Tested Party Selection in Transfer Pricing: How to Choose the Right Entity
Borys Ulanenko
CEO of ArmsLength AI
TL;DR - Key Takeaways
The tested party will most often be the entity with the less complex functional profile—routine functions, no unique intangibles, limited risks—because it's easier to benchmark against independent comparables.
Common tested party profiles include limited-risk distributors, contract manufacturers, toll manufacturers, and routine service providers. Entities owning unique and valuable IP are typically not suitable as tested parties.
When both parties are complex with unique contributions, one-sided methods (TNMM/CPM) may not be reliable—consider the profit split method instead.
Document your selection thoroughly: explain both parties' functional profiles, why the tested party is less complex, and confirm data availability for benchmarking.
Jurisdictional nuances matter: Germany prefers local comparables; India has mixed case law on foreign tested parties—some tribunals accept, others reject.
Get the latest transfer pricing insights, AI benchmarking tips, and industry updates delivered straight to your inbox.
Quick Answer: How to Select the Tested Party
The tested party is the entity in a controlled transaction whose profitability is benchmarked against independent comparables. Select the entity with the less complex functional profile—typically the one performing routine functions, bearing limited risks, and owning no unique intangibles.
The tested party is the participant in a controlled transaction selected for analysis under one-sided transfer pricing methods like the Transactional Net Margin Method (TNMM) or Comparable Profits Method (CPM). Its financial results—typically a profit level indicator such as operating margin or net cost plus—are compared against those of independent companies to determine whether the intercompany transaction is priced at arm's length.
OECD Guidance (¶3.18-3.19): The tested party is generally the party to which the method can be applied most reliably and for which the most reliable comparables can be found—and it will most often be the less complex party. Paragraph 3.19 illustrates why entities with unique and valuable intangibles are usually not suitable tested parties.
Why Do We Need a Tested Party?
One-sided methods focus on one side of the transaction. By testing the routine party's profitability, you establish its arm's length return first. Mechanically, once the tested party's arm's length result is set, the counterparty's result becomes the residual in the model—though this doesn't by itself prove the counterparty is entitled to all residual returns. This approach:
Simplifies comparability: Routine functions are common in the marketplace; finding independent companies doing similar work is straightforward
Avoids circular logic: Trying to benchmark a unique IP owner would require finding other companies with similar unique assets—a contradiction
Aligns with economic substance: The entity with routine functions earns a baseline return; the entity taking entrepreneurial risks earns residual returns
The "Less Complex Entity" Principle
At the heart of tested party selection is the principle that you should test the routine side of the transaction. This is the entity more closely resembling an independent service provider, distributor, or contract manufacturer.
Why Less Complex?
The main reason is comparability. A less complex entity's activities can be matched with independent companies because routine functions and limited risks are common in many businesses. For example, independent distributors and contract manufacturers are prevalent, and their profit margins serve as arm's length references.
In contrast, a high-complexity entity owning a valuable patent might have no true comparables—making benchmarking unreliable or impossible.
US Treasury Regulations
The US regulations explicitly codify this principle. Treasury Reg. §1.482-5(b)(2)(i) states:
"The tested party will be the participant in the controlled transaction whose operating profit ... can be verified using the most reliable data and requiring the fewest and most reliable adjustments, and for which reliable data for uncontrolled comparables can be located."
The regulation adds: "Consequently, in most cases the tested party will be the least complex of the controlled taxpayers and will not own valuable intangible property or unique assets."
Selection Criteria: Practical Checklist
Determining the tested party requires systematic evaluation. Use this checklist after completing your functional analysis:
Functional Profile Criteria
Criterion
Question to Ask
Impact on Selection
Functions
Does the entity perform only routine functions, or unique/high-value functions?
Routine functions → strong tested party candidate. Unique functions (R&D, strategy) → likely not the tested party
Intangible Assets
Does the entity own or develop unique intellectual property?
No unique intangibles → can be tested party. Owns valuable IP → avoid as tested party
Risks
Does the entity bear significant business risks, or are risks limited/mitigated?
Limited risk (guaranteed margins, cost coverage) → good tested party profile. Major risks → not ideal
Decision-Making
Is the entity's role mainly execution, or does it make strategic decisions?
Operational role → favor testing. Strategic control → likely the principal, not tested party
Dependency
Does the entity depend on the counterparty for key assets/instructions?
High dependency on affiliate → routine role, likely tested party
Data & Comparability Criteria
Criterion
Question to Consider
Impact
Comparable Data
Can you locate reliable external comparables for this entity's activities?
If yes → viable tested party. If no comparables exist → cannot reliably test
Financial Segmentation
Can you isolate the financial results for the specific transactions?
If yes → okay to test. If mixed transactions can't be separated → problematic
Accounting Consistency
Are the tested party's financials consistent with potential comparables?
High consistency → more reliable comparison. Major differences requiring many adjustments → less reliable
Multi-Year Data
Is multi-year financial data available?
If yes → improves reliability through trend analysis
Practical Application: Create a comparison table for both parties showing functions, assets, risks, and data availability. The entity with more routine characteristics across all factors is your tested party.
Common Tested Party Profiles
Certain functional profiles appear frequently as tested parties. Understanding these archetypes helps identify the appropriate entity in your transaction.
Limited-Risk Distributor (LRD)
Characteristics:
Purchases goods from affiliated manufacturer/principal and resells to third parties
Risks contractually limited: principal covers bad debts, buys back obsolete inventory
Does not own marketing intangibles (brand belongs to principal)
Executes sales strategies set by others
Benchmarking: Operating margin compared against independent distributors. An LRD earning a 3% operating margin would be benchmarked against similar distributors in the industry.
Contract Manufacturer
Characteristics:
Produces goods per affiliate's specifications under long-term purchase commitments
Principal provides product designs and guarantees to buy output
Does not undertake significant R&D; any process improvements owned by principal
Inventory and market risks mitigated by purchase guarantees
Benchmarking: Cost-plus or net cost plus mark-up compared against independent contract manufacturers.
Toll Manufacturer
Characteristics:
Processes materials without taking title to raw materials or finished goods
Principal provides all materials and retains ownership throughout
Essentially provides a processing service, not manufacturing
Extremely limited risk profile
Benchmarking: Cost-plus fees compared against tolling service providers or contract manufacturers with similar limited scope.
Contract Service Provider
Characteristics:
Provides services (IT support, back-office, contract R&D) for affiliates
Compensated on cost-plus or fixed fee basis
Owns no significant intangibles; methodologies and software belong to principal
Often operates with cost coverage guarantees from principal
Benchmarking: Operating profit to operating cost (cost-plus markup) compared against independent service companies in the same field.
Entities Typically NOT Selected as Tested Party
The counterparties to the profiles above are generally not suitable as tested parties:
Why not tested: No reliable comparables exist for entrepreneurs with unique IP—their returns vary widely based on intangible value.
IP Owner / Licensor
Primary role is owning intellectual property and licensing it
Returns derive from unique intangibles
Entitled to residual profits
Why typically not tested: As a rule of thumb, a tested party will usually not be the entity that owns unique and valuable intangibles, because finding sufficiently reliable external comparables is often difficult. US regulations state that "in most cases" the tested party will not own valuable intangible property (§1.482-5(b)(2)(i)).
Decides production strategy, bears inventory and capacity risk
May have its own brands
Why not tested: Complex profile makes benchmarking unreliable; its profit depends on unique know-how and risk assumption.
Key Principle: An entity that performs all the critical functions, assumes most of the risks, and/or owns unique and valuable assets is usually not a good candidate for one-sided methods—consider testing the other party or applying a profit split where both sides make unique contributions.
Edge Cases: When Both Parties Are Complex
Sometimes neither party is clearly routine. What do you do when both entities perform significant functions or own valuable intangibles?
The OECD Position
The OECD Guidelines (¶2.65-2.66) caution that if each party makes unique and valuable contributions, a one-sided method like TNMM is unlikely to be reliable. Instead, a profit split method is generally more appropriate.
Your Options
1. Profit Split Method
When both sides have unique intangibles or critical functions, split profits based on their relative contributions. This treats both parties symmetrically rather than forcing a single tested party.
2. Relatively Less Complex
If one party is 70% complex and the other 30%, you might designate the less-complex one as tested party—acknowledging reduced reliability and documenting the approximation carefully.
3. Test Both Sides (Practitioner Diagnostic)
As a practical diagnostic (not an OECD/IRS-endorsed requirement), some practitioners perform two one-sided analyses as a sanity check. If both cannot be satisfied simultaneously, it's evidence that profit split may be needed.
Example: A French company provides a patented component to its US affiliate, which integrates it with proprietary software to sell a final product. Both contribute unique IP. Testing either side under TNMM would undervalue the other's contribution. Profit split—allocating based on R&D investment or similar metrics—is appropriate.
When Data Issues Prevent Testing the Less Complex Party
Even if functional analysis points to one party as less complex, practical issues may prevent using it as tested party:
No Comparables Available
You've identified the subsidiary as less complex, but its business is so specialized that no independent companies do exactly the same thing. Continuing to test this entity would violate the "most reliable comparables" requirement.
Data Unavailability
The less complex party's financial data for controlled transactions isn't separately available—perhaps it also sells to third parties and can't segment intercompany results.
Alternatives
Switch the tested party: If the other party has a more common profile or better data availability, consider testing it instead—with thorough justification
Use an alternative method: If neither side can be reliably benchmarked, consider CUP method (if internal comparables exist), valuation techniques, or profit split
Broaden comparability criteria: Expand geographic scope or loosen criteria slightly, making adjustments as needed
Always document your attempts and the reasons for any change in approach.
Jurisdictional Considerations
While the arm's length principle is universal, jurisdictions have practical preferences. Note that practical enforceability and verification concerns often drive tax authority preferences, even where no formal rule exists.
Germany
No legal requirement to test the local German entity or use local comparables, but tax authorities prefer local comparables. German auditors often prefer German comparables; where not feasible, European comparables are commonly used. A comparable set with no local companies may face comparability challenges during audit.
India
India has substantial litigation on "foreign tested party" positions. Indian rules don't explicitly define "tested party," but Rule 10B mechanics have been interpreted differently across tribunals. The GE Money Financial Services case saw Indian authorities reject a foreign affiliate as tested party under a literal reading of the rules. However, case law is mixed—some tribunal decisions reject foreign tested parties as lacking statutory sanction, while other decisions have accepted them. The risk profile for using a foreign tested party in India is materially higher than in many OECD jurisdictions.
United Kingdom
UK (HMRC) guidance on benchmarking focuses heavily on data quality and management of ranges (including when statistical tools like the IQR are appropriate, and how to choose an adjustment point if results are outside the arm's length range). HMRC guidance indicates that if the filed position is outside the arm's length range, the median is often the best adjustment point where comparability defects remain. This guidance relates to range management rather than tested party selection specifically.
United States
No prohibition on testing a foreign entity as long as data quality is good. The IRS focuses on reliability of the result, not nationality of the tested party. OECD ¶3.22 similarly notes the tested party may be domestic or foreign, provided sufficient information exists to apply the method and allow review.
Documentation Requirements
Once you've selected the tested party, document the rationale thoroughly as part of your transfer pricing documentation:
What to Include
Element
Content
Functional profiles of both parties
Comparative summary of functions, assets, and risks for each party
Statement of selection
Explicitly state which entity was selected and why, citing OECD ¶3.18 or local equivalent
Complexity assessment
Document specific factors showing one side is more complex (unique IP, strategic functions, entrepreneurial risks)
Intangibles ownership
Clarify who owns valuable intangibles; if tested party has none, state so explicitly
Data availability confirmation
Note that separate financials were available and that comparable companies were found
Why not the other party
Briefly address why the other party was not tested (e.g., "owns unique IP making comparables analysis unreliable")
Sample Documentation Language
"ManufacturingCo was selected as the tested party because it performs routine contract manufacturing functions, does not own any product intangibles, and bears limited risk under its manufacturing agreement with ParentCo. ParentCo owns the product designs, patents, and brand, and bears entrepreneurial risk including market demand and product liability. This selection is consistent with OECD TP Guidelines ¶3.18, which directs that the tested party should be the entity to which a method can be applied most reliably. Independent contract manufacturers with similar profiles are readily available in commercial databases."
Common Tested Party Selection Mistakes
Mistake #1: Testing the Wrong Party
Choosing the more complex party (e.g., the IP owner) because it's the local entity or has easier data access. This produces unreliable results and is vulnerable to challenge.
Fix: Always ground selection in functional analysis, not convenience.
Mistake #2: Ignoring Intangibles in the Tested Party
Failing to recognize that the tested party has some intangibles or significant risks, then treating it as fully routine. Tax authorities may catch this discrepancy.
Fix: Be thorough in functional analysis. If the tested party has any non-routine features, either adjust for them or explain why they're not material.
Mistake #3: Data-Driven Selection Without Substance
Picking a tested party solely because comparables are easier to find, even though it's not the less complex entity functionally.
Fix: Start with functional analysis. Data availability is important but secondary to the core principle.
Mistake #4: Inconsistent Selection Year-to-Year
Testing Party A in one year and Party B the next for the same transaction without explanation. This appears arbitrary or manipulative.
The tested party is the entity in a controlled transaction whose financial results (e.g., operating margin) are compared against independent benchmarks to determine arm's length pricing. Under one-sided methods like TNMM or CPM, you analyze the tested party's profitability and compare it to uncontrolled companies with similar functions and risks. The tested party is typically the less complex entity—the one without unique intangibles and with routine functions—because its activities are easier to match with independent comparables.
How do I choose which entity to test?
Start with functional analysis: identify the party with less complex functions, no unique intangibles, and limited risks. The OECD says to select the entity for which "the most reliable comparables can be found" (¶3.18). US Treasury Regulations similarly specify choosing the party whose results "can be verified using the most reliable data." Practically, list both parties' functions, assets, and risks in a comparison table—the entity with more routine characteristics across all factors is your tested party. Also verify that comparables data and segmented financials are available for that entity.
Can I test the entity that owns valuable IP?
In most cases, no. Testing an IP owner is discouraged because finding comparables for unique situations is extremely difficult—comparables would need to own similar unique IP, which creates comparability problems. US regulations state that "in most cases" the tested party will not own valuable intangible property (§1.482-5(b)(2)(i)), and OECD ¶3.19 illustrates why IP owners are usually not suitable tested parties. Instead, test the other party (e.g., the routine manufacturer or distributor), and the IP owner's return will be the residual after the tested party's arm's length return is established.
What if both parties in the transaction are complex?
If both entities own unique intangibles or perform critical functions, one-sided methods (TNMM/CPM) may not be reliable. The OECD recommends using a profit split method in such cases (¶2.65-2.66), which analyzes combined profits and allocates them based on each party's contributions. Alternatively, if one party is relatively less complex (say, 30% complex vs. 70%), you might still designate it as tested party with careful documentation—but acknowledge reduced reliability. The key is not to force a tested party approach when it doesn't fit the facts.
Does the tested party have to be in my country?
No—the tested party can be in any jurisdiction. The arm's length principle doesn't mandate testing the local entity, and OECD ¶3.22 explicitly discusses situations where the tested party may be domestic or foreign. However, practical considerations apply: some tax authorities prefer local tested parties because they can verify local data and use local comparables. Germany, for instance, prefers local comparables. India has mixed case law on foreign tested parties—some tribunals accept, others reject—creating materially higher risk than in many OECD jurisdictions. If you test a foreign entity, ensure you have robust comparables and be prepared to defend the choice thoroughly.
What if I can't find comparables for my chosen tested party?
If the initially selected tested party yields no reliable comparables, reconsider your approach. Options include: (1) broaden the search geographically or by industry, (2) evaluate whether the other party can be tested instead (if it has a more common profile), or (3) consider an alternative method entirely (CUP, profit split). The transfer pricing process is iterative—if one route fails, document that you attempted it, explain why it was unworkable, and justify your alternative approach. Never proceed with a benchmarking study that has no valid comparables.
What documentation do I need to justify tested party selection?
Include: (1) functional profiles of both parties showing functions, assets, and risks, (2) explicit statement of which entity was selected and why, citing OECD ¶3.18 or local rules, (3) explanation of why the tested party is less complex (routine functions, no unique IP, limited risks), (4) confirmation that comparables data is available, and (5) brief note on why the other party wasn't tested (e.g., "owns unique intangibles, making comparables unreliable"). This narrative, supported by your functional analysis, demonstrates a principled, defensible approach. Lack of documentation invites auditor challenge.
Can the tested party selection differ for different transactions with the same entity?
Yes, if the entity has multiple transaction types with genuinely different profiles. For example, a subsidiary that both distributes products (routine) and provides specialized consulting (potentially complex) might be the tested party for distribution but not for consulting. However, be cautious—switching tested parties between transactions involving the same company raises questions. Document thoroughly why the facts justify different treatment. If the same entity is consistently routine across all transactions, test it consistently.
