Pricing a routine distributor used to mean a study. Now it can mean a table.
When a group sells through its own distribution companies in other countries, the tax question is old and familiar. What margin should a subsidiary that just buys finished goods from a related supplier and resells them to local customers be allowed to keep? For years the answer came from a benchmarking study: assemble a set of independent companies that look similar, work out the range of returns they earn, and defend the result to each tax authority that cared to look. It was slow, it was expensive, and two authorities looking at the same distributor could still reach for different comparables and disagree.
On 19 February 2024 the OECD published a different path for the routine cases. The report, Pillar One Amount B, sets out what it calls a simplified and streamlined approach, and the guidance now lives in the OECD Transfer Pricing Guidelines as an annex to Chapter IV. For a qualifying distributor, the arm’s length return on sales is read from a fixed matrix rather than rebuilt from comparables each year. Jurisdictions can choose to apply it for fiscal years beginning on or after 1 January 2025, and the reach is wide, because unlike much of the wider two-pillar work there is no revenue threshold that keeps smaller groups out.
The part worth dwelling on is the quiet upside for a finance team. The matrix does not ask for anything exotic. Which cell applies to a distributor is decided by three things it already measures: the industry of the goods it sells, how heavy its operating expenses are against revenue, and how much operating capital it carries against revenue. Get those figures clean and the return is close to automatic. Better still, the very data that applies the method is the data that tells management what each distribution business actually earns, so a compliance obligation and a clearer view of the operation turn out to be the same build.
None of this replaces judgement where a distributor is genuinely more than routine. The approach is careful about that: it reaches only transactions that can be reliably priced with a one-sided method, where the tested party owns no unique and valuable intangibles and carries no economically significant risk. Read as a systems problem rather than a filing, what is left is a record a finance team can own: which entities qualify, the data that places each in a cell, the return applied, and one query that shows whether any of it is missing.
What Amount B asks, step by step.
The guidance is really two steps. First, decide whether an entity is in scope at all. Second, if it is, price the transaction through the framework. That second step has three parts of its own, a pricing matrix, an operating expense cross-check, and a data availability mechanism, applied in order. It helps to read them as distinct duties, each with its own data behind it, because that is how the work actually divides. The table below sets out what each step asks, what it decides, and what it needs from the data a finance team holds.
| The step | What it asks | What it decides | What it needs from your data |
|---|---|---|---|
| Scope the transaction | Confirm the arrangement is a qualifying transaction, a buy-sell wholesale distribution of tangible goods or a sales agency or commissionaire contribution to it, and that the tested party owns no unique and valuable intangibles and assumes no economically significant risks | Whether Amount B applies to the entity at all | An accurately delineated distribution transaction, the entity’s functional profile, and a way to exclude services, commodities, digital goods, retail above the de minimis, and any manufacturing, research, or procurement that cannot be cleanly segmented. |
| Check the operating expense guardrail | Confirm operating expenses sit at or above 3% and at or below the jurisdiction’s chosen upper bound, itself between 20% and 30%, of net revenue, measured on a three-year weighted average | Whether the distributor is baseline enough to stay in scope | Three years of operating expense and net revenue for the distribution activity, weighted, per entity, and the upper bound the relevant jurisdiction has set. |
| Read the pricing matrix | Find the industry grouping of the goods and the factor intensity classification from operating expense intensity and net operating asset intensity, then take the return on sales from that matrix cell, within its range of plus or minus 0.5 of a point | The arm’s length return on sales the distributor should earn | A mapping from the goods distributed to an industry grouping, and three-year weighted operating expense intensity and net operating asset intensity for the entity. |
| Apply the operating expense cross-check | Restate the matrix return as a return on operating expense and test it against a cap and collar set by the factor intensity classification and the jurisdiction, adjusting the return on sales if it falls outside | Whether the matrix return needs correcting for an unusually lean or heavy cost base | Current-year operating expense and net revenue, and the cap-and-collar rate for the entity’s classification and country. |
| Apply the data availability mechanism | For a tested party in a qualifying jurisdiction with limited comparable data, add an upward adjustment set by the jurisdiction’s sovereign credit rating and the entity’s net operating asset intensity | The final return on sales for distributors in those jurisdictions | The entity’s country and its sovereign rating category, and its net operating asset intensity, capped at 85% for this calculation. |
Two things are worth pulling out of that table. The first is that scope is doing real work. An entity is in only if the transaction is a qualifying wholesale distribution of tangible goods, or a sales agency or commissionaire that contributes to it, and only if it clears both the qualitative test, no unique and valuable intangibles and no economically significant risk, and the quantitative one, operating expenses inside the guardrail. The distribution of services, commodities, and digital goods is out, as is retail above a de minimis of more than 20% of net revenue, and any manufacturing, research, or procurement that cannot be reliably segmented and priced on its own.
The second is that the pricing inputs are ordinary management numbers. The matrix returns a return on sales, and the cell is fixed by the industry grouping of the goods and by a factor intensity classification built from two ratios: operating expense intensity, which is operating expense over net revenue, and net operating asset intensity, which is net operating assets over net revenue, both on a three-year weighted average. The cross-check then restates the chosen return over operating expense and tests it against a cap and collar, so a return that looks fine on sales but sits oddly against the cost base is pulled back into line. And where the tested party is in a qualifying jurisdiction with thin comparable data, the data availability mechanism adds an upward adjustment set by that jurisdiction’s sovereign credit rating. Every one of these inputs is something a finance team can compute from its own ledger, once the distribution activity is cleanly segmented.
The distributor pricing record, read as one governed control.
A worked example makes it concrete. Take one entity, a regional company that buys finished goods from a related supplier and sells them wholesale to unrelated customers. It owns no valuable local brand or technology and carries no real market risk, so it clears the qualitative test. Its operating expenses run at fourteen percent of net revenue on a three-year weighted average, comfortably inside the guardrail, so it clears the quantitative one. Its goods sit in one industry grouping, and its cost and asset intensities place it in a factor intensity class. Read the matrix at that intersection and the return on sales falls out, within a defined range. Recorded that way, the entity can answer for itself, and a reviewer can see the position without rebuilding it from scratch.
Here is that logic expressed as a record. The entity carries its country and the functional profile that keeps it eligible. A qualifying-transaction block confirms the goods are tangible, the channel is wholesale, and the retail share stays under the de minimis. A scoping block holds the three-year operating expense intensity against the floor and the upper bound. A pricing block records the industry grouping, the two intensities, the factor intensity class, the matrix return, and its accepted range. Cross-check and data availability blocks record the corroborating steps. An election block notes the basis, including a United States election under Notice 2025-04 where it applies. And an exception query at the end names the gaps that would make any of it untrustworthy.
A distributor pricing record for an in-scope entity under Amount B
# Distributor pricing record for an in-scope entity under Amount B
# Names, ratings, and figures are placeholders for shape, not real values.
entity:
legal_name: "Our regional distribution company"
country: "a market where the group sells"
functional_profile:
owns_unique_valuable_intangibles: false # a scoping condition to meet
assumes_economically_significant_risks: false
role: buy_sell_wholesale_distributor # or sales_agent, or commissionaire
qualifying_transaction:
goods: tangible # not services, commodities, or digital goods
channel: wholesale_to_unrelated_customers
retail_share_of_net_revenue: 0.04 # in scope only at or below the 0.20 de minimis
segmentation: "distribution result carved out from any manufacturing, research, or procurement"
scoping_guardrail:
opex_intensity_3yr_weighted: 0.14 # operating expense / net revenue
floor: 0.03
upper_bound: 0.30 # jurisdiction picks 0.20 to 0.30; 0.30 for a US distributor
in_scope: true
pricing_matrix:
industry_grouping: 2
net_operating_asset_intensity_3yr_weighted: 0.22 # OAS
opex_intensity_3yr_weighted: 0.14 # OES, from the guardrail above
factor_intensity_class: C
return_on_sales: 0.030 # read from the matrix cell
accepted_range: [0.025, 0.035] # the cell plus or minus 0.5 of a point
opex_cross_check:
equivalent_return_on_operating_expense: "matrix return restated over current-year opex"
cap_and_collar: "set by factor intensity class and country"
adjusted_return_on_sales: 0.030 # unchanged here, inside the range
data_availability_mechanism:
tested_party_in_qualifying_jurisdiction: false # true adds an upward adjustment
sovereign_rating_category: null # drives the net risk adjustment when it applies
adjusted_return_on_sales: 0.030
election:
basis: "OECD Amount B, as adopted by the relevant jurisdiction"
us_position: "elected under IRS Notice 2025-04 for a US distributor, where chosen"
documentation: "kept to show scope, the matrix inputs, and the return applied"
exception_query: # the readiness control, run before each period
- an in-scope entity whose distribution result cannot be cleanly segmented
- a distributor whose three-year opex or net operating asset intensity cannot be produced
- an entity near the 0.20 retail de minimis or either operating expense bound
- a goods classification with no industry grouping mapped
- a tested party in a qualifying jurisdiction with no sovereign rating recordedThe exception query is what keeps it honest. An in-scope entity whose distribution result cannot be cleanly segmented is a scope claim resting on an estimate. A distributor whose three-year intensities cannot be produced is a cell chosen without its inputs. A position near the retail de minimis or an operating expense bound is a scope call that could flip next year as the weighted average moves. A goods classification with no industry grouping is a return computed on a guess. And a tested party in a qualifying jurisdiction with no sovereign rating recorded is a missing adjustment. None of this is exotic. It is ordinary readiness, measured against the method, so a gap is a query result you clear at the desk rather than a weakness a tax authority finds first.
An implementation checklist.
- 1.Name your in-scope distributors before you price anything. The approach only reaches a qualifying transaction: a buy-sell wholesale distribution of tangible goods to unrelated customers, or a sales agency or commissionaire that contributes to it. Flag each entity that fits, and record the functional profile that keeps it eligible, no unique and valuable intangibles and no economically significant risks, so eligibility is a stored fact rather than a memory refreshed each year.
- 2.Make the distribution result something you can carve out. An entity that also manufactures, researches, or procures is still eligible for its distribution activity, but only if that activity can be reliably segmented and priced on its own. Build the segmentation into the management accounts, so the distributor’s revenue, operating expense, and operating assets exist as their own view rather than as a share someone estimates under audit pressure.
- 3.Compute the three-year weighted intensities once, from the ledger. Both the scoping guardrail and the matrix cell depend on operating expense intensity, and the cell also depends on net operating asset intensity, each on a three-year weighted average. Derive them from the same segmented figures every time, so the number that decides scope and the number that decides the return are the same number, traceable to the accounts.
- 4.Map every product line to an industry grouping. The matrix return varies by the industry of the goods distributed, so the classification is not cosmetic, it moves the answer. Hold the grouping on the item or product-line master, resolve a mixed portfolio the way the guidance describes, and treat an unmapped classification as a data gap to close rather than a judgement to improvise.
- 5.Run the cross-check and the data availability mechanism as steps, not afterthoughts. After the matrix, restate the return over operating expense and test it against the cap and collar for the entity’s classification and country. Where the tested party sits in a qualifying jurisdiction, add the upward adjustment its sovereign rating sets. Encode both so the final return is computed the same way for every entity and can be replayed on demand.
- 6.Record the election and keep the working. Where a jurisdiction offers the approach and you choose it, note the basis, and for a United States distributor note the election under Notice 2025-04. Keep the scope test, the matrix inputs, and the return applied together, because the documentation that proves eligibility is exactly what a reviewer will ask to see.
- 7.Close with one exception query before each period. List every in-scope entity whose distribution result cannot be segmented, whose three-year intensities cannot be produced, that sits near the retail de minimis or an operating expense bound, whose goods have no industry grouping, or that is in a qualifying jurisdiction with no rating recorded. While that list has entries, the pricing is not ready, and the list is far cheaper to clear at the desk than to reconstruct in a controversy.
Failure modes, framed so you can plan around them.
- Treating Amount B as automatic for every distributor. The approach is optional for jurisdictions and reaches only qualifying transactions that clear both the qualitative and the quantitative criteria. An entity that owns valuable local intangibles, carries real market risk, or distributes services or commodities is not in scope, and assuming otherwise applies a routine return to a business that is not routine.
- Leaving the distribution activity tangled with the rest. The single most common reason an eligible entity falls out is that its distribution result cannot be separated from its manufacturing, research, or procurement. If the segmentation only exists as a hurried allocation at year-end, the eligibility that depends on it is fragile. Built into the accounts, it holds.
- Reading the guardrail once and forgetting it moves. The operating expense intensity that decides scope is a three-year weighted average, so it drifts as the mix of years changes. An entity comfortably inside the band this year can approach a bound next year. Watching the trend is the difference between a planned conversation and a surprise at filing.
- Guessing the industry grouping. The matrix return depends on the industry of the goods, so a casual or inconsistent classification quietly changes the arm’s length answer. Held as master data and resolved the way the guidance sets out for mixed portfolios, the grouping is defensible. Chosen afresh each year, it is a soft spot a reviewer will find.
- Skipping the cross-check because the matrix looked reasonable. The operating expense cross-check exists precisely to catch a matrix return that, restated over an unusually lean or heavy cost base, sits outside the cap and collar. Omitting it can leave a return that looks fine on sales but is indefensible on cost, and the adjustment it would have made is not optional where it triggers.
- Ignoring the data availability mechanism for a low-data jurisdiction. For a tested party in a qualifying jurisdiction, the mechanism adds an upward adjustment set by the sovereign rating and the entity’s net operating asset intensity. Treating every jurisdiction the same understates the return where the mechanism applies, and the correction is not a matter of taste, it is part of the method.
- Assuming one country’s adoption binds the other side. A jurisdiction chooses whether to apply the approach, and the two countries in a transaction may not align. A return that one side accepts as a safe harbour is not guaranteed relief on the other, so the counterparty position and the risk of a residual dispute belong in the record rather than in an optimistic assumption.
- Confusing a simpler method with a lighter evidence burden. The approach reduces the search for comparables, not the need to show your working. Eligibility, the segmentation, the intensities, the classification, and the return applied all still have to be documented. A team that hears simplified as unminuted will have the harder conversation later.
What this asks of the data model.
- The tested party is an entity with a functional profile, not just a legal name. Each in-scope distributor carries its country, its role as a buy-sell distributor, sales agent, or commissionaire, and the flags that keep it eligible, so scope is a property of the record rather than a yearly recollection.
- The distribution result is a segmented view, not a whole-company total. Revenue, operating expense, and operating assets for the distribution activity exist in their own right, carved out from any manufacturing, research, or procurement, so the figures that drive scope and pricing describe the activity Amount B actually covers.
- The intensities are derived, weighted, and traceable. Operating expense intensity and net operating asset intensity are computed from the segmented accounts on a three-year weighted average, so the same numbers decide the guardrail, the factor intensity classification, and, where relevant, the data availability adjustment.
- The industry grouping lives on the product master. The classification that moves the matrix return is held against the goods rather than chosen per calculation, and a mixed portfolio is resolved by rule, so the return is reproducible and an unmapped item is a visible gap.
- The matrix cell and its range are stored, not just the point. The record keeps the industry grouping, the factor intensity classification, the return on sales, and the accepted range around it, so the position taken can be read back in full rather than inferred from a single percentage.
- The cross-check and the mechanism are encoded as steps. The restated return on operating expense, the cap and collar, and any sovereign-rating adjustment are computed the same way for every entity, so the final return is a replayable calculation rather than a spreadsheet rebuilt by hand.
- The election and its jurisdiction are recorded facts. Whether a jurisdiction offers the approach, whether the group chose it, and, for a United States distributor, the election under Notice 2025-04, are stored against the entity, so the basis for the return is documented rather than assumed.
- Readiness reduces to one exception query. Entities that cannot be segmented, intensities that cannot be produced, positions near a bound, unmapped goods, and missing ratings are the exception list, and it can run before each period rather than surfacing when a return is questioned.
The evidence to keep.
- The scope determination per entity, showing the qualifying transaction, the functional profile, and the exclusions considered, so eligibility is documented rather than asserted.
- The segmentation working, showing how the distribution result was carved out from any manufacturing, research, or procurement, so the figures that drove scope and pricing describe the right activity.
- The intensity calculations, showing the three-year weighted operating expense intensity and net operating asset intensity tied back to the segmented accounts, so the numbers that set scope and the matrix cell are traceable.
- The industry grouping mapping, showing how the goods were classified and how a mixed portfolio was resolved, so the input that moves the matrix return is defensible.
- The pricing record, showing the matrix cell, the return on sales, the accepted range, the cross-check outcome, and any data availability adjustment, so the whole calculation can be replayed.
- The election and jurisdiction record, showing whether the approach was offered and chosen and, for a United States distributor, the election under Notice 2025-04, so the basis for the return is evidenced.
- The counterparty position, showing whether the other jurisdiction applies the approach and how a residual difference would be handled, so a cross-border mismatch is a known risk rather than a later surprise.
- The exception query output, ideally empty, listing entities that cannot be segmented, intensities that cannot be produced, positions near a bound, unmapped goods, and missing ratings, so the gaps were checked before any period was relied on.
Questions worth asking in your own review.
- Can we name every in-scope distributor in the group today, with its functional profile recorded, or is eligibility still worked out by hand each year?
- Does the distribution result exist as its own segmented view, or is it an allocation someone assembles from a whole-company total under time pressure?
- Are operating expense intensity and net operating asset intensity computed from the same segmented accounts, on a three-year weighted average, and traceable back to the ledger?
- Is every product line mapped to an industry grouping on the master data, and is a mixed portfolio resolved by rule rather than by improvisation?
- Do we run the operating expense cross-check and the data availability mechanism as encoded steps for every entity, or only when someone remembers?
- For a United States distributor, have we recorded the election under Notice 2025-04, along with the scope test and the matrix inputs behind it?
- Where the two jurisdictions in a transaction do not align on the approach, is the counterparty position and the residual dispute risk written down?
- Could we show a reviewer, in one query, that no in-scope entity is missing a segmentation, an intensity, an industry grouping, or a rating for the period?
What this adds up to.
Amount B turns the pricing of a routine distributor from a study into a table. For a qualifying entity, the arm’s length return on sales is read from a matrix set by the industry of the goods and by two ordinary intensity ratios, corroborated by an operating expense cross-check, and lifted where a low-data jurisdiction calls for it. It is optional for jurisdictions, live for fiscal years from 1 January 2025 where adopted, and already an election for United States taxpayers under Notice 2025-04. Read as a chore, it is a new set of rules. Read as an opportunity, it is a more predictable return and less time spent arguing about comparables.
The build that makes it real is a record and a few deliberate choices. Name the in-scope distributors and hold their functional profiles, segment the distribution result so it stands on its own, compute the two intensities from the ledger on a three-year weighted average, map every product line to an industry grouping, and encode the cross-check and the data availability mechanism so the final return is a calculation you can replay rather than a spreadsheet you rebuild. Record the election and keep the working, because the documentation that proves eligibility is exactly what a reviewer will ask for.
If there is one place to start, run the exception query against your own distributors before you trust them priced: every in-scope entity that cannot be segmented, every one whose three-year intensities cannot be produced, every position near a bound, every unmapped classification, and every qualifying-jurisdiction party with no rating recorded. For a tidy group that list is short and the rest is scheduled. For a sprawling one it is the work, and it is far better found now, at the desk, than in the first examination that reaches for the same numbers you already hold.
Sources.
- OECD, Pillar One Amount B, Overview of Amount B (fact sheets, December 2024), downloaded and read directly. Source for the two-step framework: Step 1 scoping, which identifies qualifying transactions and applies the qualitative and quantitative scoping criteria, and Step 2 pricing, which runs the pricing matrix (Step 2-1), the operating expense cross-check (Step 2-2), and the data availability mechanism (Step 2-3). Source too for qualifying transactions being buy-sell wholesale marketing and distribution of tangible goods to unrelated parties and sales agency and commissionaire transactions that contribute to that distribution; for the excluded transactions of services, commodities, digital goods, retail sales above a de minimis of more than 20% of total net revenue, and non-distribution activities such as manufacturing, research and development, and procurement unless they can be reliably segmented; for the quantitative guardrail that operating expenses must not be lower than 3% nor greater than an upper bound between 20% and 30% of net revenue on a three-year weighted average; for the pricing matrix returning a return on sales across three industry groupings and five factor intensity classifications built from operating expense intensity and net operating asset intensity, with a plus or minus 0.5 of a percentage point range; for the operating expense cross-check that restates the matrix return over operating expense and tests it against a cap and collar; and for the data availability mechanism that adjusts the return upward by reference to the sovereign credit rating of a qualifying jurisdiction, with net operating asset intensity capped at 85% for that calculation.
- OECD, Reader’s Guide for Pillar One Amount B, Inclusive Framework on BEPS (2024), downloaded and read directly. Source for the approach being optional and published in the OECD Transfer Pricing Guidelines as an annex to Chapter IV; for jurisdictions being able to apply it to eligible baseline distributors for fiscal years starting in January 2025, and for the report potentially reaching jurisdictions that are not members of the Inclusive Framework but have implemented the Guidelines in domestic law; for the two implementation options, a taxpayer safe harbour and a version a tax administration can also assert; for the Transactional Net Margin Method forming the basis of the pricing mechanism, with an exception where an internal comparable uncontrolled price is available; and for the underlying economic analysis being updated every five years while the financial data is reviewed annually.
- OECD, Pillar One Amount B, full report incorporated into the Transfer Pricing Guidelines. Source for the complete text of the scoping criteria, the pricing matrix, the operating expense cross-check, and the data availability mechanism summarised in this article, and the place to read a paragraph in full rather than in précis.
- OECD, Pillar One Amount B topic page. Source for the report having been released on 19 February 2024 and for the supporting materials that followed, including the June 2024 statements on the definitions of qualifying jurisdiction and covered jurisdiction that replaced placeholder definitions in the report.
- Internal Revenue Service, Notice 2025-04, Application of the Simplified and Streamlined Approach under Section 482, downloaded and read directly. Source for the Treasury Department and the IRS intending to issue proposed regulations that implement the substance of the OECD report; for United States taxpayers being able to rely on the approach for taxable years beginning on or after 1 January 2025 and before those regulations are published; for the approach being an election available to United States distributors and United States related suppliers, consistent with the first implementation option; for the approach being similar to the comparable profits method, known in the Guidelines as the transactional net margin method; and for the operating expense scoping upper bound being 30% for a United States distributor.
- OECD/G20 Inclusive Framework on BEPS, Statement on a Two-Pillar Solution to Address the Tax Challenges Arising from the Digitalisation of the Economy (October 2021). Source for Amount B being designated a component of Pillar One and for the Inclusive Framework comprising over 140 countries and jurisdictions working on an equal footing.
The two-step framework, the scoping criteria, the pricing matrix built on industry grouping and factor intensity, the operating expense cross-check, and the data availability mechanism were read from the OECD Overview of Amount B fact sheets and the Reader’s Guide for Pillar One Amount B, both downloaded and read directly rather than from a summary, and the United States position was read from IRS Notice 2025-04, downloaded and read directly. No embedded posts from X appear here, because no public post on this topic could be verified as current, relevant, and authentic at the time of writing, and an unverified embed is worse than none. Nothing here is tax, legal, or accounting advice for a specific business. The names, ratings, and figures in the worked example and the record are placeholders for shape rather than real values, and whether and how a given entity applies Amount B depends on the jurisdictions involved, the facts of the transaction, and the elections available to it.