Introduction
A claim against a state for breach of an investment treaty requires satisfaction of two preliminary conditions. First, there must be conduct that breaches an international obligation. Second, that conduct must be attributable to the state1.The first condition has generated extensive arbitral jurisprudence, particularly on standards such as fair and equitable treatment, expropriation, and full protection and security. By contrast, the second condition has received relatively limited analytical attention, despite its practical importance. This chapter focuses on the second requirement.
This relative lack of attention is difficult to justify, as the question of attribution is rarely straightforward. Modern states act not only through ministries, courts and central banks, but also through a wide range of other actors. These include state-owned enterprises (SOEs) (often separately incorporated and operating on a commercial basis), regulatory agencies, former public monopolies, municipalities, regional authorities and private concessionaires. In some cases, state activity may also involve paramilitary groups, occupation authorities or successor regimes with uncertain legal status.
In each of these situations, an investor must show that the conduct in question is attributable to the respondent state as a matter of international law. This requirement is often decisive. Where attribution is not established, the claim cannot proceed to the merits. For that reason, attribution is rarely a matter that can be resolved briefly or without careful analysis.
The governing framework is provided by the International Law Commission's Articles on Responsibility of States for Internationally Wrongful Acts (the ILC Articles), adopted on second reading in 2001 and noted by the United Nations General Assembly in Resolution 56/832.The ILC Articles are not a treaty and have not been opened for signature, and the General Assembly has repeatedly postponed consideration of their adoption in treaty form. Despite this, investment tribunals have almost uniformly recognised the ILC Articles as authoritative statements of customary international law on state responsibility3. They therefore form the principal doctrinal framework within which most modern investor–state tribunals address questions of attribution.
Within this framework, attribution operates as a secondary rule of international law. It does not define the content of a state's obligations – those are set out in the primary rules of the relevant investment treaty – but determines when conduct is treated as that of the state for the purpose of establishing responsibility4.
For present purposes, it is sufficient to note that the secondary nature of attribution rules means that they are, in principle, common throughout different areas of international law. The same rules govern attribution in investment arbitration, human rights cases and situations of armed conflict.
Whether this unity has in fact been maintained in the case law of investment tribunals is one of the central questions this chapter addresses5. In addition, the chapter examines the ILC Articles within the broader framework of state responsibility, addressing their procedural treatment in investment arbitration and the frequent confusion between attribution and related issues such as standing and contractual responsibility. It then analyses the principal attribution rules under articles 4 to 11, considers lex specialis treaty provisions, and explores less-developed areas such as judicial conduct, omissions and the interaction with umbrella clauses.
ILC framework and the limits of attribution analysis
Primary and secondary rules
Attribution under the ILC framework is a secondary rule of international law, distinct from the primary rules that define states' substantive obligations. In practice, this distinction is not always observed. Primary rules set out obligations such as the prohibition on expropriation without compensation, the duty to provide fair and equitable treatment, and the requirement to ensure full protection and security.
Secondary rules operate at a different level. They determine the legal consequences that follow when a primary obligation is breached. They address questions such as whose conduct is attributable to the state, whether a breach has been established, whether the state may rely on any circumstances precluding wrongfulness and what reparation is due. The attribution rules in Chapter II of Part One of the ILC Articles are clearly secondary rules6.
This distinction has significant practical implications for investment arbitration.
First, the rules of attribution apply across all areas of international law. As the International Court of Justice made clear in the Bosnian Genocide case, the rules governing attribution do not vary with the nature of the wrongful act in question in the absence of a clearly expressed lex specialis7. In principle, therefore, attribution is governed by a unified body of secondary rules of general application.
However, investment tribunal practice is more complex. While affirming the ILC framework, tribunals deciding cases involving SOEs and state-linked entities often emphasise the institutional and economic features distinctive to investment disputes. Tribunals applying article 8 frequently examine ownership, financing and managerial influence. The dictum in Bayindir v Pakistan suggests that the level of control required for attribution may differ in investment arbitration compared to other contexts, such as foreign armed intervention or international criminal responsibility8. Subsequent tribunals, including the Permanent Court of Arbitration (PCA) tribunal in Oschadbank v Russia, have reflected this view, although, as discussed below, the Oschadbank tribunal ultimately resolved the case on orthodox grounds without applying a distinctively lower threshold9.
Second, attribution under the ILC Articles is separate from the contractual or quasi-contractual characterisation of conduct under domestic law. The ILC Articles determine when the conduct of an entity is treated as the conduct of the state for the purposes of international responsibility. They do not determine who is a party to a contract under domestic law.
Third, attribution under the ILC Articles is distinct from the question of standing in investment arbitration. Whether an SOE qualifies as an "investor" under an investment treaty, or whether a state entity may participate in International Centre for Settlement of Investment Disputes (ICSID) proceedings under article 25 of the ICSID Convention, are separate issues.
Fourth, the secondary nature of attribution rules determines their place in analysis. A tribunal must first establish whose conduct is at issue before assessing a state's breach of primary obligations. Where attribution is disputed, the issue is often decisive: if the conduct is not attributable, the claim fails before any examination of substantive treaty standards.
Attribution, jurisdiction and standing: a persistent conflation
Case law often confuses attribution rules with two separate legal questions: whether a claimant entity should itself be treated as a state, making the dispute interstate and outside arbitral jurisdiction, and whether an SOE qualifies as an "investor" protected under an investment treaty.
These questions are analytically distinct from attribution. The first is addressed, in the ICSID context, by the Broches test, which asks whether the entity acts as an agent of the government or performs essentially governmental functions10. The second depends on the specific definition of "investor" contained in the applicable treaty11. Neither issue is determined by the ILC Articles of attribution. By contrast, the rules of attribution, as reflected in the ILC Articles, concern a different issue altogether: they determine when conduct is legally imputable to the state for the purpose of engaging its international responsibility for a wrongful act.
The tendency to conflate these distinct concepts still occurs relatively often in arbitral practice.
For example, a related issue arose in Tatneft v Ukraine, where the respondent argued that the tribunal lacked jurisdiction on the basis that the claimant – allegedly controlled by Tatarstan – was in reality an instrument of the Russian Federation, rendering the dispute effectively interstate12. The tribunal accordingly examined whether Tatneft was dependent on and controlled by the government, and whether it could be characterised as a de jure or de facto state instrument13. Although elements of this analysis echoed concepts found in the ILC framework, its primary focus was on the question of standing under the Broches test, rather than on attribution within the meaning of article 8.
This confusion constitutes a category error: attribution rules are intended to determine when conduct is attributable to the state, not to address questions of standing14.
The procedural dimension: jurisdiction, merits, burden of proof and evidentiary mechanics
Before turning to the substantive rules of attribution, it is necessary to address several procedural issues.
Procedural allocation of attribution: jurisdiction versus merits
When an investor's claim depends on attributing the conduct of a particular entity to the respondent state, the tribunal must decide whether this issue should be addressed at the jurisdictional stage or at the merits stage.
Attribution is a substantive question that properly belongs to the merits. At the jurisdictional stage, tribunals apply only a prima facie standard – provisionally accepting the claimant's factual allegations and asking whether, if proven, they could establish a treaty breach. The definitive analysis, which turns on the relationship between the state and the entity and whether the entity acted in a governmental rather than commercial capacity, is fact-intensive and is therefore deferred to the merits.
This approach was clearly expressed in Hamester v Ghana, where the tribunal emphasised that attribution should be considered separately from the question of treaty breach and is more appropriately examined at the merits stage15. In Jan de Nul v Egypt, the tribunal dismissed the attribution-based jurisdictional objection on a prima facie basis, expressly deferring the full substantive analysis of attribution under international law to the merits stage16. In Tulip v Turkey, while addressing a bifurcated jurisdictional objection regarding the mandatory notice period, the tribunal expressly stated that the full analysis of attribution under international law was an issue to be canvassed on the merits, making only a limited preliminary finding on the matter for the purposes of the immediate application17.
This allocation has important practical consequences. Treating attribution as a merits issue means that the parties will usually need to present a full evidentiary record, and the tribunal will decide the issue on the balance of probabilities rather than on a prima facie basis. It also means that attribution-based objections are unlikely to result in early dismissal of claims. Instead, tribunals will typically either join such objections to the merits or resolve them after full consideration of the evidence.
Burden and standard of proof
Attribution is a matter that the claimant must establish. This requires proof of both the relevant facts and the applicable legal standard under the ILC Articles. On the factual side, the claimant must show, for example, that the state exercised direction or control over the conduct in question, particularly under article 8. On the legal side, the claimant must satisfy the applicable test, such as demonstrating that the state exercised effective control over the specific acts, as required by the case law of Nicaragua and Bosnian Genocide18.
The standard of proof is the ordinary civil standard – namely, a balance of probabilities. At the same time, tribunals have recognised that the amount and quality of evidence required will vary depending on the strictness of the legal test. Where the applicable test is demanding, stronger evidence will be required.
Article 4: conduct of organs of the state
De jure organs
Article 4 of the ILC Articles provides that the conduct of any state organ is attributable to the state under international law. This applies regardless of the function performed – whether legislative, executive, judicial or otherwise – and irrespective of the organ's position within the state's structure, including whether it forms part of the central government or a territorial unit. Article 4(2) further clarifies that a state organ includes any person or entity that has that status under the internal law of the state.
The scope of this rule is broad. It covers not only the principal organs of the state, such as the legislature, executive and ministries, but also subordinate bodies, local authorities, courts and regulatory agencies. Attribution under article 4 does not depend on the hierarchical level of the organ, nor on whether it acted within its authority, as article 7 makes clear. In practice, investment tribunals generally treat article 4 as automatically applicable to acts of central and local government authorities, often without extensive analysis19.
Two clarifications are nonetheless important. First, although article 4(2) refers to internal law, the concept of a "state organ" is not determined exclusively by domestic legal classification20. An entity that is not formally designated as an organ under national law may still be treated as such under international law, and tribunals have, in appropriate circumstances, recognised the existence of de facto organs.
Second, whether an entity forms part of the state may turn not on its formal status but on its actual integration into the state's institutional structure – an inquiry focused on functions and role. In Lupaka Gold Corp v Peru, the tribunal held that a rural community qualified as a state organ under article 4 because its extensive statutory functions across multiple sectors evidenced sufficient integration into the state's legal and territorial organisation, regardless of formal classification21.
This approach is consistent with the ILC Commentary, which indicates that the reference to "internal law" in article 4(2) is not limited to formal designation, but extends to the functional allocation of public authority within the state22. The decision should nevertheless be treated with caution: it represents an expansive and fact-specific application of article 4 rather than a settled general rule.
De facto organs and the complete dependence test
Entities that are not formally recognised as state organs under domestic law may nonetheless be treated as such under article 4 if, in substance, they form part of the state. The ICJ made this clear in Bosnian Genocide, holding that persons or entities may be regarded as state organs for the purposes of international responsibility where they are entirely dependent on the state and act as its instrument:
persons, groups of persons or entities may, for purposes of international responsibility, be equated with state organs even if that status does not follow from internal law, provided that in fact the persons, groups or entities act in ‘complete dependence' on the state, of which they are ultimately merely the instrument23.
The complete dependence test is rigorous in its requirements. In investment arbitration, tribunals have consistently treated complete dependence as an exceptional standard and have examined factors such as the entity's legal framework, financing, governance, operational autonomy and relationship with public authorities24. The PCA tribunal (led by James Crawford) in Almås v Poland described this test as a three-part inquiry: whether the entity (1) performs essential government functions, (2) is under daily control by the central government, or (3) has no operational independence25. A similar approach can be seen in cases such as Ortiz v Algeria and Rand Investments v Serbia26. In each instance, tribunals examine the entity's legal framework, finances, governance, decision-making processes and interactions with government authorities.
When the test is met, the consequences are significant: the conduct of the entity is treated as the conduct of the state itself, even if the entity has a separate legal personality and appears to act commercially under domestic law. Deutsche Bank AG v Sri Lanka illustrates the relevant factors27. Although the tribunal found a definitive ruling unnecessary (liability rested on the acts of central state organs), it strongly suggested that Ceylon Petroleum Corporation (CPC) would qualify as a state organ or have its acts attributed to the state, citing CPC's wholly state-owned status, immunity from suit, government power to appoint and remove directors, statutory role in implementing national oil policy and the obligation to follow government directives regardless of commercial interest28.
The reasoning in Deutsche Bank is most persuasive where it focuses on the criterion of complete dependence. It becomes less clear, however, where the tribunal moves between different bases of attribution without clearly distinguishing them. In particular, the suggestion that CPC's conduct could be attributed because it was a state organ because it lacked genuine independence or because it acted under state instructions29 conflates distinct legal concepts. These correspond, respectively, to: (1) de facto organ status under article 4; (2) the disregard of separate legal personality (a concept not expressly recognised in the ILC Articles); and (3) attribution based on direction or control under article 8. Subsequent tribunals have rightly emphasised that each of these bases of attribution must be analysed separately and applied according to its own criteria.
Article 5: conduct of persons or entities exercising elements of governmental authority
Structure of article 5
Article 5 of the ILC Articles addresses a category of actors distinct from state organs. It applies to persons or entities that are not classified as organs under article 4 but are authorised by domestic law to exercise elements of governmental authority and act in that capacity in particular cases. This provision is central to attribution doctrine in investment arbitration because it covers a wide range of entities – such as SOEs, agencies, regulators and privatised former state monopolies – that interact regularly with foreign investors30.
The application of article 5 requires two conditions to be met simultaneously. First, the entity must be specifically empowered by domestic law to exercise parts of the government's authority. Second, the conduct in question must be carried out in the exercise of that authority, rather than in a purely commercial or private capacity31. The ILC Commentary makes clear that the first condition demands specific authorisation, not just general regulation: the law "must specifically authorize the conduct as involving the exercise of public authority; it is not enough that it permits activity as part of the general regulation of the affairs of the community"32.
Structural and functional analysis: the Maffezini tests in proper perspective
Much of the case law on article 5 has been influenced by the structural and functional approach developed in Maffezini v Spain33. The tribunal first examined the entity's legal status, organisation and governing rules (the structural test). It then considered whether the entity was performing a governmental function (the functional test). Because the state chose to create SODIGA in the form of a private commercial corporation, the tribunal held that the structural test was insufficient on its own for attribution, thus requiring the functional test. Ultimately, the tribunal concluded that SODIGA met both tests: the structural test of state creation and capital ownership, and the functional test of performing activities of a public nature. Consequently, its conduct involving the exercise of governmental authority was attributable to the state.
While the approach in Maffezini remains influential, its application should be carefully contextualised. The case was decided before the adoption of the ILC Articles, when the rules on attribution were less clearly articulated. In substance, the tribunal's functional test corresponds to the second requirement of article 5 – namely, whether the entity was exercising governmental authority in the specific instance.
The idea that state ownership or control creates a presumption of attribution has been rejected in later practice. In Tulip v Turkey, the tribunal clarified that a corporation does not become a state organ solely due to state ownership or control34.
Sovereign versus commercial conduct
The principal difficulty in applying article 5 lies in distinguishing between sovereign and commercial conduct. The ILC Articles Commentary recognises that what counts as "governmental" depends on the societal and historical background35. Arbitral tribunals have sought to clarify this distinction by asking whether the conduct involves powers only a state (or an entity acting on its behalf) could exercise or whether it is conduct that could equally be carried out by a private actor36.
The case law demonstrates this divide. On the sovereign side, tribunals will find attribution where the entity exercises statutory or public powers. In Rand v Serbia, for example, the tribunal found that the privatisation agency's seizure of shares under statutory powers was attributable to Serbia under article 537. A more recent example is Waleed Aljarallah v Turkey, where the tribunal concluded that the Turkish Savings Deposit Insurance Fund was empowered to exercise governmental authority by virtue of its statutory powers, including the forced exercise of shareholder rights and the liquidation of companies, granted specifically in the context of the state's response to the 2016 attempted coup38.
By contrast, tribunals have declined to find attribution where the conduct is essentially commercial. In Jan de Nul v Egypt, the alleged misrepresentations and contract breaches by the Suez Canal Authority during a dredging project were not attributed to Egypt because, even though the authority was empowered to exercise government functions, its acts in that specific context were essentially commercial39.
Similar reasoning appears in EDF Services v Romania, which concerned airport and airline contracts40, and in Hamester v Ghana, where the Ghana Cocoa Board's actions under a cocoa processing agreement were found to be purely contractual41. Professor Brigitte Stern's tribunal in Hamester emphasised that attribution under article 5 requires both legal authority to exercise governmental functions and conduct that, in fact, reflects the exercise of such authority42.
Another important case on the commercial side is Bosh International v Ukraine. The dispute involved a contract between the claimant and Taras Shevchenko National University of Kyiv. The tribunal determined that, although the University was a state institution, its termination of the contract was a commercial act rather than an exercise of governmental authority – meaning the University was acting in its capacity as a private entity (jure gestionis) rather than as a sovereign (jure imperii)43. This case shows that even when an entity performs public or regulatory functions, not all its conduct is automatically attributable under article 5; each act must be evaluated based on its character44.
Delegation versus general regulation
Article 5 applies only where domestic law specifically authorises an entity to exercise governmental authority. It is not enough that the conduct is permitted within a general regulatory framework. As the ILC Commentary makes clear, there must be a clear delegation of public powers, rather than mere participation in regulated activity45. This distinction – between specific delegation and general regulation – is essential to prevent an overly broad application of article 5. A private company operating within a regulated sector does not, for that reason alone, exercise governmental authority. Likewise, an SOE engaging in ordinary commercial transactions accessible to private firms does not automatically act with governmental authority.
Case law reflects this distinction sharply. In Ulysseas v Ecuador, the tribunal acknowledged that the state entity, CONELEC, had exercised delegated governmental authority in granting the power generation licence, but held that its subsequent seizure and temporary administration of the claimant's power barge was the "ordinary behaviour of a contractual counterparty" under the concession agreement, and therefore did not trigger attribution under article 546.
Article 7: conduct ultra vires or contrary to instructions
Article 7 provides that the conduct of a state organ or an entity exercising governmental authority remains attributable to the state under international law, even where that conduct exceeds the limits of authority or contravenes instructions. The underlying rationale is well established: a state cannot avoid international responsibility for internationally wrongful acts by relying on internal limitations that are not apparent to foreign investors and remain matters of domestic law.
Article 7 is frequently applied in investment arbitration. Tribunals have consistently held that ultra vires conduct does not preclude attribution, provided the act was carried out in an official capacity. In Maffezini v Spain, for example, the tribunal found that a funds transfer ordered by an official of SODIGA (a state entity) was attributable to Spain, even if the official's specific action was ultra vires of his mandate47.
The scope of article 7 is strictly limited. As clarified in Ortiz v Algeria, it applies only to attribution under articles 4, 5 and 6, not to conduct attributed via direction or control under article 8. There, the SOE ENPI defied the Ministry of Habitat's explicit instructions to provide land to the investor. Because ENPI acted under state instructions (article 8) rather than as a state organ, article 7 could not save the attribution; its ultra vires conduct broke the chain of effective control, rendering its acts non-attributable to Algeria48.
Furthermore, article 7 does not extend to conduct that is purely private. Where an individual or entity is acting for personal reasons rather than in any official capacity, their actions cannot be attributed to the state. This distinction is illustrated by Yeager v Iran, where the Iran–United States Claims Tribunal refused to attribute to Iran an Iran Air agent's extortion of an extra payment to let the claimant's daughter board a flight: while ultra vires acts of an organ are attributable, acts committed in a purely private capacity – such as demanding a bribe for personal profit – fall outside official capacity and beyond the scope of attribution49.
Article 8: conduct directed or controlled by the state
Structure of article 8 and its ICJ foundation
Article 8 provides that the conduct of a private individual or group may be attributed to the state where they act on the state's instructions, or under its direction or control, in carrying out the conduct in question50. This provision applies to actors who are not state organs and do not exercise governmental authority within the meaning of article 4 or 5. Even so, where their conduct is sufficiently closely connected to the state, it may be treated as conduct of the state itself.
The applicable standard under article 8 was clarified by the ICJ in the Bosnian Genocide case. In that case, the Court addressed the divergence between two approaches51. The first, developed in Nicaragua v United States, requires effective control, meaning that the state must have directed or enforced the specific conduct at issue52. The second, adopted by the International Criminal Tribunal for the former Yugoslavia in Prosecutor v Tadić, applies an overall control test, which allows for attribution based on a broader pattern of control over the group53.
The Court in Bosnian Genocide reaffirmed that the effective control test is the correct standard for attribution under article 8. It rejected the overall control approach used in Tadić as insufficient for establishing state responsibility, although it accepted that it may be appropriate in other contexts, such as the classification of armed conflicts under international humanitarian law54. The Court further emphasised that, in the absence of a specific lex specialis, the rules of attribution do not vary depending on the type of wrongful act involved55.
Effective control in investment arbitration
The implication for investment arbitration is important but must be stated with some care. Tribunals applying article 8 generally begin from the effective-control standard articulated in Nicaragua and reaffirmed in Bosnian Genocide. This requires that the state exercise control over the specific conduct constituting the alleged breach, rather than over the entity's activities in general56. This formulation remains the central test and should be applied consistently.
In investment awards, this formulation is sometimes expressed as requiring the claimant to prove both general control over the person or entity and specific control over the conduct in question. For example, in Gavrilović v Croatia, the tribunal stated that the state must demonstrate "both general control over a private party and specific control over the conduct in question"57, and that in the absence of evidence of both general and specific control, the conduct cannot be attributed to the state.
Although this distinction can be useful, it must be applied with care. The ICJ's view is that specific control – meaning effective control over the particular action that caused the violation – is the decisive legal test. By contrast, general control is not an independent requirement under the ILC Articles, but rather a form of contextual or circumstantial evidence from which specific control may be inferred. As the tribunal in Bayindir v Pakistan observed, the key question is not whether the state controls the entity as a whole, but whether it directed or controlled the conduct giving rise to the alleged breach58. The same tribunal also suggested that the level of control required for attribution under article 8 may vary depending on the context (for example, between cases of foreign military intervention, international criminal responsibility and investment disputes)59. This view has been followed by other tribunals, including the tribunal in Oschadbank v Russia, which accepted that the standard of control in investment arbitration may differ from that applied in other areas of international law60. Similar reasoning has also been endorsed in some academic literature61.
Whether this departs from the ICJ's unified framework or merely applies effective control to a different factual setting is contested. The most conceptually sound way to resolve this tension – and the position defended in this chapter – is to maintain a strict analytical distinction between the substantive legal standard and the evidentiary mechanisms used to satisfy it. As a matter of substantive international law, article 8 imposes a single, unified standard: the claimant must prove that the state directed or controlled the specific conduct in question. There is no separate, lighter "investment arbitration standard" of attribution. To the extent that dicta in cases such as Bayindir or Oschadbank suggest that the legal test itself is different, that implication risks contributing to the fragmentation of international law.
However, as a matter of evidence, the factual context inevitably dictates how that specific control may be proven. In the context of armed conflict, proving specific control over a paramilitary group typically requires evidence of direct operational orders. In the corporate context of investment arbitration, a state exercises power differently. Specific control over a commercial transaction or a contract termination by an SOE might be proven circumstantially (for instance, through a combination of absolute state ownership, the presence of government ministers on the board, and overarching policy directives that leave the entity with no genuine commercial discretion in the specific instance).
Thus, the unity of the ILC Rules is preserved. Tribunals do not need to lower or alter the legal standard of article 8; they simply draw different evidentiary inferences based on the corporate and regulatory realities of the dispute. Even in cases referencing a somewhat flexible approach, tribunals ultimately base their findings of state responsibility on evidence that the state directed or controlled the specific conduct, fully in line with the traditional ILC framework.
Instructions: imperative, not permissive
Article 8 identifies three alternative bases for attribution: conduct carried out "on the instructions of" the state or "under its direction or control". Although these concepts are related, they are distinct, and each may independently establish attribution if the required link to the state is shown in relation to the specific conduct62.
The ILC Commentary and arbitral practice indicate that "instructions" refer to a clear and binding directive or order issued by the state that determines the conduct of the non-state actor. The instruction must be mandatory in nature63. Mere encouragement, support, invitation or approval is not sufficient. In Ortiz v Algeria, the tribunal explicitly made this point, holding that a recommendation or encouragement by the state does not meet the threshold for an instruction64. In reaching this conclusion, the Ortiz tribunal relied on the earlier decision in Electrabel v Hungary, which established that an "invitation to negotiate cannot be assimilated to an instruction" under article 865.
These decisions confirm that the concept of "instructions" sets a high threshold. Establishing attribution on this basis requires clear evidence that the state issued a directive compelling the conduct in question. Ultimately, the strict requirement that instructions must be imperative – rather than merely permissive – highlights the significant evidentiary burden placed on claimants seeking to establish attribution under this limb of article 8.
Critique of conflation: article 4 (de facto organ) versus article 8
A recurring difficulty in the case law is the tendency to confuse the test for de facto organ status under article 4 with the test for attribution based on direction or control under article 866. The two standards are conceptually distinct and should be kept strictly separate, as originally articulated by the ICJ in Bosnian Genocide67.
Under article 4, an entity may be treated as a de facto state organ only where it is completely dependent on the state. This is a strict and exceptional standard. Where it is met, the entity is treated as part of the state itself and all of its conduct is attributed to the state.
Article 8 operates differently. It does not require the entity to be treated as a state organ. Instead, it attributes only those specific acts that are carried out on the state's instructions or under its effective control. The entity may remain otherwise independent and retain autonomy in its general operations.
These differences have important consequences. A finding of attribution under article 8 in respect of a particular act does not mean that the entity is a de facto organ of the state. Conversely, the fact that an entity is not completely dependent on the state does not prevent a finding that the state directed or controlled the specific conduct at issue68.
Article 11: conduct acknowledged and adopted by the state as its own
Structure of article 11
Article 11 provides that conduct that is not attributable to a state under articles 4 to 10 may nevertheless be treated as an act of that state under international law if the state later acknowledges and adopts the conduct as its own. This rule allows for retrospective attribution: conduct that was not initially attributable to the state at the time it occurred may later become so if the state clearly and unequivocally embraces it as its own.
A well-known example is the Tehran Hostages case. The initial seizure of the US Embassy by private militants was not attributable to Iran, but Ayatollah Khomeini's subsequent endorsement transformed the continued occupation into an act of the Iranian state. (The ILC Commentary notes that, because Iran was already responsible for its failure to prevent the seizure, the ICJ did not decide whether adoption had retroactive effect ab initio; where adoption is unqualified, however, it generally does)69.
Elements of article 11
Four key points emerge from the ILC Commentary and the case law. First, article 11 concerns retrospective attribution: the conduct must not have been attributable to the state at the time it occurred under articles 4 to 1070. Second, acknowledgement and adoption are cumulative requirements. As the Commentary makes clear, the use of the word "and" reflects that both elements must be satisfied71. Third, the state's acknowledgement and adoption must be clear and unequivocal, whether expressed through words or conduct72. Fourth, a distinction must be drawn between acknowledgement and adoption, on the one hand, and mere endorsement or support, on the other. A state does not adopt conduct simply by expressing approval or by recognising that it occurred73.
Recent case law
For many years, article 11 played a minor role in investment arbitration practice. Its recent importance is mainly due to two lines of cases in which private or paramilitary actions became attributable to a state because they were incorporated into the state's official action, even where attribution under articles 4, 5 or 8 remained uncertain.
In Saint-Gobain v Venezuela, the tribunal considered whether the May 2010 takeover of a proppants plant by employees and union members was attributable to Venezuela. Although not state-ordered (failing article 8), the state-owned PDVSA immediately exploited the situation to advance its nationalisation agenda. The tribunal rejected Venezuela's argument that attribution arose only with the March 2011 Expropriation Decree: internal documents showed PDVSA had already integrated the takeover into the Socialist Guyana Plan – establishing a permanent presence, replacing flags and uniforms, and readying the plant for state-aegis production – thereby acknowledging and adopting the private conduct as its own under article 1174.
А recent arbitration award concerning assets in Crimea also illustrates the growing relevance of article 11 in situations of armed conflict and territorial transition.
In Stabil LLC v Russia, the tribunal examined whether Russia could be held responsible for the actions of paramilitary forces that seized petrol stations owned by Ukrainian investors. The tribunal first found that the forces' actions could be attributed under article 5 because they exercised government functions. Alternatively, under article 11, the tribunal highlighted three points:
- the Crimean authorities' later formal nationalisation of the properties showed Russia adopted the forces' conduct;
- the Russian President publicly acknowledged support from Russian forces; and
- Russia did not attempt to prosecute or hold the paramilitary forces accountable75.
Taken together, these cases show that article 11 has gained practical significance in situations where private conduct is later incorporated into state policy. They also illustrate the high threshold for its application; attribution depends not merely on support or acquiescence, but on clear and subsequent adoption by the state.
Observations on the scope of article 11
Recent case law is generally well-reasoned, but it is important to maintain a clear and limited understanding of article 11 and avoid overexpansion in its application.
Article 11 does not operate as a general rule of ex post approval. It does not permit attribution of any private conduct that the state later supports or tolerates. The state's acceptance must be explicit and unequivocal; mere acquiescence or passive endorsement is not sufficient76.
Article 11 is residual in structure. It becomes necessary only where attribution is not otherwise established under the preceding articles. Where those provisions are satisfied, there is no need to rely on article 11.
Although the tribunal in Stabil v Russia relied on article 11 in the alternative, this approach is best understood as fact-specific. A more accurate analysis would have resolved attribution primarily under article 5, with article 11 addressed only as a subsidiary argument.
Articles 9 and 10: conduct in the absence of official authorities and the conduct of insurrectional movements
Although articles 9 and 10 of the ILC Articles have historically been less frequently applied in investment arbitration than articles 4, 5, 8 and 11, they still merit brief consideration. Their importance is likely to increase in cases involving armed conflict, territorial administration in occupied territories or changes of government.
Article 9 states that conduct by an individual or group can be attributed to the state if that person or group is effectively exercising government functions in the absence or failure of official authorities, under circumstances that call for such exercise. In essence, it addresses situations in which private actors temporarily assume public functions because the state is unable to act77.
Article 10 deals with the attribution of conduct by insurgent or other movements. Paragraph (1) considers the actions of an insurgent movement that becomes the new government to be acts of that state. Paragraph (2) extends this to movements that establish a new state within part of the territory of the original state or its administration. Paragraph (3) confirms that conduct attributable under articles 4 to 9 remains attributable in these situations78.
Lex specialis: special rules of attribution in treaty practice
As article 55 of the ILC Articles makes clear, the general rules of attribution can be overridden by specific rules found in treaties, known as lex specialis79. Numerous examples from state treaty practice illustrate this. For instance, article 1503(2) of NAFTA attributes to a state the conduct of a state enterprise when it acts under governmental authority. This provision is generally understood as a special rule that modifies the general ILC analysis, particularly the article 5 inquiry, and may also reduce the need for a separate article 8 analysis where the treaty rule is satisfied80. This rule has been included, with some changes, in USMCA article 22.3 and in recent free trade agreements involving the United States and Canada.
Recurrent themes
Attribution of judicial conduct
It is well established that judicial conduct is attributable to the state under article 4 of the ILC Articles. The ILC Commentary makes clear that the term "organ" includes bodies exercising judicial functions, and tribunals have consistently applied article 4 to acts of courts81.
Two aspects of judicial conduct frequently arise in practice and require careful distinction between the attribution of an act and the completion of a breach.
First, questions have occasionally arisen regarding the acts of lower courts. As clarified in Loewen v United States, while a lower court decision is imputable to the state under article 4, it does not generally constitute an internationally wrongful act (ie, a denial of justice) until the judicial system as a whole has spoken – the requirement of judicial finality affording the state an opportunity to redress the inchoate breach through its appellate mechanisms82.
Second, state conduct connected with judicial proceedings – post-judgment enforcement, non-enforcement or excessive delays – raises the question of whether it meets the high threshold for customary denial of justice or breaches a more specific, less demanding treaty obligation. In White Industries v India, the tribunal found that a nine-year judicial delay, although not reaching the denial of justice threshold, was attributable to the state and breached the distinct treaty obligation to provide "effective means" of asserting claims83. This approach confirms that while all court acts are attributable under the ILC framework, their substantive wrongfulness depends entirely on the primary rule being applied.
Attribution of omissions
The ILC Articles do not generally distinguish between acts and omissions84.
In practice, however, omissions raise particular conceptual difficulties. Where the failure emanates from a state organ – such as a court's failure to deliver a timely judgment – attribution is straightforward. In White Industries v India, for example, the tribunal treated the Indian Supreme Court's seven-year delay as conduct of the state, without the issue of attribution arising85.
More complex issues arise when the omission consists of a state authority's failure to intervene in private conduct. Here it is vital to distinguish between attribution of the private actor's conduct and attribution of the state's own omission; a mere failure to issue instructions or intervene is generally insufficient to attribute the private conduct to the state under article 8, which, as the ILC Commentary emphasises, requires a real link established by "specific directions or by exercising control"86.
Instead, the proper legal approach is to attribute the state's own inaction to the state under article 4 or 5, and then assess whether that omission breaches a primary obligation (such as the duty to provide full protection and security). As the ILC noted regarding the United States Diplomatic and Consular Staff in Tehran case, the responsibility of the state was entailed specifically by the "inaction" of its authorities in failing to prevent the seizure of the embassy by private militants87.
Greater attention must be paid to distinguishing whether an omission constitutes a failure by the state to exercise its public authority (attributable under article 4 or 5) or whether it is improperly used as a basis for attributing purely private conduct to the state under article 8.
Conclusion
The law of attribution in international investment arbitration has developed significantly over the past 25 years. Its basic structure is now largely settled. The ILC Articles provide the governing framework, interpreted in light of general international law – particularly the case law of the ICJ in the Nicaragua and Bosnian Genocide cases – and, where applicable, supplemented by specific treaty rules (lex specialis). Within this framework, the main rules of attribution (covering state organs (article 4), entities exercising governmental authority (article 5), ultra vires conduct (article 7), direction or control (article 8), and acknowledgement and adoption (article 11)) can be applied in a principled and coherent way. The case law discussed in this chapter, including Lupaka Gold, Ortiz Construcciones, Rand Investments, Waleed Aljarallah, MTS v Turkmenistan, Saint-Gobain and Stabil, highlights that tribunals are generally engaging carefully with this framework and reaching results that are, for the most part, doctrinally sound.
Three concluding observations follow.
First, the variety of "tests" found in arbitral decisions – such as structural and functional tests, complete dependence, effective control and distinctions between general and specific control – does not reflect a fragmentation of the law. Rather, it shows how tribunals apply the same underlying ILC framework to different factual situations. The unity of attribution rules throughout various areas of international law, as emphasised by the ICJ in Bosnian Genocide, should remain a guiding principle for investment tribunals.
Second, it remains important to distinguish attribution from related but separate legal questions. These include issues of jurisdiction and standing, the identification of the contractual obligor, and the scope and elevating effect of umbrella clauses. Confusion between these questions continues to appear in practice and can lead to unclear reasoning. The solution is not to develop new attribution tests, but to maintain a clear analytical separation between these different areas.
Third, the future development of attribution law is unlikely to come from refinement of the core tests, which are now relatively settled. Instead, it will arise in areas that remain underdeveloped, including:
- the treatment of omissions, particularly the need to clearly distinguish between the attribution of private conduct under article 8 and the attribution of a state's own failure to act under articles 4 or 5;
- the evidentiary dimension of attribution, particularly the role of adverse inferences and document production where key evidence is controlled by the respondent state; and
- the application of article 11 in situations where private or paramilitary conduct is later incorporated into state policy, including in contexts such as territorial transition or armed conflict.
Recent cases such as Saint-Gobain and Stabil demonstrate the growing practical importance of these issues.
From a practical perspective, investors should carefully address attribution from the outset of a dispute, as the burden of proof lies with the claimant and outcomes often depend on the quality of the evidence. For states, the lesson is equally clear: when utilising SOEs, states should maintain a strict separation between ordinary commercial activities and the exercise of delegated governmental authority to avoid unintended attribution under article 5. Furthermore, they must be mindful of the level of direction and control exerted over such entities' specific operations to avoid triggering article 8. Otherwise, the state risks being held internationally responsible for conduct that it did not intend to assume as its own.
Ultimately, the rules of attribution reflect a fundamental principle of international law: a state cannot avoid responsibility for the exercise of its power simply by acting through separate entities or formal structures. Preserving and applying that principle consistently remains the central task of attribution doctrine in investment arbitration.