Caught in the Chain Unknowingly: Missing Trader, Buffer, Broker

A position in a contaminated supply chain is not guilt — what you knew, or should have known, at the time is what counts.

Businesses that get caught up in VAT carousel fraud almost never find out at the moment of the deal — they find out months or years later, when the German tax office starts asking questions. Then a word from fraud analysis suddenly sounds like an accusation: buffer company. This page takes the perspective that public debate routinely leaves out — that of the honest intermediate trader. It explains what the role labels actually mean, which CJEU case law protects the trader acting in good faith, why Germany’s Section 25f turns everything into a question of evidence — and what to do at each stage.

The scenario: a normal deal — and a letter two years later

The pattern repeats across case files. A trader buys goods from a supplier it has checked the way one checks: commercial register, VAT ID validation, perhaps a site visit. The goods are real, delivery runs, payment flows, the margin is ordinary. The business repeats. At some point the supplier goes quiet — gone from the market. Much later it emerges: the supplier never remitted the VAT it collected. In the eyes of the German tax fraud investigation unit (Steuerfahndung), the chain is now contaminated. And in the middle of it: a company that did nothing but trade.

This is not a textbook hypothetical. In May 2026, the European Public Prosecutor’s Office (EPPO) ordered searches in a Frankfurt-led investigation into an €18 million small-electronics carousel; German suppliers allegedly acted as buffers, and the allegation is knowing participation. The presumption of innocence applies to those suspects. But the structure of the allegation shows precisely what every one of these cases turns on: the buffer position is where knowledge — or the absence of it — gets decided. If you stand there, indignation will not help you. Evidence will.

The roles — and why they do not describe guilt

Fraud analysis distinguishes three functions. The missing trader acquires goods VAT-free from another EU state, collects domestic VAT and disappears without remitting it. Buffers are the intermediate traders the goods then pass through — they give the chain the face of normal commerce. The broker exports the goods VAT-free and reclaims input VAT. How the whole circuit spins is explained on the VAT carousel fraud page.

What matters is what these labels do not do. They describe positions in a diagram — not knowledge, not intent, not participation. The foundational international research that coined the buffer figure says so itself: a buffer can sit between a missing trader and an exporter without acting fraudulently, and the economic literature expressly describes the intermediaries in carousel chains as mostly unwitting. More than that: the European Parliament’s own research service now works with the figure of the “honest buffer” in its 2026 analysis of fraud countermeasures — the trader who objectively stands in a contaminated chain while subjectively acting in good faith. If the EU’s legislative analysts recognise this figure, national enforcement practice cannot pretend it does not exist.

Hence the rule of thumb: being called a “buffer” in an audit report is a description, not a finding. What incriminates a trader is not the diagram — it is facts proving that the trader knew or should have known of the evasion.

The courts differentiate too

What the research describes, the courts now confirm — most recently with remarkable clarity. In December 2025, in MS KLJUČAROVCI, the General Court of the European Union held that the simplification for intra-Community triangular transactions survives even where the goods are delivered directly to the customer’s customer; that the reseller knew of the divergent transport route does no harm. Economic substance instead of formalism. And in April 2025, in Cityland, the Court of Justice ruled that striking a business off the VAT register is contrary to EU law where the authority does not examine the nature of the infringements and the conduct of the trader concerned. The equation “deregistered partner = buyer in bad faith” does not hold.

Even where the Court has recently widened the administration’s instruments, the filter stays sharp: the joint and several liability it extended in Vaniz in December 2025 — beyond the very dissolution of the tax debtor — applies only where the authority establishes — proven, not presumed — that the person concerned knew or should have known that the tax debtor would not pay the tax. The burden of proof rests entirely with the authority. This is exactly where the documented checking trail pays off.

What the case law protects: Optigen, Kittel, PPUH Stehcemp

The Court of Justice built the protective line before most of today’s allegations were ever raised — and it built it clearly. These judgments bind tax authorities in every EU Member State, including Germany.

In Optigen (2006), decided for traders who had unknowingly stood in carousel chains, the Court held: each transaction must be assessed on its own merits. The right to deduct input VAT is not affected by the fact that another transaction in the same chain — earlier or later — is vitiated by VAT fraud, where the taxable person neither knew nor could have known of it.

In Kittel, later the same year, the Court confirmed the protection before drawing the boundary: traders who take every measure that can reasonably be required of them to ensure their transactions are not connected with fraud may rely on the legality of those transactions — without risking their right to deduct. Only the trader who knew or should have known loses that reliance. The line, in other words, does not run between “inside the chain” and “outside the chain”. It runs between knowledge and the absence of any means of knowing. UK readers will recognise this as the Kittel principle applied in HMRC assessments; it is the same test, EU-wide.

In PPUH Stehcemp (2015), the Court sharpened the protection further: even where the invoicing supplier appears, by the administration’s criteria, to be a non-existent trader, the good-faith customer keeps its input VAT deduction, provided the substantive conditions of the supply are met and the customer neither knew nor should have known of the fraud. The message: a supplier later unmasked as a letterbox entity does not retroactively turn its honest customer into a bad-faith one.

Fighting fraud is legitimate — the Court stresses it in each of these judgments. But it has never allowed the chain position itself to become the accusation. Germany translated this line into Section 25f of the German VAT Act in 2020 — the provision under which input VAT deduction and zero-rating are denied to those who knew or should have known. What the provision says, and where its limits lie, is set out on the Section 25f page.

The evidence question: timing, burden of proof — and practical reality

Three rules decide the outcome of almost every buffer case in Germany.

First: timing. What counts is the state of knowledge at the moment of supply. A right to deduct that arose lawfully does not lapse because the trader later learns of the chain’s contamination. German case-law commentary states it without qualification: knowledge acquired afterwards does not retroactively destroy the deduction — and indicia drawn from conduct after the event must be treated with the greatest caution.

Second: the burden of proof. It lies with the tax authority. The CJEU reaffirmed in Global Ink Trade (2024) what the presiding judge of the competent German Federal Fiscal Court senate distils in his annual case-law review: the tax administration must precisely establish the elements of the evasion, prove the fraudulent conduct, and demonstrate that the taxable person actively participated or knew or should have known. Sweeping chain logic does not meet that standard.

Third — the uncomfortable truth: practical reality. In German proceedings, the authority assembles objective anomalies, and the business must explain why it could not have recognised them. A company that can only say “we knew nothing” stands weakly. A company that can show “we checked, we assessed, we documented — and on the information then available the deal was unremarkable” stands differently. The defence can present that diligence in a structured way, up to and including expert reports on the standard of care observed. But it can only present what exists. Evidence built at the time of trading is not bureaucracy. It is the only currency this type of proceeding accepts.

Limitation: why a 2015 transaction can still catch up with you in 2053

Few misconceptions are as expensive in carousel cases as the sentence: “That is long since time-barred.” The limitation periods of German tax law and criminal tax law are longer than most business owners believe possible — and they extend one another.

On the tax side, the assessment period where tax evasion is alleged is ten years instead of four (Section 169(2) sentence 2 of the German Fiscal Code (Abgabenordnung – AO)). Nor does it begin with the supply: it starts only at the end of the year in which the tax return was filed — and where no return was filed, only at the end of the third year after the tax arose (Section 170(2) AO). A transaction from 2015 can therefore quite comfortably still be assessed until the end of 2028. And even that end is no end: if the tax fraud investigation unit begins its enquiries before the period expires, the period does not run out until the assessments based on those enquiries have become final and unappealable (Section 171(5) AO). Above all, Section 171(7) AO couples the tax clock to the criminal law: in cases of evasion, the assessment period does not end for as long as the offence can still be prosecuted.

On the criminal side, ordinary tax evasion becomes time-barred after five years. In the particularly serious cases under Section 370(3) AO — which include evasion “on a large scale”, in the case law from EUR 50,000 per offence — the prosecution limitation period is fifteen years instead (Section 376(1) AO). That period starts only upon completion of the offence — for assessed taxes, as a rule upon notification of the incorrect assessment. Every search warrant, every notification that proceedings have been opened, every arrest warrant interrupts it — and starts it running afresh (Section 78c of the German Criminal Code (Strafgesetzbuch – StGB)). The absolute ceiling in these cases lies only at two and a half times the statutory period, that is at 37.5 years (Section 376(3) AO); and from the opening of the trial before a regional court, limitation can additionally be suspended for up to five years (Section 78b(4) StGB).

Added together, the result is uncomfortable but real: a supply carried out in 2015 that surfaces years later in a chain investigation can remain open — through Sections 171(5) and (7) AO on the tax side, and through Section 376 AO in conjunction with Section 78c StGB on the criminal side — until well beyond 2050. More than two decades of uncertainty is not a theoretical construct in carousel complexes; it is procedural reality.

The defence conclusion is just as sober: the statutory retention periods — in principle ten years for books and records, eight years for accounting vouchers (Section 147 AO) — expire long before the risk does. A business that relies on them destroys its own exculpatory evidence on schedule, while the authority works with cross-check notifications, databases and third parties’ criminal files that know no deletion policy. The transaction files on EU supply chains — VAT ID queries, verification notes, transport evidence, payment records, sign-offs — therefore belong in an archive that keeps them legible, complete and evidentially robust for twenty years and more: audit-proof, migration-safe, time-stamped, and separated from the operational system. Limitation rarely protects in these proceedings. Evidence always does.

What the honest buffer can do — at each stage

Before any suspicion: make your position provable. A risk-tiered supplier due diligence — baseline checks for every counterparty, enhanced review on concrete cause — is exactly the standard the case law demands: no more, no less. Record every check with date, source, result and decision-maker. It takes hours to organise — and is priceless in a dispute. Where your supply chain stands today is shown by the carousel risk self-check.

At the first sign: do not wait. An audit query about a vanished supplier, a cross-check notification, an information request — these are the moments the case is shaped. Secure the transaction file now, change nothing retroactively (file cosmetics backfire), bring in advisers.

In the acute case: a dawn raid, an assessment denying input VAT, an asset freeze — the first 72 hours decide the evidence position and the company’s liquidity. Banks, credit insurers and key suppliers often react to an investigation faster than any court rules on it. That is why defence begins at first contact, not at indictment. For international groups: German proceedings run on German documentation expectations — your German evidence should match them.

An honest trader has no right to be blind to red flags. But neither may system risk be reinterpreted as individual knowledge. Between those two sentences lies the whole case — and usually its solution.

FAQ