compare the educational attainment of Dutch children with different shares of immigrant children in classrooms within the same school. Within the same school, there may still be a selectivity issue due to potential non-random allocation of students to classes or higher allocation of teaching resource to classes with more immigrant children. However, we find no evidence that support either of the concerns.

Our analysis indicates that native Dutch students may be experiencing a worse learning environment as we find, for example, increased incidents of bullying with more immigrant students in the classroom. Nevertheless, overall we do not find strong evidence of negative spill-over effects on the test scores from immigrant children to native Dutch children. In our baseline and preferred specification, all parameter estimates are small and insignificant. For the mathematics scores, the spillover effect is positive; for the reading test score and the science and mathematics scores, the parameter estimates are negative but small. A 1 percentage point increase in the classroom share of immigrant children leads to an insignificant drop of less than 0.5 points in the test scores. We conclude that for native Dutch students, there is no urgent need to redistribute immigrant children more evenly across classrooms as their educational attainment does not seem to be strongly affected by the presence of these children.

Appendix A. Details on the Data

The two data sets employed in this article were collected in the Netherlands through the 2001 and 2006 PIRLS and the 1995 and 2007 TIMSS. They were both designed and conducted by the IEA to study the reading, mathematics and science achievements of nine and ten-year-old students. The samples of students in both surveys were selected using a two-stage sampling design. In the first stage, schools were selected using a probability-proportional-to-size sampling

F326 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

T a b le

A 1

S u m m a ry

S ta ti st ic s

V a ri a b le s

P ro p o rt io n o f im

m ig ra n t st u d e n ts

in cl as s (%

)

(a ) R e a d in g

(b ) S ci e n ce / M a th e m a ti cs

(1 )

(2 )

(3 )

(4 )

(5 )

(6 )

(7 )

(8 )

T o ta l

0 – 4 %

5 – 9 %

1 0 %

o r m o re

T o ta l

0 – 4 %

5 – 9 %

1 0 %

o r m o re

R e a d in g te st

sc o re s

5 6 3 .6 9

5 6 5 .3 0

5 6 6 .7 1

5 4 5 .1 9

S ci e n ce

te st

sc o re s

5 4 0 .0 6

5 4 2 .4 3

5 3 8 .2 7

5 2 5 .0 6

M a th e m a ti cs

te st

sc o re s

5 4 7 .1 4

5 4 9 .8 3

5 4 5 .4 8

5 2 9 .5 7

A g e o f st u d e n t

9 .7 5

9 .7 3

9 .7 2

9 .3 2

9 .3 1

9 .3 5

9 .3 3

G e n d e r o f st u d e n t: 1 if fe m a le

0 .4 9

0 .4 6

0 .5 1

0 .5 0

0 .5 2

N u m b e r o f b o o k s a t h o m e : 0 – 1 0

0 .1 0

0 .1 4

0 .0 4

0 .0 3

0 .0 4

0 .0 5

N u m b e r o f b o o k s a t h o m e : 1 1 – 2 5

0 .1 6

0 .1 3

0 .1 2

0 .1 5

0 .1 4

0 .1 7

0 .1 6

N u m b e r o f b o o k s a t h o m e : 2 6 – 1 0 0

0 .3 1

0 .3 0

0 .3 3

0 .3 2

0 .3 8

N u m b e r o f b o o k s a t h o m e : 1 0 1 – 2 0 0

0 .1 8

0 .1 7

0 .2 7

0 .2 1

0 .2 3

0 .2 4

0 .2 0

0 .1 9

N u m b e r o f b o o k s a t h o m e : m o re

th a n 2 0 0

0 .2 5

0 .2 6

0 .2 0

0 .2 3

0 .2 6

0 .2 7

0 .2 1

T e a ch

in g e x p e ri e n ce

(i n ye a rs )

1 6 .9 8

1 7 .7 8

1 2 .6 3

1 5 .0 3

1 7 .1 8

1 6 .8 9

1 8 .3 2

1 7 .7 7

1 if fe m a le

te a ch

e r

0 .6 8

0 .6 4

0 .8 6

0 .8 4

0 .4 3

0 .4 0

0 .5 3

A g e te ac h e r le ss

th a n 3 0

0 .2 8

0 .2 6

0 .4 4

0 .3 2

0 .1 7

0 .1 6

0 .2 1

0 .1 9

A g e te ac h e r 3 0 – 3 9

0 .1 7

0 .1 8

0 .0 7

0 .2 2

0 .2 5

0 .2 8

0 .1 1

0 .1 8

A g e te ac h e r 4 0 – 4 9

0 .2 2

0 .2 1

0 .2 9

0 .2 4

0 .3 8

0 .4 3

0 .3 4

A g e te ac h e r 5 0 o r a b o ve

0 .3 3

0 .3 5

0 .2 0

0 .2 2

0 .1 9

0 .1 6

0 .2 5

0 .2 9

C la ss

si ze

1 – 1 9

0 .1 8

0 .1 9

0 .1 4

0 .1 9

C la ss

si ze

2 0 – 2 6

0 .5 8

0 .5 5

0 .7 0

0 .6 5

C la ss

si ze

2 7 a n d m o re

0 .2 4

0 .2 6

0 .1 5

0 .1 6

N u m b e r o f o b se rv a ti o n s

1 ,3 2 0

1 ,0 6 1

1 4 1

1 1 8

3 ,6 6 4

2 ,7 5 9

5 2 7

3 7 8

N ot e. T h is T a b le

p re se n ts

su m m a ry

ch a ra ct e ri st ic s b y im

m ig ra n t p ro p o rt io n u si n g th e 2 0 0 1 , 2 0 0 6 P IR

L S a n d 1 9 9 5 a n d 2 0 0 7 T IM

S S d a ta

se ts .

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F327

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

scheme. In the second stage, one or multiple classes were randomly sampled from each of the selected schools.

Table A1 presents summary statistics for the observations from PIRLS and TIMSS by the proportion of immigrant students in the same classroom. The Table indicates that the average test scores decline as the proportion of immigrant students in classroom increases. This is true for all subjects. Dutch students, who attend schools with a limited number of immigrant students, come from households with more books at home. This may indicate that students studying with fewer immigrant students in the same classroom come from higher socioeco- nomic backgrounds. Table A1 also provides evidence on the potential impact of the extra funding for the immigrant students. In PIRLS, schools with a large proportion of immigrant students have smaller reading class sizes. In addition, reading classes with more immigrant students are more likely to be taught by female teachers who are younger. Such patterns do not exist for mathematics and science classes.

Finally, Table A2 presents the number of schools in PIRLS and TIMSS with the corresponding number of classes. The regression analysis in this article only use student data when they study in schools with two or more classes.

Appendix B. On the Random Allocation of Immigrant Students

To investigate whether immigrant students in our data were allocated randomly across classes within a school, we performed the following exercise. Consider a case where all schools have two classes each. Let A and B be the first and second class in each school, respectively. Moreover, define n to be the number of immigrant students in the age 9–10 cohort of each school. d denotes the difference in the number of immigrant students between the classes.

(i) When n = 1, the student can be allocated to one class only:

Immigrant student d

A 1 B 1

Therefore, the difference in the number of immigrant students between the two classes will always be equal to 1 (i.e. Prob(d = 1)= 1). (ii) If there are two immigrant students, they may be in the same class or in different classes:

Table A2

Number of Schools with one or More Multiple Classes

PIRLS TIMSS Number of classes Number of schools Number of schools

1 106 71 2 47 102 3 7 12 4 1 14 5 0 1

F328 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

Therefore, Prob(d = 0)= 0.5 and Prob(d = 2) = 0.5. There is a 50% probability that there is an equal number of students in both classes and a 50% probability that the difference is two.

The above two examples illustrate that

(i) If there are n immigrant students in a school, there are 2n ways to allocate the n immigrant students to either of the two classes. (ii) If n is an odd (even) number, d also takes a series of odd (even) numbers.

Generalisations of the above examples are as follows.

(i) If n is an even number

d ¼ 0 : Probðd ¼ 0Þ ¼ n n

! 1

� �n ;

d � 2 : Prob � d ¼ 2kjk 2 Nþ; k � n

� ¼ 2

n n 2 � k

! 1

� �n :

(ii) If n is an odd number

8d : Prob � d ¼ 2k þ 1jk 2 N; k � n � 1

� ¼ 2

n � 1 2

� k

! 1

� �n :

We use the probabilities calculated with the above formulae together with the number of schools with each number of immigrant students in the cohort. The information allows us to obtain the expected number of schools for each difference in the number of immigrant students between classes within the same school. In the Table above, we show the predicted numbers of schools

Immigrant student

1 2 d

A A 2 A B 0 B A 0 B B 2

PIRLS TIMSS

Differences in the number of immigrant students

(1) Predicted

(2) Actual

(3) Predicted

(4) Actual

0 23.3 29 62.4 65 1 14.9 9 23.7 24 2 5.9 2 10.7 9 3 1.7 4 3.0 3 4 0.7 2 1.6 1 5 0.3 0 0.3 0 6 0.1 1 0.3 0

Number of schools 47 47 102 102

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F329

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

(column (1) for PIRLS and column (3) for TIMSS), for corresponding differences in the number of immigrant students between two classes ranging from 0 to 6. Columns (2) and (4) each shows the actual observed numbers of schools for PIRLS and TIMSS, respectively (For the purpose of keeping the calculations simple, we are conducting this exercise only for schools with two classes; see Appendix A for the tabulations of the number of classes in PIRLS and TIMSS).

We can compare these distributions by calculating Fisher’s exact tests to take account of the small numbers in the cells. We consequently find a p-value of 0.271 for PIRLS and 0.988 for TIMMS. From these p-values, we conclude that both predicted distributions do not differ significantly from the actual distributions. The results, therefore, suggest that the allocation of immigrant students over classes within the same school indeed occurred randomly.

Tilburg University Tilburg University, University of Melbourne and CESifo CEPR and IZA

Submitted: 30 April 2012 Accepted: 1 April 2013

References Ammermueller, A. and Pischke, J. (2009). ‘Peer effects in European primary schools: evidence from the

Progress in International Reading Literacy Study’, Journal of Labor Economics, vol. 27(3), pp. 315–48. Betts, J.R. (1998). ‘Educational crowding out: do immigrants affect the educational attainment of American

minorities?’, in (D. Hamermesh and F. Bean, eds.), Help or Hindrance: The Economic Implications of Immigration for African Americans, pp. 253–81, New York: Russell Sage Foundation.

Betts, J.R. and Lofstrom, M. (2000). ‘The educational attainment of immigrants’, in (G. Borjas, ed.), Issues in the Economics of Immigration, pp. 51–116, Chicago, IL: University Of Chicago Press.

Borjas, G.J. (2004). ‘Do foreign students crowd out native students from graduate programs?’, National Bureau of Economic Research Working Paper Series.

Brunello, G. and Rocco, L. (2011). ‘The effect of immigration on the school performance of natives: cross country evidence using PISA test scores’, Economics of Education Review, vol. 32, pp. 234–46.

Bui, S.A. (2012). ‘How do limited English proficient students affect each others educational outcomes? Evidence from student panel data’, University of Houston Working Paper.

Card, D. (2013). ‘Peer effects of immigrant children on academic performance of native speakers: introduction’, ECONOMIC JOURNAL, vol. 123(570), pp. F279–80.

Ersanilli, E. (2007). ‘Focus migration-country profile: the Netherlands’, mimeo, Hamburg Institute of International Economics (HWWI).

Friesen, J. and Krauth, B. (2011). ‘Ethnic enclaves in the classroom’, Labour Economics, vol. 18(5), pp. 656–63. Geay, C., McNally, S. and Telhaj, S. (2013). ‘Non-native speakers of English in the classroom: what are the

effects on pupil performance?’, ECONOMIC JOURNAL, vol. 123(570), pp. F281–307. Gonzalez, E.J. and Kennedy, A.M. (2003). ‘PIRLS 2001 user guide for the international database’,

International Study Center, Lynch School of Education, Boston College, Boston. Gould, E.D., Lavy, V. and Paserman, D.M. (2009). ‘Does immigration affect the long-term educational

outcomes of natives? Quasi-experimental evidence’, ECONOMIC JOURNAL, vol. 119(540), pp. 1243–69. Hoxby, C.M. (1998). ‘Do immigrants crowd disadvantaged american natives out of higher education’,

in (D. Hamermesh and F. Bean, eds.), Help or Hindrance: The Economic Implications of Immigration for African Americans, pp. 282–321, New York: Russell Sage Foundation.

Hunt, J. (2012). ‘The impact of immigration on the educational attainment of natives’, IZA Discussion Paper No. 6904.

Jensen, P. and Rasmussen, A. (2011). ‘The effect of immigrant concentration in schools on native and immigrant children’s reading and math skills’, Economics of Education Review, vol. 30(6), pp. 1503–15.

Karsten, S., Elshof, D., Felix, C., Ledoux, G., Meijnen, W., Roeleveld, J. and Schooten, E. (2003). Onderwijssegregatie in Amsterdam, Amsterdam: SCO-KI.

Ladd, H.F. and Fiske, E.B. (2009). ‘Weighted student funding for primary schools: an analysis of the Dutch experience’, Sanford School of Public Policy Working Paper, Duke University, SAN09-02.

Ladd, H.F., Fiske, E.B. and Ruijs, N. (2010). ‘Parental choice in the Netherlands: growing concerns about segregation’, Stanford School of Public Policy, Working Paper Series Duke University, SAN10-02.

Lucassen, J. and Penninx, R. (1997). Newcomers: Immigrants and Their Descendants in the Netherlands 1550–1995, Amsterdam: Het Spinhuis.

F330 T H E E C O N O M I C J O U R N A L [ A U G U S T

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019

Manski, C.F. (1993). ‘Identification of endogenous social effects: the reflection problem’, Review of Economic Studies, vol. 60(3), pp. 531–42.

Ohinata, A. and van Ours, J.C. (2011). ‘How immigrant children affect the academic achievement of native Dutch children’, Discussion Paper CentER 11-136, Tilburg University.

Ohinata, A. and van Ours, J.C. (2012). ‘Young immigrant children and their educational attainment’, Economics Letters, vol. 116(3), pp. 288–90.

van Ours, J.C. and Veenman, J. (2003). ‘The educational attainment of second-generation immigrants in the Netherlands’, Journal of Population Economics, vol. 16(4), pp. 739–53.

van Ours, J.C. and Veenman, J. (2005). ‘The Netherlands: old emigrants-young immigrant country’, in (K. Zimmermann, ed.), European Migration: What Do We Know?, pp. 173–96, Oxford: Oxford University Press.

Penninx, R., Schoorl, J. and Van Praag, C. (1994). The Impact of International Migration on Receiving Countries: the Case of the Netherlands, The Hague: Netherlands Interdisciplinary Demographic Institute.

Rivkin, S.G., Hanusek, E.A. and Kain, J.F. (2005). ‘Teachers, schools, and academic achievement’, Econometrica, vol. 73(2), pp. 417–58.

2013] A C A D E M I C P E E R E F F E C T S O F I M M I G R A N T C H I L D R E N F331

D ow

nloaded from https://academ

ic.oup.com /ej/article-abstract/123/570/F308/5080403 by U

niversity of O klahom

a user on 21 N ovem

ber 2019