What Makes a Labour Market System Mature?

We talk a great deal about best practice in employment services. We compare models, benchmark outcomes, and borrow instruments from systems that appear to be working. But we rarely stop to ask a more fundamental question on what does a mature employability system look like? How do we know, from the inside, whether the system we are working in is truly capable or simply going through the motions?

Rather than drawing on secondary literature or policy documents, I spoke with practitioners and researchers who have worked inside very different employment systems across Europe and beyond, deliberately avoiding the Anglophone world, which tends to dominate these conversations. I also draw on field research I conducted in Mongolia in 2025 as part of an Asian Development Bank employment services assessment. Collectively, these conversations span Italy, Poland, Greece, South Korea, France, Benelux and a range of developing economy contexts. What they produced was a remarkably consistent set of convergent findings, the same structural failures, the same traps, the same gaps, identified independently by people from those different systems.

"If you need support, just go to the shop. You will find the support. That is actually the definition I'm still using of service in labour market maturity." — Giampaolo Montaletti, Italy

What maturity really means

Before we can ask whether a system is mature, we need a working definition. This turns out to be less straightforward than it appears.

Venetia Koussia, who led ManpowerGroup Greece for over fourteen years before moving into competitiveness and governance work, challenged the concept from the outset. "We are mature in terms of years," she said of the Greek system but not in terms of how the system response and how it serves its citizens. Her distinction is clear, chronological maturity, age, institutional presence, years of accumulated regulation, is not the same as developmental maturity. A system can be old without being capable. It can have institutions without having learning. It can have processes without having a purpose.

Eamonn Davern, who has led EU PES benchlearning exercises across twenty member states and worked with the ILO, World Bank and numerous national governments, based the maturity on the criteria developed through the European benchlearning process. It has been in itself a significant undertaking, since it required dozens of public employment services of very different sizes and resources to agree on a shared definition. The key dimensions that emerged:

• a clear organisational strategy,

• robust performance management,

• strong audit of operational processes,

• system sustainability (defined not as survival but as maintained relevance), a genuine employer engagement agenda,

• systematic change management, and critically,

• recognition that the public employment service is part of a wider ecosystem, not the centre of its own solar system.

To that framework, Maëlle Blanchard, who works in European affairs and benchlearning at France Travail, adds one of the most sophisticated accountability tools in use anywhere: the impact indicator. France Travail now asks, for every new initiative, not just what the results are, but what the results would have been without the intervention. "If France Travail was not there, what would be the result? What is the added value of France Travail in the process?" It is a deceptively simple question. Most systems do not ask it.

But the most striking definition came from Anna Karaszewska, who runs Jobs First, a Polish employment organisation specialising in activating people furthest from the labour market, and who has spent years fighting a system that, in her assessment, has never come close to maturity. For Anna, a mature system starts with a compass bearing, she claims that someone must define which challenges the system is actually trying to solve. From there the elasticity to respond to crises, outcomes measured in sustainable employment rather than job starts, integration across social protection, and continuous evaluation. But the most fundamental criterion is the simplest, when the labour market is treated as serious economic policy, not a political tool. "The labour market has never been treated seriously as an important part of the economy," she said. "It is treated as political playground."

"The system must have an azimuth, a compass bearing. Someone has to first create a direction: which challenges are we addressing, what problems do we want to solve." — Anna Karaszewska, Poland         

The cost-effectiveness trap

That political framing has a practical consequence. When the labour market is treated as a political tool rather than an economic one, the metrics that govern it tend to follow the same logic, rewarding what is visible, measurable and quick. And nothing is more visible, measurable and quick than placing someone who was already close to employment back into work. If there is a single finding that emerged from every conversation, unprompted and independently, it is that systems consistently optimise for the people easiest to serve and structurally exclude those who need the most help.

Nobody designs a system to fail the people who need it most. But when cost-effectiveness is the governing metric, that is what happens. When a programme is judged by the number of people placed in employment divided by the money spent, the rational response of any provider or job centre is to concentrate on those closest to the labour market. They cost less to activate, they generate outcomes quickly, and they make the numbers look good. Those furthest from the market, the long-term unemployed, people experiencing homelessness, ex-offenders, those with complex barriers, cost more, take longer, and generate outcomes that are harder to sustain. So, they get left behind.

Giampaolo Montaletti, who was centrally involved in building the Lombardy employment services model and a Visiting Research Fellow at Manchester University, argued that the Italian system has built a working paradigm for roughly 30% of the labour market, those who are close to employment and can navigate the standardised support on offer. For the other 70%, no working paradigm exists. Two structural design flaws drive this. First, employer selection blindness as programmes are designed without accounting for how different employers in different sectors and regions select people. Second, entry barrier selection, where digital forms, online registrations and administrative processes are required to access support, demand exactly the skills that, if someone possesses them, mean they can probably find a job without support. "If these people are able to do all these things, they are able to find a job themselves." Giampaolo explained.

Dr Sang Hyon Lee, Director at the Korean Employment Information Service, confirmed this pattern from a very different context, and with striking candour about his own system. Korea's employment services are among the most digitally advanced in the world. Yet the most vulnerable, homeless people, ex-offenders, the severely disabled, are structurally excluded from mainstream support. "For me, I want to go for the hardest to serve in the labour market. But it's not cost-effective. It takes too much investment. So, they are excluded from our system. The priority is wrong, but our whole society is focused on cost-effective," he said.

The same dynamic appeared in my field research in Mongolia in 2025, conducted for the Asian Development Bank. Employment services were visibly drifting toward easier-to-serve clients, university students and graduates, who could demonstrate outcomes quickly, rather than the vulnerable populations the programmes were mandated to support. Nobody had decided to abandon the mandate. The funding model had simply made serving it too costly.

It appeared in Greece the drift takes a different form. Venetia noted that government subsidy system, which runs to approximately eighteen different schemes, fails to attract the employers who drive the economy. "Real employers do not deal with this," she said. The businesses that do engage are often those looking for a way to reduce their wage bill rather than those truly seeking to expand their workforce. The subsidy reaches the wrong employers for the wrong reasons, and the right ones stay away.

In Poland, Anna offered perhaps the starkest illustration. She had delivered an outstanding activation programme for Ukrainian refugees, run with the UN and UNHCR, achieving employment outcomes that the mainstream system could not come close to matching. She presented the results at a series of meetings with humanitarian NGOs, distributed a report, walked through the methodology and explained what had worked and why. "Not one question," she said. "Not one expression of interest. Nobody." When a solution exists entirely outside the system, there is no institutional framework to receive it, no channel through which it can travel, no one with the mandate or the incentive to act on it. "There is no market," she said. “Nie ma rynku.”

What makes this trap particularly deceptive is that it is self-concealing. Systems that serve only the easily served can still report respectable employment rates. Anna was very direct about the headline figures. "Looking only at the unemployment rate is, for me, a kind of irresponsibility. You could even call it a delusion." She said. In Poland, some two million people are completely outside the labour market, not registered, not counted. Venetia reported the same dynamic in Greece: official unemployment has fallen from 17-18% to 7-8%, but 26% of labour is undeclared and 68% of people are on the verge of poverty. "I don't care about that number. Fewer people have been born, fewer people have been asking for a job. Nobody talks about that."

"I don't care about that number. Fewer people have been born, fewer people have been asking for a job. But 26% of labour is undeclared. Nobody talks about that." — Venetia Koussia, Greece

The evaluation paradox

What makes this trap so durable is not ignorance. In most of the systems described here, the evidence of who is being left behind, and why, exists. It is collected, analysed and sometimes published. The problem is not that systems do not know. It is that knowing and acting on what you know are, it turns out, entirely separate things.

Every system in this study has some version of an evaluation function. And yet across almost every case, the same pattern appears, the infrastructure for learning exists, the political willingness to act on inconvenient findings does not.

In Poland, Anna described the contracting out / commissioning programme, a public-private partnership instrument for activating the long-term unemployed that produced, in her own Małopolska region programme, employment retention rates of 72-73%, with more than half of participants moving into permanent employment contracts. The programme was assessed as cost-ineffective by the government and effectively dismantled, not because the evidence was wrong, but because the evidence was inconvenient. The successive governments that killed it, focused on the regions where the model had been poorly implemented, not on the regions where it had worked. "I can say with complete responsibility there is no policy. Not even an immature system. There is no system in the sense we understand it, one that responds to challenges and produces lasting results." She explained.

In Korea, Dr Lee described being asked to monitor a new employer services programme and being told explicitly that the Ministry wanted compliments, not findings, a success story to report to the President by year end, regardless of what the evidence showed. The pattern repeats annually, a programme launch, declare success, replace with something new. "I wonder," he said, "from all value for money, the real value is someone who is left behind in society. We should spend more money. But politically, it's not working well in Korea."

Greece, in Venetia's view, has not yet reached the evaluation paradox. The problem sits one step earlier. In her words: "First of all, willingness, then capability. Willingness to measure. Because now if you ask them, you might not hear this." Even company registration data in Greece is, she noted, "proven fake." Is the absence cultural, before it’s technical then?

Elsewhere the problem is more institutional. In Mongolia, funded programmes for workers affected by mine closures were designed with output indicators only as the number of people trained and no employment outcome indicators at all. When asked why, regional officials said it was "too hard" and "too much work." The explanation is revealing, not about Mongolia specifically, but because of what it says about the accountability framework above it. Anna reflecting on the same pattern in Polish EU-funded programmes, put it plainly: "Because nobody requires it. Not at regional level, not national level, not EU level." No authority at any of those levels requires programmes to demonstrate whether participants found and kept work. The indicator is simply absent from the system that is supposed to ensure it exists.

France Travail is the partial positive exception. The impact indicator approach, asking what would have happened without the intervention, is an open attempt to break the evaluation paradox by asking the right question rather than just measuring the most convenient one. But Maëlle Blanchard was careful not to overclaim, saying "I would not be able to reply with a yes or no answer because I don't have all the data available." The institutional honesty in that response is itself a maturity signal.

The political insulation problem

One finding that was not anticipated, and that emerged with unexpected force, is that mature employment systems require some degree of insulation from short political cycles. This is not simply about stability or continuity, it is about whether the institutional framework can survive a change of government without being dismantled, repurposed, or hollowed out.

Having watched successive Polish governments, of the right, the centre, the left, each treat the employment system as political territory rather than economic infrastructure, Anna reached a conclusion that stops many listeners short: the system may require constitutional protection. "I don't know how to create a mechanism to stabilise certain things without anchoring them constitutionally." She is clear about the implications because this shows "not just the immaturity of the labour market system, but the immaturity of the political system itself".

Giampaolo's experience in Lombardy suggests a different possibility. The Lombardy model achieved something rare as it became hegemonic across the political spectrum. Both left-wing and right-wing regional governments adopted and developed it, because by the time there was a change of administration, the model had built enough institutional legitimacy to survive. What made that possible was not constitutional protection but something more fragile, a period of consistent, evidence-based development that built cross-party consensus. The lesson is not that it always works. It is that it sometimes can.

Whether that kind of legitimacy can be engineered, or whether it has to be earned, is an open question. Eamonn pointed to a structural condition that makes it more likely. Public employment services internal to a government ministry are more politically exposed than those with independent legal status. The German Federal Employment Agency, with a board that includes ministry representatives, unions, employers and PES representatives, has considerably more operational freedom than a PES that reports directly to a minister. Austria's long-standing independent job centre structure works on the same principle.

France, on the other hand, illustrates the territorial dimension of this problem. Maëlle described working with departmental councils led by elected presidents from different political parties, each with the legal authority to simply ignore national direction. "If they did not agree with the policy that we were trying to implement, it was quite hard to really convince them, because we are just an agency related to the Ministry of Labour. We have not the authority to say to a president of a departmental council what they do."

Venetia named the consequence of inaction. "The political cost might be huge. It is better to do nothing than to get involved in everything." In a system where every reform creates winners and losers and accountability mechanisms are broken, standing still is often the rational choice. Through her competitiveness work, she has sat in high-level discussions under Chatham House rules where government representatives said things about their own systems, they would never say in public. "These discussions were very open, very, very open. The next step — nothing."

Digital sophistication is not the same as maturity

Not every system responds to political constraint with paralysis. Some respond with ingenuity. Korea's response was to build some of the most advanced digital employment infrastructure in the world. What drove that is more complicated than it looks. The Employment 24 platform, AI job matching, AI career counselling, mobile-first design, real-time labour market intelligence, the infrastructure is world-class. Dr Lee noted that the COVID-19 was a real stress test for this infrastructure. While the UK, US and Germany struggled to deliver basic unemployment benefits for months, and Japan's online services required an in-person visit to activate, rendering them effectively useless, Korea never shut its job centres. Mobile and online services were already fully operational.

And yet by Dr Lee's own assessment, the Korean system is not mature. The coverage gap for the most vulnerable is its central challenge. And the digital sophistication turns out, on examination, to be substantially a response to a budgetary constraint rather than a positive strategic vision. Dr Lee explained that Korea ideally needs 50,000 employment service officers but it has 5,000. The Ministry of Planning and Budget has consistently declined to fund more so the digitalisation filled the gap. Technology compensating for underinvestment in human capacity is a very different story from technology enabling better services.

This is not an argument against digital infrastructure but for understanding what it is and what it is not. Eamonn made the broader point about legacy systems. Many European PES have spent decades accumulating costly office infrastructure built on the assumption that unemployed people will come to them. Countries building employment services from scratch have no such legacy to manage, and that, paradoxically, is an advantage. They can go straight to mobile-first, digital-by-default delivery, reaching citizens in remote communities without a single job centre.

Dr Lee has seen this directly. Korea's ODA programmes have built mobile-only employment systems in Mongolia, Rwanda, Vietnam and other countries that lack the physical infrastructure for traditional job centres. For him the argument is not primarily technical. "For me, all human beings need such support. Not just European countries, all human beings globally. That's humanity."

While, Dr Lee is talking about access, Eamonn's concern is also with what happens once people are in the system. When it comes to AI profiling, he is not convinced. Despite increasingly sophisticated algorithmic systems, most of the advanced employment services in Europe still maintain counsellor determination as the final step. "Notwithstanding all the debates about AI and the future implications for jobs, maybe for a long time there may be something quite innate that still means that a very trained, professionalised counsellor can bring something to the dialogue that you're not going to get from a purely digital kind." He observed. The three types of profiling, rules-based, statistical modelling and counsellor determination, each have their limits. Administrative designations produce false positives. Statistical models produce false positives and false negatives. Counsellor override remains essential. Twenty years ago, experts at Germany's first large profiling conference predicted that by now, employment services would be run by robots. They were wrong.

"For me, all human beings need such support. Not just in European countries or South Korea — all human beings globally. That's humanity." — Dr Sang Hyon Lee, South Korea

The ecosystem gap

If the internal challenge is keeping the human relationship at the centre of service delivery, the external one is persuading institutions to have any real relationship with each other at all.

Venetia's formulation was this: "All these partners, they never look at each other. Never. Not even once. At best, individual meetings with the ministry or a union. Never together. We don't miss the body. We miss the commitment." Creating another institution, another committee, another legal framework is therefore, not the answer. What is missing is not structure but the will to use it.

For Eamonn, a partnership and an ecosystem are not the same thing. A partnership is a legal requirement, the right organisations meeting at the right intervals. An ecosystem is something different, it’s a honeycomb of organisations each with mutual reciprocity, where the PES is one node among many rather than the centre of its own solar system. "If two organisations focused on helping citizens come together in that way, you increase the kind of potential benefit from the ecosystem."

In the conversation about the French experience, Maëlle identified three distinct barriers to true collaboration, each requiring a different response.

• The political barrier: departmental councils with different political mandates from the national government resist direction they disagree with, and no agency-level power can override elected officials.

• The technical barrier: incompatible digital systems across partner organisations make even basic data sharing difficult, compounded by GDPR and data protection requirements.

• The professional culture barrier: teams trained as social workers, HR specialists and business counsellors have fundamentally different visions of what support means and speaking the same language across those backgrounds takes deliberate investment.

France Travail's answer to the culture gap was the France Travail Academy, a shared training platform open to all partners. "I think this really helps us trying to speak the same language so that when we try to reach out to each other, we really can have a core understanding of each other and how we can work together."

The French challenge assumes partners who are at least structurally separate but that is not always a given. In Mongolia the National Association of Private Employment Agencies (NAMEA) simultaneously holds roles as provider, trainer, quality assurer and sector representative. That is a concentration of functions that removes the possibility of independent quality assurance from the outset. Job seekers and employers have no way of knowing whether the standards they are told are being met actually are.

Looking in the mirror

Across all the conversations, what became clear to me, was how readily the most thoughtful practitioners named their own system's challenges and how rarely that critique travels upward into policy.

Dr Lee named the Korean shortcoming when the system excludes the people who need it most because serving them is not cost-effective, and he knows it. Eamonn described what happens to employment services that lose the confidence of the public they are meant to serve as a gradual spiral. Low reputation leads to low usage, lower usage makes the service less valuable, and a less valuable service attracts fewer people still. He has watched it happen in both developing and established systems. Venetia was more personal about it: "I am optimistic about me doing the work. But I am not so optimistic about seeing things happening." And then there was Giampaolo speaking about relational approach: "Relationality is not out of the office. Look at your employees in the public employment service, some of them are compassionate about their clients. Talk to them. You don't need a guru."

What all of these have in common is a refusal to pretend. And that is worth noting. The systems that seem to be improving tend to be the ones where people inside them can say clearly what is not working, not as a management problem, not as a resource problem, but as something built into the way the system is designed, funded and held to account.

Anna's story about Poland is a warning that travels beyond its own context. Within a decade, the commissioning out experience, the methodology, the results, the evidence, has effectively vanished from institutional memory. Practitioners who joined the field after 2016 simply do not know it happened. "Today there is no discussion. None. Not the kind where you say: what are our problems, what instruments should we look for, how should the system work. That does not exist. Anywhere."

Systems that cannot hold their own history are unlikely to learn from it.

What a mature system does

The conversations are not only accounts of challenges. Alongside the critiques, each practitioner described something they had seen work or believed could. The principle, articulated by Eamonn is that mature systems adapt the process to meet the needs of the individual, rather than retrofitting the individual into a process. This is the foundation of the Estonian and Norwegian approach to counselling, a level relationship between counsellor and job seeker, where the counsellor is a mirror enabling reflection rather than an official doing things to the client. Actiris in Brussels uses the degree of autonomy identified at the initial interview as the primary yardstick for support, not skills or education or physical distance from the labour market, but how autonomous is this person in managing their own agenda?

"The cardinal principle that works is a mindset where, as far as possible, you're not seeking to fit somebody and retrofit an individual job seeker into a process, but you're seeking to amend the process to meet the needs of the job seeker." — Eamonn Davern, international expert

Giampaolo saw the same principle at work in Lombardy, though from a different angle. A mature system absorbs the complexity of multiple funding streams, regulatory requirements and governance layers behind the curtain, presenting a coherent, seamless service to citizens. "The complexity of managing public funding should be hidden behind the curtain. It shouldn't be a problem for the citizens." The bureaucratic architecture, the European Social Fund, national funding, regional programmes, is the government's problem, not the job seeker's. A system that discharges its own administrative complexity onto its users is not yet mature.

The same approach applies to employer engagement, though most systems have not yet followed it there. The shift most are making (slowly still) is away from transactional contact, where subsidies are offered, vacancies listed, candidates referred. What it needs to become is more like a standing conversation with employers about where their workforce is headed. Luxembourg's employment service is already doing this, getting into companies upstream to understand their trajectory and asking whether early support for reskilling might prevent redundancies rather than manage them once they have happened.

How success is measured matters just as much as what services are offered. "The biggest misconception is that public employment services are there to find people work," Eamonn said. “They're there to ensure that people are equipped to actually find work." That shift from placement to preparation, from job start to labour market attachment, changes everything that gets counted. Most systems currently reward a job start. Far fewer track whether that job is still held six or twelve months later. “A revolving door could be very adverse for productivity.” He noted. Ireland has done interesting work on distance-travelled models that base success on self-actualisation and counsellor-negotiated goals, an approach that is sometimes dismissed as soft, but Eamonn does not think that is fair.

The difference between less and more mature labour market systems

Context shapes the model

Drawing on Labour Intermediation Services in Developing Economies (Palgrave Macmillan, 2016) by Jacqueline Mazza, former Principal Labour Markets Specialist at the Inter-American Development Bank and now Professor at Johns Hopkins University SAIS, Eamonn argued that there is no single trajectory for building a mature employment system. There are at least three, depending on context.

Where a country is moving toward a formalised labour market in a foreseeable timeframe, the goal is a comprehensive PES on the broadly European model, evidence-based and person-centred. Where one or two dominant sectors define the economy, mining, IT, hospitality, the employment service's most effective contribution may be to focus specifically on improving labour supply for those sectors, building the pipeline from education and training into the occupations where demand is strongest and where labour market information asymmetry is widest. And where the informal economy is likely to remain dominant for the foreseeable future, a large agrarian economy, a dispersed rural population, a weak formal sector, the employment service's role may be fundamentally different: registration and data infrastructure, entrepreneurship support, microfinance linkages, social integration. Not a scaled-down version of a European PES, but a different institution entirely.

A good examples comes from Colombia, where two provinces facing a declining PES reputation developed a quality stamp system, a provincial accreditation scheme that recognised any organisation, NGO, public or private, as an authorised employment service deliverer for a given geography, provided it met minimum service standards. "The rising tide can raise lots of different boats," as Eamonn observed. The lesson is about dispelling the assumption that any non-public provision is a threat to the public employment service. The interests of citizens are served by quality, not by institutional ownership.

"The biggest misconception is that public employment services are there to find people work. They're there to ensure that people are equipped to actually find work." — Eamonn Davern, international expert

Conclusion

The conversations in this article do not point toward a technical answer. The systems that are truly maturing are not the ones with the most sophisticated digital platforms or the most elaborate governance structures. They are the ones that have found a way to stay truthful about what they are for, who they serve, and whether they are doing it.

That means a strategy that outlasts electoral cycles. Evaluation that asks the right questions rather than the convenient ones. Resources directed toward those who need them most, not those who are easiest to help. An employer relationship built on true understanding and collaboration rather than subsidy. Digital infrastructure that extends human capacity rather than replaces it.

None of that is complicated in principle. What makes it hard is that every item on that list requires the system to work against its own short-term incentives, to serve people who are costly, to publish findings that are inconvenient, to build relationships that take years to bear fruit. The systems that manage it tend to have one thing in common, people inside them who are willing to say clearly what is not working, and institutions structured well enough to act on what they hear.

None of the systems described in these conversations have fully achieved that. Several are on a plausible journey toward it. And several are not yet asking the question.