Immigration and Swedish Schools

Hanif Bali investigates the share of pupils in Sweden who fail to quality for high-school (“obehöriga”),  concluding that recently arrived immigrants cause declining outcomes. Bali is however wrong in one regard. Since outcomes among non-recent immigrants is unchanged, he concludes that “it is only the results of children who have migrated after the school start which are lowering school results”

When analyzing the impact of immigration on the national average, it is not sufficient to look at the mean of immigrant outcomes. It is also important to look at how changes in the immigrant population share affects the national average.

Between 2000-2010 the share of Swedish pupils who do not quality for high-school increased somewhat, from 10.6% to 11.8%.  Interestingly the share of native Swedes who failed remained constant at 9%. The decline in the national average is entirely driven by students who have a foreign background. 

The increase in the population share of second generation immigrants alone appears to have caused about 30% of the decline in aggregate outcomes, even though the group did not perform worse than they did ten years ago

DNs editorial page writes that it is wrong to link immigration to the education debate, noting that “recently arrived immigrants are too few to have a decisive impact” and that citing immigration as the cause of declining school results would “lead the debate into a side track”. 

A variable hardly has to be “decisive” to be discussed. As long as immigration is causing part of the decline in Swedish outcomes it should be part of the debate. This especially because we are not aware of a single “decisive” cause of declining school results, a phenomenon which may have multiple causes. If you want to explain the decline of high-school pupils who qualify for high-school between 2000-2010, immigration was indeed “avgörande”, causing 100% of the decline. Immigration is an non-negligible part of declining PISA scores and is a major driver (perhaps the main driver) of increased inequality in Swedish school outcomes. 

It is not a valid argument to say that immigration cannot drive education outcomes because “recently arrived immigrants” are too few. You also have to look at non-recent and second generation immigrants, who also have lower average scores than native Swedes.

Another thing DN doesn’t appear to understand is that the test scores of native Swedes is not unrelated to immigration. Immigrant children on average need more help and take educational resources from native children. If immigration increases the level of disruption in schools it significantly hurts native education outcomes. Some education researchers believe that students who are in classes with worse pupils perform worse themselves. There is some empirical evidence that immigration lowers the test-score of natives, thought it likely depends on country of origin and the type of immigrants. 

The reason the DN editorial page doesn’t want the role of immigration for education outcomes to be discussed is not that it is not relevant. It’s that they championed large-scale immigration and don’t like it when voters find out about the negative effects of these policies.


In the comments Per Gudmundson point out this study, which finds that native Swedes do worse in schools with a higher share of immigrant pupils. This is a correlation study so we should not interpret this is having established causality. There exists data on exogenous placement of refugees in municipalities in Sweden during 1985–1994, used in several high-quality studies. Someone should use it to look at school-results.

As I have written about previously, the poor performance of immigrants does not explain more than a part of the huge decline in Swedish Pisa-scores. However it explains far more of the increase in the share of pupils who fail primary school. Immigration appears to be a more important explanation for what is happening among the worst-performing pupils, and not as important for the average, since immigrants are still few. In Pisa non-native students are a disproportionate share of the worse performers. My guess is that poor language skills and disruptive school-environments are the cause for this.


Comments are closed.