Transfer pricing methods.
The 5 OECD methods in practice.

The selection rule: the most appropriate method

Unlike the old fixed-margin regime, Law 14,596/2023 does not let the taxpayer freely pick the method that generates the smallest adjustment. The head of art. 11 requires selecting the most appropriate method — and § 1 sets out the criteria for that choice:

1. Facts and circumstances of the transaction and the suitability of the method to its nature, based on the functional analysis (functions performed, assets used and risks assumed — FAR);
2. Availability of reliable information on comparable transactions between unrelated parties; and
3. Degree of comparability between the controlled transaction and the independent transactions, including the reliability of any necessary comparability adjustments.

There is an explicit legal preference: under § 2 of art. 11, where reliable information on comparable prices exists, PIC is deemed the most appropriate method — unless it is shown that another method applies better to the case. The choice must be justified and documented in the Local File: a “convenient” method without functional justification is one of the most heavily challenged points in an audit.

This is a deliberate break with the past. The rule mirrors the OECD Transfer Pricing Guidelines (2022), chapter II, and shifts the burden onto the taxpayer: it is no longer enough to land on a defensible number; the taxpayer must show why the chosen method best measures the arm's length result for that specific flow. In the TaxUp team's experience, the order of the analysis matters as much as the result — the functional analysis (FAR) drives the method, never the reverse. Reasoning backward from a target margin to a method is precisely the pattern auditors look for, and the most appropriate-method test is what gives the tax authority the legal footing to disregard a study that does it.

Before the method: delineation, tested party and range

Choosing the method is not the first step. Before it, Law 14,596/2023 requires delineating the actual transaction (art. 8): identifying, from the parties' effective conduct, who performs which functions, uses which assets and assumes which risks — because it is substance, not the contract, that defines what is being priced. If the contractual structure does not reflect reality, the tax authority can recharacterize the transaction even before discussing the method, bringing down the entire study.

Two technical concepts complete the design. The tested party (art. 15) is the entity in relation to which the method applies most reliably — as a rule the less complex one, without unique intangibles, whose comparables are easier to find (typically the distributor or the routine manufacturer, not the technology owner). Picking the wrong tested party — for instance, testing the entity that holds the valuable intangible because its books are at hand — is a common error that quietly invalidates an otherwise careful study. And the range of comparables (art. 16): the arm's length result is rarely a single number, but rather a range of values from comparable companies — hence the use of the interquartile range, which discards the extremes and concentrates the analysis on the middle of the market. If the operation's result falls within the range, it is arm's length; if it falls outside, the adjustment is usually to the median, not merely to the nearest edge of the range — which is why a result sitting just outside the lower quartile can produce an adjustment far larger than the small distance to the boundary would suggest.

Before all of this, there are still the comparability adjustments (art. 9): corrections to neutralize differences (volume, term, function, market) between the operation and the comparables. The right method built on poorly adjusted comparables is a fragile defense.

From the fixed margin to the benchmark: what changed

The change of regime is one of nature, not of degree. The old system (Law 9,430/1996, arts. 18 to 23) worked with profit margins set by law — the famous 20%, 30% and 60% by type of method and activity. In practice, the taxpayer could pick the method that produced the best result, and the “proof” was simply applying the catalog percentage. Those provisions were revoked by Law 14,596/2023 (art. 46), and the fixed margins ceased to exist from calendar year 2024 onward.

In place of the ready-made percentage came the benchmark: the market margin must be proven by a study of real comparables — independent companies with a similar functional profile, from which the interquartile range of profitability is extracted. The effort shifted from calculation to proof: the assessment risk stopped being “I got the margin wrong” and became “I did not justify the choice of method and the adjustments”.

In practice, this changes who does the work and when. Before, the tax department only had to apply the percentage at year-end closing; now, throughout the year, it is necessary to map the transactions, choose the method per operation and produce the comparables study — dated and archived. The defense is born in timely documentation, not in a spreadsheet thrown together when the audit arrives.

PIC (CUP) — Comparable Independent Price

It directly compares the price charged in the controlled transaction with prices of comparable transactions between independent parties — internal comparables (the company itself selling to third parties) or external ones (the market). It is the most direct method and the preferred one when a reliable comparable exists (art. 11, § 2).

Illustrative example

• A Brazilian subsidiary imports an input from the parent at US$102/unit;
• The same parent sells the same input, under comparable conditions (volume, term, market), to an independent distributor at US$95/unit;
• A difference of US$7/unit with no comparability justification → an indication of overpricing on the import (profit shifted abroad) → adjustment to the IRPJ/CSLL base on the imported volume.

Typical application: commodities (with a quoted price — see the dedicated section), standardized inputs, interest on intercompany loans (market reference), royalties with public comparables.

Where it exists, an internal comparable — the same group selling the identical product to an unrelated third party — is usually more reliable than an external one, because product, contract terms and market conditions line up almost perfectly and demand fewer comparability adjustments. The catch is sensitivity: PIC compares prices, so even modest differences in volume, delivery terms, warranty or the point in the supply chain can move the price enough to require an adjustment under art. 9. When those differences cannot be reliably quantified, PIC loses its edge and the analysis tends to migrate to a margin-based method such as PRL or MLT, which absorb product-level noise more gracefully.

PRL (RPM) — Resale Price less Profit

It starts from the resale price to independent parties and tests the reseller's gross margin against the gross margins of comparable independent distributors. Suitable for distributors that do not add significant transformation to the product.

Illustrative example

• A Brazilian distributor resells to the local market for R$100 a product bought from the parent;
• Benchmark of independent distributors in the sector: arm's length gross margin of 25%;
• Maximum acceptable intercompany price: R$100 − 25% = R$75;
• If the parent charges R$82, the excess R$7 per unit is a taxable adjustment in Brazil.

Sensitive to differences in accounting classification (cost vs. expense) between the tested party and the comparables — one of the central concerns of the benchmark.

MCL (Cost Plus) — Cost plus Profit

It starts from the supplier's costs in the controlled transaction and tests the gross markup against margins obtained in comparable transactions between independent parties. Suitable when the tested entity is a contract manufacturer or a provider of routine intragroup services.

Illustrative example

• A Brazilian factory produces to order for the parent with a production cost of R$80/unit;
• Benchmark of independent manufacturers: arm's length markup of 12% to 18% on costs (interquartile range);
• Compliant intercompany price: between R$89.60 and R$94.40;
• If the sale to the parent is set at R$84 (5% markup), the Brazilian remuneration is below the range → adjustment into the range.

The critical point is the definition of the cost base (what is included and what is not) — discrepancies here distort the markup and weaken the defense.

MLT (TNMM) — Transactional Net Margin

It tests the net margin of the transaction against the net margin of independent comparables, measured by a profitability indicator (PLI — Profit Level Indicator): operating margin on revenue, return on total costs, return on assets. It is the most widely used method in Brazil and worldwide, given the abundance of comparables in public databases (Orbis, Compustat and similar).

Illustrative example

• A Brazilian distributor of a multinational group reports an operating margin of 1.8%;
• Benchmark with 9 comparable independent distributors: interquartile range of 3.1% to 6.4%, median 4.6%;
• 1.8% is outside the range → the tax authority adjusts the margin to the median (4.6%), and the 2.8-point difference on the transaction revenue becomes an additional IRPJ/CSLL base — with an ex officio penalty of 75% on the tax, increased to 100% (fraud/collusion) and 150% only in recidivism (Law 14,689/2023).

The choice and consistency of the PLI are decisive: the indicator must reflect the function tested (distribution, service, manufacturing) and be applied uniformly across the tested party and the comparables.

It is the method that appears most often precisely because it tolerates differences in product and accounting that would sink a PRL or an MCL — provided the functions are comparable. The trade-off is the dependence on databases and on comparable selection: a poorly assembled set, or a PLI ill-suited to the function tested, undermines the reliability of the entire study.

MDL (Profit Split) — Profit Split

It splits the combined profit (or loss) of the controlled transaction between the parties according to each one's relevant contributions — functions performed, assets used and risks assumed. It is the method for highly integrated operations, where both ends contribute intangibles or unique functions and neither can be tested in isolation by a one-sided method.

Illustrative example

• The parent owns the technology; the Brazilian subsidiary owns the local brand and the distribution network that builds the market;
• Combined profit of the product line in Brazil: R$40M;
• The contribution analysis (FAR + intangible assets) attributes 55% to technology development and 45% to local market exploitation;
• Arm's length result: R$22M for the parent and R$18M taxable in Brazil. If the pricing structure left only R$9M in the subsidiary, the difference is an adjustment.

The allocation key is the sensitive point — it requires documented economic grounding, not an arbitrary percentage.

There are two ways to apply MDL: the contribution analysis, which splits the total profit by each party's relative contributions; and the residual analysis, which first remunerates the routine functions through a one-sided method and only then splits the residual profit — the one tied to the unique intangibles — between the parties. It is the most labor-intensive method, and for that reason reserved for operations in which neither end can be tested in isolation.

Commodities and the “other methods”

Commodities have their own rule. Under art. 13 of Law 14,596/2023, where reliable information on comparable independent prices exists — including quoted prices on exchanges, research agencies or government agencies (art. 12) —, PIC is deemed the most appropriate method for the commodity transacted, unless robustly shown otherwise. The applicable quotation date must be supported by timely documentation; without it, the tax authority may adopt the quotation on the shipment date or on the date the import declaration was registered (art. 13, § 4).

Other methods (art. 11, VI). The law allows alternative methodologies when the five traditional methods cannot be reliably applied, provided the alternative produces a result consistent with what independent parties would do. But it is not a free sixth option: § 3 of art. 11 requires the documentation to demonstrate that the five methods (items I to V) are not applicable to the transaction. It is the safety valve for transactions without direct comparables (unique intangibles, restructurings), used sparingly and with reinforced grounding.

A worked case: one method per operation

Take an illustrative group, as if it were one of our clients, with four or five distinct intercompany flows. For each one, the most appropriate method is different — and that is the essence of the new regime.

• The subsidiary imports a commodity (resin) from the controlling company: there is a public quoted price → PIC (art. 13).
• It also resells finished products bought from the group, without transforming them: the gross resale margin is tested → PRL.
• Another entity manufactures to order for the group: the markup on costs is tested → MCL.
• The routine distributor, without reliable gross-margin comparables, is tested by its net margin → MLT, since it is the least complex tested party.
• And the product line in which parent and subsidiary contribute intangibles (technology and local brand) cannot be tested by a one-sided method → MDL.

Note that the same group uses several methods at once — one per operation. The mistake of those coming from the old regime is to try a single method for everything. The rule is: delineate each flow, identify the tested party and choose the method that the flow supports — justifying each choice in the Local File. (Illustrative figures.)

The practical takeaway is that transfer pricing under Law 14,596 is no longer a year-end calculation but a year-round discipline. Each new intercompany flow — a fresh product line, a financing arrangement, a service agreement signed mid-year — opens its own method question, and the answer has to be settled and documented while the facts are fresh, not reconstructed once an audit notice arrives. The TaxUp team treats the method matrix above as a living map of the group: every flow is tagged with its tested party, its chosen method and the comparables that support it, so that when the tax authority asks “why this method, and why this margin?”, the file already answers — flow by flow, in the taxpayer's own words rather than the auditor's.

Documentation: thresholds and penalties

The chosen method is only worth what is proven — and the proof has its own form and threshold. The documentation obligation (the Local File) scales with the volume of controlled transactions: below R$15 million in the year there is an exemption; between R$15 million and R$500 million the simplified version applies; from R$500 million onward, the complete version. The Master File and the Country-by-Country Report reach groups with consolidated revenue equal to or above €750 million.

Non-compliance has a price: the penalties of art. 35 of Law 14,596 run from a floor of R$20 thousand to a ceiling of R$5 million, depending on the failure (delay, omission, inaccuracy). And, on the base adjustment resulting from a poorly applied method, the ex officio penalty of 75% applies (Law 9,430/96, art. 44), raised to 100% when qualified and to 150% in case of recidivism (wording of Law 14,689/2023), in addition to Selic interest. In short: the method is the thesis; timely documentation is what sustains it in an audit.

It is worth being precise about what each threshold triggers. The R$15 million floor is an exemption from the Local File itself, not from the arm's length principle — a group below it still owes correct intercompany pricing and can be adjusted; it simply is not required to assemble the formal file. The simplified band (R$15 million to R$500 million) lightens the documentary burden but does not lower the substantive standard: the method still has to be the most appropriate one and the margin still has to be defensible. The complete file (from R$500 million) is where the full functional analysis, the comparables search and the quantitative demonstration of each method are expected in depth. The €750 million consolidated-revenue trigger for the Master File and the Country-by-Country Report is a separate, group-level test, independent of the per-entity Brazilian transaction volumes — a point frequently confused in practice.

An honest caveat is in order: there is still no consolidated CARF case law on the selection of the most appropriate method under Law 14,596 — the regime is recent (mandatory since 2024) and the existing precedents come from the old system (fixed margins), which do not carry over. For that reason, more than betting on a thesis, what protects today is the quality of the documentation: delineation, justified choice of method and a timely comparables study.