The Price of a Sale to Yourself
Every large multinational faces a question that sounds absurd until you sit with it: what should one part of the company charge another part for goods, services, or the use of an idea? A carmaker's Slovak assembly plant receives engines from the group's German engine works. A retailer's national subsidiaries pay the group's Luxembourg entity for the use of the brand. A software group's Irish company licenses code owned by a Bermudan affiliate. In each case money crosses a border between two entities under the same ultimate ownership, and the price attached to that crossing decides which country gets to tax the resulting profit. This is transfer pricing, and it is the machinery beneath everything in the last module.
The reason a made-up-sounding internal price has real teeth is that tax is levied entity by entity, country by country. Raise the price at which the German works sells engines to the Slovak plant and you move profit from Slovakia to Germany. Lower it and profit moves the other way. Multiply that across thousands of internal transactions in a group spanning fifty countries, and the aggregate placement of profit — and therefore the aggregate tax bill — is substantially a function of the prices the group sets on trades with itself. Tax authorities know this, which is why transfer pricing generates more cross-border disputes than almost any other issue in the field. Two treasuries are, in effect, arguing over the same slice of profit: each wants it taxed on its side of the line.
The agreed referee, as we saw, is the arm's-length principle: price the internal trade as unrelated parties would. For commodities and standard components this is checkable. Auditors assemble comparables — records of what independent firms charged for similar things — and test the internal price against that range. The OECD publishes methods with plain enough logic: compare the price directly against market transactions where you can; where you cannot, compare the margin the entity earns against margins earned by independent companies doing similar work; in the hardest cases, split the combined profit between the entities according to their relative contributions. None of this is mysterious. It is the everyday grind of the discipline, and for tangible goods it produces defensible answers.
Where it strains is the same place the whole system strains: functions that are unique, mobile, and hard to value. Three keep recurring. First, intangibles — patents, trademarks, copyrights, know-how — which we have met and will return to, because there is no market price for a one-of-a-kind asset. Second, intra-group services: management, legal, technical support, the shared overhead of running a global firm. What is the arm's-length fee for the head office's oversight of a subsidiary? Reasonable people, and their advisers, differ by a lot. Third, intra-group finance: when one entity lends to another, the interest rate on that loan is a transfer price too. Route a group's borrowing through a high-tax country and the interest deductions shelter income there; park the lending entity somewhere low-tax and the interest income arrives lightly taxed. Debt, priced internally, becomes another lever on where profit sits.
Notice that none of these levers requires deception. A group can document every transaction, retain expensive advisers to defend every price as within a plausible arm's-length range, and still end up with profit concentrated in low-tax entities — because the ranges for unique intangibles, discretionary services, and internal loans are genuinely wide, and a firm is entitled to sit at the end of the range that suits it. That is the uncomfortable centre of the subject. The scandalous cases involve prices no honest analyst would defend. But a great deal of profit shifting happens entirely within the rules, using the latitude the rules cannot help but leave. This is why patching the arm's-length system yielded diminishing returns, and why the eventual response changed the payoff rather than the pricing.
For your own analysis, the practical instruction is to decompose. When you encounter a multinational's structure, separate the physical from the fiscal. Ask where the tangible functions sit — the plants, the data centres, the staff who face customers — and then, separately, ask where the intangibles are owned and where the internal financing runs. The distance between those two answers is the transfer-pricing story. It will not always indicate abuse. But it will always tell you how the company has arranged, through the prices it charges itself, to have its profit recognised somewhere other than where its concrete stands.