The Court of Justice’s July 2026 judgment in Hortis lands squarely on the fault line running through every remote and cross-border hire: does the law that protects a worker follow the place they actually work, or can it be engineered away through the paperwork built around a choice-of-law clause? For a long time the answer leaned toward the worker. This ruling seems to tilt it back. To my mind, it exposes just how ill-suited Article 6 of the Rome Convention is to the realities of remote working arrangements. Here is the point from the judgment that everyone will quote, and the one that deserves more scrutiny than it has had so far; a choice-of-law clause cannot by itself prove that a contract is more closely connected with the chosen country, but the consequences of that choice – where the salary lands, which social security system the worker is enrolled in, which tax authority collects – can. Commentators have mostly read this as a sensible line between form and substance. It ought to be read as a court quietly asking parties not to lie about their intentions, while handing employers a near-complete list of the facts they need to make the lie true.
A French employee, a Swiss employer, and a very tidy paper trail
JA worked from France for Hortis, a Swiss IT services company. His contract stipulated that Swiss law would apply. His salary was paid in Swiss francs into his Swiss bank account; he was affiliated to the Swiss social insurance system; his tax treatment, email domain and phone number, being also Swiss, all pointed in that same direction. When Hortis dismissed him under Swiss employment rules, which do not require a pre-dismissal interview or the provision of reasons required in the dismissal letter, JA argued that French law, as the law of his habitual workplace, should have applied instead. French courts sided with him on the ground that he had worked there throughout the employment relationship, but Hortis appealed and the French Cour de Cassation referred two questions to Luxembourg. First, can the parties’ chosen law also be the law identified by the Rome Convention’s Article 6(2) escape clause as the country most closely connected with the contract, such that the mandatory rules of the habitual workplace fall away entirely? Second, in answering that question, may a court rely on objective factors that were themselves generated by the contract’s performance under the chosen law – the Swiss salary account, the Swiss social security affiliation – or would that make the analysis circular?
The Court of Justice replied yes on both counts, ultimately doing little more than making explicit what had long been implicit in its 2013 ruling in Schlecker: the chosen law can be the law of closest connection, and in deciding whether it is, a court may look at exactly the kind of downstream facts JA’s contract generated – so long as it does not treat the contractual clause itself as one of those facts. The Court’s decision was probably justified as a matter of doctrine. Whether it was justified as a matter of policy, particularly for the growing population of genuinely remote cross-border workers, is a different question.
Facts that used to be hard to fake
Article 6 of the Rome Convention was built around one genuinely stubborn fact, namely, where a person’s body is most of the working week. This cannot be manufactured. Article 6, like Article 8 of Rome I after it, is a proximity rule, not a most-favourable-law-for-the-employee rule. An employer can dress up a contract however it likes, but it cannot make it seem like an employee physically works in Zurich while she lives and sits at her desk in Lyon. That immovability is precisely what made the habitual-workplace rule useful as a floor – it was the one variable in the relationship that the stronger party could not quietly adjust.
Salary currency, social security affiliation, tax domicile, and email domains are all features which lack that property. They are administrative settings. A modern employer running international payroll through an umbrella company or an employer-of-record platform can set every one of them to whatever jurisdiction it prefers in the time it takes to onboard someone, entirely independent of where that person’s laptop actually is opened each morning. Hortis tells national courts to treat the immovable fact (habitual workplace) as displaceable by a set of movable ones (payroll geography), on the theory that the movable facts are more ‘objective’ because they are not the contractual clause itself. But objectivity is not the same as immunity to manipulation. Tax residency and social security registrations are objective in the sense that a court can verify them. But since they can now, unlike in the past, differ from the physical place of work, they are not objective in the sense of being outside the employer’s control, which is the sense that actually matters if the point of the exercise is to stop an employer from imposing its own law by decree.
The consent question Hortis asks but cannot really answer
To its credit, the Court of Justice did seem alive to the obvious risk created by using connecting factors generated by a choice-of-law clause to validate that same clause, and so cannot be said to have missed the aforementioned issue entirely. Paragraph 46 of its decision instructs national courts to ask whether the connecting factors reflect a genuine, mutual choice or were simply imposed by the employer. Though a meaningful instruction from the Court, this safeguard will, in all likelihood, have little impact in the majority of cases where it is most needed, and for a reason that has nothing to do with bad faith on anyone’s part. Standard-form remote employment contracts are not negotiated fact by fact, and bargaining power between an international employer and individual employee is rarely equal to begin with. A new hire is handed a contract that already specifies the governing law, the payroll entity, the benefits provider and the equipment policy, and she signs all of it as a package, in the ordinary way that employment contracts get signed everywhere. Asking a court to determine whether that package was ‘genuinely agreed’ is asking it to find bad faith in an onboarding process that looks, formally, exactly like every other onboarding process – including the ones with no manipulation in them at all. The safeguard needs a visible seam between genuine and engineered consent, and remote hiring, by its nature, does not leave one.
Of course, this is somewhat a feature of the drafting of Article 6 of the Rome Convention itself, which was created on the 1980s assumption that the habitual place of work would ordinarily coincide with the economic and social environment of the employment relationship. Remote work severs that link as a matter of course, and this matters more for some remote workers than others. JA was a mid-career IT professional with enough leverage to have negotiated an individual contract; his circumstances are not obviously the circumstances of someone with no real say in the matter. But the doctrine Hortis produces does not distinguish between him and the far larger population of workers hired in bulk through Employer of Record (EOR) arrangements with take-it-or-leave-it terms. The same reasoning that vindicates a negotiated Swiss contract for an experienced consultant will, on its own logic, vindicate an identically structured contract imposed on a call-centre worker with no bargaining power at all – because the inquiry established by the Court in paragraph 46 of Hortis has no reliable way of telling the two apart from the documents alone.
Where we ought to look for a fix instead
If the objective-factors test is going to keep expanding its reach into remote work, courts applying it should stop treating all ‘objective’ connecting factors as equally probative and start ranking them by how expensive they are to fake. Duration and pattern of actual physical presence, the location of a worker’s direct reporting line and day-to-day supervision, and infrastructure genuinely rooted in one place, such as a leased local office, cost real money and real logistics to relocate. Salary currency and social security registrations, by contrast, are close to free to set wherever an employer’s payroll vendor happens to operate. A hierarchy of that kind would not require new legislation but is simply a more honest reading of what ‘objective’ should mean when the objects in question can be reconfigured from a laptop.
The more durable fix, though, probably is not judicial at all. Grušić already flagged the underlying problem back in 2022, well before this case existed, stressing that as remote work stretches the connection between an employment relationship and any single legal system, the private international law toolkit (built for a world of physically anchored workplaces) increasingly has to improvise. His proposed answer was not a smarter choice-of-law test but a substantive floor: overriding mandatory EU employment standards, applied regardless of which national law wins the closer-connection contest, so that the outcome for the worker does not turn on how skilfully an employer’s payroll has been arranged. Hortis is, if anything, a demonstration of why that recommendation has aged well. A conflict-of-laws test this permeable to administrative engineering was never going to hold a protective line on its own. It was always going to need something behind it.
A narrower point, and a broader one
For employers, Hortis offers real certainty: a choice-of-law clause that is actually reflected in how the relationship operates in features such as pay, benefits, tax, social security, stands a good chance of being upheld even against the mandatory rules of the country where the employee happens to sit. For mobile and remote employees however, the judgment is a reminder that the habitual workplace is not an unconditional trump card, and that the facts surrounding how a contract is actually administered may end up mattering more than where the desk is. The judgment’s real significance lies in what it confirms about the purpose of these provisions, and in the questions it leaves open about how far courts will let genuinely engineered connecting factors go, in an era where remote work makes manufacturing a ‘closer connection’ easier than the drafters of the Rome Convention ever anticipated.
Read narrowly therefore, Hortis is unremarkable. It is a fairly mechanical extension of Schlecker and last year’s ruling in Locatrans, confirmed by early commentary on the case (see the EAPIL case note, and the practitioner’s take). Read broadly though, it is a small case asking a question the drafters of the Rome Convention never had to face – what happens to a proximity-based conflict rule once nearly all the facts it relies on to establish proximity become facts an employer can set unilaterally. The Court of Justice answered the narrow question competently. The broader one still remains, and it is not going away. It will keep resurfacing, case by case, for as long as remote work keeps making the ‘objective’ side of employment relationships easier to design than the physical side is to change.