This is our mid-year read of cross-border structuring, drawn from work completed in the first half of 2026 across holding, trading, fund and private-wealth mandates. It is deliberately specific about what has changed rather than restating principles.
The headline
The centre of gravity has moved from tax rate to acceptance. For most clients, the binding constraint on a structure is no longer the effective rate it produces but whether banks, counterparties, auditors and registries will accept it without friction. Structures that optimise rate and fail acceptance are being unwound at the client''s cost.
What is working
Substance-backed UAE structures. Free-zone and mainland companies with a real office, resident management and activity matching the licence continue to perform well on banking, treaty access and counterparty acceptance. The 0% qualifying-free-zone rate is available but conditional, and the groups that treat the conditions as a live annual test rather than a one-off answer are the ones that keep it.
Luxembourg holding and fund platforms for European capital. Expensive to run, but they clear institutional diligence and access the treaty and directive network. For groups with European institutional investors, nothing else does the same job.
Cayman for pooled investment. Unchanged in function. Substance and directorship expectations have hardened; the vehicle itself remains the standard.
Simple two-tier operating structures — an operating company where the people are, a holding company where the shareholders and treaty access sit, with real board activity in both. Unfashionable and effective.
Singapore and Hong Kong for Asian operating groups, where the operating substance genuinely sits there.
What has stopped working
Passive holding companies with no board activity. Failing on substance review, failing on banking, and increasingly challenged on treaty benefits through principal-purpose tests. If the only activity is receiving dividends, the structure needs either real governance or removal.
IP boxes without development substance. The DEMPE analysis and the modified nexus approach have made rate-only IP migration untenable. IP should sit where the people who develop and manage it sit.
Nominee-director arrangements as substance. Banks now treat a formation-agent director in the registered-office jurisdiction as a negative signal, not a neutral one.
Multi-layer stacks assembled to obscure ownership. Beneficial-ownership registers, exchange of information and bank UBO requirements have made opacity expensive and ineffective. Every additional layer now has to justify itself functionally.
"Move the invoice" transfer pricing. Routing revenue through a low-tax entity that performs no function is being adjusted on audit with penalties. Functions, assets and risks have to move with the margin.
Where friction has moved
Three years ago the hard step was incorporation. Today incorporation is trivial and the hard steps are:
- Banking. Now the primary determinant of feasibility. See our August briefing on corporate banking appetite.
- Substance evidence. Registries are inspecting, not just collecting returns.
- Reporting visibility. CRS, CbCR, DAC6-style disclosure and now CARF mean the structure will be seen. Design assuming it is read by the tax authority of every jurisdiction involved.
- Counterparty diligence. Customers and suppliers run their own UBO and sanctions checks; a structure that fails a customer''s onboarding costs revenue directly.
What we would build differently than in 2023
- Fewer entities. The cost of an entity — filings, substance, audit, bank maintenance, director time — has risen faster than its benefit. We routinely remove two or three entities from inherited structures without changing the outcome.
- Substance first, rate second. Choose where the people and decisions can genuinely be, then optimise within that constraint.
- Banking pre-cleared. Soft indications from banks before incorporation, not after.
- Documentation contemporaneous. Board minutes, transfer-pricing files and substance memoranda written when the decision is taken, not reconstructed under enquiry.
- A written exit plan for each entity: how it would be wound up, what that costs, and what it would trigger.
Cost of ownership, honestly stated
For planning purposes, a properly run cross-border entity costs meaningfully more than the incorporation fee suggests once registered agent, accounting, audit where required, substance filings, director fees, bank maintenance and professional time are included. Groups budgeting only for formation are always surprised in year two. We now quote annual cost of ownership on every structure proposal for exactly this reason.
Second-half watch list
- Continued rollout of CARF data collection ahead of first exchanges.
- Further national implementation and refinement of the global minimum tax, including the interaction of qualified domestic top-up taxes with substance-based carve-outs.
- Tightening of beneficial-ownership access regimes in several European jurisdictions.
- Continued consolidation of banking appetite into a smaller number of jurisdictions with stronger substance requirements.
How to use this
If your structure was designed before 2023, the useful exercise is not "is it legal" — it usually is — but "would we build it this way today, and can we evidence what it claims?" Where the answer is no, the cost of change is lowest while nothing is under enquiry.
