“75,000 fewer permanent contracts”: how AI is hindering youth employment in France

A joint report from the Bank of France and HEC points to an unprecedented freeze in hiring of entry-level employees since the rise of generative AI.

Published on October 8, 2026 at 11:27 AM

The job of developer is among the professions most affected by this decline in permanent contract hiring.

A new study, published jointly by the Bank of France and the Laboratory for Innovation and the Future of Work (LIFT) at the HEC business school on Thursday, October 8, reveals that permanent contracts for 21-29 year olds in France have been declining for the past four years in professions most exposed to AI. What was previously only an intuition among recruiters has become concrete thanks to the figures published by economist Antonin Bergeaud, associate professor in the economics department at HEC Paris, and his colleagues.

However, they warn, it is imperative to distinguish between the destruction of jobs caused by AI – current French data does not allow us to verify this phenomenon – and the closing of entry doors to employment, the first signs of which are appearing.

75,000 fewer permanent contracts per year on average

In professions most exposed to generative AI, permanent contracts for 21-29 year olds have fallen by 28.3% since 2022, the year ChatGPT became publicly available, compared to 7.5% in professions least exposed. More precisely, this drop represents approximately 6,200 fewer permanent contracts each month in these sectors, or roughly 75,000 fewer permanent contracts over a year. While experience seems to protect older employees, it is those at the beginning of their careers who are suffering the most, even though "the gap remains substantial up to age 49, so the decline is not limited to new entrants," the report's authors point out. This decline in permanent contracts has not been offset by fixed-term contracts, the study notes: "across all contracts, hiring of 21-29 year olds fell by 22.1% between November 2022 and March 2026" in the most exposed professions.

Developers on the front line

Software developers and engineers are a major factor in the workforce most at risk. A developer's job involves designing, programming, testing, and maintaining software or websites using computer code. And AI can code. In March 2026, permanent contracts for 21-29 year olds in these two professions fell by 42% compared to November 2022. The drop was only 11% in jobs less exposed to generative AI. "The slowdown also affects experienced developers," the report emphasizes. They have suffered from the economic climate: the digital sector recruited heavily between 2021 and 2022 but had to slow down when interest rates rose and hampered funding for tech companies. However, while this is the clearest finding of the study, it is also the most complex: developers exacerbate the situation without solely explaining it. Even without including them, hiring in the most exposed professions still falls by 24.5%.

AI is not the only culprit

The slowdown in long-term hiring for jobs exposed to generative AI is compounded by the increasing job security of young employees: not only are companies recruiting less, but the turnover of young permanent employees is also slowing. "The market is freezing, asymmetrically, because hiring is declining more than departures (...) Turnover is slowing and workforce renewal is weakening," the study notes. Several factors explain this: fewer hires mean fewer new employees, and therefore fewer departures in the first few months of a contract. Job offers are also becoming scarcer, and employees are changing companies less frequently. In this regard, Antonin Bergeaud and his team point out that "the increase in the ECB's deposit rate, from -0.5% to 4% between July 2022 and September 2023, weighed on investment and hiring in the technology sector."

Comparison between countries is impossible

The results obtained by other countries do not align with those of the Banque de France and HEC study due to methodological differences (data used, characteristics of the labor markets studied, etc.). In the United States, the authoritative study (Chandar and Chen, 2026) observes a relative decline in youth employment in high-risk occupations but does not cover "the entire private sector or the reallocations related to business creation and closures." In Norway, the benchmark study (Hernæs and Kostøl, 2026) finds that the evolution of youth employment is not significantly different between the most and least high-risk occupations. Developers, the most affected, do experience a decline of 18%, but this situation is not widespread across other high-risk occupations. The contrast with France may stem from the composition of the occupational groups studied or from the labor market itself, suggest Antonin Bergeaud and his team.

Redefine the missions of the first positions

Beyond this undeniably worrying observation, the study by the Bank of France and HEC suggests areas for improvement and warns that companies that save money in the short term by outsourcing junior employee tasks to AI will ultimately suffer the consequences. "Automating junior tasks can thus reduce long-term production when the cost savings it provides remain modest, precisely because it eliminates the first rung of the ladder," the document emphasizes. "A company that stops recruiting entry-level employees shifts the problem to its recruitment of experienced staff," and the effects could be felt within a decade. To avoid any unpleasant surprises, the study's authors suggest that employers, both public and private, "redefine entry-level positions around what AI does poorly, such as performance monitoring, customer relations, or making judgments on atypical cases, and maintain on-the-job training programs." Finally, these findings highlight the need to distinguish between two issues, often conflated in the public debate.