The program has recorded exits from long-term unemployment since 2016, with approximately 4,000 employees in EBEs by mid-2025. The program has demonstrated that it is possible to durably reintegrate people most distant from the labor market. The argument of budget neutrality, which made the program politically viable, did not withstand the data. In 2026, France will need to decide if it is ready to finance inclusion explicitly, without claiming that it pays for itself.
The essentials
- Positive social effects are reported, but robust causal improvement in health and dignity is not demonstrated (L’Horty scientific committee, 2025).
- 83 territories participate in the program, which employs 4,000 employees in employment-purpose enterprises (EBEs) by mid-2025.
- The subsidies and direct costs of the program exceed estimated costs avoided and revenues, according to iFRAP’s budgetary analysis.
- A minority share of beneficiaries sustains themselves durably outside the program, raising the question of structural dependence on public funding.
- The 2026-2027 debate concerns collective consent: accepting to pay explicitly for inclusion, or redirecting these resources toward other integration levers.
A gamble on the invisible workers of the labor market
TZCLD was born from a simple observation: some people cannot find employment because the ordinary labor market does not want them, regardless of training, incentives, or sanctions. After long periods of inactivity, health problems, lack of mobility, criminal records, age—each territory has hundreds of people that no one will hire. The idea comes from Patrick Valentin, taken up by ATD Fourth World in the early 2010s and subsequently supported notably by Louis Gallois. It is radical in its simplicity: create jobs adapted to these people, financed by public contributions, relying notably on avoided social costs in its economic reasoning.
The program works like this: employment-purpose enterprises (EBEs) hire on permanent contracts, on a flexible schedule, people voluntarily without employment for more than a year. EBEs develop real economic activities—market gardening, carpentry, personal services, recycling—partly market-based, partly subsidized. The State finances the Fund through a contribution to employment development, while avoided allocations constitute a cost-benefit calculation hypothesis, completed by a contribution from local authorities.
The gamble was twofold. The first, human: proving that these people can work if adequate conditions are created for them. The second, budgetary: showing that the program finances itself through the reduction of social spending. It is this second gamble that is being questioned today.
Human benefits are real and documented
The scientific committee chaired by Yannick L’Horty published its final report in September 2025, following a progress note in 2024. Its conclusions indicate effects for the people concerned: the health of employees was measured, self-esteem was observed in a survey, and their financial situation improved, without robust demonstration of causal improvement over time through the program for health and self-esteem. Access to stable income, a collective framework, and social recognition produces effects that allocations alone do not generate.
These results align with what international literature on job guarantees has documented for decades. Work structures time, creates social bonds, and confers status, beyond the income it provides. For people excluded from the market for five, ten, or fifteen years, accessing a permanent contract, even within an EBE, significantly alters their life trajectory.
A significant number of exits from long-term unemployment recorded since 2016 represent just as many people who have regained this structure. In territories often rural or peri-urban, marked by deindustrialization, this figure is not negligible. EBEs have become full economic actors: they produce goods and services, they participate in local life, they recruit suppliers. The experience in Pipriac, in Brittany, or in Colombey-les-Belles, in Lorraine, shows that entire territories can be revitalized by this type of initiative.
This data does not show that the program constitutes an exit door to the ordinary labor market. A minority share of beneficiaries stabilizes durably outside the program. For the majority, the EBE is not primarily designed as a springboard and exits are infrequent. The program’s promoters acknowledge this themselves: TZCLD is not a classical reinsertion program, but a form of job guarantee for people structurally excluded by the market.
The budgetary promise that did not hold
One must understand why the promise of neutrality was formulated with such conviction initially. The program’s designers reasoned from a stock logic: a person durably distant from employment mobilizes recurring social spending, and if one substitutes this spending with productive wages, the net balance should tend toward zero. This reasoning is coherent in theory, but it rests on a hypothesis that proved fragile in practice—that the targeted people would be precisely those costing the most to the social protection system. Yet EBE employees are less often RSA recipients before hiring than employees of economic activity insertion programs. The lower initial exposure to RSA can limit potential savings on this benefit, without isolating its effect on all transfers.
The neutrality calculation also assumed that operating costs would remain controlled once the program was running smoothly, but support needs proved more significant than expected, notably because the quality of support is precisely what produces the documented human effects.
Reducing management costs would therefore amount to cutting what makes the model valuable.
Budget neutrality was the central argument for making TZCLD politically acceptable. The idea rested on redirecting unemployment allocations, RSA, housing aid, and health spending toward productive employment. According to the short-term evaluation, public finances as a whole spend more than they recover. Studies from 2016 and 2017 had advanced encouraging estimates.
iFRAP, which examined the accounts more closely, reaches a different conclusion. The overall net cost of the program remains high. Public spending considered exceeds estimated costs avoided and additional revenues in the short term. For each EBE employee FTE studied, public finances as a whole spend more than they recover in the short term. The exact magnitude of this gap is difficult to stabilize, as the calculation perimeters vary depending on the studies, but the direction is consistent.
Several factors explain this gap. A significant share of people hired in EBEs had reduced allocations: their RSA was low, some had not received unemployment benefits for a long time. Expected savings on allocations thus proved more modest than anticipated. Simultaneously, the management cost of an adapted employment structure, with high supervision rates, activities low in productivity by nature, facilities to finance, proved structurally high.
This finding does not invalidate the program, but it imposes a change in framework. TZCLD redirects funding and produces documented positive effects, at the price of significant public cost. It remains to determine what collective value society accords to these outcomes.
Learning from the territories
Beyond aggregate figures, ten years of experience has produced considerable practical knowledge. The 83 territories participating in the program have developed highly varied activity models, adapted to local needs and available skills. Some EBEs have succeeded in generating significant market revenues, up to 40% of their budget in the most successful cases. Others remain highly dependent on subsidies, because they welcome profiles most distant from employment.
This diversity is both a strength and a limitation. A strength, because it shows the model’s capacity for adaptation to very different contexts. A limitation, because it makes generalization on costs and benefits difficult. An EBE employing ten people in a town of 2,000 inhabitants is not managed like a structure with fifty employees in a medium-sized city.
What emerges from this decade is also a better understanding of who benefits from the program. EBEs can provide a stable solution to a portion of people very distant from employment, notably those over 55 facing chronic health problems and long periods of inactivity, without constituting an adapted solution for all profiles. For them, the ordinary labor market is closed, and the EBE represents real inclusion. For profiles closer to employment, other programs might be more effective and less costly. This segmentation, better understood today, could allow concentrating TZCLD where its marginal impact is strongest.
The connection to labor market transformations is not incidental. Automation that blocks median wages and concentrates gains at the top mechanically fuels the stock of people the market durably excludes. TZCLD intervenes downstream of this process, but the structural causes of exclusion are not resolved at the territorial level.
The 2027 dilemma: pay explicitly or renounce
The second experimental phase, launched by the 2020 law, covers the period 2021-2026 and has been extended until December 31, 2026. The program’s continuation depends on an ongoing legislative process, with a targeted permanence bill under parliamentary examination.
The decision depends on a choice of values that data alone cannot settle. If one accepts that inclusion in work has value in itself, distinct from its budgetary return—TZCLD is a credible and proven instrument. If one holds that social policies must be self-financing, or that job guarantee creates unwanted dependence, the arguments are weaker.
The 2025 final report of the TZCLD scientific committee, supported by the Dares and the High Commission for Strategy and Planning, opens an intermediate path: concentrating the program on territories and profiles where impact is most demonstrated, acknowledging a significant net cost and suggesting better targeting as well as changes to the program’s parameters and principles. This approach would make it possible to maintain human benefit while making visible, and democratically debatable, the real cost of inclusion.
The French experience is not isolated. Comparable programs exist in Austria (Marienthal), Finland, and the United States on a small scale. All stumble on the same tension: benefits for people are real and measurable, budgetary benefits are uncertain or negative. The difference between those who have made programs permanent and those who abandoned them often depends less on evaluation rigor than on political consensus about what the State must guarantee.
TZCLD has demonstrated something important: we now know how to create adapted jobs for those most distant from the market. The expertise exists, territories have acquired it, management tools have been refined. The question that remains open, and which is fundamentally political, is whether French society is ready to inscribe this cost in its social budget permanently, not as a promise of neutrality, but as an assumed choice for inclusion.
Sources
- France Stratégie, Toward a job guarantee: final report of the TZCLD scientific evaluation committee
- Scientific committee Yannick L’Horty, Independent TZCLD evaluation, 2024 (France Stratégie / UPEM)
- IFRAP, Budgetary analysis of the Zero Long-Term Unemployment Territories program
- CEREQ, Brief 2026 on the trajectories of long-term integration program beneficiaries
- Ministry of Labor, TZCLD experimentation monitoring data, mid-2025
