The first filter in the 2025 Commentary is arithmetic. If less than half of a person's working time for the company is spent at their home in Spain, that home is generally not the company's place of business. The difficulty is in the counting.
The Commentary accepts that people use their homes privately and for work, and draws a line: a home or other relevant place would generally not be considered a place of business of the enterprise if the individual worked from it for less than 50 % of their total working time for that enterprise, over any twelve-month period commencing or ending in the fiscal year concerned. It then says that exceptions are not anticipated in most situations.
Three elements deserve attention: the base is total working time for that enterprise; the measure is time worked from that place; and the period is any twelve months starting or ending in the tax year.
The denominator is all the time the person works for the company, wherever: at the company's offices abroad, at client sites, travelling, and at home.
The numerator is the time worked from the home or other relevant place in question. The Commentary frames the test place by place. Time spent at a client's premises in Spain, or on a business trip to Lisbon, is working time but not time worked from the home.
That has a practical consequence. A sales manager who lives in Madrid and spends three days a week on the road may work from home for well under half of her time, even though she is in Spain all year. The home passes the first test. Her activity with clients is then looked at under other rules, including the dependent agent provision.
The Commentary does not prescribe a unit. Days are the natural measure, with part-days counted consistently. Israel entered an observation reserving the right to measure the threshold in its own way, by comparing days of presence or days worked inside and outside the country and taking whichever gives the higher figure, which shows that the method is not self-evident.
The period is not the calendar year. It is any twelve months that begin or end in the fiscal year. For a company with a calendar tax year, the 2027 analysis can therefore look at windows running from early 2026 to late 2028.
This closes an obvious gap. An employee who works from Spain for the last five months of one year and the first five of the next has spent ten consecutive months there, although less than half of either calendar year. A window drawn across the year-end captures it.
Anna works for a German company. From January to June 2027 she is at the Munich office four days a week. On 1 July she moves to Seville and works entirely from home. For calendar 2027 she is at home in Spain for roughly half her time. For the window from 1 July 2027 to 30 June 2028, it is close to 100 %. The second window is the one that matters.
Paragraph 44.9 says that the actual conduct of the individual determines the calculation. The contract and the company's policies are useful to the extent that they correspond to what happens. A policy that says "two days a week from home" is of little help if badge records and travel bookings show the person has not been to the office in a year.
Useful evidence, built up as you go and not reconstructed later:
Reaching 50 % does not create a permanent establishment. It removes the presumption. Paragraph 44.10 says that the question is then decided by the facts and circumstances, and the next paragraphs identify the commercial reason as the prominent one. A fully remote engineer in Spain with no local role can be at 100 % and still not give the employer a place of business. See the commercial reason test.
It is also worth remembering the earlier question. Below a certain duration the place is not fixed at all, whatever the percentage: three months in a holiday rental does not qualify. The percentage test only matters once the place has the necessary permanence.
Some companies will be tempted to cap remote days at 49 %. That can work where it reflects reality, for example for a commuter who really is in the office most of the week. It does not work as a paper exercise, because the test is one of actual conduct.
It is also not a universal answer. The 50 % test protects the home from being a place of business. It does not protect against a person who concludes contracts in Spain, and it does not apply in the same way to a founder who is the main person in the business. For those, see the dependent agent route and founders and one-person companies. The overall picture is in the main guide.
Source: OECD (2026), Model Tax Convention on Income and on Capital 2025 (Full Version), Commentary on Article 5, paragraphs 44.1 to 44.21, added on 18 November 2025. The Commentary is described here in our own words; the paragraph numbers are given so that it can be checked against the original.
Book a free first call. We will tell you where your company stands under the 2025 rules and what, if anything, needs doing.