Governments invest heavily in language training, but integration policies are often designed as if they work equally well everywhere. As this research shows, they do not: the value of language training depends on local labour-market conditions, and the same national programme produces different outcomes depending on where immigrants settle. Swedish for Immigrants (SFI) is one of Sweden’s main publicly funded language programmes and a cornerstone of integration policy, and language proficiency has recently become a mandatory requirement for citizenship as of June 2026. Yet despite SFI’s prominence, there is surprisingly little evidence on whether participation actually helps immigrants enter employment, and almost no research on whether that effect varies by region of settlement. A new register-based study addresses both questions, following immigrants in Swedish register data from arrival until first employment and uses discrete-time hazard models with inverse-probability weighting to account for the fact that SFI participants differ systematically from non-participants based on observable characteristics including education, income, household circumstances and migration background (Khaef, 2026). Municipalities are grouped by population size and labour-market type, from metropolitan areas through to large cities, small towns and sparsely populated areas.
Does Swedish for Immigrants Programme speed entry into work?
The first finding is clear: immigrants who attend SFI move into their first job faster than those who never attend, once differences between the two groups are accounted for through inverse-probability weighting. This finding contributes to earlier studies that documented an association between SFI and labour-market activity (Khaef, 2025): the present analysis provides evidence of a genuine acceleration of employment entry, in line with international findings on the value of formal language training.
Where you live shapes the return to language training
The study’s central contribution is to show that this return is not spatially uniform. Value of SFI participation is greatest in metropolitan and large cities. There, SFI participants show the strongest employment prospects of any group, while immigrants in the same cities who do not attend SFI programmes face a distinct penalty relative to peers elsewhere. The metropolitan advantage that earlier research has noted is therefore conditional on language training: it accrues to SFI attendees, whereas non-participants are actually disadvantaged in metropolitan areas and larger cities. In sparsely populated areas the picture differs: both non-participants and participants enter employment faster than their peer in small towns (reference), so the labour market is accessible regardless of language training, yet SFI still adds a meaningful advantage on top of that already favourable baseline.
What local conditions matter?
This study also asks what it is about a place that makes language training pay off. Three local conditions were tested: the unemployment rate, the share of foreign-born residents, and the share of high-skilled jobs. Only one clearly matters. Language training pays off most precisely in metropolitan labour markets where many jobs require strong communication skills; there, the gap between participants and non-participants widens sharply, and immigrants who complete SFI gain a clear advantage over those who do not. By contrast, local unemployment and the size of the foreign-born population leave the value of SFI broadly unchanged. In sparsely populated regions facing labour shortages, employers likely recruit more readily regardless of language training, meaning SFI still helps but is less decisive for obtaining a first job. In short, it is the local occupational structure — not labour-market competition or the concentration of migrants — that determines how much language training is worth, confirming that region and language training cannot be understood in isolation.
Implications for policy
The results point to the value of a more spatially oriented approach to SFI programmes, rather than treating the programme as uniform across the country.
- Adapt language training more closely to local labour-market conditions. A single national language programme should give local areas room to shape courses around the kind of work their region actually offers. The study shows this is what matters most: language training is worth more in places full of jobs that need strong communication skills, and less where it is not. One course built for the whole country will not fit every local job market.
- Target metropolitan regions. Returns are highest, and the penalty for non-participation most pronounced, in metropolitan areas and large cities, so raising participation and completion there is likely to bring the largest employment gains. Linking SFI completion to concrete labour-market pathways, such as combined language and vocational training, may be particularly useful in these skill-intensive urban markets.
- Think beyond first employment. The relatively strong employment outcomes in sparsely populated areas, even without SFI, should not be over-read: these areas receive a disproportionate share of lower-educated migrants, and the jobs available there are concentrated in lower-skill sectors. In such regions language training may matter less for getting a first job than for progressing into more stable, better-paid employment later — a question this study, which stops at first employment, cannot answer.
- A lesson beyond Sweden: How much language training helps is shaped by the kind of work available locally. It helps most where jobs demand strong communication skills, and less where workers are scarce and entry-level jobs are easy to get. Many countries run one national language programme as if it works the same everywhere. This study is a reason to check that assumption: where regions differ in the work on offer, the same programme may pay off very differently. That makes it useful evidence for other countries that mostly follow one-size-fits-all rules — worth testing how much language training helps across their own regions before assuming a single national programme fits them all.
References
Khaef, S. (2026). Enrollment and patterns of participation in Swedish for immigrants (SFI): timing, duration, and selectivity. Genus, 82, 11. https://doi.org/10.1186/s41118-026-00288-1
Khaef, S. (2025). Unraveling Integration: Educational Enrollment Pathways of Adult Refugees in Sweden. Population, 80(2), 153-177. https://doi.org/10.3917/popu.2502.0183.