THE BENEFITS OF MIGRATION: Cultural diversity and labour market success

Like most modern societies, Germany has experienced a significant inflow of migrants during the last decades.¹ In the years between 1987 and 2001, Germany even absorbed more immigrants in absolute numbers than the traditional immigration countries Australia and Canada together (Bade 2001). Regarding studies of migration and integration, two alternative approaches dominate the agenda: (i) the deficit-approach, emphasising that migrants are deficit actors compared to the native population, and (ii) the difference-approach, focussing on cultural differences.

We rely in this paper on the latter approach and assume that migration triggers cultural diversity in the host region. In turn, cultural diversity affects the host region positively. Following Ottaviano and Peri's (2006) model of multicultural production, different cultural groups provide different skills and services with positive impacts on regional growth and income. Niebuhr (2010) tested the assumption that innovations are more likely to occur in a culturally diverse environment because of different sets of knowledge and abilities. Her analysis of regional patent registrations in German regions finds a significant influence of a cultural diverse labour force on innovation output. Finally, in their study on the economic impact of diversity, Berliant and Fujita (2004) develop and test a model of knowledge creation and transfer that emphasises positive effects on production and innovation that derive from complementarities between different ideas and skills associated with immigrant diversity.

Adding to this literature, we extend the economic concept of cultural diversity by assuming that social processes are also affected by immigration. A combination of economic and social perspectives offers a new direction for migration research (Boswell and Mueser 2008). One particularly good example is the work of Florida (2002), who considers openness and tolerance as important factors in extending opportunity structures and achieving location advantages.

Against this background, we ask if a culturally diverse environment facilitates labour market chances for both foreigners and natives. Given the challenge in research regarding the different labour market performance of foreigners and natives, namely the (mostly) unobservable unequal initial conditions – loss of productive human capital due to migration, insufficient language skills or missing networks – we have chosen a quasi-experimental approach: our focus is on transitions after apprenticeship training. Because the German dual system distributes highly standardised certificates, apprentices who passed this barrier face comparable opportunities at the time of labour market entry. Analysing these transitions allows us to assess labour market success at an early but crucial stage of the occupational career.²

We define cultural diversity as the percentage of foreigners and the variety of different nationalities. On the basis of this definition, the amount of cultural diversity varies considerably in Germany, and it is to a large extent a result of past immigration. By using small administrative units to accurately illustrate spatial differences, we show that most rural areas are seldom culturally diverse, while the population of larger cities is characterised by a high share of foreigners from numerous countries. Besides, labour market chances also vary considerably across space and it is striking that most studies in the field of integration research neglect the regional dimension. This is particularly astonishing since foreigners are overrepresented in large cities (Damelang and Steinhardt 2008). In this paper, we measure cultural diversity at the county level and take into account that the local employment opportunities strongly depend on regional labour market conditions.

The remainder of this paper is organised as follows: the next section informs about immigration to Germany and the immigrant population in Germany. Secondly, we offer theoretical arguments on cultural diversity and its influence on labour market entry. Next, we present our quasi-experimental design including the institutional setting of apprenticeship training in Germany and the derived hypothesis. This is followed by a description of the data and our estimation strategy. Then, we show that a culturally diverse environment reinforces the transition from education to work. We conclude by relating our results to migration research in general.

The transformation of Germany into an immigrant-receiving country began within the framework of guest worker recruitment and subsequently accompanied by their family members. Labour recruitment from Italy began in 1955, from Spain and Greece in 1960, from Turkey in 1961, from Morocco in 1963, from Portugal in 1964, from Tunisia in 1965 and from Yugoslavia in 1968. By 1973, when Germany curbed its guest worker recruitment, more than 4 million people migrated to Germany. In the following years family reunification dominates immigration to Germany. The next phase in the post-war immigration began by the mid 1980s with the inflow of refugees and asylum seekers. The new migration wave has been dominated by people from Eastern European countries since ethnic cleansing and wars in former Yugoslavia brought thousands of refugees to Germany. Recent immigrants come mostly from Eastern Europe, Africa and the Middle East. By now, about 7.2 million people with foreign nationality (about 8.7 percent of the population) live in Germany, from whom the largest proportion are Turkish people and people from former Yugoslavia (Statistisches Bundesamt 2010).

Regional growth, regional labour market outcomes as well as the transition from education to work depend on a variety of location factors, e.g., endowment with capital and labour, infrastructure, market potential, sector structure, and research and development activities. All of these factors and the economic development itself are influenced by demography and migration. One essential question in this context is the magnitude of the economic impact of migration, another whether it is positive or negative for the domestic population (e.g., Borjas 1995). The aspect of diversity has only recently been considered in this discussion.

Cultural diversity can affect the economy in different ways. The economic literature (Alesina and La Ferrara 2005; Ottaviano and Peri 2006; Bellini et al.2008; Niebuhr 2010) stress that migrants from different countries provide complementary skills because of their culture-specific characteristics. These complementary skills can supplement the production process. In addition, innovations are more likely to occur because differences in knowledge and capabilities of workers from diverse cultural backgrounds drive R&D performance. The argument is that opinions and behaviour are more homogenous within than between groups. However, interaction between foreign and native workers is an essential condition for knowledge spillovers. If there is communication between the groups, these differences and alternative ways of thinking and behaving can contribute to the development of new ideas and patterns of thought.³ Consequently, German counties with a diverse workforce have a higher productivity than counties with a more homogenous workforce (Suedekum et al.2009). On a business level, ethnic diversity is associated with increased sales revenue and more customers due to more resources for problem resolution and better understanding of the pulse of the marketplace (Herring 2009). Another important point is the resolution of labour shortages and recruiting and retaining high quality staff, followed by greater innovation and enhanced market opportunities (Kraal and Roosblad 2008). Hence, economic and business growth is stimulated, resulting in a higher success rate regarding the transition from education to work.

In addition to the economic impacts, we assume also that social processes are affected by a culturally diverse environment which might improve labour market opportunities: everyday life experiences with other cultures increase with the share of migrants present, and prejudices or discrimination against foreign people are reduced (Pettigrew and Tropp 2000; Steinbach 2004). Hence, an open and tolerant cultural climate may emerge, which might be a location advantage, because it attracts creative people (Florida 2002).⁴ In a knowledge economy, creative people, e.g., engineers, IT-specialists, scientists, contribute to economic growth (Boschma and Fritsch 2009). Furthermore, the larger the share of different nationalities already active in the labour market, the easier it should be for young migrants to ‘mix in’, because their presence is acknowledged. Employers are familiar with different ethnic groups and/or are open-minded to hire migrants and work together with them, either because of experiences within their own firms or information from business contacts. Moreover, if the labour pool is ethnically heterogeneous, the costs for employers to accomplish their ethnic preferences are high. Discrimination in the hiring process should be less likely to occur. Additionally, because of their proximity to customers in regions where a particular group is represented disproportionately, employers may have incentives to hire foreign people of a certain nationality. Finally, cultural diversity might also be beneficial, particularly for migrants, for example through the provision of both formal and informal information useful for searching for a job. Ties that only provide access to information and make resources available only within a particular ethnic community may not be helpful or even turn into a trap, in particular for second-generation migrants (Wiley 1970; Zhou 1997). Following the idea of weak ties (Granovetter 1973), we expect that labour market relevant ties are not necessarily restricted to a person's own ethnic group. Rather, we assume that the probability of interethnic relationships grows with increasing cultural diversity (Ganter 2003; Munshi 2003).

The downside to the positive effects of cultural diversity are higher transaction costs in form of different languages and cultural barriers. Also, the possible trade-offs among participation in the ethnic economy, the size and concentration of ethnic neighbourhoods, and social integration referring to language acquisition and interaction with non-co-ethnics are to be considered (e.g., Lazear 1999, 2000; Boisjoly et al.2006). A diverse population might also disagree about the extent and variety of provision of public goods (Alesina et al.1999; Miguel and Gugerty 2005). According to the group threat theory (Blalock 1967), natives may not enjoy living together with a high number of foreign people, if they feel that their own opportunities are being affected negatively. Further, living in a large ethnic enclave moderates the pressure for ethnic adjustment to the host country and might even stimulate emerging alternate societies. Regarding positive and negative effects of diversity, there seems to be a range of diversity that should be beneficial for labour market outcomes depending on economical and social location conditions. Furthermore, Ottaviano and Peri (2006) emphasise the role of a core of shared norms (e.g., integration, constitutional legality) that might constitute a prerequisite for realising the potential benefits of diversity.

To test the impact of a culturally diverse environment on labour market success, we focus on labour market entry, a crucial transition for a person's employment history (Dietrich and Abraham 2008). We restrict our analyses to young people passing through vocational training, the most important track in Germany regarding labour market entry: over 50 percent of each cohort attends vocational training following school (Bundesinstitut für Berufsbildung 2009). Successful vocational training provides graduates with standardised occupational certificates that serve as an objective measure of skills and abilities to potential employers nationwide. Because the link between certificates and the access to qualified work is particularly pronounced in Germany (Shavit and Mueller 1998), the dual system of vocational training constitutes a major gateway in the economic integration of young people. There are two tracks for initial vocational training in Germany: first, the so-called dual apprenticeship, which takes place both ‘on the job’ and at vocational schools, and second, full-time training at vocational schools or ‘Berufsfachschulen’ as a specialised vocational school. In our study, we concentrate on the first track, the vocational training system (VTS) in companies.⁵

If we assume that labour market opportunities of young foreigners and Germans who have completed training in the VTS are comparable, selection effects regarding participation are important. Relative to Germans, the participation rate of foreigners in the VTS is low. In 2007, only one-quarter of non-Germans between 15 and 24 years of age participated in vocational training courses within the dual system, but more than half of the Germans of the same age did (Bundesinstitut für Berufsbildung 2009). One reason for this low participation rate is that young foreigners leave school to a disproportionate extent without school-leaving qualifications and subsequently find themselves in the so-called ‘transition system’ (Uebergangssystem), a part of the active labour market policy in Germany including (state-paid) periods of work experience and preparatory courses. Other reasons for the lower participation rate include pre-selection at school (Gomolla and Radtke 2002), where institutional barriers as well as effects of social background (Boudon 1974) result in fewer years of general education and lower levels of achievement among foreigners (Heath et al.2008). Controlling for test grades, Diehl et al. (2009) reveal a lower likelihood of young foreigners securing a vocational training position, probably due to poorer language skills and less social capital (Bommes 1996; Portes and Rumbaut 2001) or because employers ascribe potential for disturbance to young foreigners (Imdorf 2008). Relative to their German counterparts, their access to attractive VTS positions has been found to be restricted (Haeberlin et al.2005; Seibert et al.2009).

Against this background, foreign apprentices in the VTS are most likely not only better endowed with skills and more highly motivated than their ethnic peers – if they survived the selection process into the VTS, their assets are very likely to match or surpass those of young Germans. Bender and Seifert (1996) have shown that the career histories of foreign and German nationals completing training do not differ significantly in terms of occupational mobility and spells of unemployment. Thus, comparing foreign and native workers in a non-experimental setting, this approach provides labour market starting positions, which can be assumed to be almost equal.

In view of these considerations, we assume that a culturally diverse environment fosters the transition from apprenticeship training to work because cultural diversity

• a.

  is closely associated with economic growth and, thus, increasing employment opportunities due to innovations and complementarities,

• b.

  reduces reservations against migrants and foreigners and is associated with an open and tolerant climate whereby social barriers are overcome and employment opportunities arise, and

• c.

  supports interethnic networks which facilitate the transmission of information essential to labour market entry.

We expect that the higher the level of cultural diversity in the social environment, the greater the chance of a successful transition into the primary labour market should be.

The data for this analysis originate from the German Integrated Employment Biographies (IEB). The IEB database is generated at the Institute for Employment Research (IAB) by merging different sources of individual data collected by the German Federal Employment Agency for administrative purposes. The IEB includes life-course information about employment spells subject to social security contributions, unemployment benefits, participation in active labour market policy schemes, and job search. Although no information is available on other forms of employment, e.g., civil servants or the self-employed, the IEB covers more than 80 percent of the labour force in Germany. Hence, the IEB has become an important data base for research.⁶ It is a unique data base because of huge case numbers that allow for differentiation due to nationality, occupation, and regional labour market.

To analyse the impact of cultural diversity on labour market success, we select the first labour market episode following apprenticeship as the dependent variable.⁷ We differentiate two states: ‘employed’, which indicates a successful transition into the labour market, versus ‘unemployed’. Because they are not frequent, other possible episodes such as internships or participation in active labour market programmes are not considered.⁸

To cover the transition from apprenticeship to the labour market, we select all persons younger than 30 years of age in Western Germany and Berlin, who completed their apprenticeships in 2000.⁹ The eastern part of Germany is excluded from the analysis because the number of foreign people living and working there is small and diversity is not likely. The cohort of 2000 is selected, because the data allow for differentiating between subsidised and non-subsidised training from that year onwards. Subsidised training is an element of active labour market policy and is aimed at young people, who have no ‘regular’ training positions. Young people, who have completed subsidised training, have poorer labour market opportunities compared to those completing non-subsidised training (Damelang and Haas 2006). For this reason, they are excluded from the analysis.

Because information on completed apprenticeships in the data is unreliable, we take the duration of the training as a proxy. According to von Wachter and Bender (2006), who used the same data source, an apprenticeship is successful, if it is completed after a minimum duration of 450 days. The IEB provides no direct information about migration background or ethnicity, but it allows the identification of foreigners through citizenship. If foreign people become naturalised during the observation period, the original nationality is maintained for the analysis. The data allow for a differentiation of more than 180 nationalities (based on citizenship). We collapse this variety to five groups: ‘Germans’, ‘EU 15 Europeans’,¹⁰ ‘Ex-Yugoslavs’, ‘Turks’ and ‘Other Nationals’¹¹ for the rest of the sample representing the most important foreign groups in Germany.¹²

Table 1 describes all independent variables. These are, first of all, the five nationality groups as discussed above. Besides individual characteristics (sex, age and wage), two human capital indicators have been included: level of education and the number of apprenticeship trainings. For the latter, we control for switching the training firm as well as for second apprenticeship training. For the level of education, we distinguish between lower and upper secondary school as well as university.

In order to investigate whether the transition from apprenticeship to the labour market is hampered by labour market segmentation, we consider, first, the eight most popular occupations separately. Secondly, we control for the size of the training firm and other firm-level characteristics to observe if firm performance results in different training opportunities (average wage in the firm, proportion of highly qualified employees, share of foreigners, and share of apprentices). Thirdly, industrial sector is taken into account.

To capture the influence of cultural diversity spatially, we use small scale administrative units at the level of NUTS 3 regions, which more or less correspond to city regions or labour market areas.¹³ This delineation allows us to consider regional disparities in unemployment, productivity, and diversity for 326 regions in West Germany and Berlin. Because regional mobility is low among apprentices in West Germany (Riphahn 2002), the situation on the local labour market is crucial for success. For example, a high rate of unemployed youths lowers employment chances. In turn, a high GDP per year and the long-term growth of GDP (over 9 years) are expected to lead to a higher labour demand. We also consider the type of region (rural, agglomerated, and urban¹⁴ ) to account for differences in the ‘natural’ economic and social structure.

We use a probit model to examine the expected positive effects of cultural diversity on labour market entry. The conditional probability of entering the first labour market can be written as follows: coded with 1 for a positive outcome that means the event ‘employed after apprenticeship’ did occur and coded as 0 for a negative outcome meaning ‘unemployed after apprenticeship’.

To assess the influence of cultural diversity, we have to take into account that location choices of foreigners might be influenced by supportive conditions. We assume that foreigners stimulate employment growth, though, but the relationship might as likely be the other way round meaning that foreigners go to regions where they expect to meet favourable economical conditions. This influences the transition into the primary labour market positively, too. These considerations lend support to the notion that the share of foreigners is an endogenous regressor, which means that we have to disentangle the influence of cultural diversity from the share of foreigners. In order to identify the causal impact of share of foreigners on our dependent variable, we need a source of exogenous variation. Therefore, we apply a specific estimation procedure, the so-called Instrumental Variable (IV) estimation. The aim of the IV-method is to eliminate asymptotically the correlation between the error term and the endogenous variable by replacing it with other variables that are closely connected to it (relevance of the instrument), but which are not correlated with the error term (exogeneity of the instrument). In other words, the instrument z has no direct effect on the dependent variable y, but only indirectly through its impact on the endogenous variable x_k (Wooldridge 2002).¹⁵

One shortcoming of the IV technique is that the exogeneity of the instrument cannot be tested directly, but should be considered by the choice of the instrument. Hence, we applied different instruments like ‘share of foreigners (time lagged)’, ‘predicted change of the number of immigrants of each country’ like Ottaviano and Peri (2006), and ‘birth rate’. But for each of these instruments, either correlation with the error term cannot be ruled out or their correlation with the endogenous regressor ‘share of foreigners’ is not satisfying. Finally, we decided to use a political measure as a proxy for the ‘open-mindedness’ of a region towards migrants and foreigners, namely the share of Green Party¹⁶ voters at the Bundestag elections 2002. The manifesto of the Green party advocates ‘multi-culture’ and emphasises a policy of open-mindedness and tolerance towards minorities. This variable is sufficiently correlated with the ‘share of foreigners’ (R²=0.220¹⁷ ), but is per se unlikely to have an effect on labour market outcomes and, thus does not affect the transition from education to work.¹⁸ Furthermore, it is unlikely to be affected by reverse causation, because persons without German citizenship are not allowed to vote in Germany. That means that the instrument is exogenous, because the share of foreigners is not causally determined by the share of Green party voters.

To measure cultural diversity, we follow the approach of Ottaviano and Peri (2006). They define cultural diversity as the quantity and variety of groups of different cultural heritages in a region. Being unable to define cultural heritage properly with our data, we use nationality as a proxy. Although we may underestimate the magnitude of diversity, the advantage of our data is to distinguish between more than 180 different nationalities. Extending the approach of Ottaviano and Peri (2006), we calculate two indicators: one on the basis of the whole foreign labour force and, a second one, which is restricted to foreigners who are employed.¹⁹ The influence of both indicators might differ: The second indicator is purely associated with positive economic and social effects induced by a culturally diverse social environment of employed foreigners. Positive effects are more likely to occur if foreigners are integrated into the labour market. In contrast, a diverse environment with a relative large proportion of foreigners not integrated into the labour market makes it more difficult for young foreigners to access employment. We estimate two separate regressions, one for each indicator.

where s_ikt is the share of employees with nationality k in year t in county i.

A higher diversity index signifies a more diverse environment. The advantage of this indicator over similar concentration indices (e.g., the GINI coefficient) is that it accounts for both the richness of the distribution and a relatively even distribution across nationalities. Furthermore, it gives more weight to larger foreign groups. The diversity index will increase with a rising number of nationalities or when the shares of different nationalities converge. By controlling for regional labour market conditions and the (instrumented) ‘share of foreigners’, we are able to disentangle empirically the influence of cultural diversity and to examine whether cultural diversity of the workforce has a positive or negative impact on the labour market entry of people who have recently completed their apprenticeship. Due to the fact that the influence of diversity is not supposed to be a linear effect²¹ we decide to use a more flexible approach by employing quartiles that represent the whole distribution without assuming a linear function. We split the effect into four quartiles (up to 25, 25–50, 50–75, and over 75 percent) to differentiate counties with low, medium, and high level of diversity measured against the distribution for West Germany (range of the diversity index for employed foreign workers from 0.3325 up to 0.9585²² and for the whole foreign labour force from 0.5756 up to 0.9650²³ ).

We now turn to the description of our dependent variable, the transition from apprenticeship to the first labour market episode, and consider integration into the ‘primary labour market’ as the preferable status. Differentiating between ‘Germans’, ‘EU 15 Europeans’, ‘Ex-Yugoslavs’, ‘Turks’ and ‘Other Nationals’ Table 2 shows that Germans, nationals from EU 15 states, and nationals from former Yugoslavia have similar patterns regarding labour market integration: almost 85 percent of those who complete an apprenticeship enter the primary labour market. Turks have a higher risk of starting with an episode in unemployment (21 percent); only 79 percent immediately find a job in the primary labour market. The ‘other nationals’ have the lowest transition rate of all groups regarding the primary labour market. Given the large number of people who actually make it into the first labour market without delay, however, the German apprenticeship training system offers a huge capability for labour market integration of young people in general.

We argue that the transition from education to work is not only a product of human capital endowment and diminishing ethnic penalties, but also of the social environment and cultural diversity in the host region. To disentangle these effects, we analyse the conditional probability of transition into employment (vs. unemployment) for graduates. In order to cope with the endogeneity of the regressor ‘share of foreigners’, we use an instrumental variable estimation as described in Equations (1) and (2). A general probit model is likely to be miss-specified because ‘share of foreigners’ and ‘transition into employment’ might be determined simultaneously. To avoid that omitted variables drive the regressor ‘diversity’, we also included further labour market characteristics of the host region influential to the transition.

Table 3 shows the estimation results of Equation (2). Model 1 includes the diversity index based on employed foreigners and Model 2 is calculated with the diversity index based on the foreign labour force. As it is illustrated, we have controlled for human capital characteristics, occupations, as well as characteristics and sector of the training firm in both models.²⁴

First, we look at the probability of transition for the different national groups. Both estimations show that the probability of being employed in the first labour market following an apprenticeship is significantly lower for Turks and other nationals than for Germans. This result surprises as the fairly standardised German VTS with its combination of training at school and in firms should prevent young workers with insufficient language skills from qualifying. However, labour market research on Turkish youth suggest a special role among all foreign groups (Kalter 2006, 2007). Turks’ lower labour market chances result from their greater cultural distance and discrimination by employers (Steinbach 2004; Seibert and Solga 2005; Imdorf 2007). Additionally, Turkish young people have fewer social ties to Germans (Haug 2003), which restricts information on job vacancies. Since the majority of potential employers are natives, dense intra-ethnic networks lower chances for labour market integration (Montgomery 1992; Portes and Rumbaut 2001; Kalter 2006). In contrast, the transition probability of EU 15 nationals and young people from former Yugoslavia do not differ compared to Germans. However, the findings are in line with other research confirming the integration power of vocational training in Germany in general (Gangl 2003), and for foreigners in particular (Seibert 2005).

Turning to our hypothesis regarding the assumed beneficial influence of a culturally diverse environment on the transition process, it is evident that diversity has a positive impact in both specifications. The reference category representing the 25 percent regions with the highest diversity measure is the dummy variable. An extraordinarily diverse workforce is supposed to be counter-productive due to increasing transaction costs. The dummy variables for the first three quartiles show significant positive effects (Model 1), as expected. It is not appropriate to attach much significance to the small differences among coefficients for the quartile dummy variables, because their confidence intervals overlap. On the other hand, if the diversity index is calculated on both employed and unemployed foreigners (Model 2) the beneficial influence is exceeded at a lower level: In this specification, the dummy variable representing the third quartile is not significant anymore. This result is intuitive: young people experience difficulties accessing employment in a diverse labour market with a high share of unemployed foreigners. The costs exceed the benefits at an earlier stage. As a robustness check, we also used the conditional maximum-likelihood estimator.²⁵ Furthermore, we also estimate the effects of diversity excluding rural districts because of the slight extent of diversity. Results are robust due to different specifications and remain stable.²⁶

This positive influence of diversity can be explained by both economic and social processes, such as employment growth, everyday experiences, and interethnic networks. To assess the influence of diversity, the endogenous regressor ‘share of foreigners’ is instrumented by the ‘share of votes for the Green Party in 2002’. Otherwise the positive effects of migration and integration could not be interpreted causally since productive regions may act like a magnet attracting German and foreign workers. The instrumented variable ‘share of foreigners’ shows a weak significant negative effect in Model 1 indicating that diversity and not quantity is beneficial. This effect is insignificant in Model 2. We also test interaction effects between cultural diversity and nationality with the result that the influence does not differ significantly between the observed groups.²⁷ Hence, a high share of individuals with different cultural background fosters employment chances for all groups. This indicates that discrimination is (statistically) less likely in a diverse environment, and beyond that, labour market opportunities are promising.

In order to avoid that diversity is biased because of omitted variables, other factors at the regional level that matter for the successful integration of individuals in the first labour market are also included. It is obvious that a high regional unemployment rate lowers their chances, and that an increasing GDP amplifies the probability of their integration. Furthermore, finding employment in core cities is easier due to the fact of more adequate job offers (e.g., service sector) than in agglomerated regions.

Economic and social interactions depend not only on individual resources, but also on the opportunity structure influenced by migration and integration. Applying a difference-approach and considering migration as a societal resource, beneficial processes may be triggered leading to economic growth, social openness and, thus, improve the opportunity structure.

We analyse whether cultural diversity considered as the quantity and variety of foreign employees at county level extends the opportunity structure in terms of labour market possibilities. It is assumed that access to relevant information is easier, social barriers are overcome, and economic growth is stimulated if cultural diversity is high. To assess the impact of diversity, we examine the employment history of a cohort of graduates of the German dual system at the time of transition from vocational training to work.

The question of causality between economic growth and cultural diversity requires a special methodical framework: the quantity of foreign employees might not be independent from labour market entry because migrants usually settle in regions where they meet favourably economic conditions. An instrumental variable design is carried out to avoid a biased estimation: we choose the instrument ‘share of Green Party voters’ because tests confirm the validity and effectiveness as a proxy for the variable ‘share of foreigners’. To avoid omitted variable bias, we include regional labour market characteristics, since foreigners are unevenly distributed across German counties implying different labour market chances. Additionally we observe human capital endowment, occupational sorting, and labour market segmentation. Based on the estimation results we conclude that a culturally diverse environment enhances the opportunity structure for labour market success, especially the probability of transition from apprenticeship training to work rises.

This result is in line with a seminal paper by Ottaviano and Peri (2006) showing that cultural and ethnic diversity among the population increase regional productivity. Immigrants’ skills are complementary not only because they perform different tasks, but also because they bring different skills and abilities to the same task. For example, a Chinese engineer and an Italian engineer will neither provide the same ideas of style/concept because of different cultural background, ideas and various approaches. Florida (2002) and Florida et al. (2008) further argue that tolerance – specifically ‘low barriers to entry’ for individuals – is associated with the geographic concentration of talent, higher rates of innovation and regional development. A regional culture of tolerance acts on regional development by making other inputs, such as education and occupational skill, more efficient. Similar arguments by Page (2007) provide the basis for a general economic theory of tolerance and improved economic outcomes. He finds that cognitive diversity leads to a more efficient decision making process and that the associated diversity of identities enables new perspectives.

To sum up, labour market integration is considered to be a necessary prerequisite to participate fully in society. In addition to individual characteristics, the regional economic and social structures play a crucial role. From our results, the combination of economic and social processes in migration research is promising. However, and in part due to our quantitative approach, there are limits to our background story concerning the cause-and-effect-relation of cultural diversity. Further research should open the ‘black box’ of cultural diversity in order to disentangle direct and indirect effects regarding social processes.

Andreas Damelang studied social sciences at the Friedrich Alexander University of Erlangen-Nuremberg and at the Auckland University of Technology. Afterwards, he worked at the Institute for Employment Research. Since 2009 he has been a researcher at the University of Erlangen-Nuremberg. His fields of research are intergenerational mobility and labour market sociology, especially labour market integration of foreigners as well as occupational labour markets.

Anette Haas studied economics at the University of Regensburg, specializing in regional and municipal economics, econometrics and empirical macroeconomics. She has been working as a researcher at the IAB since 1997. Ms Haas' research tasks revolve around the analysis of regional labour markets and the special mobility of labour, as well as the disparities and dynamics of the labour market.