The Immigrant-Native Wage Gap in Germany Revisited

This study provides new evidence on the levels of economic integration experienced by foreigners and naturalised immigrants relative to native Germans from 1994 to 2015. We decompose the wage gap using the method for unconditional quantile regression models by employing a regression of the (recentered) influence function (RIF) of the gross hourly wage on a rich set of explanatory variables. This approach enables us to estimate contributions made across the whole wage distribution. To allow for a detailed characterization of labour market conditions, we consider a comprehensive set of socio-economic and labour-related aspects capturing influences of, e.g., human capital quality, cultural background, and the personalities of immigrants. The decomposition results clearly indicate a significant growing gap with higher wages for both foreigners (13.6 to 17.6 %) and naturalised immigrants (10.0 to 16.4 %). The findings further display a low explanation for the wage gap in low wage deciles that is even more pronounced within immigrant subgroups. Cultural and economic distances each have a significant influence on wages. A different appreciation of foreign educational qualifications, however, widens the wage gap substantially by 4.5 ppts on average. Moreover, we observe an indication of deterioration of immigrants’ human capital endowments over time relative to those of native Germans.


INTRODUCTION
The continual globalization of societies fosters cultural diversity within national borders. In 2016, roughly 23.5 percent (19.5 million) of the German population had a so-called migration background, i.e., a personal migration experience or recent migration ancestry. 1 With the recent inflow of migrants into Europe since (Eurostat, 2018b, opposing currents within the societies of European countries have become more visible. Germany as an immigration country cannot deprive from these contrary currents. As in many other countries of Europe and around the world, right-wing populist parties have recently achieved high rates of approval in elections. Sola (2018) finds a positive correlation between concerns about immigration and support for right-wing populist party "Alternative für Deutschland" (AfD) especially in eastern Germany.
Critics of immigration in Germany are supported by the fact that the benefit system is demonstrably claimed by a growing number of foreigners (Riphahn et al., 2013). This public concern is fuelled by a perception of rising levels of income inequality (Roth et al., 2017). Although the development of inequality in terms of wages has stagnated in recent years (Biewen et al., 2017), 2 research shows that especially low-skilled workers and immigrants are increasingly being negatively affected by wage inequality in Germany (e.g., Algan et al., 2010;Gernandt & Pfeiffer, 2007). Because the wage gap between immigrants and natives is a good indicator of economic integration and reflects the effectiveness of a country's immigration and labour market policies, we study wage development trends for these groups in Germany in identifying a set of causes.
Using the immigrant-native wage gap to analyse uneven remuneration is sensible because wages generally serve as an indicator of individuals' levels of labour productivity. According to human capital theory, one's productivity is determined by one's abilities and skills, which are often expressed by one's level of education (e.g., formal qualification) and work experience (Aldashev et al., 2012;Tverdostup & Paas, 2017) and therefore translated into earnings. Thus, a wage gap initially reflects a difference in productivity among workers and is not evident with discrimination against an individual or a particular group in the labour market (O'Neill & O'Neill, 2015).
Regarding importance of inclusion, the labour market integration of immigrants is a major policy concern, as immigrants' contributions to the economy depend directly on their success. Together with social and cultural aspects, income and wages are indispensable to holistic assimilation (e.g., Lehmer & Ludsteck, 2015, p. 677). In the first place, a welfare loss occurs due to insufficient job allocation. Immigrant employees may work in occupations below their qualifications and thus cannot exhaust their full production potential. In extreme cases, high wage differentials lead to larger unemployment assistance and social assistance payments in the medium run while social insurance contributions and tax revenues decrease. To identify triggers for counteracting social division, it is important to analyse whether wage differentials are due to observable differences in, for example, human capital endowments or due to unobservable influences comprising ethnic discrimination (Aldashev et al., 2012). 3 A wage differential usually originates from limited access to the labour market (Aldashev et al., 2009;Brynin & Güveli, 2012). To improve the employment and labour market prospects of foreigners, the German government become increasingly dedicated to offering courses specially designed for immigrants on language instruction, social integration, integration through apprenticeship, work, and (university) education (Federal Government, 2016;Kosyakova & Sirries, 2017). Both the total number of courses and the demand for specific courses such as those on literacy and youth integration have increased over the last decade (BAMF, 2017).
We decompose the immigrant-native wage gaps for males for the years 1994 to 2015 using data from the German Socio-Economic Panel (SOEP) as a source of information. SOEP data include a rich set of household and labour-related characteristics relevant to understanding the determinants of labour market success across groups. We consider a rich set of control variables that recognizes typically unobservable labour market influences. In particular, we examine individual personality traits and integration barriers by taking into account metrics of immigrants' proximity to Germany based on their home countries' levels of cultural distance (Kanas et al., 2012). We also consider foreign education degrees and employ the home country's economic performance as an indicator of human capital quality (Coulombe et al., 2014). We estimate the immigrant-native wage gap by applying a variant of Blinder-Oaxaca decomposition to emphasize differences in returns. To consider heterogeneous effects observed along the whole wage distribution, we apply an approach proposed by Firpo et al. (2009) based on a recentered influence function (RIF) for unconditional quantile regression (UQR) models. This approach is mainly advantageous in its more precise decomposition, which allows one to estimate the contributions of each variable to composition effects observed along the entire wage distribution (Galego & Pereira, 2014).
The variety of origins (and migration motives) involved makes it extremely difficult to depict the foreign qualifications of persons due to the presence of different education systems and requirements. Our study design is constructed to take this diversity explicitly into consideration. We differentiate between three main population groups in our analysis: Native Germans, Naturalised Immigrants, and Foreigners. We further consider (i) citizens of Turkey, (ii) citizens of the former Yugoslavia, and (iii) citizens of southern European countries as subgroups of Foreigners, as the influx of guest-workers during the 1960s and subsequent family reunification that occurred in the following decades formed large demographic groups from the Mediterranean within Germany. Naturalised Immigrants are further divided into (j) ethnic German repatriates and (jj) naturalised immigrants without ethnic immigrants. For further information on German migration history, see Appendix C.
Our empirical results show a significant gap in wages for Foreigners and Naturalised Immigrants relative to Native Germans without a migration background for the more than two decades of analysis.
Regarding individual and labour market characteristics affecting wages, on average, roughly three quarters of gaps along the wage distribution can be attributed to observable differences in individuals' human capital endowments and work-related factors but with distinct differences observed between immigrant groups.
With respect to human capital transferability across borders, a perceptible disadvantage can still be 3 A pay disadvantage or even discrimination against an equivalent job occurs when the same degree of employee labour productivity -equal qualifications and (labour market) experience, similar personal characteristics and equal overall conditions (sector, etc.) -is remunerated to varying degrees. For further details on direct and indirect discrimination see OECD (2013). attributed to education obtained abroad. This implies an insufficient adaptation of qualifications in Germany. Furthermore, we observe a rising gap in average wages for both immigrant main groups over time. We find a consistently high degree of explanation due to individual and labour market characteristics indicating that the human capital endowments of immigrants have deteriorated relative to those of native Germans over time. Given the above mentioned strong public and private efforts made to socially and economically integrate immigrants in Germany, these results raise doubts surrounding the effectiveness and efficiency of such programmes.
The remainder of this paper is organized as follows: We first review the related literature on wage inequality and the wage gap. Section 3 provides information on the data used for the empirical analysis, which is followed with a presentation of selected descriptive statistics (section 4). We introduce the econometric approach of the decomposition method in section 5. The empirical results are illustrated and evaluated in section 6. The final section provides conclusions.

RELATED LITERATURE
Wage differentials between natives and foreigners have been analysed in a number of studies. Because the convergence of immigrants' wage levels to natives' wage levels serves as an important indication of their degrees of labour market integration, a recurring contemplation of wage differences between these groups is essential to uncovering structural and persistent disadvantages (Coulombe et al., 2014). Despite current political and societal discussions, however, much of the evidence available for Germany refers to the period surrounding the turn of the millennium. A more recent account on the situation of the last decade is not available. The results from earlier studies note levels of wage discrimination against immigrants of 13 to 17 percent in western Germany for 1996 to 2005 (Bartolucci, 2014). 4 At the same time, Lehmer & Ludsteck (2011) identify a heterogeneous pattern of immigrant salary disadvantages depending on the country of origin (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006). 5 Here, even lower wages can be observed for second-generation immigrants (Algan et al., 2010). 6 Further results provided by Aldashev et al. (2012) reveal significant wage gaps for both foreigners (25 %) and naturalised immigrants (19 %) based on SOEP data for 1992 to 2009. However, Germany is not the only country in Europe experiencing wage inequality between its host and immigrant population. The majority of migrants within the European Union face income disadvantages, which tend to be even more pronounced for migrants from non-EU countries than for migrants from EU member states (Adsera & Chiswick, 2007;Lehmer & Ludsteck, 2011, 2015. For Austria, where the share of foreigners is higher than that in Germany, Hofer et al. (2017) reveal a wage gap between immigrants and natives of 15 percent for 2008 to 2010; the majority of this wage gap can be attributed to differences in human capital endowments. Moreover, wage differentials tended to be larger for higher incomes in 2008.
For Germany, related evidence indicates the opposite trend: the wage gap decreases steadily with higher incomes and may turn even positive at a wage peak (Grandner & Gstach, 2015, p. 63).
Generally, wage differences between natives and immigrants can be attributed to a lack of host country-specific human capital. Therefore, immigrants face an initial income disadvantage upon arrival 4 Bartolucci (2014) uses matched employer-employee data (LIAB) from the Institute for Employment Research (IAB). 5 Lehmer & Ludsteck (2011) use employment register data (BEH) of the German Federal Employment Agency. 6 Algan et al. (2010) use data from German Mircocensus 2005/2006 relative to natives (Fertig & Schurer, 2007;Tverdostup & Paas, 2017). To compensate for this lack of required human capital, immigrants immediately start on a path with high(er) investment costs. Hence, earnings are low directly after arrival, but high levels of human capital will guarantee assimilation into the host labour market afterwards (Borjas, 1985), leading to the diminution of the initial income gap (Fertig & Schurer, 2007). By acquiring knowledge on the language, customs, and nature of the labour market of the host country over time, immigrants can achieve supplementary and holistic assimilation. These factors can have positive effects in terms of raising immigrants' earnings. In addition, it should be noted that a positive self-selection of immigrants concerning assimilation is likely. A long period of residence in the host country may be accompanied with successful integration into the labour market and into society whereas unsuccessful integration may increase the probability of remigration (Gundel & Peters, 2008). Related to this, Gathmann & Keller (2018) show that faster access to German citizenship promotes immigrants' incentives to invest in skills, thereby causing them to enhance their labour market performance (earnings) and establish social contacts with the domestic culture. All of these processes result in deeper levels of social and cultural integration (Felfe et al., 2019). Therefore, not all immigrants will invest in host countryspecific human capital or seek jobs of higher status when investment costs exceed expected returns, i.e., when these are relatively high and the time of stay is presumably temporal (Kogan, 2011). As a result, assimilation effects may be overestimated when self-selection is not adequately regarded.
Nevertheless, due to its correlation with social and cultural assimilation, the time of residence may be an important factor shaping naturalised immigrants' and foreigners' wages (Chiswick, 1978). Descriptive statistics given by Lehmer & Ludsteck (2015) show a decline in wage differences between immigrants and natives in Germany. According to their results, immigrants assimilate through the accumulation of firm-specific human capital and by moving to better paying firms, i.e., immigrants realize search gains. The process of assimilation slows down throughout the appropriation of host country-specific human capital (Borjas, 2015). This assimilation behaviour among immigrants is tested conventionally under the framework of the assimilation hypothesis developed by Chiswick (1978). Based on this concept, Fertig & Schurer (2007) estimated a catch-up interval of wages of approximately nine years for Germany and the USA. Nevertheless, Borjas (1985:465) directly criticizes the assimilation hypothesis due to cohort effects.
A key component of host country-specific human capital is language proficiency (Gundel & Peters, 2008). Hochman & Davidov (2014, p. 352) confirm that proficiency in the host country's language is central to immigrants' labour market achievements. The effect of language on wages, however, is usually underestimated (Dustmann & Van Soest, 2002) because insufficient levels of language proficiency diminish the probability of immigrant labour market participation and therefore may not affect wages fully (Aldashev et al., 2009). Language proficiency, however, is a prerequisite to holding professions of higher standing. The results by Guven & Islam (2015) indicate that poor language skills particularly in childhood imply significant disadvantages in terms of social assimilation and academic and labour market success.
According to Christl et al. (2018), closely related literacy skills also have a significant impact on wages and explain the wage differential between immigrants and natives to a certain extent.
Whether education is obtained from the host or home country serves a further strong explanation for the immigrant-native wage gap (Fortin et al., 2016;Warman et al., 2015). Regarding the educational levels of persons of foreign backgrounds, human capital obtained in the home country may not be equivalent to that obtained in the host country due to the limited transferability of skills and due to imperfect compatibility of home and host country labour markets (Basilio et al., 2017). Indeed, Basilio et al. (2017) consider lower levels of human capital quality and the incomplete transferability of human capital to be major factors in explaining the wage differential between natives and immigrants in Germany. The returns to education and labour market experience obtained outside of Germany are demonstrably lower than those to human capital obtained in Germany (Aldashev et al., 2009). The acquisition of host country-specific skills is exacerbated further by greater linguistic and cultural distance between countries of origin and the host country. The more similar two countries are in language and culture, the easier it is to acquire these resources (Isphording & Otten, 2014). It is therefore necessary to quantify the influence of cultural differences on labour market success.
Cognitive abilities are complemented with personality traits as determinants of labour market success.
While certain personality traits result in stronger job performance, others may be unfavourable in the labour market. For example, people with certain dispositions of personality traits may gain easier access to specific occupations and positions than others (Brenzel & Laible, 2016;Heineck & Anger, 2010;John & Thomsen, 2014). Because cognitive abilities and personal characteristics influence each other, an early investment in character-shaping activities is required. The recent empirical labour literature therefore increasingly reflects the role and significance of cognitive abilities. Personality traits affect wages mostly through the channel of educational attainment and through a higher likelihood of engaging in labour market participation accompanied with more social integration (Thiel & Thomsen, 2013). Unique characteristics already lead to greater success on the educational path (Busato et al., 1999).
These and other factors influencing wage inequality have to be evaluated at different levels. For instance, Giesecke & Verwiebe (2009) show a decreasing wage differential between highly educated and less skilled employees in Germany but at the same time increasing wage differentials between occupational classes. Occupations also explain a large proportion of ethnic wage differentials in the United Kingdom (Longhi, 2017). At the same time, payment differentials within and between industries reinforce the existing wage gap between natives and immigrants, especially since immigrants are concentrated in sectors of manual activity (Antonczyk et al., 2010;Aydemir & Skuterud, 2008). Furthermore, a change in employment patterns, e.g., the growth of (marginal) part-time work, contributes to an overall increase in wage inequality (Biewen & Juhasz, 2012). Longhi (2017) concurrently highlights the spatial level of wage discrimination and stresses that estimated ethnic wage differentials are fundamentally overstated when they refer to the national level. When minorities are compared to the majority in the same local labour market while facing similar socio-economic conditions, the results reveal that ethnic wage differentials tend to be even more heavily underestimated.

DESCRIPTION OF THE ESTIMATION SAMPLE
For the empirical analysis, we use data from the German Socio-Economic Panel (SOEP). The SOEP is a wide-ranging and representative longitudinal panel study of roughly 11,000 private households where roughly 30,000 persons are interviewed annually on issues related to income, employment, education and health (see Goebel et al., 2018 for more information). We focus on the survey waves from 1994 to 2015 to exclude short-term fluctuations in the labour market occurring at the start of the 1990s. We consider strong waves of immigration occurring after the downfall of the Iron Curtain to secure sufficient sample sizes for each ethnic group and especially for ethnic German repatriates. 7 Our variable of interest 'gross hourly wage' is obtained by dividing the gross wages for each month by the reported real working hours of the last week extrapolated to monthly hours. We assume that there are 4.35 weeks in each month for the calculation. To analyse developments occurring over 22 years, we adjust wages for inflation using the GDP deflator and measure them in prices for 2010. We further apply symmetric trimming to the wage distribution by dropping the upper and lower two percent from the analysis to correct for outliers.
The comprehensive set of socio-demographic variables included in the SOEP allows for the identification of immigration status beyond the concept of citizenship. In particular, information on whether a person or one parent immigrated to Germany (immigration background) can be collected by combining a persons' citizenship, country of origin and year of immigration to Germany (see Aldashev et al., 2012). In our empirical analysis, we distinguish between Foreigners, Naturalised Immigrants and Native Germans: -Foreigners are all persons without German citizenship. We further consider three subgroups covering the main regions of origin of guest-workers from the 1960s: 'citizens of Turkey', 'citizens of the former Socialist Federal Republic of Yugoslavia (SFRY)' 8 and 'citizens of southern European countries' (Greece, Italy, Spain and Portugal).
-Naturalised Immigrants are former citizens of foreign countries who received German citizenship at or after immigration to Germany. Since Naturalised Immigrants are a highly heterogeneous group given the different origins and motivations for naturalisation, we distinguish between 'ethnic German repatriates' and 'naturalised immigrants without ethnic Germans' as two separate groups. We define 'ethnic German repatriates' as persons with German citizenship originating from countries of the former Soviet Union 9 or from Eastern Europe 10 and arriving in Germany after 1987. 11 -The remaining persons form the group of Native Germans. However, we distinguish between native Germans with and without an indirect migration background. 'Native Germans with an indirect migration background' represent the second generation of naturalised immigrants. As a reference group, we use 'native Germans without a migration background' to avoid strong cultural and language ties to (partly) naturalised parents.
Distinguishing between these groups is useful to identify potential differences and similarities between ethnic groups. We look at naturalised immigrants separately, as they clearly differ in their labour market characteristics (see below) from those of foreigners and native Germans. Legally, naturalised immigrants are not distinguishable from native Germans (the same political participation rights), but foreign roots may determine a divergent cultural and economic background. Since these people possess skills predominantly 7 Ethnic German repatriates are individuals with German ethnicity from successor states of the former Soviet Union and from other Eastern European states who returned to their ancestral homeland to settle permanently. 8 The group also includes SFR Yugoslavia's successor states: Slovenia, Croatia, Bosnia and Herzegovina, Serbia (incl. Kosovo), Montenegro and Macedonia. 9 Russia, Ukraine, Moldavia, Belarus, Kazakhstan, Tadzhikistan, Turkmenistan, Kirgizstan, Uzbekistan, Estonia, Latvia, Lithuania, Georgia, Armenia, and Azerbaijan. 10 Poland, the Czech Republic and Slovakia (formerly Czechoslovakia), Hungary, and Romania but not Bulgaria (earlier repatriation). 11 The definition of 'ethnic Germans repatriates' is imprecise to a certain extent because all immigrants from the selected countries who have acquired German citizenship are considered and not just ethnic Germans alone. As SOEP data statistics show high immigration rates for each selected country of origin only for the beginning of the 1990s, a good approximation persists.
obtained abroad, they may be valued differently in the highly regulated German labour market. In addition, naturalised immigrants can be expected to differ from foreigners in terms of their time of residence and intentions to stay in Germany. In calculating cultural distances, we use the revised measurement method developed by Kaasa et al. (2016), which is based on a revision of Hofstede's (1980) original work.
Hofstede's approach assumes that the most important cultural differences can be captured by four cultural dimensions: power distance, uncertainty avoidance, individualism-collectivism, and masculinityfemininity (see Kaasa et al., 2016, p. 234). 12 We further consider individuals' personality traits using the widely adopted Big Five personality traits. To consider occupational selection, we refer to a classification developed by Erikson-Goldthorpe-Portocarero (EGP) that clusters occupations by social status. The lower end of the scope is reflects unskilled manual occupations for which no vocational training is required, and the upper end reflects higher services covering managers and academic occupations. We augment the available data by regional information at the state level to control for the regional economic environment and for labour force supplies in the empirical analysis using statistics provided by the Federal Employment Agency (2017) and the Federal Bureau of Statistics (2017b). The incorporated regional information includes, among other data, the share of the foreign population to depict the ethnic composition. A high ethnic concentration has a significantly negative effect on immigrants' levels of German language proficiency (Danzer & Yaman, 2016) and leads in general to lower investments in human capital (Battisti et al., 2018). For homogeneity reasons, we impose a number of restrictions on the estimation sample. We only consider first generation immigrants living in western Germany (incl. Berlin) -which means persons who were born abroad and who have immigrated to Germany. To ensure a reliable comparison of groups, we concentrate our analysis on the population of prime aged males (25 to 54 years) in full-time employment.
The restriction of full-time employment is necessary because part-time jobs and atypical employment may vary between groups. For the same reason, self-employed persons, civil servants and soldiers are not regarded either. 13 Focussing on males ensures avoiding biased interpretations due to differences in labour market-relevant characteristics between females and males and in labour force participation rates of females by origin (Ñopo, 2008). The age range is limited at both ends due to different patterns of participation in 12 (1) Power distance shows the extent to which less powerful individuals of a society accept and expect an unequal distribution of power.
(2) Uncertainty avoidance reveals to what degree people feel comfortable with uncertainty. Laws, guidelines, and security measures characterize cultures with a high uncertainty avoidance. (3) Masculinity shows to what degree masculine values, such as orientation towards achievement, success, and assertiveness prevail over female values like caring, cooperation, and modesty. (4) Individualism describes the extent to which people appreciate to act as individuals rather than as members of a collectivist culture (Kaasa et al., 2016). 13 The rate of self-employment is marginally larger for Germans (10.2 %) than for foreigners (10.0 %) (Federal Bureau of Statistics, 2017b).
the educational system at the lower level and due to differences concerning (early) retirement at the upper end. With these restrictions in place, the estimation sample includes 52,165 observations of Native Germans without a migration background (76.9 %), 6,276 observations of Naturalised Immigrants (9.3 %), and 9,383 observations of Foreigners (13.8 %) (see Table A. 2 of the appendix for a detailed description).

DESCRIPTIVE STATISTICS
Before turning to the econometric methodology and empirical estimates, we discuss wage development  < Table 1 about here > Furthermore, we consider education as an indicator for qualification at three levels. Based on the CASMIN educational classification, people without formal occupational training are regarded as low-skilled, persons with occupational training are medium-skilled, and those with a college or university degree are considered highly skilled. The share of low-skilled persons is statistically higher across all immigrant groups but is the most pronounced for the group of Foreigners. Accordingly, all immigrant groups -except for naturalised immigrants without ethnic Germans -have lower shares of highly skilled workers.
Moreover, naturalised immigrants without ethnic Germans exhibit the lowest shares of persons who have completed their highest education abroad (38 %) while ethnic German repatriates -who immigrate at a comparatively higher age -present the highest ratio (68 %).
When considering the home country's economic performance in the year of immigration as a human capital quality indicator, we observe the largest economic distances to the countries of origin for ethnic German repatriates and Yugoslavs. On the other hand, the distance for southern European countries is relatively small. German language proficiency (speaking, reading and writing) is represented as a selfassessment of writing skills in the German language whereby skills are evaluated with scores of 1 (not at all) to 5 (very good). We note a slightly positive correlation with time of residence in Germany for all groups in consideration.
Furthermore, a larger cultural distance -expressed as, e.g., language, religion, and social normsbetween home and host countries could hamper social integration. The cultural distance to Germany is the largest for Turkey. Turkish culture is characterized by different epochs and ethnicities and is heavily influenced by Islam. The average cultural distance to southern European countries is considerably lower.
The culture of Naturalised Immigrants is highly heterogeneous and therefore the average value offers limited information. A comparison of Big Five traits (see Table A. 3 of the appendix) reveals significant differences in average personality traits between ethnic Germans repatriates, citizens of southern European countries and occasionally citizens of Turkey relative to Native Germans. The two latter immigrant groups are very similar in their characteristics.
As is reported extensively in the literature, occupational segmentation serves as a strong explanation for wages. As foreigners were recruited in the 1960s and 1970s predominantly for work of low status, there was a corresponding high levels of ethnic stratification across occupations (Constant & Massey, 2005).
This pattern has remained very persistent over time. While immigrants still mainly perform jobs involving manual tasks (skilled and unskilled), Native Germans are relatively more highly specialized in high and low services. 15 These differences are reflected in their distribution across economic sectors. The sectoral distribution may be explained by language proficiency whereby, e.g., in the service sector stronger language skills are generally required than in occupations mainly involving manual tasks. Furthermore, Foreigners work more often in small-and medium-sized firms than Native Germans. Overall, immigrant groups and Native Germans differ verifiably in their work-related characteristics.

WAGE GAP DECOMPOSITION
The descriptive statistics show significantly divergent log hourly wages between Native Germans and each of the immigrant groups. To quantify the underlying causes of wage differences, we apply Blinder-Oaxaca decomposition for unconditional quantile regression (UQR) models. The widely used Blinder-Oaxaca method decomposes mean wage differentials into explanatory determinants and an unexplained part. In its original settings, the decomposition technique uses a wage equation taking the form of a linear regression estimation = + for individuals of group ∈ { , }. The mean difference between groups A and B can be formulated as follows: where denotes output means while denotes sample averages of the explanatory variables for each group. Here, is the vector log hourly wage of an individual within one group, is the data matrix containing independent variables, e.g., individual and labour market characteristics (including a constant), is the vector of regression coefficients, and is the vector of random errors (Jann, 2008). The decomposition method divides the outcome difference of the wage equation into two components: The first term ( − )′̂ represents the "endowment effect" attributable to mean differences in background characteristics (e.g., education and experience). The second term � ′ (̂−̂) denotes the "coefficient effect" and represents differences in returns to similar characteristics. 16 The approach is based on a regression of the recentered influence function (RIF), which is similar to a standard OLS regression except that dependent variable Y (in our case: the log wage) is replaced by the RIF of the statistic of interest (Fortin et al., 2011, p. 76). An influence function measures the influence of a single observation on a distributional statistic. The RIF of the ℎ quantile is given by the following expression (Galego & Pereira, 2014, p. 2516): It is computed by estimating the marginal density f Y (q τ ) of Y for sample quantile q τ . This is achieved by using kernel methods and by forming a dummy variable I(Y ≤ q τ ) indicating whether the value of the outcome variable falls below q τ (

INTERPRETATION
The coefficient effect of the decomposition exposes differences in returns and is commonly appraised as a measure of discrimination investigating wage discrepancies (Firpo et al., 2018;Jann, 2008). However, this interpretation is vulnerable because the unexplained component captures both the effects of discrimination and unobserved group differences (Lehmer & Ludsteck, 2011;O'Neill & O'Neill, 2015). Unobserved causes of wage gaps may underlie individuals' soft motives (e.g., motivation, preferences, and aspirations), further unobservable skills (e.g., negotiating skills and assertiveness), or cultural and social norms in general. On the other hand, adding more control variables inevitably reduces the estimated magnitude of discrimination (Grandner & Gstach, 2015). In conclusion, the unexplained part of the decomposition serves as only an indication of discrimination and less as proof (Canal-Domínguez & Rodríguez-Gutiérrez, 2008).
In addition, Altonji & Blank (1999) emphasize that it is also deceptive to label this second component alone as the result of discrimination, as discriminatory barriers in the labour market can affect the characteristics of individuals. Regardless of the chosen model, the direct comparison of individuals or groups is limited: certain combinations of individual characteristics and job requirements are only possible for one group and may not be for others (Ñopo, 2008).

IMPLEMENTATION
The final model specification used for the estimation of wage gap decomposition is the result of a deductive process of variable selection. In the wage equation, we consider as the base set of independent variables the individual characteristics of labour market experience (and its square), a cohabitation dummy, three skill levels obtained from the international education classification, and an indicator of German language proficiency. We further control for job-related attributes such as firm size (categorical), dummy variables for industry affiliation, and dummy variables for occupational class. In addition, time and regional fixed effects are included. We augment the model with regional information at federal state level by 17 Fortin provides a Stata package rifreg to perform RIF-regressions and package oaxaca8 for enhanced Blinder-Oaxaca decompositions (Fortin, n.d.). 18 The even distribution of all observations among quantiles may lead to different ratios between immigrant groups and native Germans within the respective quantiles.
approximating the economic environment and the labour force supply: the region's settlement structure type, the share of the foreign population, real GDP per capita, and the unemployment rate. Furthermore, we use survey weights at the individual level to mitigate a potential bias due to the over-representativeness of high-income households and immigrants in SOEP data.
The wage gap decomposition is computed for each decile of the wage distribution. We consider the first to ninth quantiles because for the method to work, observations made above our highest percentile of interest are required. Endowment and coefficient effects for each of the nine wage sections are estimated.
We implement various model specifications to test for the influence of foreign degrees, human capital quality, personality, and cultural distinctness. We assume that a large cultural and economic distance as well as the limited transnational transferability of human capital prove to be a disadvantage in the German labour market. Furthermore, we review the labour market situation of immigrants over time because we assume that a rising wage gap due to various legislative amendments. The reforms have predominantly caused part-time work in manual occupations where immigrants are highly concentrated. We present the results for the two immigrant main groups of Foreigners and Naturalised Immigrants and supplement them with results for the subgroups. The derivation of the model specification precedes the respective results.

THE IMMIGRANT-NATIVE WAGE GAP
The wage gap decompositions show different results for Foreigners and Naturalised Immigrants. 19 For the period 1994 to 2015, we find substantial wage gaps for both main groups relative to Native Germans.
Naturalised Immigrants' wage gaps relative to Native Germans reach 10.0 to 16.4 percent, rising with higher wage deciles (mean: 13.1 %). 20 At the same time, the endowment effect rises from 50 to 100 percent (mean: 81 %). Therefore, a large proportion of the wage gap for low wage deciles remains unexplained when capturing unobserved factors of influence. The wage gap for Foreigners is consistently higher and less diverse between the deciles than for Naturalised Immigrants (13.6-17.6 %, mean: 14.8 %). The explanatory power of individuals' endowments of Foreigners is greater overall than it is for Naturalised Immigrants and reaches shares of 75 to 85 percent for low and middle wage deciles (see Figure 2). 21 However, the endowment effect reveals an overvaluation of high wage deciles, suggesting above-average remuneration in terms of qualification (mean: 100 %). For both Naturalised Immigrants and Foreigners, the explanation of the wage gap is mainly driven by individuals' levels of language proficiency and by occupation in high and low services (see Figure B. 2). Education has only a slightly positive effect. 19 The RIF-regression wage model estimates reveal comparable effects of the independent variables on wages for the principal groups (Table A. 4, Table A. 5 and Table A. 6 of the appendix). A person's labour market experience and higher educational level each have a significantly positive impact on wages for all groups. Here, the influence of higher education is enhanced with higher wages. Furthermore, larger firms pay significantly higher wages on average. For Native Germans and Foreigners, this impact of firm size is comparatively strong at lower wages. The industrial sectors of 'manufacturing' and 'construction' are both important factors explaining the low wages of Foreigners. While service-based occupations more heavily affect Native Germans than manual jobs, service occupations are of greater importance for Naturalised Immigrants and Foreigners especially at high pay deciles. We obtain the highest coefficients of determination for medium to high wage deciles: For Native Germans (27-30 %), Naturalised Immigrants (32-35 %) and Foreigners (32-33 %). 20 Table A. 7 of the appendix shows the results of Blinder-Oaxaca decomposition at the mean. Table A. 8 of the appendix provides corresponding results of the UQR-decomposition. 21 We classify deciles 1 to 3 as low wage deciles, deciles 4 to 6 as middle wage deciles, and deciles 7 to 9 as high wage deciles. However, the explanatory power of labour market experience is greater for Naturalised Immigrants of high wage deciles whereas for Foreigners it is stronger for low wage deciles.
Regarding wage gap development along quantiles, we obtain results opposing those of Grandner & Gstach (2015). Our findings demonstrate the advantages of decomposition for unconditional quantile regressions over mean Blinder-Oaxaca decomposition. On one hand, increasing wage gaps along the wage distribution are observed; on the other hand, we find a greater wage disadvantage for low wage deciles that would otherwise have not been discovered. 22 < Figure 2 about here > According to the descriptive statistics (see Table 1 above), the wage gap for Native Germans varies considerably among the immigrant subgroups. The wage gap for ethnic German repatriates has grown almost linearly from 11.7 to 26.8 percent with increasing wage deciles. A comparable distribution for the pay gap can be observed for citizens of Turkey (14. 8-30.5 %) and for citizens of the former Yugoslavia (17.7-31.2 %) with the exception of relatively large gaps for low and high wage deciles. The endowment effect remains at consistently low levels for citizens of Turkey (30-50 %) and increases for citizens of the former Yugoslavia (50-90 %) and for ethnic German repatriates (40-80 %). The wage gap is consistently smaller for citizens of southern European countries (2.4-14.2 %) and follows a declining course with increasing wages. For lower wage deciles, the explanation accounts for 70 percent and approximately 90 percent for higher pay deciles. The wage differential of naturalised immigrants without ethnic Germans is the smallest of the groups (4.1-7.9 %) and the only group showing a shrinking gap at higher wages.
Although the endowment effect reaches shares of roughly 60 percent only, the results imply that naturalised immigrants no longer seem to differ considerably from Native Germans in terms of personal characteristics and payoffs. 23 Crucial explanatory factors continue to include language proficiency and occupation in high and low services. For naturalised immigrants without ethnic Germans, however, these patterns are less pronounced.
These results may indicate selectivity in naturalisation, i.e., those who are more integrated into the German labour market are more likely to be naturalised. In this respect, von Haaren-Giebel & Sandner (2016) mention higher levels of integration and language proficiency and higher probabilities of staying for naturalised first-generation immigrants compared to foreigners. Overall, foreigners face stronger labour market entry barriers.
For robustness, we additionally run a RIF-decomposition where the group of Naturalised Immigrants includes foreigners who immigrated during our analysis period. We find no divergent results. The inclusion of part-time workers also leads to only a minimal shift, resulting in a slight narrowing of the wage gap for the lowest deciles (see Figure B. 3 of the appendix). Nonetheless, predominantly widening gaps observed along the wage distribution as the level of explanation increases indicate deficient human capital endowments for immigrants for better-paid occupations. Adding individual job tenure to the base model consistently enhances the explanatory content of wage gaps; however, this may be endogenously driven.

EFFECT HETEROGENEITY
We further illustrate the development of labour market conditions for immigrants over time to reveal potential effects of integration policies on wage differentials. To identify potential changes for different (1) age groups and (2) age cohort effects, we consider three age groups: 25-34 years, 35-44 years, and 45-54 years. We exclude foreigners of the first period who have been naturalized thereafter in order to minimize biases resulting from changes in the group compositions. When interpreting the results, various significant institutional changes should be taken into account for our analysis period. 24 6.2.1 AGE GROUPS OVER TIME In considering age groups over time, we equally decompose the wage gap for two periods (1994-1999 and 2010-2015) whereas an interval of 10 years between the two analysis periods is applied to exclude multiple assignments of observations to the same age group. Overall, wage gaps are perceptibly larger for the second period for both immigrant main groups. The growth observed is mainly based on a widening of the lower wage deciles. Wage gaps and explanatory rates, however, vary considerably across immigrant age groups (see Figure 3). Young Foreign workers (25-34 years) are especially affected. Rather, in the first period, the gap increases slightly from -7 to -12 percent along the wage distribution. In the second period, a complete reversal takes place and the wage gap for lower deciles escalates to -16 to -20 percent with an explanatory power value of roughly 90 percent. A different situation is observed for young Naturalised Immigrants whose wage gap rises linearly from -2.5 to -13 percent in the first period (see Figure 3). In the second period, however, the wage gap is reduced to a minimum in the lower deciles (+1 to -5 %) while it rises sharply in the higher deciles (-10 to -17 %). Foreigners (20-24 %), a continuous decline towards zero is noticeable at the highest wage deciles. The explained proportion of the wage gap decomposition is large for each of the middle deciles (70-90 %). On the other hand, the wage gap for naturalised citizens hardly changes, but a partly strong overestimation due to the endowment effect occurs. The wage gap for 45-54 year-old Foreigners is small at first but increases linearly with higher wages (-6.0 to -24.5 %). The overall expansion of the gap towards the second period is valued at 5.5 ppts on average and primarily takes place at the lower end of the wage distribution. Although a slight overestimation emerges, the model shows a high level of explanatory content overall. In the second period, naturalised citizens of this age group present a wage gap of 21 to 25 percent and therefore an increase of 7.5 ppts relative to the first period. The endowment effect levels out at ratios of roughly 75 percent.
For both immigrant main groups, we predominantly note growing wage gaps and a stronger explanation by individuals' endowments for almost all wage deciles. This indicates that the human capital endowment has deteriorated over time relative to Native Germans. Foreign low-wage earners of all age cohorts are especially affected. Additionally, we observe a shift within the explained part of the wage gap decomposition. For both immigrant main groups, the significance of language proficiency remains high but progressively declines. On the other hand, labour market experience and occupations are increasingly important in explaining the wage gap whereas economic sector affiliations are becoming less and less important ( Figure B. 2).

AGE COHORT EFFECTS
The analysis of cohort effects requires an adjustment of analysis periods. To ensure an identical composition of age cohorts over time, the ranges of the analysis period and age groups must be harmonized, producing four age cohorts from which a temporal trend can be captured for two. For example, 25-to 34-year-olds of the first period (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)  Developments are more extensive for the third age cohort (aged 25-34/35-34 years). While Foreigners undergo a massive increase in the wage gap towards the second period (from 9.2 to 19.9 % for deciles 1 to 7) and while the endowment effect rises in terms of its share, the increase is much less pronounced and more differentiated for Naturalised Immigrants. In addition to a minor widening of lower wage deciles, we observe a decline in the wage gap of the higher deciles. Particular attention has to be paid to the fourth and youngest age cohorts (aged 25-34 years in period 2). The group of Foreigners and the group of Naturalised Immigrants present a contrasting picture. Wage convergence to Native Germans is observed for Naturalised Immigrants of the lower wage deciles while this occurs for Foreigners of the higher deciles. The larger wage gaps observed at opposite ends of each wage distribution are characterized by large unexplained shares. The gains of the unexplainable wage gap for young immigrants can be influenced not only by the deterioration of their human capital but also by changes in the age cohort's soft motives and soft skills.
The growth of the wage gap observed towards the second period of each age cohort and especially for Foreigners is worrying. This means that wage disadvantages persist over time and even intensify with age and job tenure. On the other hand, wage gaps of Naturalised Immigrants tend to narrow for later age cohorts.
However, it is not clear whether these predominantly negative developments are related to institutional reforms. Furthermore, the influence of the naturalisation process on group compositions cannot be completely ruled out.

THE ORIGINS OF EDUCATIONAL DEGREES
In testing the transferability of human capital, it is necessary to distinguish whether education was obtained in the immigrant's home country or in Germany (Aldashev et al., 2012;Basilio et al., 2017;Chiswick & Miller, 2009). We therefore exclude all individuals with a foreign highest vocational or school degree.
When these restrictions apply, the immigrant-native wage gap of all immigrant groups diminishes substantially: by approximately 4 ppts for Foreigners and by approximately 6 ppts for Naturalised Immigrants in higher deciles relative to the results of our main model (see Figure 4). The endowment effect of the wage gap for Naturalised Immigrants improves by roughly 20 ppts at the lower and middle pay deciles. For Foreigners, the explained part remains nearly unchanged. Our results point to the imperfect transferability of human capital across country borders and confirm its relevance in explaining the wage differential between natives and immigrants (Basilio et al., 2017). The scope of alterations in wage differences observed when comparing the full sample to the sample of persons with an education in Germany conform with the results of Aldashev et al. (2012). We therefore can assume that comparable educational qualifications are not appreciated to the same extent. However, restrictions also exist due to a lack of formal recognition of qualifications and due to labour market regulations.

HUMAN CAPITAL QUALITY
We consider the economic distance between one's home country and Germany at the time of immigration as a cross-country proxy for the quality of foreign schooling and work experience (Coulombe et al., 2014).
We assume that the more similar a country is in its level of development to that of Germany, the more equal educational standards are and the more likely a common knowledge base is to form with respect to the level of education. For this purpose, we use the relative gross domestic product per capita (GDP p.c.) and calculate the logarithmic function of the home country's percentage GDP p.c. in terms of Germany's GDP p.c. corrected by the logarithm for Germany's economic distance to itself: The logarithm of GDP p.c. is used to denote the marginal return of countries' levels of economic performance on its human capital endowment. The indicator range runs from -2 to infinite whereas values of greater than 0.5 can be classified as a large economic distance. The closer a value is to zero, the smaller the economic distance to the country of origin. Corresponding values of the original differences can be found in Table A. 10.
In the wage regression, we use the economic distance in absolute terms instead of German writing skills and find a significantly negative influence of larger distances on wages. The share of the explained wage gap remains almost unchanged whereby economic distance becomes the key driver of explanation. An overestimation of the low wage deciles for the group of ethnic German repatriates results in an increased wage gap explanation for Naturalised Immigrants (see Figure 5). For robustness, we alternatively use the "Human Capital Index" (HCI) provided by the World Bank as the absolute distance to Germany. The index measures the amount of human capital that a child born today can expect to achieve by age 18 based on risks of poor health and poor education that prevail in the country in which she lives. The HCI scale runs from 0 (insufficient) to 1 (comprehensive) (The World Bank, 2018). The HCI confirms the validity of GDP p.c. as an indicator for the quality of foreign schooling and work experience.

PERSONALITY TRAITS
To investigate potential differences in personality composition, we consider the 5-factor model of < Figure 6 about here >

CULTURAL DISTANCE
Finally, as a final channel of influence, we examine potential barriers to integration by considering metrics of immigrants' proximity to Germany based on their home countries' levels of cultural distance. From social norms in the labour market (e.g., work behaviour), it can be assumed that a strongly divergent culture of immigrants partly induces reservations from which personnel decisions may be influenced negatively.
While cultural distance shows a consistently significant negative impact on the wages of Naturalised Immigrants, the negative impact on Foreigners' wages is significant only for low wage deciles. In applying cultural distance to the wage gap decomposition, however, we respectively recognize an overestimation of Foreigners' and Naturalised Immigrants' endowment effects for the lower and upper ends of the wage distribution in contrast to the main model (see Figure 7). When we use cultural distance without further control variables, a strong explanation rate emerges for Foreigners, but not for Naturalised Immigrants.
Therefore, it can be concluded that the wage differences of Foreigners may be attributed to their original culture to a certain extent.

CONCLUSION
The assimilation of immigrants' wage levels with natives' wage levels serves as an important indicator of labour market integration. We therefore analysed wage differentials between native Germans and two Moreover, we can identify heterogeneity of the wage gaps of further ethnic subgroups relative to native Germans: foreigners from Turkey and the former Yugoslavia as well as ethnic German repatriates have suffered a stronger wage disadvantage than southern European citizens. Again, inadequate language skills can partly explain these gaps. Our results furthermore indicate a limited degree of human capital transferability or at least a lower appreciation of foreign educational degrees. The estimated wage gap for persons graduating in Germany is smaller at approximately 4 to 6 ppts relative to the results of the basic model. When testing for human capital quality applying the economic distance between the host and home country, the influence goes hand in hand with language skills. When taking the home country's cultural distance to Germany into account, we recognize no changes in the endowment effect. Contrary to our expectations, we also find no significant influence of personal traits (Big Five).
We further determined whether immigrant wage differences might be affected by different reforms concerning the labour market. Foreigners' average wage gap rise over time mainly due to a broadening of lower wage deciles in all age cohorts whereas the increase in the average wage gap for Naturalised Immigrants has been driven by the oldest workers. Age cohort results confirm an increase in wage gaps over time, especially for Foreigners. On the other hand, the wage gaps of Naturalised Immigrants tend to narrow in later age cohorts. In addition, we predominantly ascertain a stronger explanation from individuals' endowment and labour market characteristics showing that the human capital endowments of immigrants has deteriorated towards native Germans over time and with more recent immigration cohorts.
Previous public and private programmes for the social and economic integration of migrants in Germany are proving to be insufficient in effectively tackling this long-term challenge. A stronger recognition of foreign educational qualifications would favour career decisions made based on actual qualifications while fully exploiting existing and future labour force potential and lessening economic inefficiencies. Moreover, an improvement in immigrants labour market prospects could be precipitated by adjusting vocational training, which so far has been predominantly oriented towards labour market entry (extensive margins) rather than towards the activation of individual performance potential (intensive margins). Nonetheless, immigrants' efforts towards labour market integration must be continued to improve immigrants' prospects and to diminish the social disadvantaging and rejection of ethnic groups.    (2)  a) Stars refer to t-tests conducted on the equality of means for native Germans and respective immigrant groups; significant differences are indicated at the 1 % (***), 5 % (**), and 10 % (*) levels. Survey weights are integrated to counteract sample bias. b) Calculated for immigrant groups only; no tests are provided. c) Regional information refers to the federal state level (NUTS 1). d) Foreigners also include remaining foreigners who are not regarded as citizens from guest-worker countries.

Foreigners Naturalised Immigrants (a) Citizens of Turkey (d) Ethnic German repatriates (b) Citizens of southern European countries (e) Nat. immigrants without ethnic Germans (c) Citizens of the former Yugoslavia
Survey weights are integrated to counteract sample bias.
Covariates considered in the estimation include labour market experience, labour market experience squared, marital status, three skill levels, German writing skills, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, region type, the regional share of the foreign population, regional real GDP per capita, and the regional unemployment rate. Source: DIW (2017)

45-54 years
Survey weights are integrated to counteract sample bias. Covariates considered in the estimation include labour market experience, labour market experience squared, marital status, three skill levels, German writing skills, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, region type, the regional share of the foreign population, regional real GDP per capita, and the regional unemployment rate. We exclude Foreigners who immigrated in the later period. Source: DIW (2017) Figure 4: Blinder-Oaxaca wage decomposition for UQR  -educational degree completed in Germany

Foreigners Naturalised Immigrants
Survey weights are integrated to counteract sample bias. Covariates considered in the estimation include labour market experience, labour market experience squared, marital status, three skill levels, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, region type, the regional share of the foreign population, regional real GDP per capita, and the regional unemployment rate. Source: DIW (2017), SOEP 1984-2015. Own calculations.

Foreigners Naturalised Immigrants
Economic distance

Human Capital Index
Survey weights are integrated to counteract sample bias. Covariates considered in the estimation include labour market experience, labour market experience squared, marital status, three skill levels, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, region type, the regional share of the foreign population, regional real GDP per capita, and the regional unemployment rate. Source: DIW (2017)

Nat. Immigrants
Survey weights are integrated to counteract sample bias. Covariates considered in the estimation include labour market experience, labour market experience squared, marital status, three skill levels, German writing skills, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, region type, the regional share of the foreign population, regional real GDP per capita, and the regional unemployment rate. Source: DIW (2017)

Foreigners Naturalised Immigrants
Survey weights are integrated to counteract sample bias. Covariates considered in the estimation include labour market experience, labour market experience squared, marital status, three skill levels, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, region type, the regional share of the foreign population, regional real GDP per capita, and the regional unemployment rate. Source: DIW (2017)  age "year of survey" minus "year of birth".

Foreigners Naturalised Immigrants
Decompositions also include regional fixed effects, year fixed effects, regional type, regional share of foreign population, regional real GDP per capita, and regional unemployment rate.

Foreigners Naturalised Immigrants
Survey weights are integrated to counteract sample bias. Covariates considered in the estimation are labour market experience, labour market experience squared, marital status, three skill levels, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, regional type, regional share of foreign population, regional real GDP per capita, and regional unemployment rate. Source: DIW (2017) Survey weights are integrated to counteract sample bias. Covariates considered in the estimation are labour market experience, labour market experience squared, marital status, three skill levels, German writing skills, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, regional type, regional share of foreign population, regional real GDP per capita, and regional unemployment rate. We exclude Foreigners who immigrated in the later course. Source: DIW (2017) Survey weights are integrated to counteract sample bias. Covariates considered in the estimation are labour market experience, labour market experience squared, marital status, three skill levels, German writing skills, dummy variables for firm size, dummy variables for occupational class, dummy variables for industry, regional fixed effects, year fixed effects, regional type, regional share of foreign population, regional real GDP per capita, and regional unemployment rate. We exclude Foreigners who immigrated in the later course. Source: DIW (2017)

C. GERMANY'S HISTORY AS AN IMMIGRATION COUNTRY
Germany has experienced large waves of immigration in the recent past. Each of these immigration waves were based on different migration motives and altogether, they brought a great variety of cultures from different regions of origin to Germany. We distinguish between six immigration waves since the Second World War.
The first movement took place in the last months of the war as well as in the post-war period and was characterised by war refugees and displaced persons from Eastern Europe towards Germany. Around 12.5 million citizens from Eastern provinces of the German Reich (Reichsdeutsche) and ethnic Germans living in Eastern and South-Eastern Europe (Volksdeutsche) succeeded escaping to Germany's "heartland". A