In 2024, 17.3% of French employees earn the minimum wage (SMIC) [4]. This rate, never reached since 1991, is explained by 39.4 billion euros in annual reductions on low wages [6]. These reductions make any wage increase costly for employers.

The decision to be made for 2027 is architectural: turning these exemptions into a temporary right tied to a sector-specific trajectory, rather than a permanent right tied to a wage level. In the first case, wage floor clustering continues. In the second, the State exits a logic it has maintained since the 1990s.

The 2024 Employment Record Masks Three Blind Spots

In 2024, 68.8% of 15-64 year-olds are employed, the highest level measured since 1975 [1]. This figure deserves to be broken down.

First blind spot: older workers. The employment rate of 55-64 year-olds reaches 60.4% in 2024, according to DARES, up 2 percentage points year-on-year [4]. The progress is real, but France ranks 17th out of 27 in the European Union.

The Netherlands reaches 75.3% in 2024. Sweden tops out at 78.1% [5]. The European average stands at 65.2% the same year [5]. The gap with France exceeds 4 percentage points above the European average, and nearly 18 percentage points below the best-performing country.

This lag has a precise origin. In 1980, the employment rate of 55-64 year-olds reached 58% [17]. It fell to 44% in the early 1990s, when early retirement policies tried to reduce unemployment through age-based measures. These policies did not create the jobs they claimed to free up, as documented in a Senate report from May 2025 [17]. The recovery since then has been slow, incomplete, and fragile.

Second blind spot: the quantity of work. In 2024, full-time employees worked 1,664 hours, one of the lowest levels in the European Union [16]. All employed persons worked 1,595 hours, compared to 1,667 hours in the EU [16].

On the criterion of employed persons, the French worked an average of 1,494 hours in 2024 [16]. Their German neighbors, 1,338 hours [16]. What France is missing compared to Europe is not the intensity of work among those employed, but the proportion of 55-64 year-olds working.

Third blind spot: wage quality. The share of workers earning the minimum wage rose from 12.0% in 2021 to 14.5% in 2022, then to 17.3% in 2023, meaning more than 3 million people paid at the minimum wage [4]. This rate exceeds the previous record of 16.3% recorded in 2005. Behind Greece and Portugal, France is one of the European countries where wages are most concentrated at the bottom of the scale [3].

The Low-Wage Trap Is a Mechanism Built by the State

Wage floor clustering has a precise institutional cause.

Since the 1990s, exemptions from social security contributions on low wages have created a trap. Above the minimum wage and up to 1.6 times the minimum wage, the exemptions are so large that raising a worker’s salary by 100 euros can cost the employer up to 500 euros, once the associated benefits are lost [7]. Keeping workers in this range becomes rational for the company. The Bozio-Wasmer report from France Stratégie, published in October 2024, concludes that these reductions have led to over-specialization in low-value-added jobs, at the expense of higher-productivity jobs [7].

The fiscal cost is massive. In 2024, the total exemptions from contributions reaches 88.9 billion euros [6]. General reductions on low wages alone represent 39.4 billion euros, compared to 30.3 billion in 2021 [6]. In three years, the State increased by 30% the amount it devotes to subsidizing low wages.

The logic of compensation thus finances the fiscal passivity of entire sectors whose survival depends on artificially cheap labor for employers [8]. Martinot and Morel document the perverse effects of this system on incentives for wage increases and early retirement [9]. Bruno Palier pushes the causal chain to the productive model: as long as French companies remain on a cost-competitiveness strategy, demand for low wages will be structural [10]. Reducing exemptions without a simultaneous upskilling strategy would weaken companies without improving careers [10][11].

AI Will First Strike the White-Collar Workers Who Finance Social Protection

To this structural diagnosis is added a technological rupture whose timing changes everything.

Coface and the Observatory of Threatened and Emerging Jobs published in March 2026 the most granular mapping available of French employment exposure to AI [13]. Result: 3.8% of jobs are currently vulnerable to generative AI. Over a two to five-year horizon, this rate rises to 16.3%, or roughly 5 million positions out of 30 million [13]. One in eight professionals holds a role where at least 30% of tasks are fully automatable by AI agents [13].

This shock first affects skilled intellectual work. White-collar workers pay more in taxes and social security contributions [12]. A shock concentrated on 16.3% of the labor market simultaneously creates both a jobs crisis and a social protection financing crisis [12].

This intersection is crucial to understanding why the 2027 decision cannot wait. The French social model is financed by contributions from the highest-paid workers. An AI that destroys these jobs first deprives the State of revenue at the precise moment when demand for protection increases [12].

Companies adopting AI see a relative decline in hiring in exposed professions. But jobs related to AI come with increased productivity, turnover, and total employment [15]. Creative destruction can work, but only in companies that invest in upskilling. Acemoglu, Autor, and Johnson identify in a February 2026 NBER Working Paper a major barrier: the incentives of companies and developers remain aligned with automation, not with expanding human judgment [14]. Only technologies creating new tasks are unambiguously favorable to workers [14].

Companies kept below 1.6 times the minimum wage by the exemption logic do not invest in these uses in 2024 [6]. They remain in the low-cost logic that the State subsidizes, with no reason to change as long as the subsidy continues.

Transforming Exemptions into Conditional Sector-Based Rights

The main decision for 2027 consists of transferring responsibility: the 39.4 billion euros in reductions targeting low wages stop being managed line by line by central administration and become a temporary right that sectors negotiate to maintain or renounce [6][11].

The mechanism works as follows. A sector opening genuine wage negotiations and committing to a standard agreement above the minimum wage maintains its exemptions during the transition [11]. A sector that does not loses them, gradually, over five years. The timeline for upskilling moves from the hands of the State to those of the sector. The hourly recalibration between 24 and 35 hours is the transitional mechanism of this sector-based architecture, detailed in the article “France Pays Twice for Imposed Part-Time Work.”

This shift of responsibility makes it mechanically impossible to couple State exemptions with wage freezes. Sectors that have locked their standard agreements for decades reopen negotiations or absorb their labor cost alone [8]. Sector-level social partners decide under outcome constraints, where central administration previously managed exemptions without trajectory conditions.

What this choice closes is visible. Low-productivity sectors without upskilling strategies will face bankruptcies and job losses [10]. This risk is real and documented.

The State accepts it in exchange for an inverse calculation. Remaining in the subsidized low-cost logic makes the fiscal crisis unsolvable if AI first destroys white-collar jobs that finance social protection [12]. And sectors kept at low wages do not invest in technological innovations that create new tasks [14]. The choice is therefore not between comfort and risk. It is between two risks, with two very different distributions of who bears them.

Sources

[1] INSEE, “A Snapshot of the Labor Market in 2024,” INSEE Première no. 2044, March 2025, https://www.insee.fr/fr/statistiques/8391807 (accessed 09/08/2026).

[2] INSEE, “Activity, Employment and Unemployment in 2024 and in Long-Term Series,” August 2024, https://www.insee.fr/fr/statistiques/8599938?sommaire=8578977 (accessed 09/08/2026).

[3] INSEE, “Employment Rate,” Long-Term Series, Eurostat Labour Force Survey, Extraction June 2024, https://www.insee.fr/fr/statistiques/3281596 (accessed 09/08/2026).

[4] DARES, “The Situation of the Labor Market in the 4th Quarter 2024,” February 2025, https://dares.travail-emploi.gouv.fr/sites/default/files/59606f3786704622a70c17be92952426/Dares_La_Situation_du_march%C3%A9_du_travail_T4_2024.pdf (accessed 09/08/2026).

[5] DARES / INSEE, Employment Survey 2024, Employment Rate of 55-64 Year-Olds, cited in Cercle de l’Épargne, September 2025, https://cercledelepargne.com/emploi-des-seniors-en-france-des-resultats-encourageants-mais-encore-insuffisants (accessed 09/08/2026).

[6] Commission for Social Security Accounts, May 2024, Data Reproduced by evaluation.securite-sociale.fr, “Exemptions Compensated and Non-Compensated by the State,” https://evaluation.securite-sociale.fr/home/financement/1.6.1.%20Exon%C3%A9rations%20compens%C3%A9es%20e.html (accessed 09/08/2026).

[7] Antoine Bozio and Étienne Wasmer, “Social Security Contribution Exemption Policies,” France Stratégie Report, October 2024, https://www.strategie-plan.gouv.fr/files/2024-11/rapport_vffff_241003.pdf (accessed 09/08/2026).

[8] FIPECO (François Ecalle), “Reductions in Employer Social Security Contributions on Low Wages,” https://www.fipeco.fr/fiche/Les-all%C3%A8gements-de-cotisations-sociales-patronales-sur-les-bas-salaires (accessed 09/08/2026).

[9] 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 09/08/2026).

[10] Bruno Palier (coord.), What Do We Know About Work?, Presses de Sciences Po, October 2023, https://www.sciencespo.fr/centre-etudes-europeennes/fr/actualites/que-sait-du-travail/ (accessed 09/08/2026).

[11] Bruno Palier and Christine Erhel, Working Better, PUF/Vie des idées, 2025, https://laviedesidees.fr/Palier-Bruno (accessed 09/08/2026).

[12] Axelle Arquié, “The Double Shock of AI: Employment and Taxation,” Political Economy, no. 110, 2026/2, https://shs.cairn.info/publications-de-axelle-arquie–111394 (accessed 09/08/2026).

[13] Coface / OEM (Observatory of Threatened and Emerging Jobs), Mapping of French Employment Exposure to AI, March 2026, https://anthemcreation.com/en/artificial-intelligence/ai-jobs-5-million-at-risk-france-coface-oem/ (accessed 09/08/2026).

[14] Daron Acemoglu, David Autor, and Simon Johnson, “Building Pro-Worker Artificial Intelligence,” NBER Working Paper no. 34854, February 2026, https://www.nber.org/papers/w34854 (accessed 09/08/2026).

[15] Philippe Aghion, Antonin Bergeaud, Simon Bunel, and Paul Delbouve, “Diffusion of Artificial Intelligence in France,” HEC Paris / Banque de France / Collège de France, July 2025, https://www.hec.edu/sites/default/files/documents/Diffusion%20de%20l%27Intelligence%20Artificielle%20en%20France.pdf (accessed 09/08/2026).

[16] Rexecode, “Actual Duration of Work and Its Quantity in France and Europe in 2024,” 2025, https://www.rexecode.fr/competitivite-croissance/documents-de-travail/la-duree-effective-du-travail-et-sa-quantite-en-france-et-en-europe-en-2024 (accessed 09/08/2026).

[17] Senate, Report “Impact of the Employment Rate of Older Workers on the Financial Balance of the Pension System,” no. 616, May 2025, https://www.senat.fr/rap/r24-616/r24-6162.html (accessed 09/08/2026).