Author : admin1708

CAA Toulouse, 1ère ch., 18 décembre 2025, n°23TL03074

La décision rendue par la Cour administrative d’appel de Toulouse en décembre dernier mérite qu’on s’y attarde. Non seulement par son sujet — le transfert indirect de bénéfices entre sociétés d’un même groupe — mais surtout par les enseignements qu’elle délivre sur la manière dont l’administration fiscale (et les juges) appréhendent la notion de substance économique et de dépendance de fait.
Les faits ? Une société française de maçonnerie, AFG Rénovations, faisait appel depuis plusieurs années à un prestataire espagnol : la société M A… 2014 SLU, prétendument indépendante sur le plan capitalistique.
Mais à l’examen, cette indépendance s’est révélée largement fictive.

Les éléments matériels d’un lien de dépendance… de fait

Le fisc a dressé un constat accablant :

• Aucune activité réelle en Espagne : ni personnel, ni locaux, ni moyens matériels.
• Financement intégral par la société française.
• Factures rédigées en français, sans le moindre contrat de sous-traitance produit.
• Un gérant commun aux deux structures, présent sur tous les chantiers.

À cela s’ajoute un indicateur financier particulièrement révélateur : une marge commerciale de 0,52 % pour la société française, contre 39 % pour la société espagnole.
Le doute n’était plus permis : il existait un lien de dépendance de fait entre les deux entités, et les prestations facturées étaient surdimensionnées, révélant un transfert indirect de bénéfices de la France vers l’Espagne.

La réponse du juge : réintégration et rappel à la réalité

La Cour administrative d’appel valide sans détour la position de l’administration. Elle constate que l’ensemble des éléments de la relation entre les deux sociétés révèle une dépendance économique étroite, qui ne nécessite aucun lien capitalistique pour être caractérisée (conformément à la jurisprudence constante sur l’application de l’article 57 du CGI).
En conséquence, les charges facturées par la société espagnole ont été réintégrées dans le résultat imposable de la société française, considérées comme ne répondant pas à une gestion normale de l’entreprise.

Les enseignements pratiques : fiscalité et apparences

Cette affaire vient rappeler des principes essentiels du droit fiscal, trop souvent oubliés lors de la structuration internationale des groupes :

• La fiscalité s’apprécie entité par entité, pays par pays.
Il n’existe pas d’intérêt de groupe opposable à l’administration fiscale. Chaque société doit démontrer qu’elle agit dans son propre intérêt, de manière indépendante et économiquement cohérente.
• Le lien de dépendance ne se limite pas aux liens capitalistiques.
Des éléments de fait (moyens partagés, direction commune, dépendance financière, etc.) suffisent à établir une dépendance au sens fiscal.
• L’absence de substance est fatale. Une société qui facture sans avoir les moyens réels d’exécuter les prestations est automatiquement suspecte. La charge de la preuve peut alors basculer sur le contribuable.
• La cohérence entre la forme et le fond est indispensable. Des contrats, une traçabilité, des marges réalistes et une documentation rigoureuse sont les meilleures fondations pour résister à un contrôle.

En conclusion : un montage fiscal ne vaut que s’il tient debout

En matière de fiscalité internationale, la solidité d’une structuration ne repose pas sur un organigramme joliment dessiné, mais sur la réalité économique et juridique des opérations.
Le montage d’AFG Rénovations s’est effondré, non pas à cause d’une stratégie illégale, mais parce qu’il manquait de fondations réelles.

Médiane ou moyenne ? Quand la mode rencontre les prix de transfert.

Sans le savoir, les fiscalistes, particulièrement les experts en prix de transfert, et les adeptes de la mode partagent un point commun : tous admettent que rien ne vaut un bon basique.
Il est donc assez naturel que quand les deux domaines se mêlent, on obtient parfois des décisions jurisprudentielles qui, comme une robe haute couture taillée dans une toile brute, se transforment en pièces uniques : déroutantes, inattendues, mais parfaitement ajustées à l’air du temps.
Il en est ainsi de l’arrêt de la CAA de Paris passé inaperçu malgré lui dans l’affaire Roger Vivier Paris, et dont on s’apprête à célébrer le premier anniversaire. Ses multiples dispositifs, qui forment un camaïeu rare en prix de transfert, méritait pourtant bien un défilé !
En l’espèce, la société française Rogier Vivier Paris (RGP), dont l’activité consiste en la distribution de chaussures et d’articles de luxe de la marque éponyme rue du Faubourg Saint Honoré, a été contrôlé au titre des exercices clos en 2012 et 2014. L’administration fiscale, costumes austères et dont le géni créatif se limite à la fiscalité, a entendu redresser les résultats de la société sur différents motifs, mais dont nous retiendrons l’essentiel :

Dans ce secteur le marketing est important, mais est-ce à la société d’en supporter les coûts ?

La société RVP supportait des dépenses importantes de promotion et de valorisation de la marque Roger Vivier, sans pour autant en être la propriétaire juridique. L’administration rappelle sur ce point que l’absence de refacturation de ces dépenses au propriétaire légitime constitue une libéralité, ce qui n’est pas sans réveiller des souvenirs d’une affaire déjà rendue dans le même secteur d’activité (CAA Paris, Issey Miyake, 20PA03807). L’administration a considéré au cas d’espèce que ces flux sont à considérer comme des prestations de services, à facturer selon la méthode du coût majoré de 5%.

La confirmation de l’approche classique de marge nette

La société dégageait des marges d’exploitation négatives jusqu’en 2013, soit 10 ans après sa création. De manière classique, l’administration a considéré que, compte tenu de son incapacité à définir le mix-produits et à déterminer sa stratégie de vente, la société ne devait pas supporter le risque de volatilité du marché. Corrélativement, le fisc a apposé une marge nette issue d’un panel de comparables, à défaut pour la société d’avoir présenté une étude alternative. Cela permet de rappeler l’impérieuse nécessité pour les sociétés d’être préparées à toute éventualité, et a minima de disposer d’une analyse permettant de décrire les fonctions réellement assumées localement, afin de mettre en lumière les risques corrélatifs, et de proposer une analyse de benchmark robuste. Sans quoi, le champ est libre pour l’administration d’établir sa propre version des faits.

Un revirement discret de jurisprudence sur le recours à la médiane ?

Plus intéressant encore est la méthodologie statistique pratiquée par le service pour rectifier la marge nette de l’entreprise. Dans cette affaire, le fisc avait visiblement retenu le taux de marge moyen des comparables formant son panel, et non la médiane de l’intervalle comme on l’observe usuellement. Au cas d’espèce, ce point médian semblait plus bas, dès lors que les faits présentés dans l’arrêt indiquent que le contribuable demandait, à titre subsidiaire, de se référer à cette cible. La CAA de Paris rejette les prétentions du contribuable, en précisant en son considérant n°20 : « la société RVP n’apporte aucun élément de nature à établir que le service aurait dû appliquer le taux de marge nette médian résultant du panel des comparables, et non le taux de marge nette moyen. Au demeurant, ni les Principes de l’OCDE applicables aux prix de transfert à l’intention des entreprises multinationales et des administrations publiques, dans leur version de 2022, notamment ceux énoncés au point 3.62 en cas d’application de la méthode transactionnelle de la marge nette, ni les commentaires de l’administration n’impliquent qu’une telle médiane soit appliquée. ». Il faut y lire une petite et discrète révolution, tant la référence à la médiane des intervalles est devenue une loi d’airain depuis la décision du Conseil d’Etat GE Healthcare (CE, 8ème et 3ème chambres réunies, 06 juin 2018, 409645). Pour autant, nous n’avons eu de cesse d’affirmer que ce dispositif était réemployé à tort par l’administration et procédait d’une lecture arbitraire et simpliste de cette décision. La CAA de Paris tend à nous donner raison, ce qui est heureux.

La validation implicite des phases de démarrage d’activité, véritable période de grâce en prix de transfert

Un autre considérant discret s’est glissé aussi dans la décision. On y relève que le fisc n’a pas entendu contrôler les déficits dégagés au titre des 3 premiers exercices de la société (soit, de 2003 à 2006), en considérant que RVP était alors, à ce moment-là, en phase de démarrage. Ce faisant, le service valide implicitement une tendance déjà observée par les praticiens, consistant à prétendre qu’une société, même faisant partie d’un groupe et dont l’activité est dictée depuis l’étranger, traverse une phase incompressible de stabilisation, qu’on estime arbitrairement à 3 ans, pendant laquelle la politique de prix de transfert peut être balbutiante. Ce constat est d’autant plus intéressant qu’au cas d’espèce, le service a considéré que le contribuable avait fait acte de mauvaise foi et s’est donc vu appliqué les pénalités correspondantes visées à l’article 1729 du CGI. On en déduit alors que même lorsque ladite politique de prix de transfert s’avère visiblement mal taillée, le contribuable peut bénéficier d’une période de grâce, au cours de laquelle même le fisc admet que le groupe tâtonne, comme on cherche le design parfait.

La réaffirmation de la capacité de contrôler et , au final, rectifier, des déficits prescrits
S’il fallait encore le réaffirmer, la CAA de Paris rappelle que le fisc peut tout-à-fait remettre en cause les déficits réalisés au cours d’exercices prescrits, dès lors qu’ils sont reportés sur le premier exercice vérifié. La règle, déjà ancienne, rappelle en outre que ces déficits n’ont même pas besoin d’avoir été imputés, seuls leur rémanence sur un exercice couvert par le délai de reprise suffit.

La pénalité de 40% pour manquement délibéré de plus en plus facilement confortée
Enfin, il est intéressant de noter la légèreté, devenue malheureusement usuelle, avec laquelle le juge de l’impôt valide l’application de la pénalité visée à l’article 1729 du CGI (40% de pénalités sur les impôts en cas de manquement délibéré). De manière lacunaire, la CAA souligne ainsi que « l’administration fiscale, qui se prévaut de la nature, de l’importance et de la répétition de ces insuffisances déclaratives, ainsi que des circonstances que la société RVP, qui ne pouvait pas ignorer les liens de dépendance l’unissant aux sociétés, n’avait établi aucune documentation des prix de transfert et s’est abstenue de réaliser des ajustements permettant de corriger des résultats structurellement dégradés, doit être regardée comme apportant la preuve, qui lui incombe, […] de l’intention délibérée de cette société de se soustraire à l’impôt dû ». Compte tenu des exercices en cause (2012 à 2014), l’obligation documentaire applicable à l’époque n’obligeait, de jure le contribuable à produire une analyse économique (benchmark). En dépit de cela, le dispositif nous rappelle que l’anticipation, par la production d’une documentation des prix de transfert, permet a minima de montrer à l’administration la volonté d’agir en transparence. Une documentation contemporaine et exhaustive se pose ainsi comme un gage parmi d’autres de bonne foi, ce qui peut concourir à éviter l’application de plus en plus systématique des pénalités de 40%.

Les prix de transfert, un art sur mesure

En matière de prix de transfert comme en haute couture, ce sont souvent les détails qui font toute la différence. L’affaire Roger Vivier Paris en est une démonstration éclatante : un assemblage méticuleux de considérants discrets, de coupes audacieuses et d’ajustements techniques, aboutissant à une pièce jurisprudentielle rare, presque avant-gardiste.
À l’heure où l’administration fiscale peaufine ses modèles et où les contribuables tâtonnent encore dans la confection de leur documentation, cet arrêt rappelle une vérité simple mais souvent négligée : mieux vaut une tenue bien ajustée qu’un effet de style improvisé.
Alors, à ceux qui manient la fiscalité avec autant de rigueur que de flair, une question s’impose : et vous, votre documentation prix de transfert, est-elle prête à défiler sous les projecteurs d’un contrôle ?

President Trump's tariff announcements have shaken the economic and financial world, casting doubt on tariff strategies and provoking a bidding war among the world's major economic regions. You don't need to be an expert in international taxation or geopolitics (for let's make no mistake, this is geopolitics we're talking about) to understand the consequences induced by the risks carried by these customs reforms: a company wishing to import products into the American market will now have to find sources of savings to compensate for the loss linked to the increased share left to Uncle Sam. Yet this potential source of savings is mechanically caught in a simple vice, between the final selling price dictated by the market on the one hand, and production and operating costs on the other, both parameters being relatively inelastic.

For groups of companies established on both sides of the US border, the temptation to play on transfer prices is therefore great.

Indeed, by selling products to a related company at a price reduced by the portion equal to the customs duty overtaxation, a group could, in theory, ensure that its final selling prices remain virtually unchanged, and thus safeguard its outlets and market shares. But like Ulysses, who had to maneuver skilfully between Charybdis and Silla, steering from one tax area to another could lead to shipwreck. By trying too hard to steer clear of the customs risk, these companies could be exposing themselves to heavy and severe consequences in terms of corporation tax, and to penalties as onerous as they are infamous, on the grounds that their transfer prices would thus be voluntarily and consciously reduced to nothing.

So what to do? All eyes turn to tax specialists and advisors. However, just as Ulysses didn't have Pythia on board his ship, tax specialists don't have a crystal ball in their toolkit. So we have to rely on a pragmatic as well as empirical approach, which we believe is based on concrete realities that we share with you below.

Transfer prices must reflect the practices of independent companies in comparable situations.

By way of introduction, you'll forgive us if we reiterate what is a truism for tax specialists: transfer prices must necessarily correspond to market standards, which are referred to as "arm's length". By this terminology, which is less abstruse than its original etymology in English ("arm's length principle"), we mean that intra-group transactions must be remunerated in line with practices observed between economic players who are necessarily independent of each other, functionally comparable, and placed in a similar or close economic environment.

Returning to our exports to the United States between two related parties, we need to know whether a non-American company would be prepared to drastically reduce its selling prices to its American third-party customer in an almost identical environment, in order for the latter to mechanically reduce the base subject to customs duties, and offer an iso final selling price. Indeed, this may well be the case, given that the long-term pursuit of economic activity for market players takes precedence over short-term profitability. But this assertion comes up against two obstacles:

Firstly, the lack of contemporary references. It is certainly conceivable that independent companies might agree to lower their prices in order to reduce the customs duty base. After all, these companies are constrained in the same way by the Trump reforms as business groups. Since the same causes produce the same effects, an identity of behavior between third-party players and groups of companies could be observed. But to assert is not to demonstrate, and it has to be said that the pricing practices of economic players are not available online, nor would they be precise. Only much later, when companies' fiscal years have been completed and declared, will it be possible at best to extract market trends from specialized databases. By then, however, it will probably already be too late.

Secondly, and independently of access to information over time, some markets simply have no independent comparables. Many economic sectors, such as the automotive, aerospace, defense and pharmaceutical industries, are concentrated and operated exclusively by groups of related companies, whose initial investment (capex) is too high a barrier to entry to be overcome alone. For these sectors, waiting will never be rewarded, and it will remain impossible to demonstrate ex post that independent, comparable companies in similar situations have lowered their transfer prices as a result of higher tariffs.

Fortunately, while comparability analysis is the totem of the arm's length principle, it is not the only one. When the arm's length principle comes up against the impossibility of finding comparables, the OECD has, as usual, adapted. This is when the concept of "realistically availlable options" arises, like a tax deus ex machina, which the OECD's 2022 guidelines (cf. § 1.122 to 1.125) define as the approach of assessing whether the intra-group transaction is compatible with what an independent enterprise, acting rationally, would have reasonably accepted given its economic options.

So we're no longer comparing what is, but what might have been - in this case, what two unrelated companies, acting in their own interests, would have accepted in similar circumstances. The premise is attractive. But it is fundamentally based on counter-factual reasoning: we have to imagine the alternatives that would have been reasonably envisaged by each of the parties. Such mental modeling presupposes that we can document the economic constraints of the market, the asymmetries of information, and the alternative strategies left out - in short, that we can make history speak for itself.

This concept is not purely theoretical: it is increasingly used in tax audits to reject a related transaction deemed abnormal, or to reinterpret the group's contractual choice. This is what the French tax authorities did in the Sté Issey Miyake case (CAA Paris, June 29, 2022, no. 20PA03807), when they justified a tax reassessment on the basis of a net margin method deemed more realistic than the distribution contract applied. The French administration saw this as a better expression of the conditions that an independent company would have been able to negotiate - precisely with regard to the realistic options available. In a similar vein, in Socar Trading SA (Case C-282/22, March 7, 2024), the CJEU upheld a policy of intra-group re-invoicing, on the grounds that no independent company would have accepted such a level of risk without proportionate remuneration.

The available realistic option is therefore the tool by which the tax authorities, in a speculative tax exercise, judge after the fact what a company should have done before. As far as we are concerned, the idea is appealing: it aims to correct the effects of asymmetric information or over-skilful tax planning. But its implementation remains tricky, and is often criticized.

Lessons from the COVID period

Fortunately, once the emotion has subsided, history can sometimes teach us a few lessons. We could draw a parallel between the current situation and the COVID period, during which business groups also had to adapt their business models - and therefore their intra-group transactions - to cope with government measures affecting their value chains.

At first glance, however, nothing seems to reconcile the Trump administration's geopolitical tour de force and punitive tariffs with the Covid-19 global pandemic. One is a conscious political choice, the other a global health disaster. And yet, for tax specialists well-versed in functional gymnastics, these two events share a common consequence: they are forcing multinational companies to revise their transfer pricing policies, in a hurry, outside the confines of comparability analyses.

In both cases, functions that had previously been relatively unexposed are suddenly exposed to increased risks (storage, logistics, shortage management for the COVID period; market, financial and storage risks for the customs crisis).

In both cases, groups are urged to reconfigure their functional analysis.
In both cases, there is a sudden, exogenous intrusion of a public player (sovereign state or health authority) into the value chain. Suddenly, the allocation of functions, risks and assets is no longer intangible. The margins of local entities shrink mechanically, not because their function has changed, but because the world around them has changed.

Except that French companies now enjoy a jurisprudential advantage: the reductive and arbitrary concept of limited-risk entity has meanwhile been shattered by the enlightened impetus of the Conseil d'Etat, in the SKF and (RKS CE 8 e -3 e ch. 4-10-2021 n° 443133, SAS SKF Holding France and n° 443130, SAS RKS) rulings.
We therefore believe that if the functional analysis shows that the company exporting to the United States bears and controls the market risk, the financial risk and the customer risk (of non-collection), then it should be able to decide to reduce its sales prices to the importer, without exposing itself to transfer pricing risks, based on two complementary arguments:

Firstly, it is in the company's own intrinsic interest to avoid the aforementioned risks. If lowering its sales prices (and therefore its transfer prices) enables it to secure its outlets, even at the risk of making losses, then the durability of its operations and the commercial nature of its activities (as much as its corporate form!) dictate that it should do so. Basically, the praetorian concept of "acte anormal de gestion" (abnormal act of management), which is in the DNA of our French conception of the arm's length principle (all the more so now that the idea of "available realistic option" exists), is based more on economic common sense than on tax orthodoxy. It does not require companies to pay taxes, but to conduct their economic activities rationally. So, faced with the risk of being audited and redressed on transfer pricing grounds, or losing markets and outlets, a company's "normal" choice should be to continue doing business, even if this means lowering the price of its intra-group flows.

However, this is only relevant if the company in question has the material, human and financial resources to control the risks and thus make these choices. If the decision to reduce sales prices is imposed on the company (by the group's parent company, for example), we feel that the resulting losses should not be its responsibility.

Secondly, and for these companies alone, taking a tax risk weighing on its transfer prices, however significant and strategic, can no longer be qualified as "excessive" since the judge censured this concept (CE, July 13, 2016, no 375801, Monte Paschi). Faced with the threat of a significant increase in customs duties, a related company should therefore be able to take the risk of claiming that it is realistic that an independent company would have followed the same course, even if it means being contradicted later, without being opposed for reckless conduct.

Is the answer in the contracts?

If we are still uncertain about taxation, perhaps this is a sign that the answer lies elsewhere? As a good tax expert is first and foremost a lawyer who likes numbers, he might be tempted to reason in terms of contract law, which is still one of the last cornerstones of our written law edifice. Contracts between two companies involved in an import-export transaction involving the United States may indeed have provided for the consequences of fortuitous and external events on their rights and obligations, including the remuneration of flows. There are two main types of clause: the "hardship" clause and the "force majeure" clause.

A quick reminder for those of us who prefer numbers. A hardship clause is a contractual provision enabling the parties to renegotiate the terms of a contract when unforeseeable circumstances beyond their control arise, making performance of the contract excessively onerous for one of them. The UNIDROIT Principles of International Commercial Contracts define hardship in article 6.2.2 as the occurrence of events fundamentally altering the balance of benefits, allowing the injured party to request renegotiation. Unlike force majeure, which suspends or terminates contractual obligations in the event of impossibility of performance, the hardship clause aims to maintain the contract in force by adapting its conditions to re-establish the initial balance of benefits.

In French law, it is interesting to note that this renegotiation option is now enshrined in Article 1195 of the Civil Code, since the 2016 reform of contract law: "If a change in circumstances occurs which could not have been foreseen when the contract was concluded, making performance excessively onerous for one party [...], that party may request renegotiation of the contract [...]". This provision, which is suppletive in nature, can be modified or set aside by the parties via a conventional hardship clause. In practice, therefore, many commercial contracts are inspired by the UNIDROIT Principles (art. 6.2.2 to 6.2.3) or the ICC Model Clauses, which define hardship as a contractual imbalance resulting from a fundamental event that is external, unforeseeable and beyond the control of the party invoking the clause.

In this sense, the Trump administration's customs reforms can be seen as unforeseeable events for many companies engaged in long-term international contracts. Indeed, these measures will lead to a significant increase in contract performance costs for importers, without them having been able to anticipate or control these changes.
In this context, hardship clauses could be invoked to request a renegotiation of contractual terms, particularly with regard to prices or delivery times, in order to re-establish the initial economic equilibrium of the contract. In the wake of the COVID pandemic, the International Chamber of Commerce (ICC) updated its standard force majeure and hardship clauses in 2020 to help companies cope with this type of economic upheaval.

To make the link with transfer pricing, we believe that the existence of a hardship clause may therefore strengthen the justification for an exceptional adjustment to the intra-group margin or price, at least temporarily, pending further clarification from the tax authorities. Indeed, if an entity is disproportionately impacted by an unforeseeable customs reform, an argument could be made that the price change is the result of a contractual renegotiation based on a hardship clause, and not of a desire to evade tax.

Having said this, it is imperative to stress that the implementation of a hardship clause does not automatically suspend the contract. It merely opens the door to a request for renegotiation. Moreover, the rules governing international contracts vary according to the applicable law. In the United States, for example, the doctrine of commercial impracticability of the Uniform Commercial Code (§ 2-615) could theoretically apply, but it is interpreted more restrictively than in continental law. Hardship clauses are less common, and American judges favor contractual stability, except in cases of absolute impossibility. In an attempt to avoid these pitfalls, the International Chamber of Commerce (ICC) recommends, in its 2020 model clause, explicit wording to the effect that unilateral tariff changes (e.g., "unforeseen economic sanctions or customs duties") may constitute a case of hardship justifying renegotiation. This clause is now commonplace in cross-border supply contracts.

However, if the agreement was not drafted in this way prior to the Trump administration's announcement of the customs reforms, then the parties are obviously at a loss, and the ex-post revision of their contract could amount to an abuse of rights. What's more, our experience also suggests that hardship clauses are rarely included in intra-group agreements. We'll have to wait for the next crisis; we learn from experience and failure.

Without going into too much detail, it seems to us that an alternative argument based on force majeure clauses would also be doomed to failure. A force majeure clause allows a party to a contract to suspend or be released from its obligations when an unforeseeable, irresistible and external event prevents it from fulfilling its commitments. In principle, it applies when performance becomes impossible, rather than simply more difficult or costly. In French law, force majeure is defined in article 1218 of the Civil Code (2016 reform): "There is force majeure in contractual matters when an event beyond the debtor's control, which could not reasonably have been foreseen when the contract was concluded and whose effects cannot be avoided by appropriate measures, prevents the debtor from performing his obligation". Similar wording can be found in article 7.1.7 of the UNIDROIT Principles, as well as in the ICC Rules (2020), which provide a standard clause on force majeure, including an indicative list of events (natural disasters, riots, strikes, embargoes, governmental acts, etc.).
The imposition of new customs duties, however brutal, does not, in our humble opinion as tax specialists, prevent the performance of a contract. It simply makes it more costly. The courts - in France as in most legal systems - are reluctant to admit that foreseeable or absorbable economic or regulatory measures can constitute force majeure. Indeed, the Conseil d'Etat has already ruled that an increase in tax or customs charges does not constitute force majeure, unless it can be shown to have the effect of annihilating the contractual obligation (CE, March 19, 1986, no. 49782). In practice, even COVID-19 decisions have only accepted force majeure in cases of total blockage (administrative closures, confinement, etc.). In addition, the ICC, in its standard clause 2020, also cites "acts of public authority affecting the performance of the contract" as falling within the scope of force majeure, but only on condition that they render the obligation impossible, and not simply more costly.

What if this crisis was an opportunity?

A French tax expert is necessarily an eternal optimist. Otherwise, the often absurd excess of tax regulations and our all-encompassing over-taxation would long ago have had the better of us. We can console ourselves with the belief that crises have a virtuous side: they reveal the plasticity of international tax law.

The current period, marked by the return of protectionism and the assertiveness of national economic policies, reminds practitioners that balances are never set in stone. The tug-of-war between customs logic and transfer pricing will not be resolved by a clash of doctrines, but by a detailed understanding of value chains, the functions performed and the risks assumed - all of which must be skilfully conveyed in a demonstration combining tax techniques, economic projections and legal argumentation.

Above all, this sequence teaches us that legal and fiscal rationality, when solidly demonstrated, remains the best defense against uncertainty. Rather than dreading adjustments, groups would do well to reinforce the contractual robustness of their flows, by integrating realistic adaptation clauses from the outset and rigorously documenting their decision-making.

As geopolitical tensions continue unabated, and governments deploy their tax instruments for strategic purposes, the arm's length principle must also evolve - not by becoming weaker, as the developments of Pillars I and II suggest, but by coming to terms with the realities on the ground. What if, in the future, true competition no longer lay in the quest for the best rate, but in the ability of groups to anticipate and justify their risks?

Anyone who has ever dealt with the National and International Audit Department (DVNI) knows that the inspectors who make up its ranks are meticulous, shrewd, and devious. They are also revealed to be vengeful and resourceful in the ruling handed down by the Orléans Administrative Court, which offers an unexpected twist to the legal saga pitting the ST Microelectronics group against the tax authorities.

For several years now, several subsidiaries of the eponymous group have been subject to tax adjustments conducted under Article 57 of the French General Tax Code (CGI), which until now have all been annulled by the tax judges of the various jurisdictions to which the controlled companies belonged. In all these cases, the facts were identical: French companies in the group engaged in research and development (R&D) activities on behalf of foreign affiliated companies. As such, they paid for their services using the widely proven transfer pricing method known as "cost plus" according to OECD terminology, consisting of re-invoicing the costs incurred by the services, plus a margin, in most cases equal to 7%. Since the French companies were eligible for the research tax credit (CIR), the cost base on which the 7% margin was based was, however, reduced by the amount of the CIR and the subsidies received. During successive audits, the tax authorities challenged this transfer pricing policy, considering that the deduction of the CIR and other subsidies automatically led to sharing the benefit with foreign related entities and therefore to indirectly transferring profits outside France. This strategy was, however, doomed to failure, since it had already been decided and commented on in the past.

Do the same causes necessarily produce the same effects?

In this saga, it was the Montreuil Administrative Court (TA) that first annulled the corrections. In support of its decision, it stated that "the deduction made by a French company, for the determination of the transfer price of the product of its research to be invoiced to a foreign company related to it […], of the subsidies it had received from the State for the financing of the corresponding projects, cannot be considered as allowing, in itself and independently of the level of the transfer price to which this deduction leads by application of the contractual calculation method, to presume the existence of a transfer of profits abroad, within the meaning of Article 57 of the General Tax Code, with the burden on the French company to establish the existence of a consideration."

In doing so, the judge explains to us that the mere fact of deducting the subsidies from the basis of the costs re-invoiced intra-group is not, as such, demonstrative of a liberality granted to the foreign party. In this case, he shifts the subject to the point of knowing whether independent and comparable companies, placed in an equivalent situation, would also have made such a deduction. The recital then becomes unequivocal: "it does not result from the investigation that the panel of companies on the basis of which these [administration's] rates were determined, […] would have made it possible to identify relevant terms of comparison, having regard in particular to the activities carried out, the existence or not of links of dependency, the invoicing of prices net or gross of subsidies as well as the turnover and size of these companies. Having failed to present terms enabling a valid comparison of the prices charged by the company and those charged between independent companies or to propose another alternative method which could replace this comparison, the administration does not provide any evidence which would give rise to a presumption of the existence of the transfer of profits which it invokes.

As surprising as this solution may seem, the Montreuil Administrative Court has not actually invented anything new. It has simply adopted the provisions of the Sté Philipps France ruling, which had already caused confusion, as much as it had raised some hopes of clarification by the Council of State.

In this judgment, the high court had refused to rule on the point, considering that failing to demonstrate that independent companies deducted (or not) the amount of the CIR from their increased cost base, the tax authorities could not consider that the actions of Philipps France constituted an advantage indirectly transferred by way of transfer pricing. Faced with the doubtful reflections and the strategic issue raised by the case in question, we would have liked the judge to rule on the merits, and to position himself on the legitimacy of this deductive approach, obviously to the advantage of related parties located abroad.
He could have considered the interests of French society; he preferred to reason in terms of the behavior of independent companies, thus, in our opinion, widening the gap a little further between the two theories, which are nevertheless sisters, of the abnormal act of management and the full competition price.

However, during the appeal, the public rapporteur Bruno Coudert seemed to be subtly inviting the ministry to revise its copy in order to convince the supreme judge at the subsequent stage of cassation. In inviting the Court to censure the administration's argument, the public rapporteur emphasized that "the administration's reasoning consists, it seems to us, of starting from the principle that a company which is not the owner of the industrial and/or intellectual property rights developed does not deduct the subsidies it receives for its research activity. But this seems to us to be a begging of the question with which you [the Court] will not be able to be satisfied."

It was therefore necessary to push the reflection further, and probably further develop the axiom around the fundamental interest of "independent and normally operated companies": would they give away the CIR to their partners for the sole reason of gaining market share, even if it meant endangering their financial situation and exposing themselves to tax audits on the CIR and the IS? The debate deserved to be launched, why not in the light of the theory of manifestly excessive risk.

However, although the Minister's argument before the Council of State has evolved, it has come up against the difficulty of the exercise, which is almost impossible to prove since the corrections were placed under the visa of Article 57 of the CGI. In his conclusions, the public rapporteur Romain Victor explains this with great pedagogy and humility. On the thorny issue of the deductibility of subsidies received, he begins by acknowledging that: "the OECD Guidelines applicable to transfer pricing do not help us much." He adds – and one senses a kind of disappointment – that "the administration, which bears the burden of proof, was in any case stumbling over the next step and failing to base the implementation of Article 57 on a sufficiently solid comparability analysis."

So, is transfer pricing just a matter of comparison?

This question then immediately arises, but must, in our opinion, be immediately dismissed. We are indeed among those who strongly believe that the subject of transfer pricing cannot be limited to a simple comparison report, at the risk of becoming an inert and, let's face it, deeply boring discipline. Before entering into quantitative analysis, it is therefore imperative to first estimate the purpose of the transaction, as well as the interests of the parties. To do otherwise would otherwise render meaningless a cardinal formula of our tax edifice, set in stone, according to which "the only deductible expenses are those incurred - or the shortfalls incurred - in the interest of the operation." Public rapporteur Pierre-François Racine will not deny this, and confirmed long ago that "a company (...) has as its purpose the pursuit and sharing of profits. Any action it performs to achieve this purpose is presumed to be carried out in its own interest."

The analysis of the company's interest is therefore a prerequisite and must be imposed before even looking for comparable companies against which to benchmark. The same applies to the very purpose of the transaction, which, without going into detail here on the civil law foundations of this principle, is essential to make any transaction legitimate, enforceable and enforceable in relation to the related parties.

It therefore seems curious that the tax judge did not attempt to determine whether the deduction of the CIR and appropriate subsidies from the cost base re-invoiced with a margin met these imperative conditions. By focusing exclusively on the comparability analysis, he instead adopted a dynamic inspired by OECD principles, even if it meant exempting himself from domestic law. Because, as we have had the opportunity to point out, these two sources "diverge on one essential point. While the arm's length principle establishes the comparability analysis as a cardinal point, this approach is only provided for by Article 57 as a subsidiary measure." This judgment therefore confirms an observation we made, observing that "over time, and through repetition, the arm's length principle has nevertheless influenced our positive law, and from a subsidiary concept, comparative analysis has become an alternative method. From now on, to support its corrections under the visa of article 57 of the CGI, the administration must demonstrate either an advantage by nature or an advantage by comparison.

The comparability analysis therefore indeed governed the judge's decision, both in the Philipps ruling and in its successors rendered in favor of the ST Microelectronics group. But comparing is not quantifying, and the repeated approach of the tax judge tends to indicate that neither the margin added to the cost base, nor the amount of the subsidies that burdened it were the subject. The question rather concerned whether independent and comparable companies would also have deducted these subsidies, regardless of their volume or quantum. This is evidenced by the fact that, in the ruling handed down by the Montreuil Administrative Court, the administration claimed a higher margin, resulting from an alternative analysis. Its considerations clearly state that "if the administration claims that the margin rate resulting from a situation of full competition was 12.66 for the year 2009 and 11.09 for the year 2010 [compared to 7%], it does not follow from the investigation that the panel of companies on the basis of which these rates were determined, […] would have made it possible to identify relevant terms of comparison, having regard […] to the invoicing of prices net or gross of subsidies".

The comparability analysis put forward by the judge in these decisions is therefore in reality confused with the analysis of the behaviors and choices made, or would have made, by independent companies. This is the concept of "realistic available option" developed by the OECD, and which is anchored to the arm's length principle. In its chapter dedicated to this concept, the OECD states in this regard that "All methods based on the arm's length principle are linked to the idea that independent companies examine the different realistic options available to them and, in comparing these options, take into account all the differences that have an impact on the respective value of these options."

Faced with this eminently subjective study and this almost schizophrenic reflection, the tax administration could only fail. It is indeed impossible for it to report such a demonstration. It is already forbidden to interfere in the management of companies, so how could it have known the choices?

Refusal to choose the advantage by nature

To avoid this necessarily subjective and doomed task, the administration could then have been tempted to invoke the advantage by nature, rather than by comparison. Unlike its neighboring concept, the advantage by nature does not suffer from the implacable need to demonstrate the behavior of the parties, or to enter into quantitative reasoning. However, by refusing to re-invoice the amount of the subsidies to the related party, ST Microelectronics mechanically allowed its co-contractor to benefit from the advantage from which it itself benefited. From a strictly semantic point of view, it passed on (or "transferred" to use tax verbiage) an advantage (the subsidy). The failure to invoice the tax gift it enjoys could therefore theoretically have fallen into the category of advantages by nature, not implying any comparison but where only the existence of counterparts favorable to the operation of the company counts.

In this respect, it is interesting to note that in the aforementioned Philipps case, the tax authorities had explored this avenue during the cassation. Changing their approach, they had attempted to explain before the Council of State that the inclusion of the subsidy was an advantage by nature, and that it was therefore not necessary to demonstrate an advantage by comparison. After recalling what constituted an advantage by nature according to case law, the public rapporteur Romain Victor invited the court to consider that what was at issue in this dispute was indeed the level of prices charged by SAS Philips France for the invoicing of its services to the parent company of the group to which it belongs, at a price that was not zero, thus effectively excluding any idea of selling at a loss. "However, if the sale of a service generating a profit margin may, in certain cases, conceal an indirect transfer of profits abroad, when the margin is insufficient, proof of this insufficiency can only result, when the prices charged are not zero, from a comparison with other transactions, this comparison being the only one capable of establishing that the level of prices charged to the associated foreign company differs from that usually charged to other customers or by other similar companies operating normally. Nor are we here in the hypothesis of a free transfer, which can a priori be regarded as devoid of consideration and exempting the administration from establishing a panel of relevant comparables."

Certainly, it is true that the advantage by nature induces a character of gratuity. As soon as the transaction has been the subject of a counterpart, the debate would then implicitly but necessarily be placed on the ground of the advantage by comparison, with all the subjectivity that we have pointed out previously. The system of article 57 of the CGI can then be easily summarized as follows: if the transaction has been the subject of remuneration, the only question that remains is to know if this is sufficient, that is to say if it corresponds to what independent and comparable companies would have claimed in a similar situation, even if perhaps to the detriment of their own interest. This is one of the incongruities of the principle of full competition, whose economic and Anglo-Saxon inspiration has over time overflowed our praetorian axiom of the abnormal act of management, but which we will not address in these columns.

Doesn't the comparability analysis create a double standard?
The matter is therefore settled: the Philipps France and ST Microelectronics cases are based on a comparative advantage, and we will be careful not to question minds as brilliant and enlightened as those of the illustrious public rapporteurs cited above.

However, it must be noted that comparative analysis quickly loses its supremacy when the composition of the cost base to be re-invoiced is invoked in a contract. In a SAP Holding France SAS ruling, the Marseille Administrative Court of Appeal thus validated the increases made by the administration, which, on the grounds that an intragroup R&D services contract (yet another one!) specified that all taxes and customs duties would be borne by the related party, correlatively reintegrated the amount of the contribution on the added value of companies into the basis of the cost-plus method.

Surely drawing the consequences of the Philipps France case law, the taxpayer then attempted to argue that the administration had not made comparisons demonstrating that the price paid in return for the services provided to the foreign company was higher than those charged by similar businesses normally operated with suppliers with no ties of dependency. It is a fact that, to repeat the dialectic of the rapporteur Romain Victor, the prices charged were not zero, and we were not in the hypothesis of a free transfer. The conditions for the application of the theory of comparative advantage therefore seemed to be met. However, the tax judge dismissed the debate by validating the transfer of an advantage, which he did not qualify, on the sole basis of the contract. If this decision then seems to create a two-speed situation, we justify it rather for our part by the subsidiarity of the comparative analysis which irrigates the semantics of article 57 of the CGI under the visa of which these rectifications in matters of transfer pricing are placed.

As we have almost prophetically highlighted, "Article 57 is not a carbon copy of the arm's length principle. The comparability analysis is only mentioned as a subsidiary measure and "in the absence of specific elements". It is therefore quite possible that a correction is made in the sphere of transfer pricing, even though these adjustments would lead to dissociating the remuneration of the related transaction from any arm's length reference."

This debate, which we will be able to resolve here, nevertheless has one merit: that of highlighting the immense difficulty in correctly qualifying the advantage. Since qualification forms the matrix of any demonstration in law, it is therefore not surprising that the administration failed in the dialectic of proof that weighed on it in the Philipps France and ST Microelectronics cases. However, it demonstrated real ingenuity in turning the tables on the taxpayer, by sanctioning incomplete transfer pricing documentation rather than contesting the basis of assessment.

The taxman's counter-attack, or the battle of penalties rather than the tax base

Despite the identity of facts in the various ST Microelectronics cases, the debate submitted to the Orléans Administrative Court did not focus on Article 57 of the CGI, but on Article 1735 ter of the same code. In doing so, the administration deliberately chose to avoid the issue of the basis of assessment, which, in accordance with the dialectic of proof, invited it to relate the demonstration of an indirect transfer of profit, to the question of the completeness of the transfer pricing documentation, which weighs on the taxpayer given its capital and financial dimension.

Indeed, it is useful to recall that this article 1735 ter of the CGI sets out the sanctions relating to the absence, or incomplete submission of transfer pricing documentation as described in article L13 AA of the Book of Tax Procedures (LPF). This obligation introduced into our legal corpus by the effect of the Amending Finance Law for 2009, has evolved to comply with OECD standards. Since the financial years open from 1 January 2018, the content of the documentation has thus been detailed in line with the recommendations of the OECD Committee on Fiscal Affairs and now consists of two reports, producing in turn factual and economic information (for the main file) as well as supporting, corroborative and precise elements (for the local file).

The objective of this documentation is clearly stated by the tax administration, which specifies that this documentation "therefore retains a general nature but must be sufficiently precise to allow the administration to assess whether the transfer pricing policy implemented by the company complies with the arm's length principle." It is thus in line with the OECD, which justifies its existence by stating that it is "(2) to provide tax administrations with the information necessary to enable them to make an informed assessment of transfer pricing risks; and (3) to provide tax administrations with useful information to carry out a sufficiently thorough audit of the transfer pricing practices of taxable entities in their jurisdiction, even if it may be necessary to supplement this documentation with additional information as the audit procedure progresses."

In this respect, transfer pricing documentation contributes to the tax transparency that informs the OECD's so-called "BEPS" work, particularly its Action 13. This action implicitly denounced the imbalance in the balance of power between control services and taxpayers, thereby vitiating the burden of proof on administrations. In its 2015 final report on Action 13, the OECD noted that "transfer pricing verification procedures tend to involve a very large number of elements. They often involve difficult assessments of the comparability of several transactions and markets. They may require the in-depth examination of financial, factual, and sectoral information. The availability of adequate information from various sources during the verification procedure is essential to facilitate a methodical examination by the tax administration of the taxpayer's controlled transactions with associated enterprises, and the application of the applicable transfer pricing rules." The full statement: "In situations where a proper transfer pricing risk assessment suggests that a thorough transfer pricing audit is warranted on one or more issues, it is clear that the tax administration must be able to obtain, within a reasonable time, all relevant documents and information in the taxpayer's possession."

Let us therefore make no mistake: the documentation of transfer prices at the taxpayer's expense is only intended to enable the administration to obtain, on first request, all the contextual, financial and other data enabling it to assess the existence or not of an indirect transfer of profit.
In this respect, it is interesting to note that in its initial version of 2009, Article L13 AA did not appear to require the taxpayer to produce an economic comparability analysis. More precisely, the text only specified that "when the chosen method requires it, an analysis of the elements of comparison considered relevant by the company." The latter, which constitutes the cornerstone of any demonstration of an advantage by comparison, was therefore the exclusive responsibility of the administration. The dialectic of proof was at that time unique and complete, regardless of whether the gifts presumed by the control authorities fell within the sphere of transfer pricing, or of the abnormal act of management.
On the contrary, it was only from 2018 that the new version of Article L13 AA, in its OECD-inspired form, set in stone the obligation for the taxpayer falling within its scope to produce "g) A detailed comparability analysis and functional analysis of the audited company and associated companies for each category of transactions, including any changes compared to previous financial years." This was obviously all that was needed for the administration to clarify the contours of this obligation in its doctrine, and now expects that "For each category of transactions, the comparability analysis describes the company's remuneration conditions, justifying the differences with those of independent companies."

This paradigm shift reflects one of the essential axes of tax law and a major distinction with civil proceedings: the one who can establish must prove, not the one who invokes. In the ST Microelectronics case, the administration therefore skillfully changed its strategy. By confronting the company with the documentary obligation incumbent upon it, the administration no longer had to demonstrate the existence of an advantage granted abroad; it only had to expect the taxpayer to produce proof of its absence in its place.

A new demonstration of the tightening of the burden of proof on the taxpayer
This ruling is part of a dynamic which is certainly unfavourable to the taxpayer, but which, let us recognise, aims to correct several decades of imbalance of power in the relationship to proof which opposes it to the administration.

It was case law that first brought about a tightening of the approach. The reversal initiated by the decision of the Council of State in GE Medical Systems, preferring the net margin method to the method initially put forward (and documented) by the company, is a topical illustration of this. To validate the appeal decision, the supreme judge considered that "In order to conclude, moreover, on the merits of the alternative method proposed by the administration, which was based on the study of net transactional margins […], the court dismissed the company's objections based on the lack of relevance of the sample selected by the auditor, took a position on the reference point used to determine the arm's length net margin and considered that, given the significance of the difference observed between the results declared by the company and its arm's length operating results resulting from the application of the net margin transactional method, the administration should be considered as establishing the existence of an indirect transfer of profit during the financial years audited."

On the same day, in a relevant case, the Council of State also confirmed, with a sibylline formula, the prevalence of the median of the arm's length intervals, by shifting the burden of demonstrating that another point in the interval would be more appropriate to the taxpayer. The consideration, according to which "this median constituted, in the circumstances of the case, the point in the interval that best reflected the facts and circumstances of the transactions concerned", has since been repeated in subsequent decisions, further strengthening the due diligence now imposed on the taxpayer. Thus: "the applicant does not, for its part, provide evidence of any specific circumstances enabling it to establish that the administration, taking into account the transactions in dispute, should have deviated from this median margin, the application of which is moreover recommended by the OECD in matters of transfer pricing, as enabling the amount of the corrections to be calculated and the margins of approximation to be limited in relation to a point situated at one or other of the extreme limits of this interval, which must be regarded as the point of the interval which best reflects the facts and circumstances of the transactions concerned."

This tightening was then manifested in the Finance Act for 2024, which formally made the content of the documentation binding on the taxpayer who produced it. Article 57 of the CGI now has a new weapon, and specifies that "When the method of determining transfer prices deviates from that provided for in the documentation made available to the administration by a legal entity pursuant to III of Article L. 13 AA or Article L. 13 AB of the Book of Tax Procedures, the difference noted between the result and the amount it would have reached if this documentation had been respected is deemed to constitute an indirectly transferred profit within the meaning of the first paragraph of this article, unless the legal entity demonstrates the absence of a transfer either by way of an increase or decrease in the purchase or sale prices, or by any other means."

From a hardening to a reversal, there was ultimately only a gap thinner than the Rubicon, which the administration then crossed easily by skillfully using the texts. This ruling of the Administrative Court of Orléans, far from going unnoticed, must therefore on the contrary alert us to the semantic and procedural risks that now weigh on taxpayers, in particular those falling within the scope of the documentary obligation of Article L13 AA of the LPF. Because it is likely up to them to provide proof of the full competition nature of their intragroup transactions, with all the subjectivity that we know about this concept.

Worse, since the thresholds were significantly lowered during the 2024 finance law, we can bet that the misfortunes of ST Microelectronics will now be shared by a growing number of companies.

It remains to be seen whether, by having de facto reversed the dialectic of proof on the taxpayer, the legislator and the tax judge will not have ultimately created a difference in treatment, or even discrimination, with regard to these taxpayers engaged in cross-border transactions, as opposed to those involved in strictly domestic flows and subject to the sole praetorian concept of the abnormal act of management, which remains the exclusive responsibility of the administration. Faced with this dilemma, there is only one salvation: the perfect alignment of this concept with the ever-expanding discipline of transfer pricing.

Read more:

https://www.doctrine.fr/d/TA/Orleans/2025/TAA1819EC909E54748F631

https://www.doctrine.fr/d/TA/Montreuil/2019/U64A41A86DCBC1446967A

https://www.doctrine.fr/d/CAA/Marseille/2021/CETATEXT000043799583

The Conseil d'Etat has ruled in favor of Amycel in a dispute with the French tax authorities. This case, which concerned transfer pricing issues and the discrepancy in sales prices observed by the tax authorities for the same products vis-à-vis sister companies and non-affiliated companies, is a reminder of the imperative need to carry out a detailed comparative analysis of transactions and the economic conditions surrounding them.

The background to the case

The case concerns Amycel France, a subsidiary of the American Monterey group specializing in the production and marketing of mycelium. The tax authorities had reintegrated into the company's results, for the 1998 to 2001 financial years, sums considered as profits indirectly transferred to foreign companies in the same group, pursuant to Article 57 of the CGI. To this end, the tax authorities noted that for the same products, the company charged lower prices to its sister companies in the Netherlands and the United Kingdom than to its third-party customers in France and elsewhere.

The judicial process

After a succession of unsuccessful attempts before the Orléans Administrative Court and the Nantes CAA, Amycel appealed to the Conseil d'Etat. The Conseil d'Etat not only annulled the CAA's decision on the grounds of insufficient reasoning, but also ruled in the company's favor on the merits of the case.

The decisive arguments

The Conseil d'Etat emphasized that the tax authorities had correctly established that the prices charged by Amycel France to its sister companies were lower than those charged to its other customers. However, the tax judge accepted the company's argument that neither the tax authorities nor the CAA had examined whether the independent customers to whom Amycel sold its products at a higher price were in a similar economic situation to its sister companies. This difference in situation was likely to influence the price charged, not only because of the combination of functions and risks borne by the sister companies and third-party customers, but also because of their place in a global value chain.

The implications of the decision

This is a valuable decision, as it reminds us that in the case of a remunerated transaction (i.e. one not carried out free of charge), the demonstration of an anomaly by the tax authorities must necessarily involve highlighting an advantage by comparison.
Now, in addition to reviewing the functions, risks and assets of the parties to the transaction (the functional analysis), this comparative analysis must necessarily involve a study of the economic circumstances of the transaction, as well as the economic strategy of the parties. In this case, the third parties to whom Amycel sold its products were end consumers, whereas its sister companies were wholesalers, and therefore positioned at a different stage of the value chain. This positioning most certainly explained the price differential, enabling them to generate their own margins when reselling to consumers on their markets.

For international groups, this decision is also a reminder of the importance of carefully documenting their transfer pricing and payment terms policies, taking into account the specificities of their business sector.