For groups above the EUR 750 million threshold, that changes the order of operations. Prices set during the year determine the country-by-country numbers, those numbers determine whether the safe harbour holds, and the safe harbour determines whether anyone computes a full effective tax rate at all. Documentation that arrives after the accounts close arrives after the decision.
What Pillar Two does with a transfer price
Pillar Two — the global minimum tax that closes out the OECD/G20 BEPS project — applies to multinational groups with consolidated revenue of EUR 750 million or more in at least two of the four preceding fiscal years. Within scope, the calculation starts from financial accounting profit and adjusts it, and one of those adjustments is squarely a transfer pricing rule.
Article 3.2.3 of the GloBE Model Rules has two limbs, and practitioners routinely collapse them into one.
The first limb is cross-border and imposes a double requirement. A transaction between constituent entities in different jurisdictions that is not recorded in the same amount in both entities' financial accounts, or that is not consistent with the arm's length principle, must be adjusted so that it is both. Symmetry alone is not enough — two entities can agree perfectly on a price no independent party would accept. Arm's length alone is not enough either, if the two ledgers disagree about the amount.
The second limb is domestic and much narrower: where two constituent entities in the same jurisdiction transfer an asset at a price that is not arm's length, only a loss has to be recomputed. Because Pillar Two blends income and taxes across a jurisdiction, same-country mispricing generally nets out. A loss does not, so it is caught.
A mismatch a domestic tax authority might treat as a presentation difference therefore becomes a mandatory adjustment to the minimum tax base. Intercompany accounts that never quite reconcile, credit notes booked in one entity and not the other, year-end true-ups recorded on one side — each now has a second consequence.
The safe harbour that decides whether anyone looks closer
Very few groups compute a full GloBE effective tax rate for every jurisdiction. Most rely on the transitional country-by-country reporting safe harbour, which reads the CbC report already filed under BEPS Action 13 and deems top-up tax to be zero for a jurisdiction passing any one of three tests. It is claimed, not automatic — and a jurisdiction left out in an eligible year generally cannot be brought back in later.
The transition rate is not a flat 15 %. It rises: 15 % for fiscal years beginning in 2023 and 2024, 16 % for 2025, 17 % for 2026. Under the Side-by-Side Package released by the Inclusive Framework on 5 January 2026, the safe harbour was extended by a further year — to fiscal years beginning on or before 31 December 2027, excluding any fiscal year ending after 30 June 2029 — with the rate held at 17 % for both 2026 and 2027. The same package added a permanent simplified ETR safe harbour at the 15 % minimum rate, computed from consolidated accounts rather than CbCR data, so for 2026 and 2027 the two run alongside each other.
The extension buys time, but it also means the outcome for two more years is decided by country-by-country data rather than by a full GloBE computation — which keeps the quality of that data, and of the transfer prices feeding it, in the critical path for longer than most groups planned. The package is being incorporated into the Commentary rather than the Model Rules, so most jurisdictions need domestic legislation first; several, Germany included, had not enacted it at the time of writing.
Why this is a data-quality rule, not a tax rule
The safe harbour does not accept any country-by-country report. It requires a qualified one: prepared from qualified financial statements and using accounting data consistently. Those are the accounts used for the ultimate parent's consolidated financial statements, separate financial statements prepared under an acceptable or authorised standard, or — for entities excluded from consolidation on size or materiality grounds alone — the accounts used for CbCR purposes. A group may use different types across different tested jurisdictions, but not different types within one jurisdiction.
That single consistency rule has more bite than it first appears, because group transfer pricing is often steered from consolidation data while local files are built from statutory accounts, and the two rarely agree to the euro. The December 2023 administrative guidance also added anti-arbitrage rules for deduction/non-inclusion, duplicate loss and duplicate tax recognition arrangements entered into or materially amended after 15 December 2022, whose effects must be stripped out. Financing and cost-allocation structures are the usual place they surface.
The consequence is unglamorous and important: the tests run on the numbers as recorded. A year-end true-up booked outside the accounts the CbC report was built from does not reliably repair a test that has already failed.
Where a full computation is required, the asymmetry runs the same way. Under Article 4.6.1 an adjustment that increases covered taxes for an earlier year is picked up in the year it is made, while one that decreases them forces a recalculation of that earlier year's effective tax rate and top-up tax, subject to an election for decreases below EUR 1 million. Meanwhile an upward transfer pricing adjustment raises GloBE income without any mechanism automatically lifting covered taxes to match — so the jurisdictional rate can fall even as the group concedes more profit. An audit settlement three years after the event can reopen a computation everyone considered closed.
What changes in the benchmarking workflow
None of this changes the arm's length standard. It changes when the work has to be finished and how reproducible it has to be.
- Set prices before the year, not after it. The safe harbour runs on booked results, which makes operational transfer pricing — target margins agreed early and monitored quarterly — the control that matters. A margin computed in month fifteen produces a reconciliation, not a position.
- Know where each entity sits in the range, not just that it sits inside it. A routine entity benchmarked at the upper quartile is inside the arm's length range and may still fail the routine profits test in a low-substance jurisdiction. The interquartile rangeproduced by a comparable companies analysis is now a planning input as well as a defence — and the choice between transfer pricing methods, TNMM against cost plus or CUP, moves it.
- Keep the two ledgers equal. Article 3.2.3's symmetry limb is mechanical and unforgiving; reconciling intercompany balances is a transfer pricing control now, not only a consolidation chore.
- Document data lineage, not just conclusions. The master file, the local file and the CbC report now have to tell one story. A study whose ratios cannot be traced back to a named company's published accounts is weak in a transfer pricing audit and unusable as evidence for a safe harbour position — which turns a documentation gap into a transfer pricing risk with a number attached.
Financial transactions sit at the harder end of the same scale. Intercompany pricing on loans moves with the market, so arm's length interest rates have to be current at the transaction date. Amount B, the simplified approach for baseline marketing and distribution, is optional and available for fiscal years beginning on or after 1 January 2025 where a jurisdiction has adopted it; where it applies it fixes a return the safe harbour tests then read, and where it does not, the benchmarking study still sets the number.
What this means for the data behind the study
Two requirements now sit on the same benchmarking study, and they pull in the same direction.
The first is coverage. A pan-European group testing routine distributors and service entities needs comparables from the countries the transactions actually sit in. Listed-company sets are too thin for that: most genuine comparables for a routine European entity are private companies, and a search that keeps falling back to the same handful of listed names produces a range nobody defends twice.
The second is traceability. If the range has to survive a transfer pricing audit and support a minimum tax position, every ratio has to lead back to a named company's filed accounts, with the accounting basis, the period and the rejection reasons on the record.
smartZebra built Benchmarking Pro, launched in July 2026, for that combination: a transfer pricing database of more than 500,000 company data points across 15+ European countries, covering the ratios TNMM work actually uses — EBIT margin, net cost plus, gross cost plus, adjusted NCP, gross margin, return on assets and the Berry ratio — with the original source report open behind every data point, and a flat fee rather than a charge per search.
Transfer Pricing Pro is the transfer pricing software layer that turns those comparables into a study: screening, the arm's length range and interquartile calculation, the PLI selection and an export that drops into the documentation file. The automation sits in the mechanical steps; the judgement stays with the adviser. Existing Transfer Pricing Pro licences include Benchmarking Pro, and both run on the same smartZebra platform as the cost of capital, beta factors, and credit spread modules — every output with a calculation log tracing the result back to its raw data, and the entity and methodology facts published in full.
What makes a transfer pricing database defensible in the first place is set out in our guide for tax consultants.
Pillar Two did not raise the transfer pricing standard. It moved the deadline forward and widened the audience. To see what a reproducible benchmarking search looks like end to end, the transfer pricing module shows the TNMM workflow, screening steps and calculation log on live data.
References
- OECD, Global Anti-Base Erosion Model Rules (Pillar Two), 20 December 2021 — Art. 3.2.3 (arm's length and consistency requirement), Art. 4.6.1 (post-filing adjustments)
- OECD, Safe Harbours and Penalty Relief: Global Anti-Base Erosion Rules (Pillar Two), 20 December 2022 — transitional CbCR safe harbour, three tests, transition rates
- OECD, Agreed Administrative Guidance for the Pillar Two GloBE Rules, 18 December 2023 — qualified financial statements, consistency requirement, hybrid arbitrage arrangements
- OECD/G20 Inclusive Framework, Side-by-Side Package, 5 January 2026 — extension of the transitional CbCR safe harbour, new simplified ETR safe harbour
- OECD, Consolidated Commentary to the GloBE Model Rules (2026), 28 May 2026
- OECD, Pillar One — Amount B: simplified and streamlined approach, February 2024, with pricing FAQs of 17 February 2026
- Council Directive (EU) 2022/2523 of 14 December 2022 on ensuring a global minimum level of taxation
Related pages
- Transfer Pricing Benchmarking & TNMM Analysis — the module, the TNMM workflow and the calculation log on live data
- Benchmarking Pro — transfer pricing database with 500,000+ company comparables — launch details, ratio coverage and countries
- Transfer Pricing Databases: A Guide for Tax Consultants — selection criteria and what auditors reject
- Proportionality in Transfer Pricing Benchmarking — how much comparables search the OECD Guidelines actually require
- Credit Spreads & Interest Rates — arm's length intercompany loan rates under OECD Chapter X
- Grounding / Facts — entity definition, data provenance and methodology
.png){kind=spacer}
