Loading...
Loading...
A practical framework for choosing the most appropriate transfer pricing method using transaction delineation, comparability, data availability, tested-party logic, one-sided vs two-sided methods, and jurisdiction-specific caveats.
Borys Ulanenko
CEO, ArmsLength AI
Continue exploring
Browse the full resource library or contact us if you want recommendations for your specific use case.
The best transfer pricing method is the method that gives the most reliable measure of an arm's length result for the specific controlled transaction.
Use this sequence:
frames method selection around the most appropriate method for the case, considering method strengths and weaknesses, the nature of the transaction, availability of reliable information, comparability, and the reliability of adjustments.
For a full overview of the five OECD methods, see Transfer pricing methods: the complete guide. For tested-party mechanics, see tested party selection.
Many transfer pricing files choose a method too late, after the comparables search has already started. That creates a backward workflow: the available database result starts driving the economic story.
A defensible method-selection process works the other way around. The facts come first:
Only after those questions are answered should the file decide whether CUP, RPM, cost plus, TNMM, or profit split is the most appropriate method.
A method can be easy to execute and still be wrong. TNMM is often practical because company-level net margin data is available, but it is not a default method for every transaction.
| Method | What it tests | Best fit | Main reliability question |
|---|---|---|---|
| CUP | Transaction price | Commodities, standardized goods, comparable loans, some services or licenses | Are the uncontrolled prices really comparable, and can differences be adjusted? |
| RPM | Distributor gross margin | Routine buy-sell distribution without significant transformation | Are gross margins comparable despite product, channel, and accounting differences? |
| Cost plus | Supplier gross mark-up on costs | Routine manufacturing and services | Is the cost base and gross mark-up comparable across tested party and comparables? |
| TNMM / CPM | Tested party net margin | Routine distributors, manufacturers, and service providers | Is there a less complex tested party and reliable net-margin comparables? |
| Profit split | Allocation of combined profit | Unique contributions on both sides, integrated operations | Can combined profits and contribution keys be measured reliably? |
In practice, method selection often comes down to whether direct price evidence, gross margin evidence, or net margin evidence is the most reliable evidence available for the actual transaction.
Do not start with the method. Start with the transaction.
For each controlled transaction, document:
The same legal invoice can hide different economic transactions. For example, "distribution" may mean a routine buy-sell distributor, a commissionaire, a full-risk entrepreneur, or an entity that also performs local DEMPE functions. Those profiles point to different methods.
If the controlled transaction is not clearly delineated, method selection becomes guesswork. The file should make the economic transaction legible before it argues for a method.
CUP is the most direct method because it compares prices rather than margins. It should be considered seriously where reliable uncontrolled price evidence exists.
CUP is usually strong when:
CUP is weak when:
For a detailed CUP workflow, see CUP method in transfer pricing.
| CUP question | If yes | If no |
|---|---|---|
| Is there an internal uncontrolled transaction? | Analyze it before external data | Search market evidence or move to other methods |
| Is product or service comparability high? | CUP may be reliable | CUP likely weak unless adjustments are strong |
| Are key terms comparable? | Continue CUP analysis | Quantify differences or reject CUP |
| Are adjustments reliable? | CUP may support a point or range | Consider RPM, cost plus, TNMM, or profit split |
If CUP is not reliable, the next question is whether a traditional transaction method using gross margins can be applied reliably.
The resale price method starts with the price charged to an independent customer and subtracts an arm's length gross margin for the reseller.
RPM tends to fit when:
RPM is usually weak when the reseller performs significant manufacturing, installation, technical services, brand development, or entrepreneurial market-building activity.
The cost plus method starts from the supplier's costs and adds an arm's length gross mark-up.
Cost plus tends to fit when:
Cost plus is usually weak when the supplier bears major entrepreneurial risk, owns valuable IP, or has a cost base that cannot be compared reliably with independent suppliers.
| Fact pattern | Likely gross method | Watch-out |
|---|---|---|
| Routine distributor resells finished goods | RPM | Gross margin classification and product/channel mix |
| Contract manufacturer supplies semi-finished goods | Cost plus | Cost base consistency and pass-through treatment |
| Routine service provider performs back-office support | Cost plus or cost-based TNMM | Whether gross cost data is meaningful for services |
| Distributor performs major local brand-building | Maybe TNMM or profit split | Marketing intangibles may make RPM too narrow |
| Supplier owns unique technology | Usually not cost plus alone | Residual return may belong to IP contribution |
TNMM tests the net profit of a selected party relative to an appropriate base, such as sales, costs, or assets. In US terminology, the comparable profits method (CPM) plays a similar role.
TNMM is often used because net-margin data is more available than transaction-level price or gross-margin data. That practical advantage is real, but it does not remove the need for tested-party logic.
TNMM usually fits when:
Common tested-party profiles include:
For the detailed selection criteria, see tested party selection. For US/OECD terminology, see CPM vs TNMM.
and support the practical idea that the tested party is usually the party for which the method can be applied most reliably, often the less complex party without unique and valuable intangibles.
One-sided methods include RPM, cost plus, TNMM, and CPM. They work by testing one party and treating the other party's result as residual. That can be efficient and reliable for routine transactions, but it fails when the untested side is not merely residual in an economic sense.
| Red flag | Why it matters |
|---|---|
| Both parties own or develop unique intangibles | Testing only one side may ignore valuable contributions by the other side |
| Both parties control key risks | A routine return may not reflect actual risk control |
| Operations are highly integrated | Separate-party benchmarking may not capture joint value creation |
| The tested party is not actually routine | The comparables set will understate its expected return |
| Segmentation is unreliable | Tested-party profitability may include unrelated transactions |
| The counterparty's residual profit is implausible | The method may allocate too much or too little profit to untested contributions |
The profit split method may fit where both sides make unique and valuable contributions or where the operations are so integrated that they cannot be evaluated reliably on a one-sided basis.
Profit split is common in fact patterns involving:
Do not solve a two-sided value-creation problem by calling one party "routine" because a database search is easier. That is a method-selection problem, not a benchmarking problem.
For TNMM or CPM, method selection is not complete until the profit level indicator is selected.
| PLI | Formula | Typical use |
|---|---|---|
| Operating margin | Operating profit / sales | Distributors and sales entities |
| Net cost plus | Operating profit / operating costs | Service providers and contract manufacturers |
| Return on assets | Operating profit / operating assets | Asset-intensive manufacturing or leasing |
| Berry ratio | Gross profit / operating expenses | Limited circumstances, often distribution or agency-like profiles |
The PLI should track the tested party's value driver. A routine distributor is often tested on sales; a routine service provider is often tested on costs; an asset-heavy manufacturer may require an asset-based indicator.
For cost-based returns, be careful not to confuse OECD cost plus with TNMM using net cost plus. The former is a gross method; the latter is a net-margin PLI.
Reliable data matters. A theoretically perfect method may fail if the data needed to apply it does not exist.
But data availability should be weighed with the economics, not used as a shortcut.
| Data issue | Method-selection impact |
|---|---|
| Reliable internal CUP exists | CUP likely deserves priority |
| External price data lacks contract detail | CUP may be weak |
| Gross-margin data is inconsistent | RPM or cost plus may be unreliable |
| Net-margin data is available for routine comparables | TNMM may be practical |
| No reliable one-sided comparables exist | Consider profit split or another method |
| Only aggregated tested-party financials exist | Segmentation may be needed before any method works |
Internal comparables often deserve early attention because they may provide transaction-specific evidence. A taxpayer that sells similar goods to independent customers, buys similar services from third parties, or borrows from banks under comparable terms should document why those internal references do or do not apply.
A good method-selection section does not merely state the chosen method. It briefly explains why other plausible methods were rejected.
For example:
This creates an audit trail and shows that the selected method was reasoned, not automatic.
Use this as a working method-selection flow:
| Question | If yes | If no |
|---|---|---|
| Is there a reliable CUP? | Use CUP or give it primary weight | Continue |
| Is the tested party a routine reseller with reliable gross-margin data? | Consider RPM | Continue |
| Is the tested party a routine supplier and cost base is reliable? | Consider cost plus | Continue |
| Is one party clearly less complex and benchmarkable at net level? | Consider TNMM/CPM | Continue |
| Do both parties make unique contributions or operate in a highly integrated way? | Consider profit split | Revisit transaction delineation and data |
| Do local rules provide a safe harbor or simplified approach? | Assess eligibility and documentation | Apply standard method selection |
This flow is not a substitute for judgment. It is a way to prevent the common error of jumping directly to TNMM without checking whether a more direct method is reliable.
Transfer pricing methods are broadly similar across OECD-aligned systems, but local implementation matters.
Most OECD-aligned jurisdictions apply a "most appropriate method" concept. Traditional transaction methods are not automatically required, but reliable CUP evidence and reliable gross-margin methods should not be ignored.
The US uses the best method rule. The method that provides the most reliable measure of an arm's length result should be selected based on the facts, comparability, data quality, and reliability of adjustments. US terminology can also differ: CPM is a net-profit method aligned broadly with TNMM, not the OECD cost plus method.
Amount B may simplify pricing for certain baseline marketing and distribution activities where adopted and applicable. It does not replace the method-selection exercise for all distributors, and it does not apply to entities outside the defined scope.
Some jurisdictions scrutinize foreign tested parties, local comparables, loss-making comparables, or regional datasets more closely than others. A method that is technically sound globally may need a local defense around data selection and tested-party choice.
Safe harbors for low-value-adding services, small taxpayers, or specified transactions can reduce compliance burden. But safe harbors are jurisdiction-specific and may not bind the counterparty jurisdiction. Check whether using a safe harbor creates double-taxation risk.
For multi-country files, separate the global economic method from local implementation. The group may choose one coherent policy while still documenting local adaptations for data, safe harbors, Amount B, or filing requirements.
Facts: LocalCo buys finished goods from a related manufacturer and resells them to third-party customers. It performs sales execution, warehousing, and routine local marketing. It does not own local marketing intangibles.
Method reasoning: CUP is unavailable because there are no comparable uncontrolled sales. RPM is considered because the resale price is visible, but gross-margin comparables have inconsistent COGS classification. TNMM using operating margin may be more reliable if independent distributor net-margin comparables are available.
Likely result: TNMM or RPM depending on gross-margin reliability. The file should explain why one is more reliable than the other.
Facts: ManufacturerCo produces components using group specifications. The principal owns product IP, controls market risk, and commits to buy output. ManufacturerCo bears routine production risk.
Method reasoning: CUP is absent. Cost plus fits the supplier-side profile if reliable gross mark-up data and a consistent cost base exist. If gross accounting differences are material, TNMM using net cost plus may be more reliable.
Likely result: Cost plus or TNMM with net cost plus, depending on gross data quality.
Facts: USCo and IrelandCo jointly develop platform technology. Each controls important R&D decisions and contributes unique engineering teams. The platform generates global subscription revenue.
Method reasoning: There is no reliable CUP. A one-sided TNMM analysis would treat one party as routine despite both parties making unique contributions.
Likely result: Profit split should be considered, with allocation keys tied to each party's contribution.
Facts: ProducerCo sells a standardized commodity to an affiliated trader. Independent parties use a public index adjusted for quality, timing, delivery location, and freight.
Method reasoning: CUP is likely strong if the quote is commercially relevant and adjustments are reliable.
Likely result: CUP, with detailed pricing-date and adjustment documentation.
A concise method-selection section should cover:
| Section | What to include |
|---|---|
| Transaction delineation | What is being priced and what the parties actually do |
| Functional analysis summary | Functions, assets, risks, risk control, and intangibles |
| Available data | Internal comparables, external price data, gross margins, net margins |
| Methods considered | CUP, RPM, cost plus, TNMM/CPM, profit split |
| Rejected methods | Specific reasons each plausible method was not selected |
| Chosen method | Why it provides the most reliable result |
| Tested party and PLI | If one-sided, why this party and indicator are appropriate |
| Local caveats | Jurisdiction-specific rules, safe harbors, Amount B, local data expectations |
The goal is not to write a long theoretical essay. The goal is to show a reviewer that the method follows from the facts and data.
TNMM is common, but common does not mean automatic. If a reliable CUP or gross method exists, skipping it without analysis weakens the file.
Database availability matters, but it cannot turn a complex entrepreneurial entity into a routine tested party.
OECD cost plus is a gross method. Net cost plus is a PLI often used under TNMM. The file should be clear about which method and profit line are being used.
Even in a one-sided method, the counterparty's functions and assets matter. If the residual profit allocated to the counterparty is not economically coherent, revisit the method.
A method can be economically sensible and still need local tailoring for documentation, safe harbors, Amount B, tested-party acceptance, or comparable data expectations.
Under OECD guidance, the standard is the most appropriate method, not a rigid hierarchy. However, if CUP and another method can be applied with equal reliability, CUP is usually preferred because it directly tests price. Traditional transaction methods may also be preferred over profit methods where reliability is equal.
Use CUP when you have reliable uncontrolled price data for a highly comparable transaction. This often happens with commodities, standardized products, comparable financial transactions, or genuine internal uncontrolled transactions.
Use TNMM when one party is clearly less complex, reliable net-margin comparables are available, and direct price or gross-margin methods are less reliable. TNMM is often practical for routine distributors, manufacturers, and service providers.
Profit split is usually more appropriate where both parties make unique and valuable contributions, both control key risks, or operations are highly integrated. In those cases, testing only one party may not reflect the economics.
You generally should address plausible methods, not every method at equal length. A concise explanation of why CUP, RPM, cost plus, TNMM, or profit split was accepted or rejected is usually more useful than boilerplate.
Yes. Method selection is transaction-specific. A group might use CUP for commodity purchases, TNMM for routine distribution, cost plus for shared services, and profit split for jointly developed IP.
One-sided methods require a tested party. If no party is sufficiently routine and benchmarkable, a one-sided method may fail. Tested-party selection and method selection should therefore be analyzed together.
Sometimes a safe harbor or simplified approach can reduce the analysis required for eligible transactions. But safe harbors are local, may be elective, and may not bind the other jurisdiction. Document eligibility and double-taxation implications.