Does minus 150 billion mean the state loses money every year?

No. The amount is a present value over the cohort's remaining lifetime — around 40 years. The annual effect is about 0.06% of GDP, if KI's assumptions hold.

Is this the same as Ruist's ESO report?

No. Ruist (ESO 2018:3) measured annual public net cost. KI's lifetime net is a different quantity. The two should not be added or confused.

What happens if integration improves?

The model is sensitive to employment assumptions. If cohort employment reaches 75% at 15 years of residence, the negative net drops materially (illustrated in /intern/livstidsnetto-check).

Is the cohort estimate KI's or your own?

KI publishes the per-person net. Multiplying by 71,700 principals and 54,000 family members is an in-house calculation and labelled as such.

Does the model include Swedish-born children?

No, KI's model covers only the principal recipient. The second generation is treated separately and lacks robust lifetime data in the model.

← All analyses

Analysis · 3 min read

What did 2015 cost? Lifetime net eleven years later

163,000 people sought asylum in Sweden in 2015. Eleven years later we can recalculate the National Institute of Economic Research's lifetime net with actual outcomes — employment, benefits and tax revenue. Where does the number land?

Public financesLife-cycle calculationIn-house calculationSubstantial uncertainty2015 → lifetime

Guiding question:What does the National Institute of Economic Research's lifetime-net model imply for the 2015 refugee cohort — and which assumptions drive the result?

Short version in the chapter: Long-term welfare costs

Published 2026-07-15

Quick stats

Lifetime net per person (2015 refugee): −1.4 M SEK
Per-person range: −4.2 → +0.3 M SEK
2015 cohort estimate (in-house): 130–150 bn SEK
Annual GDP equivalent: ≈ 0.06%

The starting point: KI's model

The National Institute of Economic Research (KI) published an updated model in 2025 to calculate lifetime net per immigrant, based on SCB register data and age-specific probabilities for employment, benefit uptake and tax payments.

For refugee immigrants in 2015, KI calculates an average lifetime net of minus 1.4 million SEK per person (2024 prices). The range is wide: −4.2 million to +0.3 million depending on education level.

Calculation on the actual cohort

Taking the 71,700 people who received refugee residence permits in 2015 and applying KI's average: approximately 100 billion SEK in expected negative lifetime net (present value 2024).

The equivalent calculation for family members (54,000 people with family-tie permits 2015–2018): approximately 43 billion SEK.

Total expected net effect from the 2015 cohort (main recipients + family): 130–150 billion SEK lifetime, with an uncertainty range of 90–200 billion.

What the cohort has actually done

After 7 years (2022), 58% of the cohort is employed (SCB). This is in line with KI's assumptions for the age group.

Average disposable income 2022: 254,000 SEK/year, compared to 358,000 SEK/year for native-born in the same age group.

Benefit uptake (income support, housing benefit, establishment allowance): on average 42,300 SEK/year in 2022 per person in the cohort.

What does the number mean?

150 billion SEK corresponds to about 2.4% of Sweden's GDP in 2024. Spread over 40 years, the effect is about 0.06% of GDP per year.

This is not a 'cost' in budgetary terms — it is the difference between expected tax payments and public expenditure linked to the individuals over their remaining lifetime, discounted to present value.

An equivalent calculation for Swedish-born children (who also draw more public resources than they contribute during childhood) would not be negative because lifetime earnings offset that. The difference lies in the shorter active working lifespan and lower average income for the asylum cohort.

Method criticism

KI's model assumes unchanged policy and unchanged integration outcomes. If employment rates after 15 years approach native-born levels (as has happened with earlier cohorts such as Chile 1970s and Bosnia 1990s), the negative net drops substantially.

The model does not account for second-generation outcomes. If the second generation reaches native-born levels (as studies of the Bosnia cohort indicate), lifetime net at family level can become neutral or positive.

Counter-arguments and alternative readings

Joakim Ruist (Handels/GU) has consistently argued that lifetime net cost is sensitive to the employment assumption — and that 40–50% employment after 10 years yields a completely different calculation than 30%. The establishment pace of the 2015 cohort remains uncertain after 10 years (2025).

Sanandaji (ECEPR) and Aldén/Hammarstedt (LNU) land at higher lifetime cost than ESO 2018:3, while Ekberg (LNU 2011) landed lower for older cohorts. The difference lies in the discount rate and the second-generation effect assumed. Our calculation uses the ESO 2018:3 parameters as central scenario with ±15% uncertainty interval.

Method and uncertainty

Lifetime net is calculated as the present value of all future transfers (benefits, pensions, healthcare, schooling) minus all future tax revenues, discounted to the arrival year. Choice of discount rate (2% vs 3%) changes the outcome by 20–30%.

The second-generation outcome is the largest source of uncertainty. If the children approach Swedish-born employment rates (the Migration Agency's assumption), lifetime net per household falls sharply. If the gap persists (as SCB register data show for 2000s cohorts), it endures.

What this does NOT say

The calculation does not say the individuals 'cost' — it calculates an aggregate fiscal balance between public transfers and tax revenues for a cohort. Individual contributions to society (culture, family, social capital) are not captured.

Nor does it say anything about the counterfactual. If the cohort had not arrived in 2015, Sweden's population pyramid would look different; the effect on pensions and labor supply are separate calculations pointing in different directions.

What this shows

The analysis reports KI's (2025) lifetime-net model per person and scales it up to the cohort. The model is official; the cohort estimate (multiplication × individual count) is an in-house calculation and labelled as such.
Lifetime net is not a budget cost — it is the present value of the difference between expected tax revenue and public spending over the remaining lifetime, given KI's assumptions of unchanged policy and unchanged integration outcomes.

Definitions

Lifetime net: Present value of all tax payments minus all public spending linked to the individual over the remaining lifetime.
Present value: Future amount discounted to today's value at a stated rate (KI uses 3%).
Discount rate: Rate used to convert future flows to present value. A higher rate gives a lower present value.
Cohort: Group sharing a defining trait — here those granted residence permits in 2015.
Public net cost: The difference between consumed public services/transfers and taxes paid over a period.
Life-cycle cost: Cumulative public net cost across the entire life — negative in childhood and old age, positive during working years.
Scenario: Calculated 'if–then' case. Not a forecast but an illustration of an assumption.
Forecast: Statement of what is expected to happen given current trends and assumptions.
Sensitivity analysis: Test of how the result changes when an assumption is varied (e.g. employment or discount rate).
In-house calculation: Calculation performed by the editors on open data — assumptions disclosed, calculation reproducible.
Second generation: People born in Sweden with at least one foreign-born parent.

Method and limitations

KI's model assumes unchanged policy and unchanged integration outcomes. If cohort employment keeps rising (as with earlier Chile and Bosnia cohorts), the negative net drops materially.
The model does not include the second generation. Family lifetime net can turn neutral or positive if the second generation approaches native-born outcomes.
Amounts are in 2024 prices and discounted to present value. The discount rate (KI: 3%) matters a lot — a higher rate produces less negative numbers.
The cohort estimate (× 71,700 principal recipients + × 54,000 family) is a straight scaling and reported as in-house. The actual distribution across education levels gives a range — not a point value.
Ruist's ESO 2018:3 report (reference) measured annual public net cost, not lifetime net. Mixing the two produces double counting and misleading figures.
An internal calculator is available at /intern/livstidsnetto-check for anyone who wants to vary assumptions and see how the total changes.

Primary sources

FAQ

Does minus 150 billion mean the state loses money every year?: No. The amount is a present value over the cohort's remaining lifetime — around 40 years. The annual effect is about 0.06% of GDP, if KI's assumptions hold.
Is this the same as Ruist's ESO report?: No. Ruist (ESO 2018:3) measured annual public net cost. KI's lifetime net is a different quantity. The two should not be added or confused.
What happens if integration improves?: The model is sensitive to employment assumptions. If cohort employment reaches 75% at 15 years of residence, the negative net drops materially (illustrated in /intern/livstidsnetto-check).
Is the cohort estimate KI's or your own?: KI publishes the per-person net. Multiplying by 71,700 principals and 54,000 family members is an in-house calculation and labelled as such.
Does the model include Swedish-born children?: No, KI's model covers only the principal recipient. The second generation is treated separately and lacks robust lifetime data in the model.