Selecting the Tested Party: A Structured Framework

01Why the Choice Matters and What the Rules Say

A one-sided method, cost plus, resale price, the transactional net margin method (TNMM), or the comparable profits method (CPM), tests the financial result of one party to a controlled transaction against the results of independent comparables. That party is the tested party. The other party keeps whatever profit remains once the tested party has been placed at an arm’s length return. The choice of which party to test is therefore not a formality. It fixes which entity’s return is pinned to the market and which entity absorbs the residual, and in a two-country transaction it effectively decides which tax base is being defended by reference to comparables and which is being determined by subtraction.

The formal guidance is thin. The US regulations under Section 1.482 describe the tested party as the participant whose operating profit can be verified using the most reliable data and the fewest and most reliable adjustments, and for which reliable uncontrolled comparables can be found. The OECD Guidelines, at paragraph 3.18, frame the choice consistently: the tested party is the one to which the method can be applied most reliably and for which the most reliable comparables exist, which will usually be the party with the less complex functional analysis. Both formulations point to the same instinct, that the tested party should be the simpler, more benchmarkable side. Neither explains why simplicity matters or how to proceed when the obvious candidate turns out to be unsuitable. The reasoning below fills that gap.

02The Threshold Requirement: Profit Must Depend on the Price

A one-sided method infers an unobserved price from an observed profit. That inference, developed at length in Comparability Analysis: The Five Factors and the Economics of an Inference, only works if the price being tested is a large enough driver of the profit being observed to move it materially. This is the first and most basic requirement of tested-party selection, and it applies to whichever party is proposed for testing.

The mechanics are worth stating precisely. For any party, testing its profit is a way of testing the intercompany price only because the price sits inside its profit equation. When the price is a dominant term in that equation, the profit is a sensitive reading of the price. When the price is a minor term, the profit barely responds to it, and inferring the price from the profit becomes unreliable.

Consider a contract manufacturer in Poland selling its entire output to a related principal, so that intercompany sales are essentially all of its revenue. Suppose it earns operating profit of 6 on intercompany sales of 100. A 3 percent change in the intercompany price moves its profit by half, and a 6 percent change doubles it. The profit indicator is a responsive instrument for reading the price. Now imagine the mirror situation on the other side of a transaction, a company for which the controlled purchase is only a small fraction of its total cost base. There, the same percentage change in the intercompany price would barely register in profit, and the company fails the threshold no matter how clean its comparables look.

Combining the buyer and the seller gives four cases, and they organize the initial selection.

Cases 1 and 2 resolve the choice on dependence alone, subject to the further requirements below. Case 3 is the common situation where dependence does not decide it, and the choice turns on comparable availability, functional complexity, and measurement reliability. Case 4 is the signal that a one-sided method is the wrong tool, and that a profit split or another approach should be considered.

03The Two-Transaction Problem

A subtle constraint arises when a single party’s profit is shaped by two different controlled transactions with two different related parties. A distributor might buy inventory from a related supplier in one country and license a trademark from a related party in another, paying a royalty. Its operating profit reflects both the purchase price and the royalty rate at once.

A one-sided method applied to that distributor can test its overall return, but it cannot separate that return into a verdict on the purchase price and a separate verdict on the royalty. The profit indicator is a single number, and a single number cannot simultaneously confirm two prices. Suppose the distributor’s operating margin is 4 percent and the comparable range runs from 3 to 5 percent. The margin sits inside the band, but that tells us the combined effect of the purchase price and the royalty is arm’s length, not that each is arm’s length on its own. The purchase price could be too high and the royalty too low, or the reverse, with the two errors offsetting to produce an acceptable blended margin.

The practical consequence is that a party subject to two controlled prices can serve as the tested party for at most one of them, unless its financial results can be reliably segregated by transaction, for instance through separate segmented accounts that isolate the profit attributable to each. Where segregation is not possible, the second price must be tested some other way, by making the counterparty on that transaction the tested party, or by a method that does not rely on decomposing a single margin.

04What Disqualifies an Otherwise Obvious Candidate

The party with the simplest functional profile is usually the right tested party, but simplicity on the organization chart is not the same as suitability. Two conditions can disqualify a candidate that looks ideal: a profit driver the comparables do not share, and a realized risk outcome that the comparables did not experience.

Unique or atypical profit drivers

The core principle is symmetry of profit drivers. The comparable’s profit and the tested party’s profit must respond to the same economic variables in the same way. A candidate fails when it has a material profit driver that ordinary comparables lack, or lacks one they have, or shares one but feels its effect very differently.

The most common version is a valuable intangible. Suppose a distributor has developed proprietary logistics technology that lowers its costs several points below any independent distributor’s. Its margin is structurally higher than the comparable set for a reason the comparables cannot reproduce. Testing it against them would understate its arm’s length return, because the comparables do not carry the cost advantage that the technology confers. A party that owns unusually valuable intangibles, or that has any atypical economic attribute expected to move its profit but not the prevailing price level, is for that reason a poor tested party. This is the economic content behind the regulatory preference for a tested party that owns no valuable or unique intangibles.

The same logic applies to a proprietary process, an exclusive long-term supply contract at off-market terms, or any company-specific cost structure that comparables do not share. Each introduces a profit driver that breaks the symmetry the inference depends on.

The outcome of risk, not just its assumption

Risk disqualifies a tested party through its realized outcome even more than through its expected assumption. Two companies can bear identical risk in expectation and still diverge sharply once the risk resolves, and the one whose outcome was atypical becomes hard to benchmark against companies whose outcomes were ordinary.

Take two manufacturers that both launch a new product line, bearing the same market risk with the same expected return. One launch succeeds beyond expectation, lifting volume, spreading fixed costs, and raising margin, partly at the expense of competitors whose volumes fall and whose margins compress. A year later the successful company’s realized profit reflects a favorable outcome that the ordinary comparable set never experienced. Using it as the tested party, or including its counterparts as comparables, mixes a lucky outcome with normal ones. Where a party has experienced a significant, atypical risk outcome, its profit for that period is not a reliable basis for a one-sided test, and the practitioner should look to the other party or to a longer measurement window that dilutes the one-off effect.

05The Yardstick Constraint: Can the Result Be Measured Consistently?

The final requirement is often overlooked because it sounds mechanical: even a party that passes every economic test above cannot be the tested party unless its profit can be measured on a basis consistent with the comparables. The problem is that accounting profit is an imperfect proxy for economic profit, and the imperfections do not fall equally on every company.

Economic profit is properly a matter of the timing of income and expense, discounted to a common point, closer in spirit to an internal rate of return than to any single year’s reported figure. Accounting profit, under any standard, departs from that ideal, for example by spreading a capital outlay across future years through depreciation without regard to the discounting that economic analysis would require. In a one-sided study the practitioner is almost always comparing accounting profit to accounting profit, which is acceptable only to the extent that the accounting distortions are similar for the tested party and the comparables.

This is why the choice of party and the choice of profit level indicator cannot be separated. A candidate whose reported profit is heavily distorted relative to the comparables, or for whom no indicator measures the distortion consistently, fails the yardstick test even if its functional profile is ideal. The full treatment of which indicator to use, and how asset age, accounting method, and cost-base definition affect reliability, is the subject of Selecting the Profit Level Indicator. For tested-party selection, the point is narrower: confirm that a reliable, consistently measurable indicator exists for the candidate before settling on it, because a party that cannot be measured on the same basis as its comparables cannot be tested against them.

06A Working Framework: Seven Propositions

The reasoning above can be compressed into a sequence of propositions to work through in order. Together they move from the threshold dependence test, through the disqualifying conditions, to the measurement constraint, and they double as a documentation trail for the selection decision.

Proposition 1. Because a one-sided method infers intercompany prices from profits rather than measuring them directly, the tested party’s profit must be highly dependent on the intercompany prices at issue.

Proposition 2. If a candidate’s profit depends on economic variables specific to it that are unlikely to appear in third-party comparables, that candidate probably cannot serve as the tested party.

Proposition 3. The fewer variables that drive a candidate’s profit, the more likely it is that reliable comparables exist. Simplicity is valuable because it widens the comparable pool.

Proposition 4. A variable that affects profit but not price reduces reliability even when it affects the comparables too. It may preclude using the candidate as the tested party if it affects that candidate very differently from the available comparables.

Proposition 5. If a candidate has controlled transactions with two or more related parties, so that two different controlled prices are in play, it can serve as the tested party for at most one of them, unless its results can be reliably segregated by transaction.

Proposition 6. If the outcome of a risk has significantly shaped a candidate’s profit, any comparable must have assumed the same risk and experienced the same outcome. Shared assumption is not enough; the realized outcome must match.

Proposition 7. A candidate can be the tested party only if a reliable yardstick exists for comparing its results to the comparables. Without a consistently measurable profit indicator, no valid comparison can be made.

Worked in sequence, the propositions turn a choice that the regulations leave underspecified into a reproducible analysis. Proposition 1 sets the threshold; Propositions 2 through 4 test for disqualifying profit drivers; Proposition 5 handles the multi-transaction case; Proposition 6 handles atypical risk outcomes; and Proposition 7 confirms measurability. A candidate that passes all seven is a defensible tested party, and the reasons it passed, or the reasons the more obvious candidate failed, form the core of the documentation.

07When No Party Qualifies

Sometimes the analysis shows that neither party can be tested reliably. This is Case 4 from the dependence table, or a situation where both parties own significant intangibles, or where both have experienced atypical risk outcomes, or where each is subject to profit drivers the other’s comparables cannot reproduce. The correct response is not to force a one-sided method onto an unsuitable party but to recognize that the transaction is two-sided in substance.

Where value creation is substantively shared and intangible-rich on both sides, a profit split is the appropriate method, because no independent comparable captures how the combined return should be divided. The failure of tested-party selection is itself diagnostic: a transaction in which no party can be isolated as routine is telling the practitioner that both parties contribute non-routine value, which is exactly the condition a profit split is designed for. Reaching that conclusion through the framework above, rather than defaulting to it, is what makes the eventual method choice defensible.

Comp-Press · Transfer Pricing Practitioner’s Guidance. General best-practice reference, not legal or tax advice.