THE PROBLEM OF CONSENT IN INVESTOR-STATE ARBITRATION: CAN IT TRULY BE CONSIDERED ‘VOLUNTARY’?

УДК 341.176

Abstract

This article argues that consent in investor-State arbitration is legally valid but only thinly voluntary. The law of the system still treats consent as the basis of jurisdiction: Article 25 of the ICSID Convention requires written consent; treaty offers must be accepted by investors; and once perfected, consent becomes irrevocable. Yet the operative structure of treaty arbitration differs sharply from the model of voluntary agreement familiar from commercial arbitration. In investor-State arbitration, the State makes a generalized ex ante offer to an undefined class of future investors, while the investor acquires the practical power to accept that offer after a dispute has crystallized. Jurisprudence on MFN clauses, umbrella clauses, standing, corporate restructuring, collective claims and counterclaims demonstrates that the real controversy is not whether consent exists, but how much interpretive distance tribunals may travel before consent ceases to reflect sovereign intention. The article proposes a thicker test of voluntariness based on textual clarity, temporal integrity, reciprocity, and institutional review. On that basis, investor–State arbitration should stop describing itself as simply “voluntary” and instead defend itself, if at all, as a constrained form of public-law authorization.

Аннотация

В статье обосновывается, что согласие в инвестиционном арбитраже между государством и инвестором является формально действительным, однако лишь в ограниченной степени добровольным. Несмотря на то, что система по-прежнему рассматривает согласие как основу юрисдикции — статья 25 ICSID Convention требует письменного согласия, договорные предложения подлежат акцепту инвесторами, а после выражения становятся безотзывными — структура договорного арбитража существенно отличается от классической модели добровольного соглашения, характерной для коммерческого арбитража.

В инвестиционном арбитраже государство формулирует обобщённое предварительное предложение неопределённому кругу будущих инвесторов, тогда как инвестор получает фактическую возможность акцептовать его уже после возникновения спора. Практика арбитражных трибуналов по вопросам применения оговорок о наиболее благоприятствуемой нации (MFN), «зонтичных» оговорок, процессуальной правоспособности, корпоративной реструктуризации, коллективных исков и встречных требований показывает, что ключевая проблема заключается не в наличии согласия, а в пределах его толкования без утраты связи с суверенной волей государства.

В статье предлагается более строгий тест добровольности, основанный на четырёх критериях: текстуальной определённости, временной целостности, взаимности и институциональном контроле. Делается вывод о том, что инвестиционный арбитраж не должен более описываться как исключительно «добровольный», а может быть оправдан лишь как ограниченная форма публично-правового уполномочивания.

Keywords: Investor–State arbitration, Consent and jurisdiction, Voluntariness, ICSID Convention, Treaty interpretation, Sovereign intent, MFN clauses, Public law theory.

Ключевые слова: Инвестиционный арбитраж между государством и инвестором, Согласие и юрисдикция, Добровольность, Конвенция ICSID, Толкование договоров, Суверенная воля государства, Оговорки MFN, Публично-правовая природа.

INTRODUCTION

Consent is the jurisdictional grammar of investor–State arbitration. Article 25 of the ICSID Convention makes written consent indispensable; official UNCTAD teaching materials explain that consent may arise through an investment contract, a legislative offer, or an investment treaty; and the investor ordinarily perfects treaty consent through written acceptance, often by instituting proceedings. In that formal sense, no tribunal may proceed without consent. The strongest doctrinal defense of the system begins there and insists that the voluntariness objection confuses dissatisfaction with outcomes for a defect in jurisdiction. [1.1., Article 25]

That defense is incomplete. Classical arbitration presumes privity, direct agreement, and at least rough symmetry in the parties’ decision to arbitrate. Treaty arbitration breaks with each of those assumptions. As Jan Paulsson famously described it, the system is “arbitration without privity”: the State’s consent is generalized and pre-dispute, while the investor’s acceptance is individualized and often post-dispute. [2.3.] The law thereby preserves consent in a formal sense while changing its social and political function. The question is not whether consent exists. The real question is whether it can honestly be called voluntary once it has been detached from direct bargaining and routed through treaty interpretation.

TREATY CONSENT AND THE CONTRACTUAL ANALOGY

The first place where voluntariness weakens is the relationship between treaty consent and contract consent. In contract-based arbitration, the parties negotiate the clause that binds them. In treaty arbitration, the State’s offer is made to a category of future investors, and tribunals must decide whether the treaty forum displaces, supplements, or survives contractual arrangements. The annulment decision in Vivendi is the classic statement that treaty claims are not dissolved by contract forum clauses simply because the underlying facts arise out of a concession contract. Salini v. Jordan moved along the same fault line: the tribunal distinguished between pure contractual disputes, for which it lacked jurisdiction, and genuine treaty claims, while also refusing to use MFN or the umbrella clause to manufacture jurisdiction over ordinary contract performance claims. [2.10.]

Umbrella-clause jurisprudence deepens the instability. SGS v. Pakistan read the clause narrowly and resisted converting contract breaches into treaty breaches, whereas SGS v. Philippines took the opposite path, holding that the treaty’s umbrella clause and broad dispute-settlement language could bring contract claims within the tribunal’s jurisdiction, even though it then stayed the proceedings in deference to the contract forum. This split matters for consent theory because it reveals how easily tribunals can alter the practical reach of treaty consent without any new act of assent by the host State. Once treaty consent is allowed to absorb contractual disputes, voluntariness ceases to rest on explicit agreement and begins to depend on interpretive philosophy. [2.5.]

My argument is therefore not that treaty arbitration lacks consent, but that treaty consent is qualitatively different from contract consent. It is premised on generalized sovereign pre-commitment, not transactional bargaining. Vivendi was right to preserve the autonomy of treaty claims, but that autonomy comes at a conceptual price: the more the treaty forum is separated from contractual consent, the less plausible it becomes to describe investor–State arbitration as voluntary in the ordinary arbitral sense. [2.10.]

MFN CLAUSES AND THE CONSTRUCTION OF CONSENT

No issue exposes the elasticity of consent more clearly than MFN clauses. Maffezini v. Spain held that a third-party treaty’s more favorable dispute-settlement provisions could be imported through the basic treaty’s MFN clause, subject to certain public-policy limits. Siemens followed a similar logic and allowed the claimant to bypass the treaty’s 18-month local-courts requirement by borrowing a more favorable clause from another BIT. These decisions treated jurisdictional access as part of the treaty treatment that an investor could claim through MFN language. If one accepts them, the State’s consent is real but only indirectly so: jurisdiction emerges not from the base treaty alone, but from a chain of comparative treaty obligations. [2.2.]

The restrictive response was equally forceful. Plama held that arbitration agreements must be clear and unambiguous, and that an MFN clause does not incorporate dispute-settlement provisions from another treaty unless the basic treaty leaves no doubt that the contracting parties intended that result. Wintershall and Daimler likewise rejected the use of MFN to eliminate Argentina’s 18-month local-courts precondition. The doctrinal point is larger than the MFN debate itself: tribunals were not only disagreeing about one clause, but about whether sovereign consent is a stable textual commitment or a system-wide package of comparable benefits. [2.4.]

That disagreement makes it impossible to sustain a simple voluntariness narrative. If Maffezini-style reasoning is correct, then consent may be functionally enlarged through treaty networks. If Plama-style reasoning is correct, then jurisdiction turns on strict textual closure. Either way, voluntariness is not doing the real analytical work. The real work is being done by treaty interpretation, systemic integration, and the arbitrator’s theory of sovereign intention. In methodological terms, consent in investor–State arbitration is less a fact than a constructed conclusion. [3.1., p. 38]

STANDING, RESTRUCTURING, AND INVESTOR-SIDE CHOICE

Investor-side standing further attenuates voluntariness. Tokios Tokelės and Aguas del Tunari accepted formal treaty nationality even where control or ownership pointed back toward the host State or non-protected nationals. Those decisions privileged incorporation and treaty drafting over economic reality. By contrast, TSA Spectrum pierced the corporate veil and denied jurisdiction because the claimant’s supposed foreign nationality was not supported by real foreign control. The resulting jurisprudence shows that treaty consent does not merely depend on the State’s offer; it also depends on how tribunals characterize the investor that claims to have accepted it. [2.8.]

Corporate restructuring cases push the point even harder. Phoenix Action denied jurisdiction where the investment was acquired for the purpose of bringing an international claim and therefore constituted abusive treaty shopping. Pac Rim articulated the now-standard threshold that abuse becomes relevant when a party can foresee a specific future dispute as a very high probability, not merely as a possible controversy. Philip Morris v. Australia then applied that logic decisively, holding that the claimant’s restructuring was carried out when there was a reasonable prospect that the dispute would materialize and for the principal purpose of gaining treaty protection, so the claims were inadmissible as an abuse of rights. These cases are the clearest admission in the jurisprudence that formal consent can be legally insufficient when its exercise is opportunistic. [2.9.]

Abaclat confirms the same structural tension from another angle. The majority accepted that Argentina’s treaty consent extended to mass claims by thousands of bondholders, while the dissent argued that such a transformation effectively rewrote the procedural framework without party consent. Whether one agrees with the majority or the dissent, the case demonstrates that voluntariness in ISDS does not stop at the moment of filing. It also depends on what forms of proceedings tribunals believe the original consent can bear. [2.1.]

RECIPROCITY, COUNTERCLAIMS, AND LEGITIMACY

The claim that ISDS is voluntary becomes most fragile when it confronts asymmetry. The investor usually decides whether to accept the treaty offer; the State typically cannot initiate the proceeding on the same footing. That asymmetry is visible in the law of counterclaims. Roussalis rejected Romania’s counterclaim because the majority treated the treaty’s consent as investor-centered and therefore insufficiently reciprocal. Urbaser moved the other way and accepted jurisdiction over Argentina’s counterclaim because the treaty language was broad and the counterclaim was closely connected to the same investment, though the State ultimately failed on the merits. Burlington went further still in practice: Ecuador succeeded in obtaining compensation on environmental counterclaims after the procedural basis for the tribunal’s jurisdiction had been secured. [2.7.]

These cases matter because they reveal what voluntariness conceals. A regime in which the investor alone can activate the host State’s pre-dispute consent looks less like classical arbitration and more like delegated public review. Counterclaims can partially repair that imbalance by converting one-way consent into a more reciprocal procedural relationship, but the current law remains fragmented. ICSID’s own materials still emphasize that counterclaims must fall within the scope of the parties’ consent. That requirement is doctrinally orthodox and politically telling: even at the point where reciprocity might restore legitimacy, the law returns to the same contested question of how broad or narrow consent really is. [1.2.]

A THICKER TEST OF VOLUNTARINESS

If consent is to retain normative force, it should be evaluated with a thicker test than the current binary of “there is consent” or “there is not”. I would propose four criteria. The first is textual clarity: tribunals should presume that MFN clauses, umbrella clauses, and collective procedures do not alter jurisdiction unless the treaty text clearly supports that result. The second is temporal integrity: restructurings undertaken when a specific dispute is highly foreseeable should be treated as abusive. The third is reciprocity: where a treaty grants investors direct standing, it should also address counterclaims and connected State defenses expressly, rather than leaving reciprocity to implication. The fourth is institutional control: because consent questions shape legitimacy system-wide, review mechanisms should be slightly stronger for jurisdictional construction than they are today, even if full appeal remains politically unavailable. [2.4.]

This thicker approach aligns with the reform concerns identified in UNCITRAL Working Group III, which has grouped the core complaints around inconsistency, correctness, independence, and procedural cost, and which is considering structural devices such as advisory support, procedural reform, and standing mechanisms. The broader legitimacy literature reaches a similar conclusion from different starting points: defenders argue that arbitration can bolster the rule of law, while critics argue that foreign-investor privilege and thin consent undermine democratic authority. Both sides are really debating the same thing – the adequacy of consent as a foundation for public power. [1.3.]

CONCLUSION

Investor–State arbitration can still claim consent as its legal foundation. It cannot persuasively claim voluntariness in the thicker sense associated with direct bilateral agreement. The State’s will is expressed generically and in advance; the investor’s will is expressed specifically and after the dispute has matured; and tribunals determine how far the resulting jurisdiction extends. Consent is therefore not fictitious, but transformed. The more candid description is that ISDS rests on sequenced public-law authorization rather than classical private ordering. Once the debate is framed that way, cases such as Maffezini, Plama, Tokios, Phoenix, Philip Morris, Urbaser and Roussalis stop looking like isolated doctrinal skirmishes and instead become part of one larger contest over sovereignty, standing, and legitimacy. [1.1., Article 25]

References:

1. Convention on the Settlement of Investment Disputes between States and Nationals of Other States (opened for signature 18 March 1965, entered into force 14 October 1966) 575 UNTS 159 (ICSID Convention).
2. ICSID, ICSID Convention Arbitration Rules 2022 (Ancillary Claims).
3. United Nations Commission on International Trade Law, Report of Working Group III UN Doc A/CN.9/970.
4. Abaclat and Others v. Argentine Republic (Decision on Jurisdiction and Admissibility, 4 August 2011) ICSID Case No ARB/07/5.
5. Emilio Agustín Maffezini v. Kingdom of Spain (Decision on Jurisdiction, 13 November 2000) ICSID Case No ARB/97/7.
6. Pac Rim Cayman LLC v. Republic of El Salvador (Decision on Jurisdiction, 1 June 2012) ICSID Case No ARB/09/12.
7. Plama Consortium Ltd v. Republic of Bulgaria (Decision on Jurisdiction, 27 August 2008) ICSID Case No ARB/03/24.
8. SGS Société Générale de Surveillance SA v. Islamic Republic of Pakistan (Decision on Objections to Jurisdiction, 6 August 2003) ICSID Case No ARB/01/13.
9. SGS Société Générale de Surveillance SA v. Republic of the Philippines (Decision on Jurisdiction, 29 January 2004) ICSID Case No ARB/02/6.
10. Spyridon Roussalis v. Romania (Award, 7 December 2011) ICSID Case No ARB/06/1.
11. TSA Spectrum de Argentina SA v. Argentine Republic (Decision on Jurisdiction, 8 April 2008) ICSID Case No ARB/05/5.
12. Tokios Tokelės v. Ukraine (Decision on Jurisdiction, 29 April 2004) ICSID Case No ARB/02/18.
13. Vivendi v. Argentina (Annulment Decision, 20 August 2007) ICSID Case No ARB/97/3.
14. Davy Karkason, ‘The Future of ISDS: The Reform Initiatives’ (Transnational Matters, 6 December 2023).