By Chenhaojie (Audrey) Sun [1]
Introduction
Mikhail Fridman’s investment treaty claim against Luxembourg is straightforward to describe, but much harder to classify. After his inclusion on the EU sanctions list, assets connected to him in Luxembourg were frozen. He subsequently commenced arbitration against Luxembourg before an arbitral tribunal constituted and administered by the Permanent Court of Arbitration (PCA Case No. 2025-42) under the 1989 Belgium/Luxembourg–USSR bilateral investment treaty. UNCTAD’s Investment Dispute Settlement Navigator records the claim at USD 16 billion and describes it as arising from Luxembourg’s freezing of Fridman’s assets following EU sanctions. The dispute raises a difficult question for investment law: when does an asset freeze stop being a temporary regulatory restriction and become a compensable taking?
Part of the difficulty comes from the way EU sanctions operate. The Council of the European Union adopted the listing decision; Luxembourg implemented the resulting restrictions within its territory; banks and other market actors then gave those restrictions practical effect. The loss complained of therefore cannot readily be traced to one autonomous act by the respondent State.
Three issues follow. First, does Article 10 of the BIT allow the tribunal to decide whether an expropriation occurred at all, or only the amount and method of compensation once expropriation has already been established? Second, if that threshold can be crossed, when can a sanctions freeze move from temporary immobilisation to indirect expropriation? Finally, where an EU measure is implemented through national authorities and private actors, how should responsibility and loss be allocated?
This post argues for a relatively simple organising principle: responsibility should follow the discretion Luxembourg actually exercised, while compensation should follow the loss caused by that exercise of discretion.
The Jurisdictional Threshold
The tribunal may never reach the merits unless it first overcomes a significant problem of consent.
Article 5 of the 1989 BIT protects investments from expropriation, nationalisation and measures having similar effects, subject to conditions including public interest, legal procedure, non-discrimination and compensation. Article 10(1), however, is considerably narrower than the investor-State arbitration clauses found in many modern treaties. It covers disputes relating to the “amount or method of payment of the compensation due under article 5”.
The difficulty is apparent. A tribunal authorised to determine compensation may need to decide whether compensation is due in the first place. Yet the wording can also be read as presupposing that an expropriation has already been established elsewhere. August Reinisch’s study of narrow investment-treaty dispute settlement clauses confirms that tribunals have been divided over precisely this question.
This is not a purely theoretical problem. The same BIT was considered in Berschader v Russian Federation. The majority adopted a narrow reading of Article 10 and concluded that the clause did not extend to determining whether an expropriatory act had occurred. On that approach, arbitration becomes available only after expropriation has already been established and a dispute remains over compensation.
For Fridman, Berschader is therefore a serious obstacle, though not necessarily the end of the matter. In Quasar de Valores (Renta 4) v Russian Federation, a tribunal interpreting a similarly narrow Soviet-era clause in a different BIT reached a different conclusion. It treated a dispute over whether compensation was “due” as necessarily capable of raising the anterior question whether an expropriation had occurred. On that reasoning, a clause referring to the amount or method of compensation did not confine the tribunal to quantum alone where the respondent denied that any compensation was payable in the first place. The treatment of such clauses has therefore not been uniform, a divergence also noted in Mara Valenti’s analysis of “narrow-looking” investment treaty dispute-resolution clauses.
That limitation also defines the proper scope of the present analysis. The same factual allegations could be expressed in the language of fair and equitable treatment, due process or arbitrariness. Freestanding claims of that kind, however, are difficult to reconcile with the restricted wording of Article 10. Expropriation is not simply one treaty claim among several here; it is the gateway through which the case must pass.
The EU Judgment and the Limits of Its Relevance
The immediate EU-law background is the General Court’s judgment in Fridman v Council (Case T-304/22). The Court annulled the relevant listing measures because the Council had not substantiated the grounds relied upon for Fridman’s designation to the required legal standard.
The judgment was nevertheless much narrower than a rejection of the EU sanctions regime. The Court did not question the legitimacy of restrictive measures adopted in response to Russia’s invasion of Ukraine. It required the Council to establish that the individual before it actually satisfied the relevant listing criteria.
It would therefore be a mistake to treat the annulment as establishing Luxembourg’s liability under the BIT. The General Court was reviewing the legality of the Council’s decision under EU law. The investment tribunal, assuming jurisdiction exists, faces a different task: whether conduct for which Luxembourg is responsible amounted to an expropriation under the BIT.
Still, the annulment is not irrelevant. Once the legal basis of the restrictions changed, so too did the significance of any authority Luxembourg retained over their implementation. The more useful question is not whether Luxembourg designed the sanctions regime—it did not—but whether any part of the alleged deprivation resulted from Luxembourg’s own exercise of public authority.
When Does an Asset Freeze Become a Taking?
An asset freeze does not look like a conventional expropriation. There is normally no transfer of title, no acquisition by the State and no formal confiscation. The property remains legally with the owner, but cannot be used, transferred or disposed of.
Investment law has never treated legal title as conclusive. In Tippetts, Abbett, McCarthy, Stratton v TAMS-AFFA, the Iran-US Claims Tribunal held that a taking may occur through interference with the use or enjoyment of property even where legal title remains unaffected, and stressed that deprivation must be more than merely ephemeral. That supports looking to the practical reality of ownership rather than its formal shell.
At the same time, the threshold is high. In Pope & Talbot v Canada, the tribunal rejected the expropriation claim because the regulatory interference did not substantially prevent the investor from using, enjoying or disposing of its property. Serious economic loss, without substantial deprivation, is not enough.
A sanctions freeze sits uneasily between those propositions. An investor may retain title while losing the ability to use, transfer, manage or realise the asset. As restrictions continue, the gap between formal ownership and meaningful ownership can become increasingly difficult to ignore.
Duration, Safeguards and Effective Reversibility
Duration matters, although it should not be treated as an independent test. Tippetts itself distinguished enduring deprivation from interference that is merely ephemeral. A freeze imposed for a limited period and accompanied by realistic avenues for review or release has a different character from one that continues for years with little practical prospect of modification.
The distinction should therefore be based on effective, rather than merely formal, reversibility. The fact that legislation allows a freeze to be lifted at some future point says relatively little if the investor has no meaningful route to reconsideration in practice. Conversely, a functioning system of review and derogations may reinforce the genuinely temporary character of the measure.
The European Commission’s 6 May 2026 FAQ on asset freezes under Regulation 269/2014 shows why implementation cannot always be described as wholly mechanical. The framework assigns functions to national competent authorities in relation to derogations, releases and the application of ownership and control rules.
Those procedures matter because they identify where national judgment may enter an otherwise uniform EU regime. Were applications for derogations or authorisations considered individually? Were decisions taken within a reasonable period? Were reasons given, and was effective review available? If Luxembourg had authority to permit the release or use of frozen assessts in circumstances recognised by Regulation 269/2014, the way in which that authority was exercised may be relevant. A formal possibility of obtaining a derogation or authorisation carries less weight if, in practice, it provided no realistic route to relief.
Regulatory Purpose, Proportionality and Deference
The economic effect of a measure is not the sole consideration in assessing indirect expropriation; its regulatory character and purpose are also relevant.
In Saluka v Czech Republic, the tribunal recognised that bona fide and non-discriminatory regulation adopted in the normal exercise of State regulatory powers for the general welfare will ordinarily not give rise to compensable expropriation. In Philip Morris v Uruguay, the tribunal likewise relied on Uruguay’s exercise of regulatory police powers as an additional basis for rejecting the expropriation claim.
Those cases are relevant to sanctions because they show that the public purpose and regulatory setting of a measure form part of the expropriation analysis alongside its practical effects. Security considerations therefore matter, but they cannot by themselves determine the outcome. A measure adopted for a legitimate public purpose may still require scrutiny of the severity and duration of the interference and of the way the measure operated in practice.
Proportionality can help organise that inquiry, provided it is used cautiously. Tecmed v Mexico remains a prominent, if much-debated, example of proportionality entering the indirect-expropriation analysis. The tribunal considered the relationship between the burden imposed on the investor and the public interest said to justify the measure.
That does not give an investment tribunal a mandate to decide whether the EU selected the best possible response to Russia’s invasion of Ukraine. The inquiry can remain narrower: did the freeze remain connected to its regulatory purpose; did its practical effect change over time; were safeguards genuinely accessible; and which of those matters were within Luxembourg’s control?
Responsibility Should Follow the Exercise of State Discretion
The central problem in Fridman is therefore not simply whether there was a taking. It is identifying the conduct for which Luxembourg can properly be held responsible.
The sanctions regime worked through several layers. The Council adopted the listing decision. Luxembourg implemented the restrictions within its territory. Banks and other operators applied them in individual cases, while counterparties and markets responded to the broader sanctions environment.
It would be artificial to collapse that entire chain into a single Luxembourg measure. But the opposite approach—treating the existence of an EU regulation as a complete answer—would also be unsatisfactory where national authorities retained genuine room for judgment. The Commission’s Regulation 269/2014 FAQ provides concrete examples of national functions that can matter in practice.
The better approach is to ask not simply whether Luxembourg had discretion, but what discretion it had and how it used it. Where Luxembourg had no lawful option other than to give effect to a directly applicable EU rule, the interference is difficult to describe as an independently designed national measure. Different considerations arise where national authorities had to assess the scope of restrictions, ownership or control, derogation applications, releases, or the continued operation of measures after the underlying legal position changed.
Care is also needed with terminology. The issue is not simply attribution in the technical sense. Conduct of Luxembourg authorities may be attributable to Luxembourg without establishing that Article 5 was breached. The harder questions concern the legal character and consequences of that conduct: did Luxembourg’s own exercise of authority contribute to a deprivation sufficiently serious to amount to expropriation, and did it cause the loss for which compensation is sought? The proposition that responsibility should follow the exercise of discretion is therefore an analytical argument advanced in this post, not a freestanding rule already established by arbitral precedent.
Compensation Should Follow Causation
The same problem becomes sharper when the analysis turns to damages.
The reported USD 16 billion claim did not arise in an economic vacuum. UNCTAD records the amount claimed at USD 16 billion, but Fridman’s economic position was affected by the EU listing, Luxembourg’s implementation of the sanctions, decisions of banks and counterparties, reputational consequences and measures taken in other jurisdictions.
Even if Luxembourg contributed to an expropriatory deprivation, it does not follow that every loss associated with Fridman’s sanctioned status should be recoverable from Luxembourg. The causal inquiry therefore matters. Which losses followed directly from the EU listing? Which arose from decisions made independently by Luxembourg authorities? Which resulted from private actors’ risk assessments or wider market reactions? And which were caused by sanctions adopted elsewhere?
Those distinctions are particularly important where the alleged taking is an asset freeze rather than permanent confiscation. A temporary loss of use is not economically equivalent to destruction of an investment. Valuation may therefore turn on the duration of the restriction, the rights that remained with the owner, licensing or derogation possibilities, mitigation and the realistic prospect of release.
There is now sanctions-specific scholarship directly addressing this remedial problem. Toni Marzal’s 2025 article on damages and non-pecuniary remedies for sanctions in breach of international investment law expressly discusses the Fridman claim and argues for close attention to compensable harm and the quantification of sanctions-related losses. That supports a cautious separation between losses caused by the respondent State and the broader economic consequences of a sanctions regime.
Nor should the size of the claim determine the merits. A large potential award is not a legal reason to reject an otherwise valid claim. It is, however, a reason to be especially careful about what loss was actually caused by the respondent State. If a freeze that remained reviewable and potentially reversible is valued as though the investment had been permanently confiscated from the outset, the damages exercise risks collapsing the distinction between regulation and taking that the merits analysis is meant to preserve.
The better position is therefore that Luxembourg should compensate losses caused by its own legally relevant conduct—not losses generated by decisions over which it had no meaningful authority. Equally, if Luxembourg authorities independently prolonged or aggravated the deprivation, consequences caused by those decisions may properly enter the quantum analysis.
The two propositions are closely connected: responsibility should follow the exercise of discretion, and compensation should follow causation.
Conclusion
Fridman v Luxembourg is tempting to view as a larger confrontation between sanctions and investor-State arbitration. Legally, however, the case is likely to turn on narrower questions.
The first is whether the tribunal has jurisdiction to decide the dispute at all. Article 10 of the 1989 BIT is unusually restrictive, and Berschader offers substantial support for the view that it does not extend to determining whether an expropriation occurred in the first place. Quasar/Renta 4 and the wider scholarship on narrow clauses show that the interpretative issue is not entirely settled, but it cannot be treated as incidental.
If the tribunal crosses that threshold, neither obvious shortcut is satisfactory. A sanctions freeze is not confiscation merely because it causes severe economic harm. But neither should the language of security automatically place substantial interference with property outside investment-law scrutiny.
What matters is what happened in practice. What meaningful rights remained with the investor? How long and how severely were they restricted? Were review, derogation and release procedures realistically available? Which consequences flowed directly from EU law, and which depended on choices made by Luxembourg?
The same questions shape compensation. The Council adopted the listing decision, but Luxembourg is the respondent before the tribunal. A tribunal therefore has to separate Luxembourg’s own exercise of public authority from the much wider economic consequences of an international sanctions regime.
That may ultimately be the broader importance of Fridman. Modern sanctions operate through overlapping supranational, national and private decisions. Investment law cannot sensibly allocate responsibility without paying attention to those distinctions.
For this case, a useful starting point is also a simple one: responsibility should follow the discretion actually exercised by the respondent State, and compensation should follow the loss caused by that exercise of discretion.
[1] Chenhaojie (Audrey) Sun is an LLM graduate of the University of Cambridge with experience in international arbitration, investment law and cross-border disputes, and a background spanning both civil and common law systems.