Einiö and Nivala (2025) summarise several important lessons from the Finnish Recruitment Subsidy Experiment (FRSE), an intervention they implemented in collaboration with the Ministry of Economic Affairs and Employment of Finland in 2022–23.
The objective was to turn companies without employees into employers by offering a temporary wage subsidy for their first employee, and the tool was a large-scale, nationwide randomised controlled trial (RCT). Based on evidence from previous Finnish wage subsidy policies, the risk of imperfect take-up was high.
The main challenge faced by the research team in designing the experiment was, therefore, how to deal with the uncertainty surrounding take-up vis-à-vis offering a very costly treatment, in other words, striking the right balance between undersubscription with the implied risk of an underpowered statistical analysis and oversubscription with the implied risk of a budget overrun.
The research team took several steps to overcome these challenges. First, they agreed with the agencies involved to implement a relatively flexible policy offering subsidies of up to 50% for hiring a part-time or full-time worker with either a temporary or permanent contract for up to 12 months, up to a ceiling of €10,000. However, the window for take-up was limited to four months.
Second, they took steps to maximise awareness of the policy among treated companies by sending invitation letters to participate in the RCT directly to the owners.
Third, they took steps to minimise the costs of compliance, defined broadly as any paperwork involved in signing up for the programme, obtaining refunds for wage costs from the programme, and ensuring that refunds were paid out promptly and automatically.
To overcome any remaining uncertainty about take-up rates, the authors settled on a staged rollout, in which the initial pilot phase cautiously invited a small proportion of the participating population to generate direct evidence of expected take-up before implementing the more costly main arm of the experiment.
The pilot stage proved crucial, as it turned out that more than half of the target population could be invited to participate in the programme with practically no risk of oversubscription due to very low take-up rates, even with direct invitations. As a result, during the main experiment, the researchers focused solely on achieving optimal group sizes for the control and treatment groups to ensure statistical power.
Based on the intervention, the authors derived insights into running population-wide RCTs in collaboration with official bodies in a Nordic setting, which I summarise and discuss below.
The authors propose that the Nordic countries offer an ideal setting for population-wide interventions that support smooth, efficient implementation. First, researchers and authorities can rely on a rich digital infrastructure comprising platforms for direct digital communication with, e.g., company owners, employees, job seekers, taxpayers, and other targeted population groups.
These online platforms facilitate the direct delivery of invitations to subsidy programmes and targeted information interventions. In some instances, platforms even require participants to have access to and regularly read emails to remain eligible for benefits (Cairo & Mahlstedt, 2022).
Relying on existing infrastructure has multiple advantages, including lower costs and shorter implementation and running times, as well as more efficient treatment uptake.
In the case of FRSE, a dedicated platform was created for fast, smooth sign-up, while the prompt, automatic payout of wage refunds to compliers was facilitated through an existing platform used by Finnish employers to report all employee wage costs.
While implementation costs, including invitation costs, were fairly low, the policy intervention clearly had high costs per treated complying unit, since it relied on subsidies of up to €10,000 per complying company. By contrast, the marginal cost for each invited unit, whether complying or not in the case of information treatments, approaches zero when existing platforms are used.
Second, access to administrative registers across the Nordic countries supports perfect stratified randomisation based on pre-intervention observables such as industry codes and company size. This guarantees balanced treatment and control groups.
Another aspect that could have been explored in the FRSE is general equilibrium effects. While the hiring subsidy generated only approximately 207 jobs across Finland, related wage subsidies available to a larger target population could increase competition for qualified labour, resulting in higher overall wage costs. In addition, reallocating labour from more productive large companies to less productive sole proprietorships could lower overall economic growth.
The Nordic setting is ideal for exploring such spillover effects, thanks to the fine-grained geographical information in the registers, which allows for the delineation of local labour markets by location and industry. This implies that intervention designs can vary the degree of treatment across local labour markets and investigate the general equilibrium effects. One example of this is found in Altmann et al. (2022), which offered online job search advice to jobseekers and varied the proportion of those treated in local labour markets.
Access to administrative registers also facilitates the collection of evidence on outcomes at the evaluation stage of the intervention, such as wage costs, profits and revenues, and the number of employees over time. This is clearly an advantage over surveys, which are both costly and subject to attrition.
The authors note that they would have preferred to exclude hiring family members at the design stage. On the other hand, in a Nordic context with access to population registers, examining the proportion of hired individuals who are family members appears straightforward, as registers can usually be merged using personal identifiers. In fact, the registers facilitate a full analysis of employee characteristics, such as family ties, employment status at the time of hiring, wage income after hiring, compared to the year before. Similarly, the researchers can investigate whether employees hired with these subsidies are more likely to start their own businesses.
The FRSE intervention benefited from both policy support and institutional support. It was based on cooperation and collaboration with national agencies, and it had a special advantage: political preapproval for both the budget and design. It is impossible to exaggerate the importance of long-lasting alliances between research teams and public institutions for the successful implementation of field experiments. Unfortunately, many public and private interventions depend on the goodwill and inquisitive nature of individuals within organisations, rather than on the organisations per se. This remains a source of vulnerability for interventions, especially given the research time horizon relative to the typical length of time public employees remain in post. On this aspect, the Nordic countries are not so different from other settings.
One limitation of the FRSE, as recognised by the authors, is that no presurvey or postsurvey was distributed to investigate perceived barriers and enablers to hiring amongst founders of non-employer companies. The distribution costs of such surveys are typically relatively low in the Nordic countries due to digital communication platforms, and the survey evidence would have supported a richer investigation of low take-up mechanisms, as well as perceived benefits to the company, to innovation, and to employees from hiring subsidies.
Without such evidence, the authors are left to guess about the barriers to hiring in non-employer companies, such as uncertainty about worker productivity, lack of regulatory knowledge, or perhaps a preference for working alone. Evidence already exists for the positive effects of hiring via the offer of an infrastructure for posting jobs and screening candidates (Algan et al., 2020) and via the offer of information on employment regulations (Bertrand and Crépon, 2021).
The intervention also suffers from a classic policy-research dilemma. Policies proposed by politicians are often political compromises rather than economically efficient solutions. It is therefore commendable that official bodies in Finland approved testing the policy before its permanent adoption. Nonetheless, before designing the policy, it would have been advisable to clearly define the intended effects – on company founders, employees, and overall economic growth – and to assess them in light of existing evidence.
A central question is whether hiring subsidies are the best or cheapest policy to achieve the goal of turning non-employers into employers, and whether they may risk hindering economic growth. Essentially, size-based policies are likely to encourage the reallocation of labour from larger, more productive companies to smaller, less productive ones, thereby risking harm to overall economic activity. Also, limited liability companies are more likely to experience growth when hiring than non-employer sole proprietorships.
Regarding evaluation, the authors refer to their related paper, Einiö and Nivala (2026), which presents evidence that the policy increased hiring by 0.6 percentage points, a 20% increase relative to the baseline. This is equivalent to 207 additional employers during the subsidy period, and about 138 after the expiry of the subsidy, out of the 34,500 invited companies. As such, the policy was successful overall in achieving its goal of increasing hiring, although it indicates that the appetite for hiring among non-employers – even under a generous subsidy scheme – remains very low. Qualitatively, the effects on job creation are in line with prior evidence on general hiring subsidies, just as job-training programmes tend to increase employment to the benefit of jobseekers (Dahlberg et al., 2024; sectoral job training Katz et al., 2022).
The authors present a partial cost-benefit analysis showing that one euro of subsidy generated €2.3 worth of labour input. However, they also recognise that most employees were already hired before the treatment invitation and would presumably have continued in employment even without it.
Importantly, the authors’ related study finds no evidence that companies without employees who benefit from the subsidy experience increased growth in profits, revenues, or added value. This is in line with the findings of Cockx and Desiere (2023) in a Dutch context.
Even when hiring subsidies target highly skilled labour such as scientific researchers, Kaiser and Kuhn (2016) find that while the number of employees increases temporarily while companies take part in the programme, there are no statistically significant effects on company performance, including on added value, net income, return on assets, wages per employee and labour productivity. Neither during the program nor in the post-program period.
Given the insignificant impact on company growth, it is worth asking who actually benefits from the FRSE policy? In the absence of pre- and post-survey evidence to shed light on the perceived benefits to the compliers, this question remains unanswered.
The FRSE study illustrates the benefits of the Nordic setting for the implementation of policy interventions in the form of population-wide field experiments. Particularly, the use of existing digital infrastructures and platforms can reduce or even eliminate the costs of rolling out population-wide interventions. The platforms can facilitate direct communication with individuals in the target population, support smooth implementation, and reduce sign-up and compliance costs.
Access to administrative registers enables stratified randomisation based on pre-intervention observables to ensure balanced treatment and control groups. The registers also facilitate the collection of evidence on outcomes at the analysis stage, such as company wage costs, profits and revenues, and the number of employees over time.
It remains the gold standard for politicians to work with research teams on RCTs to test the impact and effects of their policies before population-wide rollout, and the FRSE is a prime example of such a collaboration. However, while RCTs are an efficient tool for examining the bite and effects of policies, they may also expose weaknesses in the policies in question.