Transfer pricing and profit shifting

Transfer pricing is not, by itself, illegal. Every multinational enterprise (MNE) with related entities in more than one country must price the transactions between those entities — the sale of goods, the licensing of a trademark, the provision of management services, the extension of an intercompany loan, and tax law in almost every jurisdiction requires that price to be set as if the parties were unrelated: the arm's length principleArm's length principleThe requirement that related-party transactions be priced as if between independent parties dealing at market terms.. What I teach officials to internalise on day one is that transfer pricing abuse is not a distinct crime with its own smoking gun; it is ordinary commercial documentation, professionally prepared, that quietly prices intra-group transactions to move profit from a high-tax or high-risk jurisdiction to a low-tax one. The abuse lives in the assumptions behind a valuation, not in a forged invoice.

The three classical channels are services, royalties, and debt. Intra-group SERVICE FEES; management fees, technical assistance fees, "head office recharges", are the most common and least policed leakage point in developing-country tax administrations, because a genuine cost is genuinely incurred somewhere in the group and the only live question is how much of it, and at what mark-up, should be allocated to the local subsidiary. A parent company in a low-tax jurisdiction can charge a mining subsidiary in Zambia or the DRC a management fee of 3-5% of revenue for "strategic oversight" that in substance duplicates functions the local subsidiary already performs, or that confer no identifiable benefit at all. I have reviewed files where the same three-page "global services agreement" was used, word for word, to justify management charges to a dozen unrelated subsidiaries across three continents — a strong signal the fee was set to a target profit outcome rather than to a benefit actually received.

ROYALTIES for the use of trademarks, patents, know-how and software are the second channel, and the one that has done the most long-run damage to source-country tax bases because intangible property is mobile in a way a mine or a factory is not. A group can develop intellectual property centrally, migrate legal (though rarely economic) ownership of that IP to a low-tax entity through a cost-sharing or buy-in arrangement, and then charge every operating subsidiary a royalty — commonly 2-8% of net sales, for the right to use a brand or process the operating subsidiary itself helped build. The OECD's BEPS Actions 8-10 (Aligning Transfer Pricing Outcomes with Value Creation, finalised 2015 and folded into the 2022 OECD Transfer Pricing Guidelines) were a direct response to this: they insist that legal ownership of IP is not, by itself, sufficient to earn the IP return; the entity performing the DEMPE functionsDEMPE functionsDevelopment, Enhancement, Maintenance, Protection and Exploitation of an intangible — the OECD's test for which entity in a group is entitled to the IP return.; Development, Enhancement, Maintenance, Protection and Exploitation of the intangible, is the one entitled to the residual profit. An IP-holding company with no scientists, no marketing staff and no decision-making capacity is, under the DEMPE framework, entitled to no more than a routine return on the limited functions (typically custodial or financing) it actually performs, however impressive its royalty invoices look on paper.

THIN CAPITALISATIONThin capitalisationFunding a subsidiary disproportionately with related-party debt rather than equity to maximise deductible interest. and interest stripping are the third channel. A group funds its local subsidiary disproportionately with related-party debt rather than equity, because interest is a deductible expense while dividends are not. A subsidiary that would never obtain a comparable loan from an independent bank — because its balance sheet cannot service the debt, or because the loan terms bear no relationship to market conditions — nonetheless carries an intercompany loan at 12-15% interest from a treasury company in a jurisdiction that taxes interest income lightly or not at all. The interest deduction erodes the local tax base every year for the life of the loan, without a single cross-border cash movement that looks unusual on a bank statement. Most African VAT and income tax statutes now contain a fixed-ratio interest-limitation rule (commonly capping deductible related-party interest at 30% of EBITDA, following BEPS Action 4), but enforcement requires the auditor to actually test whether the debt itself would have been extended, and on what terms, between independent parties, a debt-versus-equity characterisation exercise that goes well beyond checking the interest rate against a benchmark.

COMMODITY MISPRICING is the channel with the deepest resonance for African revenue authorities, because it attaches directly to the extractive sector that dominates so many national tax bases. A mining company in Zambia or the DRC sells copper concentrate or cobalt to a related offshore marketing hub at a price below the prevailing exchange-quoted benchmark, or on terms (quality discounts, deductions for "penalty elements", freight allowances) that are opaque and difficult for a resource-constrained tax administration to verify against the actual physical characteristics of the shipment. The marketing hub then on-sells the same cargo, often within days and without taking physical possession, to the ultimate buyer at the full market price, booking the margin in a jurisdiction with negligible tax. South Africa's SARS and the Zambia Revenue Authority have both run high-profile transfer-pricing audits against exactly this structure in copper and platinum group metals; Zambia's introduction of a mineral-royalty regime with reference pricing, and its adoption of the London Metal Exchange (LME) quotational period methodology as a default benchmark, were direct legislative responses to the difficulty of contesting mispriced concentrate sales transaction by transaction.

The unifying analytical point across all four channels is that transfer pricing abuse rarely requires deception in the conventional AML sense; no fake invoice, no shell company hiding beneficial ownership, sometimes not even a jurisdiction most people would call a "tax haven". It requires only a valuation assumption favourable to the group, defensible on paper by an expensive advisory firm, and difficult for an under-resourced revenue authority to contest without its own comparably sophisticated economic analysis. That asymmetry of capacity, Big Four transfer-pricing practice on one side of the table, a handful of trained transfer-pricing auditors on the other — is, in my own casework across the region, by some distance the most important explanatory variable for why profit shifting persists at the scale it does in resource-rich developing economies, far more than any gap in the legal rules themselves.

The policy response has evolved along two tracks. The first track, the arm's length principleArm's length principleThe requirement that related-party transactions be priced as if between independent parties dealing at market terms. itself, has been reinforced and refined through the OECD Transfer Pricing Guidelines (2022 edition consolidates all BEPS-related revisions) and through the UN Practical Manual on Transfer Pricing for Developing Countries, which explicitly addresses commodity transactions, intra-group services and limited-risk distribution structures from a source-country perspective. The second track accepts that the arm's length principleArm's length principleThe requirement that related-party transactions be priced as if between independent parties dealing at market terms. is administratively unworkable for some transaction types in low-capacity environments and substitutes simpler formulary or presumptive approaches: Zambia's mineral royalty reference pricing, the Sixth MethodSixth MethodA presumptive commodity-pricing method benchmarking export prices to quoted exchange prices on the shipment date, used widely in Latin America and referenced by ATAF. used across much of Latin America and increasingly referenced by ATAF for commodity exports, and the safe-harbour regimes several African revenue authorities have adopted for routine service and distribution transactions. Officials need fluency in both tracks, because the choice between litigating an arm's length adjustment and applying a statutory presumptive rule is itself a strategic decision about the burden of proof, the resource cost of the audit, and the durability of the outcome on appeal.

The OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations, most recently consolidated in the 2022 edition; remain the reference architecture almost every jurisdiction, OECD member or not, uses to structure its domestic transfer-pricing rules. The Guidelines set out five accepted methods for testing whether a related-party price is arm's length: the Comparable Uncontrolled Price (CUP) method, which benchmarks the tested transaction directly against a comparable transaction between independent parties; the Resale Price Method, used mainly for distribution functions; the Cost Plus Method, used mainly for contract manufacturing and low-risk service provision; the Transactional Net Margin Method (TNMMTNMMTransactional Net Margin Method, tests a tested party's net profit indicator against a range derived from comparable independent companies.), by far the most commonly applied in practice because it tests a net profit indicator (such as operating margin or return on costs) against a range derived from a database of comparable independent companies rather than requiring a comparable transaction, and the Profit Split Method, reserved for cases where both parties to a transaction make unique and valuable contributions that cannot be reliably tested on a one-sided basis, such as an integrated global commodity-trading and processing operation.

For a revenue auditor, method selection is itself a battleground. Taxpayers overwhelmingly favour TNMMTNMMTransactional Net Margin Method, tests a tested party's net profit indicator against a range derived from comparable independent companies. because it is forgiving: a subsidiary can post a profit margin anywhere within an "interquartile range" of comparable companies and be defensible, and the comparable-company searches that generate that range are opaque, expensive to replicate, and easily skewed by the choice of search criteria, geographic scope and rejection matrix. I tell officials never to accept a TNMMTNMMTransactional Net Margin Method, tests a tested party's net profit indicator against a range derived from comparable independent companies. benchmarking study at face value; always re-run the comparable search yourself, using the same commercial database if your administration has a licence (Bureau van Dijk's Orbis and Moody's/ORBIS successors are the most common), and interrogate every company the taxpayer's advisor excluded, exclusions are where the manipulation usually lives, far more than in the median calculation itself.

The SIXTH METHOD (sexto método) originated in Argentina in the early 2000s specifically to address commodity export mispricing, and has since been adopted in some form by Brazil, Peru, Ecuador, Bolivia, Uruguay and Paraguay, among others. Instead of requiring the tax administration to find comparable transactions or comparable companies — an exercise that is close to impossible for globally traded commodities where the "market" is the commodity exchange itself — the Sixth Method deems the arm's length price to be the quoted price of the commodity on a recognised exchange (LME, COMEX, ICE) on the date of shipment, subject to adjustment for quality, volume and delivery terms that the taxpayer must substantiate. This inverts the burden of proof: rather than the revenue authority having to prove the related-party price was wrong, the taxpayer must justify any departure from the quoted benchmark. ATAF's technical notes on transfer pricing for the extractive sector, published for its African member administrations, explicitly recommend variants of this approach for administrations with limited capacity to run full functional and comparability analyses on every commodity export, while cautioning that a poorly drafted Sixth Method provision can be challenged as inconsistent with tax treaty non-discrimination or associated-enterprise articles if it does not preserve a genuine arm's length outcome.

PILLAR TWO of the OECD/G20 Inclusive Framework's Two-Pillar Solution introduces a wholly different, and newer, constraint on profit shifting: the Global Anti-Base Erosion (GloBE) rules, which impose a minimum effective tax rate of 15% on the profits of large MNE groups (those with consolidated revenue above EUR 750 million) on a jurisdiction-by-jurisdiction basis. GloBE rulesGloBE rulesGlobal Anti-Base Erosion rules under OECD Pillar Two, imposing a 15% minimum effective tax rate on large MNE groups, in force from 2024. began entering into force from 2024 in the EU (via the Pillar Two Directive) and a growing number of other jurisdictions, with the Income Inclusion Rule (IIR) and the Undertaxed Profits Rule (UTPR) operating as backstops: if a low-tax jurisdiction does not itself impose a top-up tax to bring the effective rate to 15%, another jurisdiction in the group's structure can collect the shortfall instead. For a developing-country administration, Pillar Two changes the profit-shifting calculus in an important but limited way, it constrains the incentive to book profit in a zero-or-near-zero-tax jurisdiction, because a top-up tax will likely be collected elsewhere in the group regardless, but it does nothing to protect a SOURCE country's own tax base if the group simply reallocates the shifted profit to a jurisdiction taxing it at, say, 16%, still well below what many developing-country statutory rates would have captured domestically. This is precisely why ATAF, the UN Tax Committee and a wide coalition of developing countries pushed hard, and continue to push through the UN Framework Convention on International Tax Cooperation negotiation process (terms of reference adopted 2024, substantive negotiations running to 2027), for a genuinely inclusive successor process to the OECD-led Inclusive Framework; one in which source-country taxing rights, not just minimum-tax backstops, are centred in the design.

A related and often underappreciated instrument is the Qualified Domestic Minimum Top-up Tax (QDMTTQDMTTQualified Domestic Minimum Top-up Tax — allows a low-tax jurisdiction to collect the Pillar Two top-up tax itself rather than ceding it to another jurisdiction.), which allows the LOW-TAX jurisdiction itself, rather than a foreign parent or sibling jurisdiction under the IIR/UTPR — to collect the Pillar Two top-up tax on profits booked within its own borders. Several jurisdictions historically used as profit-shifting destinations have adopted QDMTTs specifically to keep the top-up tax revenue for themselves rather than ceding it under the backstop rules; this is a legitimate and, from a source-country perspective, largely neutral development, but officials should understand that a QDMTTQDMTTQualified Domestic Minimum Top-up Tax — allows a low-tax jurisdiction to collect the Pillar Two top-up tax itself rather than ceding it to another jurisdiction. collected in the destination jurisdiction does nothing to restore lost taxing rights to the source country where the underlying value was actually created.

Practically, when I train a revenue authority's transfer-pricing unit to build an adjustment case, I insist on five sequential steps, in this order and no other. First, accurately delineateAccurately delineateThe OECD's requirement to characterise a related-party transaction by its actual substance and conduct, not merely its contractual label. the actual transaction — not the transaction as labelled in the intercompany agreement, but the transaction as it is actually conducted in substance, including who performs which functions, who bears which risks, and who controls the risk-mitigation decisions (the OECD's 2017 "accurately delineateAccurately delineateThe OECD's requirement to characterise a related-party transaction by its actual substance and conduct, not merely its contractual label." guidance, carried into the 2022 edition, is decisive here and has been used to recharacterise entire structures where the contractual allocation of risk did not match the substance). Second, select the most appropriate method given the delineated transaction, not the method the taxpayer's study assumed. Third, build or independently verify the comparable set. Fourth, quantify the adjustment with a clear, reproducible calculation the tribunal can follow. Fifth, and this is the step most audits skip, test the adjustment against the group's own internal documents, board minutes, budget variance analyses, internal transfer-pricing policy memos obtained through information requests or exchange of information; because taxpayers' own internal commercial reasoning frequently contradicts the arm's length narrative constructed for the tax file, and that contradiction is the single most persuasive piece of evidence in a contested transfer-pricing case.

Every transfer-pricing adjustment a revenue authority raises must eventually survive one of three tests: a negotiated settlement, a domestic tax tribunal or court, or a Mutual Agreement Procedure (MAPMAPMutual Agreement Procedure — a treaty mechanism for resolving double taxation disputes between two states' competent authorities.) under a double tax treaty where the taxpayer alleges double taxation. Building a case that survives all three requires discipline well beyond running a comparable-company search; it requires a case file that can be handed to a judge who has never studied economics and still be persuasive.

The starting point in any real audit is the FUNCTIONAL ANALYSISFunctional analysisThe process of identifying which party to a transaction actually performs functions, bears risks, and controls risk-mitigation decisions. INTERVIEW, conducted on-site wherever possible, not by correspondence. I insist that trainee auditors physically walk the operation before touching a spreadsheet: stand in the concentrate weighbridge, sit in on a pricing call between the local subsidiary and the offshore marketing hub, read the actual emails (not the summarised minutes) in which pricing and quality-discount decisions are made. The functional analysisFunctional analysisThe process of identifying which party to a transaction actually performs functions, bears risks, and controls risk-mitigation decisions. interview should establish, concretely, who negotiates the sales contract, who bears the price risk between contract date and shipment date, who insures the cargo, who decides on quality-discount disputes with the buyer, and who absorbs the cost when a shipment is rejected or downgraded on assay. In case after case I have reviewed, the local subsidiary performs every one of these functions and bears every one of these risks, while the offshore marketing entity does nothing but issue an invoice — precisely the fact pattern the OECD's 2017 risk-and-substance guidance was designed to unwind, because a party that does not actually control risk cannot be allocated the return associated with bearing that risk, however the contract is drafted.

Once the functional reality is documented, the audit team must construct an INDEPENDENT BENCHMARK, not just critique the taxpayer's study. Passive rebuttal, arguing that the taxpayer's comparable set is wrong without proposing a superior alternative; rarely survives appeal, because tribunals in most jurisdictions place at least an evidential, and sometimes a formal, burden on the tax authority once the taxpayer has produced a facially compliant transfer-pricing study. For commodity transactions this benchmark is the quoted exchange price on the appropriate quotational periodQuotational periodThe period around shipment date over which a commodity's exchange-quoted price is averaged to set the settlement price., adjusted only for verified, arm's-length-consistent deductions; auditors should obtain the exchange settlement data directly (LME publishes historical settlement prices publicly) rather than relying on the figure the taxpayer's own pricing desk supplies. For service and royalty transactions, the benchmark is a properly constructed comparable search using the same commercial database class the taxpayer used, with documented search criteria, and, crucially — with the auditor's own judgment applied to inclusion and exclusion decisions rather than simply adopting the taxpayer's rejection matrix.

QUANTIFICATION must be transparent and arithmetically reproducible from primary data: shipment-by-shipment tonnage, assay results, exchange settlement prices for the correct quotational periodQuotational periodThe period around shipment date over which a commodity's exchange-quoted price is averaged to set the settlement price., and the resulting price differential multiplied through actual sales volumes. I have seen strong technical cases collapse at tribunal because the quantification schedule presented in court could not be reconciled, line by line, back to the shipping documents and exchange data — the tribunal simply could not verify the number, and where a tribunal cannot verify a number it will not uphold it, however sound the underlying economics.

The taxpayer's most common and most effective defence is the COMPARABILITY CHALLENGE: arguing that the auditor's benchmark comparables differ materially in function, risk, market, or contractual terms from the tested transaction, such that no reliable adjustment can be derived. The correct response is not to abandon comparability analysis but to narrow it: apply comparability adjustments (for example, for differences in payment terms, volume, or contractual delivery point) rather than discarding an otherwise sound benchmark, and to document, contemporaneously, why each adjustment was made and how it was quantified. A second common defence is the DOUBLE TAXATION / MAPMAPMutual Agreement Procedure — a treaty mechanism for resolving double taxation disputes between two states' competent authorities. THREAT: the taxpayer signals it will invoke the Mutual Agreement Procedure under the relevant double tax treaty, arguing the adjustment creates double taxation because the offshore marketing hub has already been taxed, in its own jurisdiction, on the profit the source country now seeks to reallocate. Officials should not be deterred by this threat, but should prepare the case file to withstand a MAPMAPMutual Agreement Procedure — a treaty mechanism for resolving double taxation disputes between two states' competent authorities. process from the outset: MAPMAPMutual Agreement Procedure — a treaty mechanism for resolving double taxation disputes between two states' competent authorities. competent authorities on both sides will expect exactly the functional analysisFunctional analysisThe process of identifying which party to a transaction actually performs functions, bears risks, and controls risk-mitigation decisions., benchmark methodology and quantification discipline described above, and a case built loosely for domestic litigation alone frequently fails when it reaches the more technically demanding MAPMAPMutual Agreement Procedure — a treaty mechanism for resolving double taxation disputes between two states' competent authorities. forum.

South Africa's SARS, working within the ESAAMLG mutual-evaluation framework and drawing on ATAF technical assistance, has developed a reasonably mature transfer-pricing audit capability in the extractive and financial-services sectors, and its published case outcomes (where public, given taxpayer confidentiality constraints) are instructive for training purposes even where full facts are not disclosed. Zambia's revenue authority has pursued a hybrid strategy, combining Sixth-Method-style reference pricing in its mineral royalty regime with conventional arm's length transfer-pricing audits for services and financing transactions; precisely because a single methodology cannot address the full range of profit-shifting channels present in a mining-dependent economy. The DRC, by contrast, still faces acute capacity constraints in its cobalt and copper sectors, and ATAF's regional technical assistance programmes have prioritised exactly the kind of functional-analysis and benchmark-construction training described in this lesson as the highest-value intervention available, because legal reform without audit capacity produces rules that exist on paper only.

Finally, officials should treat a transfer-pricing adjustment as the beginning, not the end, of the analytical chain. A profit-shifting structure detected through a transfer-pricing audit frequently sits alongside, and sometimes conceals — outright laundering or corruption exposure: the offshore marketing hub receiving the mispriced margin is very often beneficially owned, in whole or in part, by individuals connected to the local operation, including politically exposed persons. Every transfer-pricing case file should therefore be cross-referred, as a matter of routine practice, to the financial intelligence unit and to beneficial-ownership registry checks under the FATF R.24/R.25 framework, because the tax adjustment recovers revenue for the treasury, but the beneficial-ownership trace is what may ultimately expose the underlying corruption or laundering scheme the mispricing was designed to fund.