The privatization of a segment of the Swedish Public Employment Service (Arbetsförmedlingen, or Af) operations, a key component of the January Agreement of 2019, stemmed from the Centre Party’s initiative and culminated years of criticism, particularly from business organizations, regarding perceived failures in labor market policies. This restructuring shifted the responsibility of guiding job seekers with moderate employment prospects into employment solely to external providers. Since 2020, these private providers have been delivering matching services under the largest labor market program, ”Rusta och matcha” (Equip and Match). The privatization was met with immediate criticism due to the complete handover of certain functions to these independent providers without prior experimentation to determine an optimal mix of public and private involvement. Furthermore, there was no research supporting the claim that private providers would outperform the public agency.
A recent study conducted by Af and the Institute for Evaluation of Labour Market and Education Policy (IFAU) examined the ”Rusta och matcha” program. The findings revealed that these privately delivered services were significantly more expensive than Af’s own initiatives, yet yielded no improved employment outcomes for participants. Given the intensity of the debate surrounding the 2019-20 privatization, the study has received surprisingly little attention. Comparing labor market programs is inherently complex due to potential differences in participant composition. Superior outcomes for one program over another could simply reflect pre-existing advantages among its participants. Randomized controlled trials, where participants are randomly assigned to different programs, offer the most robust method for mitigating this selection bias.
The Af-IFAU study employed such a randomized approach. Job seekers’ proximity to employment was initially assessed using Af’s statistical evaluation tool. Those deemed to have moderate employment prospects were then randomly assigned to either Af’s standard services or the external providers. Similarly, those with lower employment prospects were randomized between the external services and Af’s specialized programs for individuals facing significant employment challenges. This rigorous design, coupled with the large sample size, lends considerable credibility to the study’s findings.
Robust research often attracts criticism from stakeholders whose interests may be threatened by the findings, as well as from individuals with strong ideological convictions. This dynamic played out in the 1990s when the National Labour Market Board (Ams) attempted to discredit research that revealed weak outcomes for its programs. A similar pattern has emerged with the Af-IFAU study, which has been heavily criticized by representatives of employer organizations that include private matching providers. A major point of contention raised by critics is the study’s comparison of outcomes based on initial program assignment, regardless of actual participation, rather than comparing outcomes solely for those who actively engaged in the programs. However, this criticism overlooks the established scientific validity of the intention-to-treat analysis employed in the study. This method is preferred because actual participation is not random, even if the initial program assignment is. Provided there are substantial differences in actual participation rates, which is the case here, this method provides reliable insights into the relative effectiveness of private versus public programs.
Another critique centers on Af’s partial involvement in the study’s execution, suggesting a potential conflict of interest. While having competent internal researchers scrutinizing operations should be welcomed, it is theoretically possible that the analysis could be influenced by the institutional environment. The collaboration with IFAU, an agency with internationally renowned researchers, mitigates this concern. Af had a government mandate to ”consult with IFAU on how to create good conditions for evaluation.” Nevertheless, entrusting IFAU with sole responsibility for the study, albeit with necessary support from Af, would have further minimized the risk of both groupthink and suspicion. This alternative approach was not adopted, possibly because a collaborative project was deemed more effective. Another potential explanation is Af’s desire to showcase the capabilities of its internal analytical unit in conducting advanced research.
As a member of Af’s scientific advisory board, I was presented with an early version of the report. However, the study’s design predated the establishment of the board. Had the board been involved in the initial planning, I would have advocated for IFAU to lead the study independently, with appropriate support from Af. Despite this, the study maintains a high level of quality. Regarding policy implications, another radical overhaul based on a single study, however well-conducted, would be imprudent. Instead, Af should initiate smaller-scale pilot programs delivering matching services to individuals with moderate employment prospects, allowing for further comparisons with private providers. Finally, systematically phasing out the least effective private providers should be considered to gauge the potential for overall improvement in program outcomes.
