Impact Evaluation of Wage Subsidies and Training for the Unemployed in Slovenia

1. Assessment and recommendations

Abstract

Slovenia’s labour market has performed well in recent years, with unemployment reaching record lows and employment record highs. Nevertheless, a high share of long-term unemployed and low employment rates of individuals with low education remain persistent challenges. Active Labour Market Policies (ALMPs) are vital for addressing these challenges, but Slovenia spends only around two‑fifths of the OECD average on these policies. This evaluation examines two of Slovenia’s main ALMPs: training and wage subsidy programmes. Both programmes are effective in connecting individuals to jobs and improving most of the many labour market outcomes examined, though wage subsidies are usually relatively costly. While these ALMP programmes are generally well-targeted, additional efforts are needed to increase the uptake among low-educated workers and better integrate additional in-work support for disadvantaged groups. Building on the work in this report, Slovenia can strengthen counselling and better harness digital tools to provide more targeted support to jobseekers, especially vulnerable groups. Moreover, Slovenia can do more to link data and support regular impact evaluations of ALMPs.

1.1. Slovenia should continue to strengthen its system of active labour market policies

1.1.1. Slovenia’s labour market has performed well in recent years but active labour market policies remain crucial to help address long term unemployment and reduce labour market disparities

Over the last decade in the context of numerous global economic shocks, Slovenia’s labour market has demonstrated remarkable resilience and experienced a dramatic recovery. Following a sharp rise in unemployment in the aftermath of the 2008 global financial crisis, the internationally comparable survey-based unemployment rate peaked at 10% in 2013. Since then, Slovenia has experienced sustained improvements, with the unemployment rate declining by 6.5 percentage points through 2023 (with a slight uptick in unemployment in late 2024). Despite a temporary setback during the COVID‑19 pandemic, the labour market quickly rebounded, with unemployment returning to pre‑pandemic levels by late 2021 and continuing its decline.

Employment rates have also been robust. In 2023, 77.5% of Slovenians aged 20‑64 were employed, surpassing the OECD and EU averages of 74.4% and 75.3%, respectively. Similarly, Slovenia’s labour force participation rate of 80.2% was higher than the OECD and EU averages of 78.1% and 80.0%, respectively, highlighting a strong labour market engagement. Nevertheless, considering the employment and participation rates in OECD and EU’s top-performing countries, there is room to further improve the labour market situation in Slovenia’s.

While unemployment levels are low, challenges persist. Tight labour market conditions have led to rising job vacancy rates, with employers struggling to fill open positions. In 2022, Slovenia’s job vacancy rate peaked at 3.3%, slightly above the EU average of 3.0%, although it has moderated since then. At the same time, long-term unemployment remains a significant issue, with 38% of unemployed persons in Slovenia in 2023 being long-term unemployed. This is a considerable decrease in the share from several years ago but remains far above the OECD average of 23%. This points to the presence of a group of jobseekers with significant barriers to re‑entering employment, even as overall labour market conditions improve. Strong active labour market policies tailored to their needs can play a key role in addressing this challenge.

Certain population groups face greater barriers to employment in Slovenia. For older workers (aged 55‑64), the employment rate remains low at 54%, compared to the OECD average of 64%. People with low educational attainment also face substantial difficulties in accessing employment. The employment rate for those with only primary education is 55%, significantly below the OECD average. In contrast, Slovenians with tertiary education have exceptionally high employment rates (93%). These disparities underline the need for targeted policies to support lower-educated and older individuals, including through the increased provision of training. As in many other countries, such individuals are under-represented in training for the unemployed in Slovenia.

1.1.2. Slovenia’s institutional and regulatory setup for ALMP provision works well in practice

The key institutions managing Slovenia’s active labour market policies (ALMPs) are the Ministry of Labour, Family, Social Affairs, and Equal Opportunities (MoLFSA) and the Employment Service of Slovenia (ESS). While MoLFSA oversees policy design, the ESS implements measures and administers job-brokerage, career counselling, unemployment benefits administration, and work permit issuance. In the past, MoLFSA has also implemented measures such as contracted-out employment services through a programme of lifelong career guidance.

Slovenia’s ALMP framework is governed by a detailed regulatory structure centred on the Labour Market Regulation Act (LMRA). This law establishes the overarching principles for ALMPs, while further regulations define their implementation. Specifically, the government adopts ALMP Guidelines to shape its strategy, and MoLFSA prepares an implementation plan and a catalogue of ALMP measures. The plan is developed based on the ALMP Guidelines for the budgetary period and is adopted by the Labour Minister.

Slovenia’s ALMP system has demonstrated adaptability in the past, particularly during the swift policy response to the COVID‑19 pandemic in 2020. This adaptability stems from the MoLFSA’s flexibility within the legal framework, including the ability to propose amendments to the LMRA for significant ALMP changes, which can be implemented rapidly when there is political consensus. MoLFSA can also develop new ALMP measures or expand target groups in the ALMP catalogue to address economic shocks, though this option has not yet been used in practice.

1.1.3. Slovenia spends relatively little on active labour market policies, with spending heavily focused on public sector direct job creation programmes

Slovenia’s spending on ALMPs remains well below the OECD average, limiting the resources available to support jobseekers. In 2022, Slovenia spent only 0.18% of GDP on ALMPs, which is about two‑fifths the OECD average of 0.43%. This represents a downward trend from previous years, as spending was 0.24% in 2015 and 0.20% in 2019. While declining unemployment rates partly explain this trend, Slovenia’s ALMP expenditures per unemployed person also rank among the lowest in the OECD. Participation in training, particularly institutional training, is low, with many adults facing barriers to accessing further education. The result is that a sizable proportion of the working age population is poorly equipped to adapt to changes required in the modern labour market. For example, 48% of adults are lacking problem-solving skills in technology-rich environments in Slovenia compared to 30% in the OECD. Greater investment in adult education and training could better address emerging skills needs and foster high-quality jobs.

A significant portion of Slovenia’s ALMP spending is directed towards direct job creation programmes (17% in 2022, compared to the OECD average of 8%). Such programmes face concerns relating to their cost-effectiveness and longer-term labour market outcomes. Slovenia spends significantly more per participant on direct job creation than on training programmes or employment incentives. Furthermore, costs per participant for direct job creation are among the highest in the OECD. This underscores the need to balance their potential benefits with their high costs. A past evaluation of direct job creation programmes has suggested that they yield limited success in improving participants’ employment prospects, but an additional ongoing evaluation being conducted by the OECD will establish whether this is still true for newer programmes and whether they have positive effects on broader social outcomes.

1.1.4. The ESS needs to increase its capacity to counsel jobseekers, especially those with significant barriers to becoming employed

Despite improvement in recent years, counsellor caseloads in Slovenia remain high. ESS counsellors were responsible for an average of 256 clients in April 2024, representing a significant reduction from their peak in 2013 when the caseload exceeded 400 clients per counsellor. However, caseloads in Slovenia are still much higher than in many European countries. For instance, on average each counsellor in Denmark, Germany, and Flanders typically manages 100 jobseekers furthest from the labour market, with similar or slightly higher figures for jobseekers closer to employment. Evidence from countries such as Germany and Austria demonstrates the potential effectiveness of lower caseloads – often below 100 per counsellor – to provide intensive support and achieve better employment outcomes for vulnerable groups.

High caseloads in Slovenia make it challenging to provide in-depth support to jobseekers, particularly those classified as needing intensive support. While the frequency of meetings tends to be higher in the early phase of unemployment spells, it is in fact relatively less frequent for the groups of jobseekers who would require the most intensive support. On average, such jobseekers had only one meeting every 3.9 months in 2023, compared to once every 3.6 months for those classified as slightly more employable and once every 2.2 months for those classified as “Directly employable”. This imbalance in counselling intensity reflects broader capacity challenges within the ESS.

These challenges could be addressed by increasing the number of counsellors and by freeing up counsellors’ time for more in-depth work with specific client groups. Indeed, the ESS is currently undertaking significant investments into digital platforms and tools to increase workflow efficiency. One key initiative is a competencies-based job matching tool employing AI which considers a broader set of factors, such as professional and personal skills, to facilitate more efficient and targeted job placements. Following successful regional pilots, the new matching system is set to be rolled out nationwide in 2025. Furthermore, the ESS is developing a new statistical profiling tool to assess jobseeker skills and needs more consistently and comprehensively. This tool will replace the current, discretion-based profiling system, aiming for a more data-driven approach to categorising jobseekers.

The ESS should leverage these investments into its digital infrastructure to improve its service delivery and support jobseeker integration into the labour market. Specifically, the ESS should consider conducting statistical profiling of jobseekers prior to the first meeting based on registration and administrative data. Readily-employable, digitally literate jobseekers could initially opt into a service stream with enhanced self-service digital tools. These could include the enhanced self-service job search platform, additional online training options as well as automated individual action plans. If individuals in the self-service stream remained unemployed for longer than anticipated – e.g. four months – they could be mandated to attend in-person counselling sessions. Such an approach would free up resources for counsellors, allowing early meetings and intensive services to be targeted better at those who require in-depth support and are at risk of becoming long-term unemployed. It would also allow services to be tailored more closely to the individual needs and preferences of jobseekers.

Against this context, this report uses linked administrative data and presents the results of a counterfactual impact evaluation of two main types of ALMP programmes currently implemented in Slovenia: training programmes and wage subsidies.

1.2. Training programmes for unemployed people are found to be effective and have the potential to promote upskilling for a larger population

1.2.1. Training programmes are continually adjusted to labour market needs and are effective at connecting people with jobs

The Slovenian ALMP system is characterised by its institutional flexibility, which enables both MoLFSA and ESS to revise target groups and programme parameters – such as eligibility criteria and subsidy amounts – effectively. This is reflected in the continually updated ALMP catalogue to reflect changing labour market needs. Between 2017 and 2022, for instance, the catalogue underwent 47 iterations, averaging 8 updates per year. While some of these revisions were minor, they demonstrate the system’s capacity for frequent and targeted refinements. Moving forward, the system’s ability to provide effective reskilling and upskilling opportunities tailored to labour market needs will depend on preserving this flexibility.

One of the main types of ALMPs employed in Slovenia are training programmes. Among these are National Vocational Qualification (NVQ) programmes, which promote the validation of prior learning, allowing jobseekers to certify skills and experience gained outside formal education and obtain nationally recognised certificates. Additionally, Institutional Training and Non-Formal Education and Training programmes provide targeted reskilling and upskilling opportunities across various fields, including business, IT, construction, and production technologies. These programmes are primarily targeted to individuals further from the labour market, such as older jobseekers, those with low levels of education, and the long-term unemployed. Additionally, specific subprogrammes are tailored to address the needs of other groups, including young jobseekers.

The counterfactual impact evaluation presented in this report shows that these training programmes are effective in supporting jobseekers into employment. Twelve months after the start of the training, training participants are roughly 10 percentage points more likely to be employed than comparable non-participants, an effect ranking in the highest quartile of effect sizes observed in international meta-studies. This positive employment effect persists over the long term, with NVQ preparation programmes and local Non-Formal Training and Education initiatives proving particularly effective.

Although training programmes are effective for jobseekers across various individual attributes, including gender, age, education, unemployment duration, and nationality, the overall positive employment effect masks a certain degree of heterogeneity. Specifically, training programmes appear to be particularly effective in enhancing the employability of men and jobseekers with secondary-level education. Many of the training courses seem geared primarily towards such individuals – indeed, in many cases, having at least secondary education is a precondition for attending the training courses. Encouragingly, the training appears to be helping jobseekers find employment in growing and high-value added sectors such as professional services.

1.2.2. Training programmes also improve earnings and reduce inactivity

The findings of this report show that the training programmes improve a wide range of labour market outcomes. In addition to higher employment rates, 12 months after the start of the programme, training participants exhibit a 4.7 percentage point lower likelihood of being unemployed and a 5.3 percentage point lower probability of becoming inactive. When compared to other internationals studies, these estimates are relatively large and rank in the top quarter of the distribution.

Moving beyond labour market status, the report finds that training programmes are effective in increasing cumulative days worked and cumulative earnings, while reducing occupational mobility. Thirty-six months after the programme’s start, training participants have accumulated an additional EUR 4 500 in earnings and worked 100 more days than non-participants. However, training programmes also appear to reduce upward occupational mobility. This finding suggests that training participants who found employment after completing the programme entered occupations that offered slightly lower wages compared to non-participants. Although the magnitude of this statistically significant effect is relatively small (2.5%), it does indicate that while training participants are more likely to secure employment than non-participants, they often enter lower-paying occupations than those of non-participants. Examples of such occupations include drivers and security guards, occupations which are greatly in demand, require skills or certifications that can be acquired through the short-term training programmes offered through the ESS, but that do not generally pay higher wages.

Nevertheless, the negative effect on upward occupational mobility is outweighed by the positive impacts of training on other labour market outcomes. Training programmes help jobseekers secure stable employment, characterised by more days worked and higher cumulative earnings, even though the average wages in the occupations they enter tend to be lower than those of non-participants.

1.2.3. Expand training programmes to reach more people and consider options to make training more attractive

The analysis of take‑up rates shows disparities in how different groups of jobseekers access training opportunities. For instance, young jobseekers (both women and men) are overrepresented among training participants compared to their proportion in the overall unemployed population. In contrast, jobseekers over 50, those unemployed for over 23 months, and individuals with only primary education are underrepresented (in a small number of cases for the latter, due to minimum education requirements for training participation). This suggests that groups generally more distant from the labour market – such as older jobseekers, the long-term unemployed, and low-educated jobseekers – are less likely to engage in training than other groups of jobseekers. However, when they do participate in training, some of these groups tend to benefit more from it. Specifically, training appears to be most effective in improving the employment probabilities of those unemployed for over 12 months and jobseekers aged above 50.

While the current offer of training programmes is comprehensive and covers various fields, there is potential to expand it to better reach vulnerable jobseekers and those more distant from the labour market. One promising approach is to increase the use of micro-credentials. These credentials (e.g. a training programme in CNC welding) differ from traditional education and training programmes by being more targeted, smaller in scope, and highly flexible. They can serve as standalone qualifications or be integrated into broader learning pathways. Micro-credentials offer the ability to quickly equip individuals with the skills needed to adapt to the dynamic and ever-changing demands of labour markets.

In addition to micro-credentials, training programmes that contribute to building basic skills may be useful. While the ESS is no longer responsible for offering basic skills courses for adults, close collaboration with institutions tasked with delivering these programmes is crucial. Such co‑operation is essential to ensure the creation of effective upskilling and reskilling opportunities for low-educated jobseekers.

1.2.4. Implement systemic changes to ensure the training is consistently of high quality

Although all training programmes are found to be effective in supporting jobseekers into the labour market, their effectiveness varies across programmes. This variation may partly result from inconsistent training quality among providers, reflecting the lack of comprehensive quality criteria and quality assurance mechanisms in the current system of training.

To ensure consistently high-quality training, the system could be enhanced by introducing additional quality criteria for training providers and strengthening the monitoring of their performance. For instance, accreditation requirements, which are currently in place for some training programmes, could be expanded and made more rigorous for all providers seeking inclusion in the training registry. Additionally, a system of key performance indicators (KPIs) could be established to track providers’ performance. These KPIs – such as client satisfaction and sustained employment outcomes – could be integrated into the bidding process, enabling the selection of providers based on both quality and cost considerations while the administrative burden on training providers.

The ESS could also explore introducing a system of client choice for certain training programmes. Under this approach, the ESS would still maintain a registry of approved providers, but jobseekers would have the freedom to select a provider from the list. This choice could be guided by factors such as provider performance, feedback from other clients, and the advice of ESS counsellors. Evidence from similar systems in employment services suggests that introducing client choice can have positive effects, empowering jobseekers and encouraging providers to improve their services.

1.3. The employment incentives (wage subsidy) programme for the unemployed is also highly effective

1.3.1. Employment incentives appear effective in placing individuals in employment

Along with training, the employment incentives (wage subsidy) programme is one of the main ALMPs used to connect unemployed people with jobs in Slovenia. By providing employers with a financial incentive to hire certain categories of jobseekers, employment incentives can facilitate jobseekers’ integration into the labour market. The programme provides a wage subsidy for one year to participating private sector employers, with the amount of the subsidy varying based on jobseekers’ individual characteristics. During the period examined, the subsidy would cover roughly 50‑70% of the gross earnings of an individual earning the minimum wage. Jobseekers with greater barriers to employment – such as those with lower education levels, older workers, and individuals experiencing long-term unemployment – are eligible for higher subsidies.

The results of the counterfactual impact evaluation (CIE) suggest that participation in the employment incentives programme has large and statistically significant effects on the probability of individuals being in employment. After 36 months from the start of the programme, employment incentive participants were 26 percentage points more likely to be in employment than their matched controls. These effects capture unsubsidised employment, given that the subsidy period lasts for 12 months. The positive effects of employment incentives on employment probability are also reflected in the employment duration results: over the two‑year period following the end of the employment incentive (wage subsidy), participants were employed for 195 days more than comparable non-participants who form the control group. Comparing the outcomes of participants with the comparable non-participants shows that roughly half of the net increase in employment among employment incentive participants can be accounted for by individuals who would have otherwise been registered as unemployed, while half would have been considered inactive (not in formal employment or registered as unemployed).

When compared with results from similar types of studies in other countries, Slovenia’s results are notably positive. The estimated boost in employment rates attributable to participation in the wage subsidies in the short-term, of 34 percentage points, exceeds the effects reported in most of the comparison studies, which typically range from 0 to 10 percentage points (this difference in short-term effects may be partly attributable to methodological differences). At the same time, the long-term effect of 23 percentage points (after four years) is also above or in line with the median long-term effects from similar international studies. It is important to note that, while the point estimates in the comparison studies are generally positive, they are not always statistically significant. In fact, a slight majority (58%) of the studies in the international meta‑analysis do not report effects that are both positive and statistically significant effect over the long term. This makes the Slovene results more remarkable.

The employment effects found in this report are also roughly in line with the results of an older evaluation of Slovenia’s employment incentives programme, which looked at entrants into the Employ.me programme during the 2009‑12 period (when employment in Slovenia was contracting in the aftermath of the global financial crisis). In that evaluation, the estimated employment effects were slightly higher: the effects on employment at 36 months were approximately 30 percentage points, compared to 26 percentage points in this report’s evaluation. At that time, the employment subsidy amount was relatively more generous, amounting to roughly 50% of programme participants’ gross annual wages (compared to 38% in the current evaluation) and lasting up to two years instead of one.

1.3.2. Employment incentives have positive effects on other outcomes as well, such as retaining workers in Slovenia

The positive employment effects of employment incentives in Slovenia are especially remarkable in light of the additional finding that they did not adversely affect individuals’ occupational mobility in general. The average occupational index for jobseekers entering subsidised employment was not statistically significantly different from zero. This occupational index, calculated for each detailed occupational code using data on the wages and employment of individuals in Slovenia, measures the extent to which individuals move into generally better or worse‑paid occupations compared to the control group of similar individuals. While these aggregate results do mask substantial heterogeneity across different groups of individuals, they nonetheless indicate that jobseekers are not, overall, entering lower-paying occupations in exchange for employment.

The increased days worked by programme participants contributed to substantial additional earnings. Three years after beginning the programme, participants’ earnings amounted to EUR 21 086 more than their counterparts in the control group during the same period. Excluding the first year, during which employers were paid the subsidy – i.e. in the two years after finishing the employment incentive programme – participants earned a total of EUR 9 356 more than their counterparts in the control group during the same period. In relative terms, this means that participants had almost double the earnings of their counterparts in the control group during the two‑year period following the end of the subsidy. While a detailed cost-benefit analysis is outside the scope of the report, these figures suggest that the employment subsidy cost less than one‑third of the additional earnings attributable to the programme in the first three years after participation, but other, less costly ALMP measures could be effective for certain target groups.

This evaluation also examines the impact of the employment incentives programme on the likelihood of migration abroad based on registry data. The findings show that participation in employment incentives reduces migration probability. While the absolute effect is very modest in magnitude when compared to the employment effect, resulting in 0.5 percentage points fewer participants migrating abroad 36 months after entering the scheme, the relative effect is significant. Specifically, at 36 months, 0.80% of individuals in the control group have emigrated, compared to just 0.32% in the treatment group. This means that 2.6 times as many individuals in the control group emigrated relative to the treatment group. This effect was exclusively driven by Slovenes (or other EU citizens) who remained in Slovenia instead of moving abroad following participation in the programme.

1.3.3. While employment incentives are generally well-targeted, uptake amongst low-educated workers should be encouraged more

The employment subsidies are found to provide an employment boost to all individuals across the personal attributes examined (gender, age, level of education, nationality, and unemployment duration upon entry). However, there are differences in the magnitude of this boost. Notably, significant differences exist within age groups: older jobseekers see smaller employment boosts than their younger counterparts, with the disparity particularly pronounced for women over 50. Furthermore, non-EU nationals get a considerably larger boost from employment subsidies than Slovene and other EU-nationals, with a 7-percentage point difference in the effect. One explanation for the latter result is that employment incentives are particularly effective at helping overcome the uncertainties employers face when hiring foreign workers, particularly regarding the quality of their education, language or other skills, and prior experience.

The employment effects are notably consistent across individuals with different levels of education and unemployment duration. In all examined categories, the estimated employment increases range from 22 to 24 percentage points. This limited variation suggests that the subsidies are effectively targeted for these groups. Employment incentives aim to encourage employers to hire individuals they might otherwise hesitate to employ, due to barriers such as insufficient experience or other significant obstacles. These barriers can affect individuals regardless of education level or unemployment duration, despite the fact that, on average, those with higher education or shorter unemployment spells are more likely to re‑enter employment. The lack of variation in effects by education aligns with findings from impact evaluations conducted by the OECD of similar programmes in Lithuania and Greece.

In terms of take‑up rates, certain categories of individuals are disproportionally likely to participate in the employment incentives programme (relative to their share of the unemployed). This includes older workers, those with longer unemployment spells, those with tertiary education, and Slovene nationals. Having higher take‑up rates amongst older workers and the long-term unemployed makes sense: employment incentives are intended to boost the job prospects of otherwise harder-to‑employ individuals, and these two groups of individuals are disproportionally likely to have difficulties finding a job.

The high uptake of employment incentives amongst individuals with tertiary education is slightly less justified, as these are individuals who generally have better employment prospects. On the other hand, in the ALMPs studied, the programme parameters include highly-educated individuals only if they are long-term unemployed (and aged at least 30), thus making them otherwise statistically less likely to have very good job prospects; furthermore, the highly-educated experience the greatest boost to their employment from participation in employment incentives. Nevertheless, the disparities in terms of the effects on additional earnings by education status are smaller than in terms of employment, providing a stronger argument for increasing the share of lower-educated individuals.

One option for improving uptake amongst lower-educated individuals would be to modify the programmes’ parameters to make them more attractive to employers to hire them. While the existing subsidy amounts already vary based on several factors related to their employment prospects (unemployment duration, age, education, and social benefit receipt), the parameters could be further refined to encourage participation of jobseekers with low-education. Establishing the effectiveness of such changes would ideally be done in the context of a rigorous evaluation which could more broadly establish the importance and effectiveness of various parameters.

1.3.4. Wage subsidy programmes could be augmented with in-work support to help disadvantaged groups

One of the ESS’s key challenges is supporting individuals with significant barriers to employment. Subsidised employment programmes with ongoing, in-work counselling support while in employment can serve as a crucial step in an individual’s reintegration into the labour market. Slovenia currently offers two such small-scale programmes: a private sector wage subsidy programme aimed at integrating youth into the labour market, and Learning Workshops+, a specific implementation of direct job creation. The first programme is intended to foster intergenerational learning by having older workers employed at a firm counsel newly-employed, subsidised hires. The latter programmes incorporate professional mentoring but with employers in the social economy (although subsequent subsidised employment may be in the private sector). In the case of the first programme, although these programmes provide valuable pathways for disadvantaged jobseekers, reliance on employer-provided support may limit the quality of the support provided and the scope of employers willing to participate. In the case of the second programme, the programme has many promising features but, to the extent that participants are then becoming employed in the social economy, experience with similar types of programmes suggests that in contrast with private sector subsidies, subsidised employment in the social economy does not lead to stable, long-term jobs.

To address the limitations in these existing programmes, Slovenia could complement the private sector wage subsidy programmes with external professional in-work support. The professional in-work support could address common challenges experienced by individuals with multiple employment barriers by providing help with onboarding, mentoring for personal and professional issues, and serving as a mediator between workers and employers. Many countries employ programmes where disadvantaged individuals receive such in-work counselling support from professional counsellors (though they are not always combined with wage subsidies). One approach in such a programme in Slovenia would be to have ESS staff deliver the support, by either leveraging the expertise of ESS specialist counsellors experienced with vulnerable groups or by expanding the role of counsellors who collaborate with employers. Another approach would be to have external providers deliver the support. This could include professionals experienced in counselling, training providers, or social economy actors. Payments to these providers could be tied to clients’ sustained employment, creating incentives for effective support. This second approach could draw inspiration from a pilot programme planned in Slovenia in 2020 that was ultimately cancelled due to the onset of the COVID‑19 crisis.

1.4. Slovenia should do more to link data and support regular impact evaluations of ALMPs

1.4.1. Impact evaluations of ALMPs identify what works and for whom

Conducting regular CIEs provides valuable insights into the effectiveness and cost-efficiency of ALMPs. By comparing outcomes between programme participants and a carefully selected control group, CIEs can isolate the true impact of interventions such as participation in training or receiving more intensive counselling. Specifically, they can precisely address the question of whether participants benefited from an intervention taking into account what their outcomes would have been had they not participated. CIEs thus allow policymakers to gauge programme success, quantify impact, and conduct cost-benefit analyses.

Another important feature of CIEs is that they can identify which groups benefit most from specific ALMPs, helping to tailor policies. As the evaluation in this report has shown, the effects of an intervention can vary considerably across jobseekers depending on characteristics such as their age, education, gender and unemployment duration. Understanding these differences can help ensure that support is targeted efficiently, taking into account which individuals benefit the most from which types of support. Moreover, conducting evaluations regularly helps policymakers adapt to changing circumstances, refine programmes, and optimise interventions, creating a feedback loop for continuous improvement.

CIEs can employ various approaches. Randomised Controlled Trials (RCTs) are often considered the “gold standard”, as properly conducted random assignment eliminates other sources of variation, ensuring that differences in outcomes between treatment and control can be attributed purely to their participation in an intervention. However, in the context of ALMPs, RCTs are often impractical due to implementation difficulties or inappropriate due to limits to the types of questions they can realistically address. In such cases, quasi‑experimental methods offer effective alternatives. One approach, used in this evaluation – propensity score matching – pairs participants with non-participants with similar characteristics. Another promising alternative approach -not used in this report- for evaluating ALMPs is a regression discontinuity design. This approach exploits eligibility cut-offs to compare individuals just above and below thresholds, for example one based on a statistical profiling score estimating an individual’s probability of becoming employed within 12 months. Such non-experimental methods are particularly suited to applications when detailed and accurate administrative data are available, as is the case in Slovenia.

1.4.2. Slovenia’s linked administrative data offer excellent opportunities for conducting regular, comprehensive evaluations of ALMPs

As this evaluation demonstrates, Slovenia’s high-quality administrative databases offer significant potential for conducting detailed CIEs of ALMPs. The availability of unique individual identifiers and rich, high-quality data minimises technical challenges related to linking and analysing data. Instead, legal and procedural barriers remain the main obstacles to conducting regular and comprehensive evaluations. While the current system of accessing data through the Statistical Office functions well for occasional analyses, it is not suited for systematic evaluations using databases that may need to be regularly updated. For example, the data used in this evaluation were compiled specifically for the purpose of this evaluation, meaning it cannot be readily updated and in principle cannot be used for other purposes, such as additional evaluations of related questions. This means that, under the existing system, the considerable upfront costs associated with accessing and using the data needed for the analysis – establishing the legal basis and contracts for data access, acquiring the raw data from various sources, and compiling it into analysis-ready databases – cannot be avoided for future evaluations.

To regularly conduct comprehensive CIEs, the ESS and/or MoLFSA would need access to several key databases spanning multiple years before and after individuals are unemployed. Essential data sources include the unemployment registry, which provides information on registered unemployed individuals necessary for defining treatment and control groups, and ALMP participation data, which details programme enrolment and dates to identify intervention effects. Data on employment contracts from the health insurance institute would enable analysis of employment outcomes such as job stability and contract type. To enhance these evaluations, additional data on monthly earnings, unemployment benefits, and social benefits – a database maintained by MoLFSA – would allow for an assessment of job quality and facilitate cost-benefit analyses by linking outcomes to public expenditures. Supplementary sources, such as income tax records and employer data, could provide insights into financial outcomes and employer engagement with ALMPs, while detailed ESS records could help analyse the effectiveness of different reintegration approaches for specific jobseeker subgroups. Together, linking these datasets offers the potential for generating nuanced, credible evidence that can facilitate monitoring and provide a sound basis for policy reforms.

1.4.3. Slovenia should explore several options for moving towards more evidence‑informed policy making

Slovenia has made significant progress in advancing an evidence‑informed ALMP monitoring. MoLFSA and the ESS have established an extensive monitoring framework focusing on the programme activities (e.g. participation), outputs (e.g. completion) and gross outcomes. Examples of relevant analyses include the regular internal reports and presentations produced by the ESS analytics office as well the Annual Report on Labour Market Policy measures published by MoLFSA. Such analyses are valuable for monitoring purposes. They support short-term decision-making by providing early indications of how effective a programme may be, helping identify anomalies across programmes or target groups to make sensible modifications. The analyses are also integral to the ESS accountability framework, helping to identify high-performing regional offices and investigate under-performing ones.

While these monitoring efforts are helpful, they cannot replace the rigour and depth of CIEs, which are essential for determining the causal effects of policies. ALMPs in Slovenia are periodically subject to such CIEs, such as the one presented in this report. Nevertheless, Slovenia could take several steps to strengthen its commitment to evidence‑informed policy making and ensuring that such evaluations are conducted regularly and systematically. First, the government could mandate in legislation the regular and systematic conduct of rigorous CIEs, with provisions for their publication to ensure transparency and accountability. For example, Canada’s legislative framework requires such evaluations, fostering a culture where the expectation of subsequent evaluations drives improvements in the design and implementation of ALMPs. Second, Slovenia could increase the resources available to the research units within MoLFSA or the analytical unit within the ESS, enabling them to have sufficient staff to conduct such evaluations (and hiring external experts for peer review and quality control purposes). Conducting such research in-house has several benefits, including facilitating more regular analyses in the longer-term and building internal institutional support for evaluations. An alternative approach would be to have longer-term research contracts with outside contractors. However, even with the latter approach, MoLFSA should dedicate sufficient resources to ensuring an institutional capacity for competently assessing the quality of the evaluation provided. In either case, dedicated and stable funding should be secured. Third, consulting with evaluation specialists during the programme design phase could help embed robust evaluation methodologies, such as randomisation or regression discontinuity design, directly into a pilot programme or policy reform. For example, this could involve simple rules for which jobseekers are referred to a specific intervention, such as a mandatory ALMP at a pre‑determined point in their unemployment spell, facilitating subsequent evaluations of its effects. In addition to facilitate subsequent evaluations, this could help build awareness among stakeholders such as PES counsellors of the existence of the evidence‑based approach, facilitating their buy-in for eventual policy changes. Fourth, Slovenia should invest in careful communication strategies for presenting its evaluation results. Clearly presenting findings can foster a culture of accountability and make evidence more accessible and actionable for senior decision-makers.

More broadly, at the governmental level, establishing a dedicated cross-ministry agency tasked with promoting and facilitating evidence‑based policy making would represent an important step to more evidence‑informed policy making. Similar to how Data Protection Officers champion data protection and Fiscal Councils oversee public finance, such an agency could focus on embedding evaluation practices and evidence use across government programmes. This would follow the successful examples of such agencies in Australia, the United Kingdom and Spain.

Key policy recommendations

Increase the reach and effectiveness of ALMPs by ensuring stable and sufficient funding and evidence‑informed policy design

• Secure continuous and sufficient funding for ALMPs.

• Consider introducing mandatory ALMP participation for certain groups of jobseekers at predetermined unemployment duration milestones.

• Increase funding for training especially for vulnerable groups, as this evaluation has shown they are highly effective measures.

• Adjust and expand upskilling programmes for jobseekers with low levels of education, where the existing programmes have been shown to not be as effective as they are for more highly educated jobseekers.

• In the future, prioritise ALMP spending towards measures and services that have been evaluated and been found to be effective.

Implement systemic mechanisms to ensure training quality

• Systematically collect and monitor – and ideally, publish – information on training providers’ outputs and outcomes, including for specific courses.

• Adjust legal framework to allow quality criteria to be easily integrated into training course and training provider selection process.

• Allow clients to choose their training provider for the subset of training programmes where this is feasible.

Tailor/adjust design of employment incentives (wage subsidy) parameters to make them more attractive to difficult-to‑employ individuals

• Modify programme parameters to increase uptake amongst underrepresented groups such as individuals with low education, e.g. by redirecting subsidies towards such individuals.

• Complement the programme by adding professional in-work counselling support for especially disadvantaged groups.

Establish a fully-fledged ‑evidence‑based approach to ALMP design and implementation

• Allocate additional financial and human resources to MoLFSA and/or ESS to be able to conduct in-house evaluations and contract them out when needed to help meet operational and strategic needs for evidence.

• Establish a legal basis for MoLFSA to access additional data needed for research and evaluation purposes.

• Introduce a legal requirement to regularly evaluate Slovenia’s most important ALMPs via counterfactual impact evaluations.

• Embed possibilities for quasi‑experimental counterfactual impact evaluations into policy design, for example by carefully setting eligibility or assignment criteria and retaining necessary data for subsequent evaluation.

• Consider complementing evidence base, where feasible, with randomised controlled trials, especially in the numerous pilot programmes implemented in Slovenia (for example, to examine cost-effectiveness of increased intensity of counselling).

• Regularly conduct comprehensive cost-benefit analyses of ALMPs to evaluate whether different types of ALMP measures are cost-effective.

• Consider establishing a dedicated cross-ministry agency tasked with promoting and facilitating evidence‑based policymaking

Regarding other policies, which have not been covered by counterfactual impact evaluation:

Improve the provision of guidance and counselling programmes for vulnerable groups

• Increase counselling frequency for ESS clients requiring intensive support to facilitate an effective reintegration into employment.

• Establish a legal basis for paying outcome‑based payments to outside employment service providers – with options to let clients choose their provider – to complement the existing services provided by ESS.

• Expand mandate of ESS to include time‑limited support for the most vulnerable groups after they have become employed.

Streamline the customer journey with the help of digital tools

• Prioritise the introduction of a statistical profiling tool for the ESS to target and tailor services more efficiently

• Consider legal changes to postpone requirement for first counsellor-client meeting to be conducted within 15 days of registration for readily-employable and digitally literate jobseekers, which could be supplemented with automated individual action plan.

• Increase self-service options for readily-employable jobseekers in addition to the imminent rollout of a competency-based job matching tool.