In the wake of the 2015/16 refugee migration from Syria to Europe and North America, Dustmann et al. (2016) and Hatton (2018) called for a renewal of the European refugee resettlement system. However, what an optimal allocation mechanism should look like remains an open question. The strict allocation in Germany into high-unemployment regions, for instance, negatively affects refugees’ economic and social integration (Aksoy et al. 2020). Fasani et al. (2016) argue that free residential mobility might foster labour market integration. Importantly, existing research has investigated the integration of the refugees themselves. Whether and how backgrounds and local conditions affect not only the first generation but also the children of refugees are thus far unknown.
The ability to communicate is especially important for mothers-to-be. Communication can be facilitated by co-nationals that provide important childbearing information or by proficiency in the local language (Lazear 1999). Language proficiency is regarded as one of the most critical components of immigrants’ human capital, decisive for their successful participation in the host society. Communication ability is associated with labour market integration (Auer 2018), electoral participation (Houle 2019), and social capital (Cheung and Philimore 2014). In the context of forced displacement, language training for recently arrived refugees is particularly important. The understandable lack of preparation and often the absence of ties and social networks require that most refugees learn the language of their destination country (Foged et al. 2021).
In a recent paper (Auer and Kunz 2021), we exploit a peculiarity of Switzerland’s asylum policy: newly arriving refugees are randomly allocated across states and, by consequence, across the country’s three distinct language regions (German-, French-, and Italian-speaking regions). This exogenous distribution results in refugees – who might originate from countries with a large French-speaking or Italian-speaking population – being by chance allocated to a familiar or unfamiliar language environment (see Figure 1). The additional fact that the allocation is binding – i.e. refugees are not allowed (with very few exceptions) to move to another region – allows us to follow their integration trajectories over time.
Figure 1 Swiss language regions and significant French- and Italian-speaking populations across the world
We study the wellbeing of infants born to refugee mothers from the same country of origin and with otherwise comparable characteristics allocated to familiar versus unfamiliar language environments. For example, a pregnant refugee from Cote d’Ivoire (French-exposed) who is randomly allocated to Geneva in the French-speaking region of Switzerland might be more likely to acquire information on where to get pregnancy check-ups and be able to follow the doctor’s advice compared to her pregnant co-national who is randomly allocated to Zurich in the German-speaking region. In addition, she might benefit from other aspects of her local surroundings; being able to read the ingredients on food packaging, for example, may facilitate positive health behaviours.
Our data comprise administrative accounts of all refugees who arrived in Switzerland between 2010 and 2017 and all childbirth events in the country with detailed health information. We find no evidence of any compositional differences, selection imbalances, or differential fertility choices between refugees whose language matches that of the region to which they are allocated and those whose language does not. In other words, being randomly assigned to a familiar language environment does not make someone more likely to give birth, nor does it change the timing of pregnancies.
Our primary outcome of interest is the child’s birth weight. Birth weight predicts educational attainment, income, credit default, and health later in life, among many other outcomes. Comparing refugees from the same origin across destinations regarding their babies’ birth weight (Figure 2) shows a remarkably consistent pattern. The grey line shows the birth weight of children of refugees without language correspondence in Switzerland (e.g. Farsi-speaking refugees from Afghanistan). The average birth weight is almost flat, with slight benefits for those in the German-speaking region, which is the economically most prosperous. Conversely, the blue lines (light: any French exposure; dark: French as an official language) show large birth weight gains, both in bilingual but predominantly French-speaking and exclusively French-speaking regions in Switzerland.
Finally, the green line shows that refugees from countries with sizeable Italian-speaking communities (Libya and Somalia) benefit substantially from being allocated to the Italian-speaking region (this is a small group consisting only of two sending countries and one region in Switzerland; thus, the results from these groups should not be overinterpreted).
Figure 2 Birth weight of refugees’ children by parent’s origin language exposure and allocated birth region main language
This analysis shows that the benefits of being able to communicate are large. We find that, on average, the children of mothers who by chance were allocated to a familiar language environment weigh 72 grammes more than children of co-national mothers who arrived in Switzerland at the same time but were allocated to an unfamiliar language region. Relative to the average birth weight in our refugee population of about 3,200 grammes, this amounts to a 2.2% increase in birth weight. (In the paper, we show that these conclusions hold in various regression specifications that hold fixed the local environment, origin country, and arrival time differences, accounting for well-known risk factors for child health).
The effect is present not only at the average of the birthweight distribution but also at the lower tail, where weight changes can substantially alter infant and later-life wellbeing: the clinical Low Birth Weight indicator (weight < 2,500 grammes) decreases by 2.9 percentage points (from a mean of 6.98%). These effects are substantial when compared to targeted interventions; for example, they correspond to a $3,867 yearly tax credit for low-educated Black mothers in the US (Hoynes et al. 2015) or are two to three times more effective than attendance at nutritional food programmes for low-income mothers in the US (Rossin-Slater 2013).
In addition, we find evidence that a more extensive local network substitutes for part of the benefits of being placed into a familiar language environment, mainly when this network comprises relatively well-informed refugee mothers with newborns. In other words, the mother-to-be’s reliance on communication skills goes down when informed co-nationals are present. While these networks are less likely to influence the quality of interactions with doctors or the language match with doctors and health personnel, they seem important for sharing knowledge about health services and health-related behaviour.
We find no evidence that possible earnings differences drive our observed effects through employment nor by (language) assimilation through more extended residence in Switzerland. Importantly, we already observe higher birthweights for mothers-to-be who arrive in Switzerland pregnant, indicating that assimilation cannot explain our findings. Conversely, we observe no differences in gestational age, which is often associated with exogenous shocks occurring in pregnancy (such as losing a partner or experiencing a natural disaster). Moreover, the positive effect of a familiar language environment on infant health is greatest when mothers originate from countries in which deliveries are usually not attended by medical staff – that is, when they are less likely to be familiar with the services of sophisticated healthcare systems, such as regular check-ups.
Implications of our findings
On an immediate policy level, our results highlight the importance of early policy interventions to counter the systematic disadvantages of vulnerable groups. Measures include language training, comprehensive interpretation services for recently arrived refugees, and language-adequate information campaigns to raise awareness about available welfare and health services.
On a more general level, we echo arguments against the current form of ‘spontaneous’ allocation of asylum seekers and in favour of a more comprehensive resettlement programme (Hatton 2016). In a large survey, Bansak et al. (2016) found that European citizens support a fair allocation across countries of the arriving refugees. From a policy perspective, recent attempts to improve the allocation of refugees are manifold. Bansak et al. (2018) and Acharya et al. (2019) suggest an algorithmic approach to maximise the potential earnings of refugees; Delacrétaz et al. (2016, 2019) and Jones and Teytelboym (2018), on the other hand, favour a preference-based matching algorithm respecting refugees’ preferences. Neither of these approaches, thus far, has taken the next generation into account. Our results, however, suggest that fostering the wellbeing of children is significantly affected by how their parents are allocated across countries and regions and thus should be part of the discussion on fair and optimal allocation procedures.
Acharya, A, K Bansak and J Hainmueller (2019), “Matching refugees to host country locations based on preferences and outcomes”, eScholarship, University of California.
Aksoy, C G, P Poutvaara and F Schikora (2020), “First time around: Local conditions and the integration of refugees”, VoxEU.org, 11 December.
Auer, D (2018), “Language roulette – the effect of random placement on refugees’ labour market integration”, Journal of Ethnic and Migration Studies 44(3): 341–62.
Auer, D, and J S Kunz (2021), “Communication barriers and infant health: Intergenerational effects of randomly allocating refugees across language regions”, No. 867 GLO Discussion Paper.
Bansak, K, J Ferwerda, J Hainmueller, A Dillon, D Hangartner, D Lawrence and J Weinstein (2018), “Improving refugee integration through data-driven algorithmic assignment”, Science 359(6373): 325–29.
Bansak, K, J Hainmueller and D Hangartner (2016), “How economic, humanitarian, and religious concerns shape European attitudes toward asylum seekers”, Science 354(6309): 217–22.
Cheung, S Y, and J Philimore (2014), “Refugees, social capital, and labour market integration in the UK”, Sociology 48(3): 518–36.
Delacrétaz, D, S D Kominers and A Teytelboym (2016), “Refugee resettlement”, University of Oxford Department of Economics Working Paper.
Delacrétaz, D, S D Kominers and A Teytelboym (2019), “Matching mechanisms for refugee resettlement”, No. 2019-078.
Dustmann, C, F Fasani, T Frattini, L Minale and U Schӧnberg (2016), “On the economics and politics of refugee migration”, VoxEU.org, 18 October.
Fasani, F, T Frattini and L Minale (2018) “(The struggle for) refugee integration into the labour market: Evidence from Europe”, VoxEU.org, 9 April.
Foged, M, L Hasager and G Peri (2021), “Language training and refugees’ success”, VoxEU.org, 20 March.
Hangartner, D, and L Schmid (2020), “Migration, language, and employment”, unpublished manuscript, ETH Zurich and University of Lucerne.
Hatton, T (2016), “The migration crisis and refugee policy in Europe”, VoxEU.org, 23 May.
Houle, C (2019), “Religion, language, race and ethnic voting”, Electoral Studies 61: 1–11.
Hoynes, H, D Miller and D Simon (2015), “Income, the earned income tax credit, and infant health”, American Economic Journal: Economic Policy 7(1): 172–211.
Jones, W, and A Teytelboym (2018), “The local refugee match: Aligning refugees’ preferences with the capacities and priorities of localities”, Journal of Refugee Studies 31(2):152–78.
Lazear, E P (1999), “Culture and language”, Journal of Political Economy 107(S6): S95–S126.
Rossin-Slater, M (2013), “WIC in your neighborhood: New evidence on the impacts of geographic access to clinics”, Journal of Public Economics 102: 51–69.