The Swedish Public Employment Service (PES) offers Swedish language classes to newly arrived refugees who register as unemployed. This amounts to 15 hours of education per week over the course of a two-year span.
Dahlberg et al. (2023) analyzes a randomized control trial designed to test whether additional labor market integration services can impact employment and self-assessed societal integration in the long run. In this RCT, the control group received the baseline service described above, while the treatment group received the baseline service as well as intensive language training, work practice, and job search assistance.
In particular, refugees in the treatment group received three months of intensive exposure to Swedish through additional language classes (on top of the 15 hours) as well as general classes that were taught in Swedish. This amounted to about 40 hours a week of classroom time in Swedish. The second phase was a six-month internship where participants were assigned to a supervisor who did not speak their native language in order to practice speaking Swedish in a workplace environment. In the final phase, PES caseworkers helped refugees find jobs through searching for appropriate open positions, preparing CVs, and training for interviews.
The paper links various sources of administrative data to track participants’ attendance at Swedish classes, employment status, labor income, and social assistance payments over four years (2017-2021). Additionally, the paper has access to individual characteristics including age, gender, education, country of origin, date of entry to Sweden, and crucially- treatment status in the RCT.
The paper finds that the program successfully speeds up labor market integration: despite imperfect take-up, the treatment group has an employment rate of 45 percent directly after the program, while the control group does not reach this rate until almost four years later.
Pays for Itself
The paper accounts for both the direct costs of administering the program as well as indirect effects on the governments’ budget as a consequence of the program’s positive effect on employment. These consequences include decreased use of social assistance, as well as increased income tax and employer contributions to social security.
The paper estimates that the program has a direct cost of 27,030 SEK per participant which can be approximately broken down into the costs for increased teaching (12,000 SEK), paying intern supervisors (4,000 SEK), paying wage costs for the person running the work practice portion of the RCT (8,000 SEK), and paying wage costs for the intensified job search assistance (3,000 SEK).
There are no direct costs after the first year. However, the indirect costs include impacts during the program as well as within each of the four years following the program. The paper uses the instrumental variables approach to estimate the average treatment effect on the treated for changes in earnings and social assistance. From the estimated effect on earnings, the paper calculates the changes in taxes and employer contributions. In total, the paper finds that the program decreases social assistance by 16,636 SEK over this period, increases taxes by 15,757 SEK, and increases employer contributions by 15,187 SEK.
This leads to a total net cost per person of 27,030 – 16,636 – 15,757 = 15,187 = -20,550 SEK.
The authors also separate out the net costs to the specific municipality that ran this intervention and net costs for the rest of the public sector, and show that the net cost is negative for each of these government entities separately.
The study measures the willingness to pay for the program as the effect on earnings, taxes, and social assistance that individuals experience due to the intensive services. The IV approach estimates that individuals experience a net increase in earnings of 48,335 SEK over the program and the four years following. However, this also leads to paying 15,757 SEK more in taxes over this period and gaining 16,636 SEK less from social assistance.
Thus in total, individuals have a willingness-to-pay of 48,335-15,757-16,636 = 15,942 SEK for the net monetary benefits they experience over the four years following the program.
As the program has a negative net cost and beneficiaries have a positive willingness-to-pay of 15,942 SEK for intensive integration services, the program has an infinite MVPF.
Dahlberg, Matz, Johan Egebark, and Ulrika Vikman (2023). “Long-run integration of refugees: RCT evidence from a Swedish early intervention program.” Working Paper. https://www.ifau.se/globalassets/pdf/se/2023/wp-2023-23-long-run-integration-of-refugees_rct-evidence-from-a-swedish-early-intervention-program.pdf