In France, value sharing remains window dressing for many SMEs

Three years after the agreement that was supposed to generalize profit-sharing in all French small businesses, the CFDT qualifies the outcome as “globally catastrophic.” This verdict does not invalidate the ambition of the 2023 national inter-professional agreement, but it says something essential about the distance between a legal framework and actual practice. And it comes at the wrong time: the cognitive automation sweeping through the tertiary sector will produce massive productivity gains in the years to come. If the sharing mechanism is already failing in an ordinary context, it will be even more so when these gains are to be distributed.

The 2023 ANI imposed on companies with 11 to 49 employees showing at least 1% net profit for three consecutive fiscal years the obligation to open negotiations on value sharing. The measure potentially affected several hundred thousand employees not previously covered. The government follow-up committee established to evaluate implementation documents, in 2026, a heterogeneous reality: where social dialogue exists, mechanisms have been put in place. Where it did not exist, the law was not sufficient to create it.

The essentials

  • The 2023 ANI requires SMEs with 11 to 49 employees to negotiate starting at 1% net profit for three consecutive fiscal years, a deliberately low threshold to include the maximum number of profitable companies.
  • The CFDT qualifies the 2026 outcome as “globally catastrophic”: the mechanisms deployed often remain formal, with no verifiable link to actual productivity gains.
  • Profit-sharing, the dominant tool, leaves the employer full discretion over the calculation formula, making it possible to respect the letter of the agreement without respecting its spirit.
  • Cognitive automation will amplify this structural problem: without a verified sharing mechanism, AI gains in services will fuel capital concentration rather than wages.

What the 2023 ANI really changed

The February 2023 agreement was, on paper, a substantial step forward. Before it, the obligation to share value applied only to companies with more than 50 employees, the threshold above which legal profit participation is mandatory. In 1967, this participation became mandatory for companies with more than 100 employees; it wasn’t until 1990 that the threshold was lowered to 50 employees. Below that, nothing. Small structures could show profits for decades without ever associating their employees with the fruits of their growth.

The ANI significantly reduced this threshold. Profitable companies with 11 to 49 employees for three consecutive years had to, starting in 2025, implement at least one mechanism among four options: legal participation, profit-sharing, employer contributions to a savings plan, or a value-sharing bonus payment. The menu was wide, precisely to facilitate adoption. The negotiators’ idea was not to block SMEs on a particular technique, but to force them to the table.

This flexibility came at a price: it opened the door to symbolic mechanisms. A profit-sharing agreement whose calculation formula systematically yields zero complies with the law without costing a euro. A value-sharing bonus paid once below inflation does not constitute sustainable sharing. The legislature had bet that the obligation to negotiate would be sufficient to create momentum. This is what the 2026 outcome questions.

Why profit-sharing alone guarantees nothing

Profit-sharing is by far the mechanism most chosen by companies that have satisfied the obligation. Its appeal to managers is real: the amounts paid are exempt from employer social contributions, they create no wage precedent, and above all, the calculation formula is freely negotiated. In theory, this freedom allows the mechanism to be adapted to each company’s reality. In practice, it also makes it possible to construct a formula that mechanically caps payments at a very low level, or even zero.

The CFDT documents this phenomenon in its assessment. Agreements signed in SMEs without union delegates—that is, the majority of structures covered by the ANI—are often subject to validation by referendum. Yet in an SME of fifteen people where the manager is the daily interlocutor of his employees, negotiation is structurally asymmetrical. This is not bad faith: it is the reality of the balance of power in structures where collective labor law has never had concrete anchorage.

This structural imbalance explains why formal coverage rates have progressed without the substance of sharing following suit. Thousands of agreements have been signed, deposited, and administratively validated. But the central question—that of the link between the company’s actual productivity gains and what returns to employees—remains largely without satisfactory answer in small structures.

SMEs that do better than the legal requirement

The picture is not uniformly bleak. Where actors have invested in support, results differ. Professional federations in crafts, metallurgy, and personal services have developed sector-wide agreements that serve as turnkey models for SMEs in their sectors. These sector agreements, endorsed by social partners and adapted to the economic realities of each field, offer coherent calculation formulas and verifiable objectives.

Bpifrance, as part of its programs supporting SME managers, has integrated modules on value sharing into its training. The objective is not merely legal compliance: it is to show that well-constructed profit-sharing is a tool for retention and performance. Nicolas Dufourcq, chief executive officer of Bpifrance, has publicly defended this interpretation: value sharing, when it is real, reduces turnover and increases engagement. For an SME, this gain often exceeds the cost of the mechanism.

Companies like Materne, Lippi, and Chèque Déjeuner have long demonstrated that it is possible to construct ambitious profit-sharing agreements in structures of modest size. These examples exist, they are documented, and they illustrate what the legal framework can produce when a culture of dialogue precedes legal obligation. But they remain exceptions in a landscape of SMEs where such culture is in the minority.

The most promising avenue identified by the follow-up committee is the strengthening of on-site human resources consulting. Some regions have put in place social dialogue advisors to support SMEs in negotiation. Where this support exists, signed agreements are more substantive, formulas are more transparent, and employees are better informed of what they are entitled to.

The measurement problem that poisons everything

Behind the sharing question lurks a more technical problem, rarely discussed: that of measuring productivity in SMEs. In a large company, accounting data are available, audited, and accessible to employee representatives. In an SME of twenty employees, accounts are often presented by the manager at an annual meeting, without any real possibility of independent verification.

The law provides that employees or their representatives can access the accounting elements serving as the basis for calculating profit-sharing. But formal access to a document is different from the real capacity to understand and contest it. An employee of an SME without accounting training facing a tax return cannot verify whether the profit-sharing formula faithfully captures productivity gains or systematically underestimates them.

This information asymmetry is at the heart of the problem. It is not corrected by the 2023 ANI. It was not corrected by the 2001 Fillon law or by the 2019 PACTE law, which had already attempted to expand sharing. It is a structural limit of labor law in small structures, and it precedes the AI problem.

Why cognitive automation will worsen the problem

This context raises the following question: what will happen when productivity gains no longer come from human effort measured in hours, but from the automation of cognitive tasks?

Generative AI is beginning to modify workflows in tertiary SMEs, accounting firms, law offices, communications agencies, and consulting firms. These sectors concentrate a large share of employment in companies with 11 to 49 employees. The productivity gain expected from these tools is documented by several recent studies: McKinsey Global Institute estimated in 2023 that the automation of cognitive tasks in services could represent a significant portion of work volume in these sectors by 2030, depending on the pace of adoption. These projections must be read with caution: they describe tasks susceptible to automation, not jobs eliminated. But they give a measure of the magnitude of the potential for gains.

The Midjourney model illustrates to the extreme what AI enables in terms of concentrating gains in very few hands: the company had generated $50 million in revenue with just 11 employees in 2022, then $200 million in 2023 with approximately 40 to 50 employees. This case is extreme, but it mirrors less spectacular companies where two developers augmented by AI progressively replace ten consultants, or one accountant equipped with tools handles in an hour what once took a day.

In this scenario, the question of value sharing becomes both more urgent and more complex simultaneously. More urgent, because gains will be massive and concentrated on balance sheets without employees naturally seeing the trace. More complex, because the productivity gain linked to a software tool is even harder to measure and attribute than a gain linked to human work. How does one calculate the share of profit-sharing for an employee whose productivity has tripled thanks to a tool paid for by the company? None of the existing frameworks answer this question.

The tendency in American tech to concentrate gains in senior skills at the expense of juniors illustrates a similar dynamic: when technology amplifies the best, it does so first to the employer’s benefit, not the teams’. The sharing of these gains is not automatic; it is the result of a balance of power or company culture.

What the 2023 ANI missed and what can be built now

The CFDT’s diagnosis does not argue against the 2023 agreement. It argues for a second generation of obligations, better equipped to produce real rather than formal sharing.

Several avenues are on the table in discussions between social partners. The first is strengthening sector-wide agreements as the primary vector, by making recourse to a sectoral model mandatory in the absence of company-level social dialogue. This approach has demonstrated its effectiveness in metallurgy and in certain service sectors. It makes it possible to correct negotiation asymmetry in small structures by setting a floor that the manager cannot circumvent with an opaque formula.

The second avenue is the creation of a right to audit the profit-sharing formula, accessible via employee representatives or, in their absence, via a mandated external body. This measure exists in other European countries. In Germany, the works council has substantial economic information rights, including in medium-sized structures. This transparency is not secondary: it is what gives credibility to sharing.

The third avenue, the most forward-looking, consists of anticipating now the question of productivity augmented by AI. The debate on the four-day week in Iceland and other countries shows that real productivity gains can translate into freed time rather than increased wages. These two forms of sharing do not exclude each other, but they suppose different legal frameworks. If social partners wait for AI gains to become visible in balance sheets before negotiating, they will have accumulated several years of delay.

The objective is not to slow the adoption of tools that improve productivity. It is to construct, while there is still time, the mechanisms that will make it possible to verify that these gains do not evaporate into margins without touching wages or working time. A legal framework alone is not sufficient: the outcome of the 2023 ANI demonstrates this. But the absence of a framework guarantees capture. The question is which generation of social partners will take the initiative.

Sources

  1. CFDT / Partenaire Entreprise, “Value sharing: CFDT calls the agreement ‘globally catastrophic’ and breaks down its implementation” — https://partenaire-entreprise.com/partage-de-la-valeur-la-cfdt-qualifie-laccord-de-globalement-catastrophique-et-en-decortique-la-mise-en-oeuvre
  2. McKinsey Global Institute, “The economic potential of generative AI” (June 2023) — no link (report available on mckinsey.com)
  3. Midjourney / Journal d’un Progressiste, “Midjourney generates $200 million with 11 employees, the cognitive company model” — https://journaldunprogressiste.fr/midjourney-genere-200-millions-de-dollars-avec-11-salaries-le-modele-de-lentreprise-cognitive/
  4. Journal d’un Progressiste, “American tech cuts juniors and pays increasingly more for its seniors” — https://journaldunprogressiste.fr/la-tech-americaine-coupe-les-juniors-et-paie-de-plus-en-plus-cher-ses-seniors/
  5. Journal d’un Progressiste, “The Icelandic 4-day week did not prevent GDP growth” — https://journaldunprogressiste.fr/la-sermaine-de-4-jours-islandaise-na-pas-empeche-la-progression-du-pib/
  6. PACTE law (2019) and value-sharing law (2023) — texts available on legifrance.gouv.fr
  7. ANI of February 10, 2023 - primary text — https://revuefiduciaire.grouperf.com/plussurlenet/complements/20230213_ANIPartagedelavaleur-10022023.pdf
  8. Law n°2023-1107 of November 29, 2023 - Digital Labor Code — https://code.travail.gouv.fr/information/entreprises-de-11-a-49-salaries-mise-en-place-dun-dispositif-de-partage-de-la-valeur-participation-interessement-prime-de-partage-de-la-valeur-ppv
  9. URSSAF - social regime for profit-sharing — https://www.urssaf.fr/accueil/employeur/beneficier-exonerations/epargne-salariale/interessement.html
  10. DARES 2023 - union representation in companies — https://www.clesdusocial.com/instances-representatives-du-personnel-toujours-a-la-baisse-en-2023
  11. Value-sharing ANI Follow-up Committee - April 9, 2026 — https://www.syndicalismehebdo.fr/article/le-comite-de-suivi-de-lani-partage-de-la-valeur-se-reunit-pour-la-premiere-fois
  12. Les Clés du social - Ordinance 1967 on participation — https://www.clesdusocial.com/aout-1967-la-participation
  13. Midjourney statistics - SEO.AI / Quantumrun — https://seo.ai/blog/how-many-people-work-at-midjourney