The price tag nobody outside the company gets to check

You are standing in a distribution warehouse outside Düsseldorf. A pallet of electronic components has just cleared customs. The paperwork is immaculate. The declared value is precise to two decimal places. And yet the number on that invoice was set by the same corporate group that will receive the goods, book the revenue, and file the tax return. No market negotiated it. Nobody outside the building had a vote. That number, quiet and administrative as it sounds, is one of the most consequential figures in international tax.

This is transfer pricing. It sits at the heart of how multinational firms legally shift profits from high-tax countries to low-tax ones. The mechanism is not exotic. It is boringly structural, which is exactly why it persists.

How the internal price becomes a tax lever

When two unrelated companies trade, the price is whatever they agree on. When two subsidiaries of the same parent trade, they must also agree on a price, but on what basis? The internationally accepted standard, codified in the OECD's Transfer Pricing Guidelines and embedded in the domestic law of most developed economies, is the arm's-length principle: the internal price should be what two independent parties would have agreed.

In practice, that standard is slippery.

Consider a pharmaceutical company that develops a drug compound at its research centre in the United States, then, before the drug earns significant revenue, transfers the intellectual property rights to a subsidiary in Ireland or Luxembourg. The subsidiary licenses that IP back to operating companies around the world. Every time a pill is sold in Germany or Australia, a royalty payment flows to the low-tax jurisdiction, reducing taxable profit where the patients are and accumulating it where the rate is lower. The arm's-length price for early-stage pharmaceutical IP is genuinely hard to establish, because by definition there is no market of independent buyers and sellers for a novel, unproven compound. That ambiguity is the opening, and sophisticated firms walk through it deliberately.

The customs angle adds a second layer. Customs duties are typically levied as a percentage of the declared transaction value of goods crossing a border. The WTO Customs Valuation Agreement, which governs how most countries assess those duties, also starts from the transaction value between buyer and seller. For related parties, customs authorities are supposed to scrutinise whether the declared price was influenced by the relationship. But customs agencies and tax authorities in most countries operate in separate silos. A firm might declare a low value for a component at the customs border, minimising import duty, while simultaneously justifying a high transfer price for the same component to its tax authority, maximising the deduction in a high-tax country. The two numbers do not have to match. Often they don't.

A worked example that shows the arithmetic

Take a plausible, simplified scenario. A multinational manufactures electronic components in Malaysia at a cost of roughly 40 dollars per unit. The Malaysian entity sells them to a group trading hub in Singapore at 42 dollars. Singapore sells them to the German distribution subsidiary at 95 dollars. Germany sells to end customers at 100 dollars.

The profit at each stage: Malaysia earns 2 dollars per unit, taxed at Malaysia's rate. Singapore earns 53 dollars per unit, taxed at Singapore's relatively low corporate rate. Germany earns 5 dollars per unit, taxed at Germany's higher rate. The group's total profit is 60 dollars per unit regardless, but the majority of that profit surfaces in Singapore, not Germany, where it would otherwise naturally accumulate given that Germany is where the customers, the sales force, and the market relationships actually live.

Now ask: is 95 dollars an arm's-length price for those components? Maybe. Component pricing in electronics is genuinely complex, and Singapore trading hubs perform real functions. The German tax authority would need comparable transaction data, transfer pricing documentation, and the will to litigate to successfully challenge it. Many don't have all three. The whole arrangement sits inside the rules like a hand inside a glove.

What the rules actually require, and where they bend

The OECD guidelines offer several approved methods for establishing arm's-length prices: comparable uncontrolled price (find a genuinely similar transaction between independents), cost-plus (add a standard markup to production cost), resale-price minus (work back from the final sale), and transactional net margin methods. Each has legitimate uses. Each has failure modes.

The comparable uncontrolled price method is the most rigorous but requires a genuinely comparable transaction, which is rare for proprietary goods or unique IP. Cost-plus is easier to apply but the choice of markup percentage is contested. Transactional net margin methods, which compare a party's operating margin against industry benchmarks, are now the most widely used and also the most gameable, because the selection of comparable companies is discretionary.

Here is the wrinkle that most coverage misses: transfer pricing disputes are not usually about fraud. The firm is not falsifying documents. It is making defensible, documented choices within a framework that genuinely permits a range of outcomes. A tax lawyer once described this range as a "pricing corridor" rather than a single correct number. The firm picks a point in the corridor. The tax authority picks a different point. They litigate or settle. This is the ordinary condition of international corporate taxation, not an exceptional scandal.

What people get wrong about this

The most common misreading is that this is simply tax evasion dressed in a suit. It isn't, and conflating the two makes it harder to understand what reform would actually require. Evasion is illegal. Transfer pricing, even aggressive transfer pricing, is mostly legal. The rules were written to accommodate it, in part because the arm's-length standard is genuinely difficult to apply to integrated multinationals whose value comes precisely from being integrated. Calling it evasion lets the actual architecture off the hook.

A second misconception: that customs valuation and corporate income tax are separate problems with separate solutions. They are not. The same internal invoice drives both calculations, and reform in one domain without coordination in the other creates new arbitrage opportunities. The OECD's Base Erosion and Profit Shifting project, which produced a global minimum tax framework, addresses the income tax side. Customs valuation reform has moved more slowly, partly because customs revenue is less politically salient than income tax in wealthy countries, and partly because import-competing industries in those countries sometimes benefit from artificially low declared values.

The third thing people get wrong: assuming size is the only factor. A mid-sized firm with a single Irish holding company and a carefully structured IP licence can shift a substantial portion of its profits just as effectively as a tech giant, and will attract far less scrutiny. The mechanism scales down. It scales down further than almost anyone in the policy conversation acknowledges.

So ask yourself: if the global minimum tax sets a floor on the rate, but leaves the valuation at the border entirely untouched, what exactly has been fixed? The rate at the destination matters less if the profit never arrives there in the first place. The shirt is still priced by the person selling it to themselves.