In the United States, the cooperation agreement, the private contract by which a listed company grants an activist shareholder one or more board seats in exchange for a standstill and a period of contractual peace, has become the standard way of resolving activist campaigns. France sketched out the model as early as the 2000s. If the practice remains embryonic, this is not a matter of legal validity but of incentives. Absent a credible threat at the general meeting (assemblée générale), a genuine capacity to do harm there: to have a director elected against the board’s recommendation, to defeat its resolutions, or at least to compel it to account publicly to the shareholders, the minority shareholder’s fallback, litigation, does not bring boards to the table, the French courts being far too slow to weigh in before the meeting is held. The 2026 proxy season has confirmed it once again. The more demanding scrutiny the AMF now exercises over tender offers may, however, shift the locus of negotiation.
I. Across the Atlantic, settling has become the norm
2025 was a record year for activist campaigns in the United States, and activism is now a year-round activity, no longer confined to the annual-meeting season: campaigns are launched outside the director-nomination window (“off-cycle”) or simply urge shareholders to vote against the board’s nominees or resolutions (“vote no”). Methods are graduated: a campaign almost always opens in a friendly register — private contact, analyses submitted to management — and, failing that, turns public and then hostile: an open letter, a critical presentation to the market, the nomination of board candidates and, ultimately, the solicitation of proxies. It is at the hinge between these two registers, when the friendly approach threatens to turn hostile, that a cooperation agreement is struck.
The scenario is by now classic. An activist builds a stake, makes demands and threatens a proxy fight to elect its candidates; rather than face the vote, the board privately negotiates the appointment of one or more directors designated by the activist, in exchange for a standstill that caps its stake and bars further hostility for a defined period. This outcome has become dominant: more than 90% of the board seats activists obtain are now secured through negotiated agreements rather than at the ballot; in the first quarter of 2026 again, forty-one of the forty-five seats won came by this route. The tipping point is well identified — the entry into force, in September 2022, of the universal proxy rules, which let shareholders mix candidates on a single voting card and thus target each director individually. Since then boards settle at a record pace, even though management prevails in nearly all of the contests fought through to a vote. The lesson deserves emphasis: it is not the probability of losing that drives boards to settle, but the cost — reputational above all — that the battle imposes on each director taken individually.
Practice has standardised around three blocks of clauses: representation (one or two activist-designated directors, most often by expanding the board, conditional on the activist retaining a minimum shareholding); the standstill (ownership cap, no proxy solicitation, support for the board’s slate save where the proxy advisers recommend otherwise, pegged to the next nomination window); and the relationship (mutual non-disparagement, confidentiality, information-sharing). The recent L.B. Foster / 22NW, Stoneridge / 22NW and Tripadvisor / Starboard agreements illustrate the template. Two types recur: the action settlement, where the activist obtains an immediate measure — a distribution, a divestiture, a CEO change — and the far more frequent board settlement, where it obtains seats, that is, durable influence rather than a one-off concession. Empirically, these agreements draw positive stock-price reactions and are followed by the changes sought, with no evidence of rent extraction at the expense of other shareholders.
II. In France, early precedents left without progeny
France did not discover the instrument; it practised it before its American standardisation. The Valeo / Pardus agreement of May 2008 already combined every ingredient: a board seat for a partner of the fund; an ownership cap at 20% of the capital and voting rights; loyalty and conflict-of-interest undertakings; and the lapse of the entire agreement upon a third-party tender offer cleared by the regulator (offre déclarée conforme). Days later, the recomposition of Atos Origin’s supervisory board (conseil de surveillance) proceeded from an agreement between the company and its two main shareholders, Centaurus and Pardus, each obtaining a representative alongside seven independent members.
Nearly twenty years on, these two precedents remain the essence of the publicly available French corpus; to our knowledge, none comparable has been published since. The campaigns that followed were resolved otherwise: at Pernod Ricard, Elliott’s intervention (2018-2019) ended in unilateral company announcements — governance enhancements, a buyback programme — never acknowledging the fund’s influence; at Lagardère, Amber Capital’s campaign degenerated into a public confrontation prolonged before the courts. Validity is not the obstacle: such an agreement can be built under ordinary contract law (droit commun), subject to the classic precautions — consistency with the corporate interest (intérêt social), one-off and specifically defined voting undertakings, a framework for the information flowing to the fund-appointed director and, above all, management of the risk of concerted action (action de concert). Nor is cooperative activism economically sterile: European data show that disclosing such interventions creates value. The blockage lies elsewhere — in what fails to happen when negotiation breaks down.
III. The minority shareholder’s hollow threat
Part of the explanation lies in ownership structure. US ownership is markedly more dispersed than Europe’s; in France the concentration of capital in family hands has, if anything, increased. But ownership structure is more than a backdrop: it selects the very conflict the law is asked to govern. Where capital is dispersed, the agency conflict runs between managers and shareholders and control of the board is the prize — the natural terrain of the cooperation agreement. Where capital is concentrated, the conflict shifts to the controlling shareholder against the minority, and the minority’s lever shifts with it: no longer board composition, which the controlling bloc forecloses, but the policing of related-party dealings (conventions réglementées) and of the terms of exit.
The American settlement is born in the shadow of a credible threat: refuse to negotiate, and the activist can genuinely cost the board seats, quickly and publicly. In France that threat is hollow wherever a controlling shareholder holds the capital — and it is the controlled company, not the widely held one, that now produces the disputes. The general meeting still offers two levers, but a controlling bloc disarms both.
The first lever — having a director elected against the board — is, paradoxically, the one French law leaves most open. The appointment of a director falls squarely within the ordinary meeting’s own competence; directors are revocable ad nutum, at any time and without cause; French law ignores the staggered board, so the whole board may be renewed, or unseated, in a single meeting; and a shareholder may even put forward a candidacy by way of an amendment in the meeting itself, without having figured on the slate circulated beforehand.
A resolution to appoint or remove a director therefore collides with no power reserved to the board and clears the competence filter: it is included on the agenda. Vincent Bolloré’s progressive conquest of Vivendi followed precisely this channel — a board seat claimed as a minority shareholder, then methodically expanded into control. Yet the lever rarely delivers, for two reasons.
The first is arithmetic: where a controlling bloc holds close to 50%, the ballot is foreclosed in advance, and what defeats the resolution is not the agenda but the vote.
The second is sociological, and it persists even where the capital is dispersed: a dissident slate presupposes credible candidates willing to stand against a sitting board, and they are scarce. Lending one’s name to a contested slate carries a reputational cost — solidarity of place, the fear of foreclosing future mandates, simple aversion to public conflict — that deters precisely the experienced directors an activist would want. The individualised reputational exposure that, amplified by the universal proxy, drives American boards to settle has, in France, little to work on: few candidates will expose themselves, and the controlling bloc ensures that those who do cannot win.
The second lever — compelling the board to account, through advisory resolutions or mere discussion items — is throttled at the gate. The right to file is itself a hurdle: under Articles L. 225-105 and R. 225-71, it is reserved to shareholders holding a fraction of the capital that declines on a sliding scale — 5% where the capital is no greater than EUR 750,000, then 4%, 2.5%, 1% and 0.5% by successive tranches — a barrier the AMF has repeatedly proposed to lower.
The statute is mandatory in its terms — qualifying draft resolutions “shall” be included on the agenda — yet commentary and practice grant the board a review of legality extending to the competence of the general meeting: it was on this ground that TotalEnergies refused, in 2024, to include an advisory shareholder resolution on separating the offices of chairman and chief executive officer, a refusal a much-noticed legal opinion found consistent with French company law.
That precedent has just been extended — and hardened. In May 2026, a concert holding more than 5% of North Atlantic Energies (formerly named Esso) requested the inclusion of four draft resolutions and two discussion items (points). The board included — while urging a vote against — the two decision-making resolutions (the appointment of an additional director and the splitting, for a separate vote, of the resolution approving en bloc fifteen related-party agreements), but refused the two advisory resolutions, relating to compliance with IAS 24 (disclosure of related-party relationships and transactions) and IAS 36 (the impairment test that caps an asset’s carrying amount at its recoverable value), on the ground that they would exceed the meeting’s competence — a bare assertion that they trespass on the separation of powers, unsupported by any text, case law or reasoning; the effect of the refusal is to shield the statutory auditors from having to account to the shareholders, since the resolutions asked precisely that they pronounce on the accounts’ conformity with IAS 24 and IAS 36.
Above all, the North Atlantic Energies board refused even the two no-vote items, although the legal opinion that had supported TotalEnergies’ refusal in 2024 turned on a different criterion: the limit drawn from the meeting’s decision-making competence bites only where an express statutory provision confers the relevant power on the board — as it does for the choice whether to separate the offices of chairman and chief executive officer — so that, absent any such provision, a resolution or item escapes it. The competence filter, conceived as a legality exception, thus operates in practice as a discretionary board veto over the shareholders’ collective expression — with no ex ante third-party gatekeeper, where the exclusion of an American shareholder proposal requires, at the very least, going through the SEC.
A second factor compounds the structural one. French governance is organised around the board; but where capital is concentrated, this board primacy is less a free-standing model than the channel through which the controlling shareholder governs — and its roots, as comparative scholarship has stressed, lie in exogenous factors, the post-war structure of capital and the weight of the State, rather than in any civil-law tradition. Onto that structure French law grafts a competing normative standard. Where US corporate law rests on shareholder primacy, French law embraces a stakeholder conception: the company is run in its own corporate interest (intérêt social), which Article 1833 of the Civil Code, since the 2019 Pacte Law, requires to be pursued with regard to the social and environmental stakes of its activity, and which is distinct from the immediate interest of the shareholders. An activist invoking shareholder value thus meets not merely reticence but a normative standard readily invoked to legitimise — and, its critics argue, to entrench — board resistance.
The principle of separation and hierarchy of the corporate organs, enshrined in the Motte ruling, supplies boards with a comfortable doctrinal foundation for confining the meeting to a rubber-stamping role — one they have pressed into service against minority consultative resolutions, from the climate ballots to the IAS items at North Atlantic Energies. The reflex is rooted in a deeper cast of mind. French law knows no general-purpose advisory resolution of the kind the American shareholder-proposal regime makes routine: the statute organises a shareholder vote only where an express text provides for one, and the single place a non-binding vote might have taken root — the say on pay, transposed from the European directive — confirms the bias, for the directive expressly allowed a purely consultative ballot and the French legislature declined it, choosing a binding vote whose negative outcome bars payment of the variable and exceptional components of executive pay. From this, boards infer that whatever no text authorises may be kept off the agenda, treating the unenumerated as foreclosed rather than permitted-because-unprohibited. But that inference is the contestable step. The prevailing doctrinal view — articulated in the very opinion invoked to justify the TotalEnergies refusal — is the opposite: a consultative resolution is in principle licit, and may be excluded only where it trespasses on a power the law expressly reserves to the board. The foreclosure is thus a practice of boards, not a rule of law — which is precisely why the refusals at North Atlantic Energies are open to challenge. And shareholder dialogue in France is readily conducted in confidence, which may yield informal arrangements — quietly negotiated board appointments — but no visible contractual practice able to set precedents and become standardised.
Litigation, the supposed fallback, is a harder road than in the United States — but not a closed one. The shareholder’s ultimate weapon — an action to hold the directors personally liable for breach of their duties (action en responsabilité), under Article L. 225-251 of the Commercial Code, the closest French analogue to the breach of fiduciary duty — remains difficult to wield: successful precedents against the directors of a listed company are still few, proceedings are counted in years rather than weeks, and the courts have traditionally appraised management decisions with a deferential, business-friendly eye.
The management investigation (expertise de gestion), which the law offers minority shareholders to probe identified operations, is likewise construed narrowly. Yet the obstacle is as much one of legal culture as of black-letter law, and culture can shift. What it takes is patient pedagogy — persuading the courts and the wider ecosystem that protecting minority shareholders is not a private favour granted to a handful of activists but a matter of general interest: by the logic of communicating vessels, a market that secures fair treatment for its minorities lowers the cost of capital, deepens its investor base and ultimately serves the very companies and controlling shareholders it is said to constrain. As minority shareholders increasingly press that case before the courts, the fallback may, in time, acquire the bite it lacks today.
All told, the asymmetry is, for now, near-complete: the French issuer faces neither a likely proxy fight, nor courts swift enough to matter before the meeting, nor the discipline of a published precedent — and none of these forums yet imposes on the directors, individually, the reputational cost that drives American boards to settle. Little credible threat, and so little settlement. Yet the qualification matters: the threat is not structurally foreclosed but culturally underdeveloped, and the ground is shifting on both fronts — as the courts and the wider ecosystem come to treat the protection of minorities as a question of general interest rather than private grievance, and as the regulator, on the terrain examined next, shows that it can say no. It is an equilibrium that an organised minority can begin to move.
IV. Where the threat becomes credible again: the AMF’s review of tender offers
One terrain is the exception, and it is telling that recent developments concentrate there: the clearance review of tender offers (examen de conformité). The reason is structural — an offer followed by a squeeze-out results in the expropriation of the minority, and securities regulation accordingly surrounds it with reinforced safeguards: a multi-criteria valuation of the target, a fairness opinion (attestation d’équité) delivered by an independent expert appointed under Article 261-1 of the AMF General Regulation, and a clearance review by the AMF Board (collège) that is no mere registration. On 2 May 2025, in three parallel decisions reached on identical grounds after more than seven months of review, the AMF refused to clear the alternative buyout offers (offres publiques alternatives de retrait), each to be followed by a squeeze-out, that Bolloré SE had filed for its three listed holding companies — Compagnie du Cambodge, Financière Moncey and Société Industrielle et Financière de l’Artois. The terms had been raised in December 2024 and offered shareholders a cash-or-shares election, and an independent expert had attested their fairness; the refusal came nonetheless. The minority contended that the valuation, anchored in the stock-market prices of Bolloré and Compagnie de l’Odet, grossly understated the holding companies’ revalued net asset value (actif net réévalué), against a consistent line of regulatory decisions since 2002; Bolloré waived any appeal. Refusals of clearance are exceedingly rare, and the decision was immediately read as a landmark signal of heightened regulatory exigency towards going-private offers: clearance is not a rubber stamp, and the requirement that a delisting valuation genuinely integrate the value of the company’s assets, rather than rest on a depressed market price, recovers an effectiveness thought to have dulled.
It is precisely in such exit situations that the organised minority regains some negotiating leverage — no longer at the general meeting, but before the regulator. The temptation, then, is to make the regulator the whole of the strategy: with the meeting closed off at the outset and the courts too slow to matter, the AMF’s clearance review is the one forum that visibly bites, and a minority will naturally be drawn to stake everything on it. The pull is understandable, but it should be resisted as a settled posture: a leverage confined to the exit leaves the conduct of the company between offers wholly to the controlling bloc, and trades the contestation of governance for a single bet on a refusal of clearance that, the Bolloré decision notwithstanding, remains the rare exception. The outcome the AMF reaches on North Atlantic France’s offer for the shares of North Atlantic Energies (formerly Esso) — announced after North Atlantic France acquired ExxonMobil’s 82% controlling stake and now seeks to buy out the float and delist — will be a telling gauge of just how far the regulator can stand in for the forums the minority can no longer reach: whether the clearance review can carry, at the exit, the scrutiny the general meeting was prevented from exercising or whether Bolloré proves an isolated stand rather than the opening of a durable substitute. A clearance, or a refusal, will say which.
V. Conclusion: the settlement migrates from the board to the exit
One conclusion governs the rest: the cooperation agreement is the instrument of a contestable board, and a board is contestable above all where capital is dispersed. No reform can conjure it where a controlling shareholder commands the vote. France’s problem was never validity — Valeo proved in 2008 that such an agreement holds under ordinary law — but structure: the threat that brings a board to the table is electoral, and an electoral threat bites only where the outcome of the vote is genuinely open.
Hence two broad regimes. In the widely held company — less common in Paris than across the Atlantic, yet a substantial share of the market — the board settlement is in principle as available as in the United States; what it usually lacks is not a legal foundation but the ecosystem that gives the electoral threat its edge: influential proxy advisers, a market that reads each director’s score as a verdict, and a pool of credible candidates willing to brave a contested slate. In the controlled company — which generates the bulk of French disputes — it is, as a rule, out of reach: no minority can carry the vote, and the consultative route that might force a public reckoning is too often blocked before it can be put to a vote.
The minority is not, for that, powerless; its leverage has shifted. Where the controlling shareholder seeks the exit, the credible threat tends to re-form before the regulator — the AMF’s conformity review, whose teeth the Bolloré decision laid bare, can withhold clearance and restore net asset value as a floor. The French counterpart of the American board settlement is then less a bargain over seats than a negotiated exit — a higher price, governance or distribution undertakings, the withdrawal of pending challenges — struck in the shadow of a possible refusal. The mechanism is broadly constant; what shifts with the ownership structure is the forum. The minority’s task is to press its case where its capital structure leaves an opening: the ballot where capital is dispersed, the conformity review where it is controlled.