Fewer than one in two young people without qualifications are employed ten years after leaving the education system. France ranks 22nd in Europe on this indicator. Meanwhile, 16.5 billion euros of public support for apprenticeships go each year primarily to young people preparing for higher education diplomas. The choice for 2027 is clear: continue to pay massively at the start of training, or tie public funding to jobs actually obtained, beginning with those who need it most.

A relative decline established over a quarter century

France awards many diplomas. With 51.7% of higher education graduates among 25-29-year-olds, it ranks 7th in Europe [6]. But its employment rate across all qualification levels places it 15th in the Union, down from 11th in 2002 [6]. The expansion of education has redistributed difficulties without reducing them.

The dividing line has remained the same for twenty-five years. In 2023, the unemployment rate reaches 42% among those with little or no qualifications in the first four years after leaving the education system [3]. For those with a master’s degree, it falls to 7% [3]. The gap narrows over time, but never disappears. More than ten years after the end of studies, those with little or no qualifications still show an unemployment rate of 10%, compared with 4% for the most qualified [3].

In 2024, INSEE estimated 1.5 million young people aged 15 to 29 without employment, education, or training [5]. In the second quarter of 2026, they represent 13.3% of this age group, up 0.3 percentage points over a year [13]. France now exceeds the European average, a reversal from the situation before the health crisis.

The current economic situation worsens a structural trend. The overall unemployment rate reaches 8.1% in the first quarter of 2026, its highest level in five years [12]. Among 18-25-year-olds, it climbs to 17.9%, more than double the national average [12].

Apprenticeships: a windfall captured by those who needed it least

The 2018 reform increased the number of apprentices from fewer than 440,000 to over one million by the end of 2023 [8]. The success is real, but it benefits primarily those who did not need it.

The rise in entries comes mainly from apprentices preparing higher education diplomas. They now represent 62% of entries, twice as many as ten years ago, five times as many as in 2003 [8]. The effectiveness of the scheme in job placement is lower, and declining, as the level of diploma being prepared is higher.

The public cost of this shift is massive. Public support for apprenticeships reaches 16.5 billion euros: 7.5 billion from the state budget, 8.7 billion from France compétences [11]. This is more than the budget for justice. Total national spending exceeds 24.9 billion euros in 2023, or roughly 26,000 euros per apprentice, twice what the state allocates to each higher education student [8]. These figures are cited in the next paragraph for comparison purposes and come from the same source [8].

The Court of Auditors notes that this change “does not correspond to the objectives historically associated with apprenticeship policy, which has so far aimed at improving job placement for young people with the lowest qualification levels” [11].

The exceptional aid created in 2020 as part of the recovery plan, continued without connection to the health crisis that had justified it, covered 100% of labour costs for most apprentices, including those in higher education [8]. Nearly half of companies would have recruited work-study students anyway, with or without the subsidy [8]. Public money therefore financed private decisions that would have been made without it, to the detriment of young people who had no alternative.

Entry-level jobs erased by automation

A technological rupture is rewriting the rules for entering the labour market. It affects junior positions first, including those that apprenticeship policy claims to promote.

Junior jobs are the first affected. Stanford researchers have found evidence of job losses among 22-25-year-olds in positions heavily exposed to automation [16]. Since 2022, the employment rate of these young workers has fallen by more than 10% in sectors such as software development and customer service [16].

Axelle Arquié describes a form of cognitive Taylorism: like the assembly line worker, knowledge workers risk no longer producing end-to-end, but validating and correcting outputs generated by AI [15]. The positions that allowed people to acquire the skills and networks necessary for a career are precisely those that automation erases first.

The unemployment rate of 15-24-year-olds reached 21.5% at the end of 2025, up 4.6 percentage points in three years [12]. This rise coincides with the acceleration of AI agent deployment in companies.

The consequence is twofold. Non-qualified young people remain stuck on a segmented labour market, where short-term contracts more often lead to unemployment than stable employment [14]. Qualified young people, meanwhile, see their early career years threatened by technological substitution. The diploma offers less and less protection: in 2024, 15% of 15-34-year-olds in employment who have completed their studies believe their skills exceed what their work requires [4].

Shifting the funding decision toward results-based contracts

The current system heavily subsidizes entry into training without tying public funding to actual job creation or its quality. The state pays for the training place without knowing whether the job at the end of the pathway exists, whether it is sustainable, and for what profile of young person.

The choice for 2027 is architectural. It is about transferring the funding decision for work-study training from professional sectors toward a results-based contract between the state, regions, and companies. This contract would be based on job placement indicators at eighteen months: salary level, contract duration, alignment with the level of training.

Hiring subsidies would no longer be paid at the start of the contract. They would be conditional on exit, with proof of job placement. For training beyond secondary school, public funding would decline structurally. For below-secondary levels, it would increase, with an additional bonus for companies located in territories with high concentrations of young people without employment or training.

This change shifts the decision-making: today, professional sectors define subsidy levels without accounting for job placement results. Tomorrow, the state and regions condition their funding on proof of results. Who decides, who pays, and on what basis shifts hands.

Dani Rodrik formalized this logic: a modern industrial policy creates “good jobs” by improving productivity and labour income in services [17]. It defines good jobs as those that invest in worker skills, give them autonomy and responsibility, and extend beyond manufacturing sectors. It rests on collaborative interaction between public agencies and companies, with a portfolio of personalized services, rather than uniform subsidies paid at entry [17].

Bertrand Martinot advocates for more modular vocational training accessible throughout working life [18]. This means breaking with the model of massive funding at entry into initial training, in favour of continuous support indexed to actual trajectories.

What this choice closes is clear: professional sectors lose control over subsidy levels, and higher education institutions that massively developed apprenticeships since 2018 would see their funding flows reduced. What this choice makes possible: concentrating public money on young people who need it, in the territories where they are, with proof of results before payment.

Sources

[1] OFCE, Bruno Coquet, “Why does youth unemployment resist new resources?”, OFCE le blog, 24 April 2024, https://www.ofce.fr/blog2024/fr/2024/20240424/ (accessed 09/09/2026).

[2] INSEE, “Job placement: France, social portrait”, France, social portrait, 2025 edition, https://www.insee.fr/fr/statistiques/8612524?sommaire=8612596 (accessed 09/09/2026).

[3] INSEE, “Unemployment in job placement phase”, Training and employment, 2025 edition, https://www.insee.fr/fr/statistiques/8305516?sommaire=8306008 (accessed 09/09/2026).

[4] INSEE, “In 2024, 15% of young people in employment consider themselves overqualified for their job relative to their skills”, INSEE Focus no. 361, September 2025, https://www.insee.fr/fr/statistiques/8642606 (accessed 09/09/2026).

[5] INSEE, “A snapshot of the labour market in 2024”, INSEE Première no. 2044, March 2025, https://www.insee.fr/fr/statistiques/8391807 (accessed 09/09/2026).

[6] INSEE, Training-Employment Assessment 2025, November 2025, https://www.insee.fr/fr/statistiques/8666376?sommaire=8666378 (accessed 09/09/2026).

[7] Céreq, Generation surveys, key figures by diploma, March 2025, https://www.cereq.fr/datavisualisation/insertion-professionnelle-des-jeunes/les-chiffres-cles-par-diplome (accessed 09/09/2026).

[8] OFCE, Bruno Coquet, “Apprenticeship: a review of the wild years”, OFCE Policy brief no. 117, June 2023, https://www.ofce.sciences-po.fr/blog/apprentissage-un-bilan-des-annees-folles/ (accessed 09/09/2026).

[9] OFCE, “Apprenticeship under pressure”, OFCE Policy brief no. 157, 2026, https://www.ofce.sciences-po.fr/pdf/pbrief/2026/OFCEpbrief157.pdf (accessed 09/09/2026).

[10] Court of Auditors, “Analysis of budget execution 2024: Work and employment mission”, April 2025, https://www.ccomptes.fr (accessed 09/09/2026).

[11] Court of Auditors / IGAS-IGF, Review of public spending on apprenticeships and vocational training, January 2025, https://igas.gouv.fr/sites/igas/files/2024-09/Rapport%20Igas%20-IGF%20Apprentissage-Formation%20(revue%20de%20d%C3%A9penses).pdf (accessed 09/09/2026).

[12] INSEE, IRES, Insight no. 34, “Youth unemployment and employment rates”, February 2026, https://ires.fr/wp-content/uploads/2026/02/Eclairage_34.pdf (accessed 09/09/2026).

[13] Le JDD, “Who are the NEETs?”, August 2026, https://www.lejdd.fr/economie/qui-sont-les-neets-ces-jeunes-sans-emploi-ni-formation-de-plus-en-plus-nombreux-en-france-181035 (accessed 09/09/2026).

[14] Banque de France, “Labour market duality in France”, https://www.banque-france.fr/en/publications-and-statistics/publications/duality-french-labour-market (accessed 09/09/2026).

[15] Axelle Arquié (CEPII / Observatory of threatened jobs), interview Comptoir IA, Maddyness, January 2026, https://www.maddyness.com/2026/01/22/impact-ia-emploi/ (accessed 09/09/2026).

[16] Stanford Digital Economy Lab, study on ADP data, summarized by AEF Info, April 2026, https://www.aefinfo.fr/depeche/748156 (accessed 09/09/2026).

[17] Dani Rodrik, “An Industrial Policy for Good Jobs”, Hamilton Project / Brookings Institution, 2024, https://www.brookings.edu/articles/an-industrial-policy-for-good-jobs/ (accessed 09/09/2026).

[18] Bertrand Martinot and Franck Morel, Work is the solution. Reconciling the French with work, Hermann, June 2025, https://www.editions-hermann.fr/livre/le-travail-est-la-solution-bertrand-martinot (accessed 09/09/2026).

[19] Senate, written question no. SQ241001662 on the budgetary cost of apprenticeships, JO Senate, 17 October 2024, https://www.senat.fr/questions/base/2024/qSEQ241001662.html (accessed 09/09/2026).