Why does the analysis follow the 2015 cohort in particular?

2015 is the largest single refugee cohort and has been in the register long enough to measure 7–10 years of establishment. Younger cohorts are too new to judge durable outcomes.

Is 58% employment a reasonable measure?

It is SCB's register-based measure and is used in all official comparisons. The measure includes short work in November — wage income, hours and duration are separate metrics.

Was the establishment programme a failed reform?

The National Audit Office found 32% in unsubsidised work after 24 months — below target. The programme was replaced in 2022. Whether it 'failed' depends on the comparator and time horizon.

Can the gap to Denmark be explained by policy?

Partly. Delmi and OECD point to lower benefits and the IGU programme as likely contributors, but cohort composition and business cycle also matter. A clean causal estimate is not available.

What explains the large gender gap?

Education at arrival, family situation, country of origin and institutional factors (childcare, language expectations in social networks). No single factor explains the entire gap.

← All analyses

Analysis · 2 min read

SFI, jobs and the years that passed — a cohort analysis

Seven years after residence permit, the employment rate among 2015 refugees is still below 60%. What happened to the establishment mission?

Labour market and integrationCohort analysisOfficial statisticsMedium confidence2015–2022

Guiding question:How have establishment, SFI completion and employment evolved for the 2015 refugee cohort — and what explains the outcome?

Short version in the chapter: Education & integration — hub

Published 2026-07-15

Quick stats

Employed 7 years after (2015 cohort): 58%
Gap vs native-born: 24 p.p.
In unsubsidised work after 24 mo: 32%
SFI C-level within three years (2018→): 42%

The 2015 cohort — seven years later

SCB register data: Of the 71,700 who received refugee residence permits in 2015, 58% were employed by 2022 (age 20–64). Men: 66%. Women: 49%.

Comparison: Among native-born in the same age group, the employment rate was 82% in 2022. The gap is thus 24 percentage points after seven years.

The establishment mission

The establishment programme (2010–2022) covered 24 months with establishment benefit and mandatory Swedish tuition (SFI). The National Audit Office's 2023 review (RiR 2023:15) found only 32% of participants were in unsubsidised work after 24 months.

SFI results: 42% of those enrolled in SFI in 2018 had reached C-level (pass) within three years (National Agency for Education, 2022). Women with short education had the lowest completion rate: 24%.

What explains the weak job outcome?

Education at arrival: Among 2015–2017 refugees, 32% had less than 9 years of schooling, 25% upper-secondary, 25% post-secondary (SCB).

Regional concentration: 68% of the 2015 cohort settled or moved to municipalities with already high unemployment among foreign-born (SCB).

Women's participation: The share of women in employment grows slowly — from 32% (year 3) to 49% (year 7). Slower than the OECD average for comparable groups.

What does Denmark do differently?

Danish Ministry of Employment follow-up: Employment rate among 2015 refugees is 51% after seven years — lower than Sweden. The difference lies in policy choices: Denmark has lower benefit levels but faster labour market entry via the IGU programme.

Sweden's Etableringsjobb (started 2024) is inspired by the Danish IGU. It is too early to evaluate the effect.

Counter-arguments and alternative readings

Parts of the research literature (e.g. IFAU 2019:24) argue that SFI establishment is not primarily held back by language proficiency — but by employer recruitment behavior, by distance to the labor market in the receiving municipality, and by job-matching programs (Etableringsjobb) launched late in the cycle. On that reading, faster SFI throughput would yield marginal effect.

Another reading from Skolverket 2024 is that SFI levels A–D are not calibrated against labor-market language requirements. A passing C is insufficient for most occupations except warehousing, cleaning, and restaurants.

Method and uncertainty

SCB's register-based employment counts a person as 'employed' at ≥1 hour of work during the reference week. As a result, many of the 'jobs' in establishment statistics are part-time, hourly, or subsidized. Actual self-sufficiency (>60% of median income) is 20–30 percentage points lower in the same cohorts (ESO 2023:6).

Cohorts are followed for different lengths of time — the 2015 class has 10 years of data, the 2020 class only 5. Early differences may reflect follow-up time, not outcome.

What this does NOT say

The analysis does not say SFI is ineffective — the average hides large variation between municipalities and countries of origin. Some municipalities (e.g. Härryda, Vaxholm) reach 60% self-sufficiency within 5 years; others stay below 25%.

Nor does it say language training is the wrong intervention. It says language training isolated from labor-market linkage yields limited results — motivating combined models (SFI + internship).

What this shows

The analysis follows a defined cohort over time with SCB register data and complements it with evaluations from the National Audit Office, Delmi and IFAU. It describes co-variation between education at arrival, SFI outcomes and employment — not an established causal effect of any single measure.
National averages hide sharp differences by sex, origin and education level. A group having lower employment does not mean individuals fail to establish.

Definitions

Cohort: Group sharing a defining trait — here those granted residence permits in the same year.
Establishment programme: State programme 2010–2022 with establishment benefit and mandatory SFI for 24 months.
SFI: Swedish for immigrants — basic Swedish tuition with levels A–D.
Employment: SCB register-based measure: at least one hour of gainful work in a reference week in November.
Unsubsidised work: Employment without state wage subsidy (e.g. new-start jobs or introduction jobs).
Time to establishment: Years from residence permit to first employed year per SCB definition.
Longitudinal analysis: Studies that follow the same individuals or cohort over time rather than separate cross-sections.
Etableringsjobb: New Swedish job form from 2024 with lower entry wage and state subsidy — inspired by the Danish IGU.
IGU (Denmark): Integration foundational education — 2-year workplace-based training for newcomers in Denmark.
Education at arrival: SCB register category: <9 years, upper-secondary, post-secondary — based on self-report and validated grades.

Method and limitations

SCB register data covers registered residents and misses people who have left without deregistering. Employment rates may thus slightly understate actual establishment.
The employment measure counts short work in November — it says nothing about wage income, hours or duration.
SFI throughput is measured as the share reaching C-level within three years. Students who drop out and return later fall in different cohorts.
Cohort comparisons between Sweden and Denmark are affected by different refugee definitions, age composition and settlement patterns.
Evaluations from the National Audit Office and Delmi are quasi-experimental and control for education, age and origin — but not always motivation or health.
The page separates descriptive statistics (cohort outcomes) from causal evaluation (what a specific measure achieves). The employment gap versus native-born is descriptive, not an effect estimate.

Primary sources

FAQ

Why does the analysis follow the 2015 cohort in particular?: 2015 is the largest single refugee cohort and has been in the register long enough to measure 7–10 years of establishment. Younger cohorts are too new to judge durable outcomes.
Is 58% employment a reasonable measure?: It is SCB's register-based measure and is used in all official comparisons. The measure includes short work in November — wage income, hours and duration are separate metrics.
Was the establishment programme a failed reform?: The National Audit Office found 32% in unsubsidised work after 24 months — below target. The programme was replaced in 2022. Whether it 'failed' depends on the comparator and time horizon.
Can the gap to Denmark be explained by policy?: Partly. Delmi and OECD point to lower benefits and the IGU programme as likely contributors, but cohort composition and business cycle also matter. A clean causal estimate is not available.
What explains the large gender gap?: Education at arrival, family situation, country of origin and institutional factors (childcare, language expectations in social networks). No single factor explains the entire gap.

More analyses

Work and benefits