A family office that moves capital, people, or vehicles across a border still lives under DAC6 MDR family office reporting. The label is European, not a private-banking product. Council Directive (EU) 2018/822, the fifth amendment to the Directive on Administrative Cooperation, requires intermediaries, and in some cases the taxpayer, to file reportable cross-border arrangements that carry at least one hallmark. The regime is a transparency file. It is not a menu of structures, and this note is not a guide to keeping an arrangement off a form.
This article is general information for families and family offices. It is not a tax opinion, a mandate, or personalised advice. Whether an arrangement is reportable depends on the facts, the hallmarks, the main benefit test where it applies, and counsel who can read the file against the directive and the French transposition.
What DAC6 and MDR actually require
DAC6 is the EU name. MDR, mandatory disclosure rules, is the OECD vocabulary for the same family of regimes. The European Commission still presents the package on its DAC6 page. The legal text is Council Directive (EU) 2018/822. France transposed it by ordonnance n° 2019-1068 of 21 October 2019, inserting articles 1649 AD to 1649 AH into the Code général des impôts. The operating rule in 2026 is still that transposition, not a new 2026 rate and not a folklore of “only marketed schemes.”
Two gates sit in front of every file. First, the arrangement must be cross-border in the statutory sense: it concerns France and another State, EU or not, because a participant is resident in more than one place, because an establishment is used, because activity is exercised without residence, or because the arrangement can affect automatic exchange of information or beneficial-ownership identification. A purely domestic French gift of French shares to French children, with no foreign participant and no CRS angle, is usually outside that gate. A Luxembourg holding, a Swiss account, a UK LLP, or a US trust in the same chart is not.
Second, at least one hallmark in Annex IV (in France, the marqueurs in article 1649 AH) must be present. Hallmarks are indicators of potential tax risk. They are not a finding that tax was avoided, and a filing is not an admission of abuse. Adjacent holding files such as ATAD interest limitation for holding companies in 2026 sit in a different chapter. Leverage, hallmarks, and charts still interact.
Which DAC6 MDR family office arrangements are reportable
In practice, family offices do not “do DAC6.” They do transactions that sometimes meet both gates. The useful question is which recurring family-office moves tend to be cross-border and to carry a hallmark, not how to design a silent alternative. Ordinary listed-portfolio rebalancing, a bank current account, or a straightforward purchase of a quoted ETF does not become reportable because a family is wealthy. Standardisation, related-party cross-border payments, CRS-undermining features, and transfer-pricing hallmarks are where the statute actually bites.
A useful inventory starts with the chart, not with the marketing name of the vehicle. Who is resident where. Which entity pays whom. Which product is sold from a shelf with the same slides to more than one family. Which step could weaken CRS or beneficial-ownership reporting. Which intangible or safe-harbour pricing sits between associated enterprises. Those are the questions. “We are a family office, not a promoter” is not an answer. An in-house tax team that designs, markets, organises, or manages the implementation can be an intermediary. If no intermediary with a reporting nexus remains, the relevant taxpayer files.
Hallmarks A to C and the main benefit test
Annex IV splits hallmarks into five categories. Categories A and B, and some of category C, are taken into account only if the main benefit test is also met. The test asks whether the main benefit, or one of the main benefits, a person may reasonably expect from the arrangement is a tax advantage. It is a purpose test with a reasonableness overlay. It is not a kitchen-table slogan that “we also had commercial reasons.”
Category A covers generic features: a confidentiality condition that stops the taxpayer disclosing how the arrangement secures a tax advantage; a fee fixed by reference to the tax advantage, including a refund if the advantage fails; and substantially standardised documentation or structure available to more than one taxpayer without needing to be substantially customised. Category B covers more specific planning features: acquiring a loss-making company to use losses, converting income into capital or into a lower-tax category of receipt, and circular transactions that round-trip funds. Category C covers cross-border deductible payments between associated enterprises in listed situations: a recipient with no tax residence, a recipient in a zero or almost-zero corporate-tax jurisdiction, a recipient on a non-cooperative list, a full exemption, or a preferential regime, plus double-deduction and double-relief patterns, and asset transfers with a material valuation gap.
The main benefit test is the filter that keeps ordinary commercial cross-border activity from flooding the directory. It is also the filter families misread. A standardised Luxembourg SOPARFI kit sold with the same pack to several households can meet hallmark A.3. Whether it is reportable still depends on cross-border status and, for A.3, on the main benefit test. Counsel applies that test to the file. This article does not.
Standardised products are not automatically innocent
Family offices often buy what banks call “standard.” Standard, in DAC6, is a hallmark when the documentation or structure is substantially the same and is available to more than one taxpayer. A wrapper that is merely common in the market is not, by that fact alone, reportable. A wrapper that is a productised cross-border arrangement, with a confidentiality clause or a success fee, and with a tax advantage as a main expected benefit, is in a different conversation. The work is to read the pack against A.1 to A.3, not to assume that “everyone uses it.”
Hallmarks D and E: no main benefit filter
Hallmarks D and E do not require the main benefit test. That is the sentence most family-office summaries omit. If the arrangement is cross-border and a D or E hallmark is present, it is reportable even if nobody’s “main purpose” was tax.
Category D concerns automatic exchange of information and beneficial ownership. It covers arrangements that may undermine CRS-style reporting of financial accounts, including arrangements that use an account, product, or investment that is not a Financial Account or that uses a non-reporting Financial Institution, that transfer accounts or assets to jurisdictions that do not exchange, that reclassify income or capital to products not subject to reporting, that shift a Financial Institution or account out of CRS through a non-transparent legal form, or that undermine identification of beneficial owners through opaque chains, nominee arrangements, or jurisdictions that do not meet beneficial-ownership standards. The list in Annex IV is illustrative (“include at least”). Families who treat a trust, a foundation, or a nominee chain as a privacy tool without reading D are reading the wrong statute. This is a reporting hallmark. It is not an instruction to build a quieter chain.
Category E concerns transfer pricing: unilateral safe harbours; transfers of hard-to-value intangibles; and intra-group transfers of functions, risks, or assets where the projected annual EBIT of the transferor, during the three years after transfer, is less than 50% of the projected annual EBIT if the transfer had not been made. A family group that moves a brand, a customer list, or a manufacturing function into a new company in another Member State can meet E without anyone calling the step a “scheme.” The adjacent IFI conversation in IFI 2026 taxable wealth is a different levy. The same chart still has to be read twice: once for wealth tax look-through, once for hallmarks.
French transposition, intermediaries, and the clock
French law tracks the directive. Article 1649 AD CGI, on Légifrance, defines the cross-border arrangement and the reportable arrangement, including the main benefit filter for A, B, and specified C hallmarks. Article 1649 AE names who files. The administration’s comments sit in the BOFiP on persons required to file. Filings go through impots.gouv.fr, in the professional space by XML for entities and in the particular space for individuals.
The intermediary with a French nexus files in France. A professional bound by criminal secrecy files only with the taxpayer’s agreement; otherwise the taxpayer is in the chain. If several Member States could receive a file, the directive’s ordering rules decide which one keeps it, so that the same arrangement is not multiplied without cause. If no intermediary remains, the relevant taxpayer files. In-house family-office staff who design or manage implementation can be intermediaries. Treating the office as “the client” does not erase that.
The clock is short. The directive’s standard is thirty days from the day after the arrangement is made available, is ready for implementation, or when the first step is implemented, whichever is first. France follows that architecture. A family that closes a cross-border step in August and opens the reporting conversation in November has already missed the operational window. Penalties sit in the tax-procedure code. This article does not recast them as a budget line.
What is usually not the story, and what still must be listed
A day-to-day custody change at the same bank, a rebalancing inside a reported financial account, or a domestic French SCI that only holds a French flat for a French-resident family, without a foreign participant and without a CRS or beneficial-ownership angle, is not the typical DAC6 file. The reverse list is shorter and more expensive to miss: a productised cross-border wrapper; a related-party deductible payment into a low-tax or listed jurisdiction; a step that can weaken CRS or beneficial-ownership identification; a unilateral safe harbour or a hard-to-value intangible moved inside the group; a standardised pack sold to more than one household with a tax advantage as a main expected benefit.
Vellum Finance keeps that list as an inventory, not as a product shelf. The Vellum Finance services model is fee-only coordination among counsel, the office, and the return preparer. The 2026 output is a schedule of cross-border steps, hallmarks considered, who files, and the date of the first step. None of that replaces the intermediary’s legal obligation. It is what makes the obligation answerable.
Conclusion
DAC6 and MDR in 2026 remain a reporting regime for cross-border arrangements that carry a hallmark. Categories A and B, and parts of C, still need the main benefit test. Categories D and E do not. Family offices are in the picture when they design, standardise, or implement, and taxpayers are in the picture when no intermediary files. French law lives in CGI articles 1649 AD to 1649 AH and in the BOFiP comments, with filings on impots.gouv.fr. Read the official pages. List the steps. Let counsel apply the hallmarks. Do not treat silence as a structuring technique.
Discretion. Stability. Prosperity.
Team Vellum
A team of passionate professionals who combine their expertise to bring knowledge through Vellum Finance & Patrimoine blog articles. Each member writes about their own field of expertise, cross referencing with our colleagues own fields to ensure the highest quality of information possible in all our content.
