When large numbers of refugee children arrive in a school system, policymakers face a difficult question. Should they be integrated directly into regular classrooms with language support? Or should they first be placed in separate preparatory classes focused on language learning? The answer is not immediately apparent, and various countries have adopted distinct strategies.
Evidence from Hamburg
Willem De Cort sent me this recent study from Hamburg, published in The Economic Journal (Schilling & Höckel, 2025). The study offers rare causal evidence on this issue. Using administrative data and the quasi-random assignment of refugee families to schools, the authors compared children who entered regular classrooms with those who started in preparatory language classes.
Their findings are sobering. Refugee children who began in preparatory classes performed significantly worse on fifth-grade standardised tests—especially in German and mathematics—than peers who were directly integrated into regular classes. They were also slightly less likely to enter the academic secondary track (Gymnasium). The effect was strongest in the early years. In third grade, the gap was even larger, although by seventh grade, it seemed to have diminished.
Why would this happen?
One explanation is straightforward: children in preparatory classes had fewer opportunities to interact with native peers. Therefore, they had less exposure to the language as it is naturally used in school settings. In addition, the heavy emphasis on German left less instructional time for maths, science, and other subjects. Socially, many children from preparatory classes continued to be clustered with peers from similar backgrounds when they transitioned to secondary school, reinforcing linguistic and social separation.
Not the same everywhere
But the picture is not entirely one-sided. The study also finds that the effects varied depending on school context. In schools with relatively few migrant students, preparatory classes were clearly detrimental to their academic progress. In schools with a high share of migrant students, however, the effects were less harmful and sometimes even slightly positive. This nuance matters: integration policies do not operate in a vacuum, and the broader peer environment appears to influence their effectiveness.
Limitations of the study
It is also important to stress the limitations. The study examines short-term outcomes, including test scores and track choice, through the seventh grade. Longer-term consequences for language mastery, well-being, or labour market opportunities remain unknown in this study. Moreover, as the authors note, the results apply to Hamburg during a period of unusually high refugee inflows. Contexts with different resources, teacher training, or class sizes may yield different outcomes.
Equally important, the study compared two clear-cut models: full-time preparatory classes versus full integration from the start. Hybrid arrangements, where children might spend part of the week in a special group and part in mainstream lessons, were not examined. The findings, therefore, cannot be automatically extended to such mixed models which I’ve seen being proposed too in other places.
What to take away
So, what should we take away from this? At the very least, the findings challenge the assumption that a separate year of language preparation automatically helps children “start on the right foot.” Language is not just learned in classrooms—it is learned through participation, relationships, and access to the whole curriculum. That does not mean preparatory classes should never be used. However, it does suggest that policymakers need to carefully weigh the trade-offs and consider complementary measures that prevent isolation.
As always in education, there is no single recipe for success. The study highlights the importance of testing well-intentioned integration policies against real outcomes. Sometimes, what feels intuitively correct, such as giving children a protected space to learn the language first, may inadvertently slow down their broader academic integration.
Abstract of the study:
We study the effect of separate preparatory language learning classes on the academic outcomes of primary school-aged immigrant children in Germany compared to their direct integration into regular classrooms. Using administrative panel data and leveraging idiosyncratic assignment of refugee children to neighbourhoods, and, consequently, schools, as well as preparatory class roll out over time, we find that primary school-aged refugees attending a preparatory class perform significantly worse on fifth-grade standardised tests and are slightly less likely to pursue an academic secondary track. While limited to short-term outcomes, our results indicate that preparatory classes could impede early academic integration by clustering migrant peers, highlighting the need to consider complementary approaches to reduce achievement disparities.
