The Arm's-Length Fiction and Why It Frayed

The organising idea of the twentieth-century international tax settlement has a name, and it is worth saying plainly: the arm's-length principle. When two subsidiaries of the same group trade with each other, the tax authorities of the countries involved agree to price that internal trade as if the two entities were unrelated parties bargaining at arm's length. A subsidiary in one country selling components to a sister subsidiary in another is supposed to charge roughly what it would charge a stranger. Tax the profit that results, country by country, and in principle each jurisdiction gets its fair share of a global enterprise's income.

The principle is old, codified through the League of Nations in the 1920s and 1930s and carried forward by the Organisation for Economic Co-operation and Development — the OECD, the club of mostly rich economies that has done more than any other body to write the rules of cross-border taxation. For manufactured goods it worked tolerably. If a stranger would pay a known price for a tonne of steel or a crate of engine parts, an auditor could check the internal price against that benchmark and object when it drifted.

The trouble is that the benchmark evaporates precisely where modern value concentrates. What is the arm's-length price for the right to use a search algorithm that has no competitor, or a brand that exists nowhere else, or a data set assembled from a billion users? There is no stranger who sells the comparable thing, because there is no comparable thing. The most valuable assets in the contemporary economy are unique by construction — that is what makes them valuable — and uniqueness is exactly the condition under which "what would an unrelated party pay" has no honest answer. Into that vacuum flowed a great deal of discretion, and discretion, applied by well-advised firms, tends to flow downhill toward the lowest rate.

Here is the mechanism in its plainest form. A group locates the ownership of its crown-jewel intangibles — patents, trademarks, code — in a subsidiary resident in a low-tax jurisdiction. That subsidiary licenses the intangibles to the operating companies in high-tax markets, the ones that actually face customers. The operating companies book large royalty payments as costs, which shrink their taxable profit at home, and those royalties arrive as income in the low-tax entity, where they are lightly taxed or not taxed at all. No factory moved. No engineer relocated. The value was always being created by the users and staff in the big markets. But the profit, following the licence agreement, resides somewhere small and lightly taxed.

Stack a few of these moves and you get the structures that became notorious in the 2010s — arrangements routing income through one low-tax European jurisdiction and out to a zero-tax island, exploiting mismatches between how two countries defined tax residence. They had cute nicknames. What matters here is not the choreography but the pattern: profit was being allocated by contract, not by geography, and the arm's-length principle lacked the grip to stop it, because the contracts priced things that had no market price.

Governments noticed, and not only for revenue reasons. There is a legitimacy cost when citizens see a company earn conspicuously in their country and pay conspicuously little there. The OECD's response, launched in 2013, went by the label Base Erosion and Profit Shifting — the idea that a tax base was being eroded and profit shifted away from where activity happened. That project ran for years and produced a great deal, most of which tightened the existing arm's-length system: better documentation, disclosure of where a group books income country by country, rules against the most artificial mismatches. Useful plumbing. But it accepted the basic architecture — tax each entity separately, price the internal trades — and therefore inherited its core weakness. As long as unique intangibles could be owned in a low-tax box and rented to everyone else, the incentive to shift profit survived every patch.

That is the unfinished business the next module's subject — a coordinated global minimum tax — was designed to address. Not by fixing the arm's-length price of the unpriceable, which may be impossible, but by changing the payoff. If profit lands somewhere lightly taxed, another country gets to charge a top-up. Suddenly the low-tax box is worth less. But before we get there, we need to sit properly with transfer pricing itself, because you cannot understand why a global floor is clever until you understand exactly what it is a floor under. Faraday's read: the arm's-length principle did not fail because auditors were lazy. It failed because it asked a question — what would strangers charge? — that has no answer for the assets that now carry the most value. The system's designers built for a world of things and inherited a world of rights.