For about forty years the most lucrative trick in international business was a filing cabinet in a quiet jurisdiction. You set up a company in a place with a low tax rate and a good treaty network, you routed your profits through it on paper, and the company itself was barely a company at all. A brass plaque in a corridor in Amsterdam or Luxembourg. A local director who sat on four hundred other boards and signed whatever was put in front of him. A bank account and an address and almost nothing else. The money flowed through a country it never really touched, picked up a tax advantage on the way, and carried on. People in the trade called them letterbox companies, and for a long time they were the load-bearing wall of cross-border structuring.
That wall is coming down, and the interesting thing is not that regulators knocked it down. It's that they did it without banning anything. They simply made the trick require something it had always pretended to have and never did: substance.
Understand what the old structure was actually selling. A treaty between two countries typically lowers or removes the tax that one country charges when money leaves it for the other. Interest, dividends, royalties — all of it normally taxed on the way out, taxed less or not at all if it's going to a treaty partner. So if you're a company in a high-tax country sending royalties to a parent in another high-tax country, you don't send them directly. You build a company in a third country that has a generous treaty with both, and you route the royalties through it. The money takes a detour through a jurisdiction it has no real business being in, purely to wear that country's treaty like a coat. This is called treaty shopping, and the conduit company in the middle was, in most cases, a letterbox. It employed no one. It decided nothing. It existed to be a coat hook.
The whole edifice rested on a polite fiction that everyone involved understood to be a fiction: that the conduit company was a real business making real decisions in the country where it was registered. Tax authorities knew it wasn't. The companies knew it wasn't. For decades the fiction held because challenging it was slow, expensive, and case-by-case, and the law gave authorities no clean tool to say "this company is a coat hook, ignore it."
Then, over roughly the last decade, they built the tool. Not one tool, several, layered.
A wave of jurisdictions, including the ones whose entire business model was hosting letterboxes, passed laws saying that a company claiming to be resident there for tax purposes had to actually do something there. Real offices. Real employees, in real numbers, making real decisions. Adequate operating expenditure that lands in the local economy. The Cayman Islands, the BVI, Jersey — the classic places — all introduced economic substance requirements, partly under pressure from the EU's list of non-cooperative jurisdictions, a blacklist with real financial consequences for the places that land on it. Suddenly the brass plaque was not enough. If your company in jurisdiction X wanted the benefits of being in jurisdiction X, it had to genuinely be in jurisdiction X, with the payroll and the lease to prove it.
A multilateral instrument, signed by most of the countries that matter, inserted a test into thousands of tax treaties at once. The test asks, in effect, whether obtaining the treaty benefit was one of the principal reasons the structure was arranged the way it was. If it was, the benefit can be denied. They called it the principal purpose test, and it is exactly as broad and as menacing to the old model as it sounds. The conduit company that exists mainly to wear a treaty coat now fails on its face — because wearing the coat was the principal purpose, and the company will say so the moment anyone reads its file.
Beneficial ownership registers, spreading across jurisdictions, mean that the question "who actually owns and controls this thing" increasingly has an answer that a tax authority can look up rather than litigate for three years. The fiction survived on opacity. Strip the opacity and the fiction has nowhere to stand.
And then the heaviest layer, still settling into place. The OECD's Pillar Two regime sets a floor: large multinational groups, those above 750 million euros in revenue, must pay an effective rate of at least fifteen per cent in every jurisdiction where they operate. If a group books profit somewhere that taxes it below fifteen, another country in the chain is entitled to charge the difference as a top-up tax. The estimated take is around 150 billion dollars a year in new revenue globally. Sit with what that does to the old logic. The entire point of routing profit into a low-tax jurisdiction was that it got taxed there at the low rate. Now, if that rate is below fifteen, somebody else simply collects the shortfall. The arbitrage you spent a fortune engineering is clawed back at the other end. The juice is gone.
Here is the part that should interest anyone who actually structures things, because it is where the conventional wisdom is wrong. The lazy reading of all this is "the game is over, structuring is dead, just pay your taxes." That is not what happened, and a structurer who believes it will give bad advice.
The regulators did not abolish the right to organize your affairs across borders. They abolished the right to do it on paper alone.
Every one of these reforms is built around the same pivot: from form to substance. The treaty test asks what you actually do. The substance rules ask who actually works there. The minimum tax asks what rate you actually pay — and even its main carve-out, the substance-based income exclusion, explicitly shelters a return on real payroll and real tangible assets in a jurisdiction. The system has stopped rewarding the fiction of presence and started rewarding the fact of it.
Which means structuring did not die. It got real. The advantage now belongs to the company that genuinely puts people, decisions, and operations in a place, because that company can claim everything the letterbox used to claim and actually defend it. If you really run a regional operation out of Singapore or the UAE or Ireland — with staff who genuinely decide things and assets that genuinely sit there — the treaty benefits hold, the substance rules are satisfied by definition, and the minimum tax carve-out shelters a slice of your profit precisely because the payroll and the buildings are real. The same structure that was a fiction for the letterbox is bulletproof for the company that means it.
So the work changed under the same name. The old structuring was a clerical art — the clever filing of papers in the right sequence in the right places. The new structuring is an operational one. It asks where a business can genuinely live, what it can genuinely run from there, and whether the commercial logic of putting real people and real functions in a jurisdiction stands up on its own, before any tax benefit is counted. If it does, the tax treatment follows and survives scrutiny. If it doesn't, no amount of paper will save it, and the structurer who tries is selling his client a liability dressed as an asset.
In June 2025 the G7 agreed to carve US-headquartered multinationals out of key parts of the global minimum tax, on the argument that America's own rules already do the job; in January 2026 the OECD's inclusive framework agreed a side-by-side system to implement it. So the floor that was meant to be global already has a large hole in it, shaped exactly like the country with the most multinationals. The pattern is old and reliable: the rules tighten, the largest players negotiate their exemption, and the structuring world adapts to a map that is never quite as uniform as the press release claimed. Substance won, broadly. But the people who said the game was over have, as usual, mistaken a change in the rules for the end of the game.
Figures and developments are drawn from OECD Pillar Two model rules and 2025–2026 administrative guidance, the Peterson Institute for International Economics, and Deloitte and EY technical summaries.