TL;DR
- The core decision: Treat permanent establishment as a tax exposure with an employment trigger, not an HR problem with a tax footnote. HR, finance and legal need a single shared list of people, places and roles, and most companies don't have one.
- When doing nothing is right: Your remote workers are sales or support, they sit in one country where you've no other footprint, they're not concluding contracts, and no fixed equipment or stock is held there. The risk is real but small enough to monitor.
- What has to be true for the lighter paths to work: No employee habitually concludes contracts locally, no fixed place of business exists or will exist, and the work can be re-routed to existing entities if challenged.
- How the options split: Monitor only, restructure the population, set up a local entity, or use an employer of record for payroll. Each trades control, cost and residual exposure differently.
- Decision rule: Apply the OECD two-part test to every remote worker on your list, then route by what fails first.
- Outcome to expect: You will identify two or three people who change the answer and a single function that has to start looking at the same list as tax.
The Engineer Who Quietly Moved Countries
A head of people opens a new starter form in late summer and sees a forwarding address in a city where the company has no office, no bank account and no contract. The engineer is senior, joining a product squad, and the recruiter flagged nothing because there was nothing in the process that asked. Payroll runs from the existing entity. Equity vests against the parent. The laptop ships from the same warehouse. Nothing on paper has changed, and yet a corporate tax exposure has just accreted in a country the company has never accounted in.
The CFO hears about it six months later, when an advisor asks a routine question during a transfer pricing review. By then the engineer has been working from that location for half a year, has joined a customer call where the company name was used, and the home office has become the place where the bulk of the work happens. Each of those facts is harmless alone. Together they're the kind of pattern that draws attention from a tax authority with the time to look.
But the real issue isn't whether the company did anything wrong, and it isn't the form that was never signed. The real issue is that the trigger for permanent establishment is tax, the facts that determine it are employment, and those two functions are looking at two different lists of people.
Best tools for Global HR
When You Genuinely Do Not Need to Act Yet
Four honest stages separate a comfortable position from a difficult one, and the first one is genuine. If your situation matches it, the right answer is to monitor, not to restructure.
Stage one, the setup that's fine. You have a small number of fully remote employees sitting in countries where you've no other footprint, no fixed office, no local contract authority and no local equipment. They're individual contributors, they don't sign on behalf of the company, and their work could be re-routed through an existing entity if a tax authority asked. In a setup like this, the exposure exists in theory and is small in practice, so the move is to put the names on a watch list and revisit the file when something changes: a new hire in the same country, a promotion into a client-facing role, or a request to fly in branded equipment.
Stage two, friction starts to appear. A salesperson wants to work from home for a quarter to be near an aging parent. A customer success lead negotiates renewals from a kitchen table. Nothing on a flowchart trips, but the pattern of behaviour starts to look like the company is doing business in that country through a person. This is the moment where the right move is a conversation between HR and tax, not a project. You have a real question, not a real exposure, and the answer is usually a small reorganisation of who does what from where.
Stage three, real risk has set in. You have an employee in a country for more than a year, the work is core to the business, and either contracts are being concluded from that location or a fixed office has appeared. At this point the cost of doing nothing isn't theoretical and a tax authority that looks closely will see the pattern. The action is structural, not procedural.
Stage four, the edge case. A senior employee holds a title that includes a country in it, the customer base in that country is growing, and the company has been paying local expenses through a personal card. By the time you reach this stage, you're in a conversation with an advisor, not a decision.
The Five Questions You Ask Yourself at 11pm
Does one person in one country actually create permanent establishment? It can, where the facts line up. The number of people is less important than what they do, where they do it, and for how long. Treat the question as a facts test, not a headcount test. If you answer it wrong, the exposure survives even if you later restructure the team.
Could the work just be moved back? That depends on the role, not the geography. A product engineer working on internal systems can be moved to a different legal entity with a payroll change. A sales lead owning a book of local customers can't. If the work is portable, the risk is lower. If the work is local by nature, the risk is structural and the move is structural too.
What counts as a "fixed place of business" in a home office? The OECD commentary draws a line based on whether the home is used as a place where the business of the enterprise is carried on, on a continuing basis, and whether the setup is at the disposal of the enterprise. A kitchen table used for video calls isn't a fixed place of business in most readings. A converted spare room with a dedicated connection, branded signage, and six months of continuous use is a different conversation. Get this wrong and the home office becomes an unplanned office.
Does the salesperson in the new market change the answer? A salesperson who habitually concludes contracts in a country where the company has no presence is one of the clearest triggers. The risk isn't the role, it's the authority to bind. If the salesperson can sign, the answer is yes. If they escalate, the answer is usually no, provided the escalation actually happens on paper. Get this wrong and a routine-looking hire has just opened a tax filing the company never made.
Is the employer of record the answer? It's one of three answers, and the wrong one in a surprising number of cases. An EOR employs the worker through its own local entity, runs local payroll, and handles statutory filings. That removes the need for you to incorporate in that country. It doesn't remove the risk that a separate activity, such as holding inventory, signing contracts, or operating a fixed office, creates a permanent establishment in its own right. It also doesn't answer corporate tax. If the role is a clean payroll relationship, an EOR is often the right move. If the role involves local commercial activity, an EOR is a partial answer that looks complete.
The Three Honest Categories the Approaches Split Into
Monitor only. You keep a list of remote workers, you reassess when the facts change, and you accept that a small theoretical exposure exists. This is right for individual contributors in single countries where the work is portable and the customer base isn't local. It fails the moment the role gains commercial authority, the tenure crosses the kind of time thresholds a tax authority will care about, or a second person joins in the same country and starts to look like a pattern.
Restructure the population. You move the work, not the people. A sales lead based abroad becomes a partner agreement with a local firm, or a contractor arrangement that shifts the work back to an existing entity. This is right where the headcount is small, the roles are flexible, and the customer relationships can survive a reorganisation. It fails where the talent is the asset, the relationships are personal, and the market won't accept a local partner as a substitute for an employee.
Set up a local entity or use an employer of record. You incorporate, or you engage an EOR to employ the worker on your behalf through its own local entity. This is right where the role is core, the tenure is long, and the work is genuinely local. It fails where the volume doesn't justify the cost, where the EOR relationship masks an activity that's itself a trigger, or where the company needs to do more than employ people in that country and has not realised it yet.
Five Diagnostic Questions You Can Self-Assess Against
Where is each remote worker physically working, and for how long have they been there? Pull the list from HR, not from the headcount system. The headcount system knows the legal employer, not the working address. If you can't answer the question for a given employee in a single sentence, the file is missing data and the file is the risk.
Does any of them have authority to conclude contracts on the company's behalf? Not "could they, in theory", but "do they, in practice, and is the authority written down somewhere". A signature on a purchase order, a click on a SaaS renewal, a verbal commitment to a customer that the company later honours. If the answer is yes for any one of them, the analysis just got harder.
Is any physical space being maintained locally, even informally? A co-working membership paid by the company, a stocked apartment used for client meetings, a small warehouse holding demo equipment. Each of these is a fact a tax authority will weigh. If you can answer "no, and we've not been asked" with confidence, the exposure is small.
What is the character of the work being done from that location? An engineer building internal tooling is different from a salesperson negotiating renewals or a consultant delivering on a local statement of work. The OECD framework turns on the character of the work, not just the location, and the same employee in the same home can produce different exposure depending on what they do there.
Does any of this interact with an existing entity, an EOR, or a local partner in the same country? The list of facts is what matters, not the labels. A local partner that signs contracts is a different fact from a local partner that only refers leads. An EOR that employs staff is a different fact from an EOR that also handles billing. If you can't describe each arrangement in one sentence, the arrangement is doing more than the label says.
What Actually Creates the Exposure
A home office used as a fixed place of business
A home office becomes a fixed place of business where the space is used on a continuing basis, the company has the space at its disposal in some meaningful sense, and the work done there's the kind of work that gives the business of the enterprise its character. A spare room used for one-off video calls doesn't meet the test. A dedicated workspace used daily for the better part of a year, on company-supplied equipment, on a connection the company pays for, with the company's branding visible to clients who visit, is a different picture. The OECD two-part framework adds a time threshold and a commercial reason filter on top of this, and both have to be read together with the older test rather than instead of it. Where this falls short as a way of thinking is that it treats the home as a single object, when in practice the home is a stack of facts: the connection, the equipment, the signage, the visiting clients, the storage of paper files, the use of the address on letterhead. A reader who applies the test only to the room and not to everything that happens there will get a confident answer that's wrong.
An employee who habitually concludes contracts
The second trigger is the dependent agent permanent establishment, and it sits in a different part of the framework. An employee who habitually concludes contracts, or plays the principal role leading to the conclusion of contracts that are then routinely accepted without major change, can create a permanent establishment in the country where they sit, even where there's no fixed place of business at all. The fact that earns this a place is that it's the trigger most companies miss, because the relevant behaviour looks ordinary. Renewals get signed. Purchase orders get approved. A verbal commitment gets made and then a contract follows. None of it looks like a tax event, and yet the pattern of it's what the framework is looking for. Where this falls short as a concept is the word "habitually", which sounds like a frequency test but is read as a substance test by most tax authorities. A single large contract in a year can be enough where the pattern is that the company relies on this person to make the local market work.
A fixed office or premises
The third trigger is the one most readers think of first, and it's the one most likely to be misunderstood. A fixed place of business is a physical location at the disposal of the enterprise, used on a continuing basis, where the business of the enterprise is carried on. The phrase "at the disposal of the enterprise" is the one that does the work. A co-working desk the company pays for is at the disposal of the enterprise. A home office is rarely at the disposal of the enterprise in the strict sense, because it remains the employee's home and the company doesn't control access. A serviced office with the company's name on the door is. The reason this earns a place is that it's the trigger most often over-reported, in the sense that a tax authority will look at any space the company touches in the country and ask whether it meets the test. Where it falls short is that the home office framework has changed how this is read, and a reader who applies the older test alone, without the OECD update, will reach conclusions that are too aggressive in one direction and too relaxed in another.
Holding inventory or equipment locally
Inventory held in a country for delivery to local customers can be a permanent establishment in its own right, where the storage constitutes a delivery operation rather than a transit one. A warehouse of finished goods waiting to be shipped is different from a container in transit. A bank of laptops held in a country to support a local team is different from a single spare unit in an employee's flat. The reason this earns a place in a list dominated by employee-driven triggers is that it's the trigger HR won't see, because the equipment list is owned by IT and the inventory list is owned by operations, and neither function is in the conversation. Where it falls short as a concept is the threshold at which storage becomes a delivery operation, which turns on volume, time and intent and is rarely a clean line. A reader who treats the test as binary will mis-classify arrangements that are genuinely in the middle.
The character of the work being done
The fifth trigger is the one that sits underneath the others. The work an employee does, and where the value of the enterprise is created, is what the framework is really asking about. A salesperson closing deals in a market is a different fact from an engineer writing code that runs globally, even if both are sitting in the same home office. A consultant delivering on a local statement of work is a different fact from a project manager coordinating between time zones. The reason this earns a place is that it's the lens through which the other triggers are read, and a reader who looks only at location and tenure will miss that the location matters because of the work. Where it falls short is that it's the hardest trigger to apply cleanly, because the same role in the same week can produce both kinds of work, and the framework has to be applied to a moving picture rather than a static one.
Using an employer of record
The sixth trigger is the one that doesn't look like a trigger at all, which is why it earns its place. An employer of record is a third party that employs the worker through its own local entity, runs local payroll, and handles statutory filings, while the worker does the work of your company. The arrangement is designed to reduce the need for you to incorporate. The exposure it doesn't address is the one that sits outside the employment relationship. A worker employed by an EOR who also holds inventory, signs contracts, or maintains a fixed place of business on your behalf can still create a permanent establishment in the country. The EOR doesn't answer corporate tax, and it doesn't answer the question of whether the activity, viewed as a whole, looks like the company is doing business in that country. Where it falls short is that it's sold as a complete answer, and a reader who treats it as one will have a clean payroll file and a dirty tax file.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| One engineer, single country, internal work, under a year | Small | Fully remote, portable work | Low theoretical exposure, no commercial authority | Monitor only, with a quarterly review of role and tenure |
| One salesperson, single country, owns renewals | Small | Remote, commercially authoritative | Habitual contract conclusion risk | Restructure so contracts are signed by an entity outside the country, or move the role to a partner model |
| Three to five employees, mixed roles, one country | Medium | No local entity, EOR for some, contractor for others | Inconsistent facts, no shared list of people and places | Build the shared HR and tax list, then apply the OECD two-part test to each person |
| Growing team in a strategic market, year-plus tenure | Medium to large | EOR for employment, local customers, growing revenue | Exposure that an EOR cannot answer, corporate tax questions | Move to a local entity, accept the cost, plan the corporate tax position with advice |
| Single fixed office, branding, local staff | Large | Deliberate local presence | Obvious exposure, late to address | You are past the decision and into the conversation: get local advice, file where required, regularise the position |
| One person, but senior title, in a country on the website | Edge case | Title on the org chart, address on a profile, no entity | Optics and substance pointing the same way | Treat the role as a permanent establishment candidate, route the work to an existing entity, and remove the local signals |
| Contractor used like an employee in a key market | Edge case | Long tenure, deep integration, no entity | Dependent agent risk through a contractor | Either convert through an EOR or local entity, or re-document the relationship so it is genuinely contractor in fact and on paper |
| Acquisition adds headcount in a new country overnight | Sudden | Inherited remote workers, no policy | Exposure you did not choose and cannot unwind quickly | Freeze the population, build the list, apply the test, decide per role rather than per country |
Applying the Two-Part Test to Your Own Population
The OECD framework, as updated on 19 November 2025, gives a reader a way to think about each remote worker as a separate question. Part one asks whether the home office is a fixed place of business at all. The threshold is time-based, and a home office is generally not one where the employee works there for less than half of their total working time over any rolling twelve-month period. Part two asks whether the arrangement has a commercial reason that gives rise to a permanent establishment, and the commentary is explicit that arrangements driven solely by employee preference, talent retention or internal cost efficiency such as reducing office space aren't commercial reasons for these purposes. Both parts generally need to be met for a home office to be treated as creating a permanent establishment, and the framework supplements rather than replaces the older test in the Model Tax Convention.
A practical way to apply this to a real list is to score each remote worker on the two parts and on the work itself, then route them by where they fail. The table below shows the shape of that exercise, run against a hypothetical population of eight remote workers. The numbers are illustrative of the framework, not the framework itself.
| Worker | Country tenure | Time at home office | Commercial reason for the arrangement | Role character | Two-part test outcome |
|---|---|---|---|---|---|
| Engineer A | 8 months | Around a third of working time | None beyond personal preference | Internal tooling, no client contact | Below time threshold, no commercial reason. Low exposure. |
| Engineer B | 14 months | More than half of working time | Retention of a key hire | Internal platform, occasional client demos | Above time threshold but no commercial reason. Worth a closer look, not a filing. |
| Sales lead C | 2 years | Effectively all working time | Local market coverage | Closes renewals, signs on a delegated basis | Above both parts, role creates dependent agent risk. High exposure. |
| Customer success D | 10 months | More than half of working time | Proximity to a strategic account | Coordinates delivery, no signing authority | Above time threshold, no commercial reason, role does not bind. Moderate exposure. |
| Field marketer E | 6 months | Around half of working time | Runs local events, holds branded stock | Holds inventory, runs on-the-ground campaigns | Time threshold marginal, inventory and character of work create a separate trigger. |
| Consultant F | 3 years | Most of working time | Local statement of work | Delivers on a local contract, signs for the company | Both parts met, role is core, exposure is structural. |
| Contractor G | 18 months | Most of working time | None, treated as a remote employee | Deep integration, signs some vendor agreements | Dependent agent risk through a contractor route. High exposure. |
| Engineer H | 4 months | Around a quarter of working time | Trial period, mutual fit | Internal product work, no client contact | Below time threshold, no commercial reason. Low exposure. |
The value of running this exercise isn't the verdicts, which depend on local law and treaty, but the conversations it forces between HR and tax. Engineer A and Engineer H look the same in the headcount system and the same to a tax advisor in isolation. Sales lead C and Consultant F look different to both. The shared list is what produces the right answer, and the right answer is rarely a single answer for the whole population.
What HR and Tax Have to Agree On
HR owns the list of people. Tax owns the list of facts that create exposure. Neither function currently owns the joined-up list, and that's the gap the framework exposes.
The first thing to agree on is the population. Not the headcount by entity, not the legal employer, but the list of every person doing work for the company from a physical location, the country, the start date in that country, and the role. The headcount system isn't the source for this, because it knows the legal employer and the payroll entity, not the working address. The source is the data the company doesn't currently collect, or collects in three different systems that don't talk to each other.
The second thing to agree on is the activity. For each person on the list, the question is what they do, who they can bind, and whether the work is portable. A shared taxonomy helps, and the table below shows the shape of one. The categories are a way to start the conversation, not a way to end it.
| Category | What it covers | Why tax cares | Why HR cares |
|---|---|---|---|
| Internal work | Engineering, product, design, internal ops | Low direct exposure, character of work matters less | Easy to relocate, low friction to change |
| Customer-facing, no signing authority | Customer success, account management, support | Moderate exposure through character of work | Harder to relocate, relationship is the asset |
| Customer-facing, with signing authority | Sales leads, country managers, contract negotiators | High exposure, dependent agent territory | Role-specific, often senior, expensive to change |
| Local operational | Field marketing, on-the-ground delivery, local partnerships | Inventory and fixed place of business territory | Often outside the standard HR template |
| Contract or EOR | Anyone not on the company payroll in that country | The label is not the answer, the underlying facts are | Owned by procurement or legal, not HR |
The third thing to agree on is the trigger list. When a person joins, moves, changes role, gains signing authority, takes on local operational work, or crosses a tenure threshold, the file has to move. That requires a process, and the process has to live somewhere. It can't live in HR, because HR doesn't own the tax question. It can't live in tax, because tax doesn't own the people data. It has to live in a shared workflow, and the workflow has to be cheap enough to actually run.
The fourth thing to agree on is the escalation path. When a name on the list fails the test, who sees it, who decides, and how fast. The answer isn't a project, it's a standing conversation between two functions that already meet for other reasons. The cost of building it once is small. The cost of not building it's the cost of finding out from an advisor six months too late.
The Cost of Getting This Wrong
The invoice is the smallest part of it. Where a company finds itself with a permanent establishment it didn't plan for, the tax authority in question can look back over the period the exposure existed, and the lookback is rarely a single year. The penalties, the interest, the cost of filing returns that were never due, the cost of professional advice to manage the response: those appear on an invoice, and the invoice is what gets budgeted for.
The second-order costs don't. A senior hire in a strategic market quietly becomes the person who has to explain the company's tax position to a new customer during a procurement process, and the deal slips a quarter. A board member asks, during a routine governance review, why the company has filings in a country where it has no office, and the answer takes two meetings to construct. A planned acquisition in a neighbouring country is delayed because the target's advisors find the filings and read them as a signal of weak controls. So the real cost of getting this wrong is rarely the tax itself. It's the cost of the company becoming a different kind of company in the eyes of the people who matter to it, at moments when it didn't choose to.
The question to ask, before deciding that the current setup is fine, isn't what the exposure costs. It's what the exposure would cost if it surfaced at the worst possible time, in front of the worst possible audience, in a country where the company had not planned to have a conversation at all.
When You Are Ready to Go Further
If the picture above matches your situation closely enough that you want a structured view of the options, HROpsLab's independent comparison work is built for it. We don't sell software, we don't supply consultants, and we don't provide legal advice. What we do is maintain a current, side-by-side view of the approaches companies take to this problem, the cost shape of each, and the failure modes that the marketing material doesn't describe. The comparison is updated as the rules change, including the OECD framework that took effect in late 2025, and the underlying sources are checked against the relevant primary text rather than paraphrased from other secondary sources.
The right time to read it's when the shared list described above exists, even in draft. The wrong time is after the tax authority has written. If you want a structured walk-through, the case studies linked below are the place to start. If you want to talk the picture through with someone who has seen a hundred versions of it, the expert line is there for that too.
Frequently Asked Questions
What does permanent establishment actually mean?
Permanent establishment is a tax concept, not a legal one, and it's the trigger that makes a company liable to corporate tax in a country where it has not incorporated. The phrase comes from the OECD Model Tax Convention, and the concept is implemented through treaties and domestic law. A company can have a permanent establishment in a country without having an office there, and a company with an office in a country can avoid having a permanent establishment there, depending on the activity. The framework isn't a checklist. It's a set of tests applied to the facts, and the facts are about what the company does, not what the company has.
Can one remote employee create a permanent establishment?
Yes, where the facts line up. The number of people is less important than what they do, where they do it, and for how long. A single senior employee who habitually concludes contracts from a country where the company has no other presence is a candidate. A single engineer building internal tooling is usually not. The analysis is per person, not per headcount, and the answer for a given person can change as their role changes.
Does using an employer of record prevent permanent establishment?
No, and this is the most common misunderstanding. An employer of record employs the worker through its own local entity and runs local payroll, which removes the need for you to incorporate for employment purposes. It doesn't address the activities that sit outside the employment relationship, such as a worker who also concludes contracts, holds inventory, or maintains a fixed place of business on your behalf. An EOR is a partial answer that's often sold as a complete one. Where the work is a clean payroll relationship, it's usually the right move. Where the work is commercial, the analysis has to go further.
How does the OECD time threshold work in practice?
The OECD commentary, as updated on 19 November 2025, draws a line based on time and on commercial reason. A home office is generally not a fixed place of business where the employee works there for less than half of their total working time over any rolling twelve-month period. Above that, the analysis turns on whether the arrangement has a commercial reason that gives rise to a permanent establishment, and the commentary is explicit that employee preference, talent retention and internal cost efficiency aren't commercial reasons for these purposes. Both parts generally need to be met, and the framework supplements the older test rather than replacing it. The exact outcome turns on the specific treaty and domestic law, so local advice is essential.
Is a salesperson treated differently under the framework?
Yes, because the character of the work is different. A salesperson who habitually concludes contracts, or plays the principal role in negotiations that lead to contracts the company routinely accepts, can create a dependent agent permanent establishment in the country where they sit. This is true even where there's no fixed place of business at all. A salesperson who only refers leads and escalates to a signing authority in another country is in a different position, provided the escalation is real and on paper. The risk is the authority to bind, not the title.
Who should own this issue inside the company?
Neither HR nor tax can own it alone, because the trigger is tax and the facts are employment, and the two functions look at different lists. The right ownership is a shared workflow between the two, with finance or legal as the escalation point, and a named owner in each function who attends the standing conversation. The cost of the structure is small. The cost of not having it's the cost of finding out from an advisor rather than from your own data.
What should we do if we think we already have a permanent establishment?
The first step is to stop guessing and get the facts on paper. Pull the list of every person who has worked from a country where you've no entity, the dates, the role, and what they did. Have a tax advisor in that country review the file. Don't assume that filing will be the worst outcome, because the worst outcome is usually the one created by the lookback, not the filing. Where the exposure is real, regularising the position early is almost always cheaper than regularising it after a query. Where the exposure is theoretical, the file becomes the basis for the monitor-only approach described above.
Does the OECD framework apply in every country?
No. The Model Tax Convention and its Commentary inform treaties, and treaties are implemented through domestic law, and domestic law varies. The two-part framework is a way of thinking that most treaty-based systems will respect, but the outcome in a given country turns on that country's own rules and the specific treaty in force. A reader who applies the framework as a checklist rather than a starting point will reach the wrong conclusion. The framework is the first step in the analysis, not the last.
We help you compare the options on your own terms, not a vendor's.
