Once asylum seekers arrive in Germany, they are distributed geographically across the German regions. The number of asylum seekers that each region receives is based on a quota system considering tax returns and population size in each region (Königstein key). The allocation of individuals across those defined regions occurs randomly. This policy is subject to much debate. The system resembles a lottery that may produce winners and losers. An emerging body of research suggests that the initial placement of asylum seekers shapes their further integration trajectories into society (Chiswick and Miller, 2002; Åslund and Rooth, 2007; Aksoy et al., 2020). Local contexts may vary substantially in terms of educational, labor market and social opportunities they provide for migrants (Edin et al., 2003; Beaman, 2012; Godøy, 2017; Martén et al., 2019; Braun and Dwenger, 2020). Several initiatives have been launched to assess the potential of taking additional characteristics into account when matching asylum seekers to localities with the aim to increase integration outcomes such as employment (Bansak et al., 2018).¹ The societal benefits of improving geographic assignment appear large in light of the long-term disadvantage that asylum seekers and refugees face in terms of employment and earnings (Dustmann et al., 2017; Brücker et al., 2019; Brell et al., 2020).

In this study, we aim to explore the question whether asylum seekers that have been allocated to rural areas experience disadvantages compared to those allocated to urban areas. Some studies have shown that urban centers with a higher share of co-ethnic residents provide advantages in terms of economic integration (Martén et al., 2019). Higher concentration of co-ethnic networks reduce initial language barriers and information asymmetries when searching jobs. Urban areas may also provide more support to newcomers in terms of language learning opportunities or other support services in multiple languages. Rural areas–due to fewer available resources and fewer previous migration–may offer less support. Several initiatives have been launched in Germany to improve access to integration courses (providing language learning opportunities) in rural areas (Ohliger and Schweiger, 2019; Rösch et al., 2020; Fachkommission Integrationsfähigkeit, 2021). Research on co-ethnic networks and integration opportunity structures suggest that asylum seekers could be disadvantaged in rural areas. The available empirical evidence, however, is still limited (Rösch et al., 2020).

In this study, we explore potential rural penalties with a focus on language acquisition. Language skills are often highlighted as the main driver of positive integration trajectories (Esser, 2006; Kristen et al., 2016; Kosyakova et al., 2021) as they facilitate job searches, social integration, and correspondence with authorities or navigation of host-country institutions (Espenshade and Fu, 1997; Martinovic et al., 2009; Alba et al., 2011). In particular, we will assess several pathways that may explain differences in language acquisition between rural and urban locations based on a causal model illustrated by Directed Acyclic Graphs (DAG) (Elwert, 2013). We derive testable hypotheses based on language learning models initially developed by Chiswick and Miller (2001) and later extended and applied by various authors (e.g., Kristen et al., 2016; Kosyakova et al., 2021).

Based on large longitudinal survey data in Germany (SOEP IAB-BAMF refugee sample; N = 13,187), we first test whether there is, indeed, a rural penalty in language acquisition of asylum seekers. Second, we explore whether potential urban-rural disparities are related to differences in social networks (i.e., exposure to German speakers) and learning opportunities (access to language courses). Research has shown that contacts with natives (Bauer et al., 2005; Heath et al., 2008; Danzer and Yaman, 2013) and participation in language courses (Clausen et al., 2009; Vroome and van Tubergen, 2010; van Tubergen, 2010; Kaida, 2013; Hoehne and Michalowski, 2016; Sarvimäki and Hämäläinen, 2016; Auer, 2018; Lochmann et al., 2019; Arendt et al., 2020; Kosyakova and Brenzel, 2020) have strong and lasting effects on integration outcomes such as language acquisition and employment.

While previous research has largely discussed individual mechanisms in isolation, we propose a broader framework that incorporates different forms of opportunity structures for language acquisition of refugees depending on their geographic location. The geographic dimension of integration of refugees was neglected previously, largely due to the lack of suitable data sources. There is ongoing discussion on how particular contexts shape integration outcomes for example with respect to concentration of co-ethnic/ migrant networks, contacts to non-migrants, local employment rates, and state-funded integration support initiatives.

Our results show that (1) there is no overall rural penalty in refugees' language acquisition in Germany, (2) both contact with Germans and participation in different forms of language courses proof to be highly effective in increasing refugees' language acquisition and (3) intergroup contact with Germans is significantly more likely in rural areas while official course participation is somewhat less likely. Overall, it can be concluded that language learning in rural areas runs to a greater extent via contacts with Germans, while in urban areas institutional services are a more relevant factor.

In addition to advancing understanding of how local contexts shape integration of refugees, these results have implications for policy. The federal government is responsible for the allocation of refugees across regions and regional authorities are responsible for allocation of refugees to districts. The findings reject the claim that refugees are disadvantaged in rural areas in terms of language acquisition, partly because higher exposure to German speakers offsets marginally lower access to formal language courses. The results also suggest that further investment in courses in rural areas and more opportunities for interactions with Germans in urban areas could accelerate language acquisition of refugees and thus maximize integration benefits for refugees and society.

According to the Chiswick-Miller language learning model, host-country language acquisition is a function of efficiency, incentives and exposure (Chiswick and Miller, 2001). Efficiency captures factors that facilitate individual language learning such as prior education attainment, young age, and cognitive skills. Incentives reflect the motivation of the language learner and are driven by expected economic (i.e., income) and non-economic (home-country attachment, social exclusion) returns. The incentive dimension incorporates costs associated with language learning such as fees for instruction, material costs or opportunity costs associated with delayed transition to gainful employment. Incentives are commonly modeled as a (rational) cost-benefit calculation by the individual migrant.

Exposure – the main dimension of interest for this study – refers to “the degree to which the new language is present in contexts that immigrants encounter” (Kosyakova et al., 2021). Exposure incorporates structural language learning opportunities such as courses and interactions with native-speakers.

In this study, we are interested in potential disadvantages of residing in a rural area with regards to language learning among recently arrived asylum seekers. In particular, we are interested how exposure to German native-speakers (through every-day interactions) and access to formal language courses mediate potential effects of location on language learning.

In the following, we put forward our theoretical arguments formalized by the means of Directed Acyclic Graphs (DAGs). DAGs are a tool to illustrate the causal model, make assumptions transparent and derive formal rules for selecting control variables (Elwert, 2013; Morgan and Winship, 2014).

The main interest of this study is the total causal effect of location (urban vs. rural) on language acquisition (in the form of skills) (path 1 in Figure 1). More explicitly, we are interested in the role of the two indirect effects (mediators) of language courses (M2) and contacts to Germans (M1). The positive effect of contacts with native speakers (path 4) and language course attendance (path 5) on language acquisition is already well established (Niehues et al., 2021; Kristen et al., 2022). The focus of this study is how refugees living in rural areas are affected by both courses and contacts relative to refugees living in urban areas in terms of language acquisition.

There are several reasons to assume a negative effect of a rural location on language course participation (path 2). First, rural communities often provide less assistance in language learning. Rural communities have fewer resources to fund language learning opportunities (Schader Stiftung, 2011; Ohliger and Schweiger, 2019; Scheible and Schneider, 2020). Second, even if resources are available, rural areas do not benefit from scaling effects due to lower population size and density. In other words, if fewer asylum seekers are present, certain investments in support measures such as integration courses may not be deemed cost-effective (Ohliger and Schweiger, 2019; Scheible and Schneider, 2020). Third, rural areas have lower levels of previous migration which indicated less experience with managing diversity and established support policies (Rösch et al., 2020). This could mean that available support is of lower quality or consistency. Fourth, courses may be available in neighboring localities but too difficult to access given the distance and lower public transport provision (Scheible and Schneider, 2020). Fifth, migrants in rural areas may be less incentivized to learn the languages because there are lower expected returns to the investment given that fewer and worse jobs are available in rural areas compared to urban areas.

In contrast, there are several reasons to believe that asylum seekers living in rural areas have more exposure to native-speakers (path 3) compared to asylum seekers living in urban settings. First, the opportunity to interact with other co-ethnics is likely smaller because the concentration of migrant groups is historically lower in rural areas compared to cities (Luft, 2011; Berlinghoff, 2018). Many migrant groups in Germany settled in cities following the economic boom after World War II. Still today the proportion of migrants is much higher in cities than in the countryside (Beauftragte der Bundesregierung für Migration, 2021). Living in urban areas may offer newcomers more employment opportunities and more inter-ethnic support in navigating host-country society, however, it may be a disadvantage in terms of language learning because of fewer interactions with native speakers (Chiswick and Miller, 1996, 2002; Bauer et al., 2005; Kanas et al., 2012; Danzer and Yaman, 2013; Chiswick and Wang, 2019). Public infrastructure in cities (in terms of mobility, basic services and health) reduces reliance on personal social contacts in general. In rural areas the principle of mutual assistance between neighbors, friends and families is more important in light of weaker public infrastructure. In rural areas, it may therefore seem more likely that asylum seekers will need to enter into contact with German speakers, e.g., with their neighbors, in order to help each other in everyday life. Finally, an opposing relationship between rurality and contact with Germans is also conceivable. It is known from previous studies that populations in cities tend to be more tolerant of migration (Bangel et al., 2017). This openness could also lead to more frequent and more intensive contact. It remains an empirical question for this study which of the opposing associations overweigh in this regard.

In sum, we suggest that contacts to Germans and access to formal integration courses condition any effect from rural location on language acquisition (full mediation). Based on these theoretical reflections, we consider the following potential effects of rural area on language acquisition:

An obvious contention that may threaten the causal interpretation of our findings is selection. Asylum seekers with particular (observable or unobservable) characteristics may be more likely to live in rural areas (Rösch et al., 2020). The same characteristics may be associated with better access to integration courses and interactions with Germany (Z1). For example, younger and more educated migrants may sort themselves into urban contexts to seek better employment and more attractive lifestyle opportunities. Parents with children may prefer rural areas with lower living costs and cheaper rents while parents, particularly women, have less availability to participate in language courses (Tissot et al., 2019). Certain asylum seekers from countries with low recognition rates facing legal obstacles to enter formal language courses could be concentrated in certain locations in Germany.

To overcome this problem, we make use of a unique feature of the German asylum seeker distribution system (sometimes referred to as settlement policy in other countries). In Germany, asylum seekers are randomly allocated to a particular region and then quasi-randomly distributed further to counties. No information about the individual asylum seekers is considered when assigning a location. This is an ideal situation for causal identification resembling a natural experiment. In addition, according to a new policy, the residence of recent asylum seekers is limited to their place of first assignment. This mobility limitation assures that the composition of asylum seekers is similar across localities which – in theory – should also render the population assigned to rural and urban areas non-selective. For any potential imperfections of this random assignment that may occur in practice, we take several measures further described in the following sections. Furthermore, we conduct additional analyses in sub-samples in which we control for factors that reflect the intention and experience of living in the countryside, thus ruling out further potentially endogenous variation (Supplementary Figure A1).

This study explored the potential disadvantage that asylum seekers and refugees may face in terms of language acquisition when being allocated to rural areas after arriving in Germany (i.e., “rural penalty”). We propose a causal model based on DAGs and established language learning models and test our hypotheses using large survey data from Germany (SOEP Group, 2020). We find that asylum seekers and refugees in rural areas do not have lower language skills compared to urban contexts (null effect). We find that asylum seekers and refugees in rural areas benefit from higher levels of interaction with German speakers while the disadvantage in terms of access to structured language courses appears minor. Overall, the results support a “compensation effect” whereas migrants compensate small disadvantages in terms of access to courses with higher exposure to Germans.

These results have implications both for academic and policy discussions. Germany received several million asylum seekers since 2012. Migrants often came from war torn countries with – on average – lower educational backgrounds. Integrating asylum seekers and refugees into society and allowing for equal participation is a major challenge. Acquiring the German language is key to integration and the largest area of public investment by the government. In this context, it is striking that the evidence on how local contexts influence integration outcomes is severely limited despite much debate regarding the issue. Rather than testing very narrow hypotheses, our approach allowed us to study how various mechanisms may offset each other within a more comprehensive causal model of language acquisition. Our findings are consistent with previous literature in the sense that we find large positive effects of both contacts to Germans and participation in language courses on language acquisition. However, we show that these mechanisms are more or less pronounced depending on the local context.

The policy debate often centers on the allocation scheme of asylum seekers in Germany (Königstein key) and the degree to which it produces winners and losers in terms of integration. We find that rural areas do not necessarily disadvantage migrants in terms of language acquisition. This finding is important as allocation of migrants to rural areas has been discussed in the context of reviving areas suffering from demographic decline. The results also suggest that policymakers can further promote language acquisition of asylum seekers and refugees by improving access to formal language courses in rural areas and by facilitating interaction with Germans in urban areas. Especially the latter informal mechanism via social contacts is often neglected in the political debate, with an overly rigid focus on language courses instead. Yet intergroup contacts can also achieve other socially desirable effects in addition to migrants' language acquisition, such as reducing xenophobic attitudes within majority groups (Savelkoul et al., 2017; Khalil and Naumann, 2021). Therefore, it could also be part of a targeted integration policy, for example, to select accommodation for refugees according to local contact opportunities and to specifically avoid too much ethnic segregation (Ziller and Spörlein, 2020).

The study faced two main limitations. First, our findings are based on observational data which limits the approach to establish causality. However, we attempted to make our causal assumptions explicit using DAGs. To further strengthen our causal claims, we benefit from the allocation policy in Germany which randomly allocates asylum seekers across regions mirroring a natural experiment. In the Supplementary Material, we also address reverse causality issues using panel fixed effects models. Second, the data only contains few years since asylum seekers arrived in Germany. Future research should study the long-term effects of being allocated to rural areas on language acquisition. In addition, we encourage further research to explore “rural penalties” with respect to other relevant integration outcomes such as employment, education, health, social exclusion and life satisfaction.

Despite these limitations, this study offers a comprehensive view of different pathways of language acquisition among asylum seekers and refugees and how they may differ between rural and more urban areas.

Publicly available datasets were analyzed in this study. This data can be found at: https://www.diw.de/en/diw_01.c.357906.en/soep_order_form_mod.html.

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

SK: concept, design, analysis, and writing. JT: concept, design, and writing. UK: design. All authors contributed to the article and approved the submitted version.

Staff funding was provided by the German Federal Ministry for Education and Research. Funding for open access provided by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Projektnummer 491466077.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.