1.1. Overview

• 12. Both international law and domestic law are relevant for attribution. Even when it is undisputed that the attribution of a certain conduct is governed by international law, domestic law remains relevant to ascertain whether a certain person or entity is an organ of the StateFootnote ¹² or an instrumentality is endowed with governmental authority.Footnote ¹³ In this scenario, the international legal rules on attribution specifically rely on domestic law. A different matter is, however, to determine whether these rules apply in the first place.

• 13. The Commentary to the ILC Articles specifies the ambit of application of such rules in two main ways.Footnote ¹⁴ First, the rules govern attribution of conduct to the State “as a subject of international law” and not as a subject of domestic law.Footnote ¹⁵ Second, they only concern attribution for the purpose of determining a State’s responsibility for internationally wrongful acts. These two specifications are more complex than they may appear at first sight. At a basic level, the first one means that in a case against a State before domestic courts for violation of domestic law, the attribution of conduct to the State will not be governed by the rules of attribution in general international law but by domestic law . But the situation becomes more complex when the State is sued for breach of a contract. In principle, such a claim would be against the State as a subject of domestic law, a contractual partner. However, such claims have often been brought before international arbitration tribunals under a composite applicable law, including domestic and international legal aspects. In such cases, the second specification [Page 20] is useful to clarify that the international legal rules on attribution only concern a State’s responsibility for an internationally wrongful act, i.e. the consequences (organised by “secondary rules”, including rules of attribution) under international law of a breach of an international rule of conduct (a “primary rule”).

• 14. The decision of the tribunal in Ampal v. Egypt offers a concise confirmation of this point when it states that “[t]he Tribunal accepts the Respondent’s submission that the rules of attribution only apply to the determination of breaches of international law. They are not applicable to contractual breaches.”Footnote ¹⁶ The conceptual distinction between primary rules and secondary rules can be applied both to international and domestic law. The breach of a primary rule of domestic law has the consequences described in the applicable secondary rules of domestic law. Only if the conduct may amount to a breach of a primary rule of international law will the international legal rules on attribution apply, alongside other “secondary rules” of State responsibility for internationally wrongful acts.

• 15. These two specifications, taken together, circumscribe the scope of application of the rules of attribution in general international law in such a way as to exclude two main sets of questions which are governed by other rules: attribution-related questions under international law for purposes other than State responsibility, and attribution-related questions governed by domestic law.

1.2. Attribution-related questions governed by international law for purposes other than State responsibility

• 16. The first set of questions is expressly acknowledged by the Commentary to the ILC Articles by reference to the international legal rules governing the organs which can enter into commitments on behalf of the State without the need to produce full powers (Heads of State or Government and ministers of foreign affairs).Footnote ¹⁷ The distinction between the purpose of State responsibility and other purposes under international law (for which the international legal rules of attribution do not apply) was recognised by the tribunal in Gavrilović v. Croatia, in the following terms:

  The ILC Articles are the relevant rules on attribution that are widely considered to reflect international law. They concern the responsibility of States for their internationally wrongful acts, given the existence of a primary rule establishing an obligation. These principles of attribution do not operate to attach responsibility for “non-wrongful acts” for which the State is assumed to have knowledge.Footnote ¹⁸

  In this case, the tribunal reasoned that because there was no conduct constituting a breach of a primary rule of international law, the question of attribution did not arise. To avoid any confusion, attribution is a necessary but not sufficient [Page 21] component of an internationally wrongful act, as characterised by Article 2(a) of the ILC Articles. It is therefore relevant for the assessment of an allegation that a State has breached its international obligations, even if the assessment concludes that there has been no breach.

• 17. The distinction made by the tribunal in Gavrilović v. Croatia served mainly to conclude that the rules of attribution cannot be used “to create primary obligations for a State under a contract”.Footnote ¹⁹ An analogy can be made with the example given in the ILC Commentary relating to the rules defining the powers to bind the State. The fact that a person is an organ of the State is clearly not enough for such a person to be entitled, under international law, to conclude a treaty or to bind a State through a unilateral act, unless there are other rules that entitle that person to do so. The rules of attribution are inapplicable in this respect. The same applies to contractual undertakings. The mere fact that a person is (or is employed by) an organ does not mean that s/he is entitled to bind the State contractually – whether the national government, a territorial subdivision, or a public agency – unless there are other rules, here of domestic law, which contemplate such binding conduct. The international legal rules of attribution, including the rule concerning acts ultra vires, simply do not concern this issue.

• 18. There are also other attribution-related questions, governed by international law, where such rules are inoperative, although their treatment in the case law is sometimes confusing. For example, when the jurisdiction of a court or tribunal is limited to complaints or claims brought by entities (physical or legal) other than States, as under Article 34 of the European Convention on Human Rights (ECHR)Footnote ²⁰ or Article 25 of the ICSID Convention,Footnote ²¹ the question may arise as to whether a claimant’s links to a State are such that they preclude its right of action.Footnote ²² This is clearly different from the question of whether a certain conduct is attributable for the purpose of State responsibility. Hence, the international legal rules on attribution are inoperative. The matter is governed by other rules of international law specifically concerning this issue, i.e. Article 34 of the ECHR or Article 25 of the ICSID Convention, for the interpretation of which the international legal rules on attribution may or may not be relevant, but they are not governing. The European Court of Human Rights (ECtHR) has developed its own test to determine whether a complainant is a governmental or non-governmental actor for the specific purpose of bringing a complaint.Footnote ²³ In the investment context, tribunals have followed different approaches. In some cases, [Page 22] the matter has been addressed under a specific test applicable under the relevant rule (Article 25 of the ICSID Convention).Footnote ²⁴ In others, perhaps due to the framing of the argument by the parties, the question has been incorrectly examined under the international legal rules of attribution, although this is clearly a purpose unrelated to State responsibility.Footnote ²⁵

• 19. A further attribution-related question governed by international law but not by the international legal rules of attribution is the assessment of breach of a primary rule. The ILC Commentary makes this point clearly:

  As a normative operation, attribution must be clearly distinguished from the characterization of conduct as internationally wrongful. Its concern is to establish that there is an act of the State for the purposes of responsibility. To show that conduct is attributable to the State says nothing, as such, about the legality or otherwise of that conduct.Footnote ²⁶

  In practice, the matter may be confusing because the facts underpinning the inquiry on attribution may be largely the same as those relevant to establish breach of an international obligation. For example, when the claim concerns conduct which is blatantly in breach of international law, its attribution would suffice to trigger responsibility. But, as noted by the commentary, “the two elements are analytically distinct”.Footnote ²⁷ A conduct may be attributable to a State but, after assessment, it may not constitute a breach of a primary rule of international law. The latter assessment is governed by the primary rule.

• 20. So far, I have discussed four attribution-related questions governed by international law but not by the international legal rules of attribution, namely: (i) legal powers to bind the State in certain ways, (ii) the scope of primary norms (entities bound by it), (iii) the right to bring a claim, and (iv) the difference between attribution and responsibility. These are but some examples. As noted earlier, the international legal rules of attribution have a specific ambit of application, beyond which they are inoperative.

1.3. Attribution-related questions governed by domestic law

• 21. Certain attribution-related questions which are beyond the aforementioned ambit are governed by laws other than international law, i.e. domestic law including contractual matters. The decisions in Gavrilović v. Croatia and Ampal v. Egypt can serve again as starting points for the discussion. In Gavrilović, the respondent [Page 23] questioned the legality under domestic law of certain actions taken by its own Bankruptcy Courts. The tribunal, after stating that “the principles of attribution operate in the context of a complaint made against the State by a third party”, reached the conclusion that “[t]he involvement of the host State in this process – for example, through the Bankruptcy Court – is not a matter of attribution because there is no third party seeking to hold the State liable”.Footnote ²⁸ The conclusion is correct but the reason is questionable. Whether or not a third party seeks to hold a State responsible is not determinative. What matters is whether the conduct must be considered State conduct for the purpose of its (in)consistency with a primary rule of international law. Given that the issue in Gavrilović was (in)consistency with domestic law, the rules of attribution were inapplicable.

• 22. The latter reasoning was followed by the tribunal in Ampal v. Egypt with regard to consistency with certain contractual obligations. The tribunal sided with the respondent in its conclusion that “the rules of attribution only apply to the determination of breaches of international law. They are not applicable to contractual breaches.”Footnote ²⁹ Breach of contract is not a matter of State responsibility for internationally wrongful acts. The ILC Articles devote a specific article to the clarification of this point: “[t]he characterization of an act of a State as internationally wrongful is governed by international law. Such characterization is not affected by the characterization of the same act as lawful by internal law.”Footnote ³⁰ A breach of a contract triggers the secondary rules defining the consequences of such breach under the proper law of the contract, whether the domestic law of the host State or a foreign law selected by the parties. Only a breach of a primary rule of international law triggers the rules of State responsibility for internationally wrongful acts. It is, of course, possible that the very facts claimed to be a breach of contract may amount to a breach of international law. This raises the classic issue of the distinction between treaty claims and contract claims.Footnote ³¹ The tribunal in Ampal recalled specifically its decision on jurisdiction, where the two matters were distinguished:

  in order for it to find that there has been a breach of those standards in relation to the Gas Supply Dispute [fair and equitable treatment and unlawful expropriation], it will need to determine as an incidental question whether the Source GSPA [General Sale and Purchase Agreement] was validly terminated. However this does not change the fact that the key issue under the Treaty in respect of a claim for unlawful expropriation or breach of the fair and equitable treatment is whether there has been a loss of property right constituted by the contract or whether legitimate expectations arose under the contract.Footnote ³²

  Taken together, these two decisions clarify that the rules of attribution in general international law are only applicable to assess the consistency of State conduct with a primary rule of international law, not with one arising from [Page 24] domestic law (e.g. bankruptcy) or a contract (and its proper law). The next step of the analysis concerns the type of questions that fall under the remit of domestic (and contractual) law.

• 23. One important question that arises in practice concerns the extension of contractual obligations undertaken by entities, public or private, which are separate from the State. This may be relevant to determine liability under domestic law but also to assess whether there is an undertaking by the State (the contract) protected by an umbrella clause, i.e. by a primary rule of international law. The basic rule is clear: it is the applicable domestic law that determines who is a party to the contract; the international legal rules of attribution cannot extend a contract to a non-party. As noted in EDF v. Romania, reported in this volume:

  [T]he attribution to Respondent of AIBO’s and TAROM’s acts and conduct [both being State-owned entities in the Romanian aviation industry] does not render the State directly bound by the ASRO Contract or the SKY Contract for purposes of the umbrella clause. … Attribution does not change the extent and content of the obligations arising under the ASRO Contract and the SKY Contract, that remain contractual, nor does it make Romania party to such contracts.Footnote ³³

• 24. Interference by a State with a contractual relationship between third parties (e.g. a foreign investor and a public entity) may, however, be assessed in the light of the rules of attribution in general international law. For example, the public entity may exercise a contractual right under the direction or control of the host State.Footnote ³⁴ That specific conduct (the exercise of a contractual right) may be attributed to the State under the international rules of attribution and, if it is in breach of a primary rule of international law, it may trigger the responsibility of the State for internationally wrongful acts. But at no point are the contractual obligations between the foreign investor and the entity extended to the State. Any potential breach of contract is merely part of the facts to be assessed when considering whether there has been a breach of an international obligation.

• 25. In this context, one may ask what rules govern whether the contractual obligations undertaken by a separate entity (public or private) can be extended to the State. Whether or not a State has entered into a contract through representatives is a matter of domestic law. The decision on jurisdiction in Khan Resources [Page 25] v. Mongolia offers a good illustration.Footnote ³⁵ The dispute concerned a uranium mining venture in Mongolia. The claims were brought under both a set of contracts and the ECT. The claimants argued inter alia that the contractual obligations were binding on the State because they had been entered into by entities which were State representatives. The respondent objected that Mongolia was not a party to the relevant contract and, as a result, the tribunal lacked personal jurisdiction. The tribunal rejected the objection. In doing so, it made two significant points. First, the tribunal stated that the claimants bore “the burden of proving the facts on which they rel[ied] in support of this proposition [i.e. that the entities party to the relevant contract were representatives of Mongolia]”.Footnote ³⁶ Second, the tribunal stated that “the relationship between a state and its alleged representative must be assessed under the law of this state and in the light of the factual background of this relationship”.Footnote ³⁷ Thereafter, it examined the text of the contract, in the light of a related contract, Mongolian law and the behaviour of the parties, and it found that the State was a party to the contract and therefore the tribunal had personal jurisdiction. At no time did the tribunal feel any need to refer to the international legal rules on attribution.

• 26. The approach followed in Khan Resources v. Mongolia is, in my view, correct. It can be contrasted with the less clear approach followed on this point by the tribunal in Devas v. India, a decision reported in volume 18 of the ICSID Reports.Footnote ³⁸ In this case, the claimants argued that the conduct of a State-owned company, including the entry into a contract with the investor, was attributable to the State on the basis of the concept of agency. This argumentation was problematic because the claimant sought to establish a certain notion of agency in international law by reference to cases where agency had been analysed, as it must be, under domestic law.Footnote ³⁹ The tribunal rightly rejected the existence of such a concept but, debatably, it examined whether the obligations undertaken by the State-owned entity under the contract could be extended to the State as a result of the international legal rules of attribution. On this point, it concluded that:

  when entering into the Agreement, [the State-owned entity] was not acting as an organ of the Respondent, whether under the provisions of Articles 4 and 5 of the ILC Articles. The Agreement itself does not constitute an obligation the Respondent has entered into within the meaning of Article 11(4) [the umbrella clause].Footnote ⁴⁰

  [Page 26] This reasoning was possibly influenced by how the parties argued their case, but it is misleading. Whether a contract concluded by a separate State-owned enterprise can be extended to the State is a matter of domestic law. By contrast, whether a State has interfered with the exercise of contractual rights by the State-owned enterprise allegedly in violation of a primary rule of international law is a matter governed by the international rules on attribution. The tribunal reasoned, on this specific point, that the issuance of a force majeure notice by the State-owned enterprise had been conduct under the direction or control of the State (Article 8 of the ILC Articles), hence attributable to it for the purpose of State responsibility.Footnote ⁴¹

2.1. Overview

• 27. The international legal rules governing questions of attribution may be derived from general international law, as partly codified by the ILC Articles, but also from the treaties applicable in a specific case. As discussed next, three instruments which have been referred to in the practice of investment tribunals are the NAFTA, EU secondary legislation and the ECT.

• 28. In addition to the application of possible special rules of attribution, another sub-question arising in this context concerns the possibility that there may be specific international legal rules of attribution for investment disputes and their wider relevance for other contexts. The ILC Articles were developed to cover State responsibility for breach of any primary rule of international law, including – but not limited to – investment protection standards.Footnote ⁴² But investment disputes may present some peculiarities requiring adjustments to at least some of the general rules codified therein. As noted in the introduction, the tribunal in Bayindir v. Pakistan suggested such a possibility in connection with the rule codified in Article 8 of the ILC Articles.Footnote ⁴³ This possibility must be assessed in the broader context of the constant reference to and refinement of the general rules of attribution in the case law of investment arbitration tribunals. On the one hand, constant reference to the general rules is evidence that they are applicable as such to investment disputes. On the other hand, the growing body of investment cases analysing these rules has resulted in increasingly specific jurisprudential tests, which may (or may not) be relevant beyond investment disputes.

• 29. The following paragraphs discuss, first, the sub-question of special treaty-based rules of attribution and, second, the one relating to the possibility of special rules of general international law applicable to investment disputes.

2.2. Special treaty-based rules of attribution

• 30. It is useful to begin the discussion of special treaty-based rules of attribution with the decision in Mesa Power v. Canada, reported in this volume,Footnote ⁴⁴ because it addresses the two main issues that may arise in this context, namely the framing of the alleged special rule and the interactions between it and the general rules of attribution. The basic context of the attribution issue in this case was the respondent’s argument that the NAFTA contains a special rule of attribution (Article 1503(2)) for State enterprises which makes certain acts of these entities (which could be attributed under general international law) not attributable to the State. Hence the two questions: are the relevant entities “State enterprises” and, if so, what is the operation of Article 1503(2)? The tribunal answered that the entities were indeed State enterprises and that Article 1503(2) was a special rule of attribution under which only certain acts of the entities were attributable to Canada.

• 31. The first part of the answer – the characterisation of the entities – was based on the text of Articles 202 and 1505 of the NAFTA. According to these provisions, “state enterprise means an enterprise that is owned, or controlled through ownership interests, by a Party” (Article 202) and “[f]or the purpose of this Chapter [Chapter 15 on Competition Policies, Monopolies and State Enterprises] state enterprise means, except as set out in Annex 1505, an enterprise owned, or controlled through ownership interests, by a Party” (Article 1505). In order to make the more encompassing rules of general international law governing, the claimant argued that the three entities at stake were not State enterprises because they did not meet the test of Annex 1505. But the tribunal rejected this argument on the grounds that Annex 1505 was only relevant for Article 1503(3) of the NAFTA, not Article 1503(2). Thus, the test was whether the relevant entities were “owned or controlled” by Canada, which in the tribunal’s view was indeed the case.

• 32. That led to the question of the operation of Article 1503(2) of the NAFTA, which is our focus here. This provision states:

  Each Party shall ensure, through regulatory control, administrative supervision or the application of other measures, that any state enterprise that it maintains or establishes acts in a manner that is not inconsistent with the Party’s obligations under Chapters Eleven (Investment) and Fourteen (Financial Services) wherever such enterprise exercises any regulatory, administrative or other governmental authority that the Party has delegated to it, such as the power to expropriate, grant licenses, approve commercial transactions or impose quotas, fees or other charges.

  The tribunal, following both the arguments of the parties and a decision of an earlier NAFTA tribunal in UPS v. Canada,Footnote ⁴⁵ framed this provision as a special attribution rule. As such, it must be understood as a secondary rule which operates [Page 28] at the same stage as other secondary rules, including the rules of attribution in general international law. This interpretation is plausible,Footnote ⁴⁶ but it may misrepresent the nature of Article 1503(2) and of analogous provisions in other treaties (e.g. Article 22(1) of the ECT).Footnote ⁴⁷

• 33. Indeed, Article 1503(2) could also be understood as a primary rule of international law specifying which primary rules formulated in the NAFTA govern the conduct of State enterprises. This alternative framing is no less plausible,Footnote ⁴⁸ particularly in the light of Article 1503 as a whole, which specifies the obligations of each State party in connection with its State enterprises. Perhaps more compellingly, Article 1116(1)(a) of the NAFTA refers to claims brought by investors for breach of obligations under Article 1503(2), further confirming that this provision states a primary rule of obligation. The fact that the jurisdiction of a tribunal under Chapter 11 does not extend to alleged breaches of another primary rule (Article 1503(3)) does not make any difference. That a treaty may limit the jurisdiction of an arbitration tribunal only to claims for breach of certain standards but not others (which may, for example, fall under the jurisdiction of domestic courts or be simply removed from the remit of arbitration tribunals) does not mean that the rules of attribution do not apply to assess whether a certain conduct is attributable to a State. Framed as a primary rule, Article 1503(2) could not serve as a lex specialis with respect to secondary rules of attribution in general international law. The result would be that more conduct of State enterprises may have been attributable to the respondent.

• 34. Leaving aside questions of framing, the tribunal in Mesa Power v. Canada pursued its analysis as if Article 1503(2) was indeed a special (secondary) rule of attribution. As mentioned, the analysis of the interactions between this special rule and the general rules followed the reasoning of the tribunal in UPS v. Canada. The Mesa Power tribunal’s position is summarised in the following excerpt:

  The NAFTA thus establishes a special regime which distinguishes between a NAFTA Party and its enterprises, specifies what control obligations the former has over the latter, and thus organises the NAFTA Party’s responsibility for acts of its enterprises. This regime cannot be displaced by the ILC Articles, which, as mentioned above, are [Page 29] residual in nature. … As a consequence, the responsibility regime arising from Article 1503(2) prevails over the residual rules of Article 5 of the ILC Articles. The acts of the [three entities] will accordingly be attributable to Canada if these enterprises were exercising regulatory, administrative or other governmental authority as specified in Article 1503(2) when they carried out the acts in question.Footnote ⁴⁹

• 35. This conclusion required the determination of whether the three entities, in their impugned acts, were exercising governmental authority. At this stage, given that the term “governmental authority” is not defined in the NAFTA, the tribunal had little choice but to revert to the rules of attribution in general international law, specifically to the rule codified in Article 5 of the ILC Articles and its commentary, to clarify its meaning.Footnote ⁵⁰ This reasoning accurately reflects the operation of the lex specialis principle, which may displace the general rule in part or fully and, even in the latter case, it does not preclude resort to other rules applicable between the parties for interpretation purposes.Footnote ⁵¹

• 36. The conclusion reached by the tribunal in Mesa Power v. Canada, i.e. the application of a special treaty-based attribution rule which made certain acts non-attributable to the State, can be contrasted with the reasoning of the tribunal in InterTrade v. Czech Republic. In this case, the claimant sought to attribute an allegedly unfair tender process managed by an instrumentality established by the respondent through different routes, including reference to the instrumentality’s status under EU law. In support of this argument, the claimant referred to two opinions of the European Commission characterising the instrumentality as a “public contracting entity” under Directive 92/50/EEC and concluding, on that basis, that the mismanagement of the tender process by the entity amounted to a breach of EU law by the Czech Republic.Footnote ⁵² Unlike the situation in Mesa Power v. Canada, the application of EU law would have meant that the relevant obligations had a wider group of duty-bearers, in that “public contracting entities” were widely defined, encompassing the Czech instrumentality. The tribunal rejected the argument, however, and found that attribution for the purpose of State responsibility for breach of international law was governed not by EU law but by the rules of attribution in general international law, specifically Article 5 of the ILC Articles.Footnote ⁵³ The result was that the acts could not be attributed.

• 37. For present purposes, the decision on InterTrade v. Czech Republic is noteworthy in relation to the issue of framing. As discussed earlier in this preliminary study, the rules on attribution cannot extend the scope – specifically the duty-bearers – of a primary rule arising from a contract, domestic law or international law. If a rule applies only to a specific category of duty-bearers (e.g. judicial institutions), it cannot be extended to apply to administrative institutions merely because both types of institutions are “organs”. By contrast, if a rule defines its duty-bearers broadly (e.g. the State), then conduct attributable to the State will be [Page 30] relevant to determine whether the duty-bearer of the rule has breached it or not. The duty-bearers of a primary rule are thus defined by the rule itself.

• 38. In InterTrade v. Czech Republic, the real inquiry was whether a rule of EU law defining its duty-bearers broadly (encompassing conduct by the Czech instrumentality) had the effect of broadening the duty-bearers of certain investment protection standards under the applicable BIT to include not only the State but also separate instrumentalities. The clear answer is that these are different primary rules, each with its own duty-bearers. The action of instrumentalities is treated as action of the State (the duty-bearer) only if certain conditions are met, described in the general rules of attribution. Thus, the tribunal correctly analysed this issue under such rules and, on the facts, a majority concluded that the conduct was not attributable. The lex specialis issue does not arise in such a context because the two rules operate at different levels, one as a primary rule and the other as a secondary rule (of attribution). There was no conflict between two secondary rules of attribution. Even if such had been the case, as the tribunal seemed to suggest,Footnote ⁵⁴ the selection of the lex specialis would have been guided by the determination of the relevant primary rules at stake. For attribution of conduct to the State to assess its consistency with the standards of the applicable BIT, the rules of attribution in general international law apply.

• 39. Another aspect of the lex specialis inquiry is raised by the decision of the tribunal in Beijing Urban v. Yemen, which is reported in this volume. At stake was whether the claimant’s links to the Chinese government precluded the tribunal from asserting jurisdiction over the claims under Article 25(1) of the ICSID Convention. The respondent argued that the tribunal lacked personal jurisdiction because Article 25(1) excludes inter-State disputes from its remit, and the conduct of the claimant was attributable to the State under the rules of attribution of general international law. The tribunal accepted that the claimant was “a publicly funded and wholly state-owned entity established by the Chinese Government”,Footnote ⁵⁵ but it rejected the objection.

• 40. For present purposes, the most relevant aspects of this decision concern the framing of the issue and the test applied to address it. It is well established that, unlike arbitration clauses in BITs, Article 25(1) of the ICSID Convention is not a standalone basis of jurisdiction. It only defines the scope of jurisdiction of ICSID tribunals. Nor is Article 25(1) a primary rule, as it does not grant a right to bring a claim. But neither does it set out a special rule of attribution of conduct “for the purpose of State responsibility”. It is, like Article 34 of the European Convention on Human Rights, a rule setting out conditions for a claim to proceed before a specific dispute settlement mechanism.

• 41. The tribunal correctly addressed the issue under Article 25(1) as such and, more specifically, in the light of the interpretation given to it by Aron Broches, one of the main drafters of the ICSID Convention. The “Broches factors” or “Broches [Page 31] test” were thus an elaboration of Article 25(1), as the governing rule.Footnote ⁵⁶ Although the tribunal expressly mentioned the conceptual link between the Broches test and the attribution rules in Articles 5 and 8 of the ILC Articles,Footnote ⁵⁷ its reasoning makes clear that these other rules were not applied. The test asks whether the entity bringing a claim before an ICSID tribunal is either “acting as an agent for the government” or “discharging an essentially governmental function”.Footnote ⁵⁸ In either case, the claim would be precluded.

• 42. Despite the resemblance between this test and certain attribution rules, Article 25(1) is clearly not about attribution “for the purpose of State responsibility”. As a result, even if it is framed as a special rule of attribution (for the purpose of determining the jurisdiction of ICSID tribunals or, more specifically, of assessing the nature of the claimant), it is not a lex specialis with respect to rules attributing conduct for the purpose of State responsibility. It is a lex specialis for the purpose of any general rule which may govern the relation between an entity and a State for the purpose of bringing an action. This is not to say that the rules of attribution in general international law may not be used to interpret this dimension of Article 25(1) of the ICSID Convention, as far as they do not operate as the governing rules.

• 43. The reference to the Broches test in Beijing Urban v. Yemen also raises a broader question for the relations between special and general rules: the status and scope of operation of the many jurisprudential developments relating to attribution in the case law of investment arbitration tribunals.

2.3. Special rules of attribution applicable in investment disputes

• 44. The issues examined in the following paragraphs can be usefully introduced by reference to an observation made by the tribunal in Bayindir v. Pakistan, reported in this volume. In this case, the tribunal had to determine whether the exercise of a contractual right (the termination of a construction contract) by a Pakistani instrumentality (the National Highway Authority or NHA) could be attributed to the State. The tribunal reasoned that the conditions of Articles 4 and 5 of the ILC Articles were not met, but that the exercise of the right had been under the direction or control of the government, therefore attributable under Article 8. To reach this conclusion, it had to take a stance on the test for attribution under this rule. It made, in this regard, the following statement:

  the Tribunal is aware that the levels of control required for a finding of attribution under Article 8 in other factual contexts, such as foreign armed intervention or international criminal responsibility, may be different. It believes, however, that the [Page 32] approach developed in such areas of international law is not always adapted to the realities of international economic law and that they should not prevent a finding of attribution if the specific facts of an investment dispute so warrant.Footnote ⁵⁹

• 45. The references to the contexts of “foreign armed intervention” and “international criminal responsibility” concern the divergence of views between, on the one hand, the International Court of Justice (ICJ) in the Nicaragua case Footnote ⁶⁰ and later in the Bosnian Genocide case Footnote ⁶¹ and, on the other hand, the International Criminal Tribunal for the former Yugoslavia (ICTY) in the Tadić case.Footnote ⁶² This divergence of views concerns the level of control required for attribution (specific to the act or general to the group). In the Bosnian Genocide case, the ICJ noted that the general control test applied by the ICTY could be suitable for that specific context, but that the rule in general international law required specific “effective control”.Footnote ⁶³

• 46. The tribunal in Bayindir v. Pakistan seemed to suggest that the realities of international economic law call for yet another approach under the same rule. It did not clarify this approach but, from the factual discussion, the main modulation was the sufficiency of giving “clearance” for a specific decision in a highly concentrated government where certain actions would not be expected to proceed without at least passive approval from the leader.Footnote ⁶⁴ The Bayindir test combines therefore a specific context (the high concentration of political authority in one person’s hands which may be found in authoritarian regimes) with an action which falls short of giving a specific direction and is more like a permission. The difference between a specific direction and a mere permission lies at two levels: impulsion and specificity. Permission assumes that the impulsion does not come from the leader and that the “clearance” granted to whomever gives the impulsion can be of a more general nature.

• 47. I will discuss the operation of the rule formulated in Article 8 of the ILC Articles later in this study (Section 4.4). For present purposes, the Bayindir test is but an illustration of what could be seen as investment-specific tests or approaches to the operation of the attribution rules in general international law. Other tribunals have referred to specific tests in relation to attribution but rarely contrasting them to the general understanding in international law. One example concerns the decision in Kardassopoulos v. Georgia, reported in this volume.Footnote ⁶⁵ The tribunal discussed attribution issues mainly in its award on the merits and concluded that:

  [Page 33] whether one applies the principles of attribution set forth in the ILC Articles on State Responsibility or the tests developed in arbitral jurisprudence to ascertain whether the acts or omissions of a particular entity are attributable to a State, the answer in these arbitrations is the same.Footnote ⁶⁶

  The “tests developed in arbitral jurisprudence” mentioned in this excerpt are those referred to in Maffezini v. Spain, a decision of January 2000, i.e. the year before the adoption of the ILC Articles in their second reading.

• 48. The Maffezini tribunal relied on “structural” and “functional” tests to determine whether an entity is a “State entity”. The content of such tests shares some common ground with the general rules of attribution, but it also presents some marked differences, most notably a presumption arising from some specific circumstances:

  Here a finding that the entity is owned by the State, directly or indirectly, gives rise to a rebuttable presumption that it is a State entity. The same result will obtain if an entity is controlled by the State, directly or indirectly. A similar presumption arises if an entity’s purpose or objectives is the carrying out of functions which are governmental in nature or which are otherwise normally reserved to the State, or which by their nature are not usually carried out by private business or individuals.Footnote ⁶⁷

  The reasoning of the Maffezini tribunal on this point is debatable, and the overwhelming majority of the case law now follows the rules codified in the ILC Articles, as discussed later in this chapter (Section 4). For example, it seems clear that State ownership of a separate entity does not entail any presumption of attribution as such.Footnote ⁶⁸ For the acts of such an entity to be attributable, it will normally have to meet the requirements of the rule codified in Article 5 of the ILC Articles. But the specificity of these “tests” and the fact that they are presented in an award rendered in 2010, Kardassopoulos v. Georgia, begs the question of the recognition of attribution rules which are specific for investment disputes.

• 49. All in all, a systematic review of the relevant investment decisions leads to the conclusion that there is no clearly formulated thesis according to which there are special attribution rules in general international law which apply to investment disputes. The few decisions that either suggest or refer to such specificities, such as Bayindir v. Pakistan and Kardassopoulos v. Georgia, proceed on the basis of the rules codified in the ILC Articles. Some earlier decisions, such as Maffezini v. Spain,Footnote ⁶⁹ RFCC v. Morocco Footnote ⁷⁰ or Nykomb v. Latvia,Footnote ⁷¹ which do not clearly rely on specific routes from the ILC Articles, can be explained by the fact that they faced a less settled body of case law on this issue. In all events, their approach is not presented as a set of investment-specific rules, distinct from the general rules of attribution. For the rest, the wide convergence of the now mature body of [Page 34] investment cases dealing with attribution summarised, for example, in Ortiz v. Algeria Footnote ⁷² makes it clear that the applicable rules are those of general international law, as codified in the ILC Articles. Any variations from them, even those presented as “tests”, must be understood as elaborations of the requirements of these rules on the specific facts of the case. This is consistent with the views of the ICJ in the Bosnian Genocide case that “[t]he rules for attributing alleged internationally wrongful conduct to a State do not vary with the nature of the wrongful act in question in the absence of a clearly expressed lex specialis”.Footnote ⁷³ The only remaining question is whether they could be extrapolated to other types of disputes. Here, the answer is likely negative, at least in the light of the ICJ’s apparent reluctance to take on board the developments made by investment tribunals on matters of general international law.Footnote ⁷⁴

3.1. Overview

• 50. The third question concerns the stage of the proceedings in which matters of attribution must be addressed.Footnote ⁷⁵ Two main related sub-questions arise in this context. The first is whether there is a legal requirement to address matters of attribution at a specific phase of the proceedings or, on the contrary, whether such stage is determined by the nature of the objection and other relevant considerations. The second sub-question concerns the treatment of attribution arguments before a full review of the evidence can be conducted and the allocation of the burden of proof.

• 51. The link between the first and second sub-questions can be illustrated by Consutel v. Algeria.Footnote ⁷⁶ In this case, the tribunal proceeded on the assumption that matters of attribution must always be handled at the merits and, as a result, at the jurisdictional level, the facts supporting attribution of conduct must be assumed:

  The Tribunal considers that the questions of attribution discussed by the parties are questions for the merits and not of jurisdiction. Consequently, subject to what follows, the Tribunal must, when examining its jurisdiction, take for granted that the acts and omissions reproached to Algérie Telecom [a telecommunications operator wholly owned by the State] can be attributed to the Respondent.Footnote ⁷⁷

  Subject to the discussion in Section 3.2, this conclusion provides a clear illustration of the link between the sub-questions identified earlier.

• 52. In what follows, I examine first the reasoning of tribunals to allocate matters of attribution to a certain stage of the proceedings in order to identify what are the [Page 35] considerations guiding such determination. I then analyse the resort to prima facie tests and the allocation of the burden of proof.

3.2. The determination of the relevant phase

• 53. The determination of the appropriate phase at which matters of attribution must be discussed has sometimes been at issue in the investment case law. Whereas it seems reasonable for attribution – as part of the examination of State responsibility – to be addressed once the tribunal has asserted jurisdiction over the dispute, i.e. when examining the merits of a claim,Footnote ⁷⁸ the question is much more complex than it first appears.

• 54. In the excerpt of Consutel v. Algeria reproduced in the previous sub-section, the tribunal considered that attribution is always a matter for the merits. However, on closer examination, the authorities on which it relies are not so conclusive. The tribunal refers, to buttress its assertion, to several decisions, the earliest of which is Jan de Nul v. Egypt.Footnote ⁷⁹ In this case, the tribunal reasoned that it was not appropriate “at the jurisdictional stage to examine whether the case is in effect brought against the State and involves the latter’s responsibility. An exception is made in the event that if [sic] it is manifest that the entity involved has no link whatsoever with the State.”Footnote ⁸⁰ In addition to this first “exception” (manifest lack of attribution), the tribunal also mentioned another “exception” by reference to Salini v. Morocco.Footnote ⁸¹ In the latter case, a tribunal addressed attribution at the jurisdictional stage because the parties had extensively pleaded the issue as one relevant to jurisdiction. A similar approach was followed in RFCC v. Morocco.Footnote ⁸²

• 55. The interest of this second “exception” to a purported “rule” is that it depends on how the parties themselves frame the facts relevant for attribution. Framed as an objection to jurisdiction, the tribunal is expected to address the issue at that stage and some tribunals have done so, either presenting their examination as an effort to satisfy the parties’ procedural expectationsFootnote ⁸³ or simply addressing it as an objection to jurisdiction ratione personae.Footnote ⁸⁴ In one of the latter examples, Teinver v. Argentina, the tribunal reasoned that there was indeed authority to “support the conclusion that matters of state attribution should be adjudicated at the jurisdictional stage when they represent a fairly cut-and-dry issue that will determine whether there is jurisdiction”.Footnote ⁸⁵ In casu, it rejected the objection on the grounds that the respondent accepted that some conduct was attributable to the State and that, as far as the conduct identified in the objection was concerned, [Page 36] “the fact-intensive nature of Claimant’s allegations” required the tribunal to “postpone adjudication of this issue until the merits phase”.Footnote ⁸⁶

• 56. Rather than an “exception” to a “rule”, this line of cases suggests that there may be circumstances under which attribution is relevant for jurisdiction. Relevant considerations to make this determination would include: (i) manifest lack of link between the conduct and the State, (ii) the framing and pleadings of the parties, (iii) the overall effect of a finding of non-attribution (whether it excludes jurisdiction altogether or not) and, of course, (iv) factual considerations (which may be addressed at the jurisdictional level if they are sufficiently “cut-and-dry”).

• 57. The decision in Teinver v. Argentina, although clearly distinct in its understanding of the issue, is one of the authorities on which the Consutel tribunal relied. A similar mismatch between the rule asserted and the authorities relied upon can be discerned in another case cited in Consutel, namely Hamester v. Ghana. In this case, the tribunal began its reasoning by stating that “[t]he question whether the issue of attribution is, in a given case, one of jurisdiction or of merits is not, in the Tribunal’s view, susceptible of a clear-cut answer”.Footnote ⁸⁷ It then provided a nuanced view of the question:

  Not all issues, however, are so discrete or easily answered. Many – as is the case with attribution – entail more complex considerations, which could be characterised both as jurisdictional and relevant to the merits (and so to be considered only if the Tribunal has jurisdiction) …

  In order to clarify the distinction between a jurisdictional question and a merits question, it is useful to consider the different burden of proof required for each. If jurisdiction rests on the existence of certain facts, they have to be proven at the jurisdictional stage. However, if facts are alleged in order to establish a violation of the relevant BIT, they have to be accepted as such at the jurisdictional stage, until their existence is ascertained (or not) at the merits stage. The question of “attribution” does not, itself, dictate whether there has been a violation of international law. Rather, it is only a means to ascertain whether the State is involved. As such, the question of attribution looks more like a jurisdictional question. But in many instances, questions of attribution and questions of legality are closely intermingled, and it is then difficult to deal with the question of attribution without a full enquiry into the merits. …

  In any event, whatever the qualification of the question of attribution, the Tribunal notes that, as a practical matter, this question is usually best dealt with at the merits stage, in order to allow for an in-depth analysis of all the parameters of the complex relationship between certain acts and the State.Footnote ⁸⁸

  This is perhaps the clearest analysis of the issue in the current state of the investment case law. It confirms a prior analysis made in Maffezini v. Spain, where the tribunal distinguished questions of attribution relevant for jurisdiction and merits in similar terms:

  [Page 37] the Tribunal has to answer the following two questions: first, whether or not SODIGA is a State entity for the purpose of determining the jurisdiction of the Centre and the competence of the Tribunal, and second, whether the actions and omissions complained of by the Claimant are imputable to the State. While the first issue is one that can be decided at the jurisdictional stage of these proceedings, the second issue bears on the merits of the dispute and can be finally resolved only at that stage.Footnote ⁸⁹

  If the jurisdiction of the tribunal rests indeed on the existence of conduct attributable to a State – irrespective of whether such conduct amounts to a breach of a rule of international law – then the claimants have to establish those facts for the tribunal to have jurisdiction. Only if this is established can the tribunal examine whether certain conduct constitutes a breach or not.

• 58. Given the need for a factual inquiry, this is usually done at the merits stage, where the facts are discussed in detail. Such was the approach followed in Tulip Real Estate v. Turkey, where the tribunal examined whether the conduct of an entity (Emlak) was attributable to Turkey and concluded that, to the extent that the conduct was not attributable, it was “outside of the remit of the Tribunal”.Footnote ⁹⁰ Such fact-intensive objections to jurisdiction could be addressed at the jurisdictional level, if the tribunal allows for bifurcation, or later, if the objection is joined to the merits. In Maffezini v. Spain, aspects of attribution relevant for jurisdiction were determined at the jurisdictional stage. The tribunal was satisfied – for jurisdictional purposes – with a prima facie showing made by the claimant that the entity at stake (SODIGA) was acting on behalf of the State.Footnote ⁹¹ The evidentiary burden that the claimant was required to discharge, although prima facie, was quite significant. This raises the additional question of what exactly is meant by a prima facie test in this context.

3.3. The operation of prima facie tests and the allocation of the burden of proof

• 59. Reference to a prima facie test to handle facts at the jurisdictional level is common in the investment case law. For example, the tribunal in Mesa Power v. Canada noted that “attribution is generally best dealt with at the merits stage. This is subject to a prima facie pro tem test being performed in the context of jurisdiction to ascertain that the acts alleged are susceptible of constituting treaty breaches.”Footnote ⁹² However, little clarity is provided on the parameters of this prima facie test. To explore such parameters, it is useful to distinguish three main aspects: (i) the level of scrutiny of the facts, (ii) the nature of the objection, and (iii) the allocation of the burden of proof.

• 60. The first issue was examined with characteristic insight by Judge Rosalyn Higgins in her Separate Opinion in the first phase of the Oil Platforms case.Footnote ⁹³ The context of the analysis was a peculiar objection to jurisdiction raised by the US [Page 38] arguing that the treaty invoked by Iran was not applicable ratione materiae to the dispute. Judge Higgins identified the fundamental tension underpinning resort to prima facie tests as follows:

  a struggle between the idea that it is enough for the Court to find provisionally that the case for jurisdiction has been made, and the alternative view that the Court must have grounds sufficient to determine definitively at the jurisdictional phase that it has jurisdiction.Footnote ⁹⁴

  One particularly important precedent colouring her entire analysis was the Mavrommatis case, a diplomatic protection dispute brought under Article 26 of the UK Mandate over Palestine before the Permanent Court of International Justice (PCIJ).Footnote ⁹⁵ The UK challenged the jurisdiction of the PCIJ, leading the Court to examine at the jurisdictional stage whether the facts alleged by Greece as constitutive of the dispute fell under the terms of the mandate. The Court conducted a thorough analysis of each claim at the jurisdictional level, noting however that its analysis did not prejudge the merits. Judge Higgins contrasted the Mavrommatis approach with the less demanding one followed by the ICJ in the Ambatielos case, where the Court seemed to content itself with some degree of plausibility.Footnote ⁹⁶ A significant difference between the two cases was that, in Ambatielos, the merits of the claim were expressly reserved to a tribunal other than the Court. But the cases epitomise two contrasting positions regarding the treatment of facts at the jurisdictional level.

• 61. Judge Higgins pursued her analysis and formulated a principle according to which the approach to be followed would lie somewhere in between the thoroughness of Mavrommatis and the flexibility of Ambatielos. Applied to the facts of the Oil Platforms case, she suggested that:

  The only way in which, in the present case, it can be determined whether the claims of Iran are sufficiently plausibly based upon the 1955 Treaty is to accept pro tem the facts as alleged by Iran to be true and in that light to interpret Articles I, IV and X for jurisdictional purposes – that is to say, to see if on the basis of Iran’s claims of fact there could occur a violation of one or more of them.Footnote ⁹⁷

• 62. This test was brought into the investment case law by a series of decisions, mainly Impregilo v. Pakistan Footnote ⁹⁸ and then Jan de Nul v. Egypt.Footnote ⁹⁹ However, in this process it underwent a transformation. The level of scrutiny of facts, if they are to be taken as “alleged”, is clearly less demanding than the prima facie test in Maffezini, discussed in the previous section. This could be explained by the fact that in the Oil Platforms case and in Impregilo v. Pakistan,Footnote ¹⁰⁰ the test was used to determine whether the facts as alleged would fall under the provisions of the treaty [Page 39] invoked by the claimant or, in other words, whether the treaty prima facie applied to them.

• 63. The latter conclusion raises the second issue mentioned above, namely the nature of the objection. Whereas in the Oil Platforms case and Impregilo v. Pakistan, the key question was whether the alleged facts a priori fell under the treaty (objection ratione materiae),Footnote ¹⁰¹ in Maffezini v. Spain and in cases where attribution is raised as a jurisdictional obstacle, the objection is ratione personae.Footnote ¹⁰² Facts relevant for the assessment of a breach of a primary rule may be taken as alleged at the jurisdictional level to determine if the treaty prima facie applies to them. But facts relating to attribution are relevant for the assessment of both jurisdiction ratione materiae and jurisdiction ratione personae. Extrapolating the prima facie test from a broad assessment of whether the facts fall under a treaty to a specific assessment of whether certain acts are attributable to a State is a non-obvious step. It was made by the tribunal in Jan de Nul v. Egypt,Footnote ¹⁰³ and it has thereafter featured in decisions relying on Jan de Nul, such as Consutel v. Algeria.Footnote ¹⁰⁴ The problem of this extrapolation is that the type of demanding prima facie assessment conducted in Maffezini v. Spain fell between the cracks. The extrapolation had the result of transforming one demanding prima facie test into a mere fiction or assumption of facts for jurisdictional purposes. In other words, it undermines the possibility to raise an objection to jurisdiction ratione personae for lack of attribution at an early stage.

• 64. The third issue, the allocation of the burden of proof, is affected by this extrapolation. If attribution matters relevant for jurisdiction or breach are conflated, assumed to be met prima facie and allocated en bloc to the merits, the evidentiary burden of the claimant at the jurisdictional level is greatly facilitated. The consequences of this conflation may be alleviated when there is no bifurcation of jurisdiction and merits. In such case, the prima facie test loses its relevance,Footnote ¹⁰⁵ and the claimant has the burden to establish the facts on which it claims attribution for purposes of both jurisdiction and breach. But in case of bifurcation the problem re-emerges. Even if the attribution allegations are defeated at the level of merits, that creates significant unnecessary costs for both parties. If the tribunal “assumes” – rather than establishing prima facie in line with the more demanding Maffezini test – the facts underpinning its jurisdiction ratione personae, it may also be overstepping its powers.

• 65. The problems arising from the extrapolation of one prima facie test developed for jurisdictional objections ratione materiae to objections ratione personae can be avoided if matters of attribution are handled in the light of the [Page 40] general considerations made in Hamester v. Ghana.Footnote ¹⁰⁶ When certain facts must be established for the tribunal to have jurisdiction, then the burden of proof is on the claimant and the question must be addressed either at the jurisdictional phase or as a jurisdictional objection joined to the merits.

• 66. Addressing it at the jurisdictional phase may be appropriate under certain circumstances, discussed earlier by reference to Maffezini v. Spain and Teinver v. Argentina, including whether: (i) there is a manifest lack of link between the conduct and the State, (ii) the parties frame or plead the question as one of jurisdiction, (iii) the overall effect of a finding of non-attribution would exclude jurisdiction altogether, and (iv) the relevant factual considerations are sufficiently distinct to be addressed separately at this stage.

• 67. A tribunal may decide, instead, to join the jurisdictional objection ratione personae to the merits. In such case, once the merits phase is reached, attribution will have to be established in full. The question then arises of how the joined jurisdictional objection is to be addressed. In both Hamester v. Ghana Footnote ¹⁰⁷ and Tulip Real Estate v. Turkey,Footnote ¹⁰⁸ the tribunals rejected bifurcation of this issue, so the objections were addressed together with the merits. Yet, whereas the Hamester tribunal appeared to convert the objection into a matter for the merits, thereby asserting jurisdiction but rejecting attribution of specific conduct under certain claims, the Tulip Real Estate tribunal concluded that conduct non-attributable to Turkey was “outside the remit of the Tribunal”.Footnote ¹⁰⁹

4.1. Overview

• 68. The fourth question concerns the main routes of attribution admitted in general international law as they are codified in Part I, Chapter II of the ILC Articles. Several routes have featured in the practice of investment tribunals, most notably attribution of conduct of State organs (Article 4), State instrumentalities (Article 5), and persons or entities acting under the instructions, direction or control of the State (Article 8). There are variations in the operation of these routes, such as the question of acts ultra vires (Article 7), and there are also other routes, such as conduct acknowledged and adopted by the State (Article 11), which are discussed in the decisions reported in this volume. Still other routes, rarely discussed in the investment case law, include Article 6 (organs of a State placed at the disposal of another State) and Article 10 (insurrectional movements).

• 69. The discussion in this section is organised under five headings, one for each main attribution route (Articles 4, 5, 8 and 11, respectively) and another for all the rarely used ones (Articles 6, 9 and 10). The question of ultra vires action [Page 41] (Article 7) is a modulation of certain attribution rules. It will be discussed in their specific context and also in Section 5.4 as one recurrent problem.

4.2. Conduct of organs of the State (Article 4)

• 70. The conduct (actions or omissions) of State organs is attributable to the State.Footnote ¹¹⁰ This is a classic rule of customary international law, codifiedFootnote ¹¹¹ in Article 4 of the ILC Articles in the following terms:

  • 1. The conduct of any State organ shall be considered an act of that State under international law, whether the organ exercises legislative, executive, judicial or any other functions, whatever position it holds in the organization of the State, and whatever its character as an organ of the central Government or of a territorial unit of the State.

  • 2. An organ includes any person or entity which has that status in accordance with the internal law of the State.

• 71. Paragraph 1 removes any doubt regarding possible variations in attribution arising from the internal organisation of a State. Whether States are organised following a tripartite separation of powers or not, and whatever the form of power allocation across sub-national entities, the conduct of organs from any governmental structure is attributable to the State.

• 72. The decision in Tethyan Copper v. Pakistan, reported in this volume, provides a useful illustration of the basic rule. In this case, an Australian company active in the copper mining sector claimed that the authorities of Balochistan, a province of Pakistan, had refused to grant it a mining lease in violation of the Australia–Pakistan BIT. The tribunal made a distinction between, on the one hand, federal and provincial organs and, on the other hand, an agency established as a statutory corporation under provincial legislation (the Balochistan Development Authority or BDA). Conduct of the first group was clearly attributable to Pakistan under the rule codified in Article 4 of the ILC Articles, irrespective of the nature of the organs (executive, legislative or judicial) or their level in the internal organisation of the State (federal, provincial or municipal, or other devolution structures).Footnote ¹¹²

• 73. This conclusion was not affected by the fact that the province of Balochistan had a separate legal personality under domestic law. According to the tribunal, given that sub-national entities have no international legal personalities, if their acts could not be attributed to the State, “it would be impossible to make the [Page 42] conduct of provincial units subject to international obligations, which would in turn discourage investments in areas that are governed by provincial law and/or in which investors have to deal with provincial authorities”.Footnote ¹¹³ Thus, in the context of State organs, it makes no difference whether the territorial subdivision has a separate legal entity, as is very frequently the case.

• 74. In addition, in the context of State organs, the nature of the acts, sovereign or commercial, has no bearing. As stated by the commentary to Article 4 of the ILC Articles, which the tribunal quoted, it is “irrelevant for the purposes of attribution that the conduct of a State organ may be classified as ‘commercial’ or as acta iure gestionis”.Footnote ¹¹⁴ It is important to distinguish the sovereign vs commercial dichotomy from whether the organ acts in an “official capacity” or in a purely private capacity. Only action/inaction in an “official capacity”, whether sovereign or commercial in nature, is attributable. Whereas the “sovereign” or “commercial” classification focuses on the nature of the “act”, the “official” or “private” capacity classification is a matter of “context” or “appearance”.Footnote ¹¹⁵ If an act (whether sovereign or commercial) is performed by a person cloaked with governmental authority or holding herself out as a State official, then it is attributable to the State, even if the act is ultra vires (see below, Section 5.4). By contrast, in the absence of the “appearance” of an official capacity, the acts are not attributable. I will return to this difference later in this study (Section 5.3).

• 75. One deceptively trivial question concerns “how” to determine whether a body or a person is a State organ. The basic first step to determine this quality is the domestic law of the State, as expressly stated in Article 4(2). At this point, two issues arise. First, what should the domestic law say for the body or person to be deemed a State organ? Second, to what extent can such body or person be deemed an organ when the domestic law is silent or unclear or, most difficult of all, when it clearly states that the entity or person is not an organ? As we shall see, the second question is much more complex and raises the issue of de facto organs.

• 76. On the first question, the commentary to Article 4 of the ILC Articles simply states that “[w]here the law of a State characterizes an entity as an organ, no difficulty will arise”.Footnote ¹¹⁶ Without raising, for now, matters of clarity or ambiguity in the designation made by domestic law, one may ask: what are the obvious inclusions and exclusions, and what should guide this analysis? On this issue, guidance in the relevant authorities points both to aspects of domestic law which support the characterisation as an organ (mainly the terminology used and the [Page 43] nature/scope of functions, i.e. judicial, legislative and executive) and to aspects which do not affect such characterisation.

• 77. The latter are easier to begin with. The general commentary to Part I, Chapter II of the ILC Articles notes that “[t]he State as a subject of international law is held responsible for the conduct of all the organs, instrumentalities and officials which form part of its organization and act in that capacity, whether or not they have separate legal personality under its internal law”.Footnote ¹¹⁷ This observation applies to all attribution routes, not only Article 4. Yet, it was relied upon by the tribunal in Eureko v. Poland to reach the correct conclusion that the State Treasury, which in Poland has a separate legal personality, is nonetheless a State organ.Footnote ¹¹⁸ Similarly, as noted earlier, in Tethyan Copper v. Pakistan the acts of the province of Balochistan were attributed to Pakistan under Article 4, irrespective of Balochistan’s separate legal personality under domestic law.Footnote ¹¹⁹ This observation is significant because other tribunals have relied on the separate legal personality of an entity to conclude that it was not a State organ.Footnote ¹²⁰ The correctness of that conclusion in the factual circumstances of those cases, which concerned instrumentalities, must not blur the line drawn by the Commentary to the ILC Articles. Ministries, provinces, municipalities, etc., may have a separate legal personality and yet they are clearly State organs. Hence the significance of the reasoning in Eureko v. Poland and Tethyan Copper v. Pakistan on this point. Both tribunals also emphasised another aspect which does not affect the characterisation of a person or entity as an organ, namely the sovereign or commercial nature of its acts.

• 78. By contrast, aspects of domestic law that have been deemed very relevant to classify an entity as an organ include the terminology used in the law and the nature of the functions (not the acts) entrusted to it. Conduct by a “Minister”, as the agent of the State Treasury, irrespective of the nature of the act, was determinative in Eureko v. Poland.Footnote ¹²¹ At the opposite end of the spectrum, the characterisation of the Ghana Cocoa Board as a “corporate body” which can be “sued in its corporate name” under Ghanaian law was important in Hamester v. Ghana to conclude that the Board was not a State organ.Footnote ¹²² The nature of the functions entrusted to the relevant person or entity are also relevant. Thus, the executive functions of the State Treasury Minister in Eureko v. Poland Footnote ¹²³ or the Ministry of Petroleum in Unión Fenosa v. Egypt,Footnote ¹²⁴ and the judicial functions of the [Page 44] Bankruptcy Council, Court and Judge in Gavrilović v. Croatia,Footnote ¹²⁵ provided a strong indication that the relevant entities were State organs. It must be noted, however, that terminology and functions provide indications, but they are not determinative, and they are even less “prongs” of a “test”. The exercise of functions akin to judicial functions (e.g. arbitration of disputes) is not as such enough to conclude that the relevant person or entity is an organ, even if the function is expressly recognised in law. The characterisation is a context-specific inquiry, which pays particular attention to how domestic law relates to the status and functions of the relevant person or entity but must also take into account the latter’s factual position in the organisation of the State.

• 79. The latter point leads to the second question, namely the issue of de facto organs.Footnote ¹²⁶ Article 4(2) of the ILC Articles states that “[a]n organ includes any person or entity which has that status in accordance with the internal law of the State”. The term “includes” must be singled out because, as explained in the commentary, “in some systems the status and functions of various entities are determined not only by law but also by practice, and reference exclusively to internal law would be misleading”.Footnote ¹²⁷ Deferring entirely to domestic law for the characterisation of organs would imply that States could “avoid responsibility for the conduct of a body which does in truth act as one of its organs merely by denying it that status under its own law”.Footnote ¹²⁸ This is the implication that the word “includes” in paragraph 2 is intended to avoid. Thus, State organs can be de jure and de facto. To understand the category of de facto organs and its legal implications, it is necessary (i) to determine its locus within the ILC Articles, then (ii) examine the conditions of such characterisation, and finally (iii) clarify the scope of the conduct that can be attributed under this route.

• 80. The locus of the category of de facto organs in the ILC Articles can be clarified by reference to the ICJ’s judgment in the Bosnian Genocide case. In this case, the Court had to determine whether the massacres at Srebrenica could be attributed to the Federal Republic of Yugoslavia. When undertaking the analysis of attribution under the rule formulated in Article 8, the Court introduced some distinctions which are particularly relevant for present purposes:

  The Court must emphasize, at this stage in its reasoning, that the question just stated is not the same as those dealt with thus far. It is obvious that it is different from the question whether the persons who committed the acts of genocide had the status of organs of the Respondent under its internal law; nor however, and despite some [Page 45] appearance to the contrary, is it the same as the question whether those persons should be equated with State organs de facto, even though not enjoying that status under internal law. The answer to the latter question depends, as previously explained, on whether those persons were in a relationship of such complete dependence on the State that they cannot be considered otherwise than as organs of the State, so that all their actions performed in such capacity would be attributable to the State for purposes of international responsibility. Having answered that question in the negative, the Court now addresses a completely separate issue: whether, in the specific circumstances surrounding the events at Srebrenica the perpetrators of genocide were acting on the Respondent’s instructions, or under its direction or control. An affirmative answer to this question would in no way imply that the perpetrators should be characterized as organs of the FRY [Federal Republic of Yugoslavia], or equated with such organs. It would merely mean that the FRY’s international responsibility would be incurred owing to the conduct of those of its own organs which gave the instructions or exercised the control resulting in the commission of acts in breach of its international obligations. In other words, it is no longer a question of ascertaining whether the persons who directly committed the genocide were acting as organs of the FRY, or could be equated with those organs – this question having already been answered in the negative. What must be determined is whether FRY organs – incontestably having that status under the FRY’s internal law – originated the genocide by issuing instructions to the perpetrators or exercising direction or control, and whether, as a result, the conduct of organs of the Respondent, having been the cause of the commission of acts in breach of its international obligations, constituted a violation of those obligations.Footnote ¹²⁹

  The distinction made by the Court is very clear. Leaving aside de jure organs, factual powers over a person or entity may be framed through two main routes, that of a de facto organ (under Article 4) and that of instructions, direction or control (under Article 8). The conditions and implications of each route are entirely different. Conduct of a person or entity which is not a de facto organ (i.e. which is not attributable under Article 4) may still be attributed through the Article 8 route. By contrast, conduct of a de facto organ is assimilated to conduct of a de jure organ. In other words, it is conduct by an organ, so the question of whether a non-organ acts under the instructions, direction or control of an organ does not arise.

• 81. With respect to the conditions for a person or entity to be characterised as a de facto organ, the Court addressed this issue earlier in its judgment. Relying on its decision in the Nicaragua case,Footnote ¹³⁰ the Court identified the relevant standard as being one of “complete dependence” of the relevant person or entity on the [Page 46] respondent State.Footnote ¹³¹ The rationale underlying the rule, as noted in the Commentary to the ILC Articles, is to prevent a State from escaping responsibility by merely denying the person or entity the status of an organ in its domestic law. The independence of the person or entity must therefore be “purely fictitious”Footnote ¹³² and, as a result, the standard to be applied is very demanding. As the Court noted: “so to equate persons or entities with State organs when they do not have that status under internal law must be exceptional, for it requires proof of a particularly great degree of State control over them”.Footnote ¹³³ The burden of proof lies with the claimant. Relevant aspects to assess whether there is “complete dependence” include: degree of autonomy of the person or entity, evidenced for example in differences of view regarding which course to follow;Footnote ¹³⁴ a distinct or at least not unified chain of command;Footnote ¹³⁵ the options available to the State to enforce its control (organisation, training, financing, supply of equipment, selection and payment of leaders or key people) and their limitations;Footnote ¹³⁶ the actual use by the State of such means of control;Footnote ¹³⁷ and the timing of the conduct at stake compared to that of any support given or other control levers.Footnote ¹³⁸

• 82. The statement of general international law on this point by the ICJ commands particular authority but, given the differences between the operation of the aforementioned conditions in a context of armed conflict and in that of a foreign investment transaction, such statement can be usefully supplemented by a review of the practice of investment tribunals. Yet, adaptation of a rule to a certain context must not amount to a departure from or a relaxation of the requirements of general international law. The test remains that of “complete dependence”, but the degree of dependence requires a fact-intensive inquiry adapted to the context of foreign investment disputes.

• 83. As a general matter, tribunals have considered that entities which are not de jure organs and which have a separate legal personality are not, in principle, de facto organs.Footnote ¹³⁹ The relevance of a separate legal personality in this context (unlike the context of de jure organs) reflects the fact that legal personality cannot be presumed to be merely fictitious. On the contrary, the “complete dependence” of a separate legal entity on the State for attribution purposes must be established by the claimant;Footnote ¹⁴⁰ as the ICJ noted in the Bosnian Genocide case, the recognition that such is the case “must be exceptional, for it requires proof of a particularly great degree of State control over [the entity]”.Footnote ¹⁴¹

• [Page 47] 84. State ownership of a separate legal entity, including majority ownership, is as such insufficient to either establish “complete dependence”Footnote ¹⁴² or create a presumption of it.Footnote ¹⁴³ The burden of proof remains with the claimant. Cases where entities with a separate legal personality, which are not de jure organs, have been deemed to be de facto organs of the State must be understood as reflecting the particular circumstances of the legal relationship between the entity and the State. The fact that State ownership is insufficient to establish or presume complete dependence does not mean that it is irrelevant. In Nykomb v. Latvia, the tribunal concluded that a public enterprise, Latvenergo, which was wholly owned by the State and, in addition, operated in a way that evidenced no commercial freedom, was a “constituent part” or a “vehicle” of Latvia.Footnote ¹⁴⁴ The tribunal did not explicitly conclude that Latvenergo was a de facto organ. In fact, it did not even clarify under which rule codified in the ILC Articles attribution was possible, merely referring to the “rules of attribution in international law”.Footnote ¹⁴⁵ In Deutsche Bank v. Sri Lanka, reported in volume 19 of the ICSID Reports, the tribunal concluded that the acts of Ceylon Petroleum Corporation (CPC), a wholly owned company of the State, could be attributed to Sri Lanka. Ownership was, however, only one of several indications of the degree of dependence, which the tribunal summarised in paragraph 405 of the award. Such dependence had been recognised even by the Supreme Court of Sri Lanka, which had depicted CPC as “a government creation clothed with juristic personality so as to give it an aura of independence” with “deep and pervasive State control”.Footnote ¹⁴⁶ The tribunal’s conclusion seems compelling on the facts, although, from a legal standpoint, it is formulated somewhat ambiguously, in the following terms: “CPC’s actions would be attributable to the State, either because CPC is an organ of the State under ILC Article 4 or because CPC lacked separate legal existence, and/or acted under the instruction of the State.”Footnote ¹⁴⁷ The tribunal did not need to reach any firm conclusion on the attribution of CPC’s conduct, however, given its finding that conduct of the Supreme Court and the Central Bank of Sri Lanka was attributable to the State and in violation of investment obligations.Footnote ¹⁴⁸ In Flemingo DutyFree v. Poland, the Polish Airports State Enterprise (PPL), was deemed to be a de facto organ under Article 4. The tribunal took into consideration that PPL was owned and controlled by Poland. Although it found guidance in the 2000 decision in Maffezini v. Spain, which treats State ownership of an enterprise as a “rebuttable presumption that it is a State entity”, [Page 48] without however equating this to the concept of de facto organ, ownership alone was far from being the only or the most important factor in the tribunal’s assessment. The decisive factor, to which the tribunal “attached much importance”, was the declaration of the Secretary of State in the Ministry of Transport emphasising that PPL operated “within the structure of the Ministry” and that the Ministry was “also responsible for all issues connected with the functioning of the enterprise”.Footnote ¹⁴⁹ In Ampal v. Egypt, the tribunal referred en bloc to Articles 4, 5, 8 and 11 of the ILC Articles to conclude that the acts of two Egyptian corporate entities, the Egyptian General Petroleum Corporation (EGPC) and the Egyptian Natural Gas Holding Company (EGAS), were attributable to the State. On the evidence reviewed in the award, this was the correct conclusion, and Egypt did not challenge it. Yet, it is unclear whether attribution under the rule of Article 4 was based on the characterisation of the entities as de facto organs or as de jure organs.Footnote ¹⁵⁰ The claimants in the case had argued that the entities were de facto organs, but the tribunal was ambiguous on this point, possibly due to doubts about how to handle the separate legal personality. As discussed earlier, legal personality is not relevant in the context of de jure organs, and it is only relevant in that of de facto organs because it suggests a certain degree of autonomy incompatible with “complete dependence”.

• 85. As this overview of the case law suggests, State ownership is only one relevant factor among several others. In order to provide further clarity to what is essentially a fact-intensive inquiry to determine the degree of dependence, some indicative factors or sets thereof have been used in the case law. At the outset, it must be observed that the “structural” and “functional” tests alluded to in Maffezini, which were not expressly developed for de facto organs, were introduced before the completion of the ILC Articles. No less importantly, they have no specific connection with the “complete dependence” standard restated by the ICJ in the Bosnian Genocide case and, more generally, they no longer reflect either general international law or arbitral practice on the issue of de facto organs, if they ever did.Footnote ¹⁵¹ More recent decisions assess the degree of dependence in the light of three main sets of factors. In Almås v. Poland, reported in this volume, another tribunal chaired by Judge Crawford considered “factors, such as the performance of core governmental functions, direct day-to-day subordination to central [Page 49] government, or lack of all operational autonomy”.Footnote ¹⁵² Looking at these factors, it rejected the allegation that the Polish Agricultural Property Agency was a de facto organ.Footnote ¹⁵³ In the more recent award rendered in Ortiz v. Algeria, also reported in this volume, the tribunal expressly referred to the Bosnian Genocide case in relation to the “complete dependence” standard as well as to the Almås award in respect of the three “criteria” to assess such dependence.Footnote ¹⁵⁴ The tribunal then examined the situation of three entities and concluded that none of them could be considered a de facto organ.

• 86. Having reviewed the locus of de facto organs within the ILC Articles and the conditions for the characterisation of a person or entity as such, it remains to clarify the scope of the conduct that can be attributed under this route. The answer to this third question is simpler: once a person or an entity has been characterised as a de facto organ, it is a State organ and, as for de jure organs, all conduct in an official capacity (as opposed to private conduct) is attributable. The scope of attributable conduct is thus wider than that under Article 5 of the ILC Articles, which is limited to specific acts in the exercise of governmental authority (therefore excluding commercial acts), or under Article 8, which is limited to specific acts under the instructions, direction or control of the State (therefore excluding, in principle, acts ultra vires). The broader scope of conduct that can be attributed under Article 4, as compared to other attribution routes (mainly Article 5), explains why claimants in investment proceedings often argue that a State instrumentality or an enterprise partly or fully owned by a State is so entirely dependent on the State that it is a de facto organ. But, as discussed earlier, such a finding must be exceptional.

4.3. Conduct of instrumentalities (Article 5)

• 87. The conduct (actions or omissions) of separate legal entities which are not State organs but exercise elements of governmental authority can be attributed to the State under certain conditions.Footnote ¹⁵⁵ This is a rule of customary international lawFootnote ¹⁵⁶ codified in Article 5 of the ILC Articles:

  The conduct of a person or entity which is not an organ of the State under article 4 but which is empowered by the law of that State to exercise elements of the governmental authority shall be considered an act of the State under international law, provided the person or entity is acting in that capacity in the particular instance.

  There is a significant body of case law from investment tribunals on the operation of the rule codified in Article 5 of the ILC Articles, which is understandable given the frequent involvement of State instrumentalities in dealings with foreign [Page 50] investors. In clarifying the scope of this rule, three aspects are important: (i) the nature of the person or entity; (ii) the requirements for the attribution of conduct; and (iii) the scope of the acts that may be attributed through this route.

• 88. Regarding the nature of the relevant person or entity, a threshold matter is that it must not be a State organ, whether de jure or de facto. It is therefore not possible for a person or an entity to be considered at the same time a de facto organ (Article 4) and a State instrumentality (Article 5). This is significant because there are sometimes ambiguities in the way tribunals reach a finding of attribution;Footnote ¹⁵⁷ stating that an entity is both a State organ and a State instrumentality is contradictory. Aside from this threshold matter, the entities whose conduct may be attributed under Article 5 are quite diverse. The Commentary to the ILC Articles notes that “[t]he generic term ‘entity’ reflects the wide variety of bodies which, though not organs, may be empowered by the law of a State to exercise elements of governmental authority. They may include public corporations, semi-public entities, public agencies of various kinds and even, in special cases, private companies.”Footnote ¹⁵⁸ The only – and crucial – requirement is that “in each case the entity is empowered by the law of the State to exercise functions of a public character normally exercised by State organs”.Footnote ¹⁵⁹ Hence the frequent characterisation of this attribution route as a “functional test”.Footnote ¹⁶⁰ This, as well as the requirement that the specific conduct of the entity which the claimant seeks to attribute to the respondent is an “exercise of governmental authority concerned” makes such conduct attributable.Footnote ¹⁶¹

• 89. These two requirements are generally understood as a two-element test recognised, expressly or implicitly, in virtually all decisions addressing Article 5 of the ILC Articles.Footnote ¹⁶² It has been deemed to be a “residual” test applicable in the absence of a lex specialis.Footnote ¹⁶³ However, the analysis of each criterion in the [Page 51] different cases varies significantly in the level of detail and, more importantly, in the level of scrutiny displayed by tribunals. It is therefore useful to flesh out the basic parameters that must guide the examination of each criterion.

• 90. With respect to the requirement that the “person or entity … is empowered by the law of that State to exercise elements of the governmental authority”, the main parameters concern the terms “empowered by the law of that State”, the meaning of “exercise” and that of “elements of the governmental authority”.

• 91. The type of persons or entities “empowered by the law of [the] State” is presented in the Commentary to the ILC Articles as a “narrow category”.Footnote ¹⁶⁴ The narrow boundaries of the category are given by the fact that “internal law … must specifically authorize the conduct as involving the exercise of public authority”.Footnote ¹⁶⁵ Thus, non-specific, broad or implicit authorisations are insufficient to meet the first criterion. There is no space for de facto State instrumentalities under Article 5. Either the internal law specifically vests governmental authority in the person or entity, or this parameter is not met. In Unión Fenosa v. Egypt, the tribunal rejected the attribution ground under Article 5 made by the claimant because the latter had failed to show a “provision of Egyptian law ‘specifically authorising’ EGPC [the national oil company] to conclude the SPA [a natural gas sale and purchase agreement] in the exercise of the Respondent’s public authority”.Footnote ¹⁶⁶ In Ortiz v. Algeria, the tribunal observed, when examining the specific situation of the entities in question, that the specific delegation of governmental authority must concern the type of acts concerned by the dispute.Footnote ¹⁶⁷ This connects the first and the second requirements of Article 5. It is not enough for the entity to have been delegated public powers in a certain sphere; the delegation must specifically concern the sphere where the acts allegedly in breach of international law were performed.

• 92. Another parameter of the first requirement is the meaning of the term “exercise”. In Tulip Real Estate v. Turkey, the claimant argued that a Turkish real estate investment trust, Emlak, was authorised to exercise elements of governmental authority because a zoning law granted Emlak certain preferential treatment with respect to construction permits and the purchase of land. The tribunal dismissed the argument because these advantages in no way empowered Emlak to actually “exercis[e] elements of governmental authority vis-à-vis any particular object or person”.Footnote ¹⁶⁸ Taken together, the first and second parameters mean that what must be “specifically authorised” is the “exercise” of public authority with respect to an object or person, as a government might otherwise perform.

• 93. That leads to the third and most important parameter, the definition of “elements of the governmental authority”. The Commentary to the ILC Articles observes that:

  [Page 52] Article 5 does not attempt to identify precisely the scope of “governmental authority” for the purpose of attribution of the conduct of an entity to the State. Beyond a certain limit, what is regarded as “governmental” depends on the particular society, its history and traditions. Of particular importance will be not just the content of the powers, but the way they are conferred on an entity, the purposes for which they are to be exercised and the extent to which the entity is accountable to government for their exercise. These are essentially questions of the application of a general standard to varied circumstances.Footnote ¹⁶⁹

  The four criteria identified in this paragraph were examined by the tribunal in Ortiz v. Algeria by reference to previous cases and scholarship. In this light, (i) the content of the powers means whether the delegation authorises the exercise of acta jure imperii (sovereign acts or public prerogatives), rather than mere acta jure gestionis (commercial acts), (ii) the mode of conferral concerns the nature of the act, legislation, regulation, a decree, a contract, whereby the authority is conferred, (iii) the purpose refers to the nature of the goals, public or private, in pursuit of which the powers are conferred, and (iv) the accountability criterion concerns the degree of public oversight to which the person or entity is subject.Footnote ¹⁷⁰ By contrast, as the Commentary to the ILC Articles notes:

  The fact that an entity can be classified as public or private according to the criteria of a given legal system, the existence of a greater or lesser State participation in its capital, or, more generally, in the ownership of its assets, the fact that it is not subject to executive control – these are not decisive criteria for the purpose of attribution of the entity’s conduct to the State.Footnote ¹⁷¹

  The difference between acta jure imperii and acta jure gestionis, which will be examined later in this study (Section 5.2), is therefore decisive already at the level of the first requirement.

• 94. Yet, even when the powers specifically delegated by domestic law include public prerogatives, it is still necessary that the specific acts allegedly in breach of international law involved an actual exercise of such prerogatives. This is the second requirement of the attribution rule codified in Article 5 of the ILC Articles. At this stage, the difference between the exercise of public prerogatives and the performance of merely commercial acts is once more essential. The focus of this analysis must be not on the overall powers of the person and entity but only on the nature of the acts specifically challenged by the claimant. This analysis is sometimes difficult because claimants tend to refer to a series of acts or to conduct in general, whereas, as noted by the tribunal in Hamester v. Ghana, the inquiry relating to the second requirement must be conducted “into each and every act”.Footnote ¹⁷² In this case, the tribunal considered that arm’s-length horizontal negotiations relating to the price of a natural resource (rather than an officially imposed price) were an indication of the commercial nature of an entity’s dealings with the [Page 53] investor.Footnote ¹⁷³ It also considered that the violation of a contractual term by an entity is not enough, as such, to be an exercise of a public prerogative.Footnote ¹⁷⁴ Similarly, a dispute among shareholders had to be seen as a common occurrence in commercial dealings.Footnote ¹⁷⁵

• 95. Arguments about the specific nature of an act are fact-intensive and case-specific. Some parameters that can guide the assessment include, first and foremost, whether the relevant act could be performed, in general, by an entity devoid of public authority. Typically, such is the case of the ability to enter into a contract or to exercise a contractual right, including termination of a contract .Footnote ¹⁷⁶ Even when a contract is entered into in a public capacity, the exercise of a contractual right (as the specific act relevant for attribution) is not necessarily an exercise of a public prerogative. In Almås v. Poland, the tribunal clearly distinguished the two:

  It is true that it [ANR, the Polish Agricultural Property Agency] entered into the relevant contract in the exercise of statutory powers to manage State agricultural property. But the key point is that vis-à-vis the Claimants, termination was not an exercise of public power but of a purported contractual right. The management of real property, including the exercise of the contractual right to terminate a lease, derives from the general law; it is a capacity of any entity that holds and rents out land. Because ANR was not exercising whatever Polish government authority it may have had when it terminated the Lease, its action is not attributable to the Polish State on the basis of Article 5.Footnote ¹⁷⁷

  The claimants tried to counter this conclusion by reference to two arguments, namely that the exercise of the contractual right had been unlawful and that it was motivated by policy reasons. The tribunal rejected both grounds.Footnote ¹⁷⁸ It must be noted that the tribunal, in reviewing these arguments, at no point expressly accepted that the unlawful exercise of a contractual right or a public policy motive underpinning the exercise of a contractual right was, as such, a sufficient ground to turn such conduct into an exercise of a public prerogative. In fact, it did not deem it necessary to determine whether the exercise of the contract had been lawful. This is consistent with the reasoning of the tribunal in Hamester v. Ghana, according to which the violation of a contractual provision does not make it an act in the exercise of governmental authority.Footnote ¹⁷⁹ Any contractual partner can breach a contractual term and any contractual partner can exercise a contractual right for a range of motives, which do not make the act an exercise of a public prerogative. A different question is whether the contractual right is exercised by the entity under the instructions, direction or control of a State organ (see Section 4.4).

• [Page 54] 96. The scope of attribution under the rule codified in Article 5 of the ILC Articles is also limited by the focus on “each and every act”.Footnote ¹⁸⁰ This is a significant difference between State organs and the instrumentalities envisioned in Article 5. In the former case, all conduct, whether sovereign or commercial, is attributable, whereas in the latter case, only specific acts in the exercise of public prerogatives can be attributed. Hence the frequent attempts by claimants to argue that an instrumentality must be considered a de facto organ, which requires a far more demanding showing of “complete dependence”. Occasionally, tribunals have overlooked this important difference by conflating the scope of the delegation of powers with the scope of acts that can specifically be attributed under Article 5.Footnote ¹⁸¹ This is incorrect because the specific delegation of powers (which must be explicitly made by law) is only one of the two requirements for attribution under Article 5.

• 97. Another aspect relevant to the scope of attribution is the question of ultra vires acts (see Section 5.4). Article 7 of the ILC Articles expressly refers to the attribution routes of Article 5 when it states that:

  The conduct of an organ of a State or of a person or entity empowered to exercise elements of the governmental authority shall be considered an act of the State under international law if the organ, person or entity acts in that capacity, even if it exceeds its authority or contravenes instructions [emphasis added].

  Thus, an act in the exercise of governmental authority which exceeds the public prerogatives conferred upon the instrumentality will remain attributable as long as (like for State organs) the person or entity is not acting in a purely private capacity.

4.4. Conduct directed or controlled by a State (Article 8)

• 98. Another attribution rule which has frequently featured in the case law of investment tribunals is codifiedFootnote ¹⁸² in Article 8 of the ILC ArticlesFootnote ¹⁸³ in the following terms:

  [Page 55] The conduct of a person or group of persons shall be considered an act of a State under international law if the person or group of persons is in fact acting on the instructions of, or under the direction or control of, that State in carrying out the conduct.

  In the Bosnian Genocide case, the ICJ clarified that, in order for a specific act or conduct to be attributable under this rule, the person or group of persons must have:

  acted in accordance with that State’s instructions or under its “effective control”. It must however be shown that this “effective control” was exercised, or that the State’s instructions were given, in respect of each operation in which the alleged violations occurred, not generally in respect of the overall actions taken.Footnote ¹⁸⁴

  This test, which had originally been developed in the Nicaragua case,Footnote ¹⁸⁵ has been taken up in a series of investment decisions.Footnote ¹⁸⁶ It is therefore well established. As before, in order to understand the operation of this attribution route, it is useful to (i) identify the locus of this route within the ILC Articles, (ii) analyse the requirements for conduct to be attributed, and (iii) clarify the scope of the conduct that may be attributed.

• 99. Regarding the locus of this attribution route, as noted earlier in this study, the rule formulated in Article 8 concerns situations which are different from the characterisation of a person or entity as a de facto organ. Whereas the latter is subject to a test of “complete dependence” and makes all conduct (rather than specific acts) attributable, the former is governed by a test of “effective control” and can only lead to the attribution of specific acts.Footnote ¹⁸⁷ Acts attributed under Article 8 can render a State potentially responsible (if there is a breach of an international obligation) for the action of its organs consisting in giving instructions or exercising effective control over certain persons or a group of persons. This means that Article 8 may operate in parallel with Article 4 (e.g. the instruction or effective control must be given or exercised by an organ under Article 4, and it must be proven that a person or group of persons acted upon the instruction under Article 8). It may also operate in parallel with Article 5 to the extent that a State instrumentality enjoying governmental authority and using it in a specific case may be giving an instruction or exercising effective control.

• 100. Moving to the requirements of the “effective control” test, in the Bosnian Genocide case, the ICJ distinguished it from the “complete dependence” test as follows:

  The test thus formulated differs in two respects from the test – described above – to determine whether a person or entity may be equated with a State organ even if not having that status under internal law. First, in this context it is not necessary to show that the persons who performed the acts alleged to have violated international law [Page 56] were in general in a relationship of “complete dependence” on the respondent State; it has to be proved that they acted in accordance with that State’s instructions or under its “effective control”. It must however be shown that this “effective control” was exercised, or that the State’s instructions were given, in respect of each operation in which the alleged violations occurred, not generally in respect of the overall actions taken by the persons or groups of persons having committed the violations.Footnote ¹⁸⁸

• 101. Attribution under the “effective control” test must be (much like a finding of complete dependence) exceptional.Footnote ¹⁸⁹ The burden is on the party alleging attribution to establish two sets of elements: the action of the State (the “control” element) and the action of the relevant “person or group of persons” (those “effectively” controlled).

• 102. The first set of elements concerns several parameters which have received attention in the case law, most notably the disjunctive character of the terms “instructions”, “direction” and “control”, their imperative character, the intended result, the level of command involved in the exhortation to act, and the form of communication.

• 103. Article 8 speaks of “instructions of, or under the direction or control of, that State”. This formulation includes three disjunctive elements, namely “instructions”, “direction” and/or “control”. The tribunal in Tulip Real Estate v. Turkey concluded from this disjunctive character that it “need[ed] only be satisfied that one of those elements is present in order for there to be attribution under Art. 8”.Footnote ¹⁹⁰ However, it must not follow from this diversity of elements that there are three different tests. In Tulip Real Estate v. Turkey, the tribunal reasoned that the “effective control” test applies to the “control” element.Footnote ¹⁹¹ This is correct but only partially so. The “effective control” is the umbrella test, some elements of which include “control” but also “instructions”, “direction” and other elements. It would be incorrect to conclude that the assessment of acts performed under the “instruction” or the “direction” of the State is different from the “effective control” test. In Tulip Real Estate v. Turkey, the tribunal did not go as far, and it actually dismissed the attribution argument under Article 8 for lack of specific control over the act allegedly in violation of international law.Footnote ¹⁹² The disjunctive character is instead meant to capture situations where “instructions”, “direction” or “control” do not fully converge. For example, a person under the effective control of a State may receive the instruction to destroy a military target without however injuring civilians or damaging civilian property. The act may unfold under the effective control of the State but, in the course of being performed, a civilian may be injured. In such a case, even though the instruction has not been strictly followed, the act under the control of the State would remain attributable.Footnote ¹⁹³ The ILC Commentary notes that close calls like this example “can be resolved by asking [Page 57] whether the unlawful or unauthorized conduct was really incidental to the mission or clearly went beyond it”.Footnote ¹⁹⁴ Yet, if a person receives the instruction to perform an act and she does not perform the act required from her, the mere issuance of an instruction would not make the non-compliant part of her act attributable.Footnote ¹⁹⁵ As further discussed in the context of the second set of elements, mere instructions without a showing of effective control are insufficient for attribution.

• 104. The rule formulated in Article 8 does not require any specific form in the nature of the “instructions” or “direction”, or in the way they are given. They may be formulated in writing or given orally, and they may appear in a variety of acts. Yet, the very meaning of the terms “instruction” or “direction” exclude pure recommendations or suggestions or general policies. In Hamester v. Ghana, a general policy on equitable cocoa distribution was deemed to be insufficient to convey an instruction or direction to the relevant entity, Cocobod.Footnote ¹⁹⁶ This is not because of the form or the medium of communication of the policy but because such a general policy could not be said to contain any specific instruction or direction. The instruction or direction must effectively command authority, whether because it is legally binding or because, in practice, it is imperative.Footnote ¹⁹⁷ What is imperative in practice depends on the specific political and administrative context of the instruction or directive. Thus, an invitation to negotiate could not be considered an instruction.Footnote ¹⁹⁸ Even a more active exhortation would fall short of what is required if the addressee has some margin for her own decision-making and could realistically select a different course of action. As long as she “owns” the decision, the act would not be attributable. This is yet another indication that the “effective control” test remains the overarching framework.

• 105. The Commentary to the ILC Articles refers to the use of control “in order to achieve a particular result”.Footnote ¹⁹⁹ In EDF v. Romania, the tribunal approached these terms as setting a subjective test, namely that there must be a mismatch between what the addressee of the instruction or direction perceives to be its commercial interest and the course of action imposed by the State.Footnote ²⁰⁰ Whereas such a mismatch would provide evidence that the State drove the decision eventually taken, it is not a requirement under the rule of Article 8. Indeed, a State may issue a clear instruction to an entity requiring it to act in a certain way which is also advantageous commercially. Although no mismatch would be present, the entity would still be acting on the instructions of the State. It is true that, in this case, the lack of autonomy of the addressee of the instruction would be less clear, but the situation would still fall under Article 8. This is particularly the case in light of the difficulties of ascertaining what is in the best interests of an entity. Many decisions are taken in a context of uncertainty where there are reasonable grounds [Page 58] to follow more than one course of action. The fact that the State’s preferred course of action is also the one perceived to be the best by the entity does not as such exclude attribution, if the entity could not have followed a different course of action.

• 106. In some cases, the level of command necessary for an exhortation to amount to an “instruction” or “direction” has been modulated to take into account a highly concentrated governmental structure. As noted earlier in this study, in Bayindir v. Pakistan, the tribunal concluded that mere “clearance” from the then military ruler of Pakistan, General Musharraf, was sufficient to consider the decision of a State instrumentality as attributable to the State. The Bayindir test is not representative of the exceptional character of attribution under Article 8 of the ILC Articles. It can only be understood, in the specific circumstances of the case, as a combination of a context (a highly concentrated political rule) with an act which, in principle, falls short of giving a specific “instruction” or “direction” and operates as permission. Yet, such permission is not an “acknowledgement and adoption” of the conduct, in the meaning of the rule formulated in Article 11 of the ILC Articles, because it takes place before the act is performed. In this conceptual area between Articles 8 and 11, such a “permission” should not in principle warrant attribution. This is because “permission” differs from “instruction” and “direction” at two levels: impulsion and specificity. “Permission” presupposes that the impulsion for the act does not come from the State organ but from the person or group of persons whose acts the claimant seeks to attribute. That implies some level of autonomy in decision-making, although a decision cannot, in practice, proceed without the endorsement of the State organ. In addition, “permission” does not imply that the State organ is aware of all the details, and it may be more akin to an “I leave it in your hands” form of clearance, which again would emphasise a certain margin of manoeuvre in the future conduct of the person or group of persons. Such a relaxation of the requirements could not serve as a standard, and it does not reflect general international law. The Bayindir tribunal was likely aware of this, and it signalled the departure by reference to the context of international economic law (quoted above, but worth recalling):

  the Tribunal is aware that the levels of control required for a finding of attribution under Article 8 in other factual contexts, such as foreign armed intervention or international criminal responsibility, may be different. It believes, however, that the approach developed in such areas of international law is not always adapted to the realities of international economic law and that they should not prevent a finding of attribution if the specific facts of an investment dispute so warrant.Footnote ²⁰¹

  However, this is not sufficient to justify a departure. As noted by the ICJ in the Bosnian Genocide case, “the rules for attributing alleged internationally wrongful conduct to a State do not vary with the nature of the wrongful act in question in the absence of a clearly expressed lex specialis”.Footnote ²⁰² The only reason why such a conclusion may be justified lies in the highly concentrated nature of political rule [Page 59] in Pakistan under General Musharraf. In that regard, it is not the permission as such which would make the act attributable but the confluence of three other elements, namely the highly concentrated political rule in one person, the fact that the permission comes directly from this person (rather than other State organs), and the manifest ability of the claimant to prove such clearance. In the absence of these clearly circumscribed conditions for the relaxation of the requirements of attribution under Article 8, the risk would be that any act in a politically concentrated regime would be deemed to be attributable on the assumption that it can only proceed with the approval of the State. However justified the conclusion may have been in the factual circumstances of Bayindir v. Pakistan, it must not be understood as a relaxation of the requirements of the rule in general international law. The tribunal in Hamester v. Ghana observed, indeed, that “being informed and discussing the case with the parties – both the Claimant and Cocobod – does not mean that the latter was under the effective control of the Government”.Footnote ²⁰³ Awareness and laissez-faire can certainly not be equated to “instruction”, “direction” or “control” (nor with “acknowledgement and adoption” under Article 11, as discussed in Section 4.5). Inaction from State organs is attributable to a State and may amount to a breach of international law but only when there is a duty to act and other conditions are met.Footnote ²⁰⁴ In all events, this is different from the rule formulated in Article 8.

• 107. Another aspect of the requisite “instruction” or “direction” – in addition to its imperative nature, its intended result and its context – concerns the form of its communication to its addressees. As noted earlier, international law does not require any specific form or communication channel. Yet, the form of communication may be such that an exhortation falls short of a governmental instruction or direction. Thus, in the Tehran Hostages case, the ICJ considered that the public declarations by Ayatollah Khomeini exhorting militants to attack the United States and Israel were not sufficient to make the taking of the US Embassy an act of the State:

  In the view of the Court, however, it would be going too far to interpret such general declarations of the Ayatollah Khomeini to the people or students of Iran as amounting to an authorization from the State to undertake the specific operation of invading and seizing the United States Embassy. To do so would, indeed, conflict with the assertions of the militants themselves who are reported to have claimed credit for having devised and carried out the plan to occupy the Embassy.Footnote ²⁰⁵

  The Court went on to conclude that Iran had engaged its international responsibility due to its inaction in the face of this first phase of events, and its subsequent acceptance of the acts that ensued, but the relevant acts and attribution routes are different from the one now formulated in Article 8 of the ILC Articles (see Section 4.5).

• [Page 60] 108. So far, I have discussed in some detail one set of elements of the “effective control” test, that concerning the action of the State (or “control” element). I now turn to the other set of elements, that relating to the action of the relevant “person or group of persons” which must be “effectively” controlled. This second set includes three main parameters, namely the nature of the addressees, the meaning of “general control” and that of “specific control”.

• 109. The expression “person or set of persons” in Article 8 is intended to encompass both persons with a legal personality (whether physical or legal persons) and groups which do not have legal personality as such. The Commentary to the ILC Articles expressly notes that “while a State may authorize conduct by a legal entity such as a corporation, it may also deal with aggregates of individuals or groups that do not have legal personality but are nonetheless acting as a collective”.Footnote ²⁰⁶ Significantly, unlike the rules codified in Articles 4 and 5, a person or group of persons under Article 8 may be acting in their private capacity (i.e. irrespective of any appearance of “official capacity”), as long as they are effectively controlled.

• 110. From the standpoint of these persons or this group of persons, effective control requires that those who performed the acts allegedly in breach of international law are under both the “general” and the “specific” control of the State.Footnote ²⁰⁷ In other words, it is not sufficient to prove that the State generally controls an entity but, in addition, it must be established that it used its controlling power to make the entity perform the specific acts allegedly in breach. The Commentary to the ILC Articles clarifies this difference by reference to an example relating to State-owned or controlled companies:

  international law acknowledges the general separateness of corporate entities at the national level, except in those cases where the “corporate veil” is a mere device or a vehicle for fraud or evasion. The fact that the State initially establishes a corporate entity, whether by a special law or otherwise, is not a sufficient basis for the attribution to the State of the subsequent conduct of that entity. Since corporate entities, although owned by and in that sense subject to the control of the State, are considered to be separate, prima facie their conduct in carrying out their activities is not attributable to the State unless they are exercising elements of governmental authority within the meaning of article 5. … On the other hand, where there was evidence that … the State was using its ownership interest in or control of a corporation specifically in order to achieve a particular result, the conduct in question has been attributed to the State.Footnote ²⁰⁸

• 111. State ownership of an enterprise may be evidence of “general control”, but the existence of a separate corporate entity creates a presumption to the contrary. Indeed, as noted by the commentary, “prima facie their conduct in carrying out their activities is not attributable to the State”. State ownership is of course relevant, particularly if the State is the sole or a majority shareholder, but it is [Page 61] neither sufficient nor necessary to establish general control. A State may own a company but allow it to act independently in the pursuit of its business. Also, a State may be a minority shareholder but, given certain arrangements in a shareholder agreement or other aspects of the enterprise’s operations, it may nevertheless exercise general control. If a minority shareholder may exercise control, that also means that majority shareholding is not, in and of itself, sufficient to establish general control.

• 112. In all events, establishing “general control” is only a first step. For the conduct to be attributable, the claimant must also establish “specific control”, namely that the specific acts allegedly in breach of international law (and not the general operation of the company or other specific acts) were effectively performed as a result of the State’s instructions, direction or control.

• 113. In EDF v. Romania, the tribunal examined the evidence on the record to conclude that the Ministry of Transport had used its ownership and control of two corporations, which were themselves shareholders of a third one, in order to make the latter terminate or refrain from extending its contractual arrangements with EDF and introduce instead a system of auctions for commercial spaces in the Otopeni Airport.Footnote ²⁰⁹ By contrast, in Gavrilović v. Croatia, the tribunal admitted arguendo that the State exercised general control over the relevant entity but concluded that there was “insufficient evidence that the Respondent exercised specific control over the relevant acts at that time”.Footnote ²¹⁰ Similarly, in Tulip Real Estate v. Turkey, the tribunal noted that the State was “capable of exercising a degree of control over Emlak to implement elements of a particular state purpose”.Footnote ²¹¹ However, it emphasised that “the relevant enquiry remains whether Emlak was being directed, instructed or controlled by TOKI with respect to the specific activity of administering the Contract with Tulip JV”,Footnote ²¹² which the tribunal rejected. Yet another example is provided by Ortiz v. Algeria. In its analysis of attribution for one of the entities at stake, the tribunal distinguished the two levels of control, general and specific, admitting the first but rejecting the second:

  While the Tribunal is ready to accept that the Algerian State exercised an overall control over [the State-owned construction company] OLA/ALRECC, at a pinch based on the fact that the State owns 100% of OLA/ALRECC’s share capital and the CPE [a government body responsible for State holdings and privatisation] authorised the [State-owned real estate company] SGP Indjab to institute the joint-venture and therefore the partnership between OLA/ALRECC and Ortiz, [the Tribunal] however detects no convincing element in the record that would allow to conclude that the State gave specific instructions or directions to OLA/ALRECC or that it allegedly exercised a specific control over OLA/ALRECC in relation to the three examples mentioned in the previous paragraph [three acts allegedly in breach]. As explained above, Article 8 requires a party to demonstrate the existence of an [Page 62] instruction or a direction or an effective control, that is both general and specific, and it does not suffice to contend that the activity of OLA/ALRECC in general was under State control, as the Claimant suggests.Footnote ²¹³

  Thus, the proof of each level of control remains distinct. General control over an entity by no means implies specific control over one or more acts performed by that entity. For attribution purposes, it is the latter that matters.

• 114. This clarification leads directly to the question of the scope of attribution under the rule formulated in Article 8. This attribution route concerns specific acts under instructions, direction or control, not overall conduct of a generally controlled entity. It is possible that a series of acts may be the result of a specifically instructed, directed or controlled act, in which case all the series would be attributable. In Bayindir v. Pakistan, the tribunal considered that “each specific act allegedly in breach of the Treaty was a direct consequence of the decision of the NHA to terminate the Contract, which decision received express clearance from the Pakistani Government”.Footnote ²¹⁴

• 115. Like in the context of Article 4, whether the acts in question are sovereign or commercial in nature makes no difference for attribution under Article 8 because they can be traced back to a State organ. This is expressly stated in the Commentary to the ILC ArticlesFootnote ²¹⁵ as well as in some decisions from investment tribunals.Footnote ²¹⁶

• 116. However, unlike Article 4, acts performed in a private capacity (by contrast to acts in an “official capacity”, i.e. “carried out by persons cloaked with governmental authority”)Footnote ²¹⁷ can be attributed under Article 8. The commentary specifies that “it does not matter that the person or persons involved are private individuals”.Footnote ²¹⁸ This is because the rationale for attribution is that the State itself is behind the act, irrespective of the front person.

• 117. Finally, unlike both Articles 4 and 5, in principle ultra vires acts cannot be attributed under Article 8.Footnote ²¹⁹ This is again in the nature of this attribution route, which is based on the fact that the person or persons performing the act are effectively doing so on the basis of a State command. To the extent that they depart from that command, their action is no longer effectively controlled. The only exception, noted earlier, concerns conduct departing from specific instructions which is “really incidental to the mission” rather than “clearly … beyond it”.Footnote ²²⁰ In Ortiz v. Algeria, the tribunal considered that a specific action that, [Page 63] according to the claimant, had violated a governmental mandate could not, as a result, be attributed under Article 8.Footnote ²²¹

4.5. Conduct acknowledged and adopted by a State (Article 11)

• 118. A fourth attribution route, which is far less frequently relied upon in the case law, concerns conduct which is initially not attributable to the State but that becomes so as a result of the latter’s acknowledgement and adoption.Footnote ²²² This rule is codifiedFootnote ²²³ in Article 11 of the ILC Articles, according to which:

  Conduct which is not attributable to a State under the preceding articles shall nevertheless be considered an act of that State under international law if and to the extent that the State acknowledges and adopts the conduct in question as its own.

• 119. Although the operation of this rule appears simple at first sight, and tribunals have applied it with little detailed reasoning,Footnote ²²⁴ on closer examination such operation may become very complex. As before, I will examine (i) its locus within the ILC Articles, (ii) the conditions for attribution of conduct under this route, and (iii) the scope of conduct which may be attributed.

• 120. Regarding the locus of the rule, the Commentary to the ILC Articles notes that, unlike all other bases of attribution in general international law (with the sole exception of conduct of insurrectional or other movements under Article 10 of the ILC Articles), Article 11 “provides for the attribution to a State of conduct that was not or may not have been attributable to it at the time of commission, but which is subsequently acknowledged and adopted by the State as its own”.Footnote ²²⁵ Specifically, whereas the instructions, directions or control under Article 8 must precede the allegedly wrongful act, as otherwise the acts would not be under the “effective control” of the State, Article 11 takes as a starting point that the relevant acts have already taken place or that their performance began without State involvement.

• 121. More generally, consistently with the nature of the rules of State responsibility for internationally wrongful acts, attribution under Article 11 says nothing about actual breach. Whereas this may seem a trite observation, the complexity under Article 11 comes from the fact that acts which may not be unlawful under international law (e.g. because the private persons performing them are not bound by, for example, international obligations relating to the protection of consular and diplomatic premises and staff) may become so if attributed to the State, which is bound by an applicable international obligation.Footnote ²²⁶ Thus, attribution changes the [Page 64] law in the light of which the lawfulness will be assessed, but it leaves the question of breach entirely open.

• 122. The two main contexts in which the rule formulated in Article 11 is likely to be relevant concern situations of State succession, where “the successor State, faced with a continuing wrongful act on its territory, endorses and continues that situation”,Footnote ²²⁷ and situations of mass revolts,Footnote ²²⁸ hence the close relationship with Article 10 on insurrectional and other movements. The difference between these two contexts lies in the possible application of specific rules of State succession to responsibility for internationally wrongful acts which, although they were reserved in Article 39 of the 1978 Vienna Convention on State Succession to Treaties,Footnote ²²⁹ have subsequently benefited from sustained clarification efforts, most notably by the Institut de Droit InternationalFootnote ²³⁰ and the ongoing work of the ILC.Footnote ²³¹ Investment disputes most often arise beyond the context of State succession.

• 123. With respect to the conditions for attribution under this route, the principle, expressly acknowledged in the Commentary to the ILC Articles, is that acts of private persons are not attributable to the State.Footnote ²³² Thus, attribution under Article 11 is the exception, not the rule, and the relevant conditions must be established by the claimant. There are three cumulativeFootnote ²³³ conditions in the expression “if and to the extent that the State acknowledges and adopts the conduct in question as its own”. The State must: “acknowledge” the specific conduct in question; “adopt” it “as its own”; to an “extent” which will determine what can be attributed. The latter condition merges with the scope of attribution, but it has some aspects relating to the definition of the extent that must be clarified beforehand. In addition, there is a question of “form” in both the acknowledgement and the adoption which also requires some comment.

• 124. The “acknowledgement” requires specific identification of the relevant conduct.Footnote ²³⁴ General acknowledgement, approval or support is not enough. As noted in the Commentary to the ILC Articles:

  as a general matter, conduct will not be attributable to a State under article 11 where a State merely acknowledges the factual existence of conduct or expresses its verbal approval of it. In international controversies, States often take positions which amount to “approval” or “endorsement” of conduct in some general sense but do not involve any assumption of responsibility.Footnote ²³⁵

• [Page 65] 125. The relevant conduct must not only be specifically identified through the acknowledgement, but the State must “adopt” it “as its own”. This is more than a general approval or endorsement, the State must deliberately and unambiguously express that the specifically identified conduct must be deemed its own. Normally, this would be done through a formal act, such as a “decree”Footnote ²³⁶ or the formal confirmation or ratification by a governing body.Footnote ²³⁷ It may also take the form of integrating an event (the taking over of a plant by the members of a trade union) into an expropriation process, if recorded clearly and unambiguously as such in internal memoranda and coupled with the establishment of a permanent presence in the plant.Footnote ²³⁸

• 126. As suggested by these examples, although the “acknowledgement” and “adoption” are not subject to specific requirements of form, as they can find expression through words (written or oral) or conduct, they must be “clear and unequivocal”.Footnote ²³⁹

• 127. Both the “acknowledgement” and the “adoption” must be performed by the State. Article 11 does not specify whether it is a State organ or another person or entity whose acts are attributable to the State. In Saint-Gobain v. Venezuela, the tribunal first concluded that the acts of a State entity, PDVSA, were attributable to Venezuela under Article 5 of the ILC Articles and then concluded that the taking over of a plant by a trade union had been acknowledged and adopted by PDVSA as its own.Footnote ²⁴⁰ Regarding the possibility that the acts of acknowledgement and adoption performed by a State organ or a State instrumentality may be ultra vires, their attribution will be determined under Article 7, as for other acts of organs and instrumentalities.

• 128. The acknowledgement and adoption will also determine the third condition, namely the “extent” of the conduct which is “fundamentally transformed”, in the words of the ICJ, and becomes an act of the State.Footnote ²⁴¹ The clarifications relating to specificity and form also apply in this context. One complex issue which may arise concerns the possibility that different organs and instrumentalities, through different acts, may acknowledge different extents of the private conduct in question. In such a case, a tribunal may consider that only the narrower set of generally acknowledged acts or those included in acts amounting to “adoption” are attributable. Otherwise, the claimant would be able to bring into a narrow acknowledgement and adoption a much wider set of general views, approvals and endorsements which fall short of the requirements of Article 11. This is consistent with the views expressed in the Commentary to the ILC Articles that:

  provided the State’s intention to accept responsibility for otherwise non-attributable conduct is clearly indicated, article 11 may cover cases where a State has accepted responsibility for conduct of which it did not approve, which it had sought to prevent and which it deeply regretted.Footnote ²⁴²

  [Page 66] Thus, if a State’s unambiguous adoption of the act as its own leads to attribution irrespective of its initial disapproval of the conduct, such unambiguous adoption must also be controlling when it only accepts some of the events it more widely approved. The rationale for this solution is simply that general approvals and endorsements are not enough; what matters for attribution is the specific and unambiguous acknowledgement and adoption of specific acts.

• 129. Moving to the scope of attributable conduct under this route, in addition to the question of the “extent” to which conduct has been acknowledged and adopted, there are four other aspects that call for comment.

• 130. The difference between sovereign and commercial acts has no direct bearing on the attribution of the act under Article 11. What matters is the acknowledgement and adoption. Even if the acts of the entity that accepts the conduct as its own are attributed under Article 5, the difference between sovereign and commercial acts will only be relevant to determine if the acts of “acknowledgement” and “adoption” were in the exercise of public prerogatives, but not with respect to acts initially performed by the private persons.

• 131. The conduct which is acknowledged and adopted is, as a rule, private conduct performed outside any appearance of official capacity. As a result, the latter requirement, which is important under Articles 4 and 5, is not relevant in the context of Article 11.

• 132. The question of ultra vires conduct presents complexities similar to those arising under Article 8. The principle is clearly that ultra vires conduct is not attributable under this route, simply because it is conceptually of limited relevance. Indeed, the State decides the extent to which it acknowledges and adopts the conduct and, as a result, any conduct beyond the scope of acceptance of the State is not, strictly speaking, ultra vires but simply prior conduct not acknowledged. However, there may be a situation in which a State acknowledges and adopts an ongoing situation, e.g. the occupation of an embassy or a plant, considering the private actors who are still performing the relevant conduct as agents. In such a case, if these actors act ultra vires (e.g. they torture or execute hostages) the conduct should be attributed to the State. The boundary between what can and cannot be attributed will be drawn by analogy to Article 8. Acts which are incidental to the situation acknowledged and adopted will be attributable whereas those that go clearly beyond (e.g. the occupation of different premises) would not.

• 133. The fourth and perhaps most complex question concerns the temporal dimension of attribution. Attribution is a constitutive element of an internationally wrongful act. When a State acknowledges and adopts a certain conduct, in principle this conduct becomes an act of the State only as of the date of such acceptance, not before. The timing of the acknowledgement is therefore important. This dimension has not been addressed in any detail in the investment case law, despite its relevance. The Commentary to the ILC Articles clarifies it by contrasting the Tehran Hostages case and the Lighthouses arbitration. In the first case, the ICJ made a distinction between the situation following the occupation of the US Embassy (first phase) and that following the decree adopting the acts as Iran’s (second phase). The private acts of the first phase had been generally supported [Page 67] and endorsed but such declarations lacked specificity and could not make these acts attributable to Iran. In this first phase, it was the inaction of the Iranian authorities, i.e. State organs, which was attributable to Iran and constituted a breach.Footnote ²⁴³ Only in the second phase was private conduct attributable as a result of its formal acceptance by Iran. The conclusion is that acknowledgement and adoption do not have a retrospective effect. Yet, as noted by the Commentary to the ILC Articles, “where the acknowledgement and adoption is unequivocal and unqualified there is good reason to give it retroactive effect”.Footnote ²⁴⁴ The commentary supports this observation by reference to the Lighthouses arbitration, where Greece was held to have assumed responsibility for the breach of a concession agreement by the then Ottoman territory of Crete, which subsequently became part of Greece.Footnote ²⁴⁵ The commentary also mentions the alignment on this point with Article 10 of the ILC Articles on insurrectional movements.

4.6. Other attribution routes

• 134. The attribution routes discussed so far are those featuring in the very large majority of investment cases. These are not the only routes accepted in general international law. Articles 6 (organs placed at the disposal of a State by another State), 9 (absence or default of the official authorities) and 10 (insurrectional and other movements) of the ILC Articles contemplate three other routes. Yet, they have very rarely featured in the investment case law, which contains at best brief references to them. Adding a general discussion of these routes in this study would be of limited use. Those interested in their operation can refer to the Commentary to the ILC Articles relating to these provisions as well as to a number of detailed studies clarifying their operation.Footnote ²⁴⁶ The following discussion is confined to an overview of the practice of investment tribunals, which has briefly addressed Articles 6 and 10.

• 135. In Electrabel v. Hungary, the tribunal considered the operation of Article 6 of the ILC Articles. This provision states that:

  The conduct of an organ placed at the disposal of a State by another State shall be considered an act of the former State under international law if the organ is acting in the exercise of elements of the governmental authority of the State at whose disposal it is placed.

  The commentary gives some illustrations, such as the situation of “a section of the health service or some other unit placed under the orders of another country to assist in overcoming an epidemic or natural disaster”.Footnote ²⁴⁷ The configuration [Page 68] addressed in Electrabel was, however, more complex. The issue was whether Hungary, in implementing a binding decision from the European Commission, could be held responsible for breach of the ECT. Responsibility supposes attribution. The tribunal framed the question under Article 6 noting, by analogy, that Hungary’s acts in implementing the decision should be attributed to the European Commission, with the EU assimilated to a State given that it was a party to the ECT. As a result, the application of the rule formulated in Article 6 led to the conclusion that:

  if and to the extent that the European Commission’s Final Decision required Hungary, under EU law, prematurely to terminate Dunamenti’s PPA, that act by the Commission cannot give rise to liability for Hungary under the ECT’s FET standard.Footnote ²⁴⁸

  The tribunal went on to ascertain what the decision actually required from Hungary. It is only to the extent of what was required that the finding of non-attribution, based on Article 6, could shield Hungary. And it was only if Hungary had added to those requirements or, specifically, interpreted them in a manner that was unreasonable or unsupported, that its own acts could amount to a breach. The tribunal considered that such was not the case.Footnote ²⁴⁹

• 136. The reasoning of the tribunal raises the key issue in this attribution route, namely the extent to which the foreign organs (by analogy Hungary’s organs as organs placed at the disposal of the EU) act as State organs or, rather, display some autonomy which would defeat attribution. The Commentary to the ILC Articles notes, indeed, that:

  what is crucial for the purposes of article 6 is the establishment of a functional link between the organ in question and the structure or authority of the receiving State. The notion of an organ “placed at the disposal” of another State excludes the case of State organs, sent to another State for the purposes of the former State or even for shared purposes, which retain their own autonomy and status.Footnote ²⁵⁰

  In addition to this link, two other conditions must be met for attribution. The organ “placed at the disposal” must “possess the status of an organ of the sending State; and secondly its conduct must involve the exercise of elements of the governmental authority of the receiving State”.Footnote ²⁵¹ Thus, the differences between organs and other entities, sovereign and commercial conduct, and private and official conduct are all relevant in this context. The Electrabel tribunal did not analyse these issues when making its analogy. Given that the act of the purported organ was the termination of a power purchase agreement, which is in principle a commercial act, the question was indeed relevant. If the tribunal did not address it, it is likely because the reference to Article 6 was mainly an argument by analogy. But this shortcoming must be kept in mind because Electrabel v. Hungary is possibly the [Page 69] only investment dispute addressing this provision, and precedents can easily be overstretched when transposed to other factual configurations.

• 137. In another investment decision, the operation of Article 10 of the ILC Articles was briefly mentioned. Article 10 states that:

  • 1. The conduct of an insurrectional movement which becomes the new Government of a State shall be considered an act of that State under international law.

  • 2. The conduct of a movement, insurrectional or other, which succeeds in establishing a new State in part of the territory of a pre-existing State or in a territory under its administration shall be considered an act of the new State under international law.

  • 3. This article is without prejudice to the attribution to a State of any conduct, however related to that of the movement concerned, which is to be considered an act of that State by virtue of articles 4 to 9.

In Cengiz v. Libya, the tribunal interpreted this rule in the context of the Libyan civil war, particularly as regards the taking over of the claimant’s main camp by the Libya Shield, a military movement subsequently integrated within the new government that evicted the Gaddafi regime. According to the tribunal:

Of note, unlike Article 11, this attribution route involves a retrospective effect as a matter of principle. The reason is the continuity “between the new organization of the State and that of the insurrectional movement”, which “leads naturally to the attribution to the State of conduct which the insurrectional movement may have committed during the struggle”.Footnote ²⁵³

5.1. Overview

• 138. This section brings together a number of observations made in the context of previous questions in order to provide clarity on five recurrent problems. The first is the distinction between sovereign and commercial acts and its relevance and implications for different attribution routes. The second problem concerns another distinction, that between acts performed in “an official capacity” and private acts, which is important for some attribution routes and the scope of attribution of ultra vires conduct. Attribution of this conduct is examined specifically as a third recurrent problem. Ultra vires acts are an extension of the scope of operation of Articles 4 and 5, in combination with Article 7 of the ILC Articles. The fourth [Page 70] recurrent problem focuses on the complex issue of attributing inaction or failure to act. Although there are several cases dealing with this issue, the matter is ordinarily discussed together with the assessment of breach of certain primary rules. The last recurrent problem brings together certain issues relating to contractual matters, already examined in connection with Question 1 (law governing attribution-related matters). The focus is on issues such as privity of contract, exercise of contractual rights, and the implications for the operation of umbrella clauses.

5.2. Sovereign vs commercial acts

• 139. In discussing the scope of attribution, i.e. the nature and extent of the conduct that can be attributed under the different routes examined in Section 4, I repeatedly referred to the difference between sovereign and commercial acts. This difference, which is sometimes formulated using a variety of terms (acta jure imperii and acta jure gestionis, or governmental authority and commercial conduct, sovereign or public prerogatives and commercial acts, etc.), has not the same relevance in every attribution route. In this section, my goal is to clarify both the meaning of these expressions and the different legal contexts in which they are used. The latter must be addressed first because guidance on meaning is conditioned upon the relevance of different legal contexts for attribution.

• 140. The distinction between sovereign and commercial acts has been studied at length in the context of immunities, both of States and State officials, including high-rank officials, heads of mission, diplomats and consular staff.Footnote ²⁵⁴ In these contexts, the distinction is relevant mainly to determine the acts for which international law grants immunity of jurisdiction before foreign courts. The focus is clearly on the nature of the acts. These contexts are relevant but distinct from that of attribution rules. Attribution rules cannot serve to determine which State official benefits from immunity. Some officials do and some others do not, despite the fact that they would all be State organs from an attribution perspective. The rules granting immunity, and the reasons underlying them, are simply distinct from attribution rules. The tribunal in F-W Oil Interests v. Trinidad and Tobago emphasised this point by reference to the commentary to Article 4 of the ILC Articles, which clarifies that the sovereign or commercial nature of an act has no bearing on the attribution of an act of a State organ.Footnote ²⁵⁵ Thus, whereas developments in the law of immunities may be useful, by analogy, to characterise the respective provinces of sovereign and commercial acts in the context of attribution, the two contexts remain clearly distinct.

• 141. Similarly, some primary norms refer to concepts which bear some resemblance to those of sovereign or commercial acts. For example, in Bayindir v. Pakistan, the tribunal dismissed the investor’s expropriation claim, noting that:

  [Page 71] even if the expulsion violated the Contract and deprived Bayindir of the economic substance of its contract rights, a finding of expropriation would only be founded if the acts at issue were sovereign acts. The evidence does not point in this direction. To the contrary, it shows that Pakistan can reasonably justify the expulsion by Bayindir’s poor performance … with the consequence that the expulsion must be seen in the framework of the contractual relationship, not as an exercise of sovereign power.Footnote ²⁵⁶

  The tribunal had previously found that the act was attributable under the rule codified in Article 8 of the ILC Articles because the exercise of the termination right by the instrumentality had been given “clearance” by the State. Article 8 (like Article 4) does not require as a condition for attribution that the act be an exercise of governmental authority. In this case, the assessment of attribution and the assessment of breach are clearly distinguished.

• 142. Had the act been attributed under Article 5, which requires a specific exercise of governmental authority for attribution, the question would have been more complex. Indeed, at that point, the issue that would arise is whether the assessment at the level of attribution predetermines the assessment, at the level of breach, that the act is sovereign. Conceptually, both are clearly distinct. In one case, the tribunal interprets and applies the conditions set in a secondary rule (codified in Article 5 of the ILC Articles). In the other, the tribunal interprets and applies the conditions of a primary rule (the relevant expropriation clause in the treaty). The tribunal in Almås v. Poland indirectly addressed this issue in its examination of attribution under Article 5. In this case, the claimant argued that the exercise of a contractual right by a separate State instrumentality to terminate a lease could be attributed under Article 5. The tribunal took as a starting point the “prima facie compelling argument” that any entity would be able to terminate a lease, hence the act was not an exercise of governmental authority. The claimant sought to counteract this starting point by arguing inter alia that the true reason for termination rested on a policy motivation. It referred in this regard to the decision in Vigotop v. Hungary,Footnote ²⁵⁷ which concerned not attribution but the assessment of expropriation. The Almås tribunal examined the test applied in Vigotop, mindful that it concerned a different question, that of expropriation. It premised its examination on the observation that:

  A short answer to the argument based on Vigotop Limited v. Hungary would be that it concerned termination of a contract with the State itself whereas in the present case the Lease Agreement was with a separate entity with its own contractual capacity and whose conduct in terminating the Agreement was not attributable to the Respondent under the law of State responsibility.Footnote ²⁵⁸

  Thus, the two planes (assessment of attribution under the secondary rule and assessment of breach under the primary rule) were clearly distinct. It nevertheless examined the Vigotop test at length and concluded that it was not met. What this [Page 72] decision suggests is that, although the two planes are distinct, their boundaries are not conceptually watertight. This is for two reasons. First, the two assessments rely on broad conceptual categories, the contours of which cannot be defined once and for all. Secondly, the facts that are relevant to determine whether an entity exercised elements of governmental authority for the purpose of attribution are also relevant to assess whether some primary norms may have been breached.

• 143. A third context where the distinction between sovereign and commercial acts has arisen is in the nature of the entity bringing a claim under the ICSID Convention. As noted earlier in this study, attribution rules are not as such applicable to determine this question, because the matter does not concern State responsibility, but the scope of entities entitled to rely on the ICSID dispute settlement mechanisms . In Beijing Urban v. Yemen, the tribunal found guidance on this issue in the so-called “Broches factors” or “Broches test”, which precludes claims from entities “acting as an agent for the government” or “discharging an essentially governmental function”.Footnote ²⁵⁹ Relying on CSOB v. Czech Republic, it noted that the distinction between a governmental function and commercial action in this context concerns the nature of the act, not its purpose.Footnote ²⁶⁰ Although the Broches factors were deemed to be “the mirror image” of the attribution routes under Articles 5 and 8 of the ILC Articles,Footnote ²⁶¹ the latter were not applied as such. This is another example of conceptual and factual proximity of the considerations which are determinative for two legally distinct assessments.

• 144. Even in the specific context of attribution, the distinction between sovereign and commercial acts is only relevant for one of the four most frequent attribution routes (Articles 4, 5, 8 and 11) analysed in Section 4. Only Article 5 makes attribution dependent on both a delegation and an actual exercise of governmental authority. For the other three routes, the distinction has no bearing. The less travelled route of Article 6 also requires that the foreign organ placed at the disposal of a State exercises a parcel of this State’s governmental authority for the act to be attributable.

• 145. The foregoing paragraphs show that the characterisation of sovereign and commercial acts in the context of attribution is distinct from those in the contexts of immunities, the assessment of breach of a primary rule, and the entitlement to bring a claim under the ICSID Convention. Moving now to the meaning of the terms used in the context of attribution specifically, the first point to be mentioned is that the ILC Articles deliberately refrained from giving a definition of “governmental authority”.Footnote ²⁶² This was emphasised by the tribunal in F-W Oil Interests v. Trinidad and Tobago:

  The Tribunal notes that the draft Articles contain no definition of the broad notion of “elements of the governmental authority” (any more than does the BIT for the equivalent phrase “other governmental authority delegated to it”). Indeed the ILC [Page 73] consciously refrained from including in the draft even elements towards defining its application in particular cases. Rather, the Commission took the view, as expressed in paragraph (6) of the Commentary to draft Article 5, that the notion had to be judged in the round, in the light of the area of activity in question, and in the light of the history and traditions of the country in question. In short, the notion is intended to be a flexible one, not amenable to general definition in advance; and the elements that would go in its definition in particular cases would be a mixture of fact, law and practice. Moreover – and the point is of some importance – it is not the case that the same answer would necessarily emerge on every occasion; in some of its activities a State enterprise might fall on one side of the line, in others on the other.Footnote ²⁶³

  However, this openness is certainly not limitless.Footnote ²⁶⁴ Some relevant parameters to circumscribe the distinction include (i) the fact that there are only two terms (i.e. there is no third category other than sovereign or commercial acts), (ii) there are some criteria which guide the determination of the category of governmental authority (content of the prerogative, mode of conferral, purpose of its exercise, degree of public oversight) (see Section 4.2), (iii) the characterisation is conducted in concreto in the light of all relevant circumstances (i.e. the focus is normally on one or more specific acts) and (iv) there are usually some narrower categories that provide useful starting points or prima facie characterisations (i.e. rather than broad categories such as “sovereign acts” or “governmental authority” or “commercial acts”, narrower categories such as “exercise of contractual rights” are used as starting points).

• 146. These parameters point to a case-by-case characterisation of acts.Footnote ²⁶⁵ Some examples derived from the decisions reported in this volume include the characterisation as commercial acts of matters such as price negotiations, non-delivery of merchandise, corporate shareholder battles,Footnote ²⁶⁶ auctions of commercial spaces in an airport,Footnote ²⁶⁷ pre-contractual negotiations, requests for time extension or the termination of a contract.Footnote ²⁶⁸ Other matters have instead been considered governmental or sovereign acts, including the modernisation of an airport,Footnote ²⁶⁹ the ranking and evaluation of applications to receive support under a feed-in-tariff scheme,Footnote ²⁷⁰ or the contracting of public works.Footnote ²⁷¹

5.3. Official capacity vs private acts

• 147. A separate distinction which must not be conflated with the one examined in the previous section is that between acts performed in an official capacity and private acts.Footnote ²⁷² Conduct in an official capacity may concern both the exercise of [Page 74] governmental authority and purely commercial acts, and it may even be ultra vires. Yet, as long as the relevant person “acts in an apparently official capacity, or under colour of authority, the actions in question will be attributable to the State” under Article 4.Footnote ²⁷³ The same reasoning applies to attribution under Article 5 which, again, requires the instrumentality to be vested with and be exercising governmental authority, which in virtually all cases will involve action “under the cover of its official status”.Footnote ²⁷⁴ By contrast, the acts of persons or groups of persons acting under the effective control of the State (Article 8) or which are acknowledged and adopted by it (Article 11) can be attributed irrespective of any appearance of official status in the persons performing the acts.

• 148. Thus, “official capacity” and “governmental authority” are clearly distinct concepts. Official capacity is a matter of appearance. The acts performed by the relevant person or entity must be “carried out by persons cloaked with governmental authority”.Footnote ²⁷⁵ Conversely, when “the act had no connexion with the official function [of an entity] and was, in fact, merely the act of a private [entity]”, it is not attributable under Articles 4 or 5.

• 149. I will discuss in the next section the attribution of acts ultra vires. For present purposes, it suffices to say that acts ultra vires and “private acts” are also distinct concepts. Only ultra vires acts of organs and instrumentalities performed under an official capacity are attributable under the rules formulated in Articles 4, 5 and 7 of the ILC Articles, whereas private acts (without any official appearance) are not attributable under these routes, whether they are consistent or not with the mandate of a person or entity.

• 150. The distinction between official capacity and private acts has been raised occasionally in the investment case law. In Gavrilović v. Croatia, the tribunal referred to it when assessing attribution under Article 4:

  The conduct of an organ of the State in an apparently official capacity may be attributable to the State, even if the organ exceeded its competence under internal law or in breach of the rules governing its operations. The corollary of this is that acts that an organ commits in its purely private capacity are not attributable to the State, even if it has used the means placed at its disposal by the State for the exercise of its function … In this case, there is no compelling evidence to support the proposition that the Bankruptcy Council or any member thereof was acting in a purely private capacity, nor for personal gain. This conclusion is unaffected by the Respondent’s contention that the conduct of the Bankruptcy Judge is not attributable to the Respondent because of his corrupt behaviour, in particular his later work as the lawyers of Mr Gavrilović and his family. This contention, without more, does not suffice to reach a finding that the actions of the Bankruptcy Judge were corrupt.Footnote ²⁷⁶

• [Page 75] 151. This observation raises the issue of whether an allegation or proof of corruption would make the conduct ultra vires, hence attributable, or instead turn it into private conduct, hence non-attributable. The tribunal in Gavrilović left the issue open. However, the Commentary to the ILC Articles makes clear that:

  It is irrelevant for this purpose that the person concerned may have had ulterior or improper motives or may be abusing public power. Where such a person acts in an apparently official capacity, or under colour of authority, the actions in question will be attributable to the State.Footnote ²⁷⁷

And it then adds that “[t]he case of purely private conduct should not be confused with that of an organ functioning as such but acting ultra vires or in breach of the rules governing its operation”.Footnote ²⁷⁸ Situations of corruption are peculiar because the relevant official uses her position – that is after all the very point of the corrupt act – in an improper manner for personal gain. Yet, the appearance is clearly that of an act in an official capacity. Such conduct can therefore be attributed, even if ultra vires, under either Article 4 or Article 5, in combination with the rule formulated in Article 7 of the ILC Articles. The conclusion that an act performed by a State organ or instrumentality as a result of corruption may be attributable does not mean that a contract concluded on that basis is valid. Attribution rules cannot create primary rules of obligation under private law; the extent of a contract’s validity will rather be governed by the applicable domestic law. It is, however, the corrupt act which is attributable and may engage the responsibility of the State’s organ (e.g. if the losing tenderer brings a claim).Footnote ²⁷⁹ When the claimant was involved in the corrupt act, the claim is in all events inadmissible, and it should be dismissed before reaching the assessment of the merits.Footnote ²⁸⁰

5.4. Ultra vires action (Article 7)

• 152. Article 7 of the ILC Articles codifiesFootnote ²⁸¹ the rule according to which acts ultra vires of State organs, including foreign organs at the disposal of a State, and of State instrumentalities are attributable to the State:Footnote ²⁸²

  The conduct of an organ of a State or of a person or entity empowered to exercise elements of the governmental authority shall be considered an act of the State under international law if the organ, person or entity acts in that capacity, even if it exceeds its authority or contravenes instructions.

• [Page 76] 153. For ultra vires action to be attributable, “the organ, person or entity” must act “in that capacity”. The reference to “governmental authority” must not lead to the wrong conclusion that the “capacity” in question is the exercise of elements of governmental authority. Such a conclusion would exclude commercial acts performed by organs, which is clearly not what the rule does. Instead, the commentary makes absolutely clear that “in that capacity” concerns the concept of “official capacity”.Footnote ²⁸³ The commentary notes indeed that:

  Cases where officials acted in their capacity as such, albeit unlawfully or contrary to instructions, must be distinguished from cases where the conduct is so removed from the scope of their official functions that it should be assimilated to that of private individuals, not attributable to the State.Footnote ²⁸⁴

  I referred in the previous section to this issue. The commentary to Article 7 further notes that the formulation of the rule in this provision is intended to indicate that “only the actions and omissions of organs purportedly or apparently carrying out their official functions, and not the private actions or omissions of individuals who happen to be organs or agents of the State”, will be attributable.Footnote ²⁸⁵

• 154. Ultra vires conduct is, in principle, not attributable under routes other than those formulated in Articles 4, 5 and 6 of the ILC Articles. Particularly, it is in principle not attributable under the rules codified in Articles 8 and 11. As noted earlier in this study, this is because the very reason for attribution is that the conduct is effectively controlled or specifically acknowledged and adopted. The only exception to this exclusion concerns acts which are incidental to the acts effectively controlled or acknowledged and adopted.

• 155. The attribution of ultra vires acts is a recurrent issue in the investment case law. Several decisions reported in this volume tackle the question from different angles. In Tethyan Copper v. Pakistan, the tribunal applied the basic principle that the acts of provincial authorities are attributable to the State under Article 4, irrespective of their sovereign or commercial nature, and even if they are ultra vires.Footnote ²⁸⁶ The same principle was stated in Gavrilović v. Croatia, with nuances, noted in the previous section, relating to an allegation of corruption.Footnote ²⁸⁷ In Kardassopoulos v. Georgia, the tribunal recalled the principle that acts ultra vires remain attributable in the context of Article 5.Footnote ²⁸⁸ In Ortiz v. Algeria, the question was approached from the perspective of Article 8 of the ILC Articles. The tribunal confirmed the principle that ultra vires acts cannot be attributed through this route except for situations of incidental ultra vires acts in an operation which remains under the effective control of the respondent.Footnote ²⁸⁹

5.5. Attribution of failure to act

• 156. The legal and evidentiary requirements necessary to establish attribution, particularly of acts of private persons or separate entities (e.g. under Article 8 of the ILC Articles), place a heavy burden on claimants. In attempting to get around this obstacle, in some cases claimants argue that it is not the act itself but the lack of effective intervention from State organs, i.e. the State’s failure to act, which is attributable to it under Articles 4 or 5 of the ILC Articles.Footnote ²⁹⁰

• 157. By way of illustration, in Bayindir v. Pakistan, the claimant relied on Wena Hotels v. Egypt Footnote ²⁹¹ to argue that despite being aware of the acts allegedly in breach of the applicable treaty, the respondent had stood by and failed to protect the investor from the action of entities under its control.Footnote ²⁹² The tribunal rejected attribution under Article 4, so this argument was ineffective. However, in other cases, such as Amco v. Indonesia,Footnote ²⁹³ AMT v. Zaire,Footnote ²⁹⁴ Wena Hotels v. Egypt and, most importantly, in the Tehran Hostages case, a similar argument has been successful. In reviewing these and other cases, including Ampal v. Egypt and Strabag v. Libya, both reported in this volume, it is important to note that the nature of the argument brings much closer the assessment of attribution to that of breach. The reason for this proximity is that attribution of failure to act under secondary rules such as Articles 4 or 5, although conceptually distinct, supposes a duty to act, i.e. a primary rule. It is true that other elements are also required, most notably knowledge of the need to act, but these aspects are normally addressed during the assessment of breach.

• 158. The requirements for a failure to act to be attributed to the State and indeed amount to a breach have been authoritatively laid out by the ICJ in the Tehran Hostages case. The Court had concluded with respect to the first phase of events (the taking of the US Embassy by groups of private persons) that the Iranian authorities had not been specific enough in their support and endorsement for the acts to be attributable to Iran. However, what was attributed and amounted to a breach was the failure of the Iranian authorities to act. At paragraph 68 of its judgment, the Court concluded as follows:

  The Court is therefore led inevitably to conclude, in regard to the first phase of the events which has so far been considered, that on 4 November 1979 the Iranian authorities:

  • (a) were fully aware of their obligations under the conventions in force to take appropriate steps to protect the premises of the United States Embassy and its [Page 78] diplomatic and consular staff from any attack and from any infringement of their inviolability, and to ensure the security of such other persons as might be present on the said premises;

  • (b) were fully aware, as a result of the appeals for help made by the United States Embassy, of the urgent need for action on their part;

  • (c) had the means at their disposa1 to perform their obligations;

  • (d) completely failed to comply with these obligations.

  Restating these requirements in more general terms, what is required is (i) a duty to act; (ii) awareness of the need to act; (iii) the means to act in the specific situation under consideration; (iv) failure to act as required by the applicable duty.

• 159. Regarding the first condition, despite the formulation used by the Court, it is important to emphasise that international obligations do not depend, for their application and operation, on the “awareness” the relevant body may or may not have of them. Members of police forces are not released from the obligation to protect diplomatic staff merely because they are not familiar with international law. Members of military forces are not released from the obligation to respect international humanitarian law merely because they may be unfamiliar with it. The formulation used by the Court was intended to stress the fact that, in that case, awareness of what international law required in the circumstances had been expressly manifested by Iranian authorities. In the investment context, primary rules entailing a duty to act include the full protection and security clause and the fair and equitable treatment clause. It is not my purpose to examine here in any length these two primary rules.Footnote ²⁹⁵ I shall limit myself to the observation that a duty to act is certainly not equivalent to strict liability for damage which occurs in the territory or under the jurisdiction of a State. As noted by a chamber of the ICJ in the ELSI case, the reference in the applicable treaty to “constant protection and security” could not “be construed as the giving of a warranty that property shall never in any circumstances be occupied or disturbed”.Footnote ²⁹⁶ Similarly, in AAPL v. Sri Lanka, the tribunal firmly rejected the claimant’s argument that a full protection and security clause entailed strict liability for damage incurred.Footnote ²⁹⁷

• 160. The second condition concerns knowledge of the need to act. In Wena Hotels, there was ample evidence that the State had knowledge of what the State-owned and government-controlled entity performing the act (the Egyptian Hotels Company) was about to do and did.Footnote ²⁹⁸ This is consistent with the requirement as [Page 79] set out in the Tehran Hostages case, where the Court concluded that Iranian authorities were very much aware of the situation. Whether constructive (rather than actual) awareness would be sufficient and, if so, whether it would be enough only for attribution or also for breach is less clear. In a different context, the European Court of Human Rights (ECtHR) has considered that actual or constructive knowledge could provide sufficient bases. Thus, in Osman v. United Kingdom, the ECtHR observed that:

  where there is an allegation that the authorities have violated their positive obligation to protect the right to life in the context of their above-mentioned duty to prevent and suppress offences against the person … it must be established to its satisfaction that the authorities knew or ought to have known at the time of the existence of a real and immediate risk to the life of an identified individual or individuals from the criminal acts of a third party and that they failed to take measures within the scope of their powers which, judged reasonably, might have been expected to avoid that risk.Footnote ²⁹⁹

  The test laid out by the ECtHR is generally compatible with that of the ICJ, although it is coloured by the specific primary rule at stake, the right to life. In the specific circumstances of the case, the argument of actual or constructive knowledge was rejected.Footnote ³⁰⁰

• 161. The third condition concerns the means to act in the specific situation. This raises the difficult and fact-intensive question of the degree of diligence that may be expected under the circumstances. One aspect of this question is whether such degree is affected by the context. As noted by the sole arbitrator in Pantechniki v. Albania, “it seems difficult to maintain that a government incurs international responsibility for failure to plan for unprecedented trouble of unprecedented magnitude in unprecedented places”.Footnote ³⁰¹ In the same vein, in Strabag v. Libya, the tribunal reasoned that:

  In light of both Parties’ extensive evidence showing circumstances of widespread conflict, violence and disorder in Libya at relevant times, the Tribunal is compelled to agree with the thrust of Respondent’s assessment: In the circumstances prevailing in Libya during and since the Revolution, it was not reasonably possible for the Libyan authorities to take consistent and effective measures to protect Claimant’s investment.Footnote ³⁰²

  But a challenging context is not as such sufficient justification for lack of action.Footnote ³⁰³ The fact that neither a claimant nor any other investor in similar circumstances received the support required by the duty to act is no excuse.Footnote ³⁰⁴ The assessment remains a case-specific one. Thus, whereas the tribunal in Ampal v. Egypt premised its assessment of the claims regarding the respondent’s failure to protect a pipeline from attacks on the fact that “the circumstances in the North [Page 80] Sinai Egypt were difficult in the wake of the Arab Spring Revolution”,Footnote ³⁰⁵ it concluded nevertheless that the full protection and security clause had been breached. The tribunal identified a pattern characterising the lack of sufficient action from the respondent:

  It is apparent to the Tribunal that, when considering the totality of these attacks, a certain pattern emerges: an attack is perpetrated, to which GASCO [a subsidiary of the national gas company] reacts months later and then adopts some measures to heighten the security of the pipeline, those measures are seldom implemented (or there is no evidence on the record that they were), another attack happens, and so on.Footnote ³⁰⁶

  The reasoning is already placed at the heart of the assessment of breach, not merely attribution. As noted earlier, the two aspects are closely intertwined in an allegation of attribution for failure to act. The failure was, above all, that of Egyptian security forces,Footnote ³⁰⁷ which are organs of the State.

• 162. The fourth condition, the actual failure to act as required by the applicable duty, concerns entirely the assessment of merits. I shall therefore confine my discussion to one observation. As noted in Wena Hotels v. Egypt, action is required both before (to prevent the encroachment) and after the acts in question (to restore the situation or compensate for the damage).Footnote ³⁰⁸

5.6. Attribution, contracts and umbrella clauses

• 163. One of the most complex problems arising in connection with the rules of attribution in general international law concerns their relevance in the context of contractual matters.Footnote ³⁰⁹ Tribunals have taken very different positions, thereby creating significant uncertainty not only on their operation but even their very application. In order to clarify some of these issues, the premise must be that the rules of attribution are only applicable to determine State responsibility for internationally wrongful acts. They are not applicable to determine, under domestic law, if a State has violated the contract or the proper law of the contract. Nor are they applicable to create privity of contract in the first place, a matter also governed by domestic law. Yet, as discussed in this section, tribunals have referred to the rules of attribution to extend a contract with a separate legal entity to the national government, or even to elevate the terms of such contract to the level of an obligation or an undertaking entered into by the State and protected by so-called umbrella clauses. These conclusions likely reflect the specific facts of the case, the way in which counsel pleads their case and, no doubt, also the vicissitudes of deliberations among tribunal members. But clarity is necessary to provide some [Page 81] consistency and bring to bear the many decisions that accurately state and apply the rules of attribution in international law.

• 164. In this regard, it is useful to start with a basic hypothetical. Let us assume that States A and B have concluded a bilateral investment treaty (BIT). State A is divided into three territorial subdivisions (provinces): A1, A2 and A3. Its administrative apparatus therefore consists, in basic terms, of the national government (which we can call A) and three provincial governments, each with its own separate legal personality in the public law of State A. Investor B1 from State B concludes a contract with the government of province A1. Clearly, there is privity of contract between B1 and A1, but it is equally clear that there is no privity of contract between B1 and either A2, A3 or A. For privity of contract to arise between B1 and A, the party to the contract would have to be A. This would be possible if an agent (under domestic law) of the national government, e.g. a Minister, concluded the contract. It would also be possible if A1 acted “on behalf” (as an agent under domestic law) of State A when it concluded the contract. But both the Minister’s and A1’s power to act as an agent of State A rests on domestic law, not on the Minister’s or A1’s character of State organ under international law. For example, a Minister of Environment will normally not be entitled under domestic law to conclude a contract relating to armed forces, despite the fact that it is a State organ. Leaving aside the hypothesis of A1 as an agent, let us now assume that, for purely political reasons, A forces A1 to exercise abusively its contractual right to terminate the contract with B1. Let us further assume that the BIT contains an umbrella clause stating that each State party, i.e. A and B, undertakes to respect all obligations entered into with investors from the other party.

• 165. In this relatively simple situation, three different attribution-related questions arise. The first is the question of the privity of contract. As anticipated in the previous paragraph, this question is governed by domestic law. The fact that A1 is a de jure organ of State A has no bearing. The rules of attribution do not create new obligations (primary rules) or privity of contract. The second question concerns the abusive exercise of a contractual right. As State A is not a party to the contract, it cannot incur contractual liability under domestic law, although it may have violated other rules of administrative due diligence or fairness under domestic law. Only A1 would incur contractual liability for violation of the contract. But because A1 is a de jure organ of State A, its abusive termination of the contract is an act attributable to State A under the rule codified in Article 4 of the ILC Articles which may engage its State responsibility for an internationally wrongful act. Privity of contract is not relevant here if the international obligation (primary rule) allegedly breached does not require it, which is the case of most investment protection standards in BITs. But, as I have noted, the BIT contains at least one international obligation of State A for which privity of contract between B1 and A is directly relevant, i.e. the umbrella clause. At this stage, the third question arises with respect to the operation of the umbrella clause. Are the rules of attribution in general international law applicable to determine whether the contract (as an obligation possibly protected under the umbrella clause) has been “entered into” by State A with B1? The answer is, again, that the rules of attribution do not create obligations. If there is no privity of contract between [Page 82] A and B1, then the fact that A1 is a de jure organ of A is immaterial. It is only if the tribunal interprets the terms “obligations entered into” (very) broadly as not requiring privity of contract that State A may be deemed to have “entered into” such obligations with respect to B1. Such obligations would not be, at least formally, the contract itself but, if at all conceivably, some broad form of obligation arising from expectations relating to non-interference with the contract. Even then, the rules of attribution will operate only to attribute the abusive exercise of a termination right by a State organ (A1) to State A. Thus, what is attributed is the act, not the contract, and attribution is for the sole purpose of the assessment of potential breach of an international obligation (the umbrella clause).

• 166. A fourth question arises when A1 is not a State organ but a State instrumentality. The difference in such a case is that the exercise of the contractual right by A1 would either have to be a specific display of public prerogatives (as a purely commercial act is not attributable under Article 5 of the ILC Articles) or, if it is a commercial act, it would have to be under the instructions, direction or control of State A (hence attributable under Article 8 of the ILC Articles). Otherwise, it would not be attributable.

• 167. The four questions identified in the preceding paragraphs (question 1: privity of contract; question 2: exercise of contractual right; question 3: operation of an umbrella clause; question 4: attributability of an exercise of contractual rights depending on the route) can guide the overall assessment of the otherwise confusing body of investment decisions. Their analysis can be combined because questions 1 and 3, as well as 2 and 4, often arise together, although they remain analytically distinct.

• 168. On questions 1 and 3, the principle that rules of attribution do not create privity of contract has been repeatedly stated, including to conclude that the umbrella clause is inapplicable to the obligations arising from the contract. In EDF v. Romania, the tribunal stated the principle and its consequences in limpid terms which deserve to be quoted in extenso:

  The “obligations entered into”, to which Article 2(2) of the BIT [an umbrella clause] refers, are obligations assumed by the Romanian State. The breach of contractual obligations by a party entails such party’s responsibility at the contractual level. There is in principle no responsibility by the State for such breach in the instant case since the State, not being a party to the contract, has not directly assumed the contractual obligations the breach of which is invoked … It is unclear whether Claimant relies on the attribution to the State of certain acts and conduct of [State-owned entities] AIBO and TAROM on the assumption of their being in breach of the ASRO Contract or the SKY Contract in order to impute to the State the responsibility for such breach. If so, this construction of the umbrella clause would be incorrect since the attribution to Respondent of AIBO’s and TAROM’s acts and conduct does not render the State directly bound by the ASRO Contract or the SKY Contract for purposes of the umbrella clause … Attribution does not change the extent and content of the obligations arising under the ASRO Contract and the SKY Contract, that remain contractual, nor does it make Romania party to such contracts.Footnote ³¹⁰

  [Page 83] In these paragraphs, the tribunal clearly recalls that without privity of contract with the State, the umbrella clause cannot apply to contractual obligations, and that attribution rules do not create privity of contract.Footnote ³¹¹

• 169. Similarly, in Gavrilović v. Croatia, the tribunal referred to EDF v. Romania in support of the principle that “attribution does not change the extent and content of the obligations arising under the contract, which remain contractual, nor does it make the State party to such contracts”.Footnote ³¹² As a result, it rejected the application of the umbrella clause to the obligations arising from a contract to which the State was not a party. When doing so, it reviewed some decisions that would appear to suggest the contrary principle, namely that attribution rules can make a State, which is not party to a contract, bound by the obligations arising therein and therefore subject to respect them under the umbrella clause. The tribunal noted, however, that in all those cases, the relation between the entity party to the contract and the State was such that the former represented (were agents of) the latter when entering the contract.Footnote ³¹³

• 170. There are other cases that, as those invoked by the claimant in Gavrilović v. Croatia, could give the impression that attribution rules create privity of contract with the State, hence obligations protected by an umbrella clause. Yet, on closer inspection, we see that in all these cases, it is the “agency” of the relevant entity (its acting as the agent of the State) which makes the State bound by the contract. The confusion in such cases comes from the fact that tribunals refer in the same sections to the rules of attribution, whereas, strictly, what creates the privity of contract is the principal–agent relationship between the State and the entity acting on its behalf, whose signature binds the State with respect to the investor.

• 171. A good illustration is offered by the comparison of two cases against Egypt, Ampal and Unión Fenosa, where the factual configuration relevant for the attribution was broadly similar. In these two cases, the claimants relied on attribution rules to extend contracts signed by the investor with Egyptian corporate entities (EGPC and EGAS) to Egypt. In Ampal, the tribunal concluded that the contracts (not just the acts leading to their termination) bound the State. It referred to Article 8 of the ILC Articles (mentioning that both the conclusion and termination of the contract were “with the blessing of the highest levels of the Egyptian Government”).Footnote ³¹⁴ Strictly, this reasoning can only be legally correct for the act of termination, which may indeed be attributed to the effective control of the State. But attribution rules cannot change the parties to a contract. Looking closer at the factual reasoning of the tribunal, the reason why, in the circumstances of the case, it extended the contract to Egypt arises from the text of the contract itself (which stated that EGPC “represented” the Ministry of Petroleum in the negotiation and conclusion of the agreement).Footnote ³¹⁵ By contrast, in Unión Fenosa, the tribunal [Page 84] rejected the extension of a contract between the investor and EGPC to Egypt because EGPC was not, in that contractual negotiation, acting on behalf (in representation or as an agent) of the State.Footnote ³¹⁶ In other words, these two cases illustrate two distinct situations where, first, the entity in my initial hypothetical (A1) acts on behalf of State A, thus creating privity of contract with B1 (as in Ampal),Footnote ³¹⁷ and secondly, it acts on its own behalf, thus excluding any privity of contract between A and B1 (as in Unión Fenosa). The confusion comes from the reference to the rules of attribution, which have no bearing on whether EGPC is empowered to represent the State in the negotiation and conclusion of a contract. Such matters are governed by domestic law.

• 172. What remains to be clarified is one aspect of question 3, namely whether the umbrella clause may operate irrespective of privity of contract between the investor and the State and whether rules of attribution have any bearing on this matter. The correct answer is the one provided in EDF v. Romania and subsequently confirmed by other tribunals.Footnote ³¹⁸ The rules of attribution do not create privity of contract and, as a result, the application of the umbrella clause is excluded. However, in some cases, claimants have sought to broaden the application of an umbrella clause by arguing that terms such as “party” or “entered into” must be interpreted in the light of attribution rules. As a general principle, this argument rests on very thin and questionable grounds. Yet, in some cases, tribunals have accepted an extension on the specific facts of the case. Thus, in Strabag v. Libya, the tribunal concluded that:

  Reviewing the overall circumstances cumulatively, including the public importance of the functions carried out by [infrastructure authorities] RBA, TPB and [HIB] and their vesting with governmental authorities, their lack of administrative and financial economy, the nature of the contracts and their being deeply bound with state interest, and the existence of overwhelming evidence that demonstrates that an array of public authorities had a major hand in the conclusion and performance of the contracts, the Tribunal is of the view that, in this case, there is an exceptional combination of circumstances compelling the conclusion that the Respondent did, indeed, “enter into” the obligations in the disputed contracts within the meaning of Article 8(1) of the Treaty [umbrella clause].Footnote ³¹⁹

  But the tribunal reached this conclusion on the specific facts of the case, particularly the domestic regime, with only a cursory reference to attribution rules. More debatable is the statement of the tribunal in Bosh v. Ukraine that “where the conduct of entities can be attributed to the Parties (under, for instance, Articles 4, 5 or 8 of the ILC Articles), such entities are considered to be ‘the Party’ for the purposes of [the umbrella clause]”.Footnote ³²⁰ This statement is, with respect, plainly wrong. The only way in which it could be supported would be that a [Page 85] peculiar wording of the umbrella clause, its context and possibly several other indications would call for a highly unusual interpretation of the term “Party”. In casu, the tribunal correctly concluded that the obligations entered into by a university were not those of Ukraine, hence the umbrella clause was not applicable. But taken out of context, as attempted by the claimants in Gavrilović v. Croatia, it could mislead a tribunal and generate significant confusion.

• 173. Whether or not there is privity of contract between the investor and the State, it is entirely possible for a State to interfere with a contractual framework. This is the focus of questions 2 and 4, which are premised on the necessary distinction between, on the one hand, the contractual relationship (including its parties, terms and governing law) and, on the other hand, the acts which interfere with this contractual relationship potentially breaching a rule of international law (distinct from the contract). Attribution of such acts to the State is plainly governed by the rules of attribution. Acts of interference can include the exercise of contractual rights but also acts external to the contractual framework which interfere with it. In both cases, the rules of attribution will determine whether and to what extent the act is attributable to the State. At this stage, it does make a significant difference whether the entity acting is a State organ (Article 4), or a State instrumentality (Article 5), or a “person or group of persons” acting under the effective control of the State (Article 8). The most basic difference is that commercial acts, such as the normal exercise of a contractual right, will only be attributable if performed by a State organ or under the effective control of the State.

• 174. This difference can be illustrated by reference to Bayindir v. Pakistan, reported in this volume. As mentioned above, the investor had concluded a contract with a State instrumentality, the National Highway Authority (NHA), which subsequently exercised its right to terminate the contract. The investor claimed that these acts were attributable to the State and in breach of the applicable investment treaty. The tribunal examined the attribution of the termination – from which other acts followed – in the light of the attribution rules codified in Articles 4, 5 and 8 of the ILC Articles. This is of course correct because it is the attribution of the act allegedly interfering with the contract, not of the contract itself. As noted by the tribunal:

  From a contractual standpoint, these actions were those of NHA and not of the Government of Pakistan. The Tribunal must therefore determine whether they are attributable to the Respondent under the international law rules of attribution reflected in Articles 4, 5 and 8 of the International Law Commission’s Articles on Responsibility of States for internationally wrongful acts.Footnote ³²¹

  The claimant had indeed acknowledged, during the proceedings, that there was no privity of contract between the investor and Pakistan itself.Footnote ³²² The tribunal went on to exclude the possibility that the NHA might be considered a State organ, given the separate legal personality of this agency. It was instead a State [Page 86] instrumentality, whose acts might be attributed under the two-element test applicable under Article 5. Yet, although the NHA was empowered to exercise elements of governmental authority, the decision terminating the contract was a commercial act. Therefore, it could not be attributed to Pakistan under this route.Footnote ³²³ Finally, the tribunal found that the exercise of the termination right, despite being a commercial act, was attributable under the rule formulated in Article 8.Footnote ³²⁴ In this regard, it expressly noted that “attribution under Article 8 is without prejudice to the characterization of the conduct under consideration as either sovereign or commercial in nature. For the sake of attribution under this rule, it does not matter that the acts are commercial, jure gestionis, or contractual.”Footnote ³²⁵ Eventually, the claims were all rejected on the merits.

• 175. This case provides a clear illustration of the inquiries involved in questions 2 and 4. Attribution rules are not applicable to determine whether there is privity of contract or not, but they clearly govern whether – even in the absence of privity of contract – a specific act (in this case of a party to the contract, the NHA) is attributable to the respondent. At this stage, the different scopes of attribution under Articles 4, 5 and 8 again play an important role. The difference between sovereign and commercial acts is relevant for Article 5, but not for Articles 4 and 8.

• 176. The case did not raise the applicability of an umbrella clause at the merits stage, but if questions 2 and 4 are combined with questions 1 and 3, the overall principle is relatively simple. If there is no privity of contract with the State (a matter determined under domestic law), there are no obligations entered into by the State which would be protected by the umbrella clause. However, the attribution of action affecting the contract, whether the exercise of contractual rights by an entity party to the contract or acts external to the contractual framework, must be assessed in the light of the rules of attribution (not of domestic law). At this point, the scope of attribution of different routes determines what acts can be attributed.