Labor market outcomes of immigrants and non-citizens in the EU

An East-West comparison

The Authors

Martin Kahanec, IZA, Bonn, Germany

Anzelika Zaiceva, IZA, Bonn, Germany, and University of Bologna, Bologna, Italy

Acknowledgements

JEL classification – F22, J15, J61, J71. Financial support from the Volkswagen Foundation for the IZA project on “The Economics and Persistence of Migrant Ethnicity” is gratefully acknowledged.

Abstract

Purpose – The purpose of this paper is to comparatively analyse the roles of foreign origin and citizenship in the labor markets of Eastern and Eestern Member States of the EU.

Design/methodology/approach – The EU Survey of Income and Living Conditions is used to evaluate the roles of foreign origin and citizenship on employment and earnings using the standard Probit and OLS econometric models. The native/non-native labor market divide is measured using Fairlie and Oaxaca-Blinder decomposition techniques.

Findings – The results indicate that, while predominantly foreign origin is of key importance in the Western EU Member States, both foreign origin and citizenship matter in the Eastern EU Member States, their roles depending on gender. Moreover, the evidence suggests that the effects of citizenship in the EU8 may be driven by the (predominantly ethnic Russian) non-citizens in Estonia and Latvia.

Research limitations/implications – Further analysis is necessary to evaluate the observed associations as causal relationships.

Originality/value – The study is the first to shed light on the role of foreign origin and citizenship in the EU8 and the EU15 in the comparative East-West perspective. The findings have noteworthy implications for the targeting of national as well as EU-wide integration policies.

Article Type:

Research paper

Keyword(s):

Immigrants; Citizenship; Earnings; Employment; Labour market; Eastern Europe.

Journal:

International Journal of Manpower

Volume:

Number:

1/2

Year:

2009

pp:

97-115

Copyright ©

Emerald Group Publishing Limited

ISSN:

0143-7720

1. Introduction

How people born outside the country of their current residence – immigrants – fare in the Western European labor markets in terms of their earnings, employment, self-employment and other labor market outcomes has been the focus of a large body of literature. A complementary literature looks at residents without the citizenship of the host country – non-citizens. In contrast, the labor market fates of non-natives[1] in Eastern Europe have received scarce attention. Yet the populations in this region are far from monoethnic, and the related labor market issues are far from trivial[2]. This paper evaluates and compares the labor market performance of immigrants and non-citizens in the new Eastern Member States of the European Union that accessed in 2004 (EU8) to that in the long-standing Western EU Member States (EU15)[3].

The histories of immigrants and non-citizens in Western and Eastern Europe in the post-war period differ significantly. There has been a large influx of foreign workers and their families into Western European countries since the 1960s, fueled by the need to sustain the post-war economic growth. This inflow has followed post-colonial linkages as in the case of France or the UK, or new immigration patterns emerged, such as those facilitated by immigrant treaties between the former West Germany and the source countries including Turkey, Spain and Yugoslavia. Although some of the immigration channels were discontinued following the 1973 oil shock, migration went on through family reunification, increased fertility rates, and asylum seekers and refugees. Whether these immigrants have become citizens or not largely depends on the specific legal arrangements in a given country. More recently, European migration has received two strong impetuses: the economic success in traditional emigration countries that turned them into countries of immigration (e.g. Ireland, Spain and Greece); and European integration and EU enlargement that removed barriers to migration within an increasing number of European countries[4].

In contrast, Central and Eastern Europe, under the yoke of communist regimes, experienced very little international migration prior to the 1990s. However, the Baltic republics of the former Soviet Union, Estonia and Latvia in particular, received substantial inflows of mostly Russian speaking people (ethnic Russians) during the forceful industrialization campaigns of the Soviet central government and as a result of circulation of military and administration personnel in the Soviet Union (Laitin, 1998). Upon the independence of these countries in 1991, the non-natives (mostly ethnic Russian) emerged as sizeable minorities, whose members frequently lack citizenship due to the restrictive citizenship laws (Brubaker, 1992). At the same time, an “ethnic unmixing” took place in many former Soviet Union countries with migration flows of ethnic Russians to Russia, ethnic Germans to Germany and ethnic Jews to Israel. Nevertheless, according to the respective national census data, the Russian minority constitutes 25.6 percent in Estonia (in 2000), 29.6 percent in Latvia (in 2000) and only 6.3 percent in Lithuania (in 2001)[5]. Leping and Toomet (2007) report that non-Estonians in Estonia are concentrated in urban industrial centers, certain industrial sectors (mining, manufacturing, energy and logistics) and certain occupational groups (crafts). Similar patterns are reported by Hazans (2007) for Latvia.

During the 1990s, EU8 countries started to receive some inflows of economic immigrants, including skilled professionals that accompanied the inflows of foreign direct investments. With improving economic conditions and after accession to the EU, EU8 countries are becoming an even more attractive destination region for migrants from less prosperous countries, predominantly further East in Europe or Asia, including Russia, Ukraine, Vietnam and China. Clearly, the different nature of the non-natives arriving to EU8 since the 1990s and the non-native predominantly ethnic Russian people residing in Estonia and Latvia make a case for distinguishing these two countries in the analysis.

Given the institutional and historical variation of migration trajectories, the legal definitions of immigrants and citizens vary across Europe (Zimmermann et al., 2007). In Germany and France, for example, most datasets identify non-natives by nationality (i.e. citizenship). The Danish statistical office defines a person as a Dane if at least one of his or her parents is a Danish national and was born in Denmark. The British census does not group people by country of origin, but by ethnicity (including “race”, skin color, national and regional origins, and language). In the EU8, the census data report self-identified ethnicity as the main criterion identifying ethnic groups. In Estonia and Latvia, large proportions of non-Estonians and non-Latvians, predominantly of Russian origin, lack citizenship.

The main question that this paper addresses in a comparative perspective is whether and how these institutional and historical differences manifest themselves in the labor market outcomes – earnings and employment – of non-natives in the EU15 and the EU8[6]. We do not aspire to evaluate the causal relationships, rather, we highlight the differences in the roles of immigrant and citizenship status in EU15 and EU8 labor markets in an exploratory and descriptive manner[7]. In the next section we summarize the literature, and then introduce the EU SILC dataset used in this study that offers a unified definition of immigrants and non-citizens across all EU Member States. We then briefly describe the main features of the native and non-native populations across Europe, and quantify and compare the effects of being an immigrant or a non-citizen in EU15 and EU8 using standard OLS and Probit models. Finally, to measure the divide engendered by immigrant or citizenship status in EU15 and EU8 labor markets we decompose the raw outcome differentials between native and non-native groups into the part that is explained by observable characteristics and the unexplained part. The latter may reflect discrimination but also differences in ethnic capital or the character of institutional and own selection of non-natives into different countries or citizenship statuses. We then discuss the roles that foreign origin and citizenship play in EU15 and EU8 and provide suggestions for further research.

2. Related literature

The early literature on the position of immigrants in the earnings distribution includes Chiswick (1978) and Borjas (1985, 1990, 1995). Chiswick et al. (2008) investigate immigrant earnings in an international perspective. Zimmermann (2005) discusses what we know about the European immigrant ethnic minorities. Adsera and Chiswick (2007) scrutinize the gender and country of origin differences in immigrant labor market outcomes across European destinations. The employment gap between immigrants and natives is evidenced, e.g. by Amuedo-Dorantes and de la Rica (2007) for Spain; and Simpson et al. (2006), and Kahanec and Mendola (2009) for the UK. That immigrant ethnic minorities with the same characteristics as natives typically have lower labor market returns is documented, for example, by van Ours and Veenman (2005) for the Netherlands and Aeberhardt et al. (2007) for France. Constant et al. (2005) discuss immigrant labor market adjustment in France.

As concerns citizenship, Bratsberg et al. (2002) find positive effects of naturalization on wages in the USA. Fougère and Safi (2006) find that naturalization has a strong positive effect on the employment probability of immigrants in France. Bevelander (2000), however, finds that naturalization has a negative effect on economic activity in Sweden. For Denmark, Constant and Zimmermann (2005) find no effect of naturalization on the probability of working but a significant positive effect on earnings, conditional on working. For Germany, they find that naturalized immigrants are more likely to work in paid-employment, less likely to go into self-employment, but they earn more in both self- and paid- employment than the non-naturalized ones. Constant (1998), however, does not find any positive effects on earnings of guest workers in Germany.

Constant et al. (2006a) measure the Russian-Ukrainian earnings divide in Ukraine, presenting a pioneering insight into the role of ethnicity for labor market outcomes in Eastern Europe[8]. This study finds a statistically significant and growing earnings premium for being a Russian speaker in Ukraine. The scarce literature on Eastern EU members includes Hazans (2007), who examines the differences in wages and occupational distribution, and Hazans et al. (2007), who analyze the differences in the duration of unemployment between the Latvian majority and non-Latvian (mainly Russian-speaking) minority. Both studies report significant negative effects of being of non-Latvian on labor market outcomes. Leping and Toomet (2007) investigate the earnings gap between Estonian and non-Estonian (mainly Russian-speaking) workers in Estonia, reporting a growing earnings advantage of ethnic Estonians.

3. The data

The data that we use in this paper come from the EU Survey of Income and Living Conditions (SILC) for 2005. This survey was launched by Eurostat in 2004 for 13 EU countries, and its 2005 wave included new EU Member States for the first time. The data collected is based on a nationally representative probability sample of the population (both households and individuals) residing in private households within a Member State. A specific regulation defines minimum effective sample sizes for each country and for both the cross-sectional and the longitudinal components of the data. Sampling design varies across countries and includes stratified two-stage design and a simple random sampling among others. Personal-level information is obtained from individual interviews with all household members aged 16 and over. The dataset contains a rich set of socio-economic variables as well as information on immigration and citizenship status for both old and new EU members, which is crucial for the purpose of our paper. Moreover, in this dataset it is possible to distinguish between EU and non-EU origins of immigrants[9].

Table I reports the proportions of foreign-born and foreign citizens in the EU countries from our dataset as well as from the Eurostat Population Statistics 2006 (Eurostat, 2006). When comparing the figures from the two sources it has to be kept in mind that the time periods as well as definitions may differ. The countries with the largest proportion of foreign-born, Luxembourg apart, are Latvia, Estonia, Sweden, Belgium and Austria. Latvia, Estonia, Belgium and Austria exhibit the largest share of non-citizens. The table also suggests that, in the majority of countries, the largest share of immigrants were of non-EU origin.

The dependent variables that we analyze in this paper are respondent's employment status and hourly earnings. The former measures whether the respondent is employed or unemployed using the information from the activity status variable on the dominant labor market activity during the reference period constructed by Eurostat. The later was generated from the reported employee cash or near cash income in Euros per year and usual weekly hours worked[10]. We use gross employee earnings in order to mitigate the potential effects of differences in tax systems between EU15 and EU8 countries[11]. Our key independent variables are immigration and citizenship statuses measuring, respectively, whether an individual is foreign-born and possesses host country's citizenship, or not. In a later stage we also distinguish between EU and non-EU origin, since, in contrast to non-EU origin, EU origin grants the individual a number of rights and services (i.e. free movement and social security) that might have implications for the role of citizenship and foreign origin on one's labor market performance in an EU country. The set of independent control variables includes human capital, measured by educational attainment and potential experience, and its square, which is calculated depending on the highest educational degree obtained following Adsera and Chiswick (2007). We also control for marital status, household size, presence of children, and the health status of the respondent. In addition, the earnings equation includes supervisory position, firm size, occupation and sectoral dummies[12].

In the final sample we include men between 18 and 60-years-old and women between 18 and 55-years-old[13]. and exclude those still in education, disabled or in military service and with missing information on the key explanatory variables. This leaves us with 154,968 observations. In earnings equations we include only those working full-time, and exclude the self-employed and outliers with respect to earnings (the lowest and highest 2 percent from the country-specific earnings distributions) and include additional supervisory position, firm size, industry and occupation controls, reducing the sample to 83,855 observations. We drop Slovenia from the analysis due to a lack of data on citizenship, and Luxembourg due to its specific immigrant population[14].

The summary statistics of the key variables are reported in Table II. The main observations are that non-immigrant citizens are the group with the highest probability of employment among males and females in the EU15, while in the EU8 it is immigrant citizens for both genders. Regarding earnings, in the EU15 immigrant citizens are at the top for both genders, while the lowest earnings are reported by immigrant non-citizens among males and non-immigrant non-citizens among females[15]. In the EU8, non-immigrant non-citizens are the lowest earners for both genders. For males it is citizens (both immigrants and non-immigrant) who earn most, while for females it is non-immigrant citizens.

Both male and female immigrant non-citizens are the youngest and least experienced in the EU15, while it is non-immigrant non-citizens in the EU8. In both regions non-immigrant non-citizens have the highest proportion with secondary education and the lowest proportion with tertiary education (except for females in EU8, for whom it is non-immigrant citizens). Finally, immigrants have higher proportions with tertiary education than non-immigrants for the most part. These findings are consistent with the hypothesis of positive selection (on observables) into migration.

4. Methodology and results

In order to evaluate the links between foreign origin and citizenship on the one hand and employment probability on the other hand in EU15 and EU8, we first estimate the probit binary choice model of the probability of being employed rather than unemployed. In a similar comparative framework, we then consider earnings as another measure of labor market outcomes and estimate Mincerian earnings equations using the standard OLS technique. For both employment and earnings we establish whether and how immigrant and citizenship status matters, and then disaggregate these effects by the EU and non-EU origin of non-natives.

Table III reports the marginal effects from the probit employment probability models. It is immediately evident that foreign origin rather than lack of citizenship bears a penalty in employment for males and females in the EU15. In the EU8, by contrast, it is the lack of citizenship that constitutes a barrier to employment for males, and both foreign origin and citizenship have a negative effect for females. The non-negligible negative effects of being an immigrant and non-citizen in the EU8 signify great employment barriers to people, especially females, which have both of these characteristics. The remaining regressors exhibit anticipated effects.

Table IV reports OLS models of the determinants of hourly earnings in EU15 and EU8 countries. For males in the EU15 the results are essentially the same as in the case of employment: it is foreign origin that matters and negatively affects earnings. In contrast, foreign origin in the EU8 loses significance after we control for citizenship. The results indicate that neither lack of citizenship nor foreign origin alone suffice to inflict a significant earnings penalty, but males having both of these characteristics suffer from a significant earnings penalty (the sum of the effects of being immigrant and non-citizen) of about 7 percent (p=0.002).

For females in the EU8 the results are also similar to those for employment, where both foreign origin and lack of citizenship are disadvantageous, and immigrant non-citizen women are especially vulnerable[16]. Interestingly, foreign origin is not associated with lower earnings in the EU15 in a statistically significant way once citizenship has been controlled for. Again, the other regressors exhibit standard effects.

The findings above, however, may hide important differences stemming from immigrants' origin. Table V decomposes the effects of non-nativity for people of EU and non-EU origins, reporting marginal effects of foreign origin and non-citizenship in models corresponding to those in Tables III and IV. In addition, we also report the analogous results for the EU6 countries, i.e. those new Member States that do not have a significant Russian-speaking ethnic minority[17].

We observe that in the EU15 the negative role of foreign origin on employment is predominantly driven by non-EU immigrants, although male EU immigrants also experience a penalty. Female EU immigrants suffer from an employment penalty only if they are also EU non-citizens (the sum of the two coefficients, p=0.003). In the EU8, the negative effect of foreign origin on female employment is mainly driven by EU immigrants. For males and females in the EU8 the negative role of citizenship is mainly due to non-EU non-citizens. As this effect disappears in the EU6 (and is not present in the EU15), one possible explanation could be that it is driven by the ethnic Russian non-citizens in Estonia and Latvia[18]. In the EU8 the effect of being an EU non-citizen is positive and cancels the negative effect of being EU immigrant for females having both of these characteristics (the sum of the two coefficients, p=0.732).

Concerning earnings in the EU15, the negative effect of foreign origin on male earnings is mainly driven by a non-EU immigrants[19]. The analysis also reveals that the overall insignificant effect of being non-citizen masks the negative role of non-EU citizenship. For females, the negative role of being non-citizen is primarily due to non-EU non-citizens. In the EU8, the negative roles of being an immigrant or non-citizen are mainly driven by non-EU non-natives and become insignificant in EU6 for both males and females. As before, this finding points to the role of non-natives, mainly ethnic Russians, in Estonia and Latvia. The positive role of being an EU non-citizen is consistent with the hypothesis of highly skilled predominantly male expatriates working at subsidiaries of Western multinationals in the EU8.

To summarize this evidence, we find that in EU15 it is especially foreign origin that matters for both employment and earnings of males, and employment of females. Looking at immigrants' origins it turns out that these effects are predominantly driven by non-EU immigrants. Females in the EU15, however, exhibit different patterns concerning their earnings profiles. In particular, for them it is the lack of EU citizenship that bears an earnings penalty.

Besides that being EU immigrant negatively affects male employment in the EU15 and female employment in EU8, foreign origin plays similar roles in the two regions. Unlike in the EU15, however, a lack of EU citizenship has a significantly negative role on employment for both males and females. All the effects of non-EU citizenship found in the EU8 become insignificant or reduced in magnitude if we drop countries with substantial Russian-speaking ethnic minorities. This points at the significant role of citizenship status for labor market outcomes of non-Estonians and non-Latvians in Estonia and Latvia, respectively, and is consistent with the evidence of a labor market disadvantage of these minorities found in other studies. Another interesting finding is the positive role of being a non-citizen of EU origin on female employment and male earnings in the EU8.

The different results found for males and females highlight the role of gender for migration trajectories and labor supply decisions[20]. That foreign origin negatively affects female but not male employment in the EU8 might be driven by various selection channels. For example, high skilled expatriates working at subsidiaries of Western multinationals in the EU8 might cancel the negative effects of foreign origin typically found elsewhere. Since such expatriates are predominantly males, however, negative effects persist for females. In fact, if women come for non-economic reasons as partners of such high-skilled males (i.e. tied movers), they may face difficulties finding employment due to their lack of host country specific skills (i.e. language). From another perspective, some immigrant women of non-EU origin in the EU8 may belong to less skilled occupational categories. Jobs for such women are relatively scarce, since there is not much tradition of female household help or carers, partly because such jobs were viewed as symbols of inequality by the egalitarian socialist regimes and, as such, eradicated. On the other hand, male-dominated sectors such as construction are thriving in the EU8 countries. As a result, low skilled immigrant women might find it harder to find jobs than their male counterparts. Discrimination of women might play role as well.

5. The measured labor market divide

In the analysis above we have assumed that the structural labor market relationships are the same for all native and non-native groups. We relax this assumption here and estimate a Fairlie decomposition of employment probabilities (Fairlie, 2005) and an Oaxaca-Blinder decomposition of earnings differentials (Oaxaca, 1973; Blinder, 1973) using the Neumark (1988) method. Namely, for the latter, we decompose earnings differentials between pairs of native and non-native groups as follows:

Equation 1 in which x represents a vector of individual characteristics, y denotes earnings, β is a vector of coefficients. Superscript p denotes vectors of coefficients β obtained from the pooled model, while superscripts A and B indicate vectors of coefficients from the respective group-wise models. EX and UN mark the explained and unexplained parts of the differential between y ^A and y ^B, respectively[21]. Fairlie decomposition decomposes binary outcome employment differentials in the same spirit. The penalty for foreign origin or non-citizenship in the labor market, measured as the unexplained part of employment and earnings differentials, signifies the role of discrimination, differences in ethnic capital, selection, or measurement issues such as those possibly introduced by imperfect transferability of the immigrants' human capital. We limit our attention to males, since their labor market outcomes are less sensitive to the selection issues related to female labor market participation decision[22].

From Table VI we see that being an immigrant bears an employment penalty beyond differentials in characteristics of about 3.6 to 6.9 percentage points vis-à-vis non-immigrant categories in the EU15. Being a non-citizen bears a 2.2 percentage point penalty within the immigrant categories and no significant effect for non-immigrants. In the EU8 being a non-immigrant non-citizen implies and employment penalty of about 9.7 percentage points with respect to immigrant citizens and 11.7 percentage points vis-à-vis non-immigrant citizens. Being a non-citizen is associated with a penalty of 5.1 percentage points within the category of immigrants. These results confirm the penalty associated with foreign origin in the EU15 and the negative role of non-citizenship in the EU8 revealed by the probit model.

Table VII draws an interesting picture about earnings differentials. In the EU15, being in any immigrant category bears an earnings penalty vis-à-vis non-immigrant citizens. The estimated penalties are similar for immigrant citizens and non-citizens, confirming our OLS results that it is mainly foreign origin that matters in the EU15. In the EU8, being a non-citizen immigrant exhibits a significant earning penalty vis-à-vis non-immigrant citizens beyond what observable characteristics can explain. The same result, albeit non-significant, holds within the non-immigrant category. Foreign origin per se does not exhibit significant penalty.

Comparing citizens and non-citizens in the EU8, while employment penalties are larger than the raw gaps, characteristics explain significant parts of earnings gaps. That there are almost no non-immigrant non-citizens in EU8, outside Estonia and Latvia, suggests that the observed employment penalties for this group are mainly driven by (non-immigrant non-citizen) Russian ethnic minorities in these two countries. The result that most of the earnings gap that this ethnic group faces vis-à-vis citizens is explained by observable characteristics probably points to the role of regional, industrial and occupational segmentation as well as limited access to jobs in the public sector of non-Estonians and non-Latvians in these countries. While the Soviet system officially promoted equality and perhaps even favored ethnic Russians, the regime change seems to have left ethnic Russians in a relatively unfavorable position, being concentrated in heavy industry and mining (and the corresponding regions) that were struck hardest during transition (Leping and Toomet, 2007).

6. Conclusions

In this paper we have mapped and compared the roles of foreign origin and citizenship for labor market performance in Eastern and Western EU Member States. While our ambition was not to identify causal relationships, we have shown that these roles are different in these two regions. In the EU15 it is foreign origin that bears an employment penalty for both genders and earnings penalty for males. Citizenship matters for female earnings. In the EU8, lack of citizenship significantly hurts male employment, inflicts a negative penalty on male earnings if accompanied by being foreign born, and, similarly to foreign origin, plays a negative role for females both in terms of employment and earnings.

Distinguishing between EU and non-EU origins, our findings confirm the important and gender-dependent role of non-native backgrounds. The decomposition analysis confirms the significance of foreign origin for earnings and employment in the EU15 and the important role of citizenship in the EU8. Moreover, our evidence suggests that the effects of citizenship in the EU8 may be driven by the (predominantly ethnic Russian) non-citizens in Estonia and Latvia. Thus, ethnic minorities who are deprived of citizenship in these countries seem to be a particularly vulnerable group in the labor market. Further analysis is necessary to evaluate the observed associations as causal relationships.

From a policy perspective, our study offers a number of insights. First, both foreign origin and citizenship remain important barriers to non-native labor market integration and, as such, require adequate policy attention in the EU15 and EU8 alike. Second, the vulnerable situation of ethnic minorities (predominantly ethnic Russian) in Estonia and Latvia calls for additional policy efforts to eradicate labor market barriers faced by non-natives. In Estonia and Latvia, in particular, employment policies seem to be most relevant for non-immigrant non-citizens. Once employed, non-nativity may play a less negative role in the labor market for some groups, although labor market segmentation remains a significant issue. Third, further policies targeting the regional, industrial, and occupational segmentation of non-natives are necessary in the whole EU. In Estonia and Latvia such policies are particularly desirable, given that the burden of economic transformation seems to hurt non-Estonians and non-Latvians disproportionally. Finally, compared to the EU15, structural and historical reasons make it even harder for non-EU immigrant women in the EU8 to integrate in the labor markets, thus necessitating additional gender equality policies.

Equation 1

Table I Proportions of foreign-born and foreign citizens in the EU

Table II Summary statistics

Table III Employment probabilities

Table IV Earnings profiles

Table V Marginal effects of immigrants status and citizenship by origin

Table VI The employment divide

Table VII The earnings divide

Notes

We use the term non-native to denote people who are either foreign born or non-citizens. We are well aware of the fact that some non-citizens are not foreign born, however. This is the case, among others, for many ethnic Russians in the Baltic States or guest workers' descendants in Germany.

The early contributions include Chiswick (1978) and Borjas (1985, 1990, 1995) on the labor market performance of immigrants, Bratsberg et al. (2002) on the effects of naturalization, Zimmermann (2005) summarizes what we know about Western European immigration, and Constant et al. (2006a) represents the literature on ethnic minorities in Eastern Europe. We summarize this literature below.
The EU8 refers to the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, and Slovenia. We do not include Cyprus and Malta here, since both their historical background and labor market situation is very different from the EU8 countries that underwent transition after the fall of the Berlin Wall. The EU15 refers to Austria, Belgium, Denmark, Germany, Greece, Finland, France, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden and the UK. We stick to this nomenclature even if some countries are dropped from the analysis.
See Zaiceva and Zimmermann (2008) for an analysis of the recent trends in European migration.
Interethnic marriages and their effects on labor market outcomes, though beyond the scope of this study, may be an important issue in the Baltic States. For example, according to the Latvian statistical office, 21.7 percent of males and 19.9 percent of females with Latvian ethnicity report having a spouse of different ethnicity.
Another interesting outcome measure, which is beyond the scope of this study, is self-employment. Our data show that whereas 10.6 percent of immigrants and 9.7 percent of non-citizens (as compared to 15.8 percent for natives) are self-employed in EU15, the corresponding numbers for EU8 are 6.8 percent and 5.0 percent (10.8 percent).
The endogeneity of migration and citizenship decisions is well documented; see, e.g. Zimmermann (2005) on the former and DeVoretz (2008) on the latter.
Another study by the same authors investigates the Russian-Ukrainian political divide in Ukraine (Constant et al., 2006b).
Unfortunately, more detailed information on the country of origin is not available. In addition, for Estonia, Latvia and Germany, non-citizens and immigrants of EU origin are grouped with their non-EU counterparts.
Gross employee cash or near cash income includes: the monetary component of the compensation of employees in cash payable by an employer to an employee; and the value of any social contributions and income taxes payable by an employer to social insurance schemes or tax authorities on behalf of the employee. Net earnings only include the first component. The number of months worked during the income reference period was accounted for in the calculations.
For Spain, Greece, Italy, Latvia and Portugal, where net rather than gross earnings are reported, we used net instead of gross earnings. Auxiliary analysis where we dropped these countries (available upon request) confirms that our results remain robust with respect to using net or gross earnings. In the estimation with net hourly earnings only (dropping Denmark, Finland, the Netherlands, Hungary and Slovakia due to missing information), the results were qualitatively the same the only exception being that foreign origin in the EU8 became significant at the 5 percent level. This result was driven by non-EU male immigrants in the EU8 that became significant at the 5 percent level in the regressions by origin.
We have also experimented with a rural settlement dummy and the results were, in general, qualitatively similar; however, due to the much smaller sample size we decided to report the results without it.
In order to overcome the potential pensioner bias, we experimented with different age thresholds as well as with dropping reported pensioners from the sample. The results were not affected.
Exclusion of Luxembourg had no noteworthy effects on our results.
Averaged over citizens and non-citizens, immigrants earn less than non-immigrants.
Note that selection into the labor force may particularly bias the results for females. However, estimating a more structural model is beyond the scope of this paper and is left for the future research.
EU6 denotes the Czech Republic, Hungary, Lithuania, Poland, Slovakia and Slovenia (dropped here).
Recall, however, that in Estonia and Latvia non-citizens and immigrants of EU origin are grouped with their non-EU counterparts.
These results for EU15 are consistent with those of Adsera and Chiswick (2007).
We focus on the role of foreign origin on employment in the EU8. The other gender differences that we find could be explained by similar or completely new arguments, but their thorough analysis is beyond the scope of this paper.
Note that the standard Oaxaca-type decomposition decomposes the results at the mean of the earnings distribution, which may mask important differences across the entire distribution. See Machado and Mata (2005) and Melly (2005) for the decompositions at different quantiles.
Note that some of the results in the decomposition analyses are based on relatively small samples. Nevertheless, as discussed below, they are consistent with the Probit and OLS analysis of the previous sections.

References

Adsera, A., Chiswick, B.R. (2007), "Are there gender and country of origin differences in immigrant labor market outcomes across European destinations?", Journal of Population Economics, Vol. 20 pp.495-526.