Picture yourself flipping through a group's annual report, past the consolidated revenue line, past the chairman's letter, past the segment breakdown. Somewhere near the back, one subsidiary is quietly absorbing losses the rest of the group never seems to feel. You almost missed it. You were supposed to.
It isn't an accident. It's architecture.
Transfer pricing is the system by which a multinational sets prices for transactions between its own subsidiaries: one entity sells components to another, licenses a brand name to a third, charges a fourth for management services. Because buyer and seller share the same ultimate parent, there is no arm's-length negotiation keeping those prices honest. The parent sets them. And the price it sets determines, almost mechanically, which legal entity ends up showing a profit and which ends up showing a loss.
The price is the decision
Consider a plausible but entirely recognizable scenario. A group manufactures consumer electronics. Its Irish subsidiary holds the group's intellectual property portfolio and licenses the brand and technology to an operating subsidiary in Germany, which actually builds and sells the products. If the Irish entity charges a royalty of 15% of German revenues, the German operation will show thin margins or outright losses even in a decent trading year. The Irish entity, meanwhile, books that royalty income against almost no costs. The loss doesn't disappear from the group. It migrates to the address on the invoice.
This is not inherently illegal. The arm's-length principle, the international standard endorsed by the OECD and written into most bilateral tax treaties, requires that intercompany prices reflect what two unrelated parties would have agreed. A legitimately valuable IP portfolio does command real royalties. The argument is always about the number, and that argument can run for years across two jurisdictions and three law firms before anyone blinks.
Tax authorities know this, which is why transfer pricing is the single most litigated area of international tax. The German finance ministry and the Irish Revenue Commissioners both have a stake in where the income sits. They will read the same royalty agreement and reach different conclusions about whether 15% is arm's-length or a polite fiction.
What most people get wrong
The popular assumption is that multinationals use transfer pricing exclusively to shift profits into low-tax jurisdictions. That happens, and it deserves the scrutiny it gets. But the loss side of the ledger is less discussed and equally deliberate.
A group facing a downturn in a particular market may have strong incentives to concentrate losses in a subsidiary where those losses generate the most valuable tax shield: a jurisdiction with a corporate rate of 25 or 30%, where a loss carryforward offsets future profits at real money rather than at 12.5 cents on the dollar. Losses, in other words, are assets. The group's treasury and tax teams think carefully about where to park them, and the parking is rarely random.
There is also the question of functional analysis. Transfer pricing rules don't just examine the price. They look at who performs the functions, who owns the assets, who bears the risks. A subsidiary contractually stripped of meaningful decision-making, operating purely as a contract manufacturer or a limited-risk distributor, is supposed to earn only a thin, stable return. It shouldn't bear market losses, because it has also been stripped of the upside. Think of it like a salaried employee being asked to cover the firm's bad bets: the risk profile and the reward profile have been deliberately decoupled. When a limited-risk distributor in France suddenly absorbs the losses from a bad product launch, a sharp-eyed examiner from the Direction générale des finances publiques will ask why an entity that bore no entrepreneurial risk is now paying for an entrepreneurial failure.
Here is how it plays out in practice. Take two colleagues, Marcus and Yuki, who joined the same multinational in the same year. Marcus runs the Swiss procurement hub. Yuki manages the Australian distribution entity. In a year when commodity costs spike and Australian sales disappoint, the group's internal pricing keeps Marcus's hub profitable, its margins protected by a cost-plus agreement locked in eighteen months earlier. Yuki's entity, structured as a full-risk distributor, absorbs the entire margin compression. Both subsidiaries belong to the same parent. Only one of them is having a bad year on paper, and the bad year was written into the contract before the year even started.
The practical consequence is serious. A full-risk subsidiary is genuinely exposed: it can go technically insolvent while the parent group remains healthy. Local creditors, local employees, and local tax authorities all find themselves dealing with a loss-making entity whose troubles were, in a meaningful sense, designed into its intercompany agreements. That is not a technicality. It is a transfer of economic consequence from one set of stakeholders to another, dressed up in pricing schedules.
So when you find the loss center in a group you are analyzing, the first number to interrogate is not the loss itself. It is the royalty rate, the service fee, the transfer price that produced it. If that rate was set without reference to any comparable uncontrolled transaction, you are not looking at a business that struggled. You are looking at a business that was structured to struggle.
Country-by-country reporting requirements, now embedded in most major jurisdictions, exist precisely to make that question answerable. They don't always make it easy. But they make it considerably harder to pretend the architecture was natural, and that harder conversation is exactly the one the loss-bearing subsidiary's local tax authority is now equipped to have.