Permanent establishment for a foreign company: what to do in the first ten days starts with three moves: freeze the facts before memories soften, secure the paper trail behind who signed and decided what, and name one person authorised to speak with Skatteverket. Delay past that window narrows the positions still open to argue.
Who this concerns
The question usually surfaces in one of three shapes. A foreign group has someone working from a Swedish home office who closes contracts without a formally registered branch. A construction or installation project has run long enough that a site presence looks less like a visit and more like a fixed place of business. Or Skatteverket, the Swedish Tax Agency, has already written to ask how activity in Sweden is organised, which is usually the point at which "we will deal with it eventually" stops being an option available to the group.
What changes when the parent company sits outside Sweden, or outside the EU or EEA altogether, is the number of moving parts that have to be reconciled at once: which entity actually instructed the person on the ground, whether a tax treaty between Sweden and the home jurisdiction narrows the domestic test, and who inside the group holds authority to respond on the company's behalf. None of that gets easier by waiting a week. It gets easier by writing it down now, while the people who were actually there still describe events the same way.
This is a fact pattern that groups in the tax disputes practice see repeatedly across sectors: the commercial arrangement was never designed with a Swedish tax footprint in mind, and the question of whether it created one only comes up once someone outside the commercial team starts asking.
What the law says
Under Swedish law as it currently stands, the question is not whether the foreign company has formally registered anything here. It is whether, in substance, the company has a fixed place through which its business is wholly or partly carried on, or whether someone habitually exercises authority to conclude contracts on its behalf in Sweden. Both branches of that test turn on facts rather than paperwork: where decisions were actually taken, who had authority to bind the company, and how long and how regularly the activity ran.
Neither branch is satisfied by a single visit or an isolated transaction. Both are satisfied more easily than most commercial teams expect once activity becomes regular, once the same person is doing the same thing month after month, or once a project overruns its original schedule and starts looking permanent rather than temporary.
A tax treaty between Sweden and the company's home jurisdiction, where one exists, typically narrows this test rather than widening it, adding conditions around the nature of the activity and how long it has to run before it counts. Purely preparatory or auxiliary activity, for instance, is generally treated differently from activity that forms part of the company's core business. Which treaty applies, and what it actually says about the specific activity in question, is a fact question for the particular pair of jurisdictions involved. It is not a default answer that travels across cases, and assuming the treaty language from one jurisdiction applies to another is one of the more common early mistakes.
How it works in practice
The first ten days are not about reaching a conclusion on exposure. They are about not destroying the ability to reach one later.
Day one: freeze the facts, not the response
Before anyone drafts a reply to Skatteverket or an internal memo assessing risk, stop and capture what already exists: calendars, expense claims, travel records, the org chart as it actually operated rather than as it appears on paper. Nothing gets deleted, archived, or "tidied up." A tidied inbox looks identical to a concealed one from the outside, and the difference matters later.
Identify what actually triggered the question
An internal review during a restructuring, a routine payroll or VAT audit that widened in scope, and a direct enquiry from Skatteverket each carry a different clock and a different level of urgency. Confusing the three leads either to overreacting to a routine question or underreacting to one that already has a deadline attached.
Map who had authority to bind the company here
The dependent agent test does not ask who had a title. It asks who could actually commit the company, in practice, without the contract being renegotiated or re-approved elsewhere. If a person on the ground quoted prices, negotiated terms, and closed deals with only nominal sign-off from head office, that pattern matters more than the job description on their business card.
Separate the fixed place question from the agent question
A home office, a warehouse, or a construction site raises the fixed place question. A salesperson or negotiator raises the dependent agent question. The two run independently, and a group can fail one while clearly passing the other. Treating them as a single combined question is a common way to either miss exposure or overstate it.
Secure the paper trail before it becomes someone's recollection
Emails, calendar entries, Slack or Teams messages, and expense reports from the relevant period are worth more now than any witness statement will be worth in eighteen months. Collect them, preserve them as they are, and resist the instinct to annotate them with an interpretation before the facts are fully assembled.
Decide who is authorised to speak to Skatteverket
One person, clearly designated, with everyone else in the group briefed to redirect enquiries rather than answer them informally. Informal answers given by well-meaning staff who do not know the full picture are one of the most common sources of positions that later have to be walked back.
Set one internal message and stop the speculation elsewhere
Internal emails that speculate about "how bad this could be" are discoverable and do not age well. A short, factual internal note describing what is being done to establish the facts is safer, and it is enough.
What to check in the first ten days
- Who signed the contracts that generated the activity in question, and did they need approval from outside Sweden to do so
- How long the relevant activity has actually run, measured from when it started rather than from when someone noticed it
- Whether a tax treaty exists between Sweden and the parent's home jurisdiction, and whether the activity fits within any preparatory or auxiliary exception it contains
- Whether the same activity is already reflected, or contradicted, in transfer pricing documentation, VAT registrations, or payroll filings
- Whether anyone in Sweden has held themselves out, in writing or in meetings, as able to bind the company
FAQ
#### Does a single home-office employee in Sweden create a permanent establishment?
Not automatically. It depends on what that person actually does: closing contracts and negotiating terms without meaningful oversight points toward exposure, while purely administrative or support work generally does not. The employment contract's wording matters less than the pattern of actual conduct over time.
#### What happens if Skatteverket has already opened an audit before day ten?
The ten-day window still applies to internal preparation, but the external clock is now set by the audit's own deadlines. The priority shifts to responding within those deadlines with a factually accurate account rather than a rushed or defensive one, since an inaccurate early submission is harder to correct than a late but accurate one.
#### Does a tax treaty override the domestic permanent establishment test?
Where a treaty applies, it generally narrows rather than replaces the domestic test, adding conditions the domestic rules do not contain on their own. Whether it applies at all, and what it actually says about the specific activity, depends on the particular treaty and cannot be assumed from how another jurisdiction's treaty reads.
The numbers
There is no fixed count of days, contracts, or site visits that converts a Swedish activity into a permanent establishment on its own. What the assessment actually turns on is the combination of duration, regularity, and the scope of authority exercised locally, not any single threshold viewed in isolation. A short but recurring pattern of contract-signing can weigh more heavily than a long but genuinely one-off project.
The time an assessment takes, once documents are gathered, is set by Skatteverket's own workload and by how complete the group's response is on first submission, not by a standard timetable that applies to every case. A submission with gaps invites follow-up questions, and each round of follow-up adds time that a complete first submission avoids.
What actually drives the scale of the work involved is the number of jurisdictions and entities in the group structure, how far back the activity in question runs, and how much of the underlying documentation already exists in usable form versus needing to be reconstructed from scratch.
Where it usually goes wrong
The ten-day urgency does not apply evenly to every situation. A group that already operates a registered Swedish branch, or whose activity clearly falls within a preparatory or auxiliary exception under an applicable treaty, is working from a different starting point than one with no registration and no treaty protection at all. Treating every case as equally urgent wastes effort on the wrong problem.
The most common trap is assuming that remote or home-office work is inherently safe because it happens outside a traditional office. Authority to bind the company travels with the person, not with the building they sit in, and a home office with genuine negotiating authority carries more exposure than a large office with none.
The opposite trap is equally common: treating every Skatteverket enquiry as evidence that exposure already exists. Many enquiries are routine and close without further action once the facts are supplied. Escalating internally before the facts are assembled, in writing that later becomes discoverable, tends to create more of a problem than the underlying activity did.
Finally, groups sometimes wait for a formal decision before taking any of the steps above, on the theory that nothing is real until Skatteverket says so. By the time a decision arrives, the window for preserving contemporaneous evidence has usually already closed, and the group is arguing from memory rather than from record.
What to do next
This material gets a group as far as understanding what the first ten days should look like. It does not, and cannot, tell you whether your specific facts cross the line, because that depends on documents, dates, and a treaty analysis specific to your structure. That is the point where a review of the actual facts against the applicable treaty and domestic test needs to happen, and it is the scoping conversation this practice runs before any wider engagement is proposed.
If Skatteverket has already issued a decision rather than an enquiry, the relevant next step changes: the question becomes whether and how to contest that decision, which is covered separately in the piece on appeal to the administrative court.