IP holding structures after the nexus rules

The structure goes like this. Put the software, the brand and the patents in a company in a low-tax jurisdiction. Have your operating companies pay it a royalty. The royalty is deductible where you earn and lightly taxed where it lands. Repeat annually.

It was a good structure. It has been comprehensively dismantled, in four separate moves between 2015 and 2020, and the writing that still recommends it predates all of them. Here is what each move did, and what is left that actually works.

The four things that broke it

1. The modified nexus approach

Preferential IP regimes — patent boxes, innovation boxes, IP boxes — were required to be rebuilt around a nexus fraction. You get the preferential rate only on the proportion of income matching the proportion of qualifying R&D expenditure you yourself incurred:

qualifying expenditure ÷ overall expenditure × overall income = income eligible for the preferential rate

Acquisition costs of IP, and R&D outsourced to related parties, do not count as qualifying expenditure. A modest uplift is allowed to soften the edges. The practical effect is exact: a company that bought or received the IP rather than developing it gets a nexus fraction near zero, and therefore no benefit at all. The holding company that does nothing but hold is precisely the entity the rule was written to exclude.

2. DEMPE

Transfer pricing was rewritten so that legal ownership of an intangible, by itself, entitles the owner to nothing beyond a funding return. The returns follow the functions: development, enhancement, maintenance, protection and exploitation. Whoever performs and controls those functions is entitled to the profit they generate.

So if the code is written by your team in Warsaw and the brand is managed by you personally in Lisbon, a company in a third jurisdiction that holds the registrations and signs the licences is entitled to a return on the capital it put in and essentially nothing else. Charging a large royalty to it is not aggressive planning; it is a transfer pricing adjustment waiting to be made.

3. Withholding tax and treaty access

Royalties crossing a border attract withholding tax unless a treaty reduces it. Getting the treaty rate now requires surviving a principal purpose test, and a company interposed to obtain the rate is close to the textbook example the test was written for. Treaty shopping doesn't work covers why in detail.

4. CFC rules

Royalty income is passive income, and passive income in a low-taxed controlled subsidiary is the paradigm case for controlled-foreign-company attribution. The profit gets taxed to you at home anyway, in the year it arises, whether or not you take it out. CFC rules will find your offshore company sets out the mechanics.

And moving the IP out has a price

This is the part that turns a bad idea into an expensive one. Transferring intangibles out of a jurisdiction is a taxable event in most developed tax systems, valued at market.

• United States. Outbound transfers of intangibles are treated under §367(d) as producing a deemed royalty stream to the transferor, over the useful life of the property, commensurate with the income the property actually generates. Sell your successful product's IP abroad cheaply and the provision reprices it upward as the product succeeds.
• Germany and much of Europe. Transfers to a related foreign party are priced at arm's length under the exit-taxation provisions, with a hypothetical arm's length range and, in some cases, a price-adjustment clause if the outcome diverges from the valuation.
• Everywhere. You need a defensible valuation of the intangible at the transfer date. That is a professional engagement with a professional's fee, and it is the floor cost of the whole exercise before any tax.

What still works

Legitimate preferential regimes survive, because the nexus approach was designed to keep them for companies actually doing the research. They reward R&D where it happens rather than relocation:

Note what all three have in common: you claim them where you already are. None involves moving anything. If you are a founder in the UK, the Netherlands, Poland or Cyprus doing genuine development, the regime you should be asking about is your own country's, not a Caribbean one.

The threshold test

Before considering any cross-border IP structure, five questions. A no to any of them means the answer is to keep the IP where the work happens.

1. 1Is the annual tax saving above six figures? Valuation, transfer pricing documentation, local substance and annual filings are the running cost. Below that the structure loses money.
2. 2Will the entity holding the IP actually perform DEMPE functions — real people making real decisions about the IP, in that place?
3. 3Can you fund it without a transfer? Building new IP in the entity from the start avoids the exit charge entirely. Moving existing IP does not.
4. 4Does the royalty survive withholding and the principal purpose test on the specific treaty you intend to use?
5. 5Do your home country's CFC rules leave anything on the table after attribution?

The thing to do instead

At founder scale, the IP question that costs real money is not which company owns it. It is whether the company owns it at all.

Absent a written assignment, work produced by a contractor generally belongs to the contractor, not to you — in most common-law systems the default vests copyright in the author, and 'work made for hire' is narrower than people assume. Founder-scale companies fail this constantly, and it surfaces at exactly the wrong moment: in diligence, where unassigned contractor IP is one of the most reliable price adjustments there is.

• Get a written IP assignment from every contractor and employee, including the ones from four years ago, and including designers and copywriters.
• Register the trademark where you actually sell, which is a cheap, boring, high-return action nobody writes offshore guides about.
• Keep the IP in the operating company until there is a specific, articulable reason for it to be elsewhere — a joint venture, a licensing business, a genuine group with real substance in more than one place.