Investment treaty arbitration is currently undergoing a backlash, precipitated in large measure by the problem of inconsistency in investment awards. Inconsistent reasoning and decisions have proved particularly problematic when different shareholders of an affected company pursue claims for reflective loss before different tribunals concerning the same dispute. This article examines two recent investment treaty cases—CC/Devas v India and Deutsche Telekom v India—that are a critical example of this very real problem.
The first part of this article assesses the inconsistent interpretation and application of essential security interest clauses in those cases. As this article shows, the disagreement between the tribunals is explained by contradictory assessments of the same facts.
The second part critically evaluates the impact of the two cases on the development of the precise criteria for successfully invoking an essential security interest clause. Such clauses comprise two elements: (i) the existence of such an interest; and (ii) the nexus between the interest and the measure adopted. These elements have hitherto been subject to inconsistent interpretations, leaving the exact contours of such clauses unclear. This article uncovers which security interests can qualify as ‘essential’ interests, focusing particularly on the question whether such an interest encompasses protection of strategic resources for military and non-military use. It also examines the meaning of the nexus requirement of ‘necessary’, and explores the difference(s) between the nexus requirements of ‘directed to’ and ‘necessary’.
The last two decades have witnessed an unprecedented growth—a ‘baby boom’—in the number of investment treaty arbitrations (ITAs) between foreign investors and host States.2 In the increased academic and public scrutiny that has followed this rapid rise, many have repeatedly questioned the legitimacy of the investment treaty regime, arguing that it is undergoing a ‘legitimacy crisis’.3
Various factors have precipitated this ‘backlash against investment arbitration’.4 They range from procedural issues—such as lack of transparency,5 ‘double hatting’ by lawyers who act as both counsel and arbitrator6 and the absence of an appeals mechanism7—to substantive issues such as expansive interpretations of investment protection standards by investment tribunals.8 Additionally, a central concern relates to inconsistency of decisions by tribunals.9 Even the staunchest defenders of ITA concede that ‘the current system of investment arbitration has not been designed … to promote uniformity and consistency’.10
As Susan Franck describes, inconsistent decisions can result when tribunals reach different conclusions regarding the meaning of the same treaty provision, or when tribunals diverge in their assessment of cases involving similar commercial situations and similar investment rights.11 Significant concerns also arise when tribunals reach divergent conclusions in situations in which the same dispute is litigated before separate tribunals by different shareholders of an affected company. The inconsistent outcomes in the Lauder and CME saga is well known.12 Dutch company, CME, and its ultimate shareholder, Lauder, commenced parallel arbitration against the Czech Republic for revoking television licences granted to a local company, TV Nova. Both arbitrations concerned the same facts, the same host State action and the same dispute. The Lauder Tribunal decided in the Czech Republic’s favour, but the CME tribunal—within two weeks of the Lauder decision—held that the Czech Republic had breached its treaty obligations.
Against this backdrop, two awards against India—CC/Devas v India (CC/Devas) and Deutsche Telekom v India (DT)—which were recently made public, are of particular significance.13 Underlying both cases are the same facts and the same dispute. An Indian State-owned company, Antrix, cancelled an agreement by which it leased electromagnetic spectrum on two satellites to an Indian commercial enterprise, Devas. The Indian government ordered the cancellation, citing increased demands for allocating the spectrum to meet the national needs of military and para-military forces and public utility services. Alleging expropriation and breach of fair and equitable treatment (FET), three Mauritian shareholders of Devas initiated the first case, CC/Devas, under the India–Mauritius BIT. The second case, DT, was commenced by German shareholder Deutsche Telekom under the Germany–India BIT. Both BITs include an essential security interest (ESI) clause, which limits the availability of investment protections when a State adopts a measure to protect such an interest. The fundamental question before the tribunals, thus, was whether India had cancelled the agreement to protect an ESI. While the CC/Devas Tribunal found in India’s favour (by a majority), the DT Tribunal reached the opposite result five months later, concluding that there was no nexus between India’s ESI, ie the needs of the military and para-military forces, and the decision to cancel the agreement.
This article critically analyses the divergent interpretation and application of the ESI clause in CC/Devas and DT. It will show that the DT Tribunal explained the divergent conclusions by relying, partly, on the different legal thresholds for establishing a nexus between the decision and the ESI in the India–Mauritius and Germany–India BITs. Under the India-Mauritius BIT, India’s decision needed to be ‘directed to the protection of’ its ESI. By contrast, the Germany–India BIT covered only decisions that were ‘necessary’ to protect an ESI. The DT Tribunal suggested that the ‘necessary’ standard requires a ‘more stringent nexus’ between the measure and the ESI than the ‘directed to’ standard.14 However, as this article will argue, it was not the difference in treaty standards that led to the split in opinion between the tribunals. Rather, the disagreement is explained by different, and contradictory, assessments of the same facts: the DT Tribunal undertook a cumulative assessment of facts preceding and following India’s decision to cancel the agreement, whereas the CC/Devas Tribunal based its finding solely on the ultimate decision.15
A second purpose of this article is to revisit the debate on the precise legal criteria for successfully invoking an ESI clause. Such clauses are pervasive in investment treaties, and have been subject to extensive academic scrutiny.16 Yet, until CC/Devas and DT, they had only been invoked in a series of five cases that assessed the necessity of measures adopted by Argentina in the wake of its 2001–2 economic crisis.17 In those cases, too, the tribunals diverged in their factual and legal assessments, leaving much to be desired in terms of guidance on the precise contours of such clauses. Through an analysis of the awards in CC/Devas and DT, this article will examine which security interests qualify as ‘essential’ interests, and what is meant by ‘necessary’, and will explore the difference(s) between the nexus requirements of the terms ‘directed to’ and ‘necessary’.
Section II recaps the conflicting jurisprudence on ESI clauses that has been developed in decisions against Argentina. Section III demonstrates the contradictory elements of the rulings in CC/Devas and DT. Section IV re-evaluates the legal elements of ESI clauses and examines the implications of these developments. Section V concludes.
II. ESI CLAUSES—A BREEDING GROUND FOR INCONSISTENCY: REVISITING THE CASES ON THE ARGENTINIAN ECONOMIC CRISIS
ESI clauses owe their origin to friendship, commerce and navigation treaties (FCN treaties) concluded by the United States in the post-Second World War era.18 They enlist the ‘exceptions’ to a State’s substantive obligations under a treaty.19 Although a significant proportion of investment treaties include ESI clauses, their formulations differ.20 Broadly speaking, such clauses comprise two elements: first, the ‘permissible objective’ in relation to which host States can take measures that would otherwise breach their investment obligations. ESI clauses are a sub-type of a ‘non-precluded measure’ clause, in which the permissible objective is the protection of an ‘essential security interest’. A non-precluded measure clause is wider in scope, and also covers objectives such as ‘public order’, ‘public health’, ‘public morality’ and situations of ‘extreme emergencies’.21 The second element requires proof of a causal link between the measure and the objective (the ‘nexus’ requirement), which varies widely, both within and across the practice of States.22 Surveys demonstrate that the required nexus varies from measures ‘necessary’ to protect the permissible objective, to those ‘required’, ‘related to’, ‘directed to’, or ‘for’ the protection of the stated objective.23
Despite their prevalence, ESI clauses have remained largely dormant in ITA. Prior to CC/Devas and DT, only Argentina had invoked them, in five cases in which investors challenged measures that that country had adopted to address its economic crisis in 2001–2: CMS (2005); LG&E (2006); Sempra (2007); Enron (2007); and Continental (2008) (collectively, the Argentinian cases).24
These disputes arose from near-identical facts. The five investors acquired shareholdings in companies that Argentina privatised in the late 1980s and early 1990s. The privatisation programme was backed by guarantees, which included the pegging of the Argentinian peso to the US dollar, and the calculation of tariffs in US dollars. By 2001, Argentina was in the grips of a severe economic crisis. In response, Argentina adopted a series of measures to stabilise its economy, which included devaluing the peso by abolishing the currency board that pegged the Argentinian peso to the US dollar, pesifying all dollar-denominated contractual and financial obligations, and freezing bank deposits.25 These measures, the investors alleged, violated Argentina’s investment obligations.
Not just the factual background, but the investment treaty invoked was also identical. Being US nationals, the investors in all five cases relied upon the Argentina–US BIT. Article XI of that BIT exempts a State from its investment obligations if the measures adopted were ‘necessary’ to protect its ESI. Argentina relied upon this ESI clause to plead exemption from its obligations, arguing that its measures were necessary to protect against the social, political and economic challenges presented by the economic crisis.
Despite these similarities, the tribunals split in their legal assessments of what could constitute an ESI and what the required nexus between the measure and the ESI under the ‘necessary’ standard was. They also diverged in their factual assessment of whether Argentina’s measure met the stated legal standards. The LG&E and Continental tribunals concluded in Argentina’s favour, but the remaining three tribunals—in CMS, Enron and Sempra—held that the economic crisis did not threaten Argentina’s ESI, nor were the measures a necessary response. Within each group, too, there were significant differences in reasoning.
A. On ESI
A primary point of disagreement between the tribunals concerned the relationship between an ESI clause and the customary international law (CIL) defence of ‘necessity’. The ESI clause itself provided little guidance on the meaning and scope of the terms ‘necessary’ and ‘ESI’. The tribunals thus had to decide whether the standards for applying the necessity defence under CIL applied also to the treaty’s ESI clause. Under CIL, ‘necessity’ precludes wrongfulness only if a State acts to safeguard an ‘essential interest’ against a ‘grave and imminent peril’.26
The CMS, Enron and Sempra tribunals pegged the ESI clause to the CIL defence of necessity. The CMS Tribunal did not clarify its basis for doing so, stating simply that the existence of a state of necessity under CIL and the treaty’s ESI clause raised ‘one fundamental issue’.27 The Enron and Sempra tribunals explicitly clarified that reference to the CIL standard of necessity was made necessary because the BIT provision did not provide a definition of ESI.28 Having pegged the ESI clause to necessity under CIL, the three tribunals agreed that whether Argentina’s ESI was at stake would depend on the gravity and imminence of the crisis.29 For the CMS Tribunal, this standard of gravity could be met only if Argentina’s crisis resulted in ‘total economic and social collapse’.30 For the Enron and Sempra tribunals, only if the very existence of the State and its independence was compromised could the crisis qualify as involving an ESI.31
The LG&E and Continental tribunals, by contrast, distinguished the ESI clause from necessity under CIL. As the Continental Tribunal explained, necessity and ESI clauses have different objectives. Necessity under CIL precludes only the wrongfulness of an act. An ESI clause is a ‘safeguard’ clause; a measure covered by it lies outside the scope of the investment treaty in so far as a State would not be held to have breached its obligations under the treaty.32 The LG&E and Continental tribunals disagreed that the ESI clause could apply only in situations of ‘total collapse’ or when a State’s existence was compromised.33 For them, a ‘threat’ of total collapse,34 or a danger that the State’s ‘economic foundation is under siege’, suffices to prove the existence of an ESI.35
On either account, all the tribunals agreed on the facts that Argentina faced a ‘severe’ crisis.36 However, the CMS, Enron and Sempra tribunals were not convinced that the crisis was sufficiently severe to compromise Argentina’s existence and independence.37 For the LG&E and Continental tribunals, the facts proved that Argentina was in the midst of a severe economic crisis that could not be addressed by ordinary measures and called for ‘immediate, decisive action’.38 As the LG&E Tribunal noted, ‘[t]o conclude that such a severe economic crisis could not constitute an essential security interest is to diminish the havoc that the economy can wreak on the lives of an entire population and the ability of the Government to lead’.39
Academics commonly explain these divergent conclusions by pointing to the different viewpoints of the tribunals on the interpretive relevance of the CIL standard of necessity.40 This provides only a partial explanation, as the starting point of the tribunals’ reasoning also varied. The CMS, Enron and Sempra tribunals began their reasoning from the point that the object and purpose of investment treaties is to protect foreign investments, including in situations of economic difficulties.41 As the Enron and Sempra tribunals articulated, ‘any interpretation resulting in an escape route from the obligations defined cannot be easily reconciled with that object and purpose’.42 The LG&E and Continental tribunals placed no such premium on the investment protection objectives of an investment treaty. Consequently, these tribunals found it imprudent to limit the application of ESI clauses to measures taken after a State’s collapse.43 Additionally, the Continental Tribunal called for a ‘margin of appreciation’ in favour of States.44
B. On the Standard of ‘Necessary’
The CMS, Enron and Sempra tribunals were also unconvinced that Argentina’s measures were ‘necessary’ to address the situation. Turning again to necessity under CIL for guidance, these tribunals declared that a measure could be necessary if it was the ‘only way’ to protect against a crisis.45 The LG&E and Continental tribunals rejected this approach but failed to provide a consistent standard for determining the necessity of adopting a measure. The LG&E Tribunal stated that a measure is necessary if ‘a State has no choice but to act’.46 Equally, the LG&E Tribunal declared that a finding of necessity would not be affected by the fact that a State could avail itself of alternative responses.47 The Continental Tribunal, borrowing from the necessity standard adopted by the WTO, articulated a two-fold test.48 First, a measure should ‘contribute materially’ to the protection of a State’s ESI.49 Secondly, a State must not have ‘reasonably available alternatives, less in conflict or more compliant with its international obligations’.50
The disagreement extended to the case facts. The LG&E Tribunal considered the measures a necessary and legitimate response, given the crisis at play.51 According to CMS, Enron and Sempra, however, the measures that Argentina adopted were not the ‘only way’ to respond to the crisis.52 None of these tribunals identified the alternatives, stating that it was not for the Tribunal to conduct such an assessment.53 The Continental Tribunal concluded, first, that there was a ‘genuine relationship of end and means’ as the measures were ‘inevitable, or unavoidable, in part indispensable, and in any case material’ to prevent a breakdown of the Argentinian economy.54 Second, unlike those in CMS, Enron and Sempra, the Continental Tribunal conducted an in-depth review of the proposed alternatives and found them unsuitable to address the crisis.55
The divergent outcomes in the Argentinian cases are repeatedly cited as examples to highlight concerns of inconsistency and unpredictability in ITA.56 Academics occasionally rationalised the initial divergence of views in the first two cases—CMS and LG&E—as the teething problems of a system in its infancy, and it was hoped that a ‘common legal opinion or jurisprudence constante’ on the requirements for invoking ESI clauses would develop as the system evolved.57 On the contrary, the cases following CMS and LG&E have left us none the wiser regarding the precise legal contours of ESI clauses, nor has the application of ESI clauses to the same facts been consistent. The tribunals agreed on two issues: that ESI clauses are not self-judging,58 and that non-military, economic crises can, in principle, qualify as ESIs.59 Furthermore, the clarifications issued by the annulment committee decisions in CMS, Enron and Sempra make clear that ESI clauses are distinct from necessity under CIL.60 These elementary issues aside, the Argentinian cases offered little clarity on the two key elements for invoking the ESI clause: what constitutes an ESI, and when the nexus between a measure and an ESI is established under the ‘necessary’ standard.
CC/DEVAS AND DT: SAME FACTS, DIFFERENT RESULTS
ESI clauses reappeared in CC/Devas and DT against the backdrop of the inconsistent and inconclusive legacy of the Argentinian cases. As the introduction foreshadowed, both cases concerned the same dispute: the cancellation of the agreement by which the Indian state-owned company, Antrix, leased the spectrum to Devas. In both cases, India argued that it was protected by the ESI clauses in the India–Mauritius and Germany–India BITs, as the agreement was cancelled to make the spectrum available for military and other strategic needs. Both tribunals spent substantial effort in interpreting and applying the respective ESI clauses, and ultimately reached divergent conclusions: CC/Devas agreed with India, while DT found in favour of the Claimant. The inconsistent interpretation and application of ESI clauses that explain these divergent outcomes are described below.
A. Facts Leading to the Cancellation of the Agreement
Antrix and Devas concluded their agreement in 2005. Beginning in that year, the Indian military made repeated demands for the spectrum. In 2009, in response to a call for a ‘consolidated proposal’ by the Indian space agency, the Indian Space Research Organisation (ISRO), the military presented a detailed projection of its spectrum requirements. At the time, ISRO expressed its ‘inability’ to meet the military’s demands and encouraged the military to ‘explore new avenues’.61
From 2009, several governmental departments scrutinised the agreement. Not just the military’s needs, but several other reasons, brought about this scrutiny. The Indian Department of Space (DoS) learned of possible irregularities in the agreement and set up a committee through the Department of Telecommunications (DoT) to review it. The committee found that the agreement imposed onerous penalties on Antrix for delays in satellite delivery, and noted that it did not include a clause that would allow the government to prioritise ‘strategic and other essential’ needs. Accordingly, the committee recommended that the agreement be revisited.62
In parallel, the Indian media’s scrutiny of the Antrix–Devas agreement led to ISRO admitting in a memorandum that the agreement did not allow ISRO to accommodate ‘strategic needs’ for the spectrum and denied other commercial enterprises a level playing field. Consequently, the ISRO queried, for the first time, whether the agreement should be annulled.63
These multiple reasons continued to inform calls from governmental departments to annul the agreement. In 2010, the DoS recommended annulment, identifying three reasons in support: (i) the need to preserve the spectrum for military and societal needs; (ii) concerns regarding certain provisions of the agreement; and (iii) denial of a level playing field to other commercial enterprises.64
Pursuant to the DoS’s recommendation, the ISRO sought a legal opinion from India’s Additional Solicitor-General (ASG) on terminating the agreement. The ASG opined that the decision to terminate should come from the government, as that would constitute a force majeure event under the agreement.65 The ISRO then asked the government’s approval to cancel the agreement. In its request, the ISRO shed its other concerns and cited only India’s military and strategic needs for the spectrum as the reason for cancelling the agreement.66
Ultimately, in 2011, the government approved the ISRO’s request and ordered the cancellation of the agreement. The government’s decision, too, did not list other concerns that had cumulatively led to a review of the agreement. Instead, it based the decision solely on the increased demand for allocating the spectrum for ‘national needs, including for the needs of defence, para-military forces, railways and other public utility services as well as for societal needs’.67 Yet, the government did not allocate the spectrum to the military until 2015. For several years after the decision, the military and the DoT made competing demands for the spectrum.68
B. Did the Dispute Involve an ‘Essential Security Interest’?
The Claimants in both CC/Devas and DT argued that a threat or a risk to a security interest was necessary for the security interest to qualify as an ESI. In CC/Devas, the Claimants contended that a security interest could be ‘essential’ only if the interest was: (i) ‘vital … absolutely necessary; extremely important’; and (ii) under a threat.69 Likewise, the DT Claimants stressed that an ESI could exist only if the interest was under an ‘imminent threat of severe consequences’.70
From a legal standpoint, the CC/Devas Tribunal initially agreed with the Claimants that ‘essential’ entails ‘important’, ‘absolutely necessary’, ‘indispensably requisite’ or ‘unavoidable’ security interests.71
At a later stage, the Tribunal diluted this standard and queried instead whether the Indian military was in ‘genuine need’ of the spectrum.72 Consequently, in its assessment of the facts, the Tribunal did not examine whether the military’s spectrum needs constituted indispensable or unavoidable security interests, nor did it inquire whether the military’s needs had become absolutely necessary in 2011 when the government annulled the agreement. Acknowledging that it must give a ‘wide measure of deference’ to India’s assessment of its ESI,73 the Tribunal noted simply that:
Even though there is nowhere in the CCS decision any specific reference to the Respondent’s ‘essential security interests,’ the Tribunal, by majority, has no difficulty concluding that the reservation of spectrum for the needs of defence and para-military forces can be classified as ‘directed to the protection of its essential security interests’.74
The DT Tribunal, by contrast, articulated a lower standard for establishing the existence of an ESI. Unlike that in CC/Devas, the DT Tribunal did not compare ‘essential’ interests to ‘absolutely necessary’ or ‘unavoidable’ interests,75 stating instead that ‘essential’ security interests are those that go to the ‘core (the “essence”) of state security’.76
On the facts, however, it is unclear whether the Tribunal applied this lower standard or effectively endorsed a standard of absolute necessity (as initially adopted by the CC/Devas Tribunal). On the one hand, unlike CC/Devas, the DT Tribunal did not assess whether the military’s spectrum needs were genuine. Instead, it summarily concluded that it would ‘of course accept that the so-called strategic needs’ of India’s military and para-military forces constituted ESI.77 On the other hand, the Tribunal doubted that the case involved a ‘true instance’ of ESI.78 First, the Tribunal noted that, in 2009, the ISRO asked the military to ‘explore new avenues’ and ‘best utilise’ the available spectrum, suggesting that military needs were not irreconcilable with the agreement.79 Secondly, the Tribunal reasoned that if an ESI did exist, the government would not engage in protracted debates about who the spectrum should be allotted to, after cancelling the agreement.80 Finally, the Tribunal found it hard to reconcile the fact that part of the spectrum was allotted to government-owned companies for commercial purposes in 2009 when the military had already stated its need for the spectrum, again suggesting the ‘absence’ of an ESI.81 This oscillation between the presence or absence of an ESI ultimately proved inconsequential, as the DT Tribunal found that there was no nexus between India’s military needs and the decision to cancel the agreement.
C. Was There a Measure Between the Measure and the ESI?
The nexus requirement differed in the applicable BITs. The CC/Devas Tribunal, constituted under the India–Mauritius BIT, had to determine whether India’s decision was ‘directed to’ the protection of its ESI.82 The DT Tribunal, applying the Germany–India BIT, had to assess whether the decision was ‘necessary’ for protecting the ESI.83
Both tribunals acknowledged the difference in the applicable nexus requirements. The CC/Devas Tribunal distinguished the two standards, but failed to articulate clearly what ‘directed to’ implied.84 It declined that ‘directed to’ required a State to demonstrate ‘necessity in the sense that the measure adopted was the only one it could resort to in the circumstances’.85 Neither was it convinced of the other extreme—that the standard had the practical effect of turning the ESI clause into a self-judging one.86 Ultimately, the CC/Devas Tribunal appeared to suggest that a measure would be ‘directed to’ the protection of an ESI if it ‘related to’ that objective.87
The DT Tribunal held, conversely, that ‘necessary’ implied that a measure must not simply be ‘related to’ an ESI.88 In the DT Tribunal's view, a measure would be necessary if: (i) it was ‘principally targeted’ at protecting the ESI and was ‘objectively required’ to achieve that protection; and (ii) the State could not avail itself of other reasonable alternatives that were less in conflict with its international obligations.89
Prima facie, these differences in the nexus requirements explain the divergent outcomes in the two cases. As the DT Tribunal emphasised:
[T]he BIT applicable in this case requires it to find that a measure was ‘necessary’ …and not merely ‘directed at the protection’ of such interests as was required under the treaty at issue in the Mauritius BIT Arbitration. In the Tribunal's view, the phrase ‘to the extent necessary’ implies a more stringent nexus between the measure at issue and the interests pursued.90
A deeper analysis of the Tribunal's reasoning reveals, however, that the divergences are attributable not to these legal distinctions but to different perceptions regarding the import of India’s decision to cancel the agreement, and the relevance of the facts preceding and following that ultimate decision.
Both tribunals were privy to the same three key facts. First, in the years leading to the government’s decision to cancel the agreement, several factors (including military and strategic needs, fears of a political scandal, onerous contractual clauses, and the lack of a level playing field for other commercial enterprises) were at play. Second, the 2011 decision to cancel the agreement was based only on military and other strategic needs. While the decision ordered the cancellation of the agreement, it did not simultaneously direct that the spectrum be allocated to the military. Finally, following the agreement’s cancellation, protracted debates regarding the spectrum’s allocation continued for several years between various governmental branches.
The majority in CC/Devas admitted that a ‘mix of factors’ had led to calls for annulling the agreement,91 and conceded that these events could help understand the ‘context’ in which the decision was reached.92 Yet, in its eventual analysis, the Tribunal majority considered only the ‘actual decision’ to be the ‘determinant factor’,93 explaining its reluctance to engage with facts other than the decision itself partly in political terms:
it is a regular phenomenon in public administration that decisions are influenced by a number of factors including, sometimes, purely political ones.94
Be that as it may, the Tribunal's approach of placing selective emphasis on the content of the decision creates the risk of States being able to escape their investment obligations by couching the actual measure in objective terms as one directed to the protection of an ESI, even when the actual object of the adopting the measure might well be different.95 In a case such as CC/Devas, it is unsurprising that the majority reasoning was met with a strong dissent, in which the majority was chided for restricting itself to the ‘formulaic’ decision of the Indian government, which did not accurately reflect the whole ‘substance’ or ‘purpose’ of the decision.96
Even on just the terms of the decision, the CC/Devas majority and dissent disagreed on the indicia for determining the purpose of the decision. For the majority, it was sufficient that India took the spectrum from Devas to meet, at least partly, its military needs.97 For the dissent, the taking of the spectrum from Devas needed to be matched by a giving of the spectrum to the military. The potential uses for the spectrum remained open for governmental deliberation at the time of the decision and several years after, indicating to the dissenting arbitrator that the decision was not contemporaneously ‘directed’ towards the protection of the military’s needs.98
This factual disagreement between the majority and dissenting members in CC/Devas regarding the ‘import and intended effect’ of the decision to cancel the agreement99 also formed the basis for the divergent outcomes in CC/Devas and DT. Like the dissenting arbitrator in CC/Devas, the DT Tribunal concluded that ‘as long as the choice among these potential usages … remained open’, India’s decision could not be held to be targeted towards protecting the military’s needs.100
Thus, the DT Tribunal's attempt to explain the divergent outcomes on the differing treaty standards of ‘directed’ and ‘necessary’ was simply a subterfuge for what was essentially its disagreement with CC/Devas regarding the factual existence of a causal link between the decision and the ESI. It was not a stricter nexus requirement that defeated India’s claim in DT. The history behind the decision, and the indeterminate status of spectrum allocation, were irrelevant facts for the CC/Devas majority. These exact facts proved to the DT Tribunal that the decision was not targeted at—or, synonymously, directed towards—addressing the military’s needs.
Having demonstrated the inconsistent approaches of the CC/Devas and DT tribunals, the next section will critically evaluate the contribution of these cases to the development of our understanding of the interpretation and application of ESI clauses.
IV. RE-EVALUATING THE INTERPRETATION AND APPLICATION OF ESI CLAUSES
Combined, the Argentinian cases, CC/Devas and DT, have left concrete, consistent guidance on two threshold issues. First, it is now settled that ESI clauses and CIL on necessity achieve different objectives and operate exclusively of each other. Second, unless expressly articulated, ESI clauses are not self-judging.101 Beyond these, a jurisprudence constante on the meaning of ESI and the nexus requirement remains elusive. The positive and negative implications of the developments generated by these decisions are discussed below.
A. Essential Security Interest, Military Needs and Protection of Strategic Resources
The term ‘essential security interests’, and its variations such as ‘national security’ and ‘public security’, appear in a wide range of investment and other treaties, but the term itself remains notoriously undefined.102 Conventional discussions on what constitutes an ESI have been limited to whether the term covers only military threats (such as war, armed attack or terrorism) or extends also to economic and other non-military crises.103 Article XXI(b) of the General Agreement on Tariffs and Trade (GATT) is a typical illustration of an ESI clause that confines ESI to actions taken for specific military or defence-related concerns.104 Outside the World Trade Organization context, there is increased judicial and academic acceptance that ESI can encompass non-military crises, including economic and environmental crises.105 As existing jurisprudence continues to tether ESI to notions of ‘crisis’,106 the central point of contention in non-military contexts concerns the level of intensity that a situation must reach to qualify as an ESI. In the Argentinian cases, this translated into disagreements regarding whether a State’s total collapse was necessary to prove the existence of an ESI, or whether a threat of collapse would suffice.
The factual matrix in CC/Devas and DT concerned neither of these scenarios. The dispute concerned the reacquisition of a strategic, limited resource. The interpretive challenge for the tribunals was, thus, whether the definition of an ESI could be further broadened to cover not just crisis situations emerging from military or non-military threats, but also situations of protecting a strategic resource when no foreseeable crisis exists. This analysis entailed two questions: first, can protection of a strategic resource constitute an ESI? If so, must the need for the resource relate only to military purposes, or can ESI protect resources that are necessary to meet other needs? Second, must there be an existing crisis in relation to the strategic resource in order to trigger ESI? The tribunals’ perfunctory analyses—accepting military, but not other needs, for the spectrum as ESI—do no justice to the importance of answering these questions clearly.
(i) Strategic resources and military needs
Since it first materialised in treaty texts, drafters have been careful to emphasise the exceptional nature of ESI clauses. It has often been reiterated that ESI clauses are to be used for ‘serious reasons’, that they are ‘not intended to be a loophole through which arbitrary actions would or could be taken so as to defeat the purpose of the Treaty’, and that their purpose is not to ‘create a basis for unduly prolonged departures from any provision of the Treaty’.107 The qualifier ‘essential’ was included to ensure that ESI clauses were not invoked in a ‘frivolous manner’, and to emphasise that only the ‘most important’, ‘serious’ or ‘vital’ security interests would be protected by the clause.108 Measures associated with the military, even when undertaken during a war, were not intended automatically to constitute ESI.109 During negotiations with the United States on an FCN treaty, Dutch negotiators queried whether ESI would cover a Dutch law that permitted military authorities to seize property during a war and defer payment of compensation until after that war. The US negotiators responded that while an ESI clause ‘would undoubtedly allow a certain amount of flexibility in the modalities of payment, in a war or emergency’, it ‘would not be regarded as derogating from the basic principle that compensation would be due and payable’.110
Ensuring that the military has the resources to defend a State’s sovereignty and territorial integrity against internal and external threats undeniably qualifies as a ‘security interest’.111 However, whether the availability of a resource constitutes an ‘essential’ security interest should depend on whether the resource is strategic, ie integral or necessary, for the military to perform its function of national defence.112 If resources integral to defence and warfare are not distinguished from non-integral resources, any measure undertaken for military supply could be justified on the ground of ESI. Such an expansive interpretation would threaten the balance that ESI clauses seek to achieve between upholding an investment treaty’s investment protection objectives and ensuring a State’s freedom to protect against security concerns.113 It would also conflict with the basic international legal principle that compensation must accompany expropriation.114 A comparative assessment of jurisprudence on ESI under article XXI of the GATT is instructive on this distinction.115 In 1975, Sweden introduced an import quota for certain varieties of shoes, arguing that the decrease in domestic shoe production risked the non-availability of shoes for military use in case of war, and thus threatened its ESI.116 Sweden revoked the quota, but not before many Contracting States raised doubts regarding whether the justification was covered by article XXI of the GATT.117 Commenting upon the case, Hahn similarly argues for a distinction between ‘key’ and other military goods.118 In Hahn’s view, ensuring the availability of shoes cannot be an ESI because ‘shoes are not likely to be a key factor in battlefields’.119 Likewise, Schloemann and Ohloff suggest that a resource must serve a defence purpose in order to qualify as ‘security-sensitive’.120
Second, whether the existence of a crisis is necessary for security interests to become essential is open to debate. An armed conflict inevitably evokes notions of crisis. In situations of non-military threats, proof of a crisis may be necessary to distinguish ordinary adversities from ones that have the potential to affect the very security of a State. Thus, in situations of economic adversity—as in the Argentinian cases—the existence and severity of an economic crisis become integral to proving an ESI to ensure that economic downturns that can be reversed by ordinary measures are kept outside the scope of an ESI clause.121 In other, wholly different situations, such as the availability of a limited, strategic resource, the existence of a crisis provides an improper constraint on the definition of the ESI. In such situations, it should suffice that the threat or risk associated with the failure to protect the resource is tangible and not speculative at the time of adopting the measure.122 Support for this view is found in the International Court of Justice’s (ICJ’s) ruling in Nicaragua. While defining ESI, the ICJ held that ‘the risk run by these “essential security interests” [must be] reasonable’.123 One of the measures challenged in Nicaragua concerned the United States’ imposition of trade embargoes upon Nicaragua in response to Nicaragua’s support of armed groups in its neighbouring countries. In holding that the embargo was not protected by the ESI clause in the US–Nicaragua FCN treaty, the ICJ appeared to emphasise the speculative nature of threat to the United States’ ESI at the time the embargo was imposed. The ICJ reasoned that ‘no evidence at all is available to show how Nicaraguan policies had in fact become a threat to “essential security interests” in May 1985’, especially because ‘those policies had been consistent, and consistently criticized by the United States, for four years previously’.124 Interpretive explanations of ESI provided by some States similarly articulate the need for a tangible risk. The United States, for instance, has repeatedly emphasised, albeit tautologically, that ESI includes ‘actions not arising from a state of war or national emergency’, if such actions ‘have a clear and direct relationship to the essential security interest of the Party involved’.125
As Section III.B demonstrated, it is this last factor that both the CC/Devas and DT tribunals endorsed to different degrees. Despite emphasising the unavoidable and indispensable nature of an ‘essential’ security interest (CC/Devas) and holding that ESI are those that go to the core of the security interest (DT), neither Tribunal examined whether ensuring effective communication is integral to the military’s operations. The ‘essential’ nature of the ‘security interest’ was presumed on that count as they were needs of the military. However, at least the CC/Devas Tribunal enquired whether the military’s needs were tangible, by assessing whether the military had a ‘genuine need’ for the spectrum.126 The DT Tribunal flagged concerns about the tangible nature of the military’s spectrum needs, suggesting the ‘absence’ of ESI in the case, but ultimately left unclear whether such an inquiry is material to proving the existence of an ESI.127
Should the government attest to the tangibility of the risk to the ESI at the time of adopting a measure? CC/Devas espoused a more lax approach. Approaching this issue from the military’s point of view, the Tribunal in CC/Devas found the military’s consistent and repeated demands for the spectrum to be concrete proof of a tangible need to protect military communications. By contrast, the Tribunal in DT viewed the issue from the government’s lens. According to it, India’s ESI appeared to be speculative, because despite the military’s sustained demands over a prolonged period, the government had not definitively agreed to allocate the spectrum to the military before, at the time, or immediately after it annulled Devas’s contract. Which of the two approaches is preferable depends not just on the facts of the case, but on more systemic concerns such as the extent to which a State’s assessment of its ESI should be subject to judicial review,128 and what the adequate balance between investment protection and State sovereignty ought to be. At the ICJ, there is a degree of agreement that the plausibility of a risk to an ESI is to be judged against the standard of reasonableness.129 Although reasonableness is a ‘notoriously ubiquitous and imprecise concept’ that is applied differently in different contexts and encompasses several distinct elements,130 it requires, first and foremost, that the ‘State’ must recognise the purpose behind enacting a measure.131 To that extent, the stricter scrutiny preferred by the Tribunal in DT—examining whether the government considered military communications to be under a tangible threat without the spectrum—appears more suited to establish whether there is a genuine ESI that needs protection.
(ii) Strategic resources and non-military needs
A more difficult interpretive question concerns the reach of ESI when a strategic resource is reacquired for non-military purposes. As previously described, India reacquired the spectrum to meet, not just its military’s needs, but also those of public utility services such as the railways and other societal needs such as tele-education, tele-health and rural communications.132 In a rare display of agreement, the CC/Devas and DT tribunals held that these non-military interests were issues of ‘public interest’ and could not constitute ESI ‘without distorting the natural meaning of such term’.133
Whether or not ESI can encapsulate the protection of strategic resources for non-military needs relates back to the charged issue of whether ‘security’ is a limited concept covering military concerns with the primary focus on protection of the State. Increasingly, ‘security’ discourse has been criticising attempts to understand security as a fixed concept, arguing instead that security is a multi-faceted and evolving concept that encompasses issues of human, political, military, socio-economic, environmental and energy security.134 For instance, Kurtz argues that ‘challenges to security are now multi-faceted and encompass events often far beyond State control, including risks of external pollution, terrorist attacks and water shortages. These changing risks require a new paradigm of “human security”’.135 States, too, recognise that ‘security’ has a broad reach.136 The United Kingdom, for instance, has noted that ‘[o]ver recent decades, our view of national security has broadened to include threats to individual citizens and to our way of life’.137
Under this broader outlook, protection of strategic or critical industries and infrastructure increasingly constitutes an important element of national security.138 Critical industries and infrastructure are variously defined by national governments, but include, in essence, industries and infrastructure that are essential for the economic and social well-being of a country and its citizens.139 Public utilities, such as energy, water, transport and communications, are recognised, near universally, as critical industries and infrastructure.140 Already, countries place ex ante restrictions on foreign investment in critical industries on the ground of national security.141
This evolving and dynamic understanding of security suggests that policies aimed at ensuring critical infrastructure security can be covered by a broad interpretation of ESI. Measures adopted to protect a resource vital to infrastructure security could, by extension, constitute an ESI under this analytical framework. Such measures could include forced reacquisition of resources, in case a host State concludes that a resource not previously considered security-sensitive, has acquired such a characteristic after an investment has been made.142 Needless to say, the threshold for such an interest to qualify as a ‘security interest’ and, more importantly, an ‘essential’ security interest remains high. On this important point, decisions of the European Court of Justice (ECJ), now the Court of Justice of the European Union (CJEU) might provide a lodestar for investment tribunals. The Treaty Establishing the European Community (EC Treaty) enlists ‘public security’ as a ground for deviating from treaty obligations relating to freedom of establishment and free movement of persons, services and capital.143 The ECJ has repeatedly held that ‘public security’ includes the protection of critical resources. In Campus Oil v Minister for Industry and Energy, the ECJ was asked to decide whether an Irish rule that required importers of petroleum products to purchase a proportion of their requirement from a State-owned refinery was justified on grounds of public security. The ECJ held that:
petroleum products, because of their exceptional importance as an energy source in the modern economy, are of fundamental importance for a country’s existence since not only its economy but above all its institutions, its essential public services and even the survival of its inhabitants depend upon them. An interruption of supplies of petroleum products, with the resultant dangers for the country’s existence, could therefore seriously affect the public security.144
Particularly in relation to sectors of public utility, the ECJ has maintained the view that the aim of safeguarding supply of public utilities is covered by the concept of public security.145 What emerges from the ECJ’s jurisprudence is that first, public security can include measures taken to protect critical infrastructure such as public utilities. Second, only those resources that are vital or of ‘exceptional importance’ for uninterrupted supply of public utilities can constitute public security. Finally, on whether public security covers only measures taken in the face of a crisis, the ECJ’s decisions make clear that an impending crisis is not a prerequisite to adopt measures to protect critical infrastructure and resources. As the decisions suggest, the intention behind adopting a measure to protect a resource/infrastructure must not be purely economic.146 While identifying whether the protection of a resource/infrastructure is a matter of public security, regard must be had to the effect that the non-availability of that resource or infrastructure would have on the country’s existence or to a ‘fundamental interest of society’.147 A measure is justified by public security if it is intended to ensure that there would be a minimum supply of a resource or a utility ‘in the event of a genuine or sufficiently serious threat’.148 To this extent, the precautionary character of protecting critical resources/infrastructure diverges from the reactive nature of measures taken to combat military or non-military crises.149
A caveat should be entered at this point. The EC Treaty refers to ‘public security’ and not to ESI. The ‘public security’ formulation might suggest that the term covers a broader range of scenarios.150 As discussed in this section, the different formulations should not exclude critical infrastructure and resource security from the scope of ESI in principle, but can have an impact on the situations in which a threat to a critical resource/infrastructure become an ESI. In that regard, guidance is available from recent investment treaties concluded by some South Asian States. The ESI clauses in these treaties specifically identify the protection of ‘critical public infrastructure, including communication, power and water infrastructures’ as an ESI, but limit its scope to protection from ‘deliberate attempts intended to disable or degrade such infrastructure’.151 A precautionary measure intended to prevent the consequences of a future threat (as in the ECJ’s case) or measures taken to develop a critical infrastructure, might be viewed as a regulatory or a strategic measure but would not constitute ESI under the stricter standard articulated in these treaties.152
The foregoing shows that protection of critical or strategic resources/infrastructure can constitute ESI. Which kinds of threat to a critical resource/infrastructure can qualify as ESI is a subject of ongoing debate. In CC/Devas and DT, the tribunals are right to suggest that societal needs, such as tele-education and tele-health, are matters of public interest: the taking of property for such reasons should be regulated by rules on expropriation. However, needs of the railways and other public utility services—which India had itself identified as ‘national’ and not ‘societal’ needs—invited discussion on the scope of ESI clauses in relation to protection of critical infrastructure/resources.153 On the facts, the case may not have met the strict threshold for classifying the protection of critical infrastructure/resources as ESI. However, the tribunals’ dismissal of measures taken for non-military needs as ones taken in the ‘public interest’ deviates not only from now established jurisprudence that non-military concerns are covered by ESI, but also presents a missed opportunity for analysing the breadth of ESI clauses in relation to protection of critical resources/infrastructure.
B. Existence of an ESI and Deferential Standard of Review
As previously adumbrated, investment tribunals agree that, in the absence of clear language, ESI-based exceptions are justiciable. Whether an ESI exists is not left exclusively to a State’s subjective judgment. Quite apart from the question of State discretion and non-justiciability, is the issue of deference in judicial review of the existence of an ESI.154
A deferential standard of judicial review, or the granting of a ‘margin of appreciation’ to a State’s assessment of its position,155 is widely recognised as a product of the European Court of Human Rights (ECtHR).156 The ECtHR routinely deploys the margin of appreciation doctrine in assessing justifications based on ‘national security’ and other objectives such as ‘public emergency’ and ‘protection of health or morals’.157 As the ECtHR held in CG and others v Bulgaria, States enjoy a ‘large margin of appreciation’ in defining ‘what is in the interests of that [national] security’.158 The rationale for according such deference, as the ECtHR famously explained in Ireland v UK, is that:
by reason of their direct and continuous contact with the pressing needs of the moment, the national authorities are in principle in a better position than the international judge to decide … on the presence of such an emergency.159
Investment tribunals have been slow to acknowledge a State’s definitional prerogative on matters of ESI. The early tribunals of CMS, LG&E, Sempra and Enron did not pronounce on the appropriate standard of review, and proceeded to conduct a full-scale ‘substantive’ review of whether Argentina’s economic crisis constituted an ESI.160 Support for deference came first from Burke-White and von Staden, who relied on the ECtHR’s jurisprudence to propose a similar margin of appreciation to States in identifying their ESI.161 In the same year that Burke-White and von Staden published their work, margin of appreciation finally found its way into investment treaty jurisprudence when the Continental Tribunal similarly drew from the ECtHR’s template to declare that ‘this objective assessment [of what is an ESI] must contain a significant margin of appreciation for the State’.162
The CC/Devas and DT decisions have further entrenched margin of appreciation into investment treaty jurisprudence on ESI. Both tribunals concurred that States are owed a degree of deference on the question of the existence of an ESI.163
Investment law’s evolution towards deferring to a State’s identification of its ESI is a step in the right direction from historical, comparative and practical perspectives. Historically, negotiations of FCN treaties between the United States and its treaty partners indicates that although ESI clauses were meant to be justiciable, States shared a belief that tribunals would accord weight to a State’s assessment on matters of security.164 Notably, in negotiations between Germany and the United States, when German negotiators queried whether ESI clauses were justiciable, the United States responded by clarifying that such clauses were in fact justiciable but speculated that ‘international tribunals would probably give very heavy weight to arguments presented by the government invoking the reservation’.165
From a comparative perspective, other international tribunals have come down in favour of a deferential review of security-based and other objectives. The ECJ has toed a line similar to that of the ECtHR. While holding that States cannot unilaterally determine the scope of public security,166 the ECJ has nonetheless recognised that States enjoy ‘a certain degree of discretion’ on such matters.167 As Advocate Jacobs has explained, the importance of security as a legitimate aim, and a diversified understanding of what might constitute a security interest for a particular State, are important for deferring to a State’s appraisal on issues of national security.168
The ICJ has not displayed the same degree of enthusiasm towards adopting a deferential standard of review on other matters, but on the interpretation of ESI the ICJ, too, has preferred a deferential approach. In the Oil Platforms case, the ICJ rejected a deferential approach on questions of self-defence, holding instead that ‘the requirement of international law that measures taken avowedly in self-defence must have been necessary for that purpose is strict and objective, leaving no room for any “measure of discretion”’.169 In the Gab?íkovo-Nagymaros Project case, while assessing the necessity of Hungary’s suspension of works on the project under CIL, the ICJ similarly held that it must establish ‘the objective existence of a “peril”’.170 Bjorge, relying upon the ICJ’s preference for an objective analysis, has contended that the ‘margin of appreciation is decidedly old hat’ and has been ‘reduced to a vanishing point’ in international law.171
The ICJ’s inclination towards an objective standard of review in these cases is distinguishable, and in no way suggests a blanket rejection of deferential review by the Court. In Gab?íkovo-Nagymaros Project, the ICJ’s choice of objective review can be explained by the fact that the CIL defence of necessity applies only ‘on an exceptional basis’.172 More importantly, in Oil Platforms, the ICJ addressed only the narrow question of the standard of review applicable to the law of self-defence. It decidedly refrained from commenting upon the standard for examining whether a State measure was exempted under an ESI clause. In that case, the United States had contended that its military attacks on Iranian offshore oil installations were necessary to protect its ESI. In support of its contention, the United States asserted that ‘[a] measure of discretion should be afforded to a party’s good faith application of measures to protect its essential security interests’.173 As the ICJ had already decided to examine the issue under the law on self-defence, it chose against providing its opinion on the standard of review applicable to ESI clauses by stating that ‘[t]he Court does not have to decide whether the United States interpretation … on this point is correct’.174
The ICJ’s approach towards reviewing the existence of an ESI is more accurately found in the Nicaragua case. In that case, as previously stated, the ICJ adopted a standard of reasonableness to assess the risk run by ESIs.175 As Judge Koojimans notes in his separate opinion in Oil Platforms, this language suggests an espousal of deferential review:
With regard to the assessment of the risk run by the essential security interests, the term ‘reasonableness’ is used; with regard to the ‘measures taken’, the Court states that it is not sufficient that they may be deemed ‘useful’ but that they must be necessary. This seems to indicate that with regard to the measures taken a stricter test must be used than with regard to the assessment that essential security interests are at risk. There seem to be good reasons for such a distinction with regard to the margin of discretion to be left to governmental authorities. The evaluation of what essential security interests are and whether they are in jeopardy is first and foremost a political question and can hardly be replaced by a judicial assessment. Only when the political evaluation is patently unreasonable … is a judicial ban appropriate.176
Practically, too, a deferential review is a sound way to avoid adding fuel to the fire that is the legitimacy crisis of investment law. As Schloemann and Ohlhoff explain, concepts such as ESI are a function of ‘contemporary sovereignty’ and demand ‘individualization, or individual definition, by the State concerned’.177 The impossibility of developing a shared understanding of what constitutes an ESI for a State attests to a challenge that a judicial tribunal might face in undertaking a fully fledged review of this question. Certainly, doing so runs the risk of an investment tribunal being perceived as ‘substituting the functions of a sovereign State’ and sitting in appeal over a State’s policy decisions—matters that go to the heart of investment law’s legitimacy crisis. Deference thus strikes an appropriate balance between judicial review and a State’s prerogative to define its security concerns.
A side consequence of affording deference to States on matters of ESI is that tribunals will be less likely to base their findings that an ESI clause is non-applicable on the question of whether an ESI exists, if they can reach that same conclusion by establishing the lack of a nexus between a State measure and the ESI. As Akande and Williams explain, international tribunals consider the issue of the existence of an ESI a mere ‘formality’, as it would be ‘exceedingly rare for an international tribunal to find that a particular interest considered by a State to be a national or essential security interest may not actually be classified as such’.178 Certainly, the DT Tribunal's reluctance to declare the absence of an ESI validates this observation. As previously described, the significant time lag between the Indian military’s demands for the spectrum and the Indian government’s ultimate allocation of that spectrum to the military, raised justifiable concerns regarding the existence of an ESI for the DT Tribunal. Ultimately, however, the DT Tribunal shied away from pronouncing upon this more political matter, as it could base its decision on India’s failure to establish that it was necessary to cancel Devas’s agreement.
C. ESI Clauses and the Nexus Requirement
Prior to the DT case, investment tribunals had adopted three different interpretations of the necessity requirement in ESI clauses: (i) the tribunals in CME, Enron and Sempra applied the framework of necessity under CIL to suggest that a measure is ‘necessary’ under an ESI clause if it is the ‘only way’ for the State to protect its ESI; (ii) the Tribunal in LG&E proposed that a necessary measure is one that is ‘legitimate’ and undertaken in a situation when a State has no choice but to act; and (iii) the Tribunal in Continental borrowed from WTO jurisprudence a two-step test of examining whether the measure ‘contributed materially’ to the protection of the ESI and whether the State could deploy a less restrictive alternative.
The latter approach of Continental requires further examination to assess the developments generated by the DT ruling. WTO jurisprudence on necessity is derived from interpretations of the standard in the general exception clause in Article XX of the GATT and Article XIV of the GATS. WTO law adopts a proportionality-based test comprising two stages. The first stage entails a holistic ‘weighing and balancing’ exercise of several factors, such as the importance of the interest at stake, the measure’s contribution to the interest, and the restrictive impact of the measure on international trade.179 If this analysis yields a preliminary conclusion that the measure is necessary, the second stage confirms the necessity of the measure by comparing it with other less trade-restrictive measures which a State can reasonably adopt to achieve the same goal (the LRM test).180
Contrary to certain academic opinion,181Continental’s two-step test does not fully mirror that of the WTO. Unlike WTO law, Continental did not propagate a proportionality-based ‘weighing or balance’ exercise. Furthermore, Continental’s assessment of material contribution requires proof that the measure was ‘apt to and did make … a material or a decisive contribution’ towards the achievement of the objective.182 The WTO does not require this stricter proof of actual contribution. Under WTO jurisprudence, a measure can be necessary even if its benefits are not ‘immediately observable’ so long as the measure is likely or ‘apt to produce a material contribution’.183 Ultimately, then, Continental cherry-picked the LRM test from WTO law and incorporated it into investment jurisprudence.
The chequered development of necessity in investment law placed the DT Tribunal in the unique position of charting the future path. Yet, without engaging with previous investment jurisprudence or with WTO law, the DT Tribunal proposed a two-pronged test involving an analysis of whether the measure was ‘principally targeted’ at protecting the ESI and whether the State could achieve the same outcome through less restrictive reasonable alternatives. The similarity in the tests adopted by the tribunals in Continental and DT are immediately apparent. However, DT does not toe entirely the same line as Continental: linguistic and standard of review-related variations are discernible.
Linguistically, the Tribunal in DT has discarded Continental’s analysis of material contribution in favour of examining whether the measure was principally targeted at the ESI. While the Tribunal did not clarify what it meant by ‘principally targeted’, one senses a similarity with the WTO standard of aptness under which a measure’s potential to contribute to the objective, and not its actual contribution is put to objective scrutiny. If DT has indeed favoured this more lax standard of assessment, it is a welcome refinement of Continental’s material contribution test. It acknowledges the temporal gap that might exist between the adoption of a measure and the manifestation of its results. As the WTO Appellate Body has explained, the results of certain measures ‘can only be evaluated with the benefit of time’.184
Shunning evidence of actual contribution also pays heed to the fact that certain government measures may be just one part of a ‘multiplicity of interacting measures’ that are designed to collectively achieve a policy objective.185 The State or the government is not a single entity. The execution of a measure or policy might require the involvement or participation of various government departments, each with their own mandate. In such situations, the isolated contribution of one government measure may be difficult to establish without accounting for the network of other contributing measures. Shaping necessity examination in terms of the principal target of the measure can allow tribunals to assess the likely contribution of a distinct measure in light of the suite of interrelated measures. By itself, the measure might not contribute towards the objective, but it can still be necessary if it can be proved that the measure forms part of a broader initiative such that there exists a genuine relationship of ends and means between the stated objective and the impugned measure. This analytical framework can explain DT’s factual analysis. To conclude that it was not necessary for India to cancel the agreement with Devas, the Tribunal in DT drew primarily upon the fact that no comprehensive policy was in place to allocate the spectrum to the military at the time that the agreement was cancelled. The government department responsible for terminating the agreement had done so when other ‘organs of the Government’ responsible for spectrum allocation had not agreed to allocate the spectrum to the military.186 India’s claim that there was a genuine relationship of ends and means between the military’s need for the spectrum and the standalone decision to cancel the agreement would have likely passed muster, were the government departments in agreement regarding the proper use of the spectrum. The taking of the spectrum from Devas, in that case, would form part of a broader established policy to make the spectrum available to the military.
From a standard of review perspective, Continental and DT differ in the rigour with which to scrutinise the availability of other reasonable less-restrictive alternatives. Having drawn from WTO law, the Tribunal in Continental implemented the LRM test by engaging in an extensive review of whether the suggested less-restrictive alternatives could achieve the same level of benefit as Argentina sought to achieve through the imposed measure.187 In contrast, the Tribunal in DT did not articulate the relevant standard for assessing when it would be reasonable to conclude that States could avail themselves of alternative measures. In DT, the Claimants proposed that India could have alternatively satisfied the military’s needs by acquiring spectrum previously allocated to other governmental departments.188 India argued in response that the alternative quantity of spectrum in the hands of other governmental departments would not be sufficient to meet the military’s needs.189 The Tribunal did not examine whether this proposed alternative would achieve India’s stated objective to the same level, concluding summarily that India had not considered ‘these reasonable, least restrictive alternative measures, although they were clearly available’.190
ESI clauses being exceptions to investment obligations, the burden of demonstrating the necessity of a measure falls upon the respondent State. However, the State need only show that the alternatives suggested by the claimant would not be equally effective in addressing the objective.191 This is not peculiar to assessments of necessity under ESI clauses. The relative effectiveness of alternative measures is also integral to establishing whether a measure was the ‘only way’ to safeguard against an essential interest under the CIL defence of necessity.192 The DT Tribunal's failure to engage with India’s argument that the proposed alternative was ineffective, insofar as it would not provide the same level of benefit (as the impugned measure), sets a dangerous precedent for future tribunals. This view is best treated as obiter dictum, as the Tribunal in DT had already determined that the measure failed to satisfy the first prong of the necessity test—namely whether the measure was principally targeted at the ESI.
Two broader points remain to be made. First, incorporation of the WTO’s LRM test into investment law has faced academic criticism.193 The different textual make-ups of ESI clauses in investment law and the general exceptions clauses in the GATT and GATS are key to this objection against convergence with WTO law. Notably, the general exception clauses in the GATT and GATS include a protective chapeau that excludes measures that are applied in a manner constituting ‘arbitrary or unjustifiable discrimination’. Critics have argued that the additional checks offered by the chapeau allow the WTO to adopt a more flexible interpretation of the necessity standard.194 Others have convincingly refuted this criticism as being overstated. Distinguishing ‘regulatory design’ from ‘regulatory application’, Kurtz helpfully explains that the chapeau performs the different function of testing the application of the measure, whereas necessity tests the design of a measure under WTO and investment law.195 Elsewhere, Mitchell and Henckels demonstrate that the same LRM test applies to necessity even under WTO agreements that do not contain a chapeau similar to that of the GATT and GATS.196 More significantly, a cross-regime comparison demonstrates that the LRM test is not unique to the WTO. Other international courts including the CJEU and the ECtHR similarly apply the LRM test as a second-level analysis to determine the necessity of a measure.197
Second, the attempt to harmonise the standard of necessity under investment law with that of other areas of international law is writ large in the recent formulations of Continental and DT. Yet, the tribunals have treaded cautiously, transplanting selectively the LRM test while discarding any relevance of a fully fledged proportionality analysis.198 This cherry-picking exercise is not without merit.199 It signals institutional awareness of the constraints within which investment tribunals operate. Proportionality analysis demands significant intrusion into a State’s policy determinations and requires ‘judges to behave as legislators do, or to sit in judgment on a prior act of balancing performed by elected officials’.200 Being far removed from the socio-political context in which governments make such determinations, investment tribunals lack the familiarity or the expertise to make the value judgments required under proportionality analysis.201 Indeed, this constraint has not prevented other international tribunals from engaging in a proportionality review. However, institutional specificities explain why proportionality has stronger footing in those other tribunals. The CJEU, for instance, is embedded in the European regional setting. Its judicial authority to apply proportionality analysis derives from the constitutional functions it performs and the goal of economic integration that it pursues.202 Investment tribunals, by contrast, do not operate in a particular supra-national constitutional order; their environment consists of a highly fragmented network of investment treaties. At the WTO, proportionality weighing, although the subject of criticism,203 is still less problematic because panel decisions are subject to appellate review. No such appellate check is available to curb unduly activist applications of proportionality analysis by investment tribunals.
(ii) ‘Directed to’ versus ‘necessary’
Until CC/Devas and DT, there was little academic or judicial clarity on the difference between the nexus requirements of ‘directed to’ and ‘necessary’. Broad agreement existed that necessity imposes a more stringent nexus requirement.204 The ICJ had emphatically stated in its ruling in Nicaragua that measures had to be more than ‘merely useful’ to qualify as necessary.205 The precise boundaries of distinction however remained unclear. Some academics adopted an intention-based test, arguing that ‘directed to’ requires proof that the State adopted the impugned measure with the intention to further the stated objective.206 Others proposed that any tenuous link between the measure and the objective would suffice. Only ‘extreme cases’ would fail to meet the ‘directed to’ requirement, as the test brings ESI clauses ‘very close to a self-judging clause’.207
CC/Devas has dismissed this latter proposal, and confirmed that tribunals must objectively assess the nexus between the measure and the objective even under the ‘directed to’ standard.208 In the remainder of its analysis, however, the tribunal appears to have adopted two divergent views of the test. On the one hand, the tribunal indicated that necessity’s stringency lay in the second-level analysis of less restrictive alternatives:
[Under the ‘directed to’ standard], the Respondent does not have to demonstrate necessity in the sense that the measure adopted was the only one it could resort to in the circumstances.209
This (first) outlook suggests that difference between ‘directed to’ and ‘necessary’ lies in the fact that ‘directed to’ requires only a first-level necessity analysis of whether the measure is ‘principally targeted’ at the objective, or has the potential to contribute materially to the objective. If this analysis yields a positive result, the measure will be found ‘directed to’ the objective even if the State has other alternatives at its disposal.
On the other hand, the tribunal subjected the facts to a significantly more lax scrutiny. As previously recalled, India’s decision to cancel Devas’s agreement contributed only tangentially to the fulfilment of its military’s spectrum needs because, at the time of the decision, the various government departments had not reached agreement on transferring the spectrum to the military. Despite the fact that the taking of the spectrum would not be matched by a giving of the spectrum to the military, the tenuous link was sufficient for the Tribunal to conclude that the measure was directed to the protection of India’s ESI. This (second) outlook places ‘directed to’ and ‘necessary’ at two ends of the nexus spectrum. It suggests that a measure need only be capable of providing the most incidental bare-minimum contribution to meet the ‘directed to’ standard. It further suggests that time is of no relevance in examining the ‘directed to’ nexus. Measures that have no immediate potential to contribute to the objective would nevertheless pass the ‘directed to’ test, so long as they are undertaken with the intention to contribute to that objective.
Between the two, it is suggested that the first outlook is preferable. The deference afforded to States under the second outlook would bring such ESI clauses dangerously close to being self-judging clauses, even though that was rightly not the (initial) intention of the Tribunal in CC/Devas.210 Admittedly, differences in nexus requirements under ESI clauses reflect differences in risk allocation between foreign investors and States.211 However, the second outlook gives significantly greater protection to a State’s regulatory freedom to adopt measures at a time when they have no potential to protect the ESI, skewing the balance unfavourably against foreign investment protection. Such an overly permissive approach could pave the way for opportunistic abuse of ESI clauses by host States.
The first outlook, by contrast, allows investment protection to be balanced with State sovereignty, while giving States greater regulatory freedom to invoke ESI clauses than available under the ‘necessary’ standard, thereby preserving the latitude States intended to offer themselves by framing the nexus requirement in linguistically different terms. More importantly, it better delineates the boundaries between the nexus requirements in ESI clauses. ‘Necessary’ and ‘for’ form the two ends of the nexus spectrum. ‘Necessary’ requires the measure’s potential for contribution to be additionally tested against the benchmark of the LRM test. ‘For’, as some academics have argued, requires a ‘relatively thin nexus’ under which measures need ‘merely further a permissible objective’.212 ‘Directed to’ lies between these extremes, but the second outlook risks conflating the standards of ‘directed to’ and ‘for’. Furthermore, the first outlook also keeps in check the risks inherent in a broad intention-based analysis. The facts of CC/Devas and DT have made it amply clear that, in certain situations, despite a State’s intention to use a measure to further a stated objective, the measure may not have the ability to contribute to that objective. These are situations wherein a measure is designed to further an objective, but the probability of its success is contingent on subsequent governmental initiatives. Given that the measure’s success is being stalled by a wing of the government, the genuineness of the relationship of means and ends between the measure and the objective is suspect until such time as there is a meeting of minds between the government departments. Seeking evidence of potential contribution akin to the first test under the necessity analysis reduces the risk of governmental overreach.
Criticism of the investment treaty regime continues to rise. Significant reforms of the regime are already under way. Inconsistent decisions on the same or similar disputes provide important fuel for these calls for recalibration, and the need for appellate review. While consistency in case law is not a virtue in itself and can impede the development of the law, the reasons for contradictory reasons and outcomes need to be adequately articulated to allow for a coherent legal system to develop. The recent decisions against India in the CC/Devas and DT cases are but another example of the very real problem of inconsistent reasoning and outcomes in investment treaty arbitration. Both tribunals were asked to resolve the same dispute, albeit under different investment treaties. As this article has demonstrated, the ultimate divergence in outcomes on whether India’s actions were exempt under the ESI clauses of the respective treaties had little to do with the differences in the formulation of those clauses. The tribunals had markedly different takes on the same facts. This was the case not only in relation to the import of India’s decision to cancel Devas’s agreement. Even on the existence of an ESI, the tribunals did not see eye to eye. This article has not taken issue with the outcome of the later decision in DT. Rather, it is the process of reasoning that raises concerns of inconsistency.213 The later Tribunal, in DT, had the award of the previous Tribunal, in CC/Devas, on record. Yet, the Tribunal in DT made little attempt to engage with the reasons presented in that previous award, citing the CC/Devas award only once in its examination of the applicability of the ESI clause. Better engagement with the previous decision could have explained the need for the contradictory outlook, thereby moulding the law in a more coherent fashion.
Paucity of engagement with previous jurisprudence on ESI clauses, especially the Argentinian cases, is a broader problem that plagues the CC/Devas and DT decisions. Indeed, investment tribunals are not bound by the doctrine of precedent. Nevertheless, tribunals themselves have frequently subscribed to the virtues of developing a jurisprudence constante. In line with this view, tribunals have opined that they ‘may pay due consideration to earlier decisions’ and have even suggested that they have not just ‘a duty to adopt solutions established in a consistent line of cases’, but also a ‘duty to seek to contribute to the harmonious development of investment law’.214 The inclusion of exception clauses, such as ESI clauses, in investment treaties are important tools for the creation of a sustainable investment law regime. They make it possible to reconcile private objectives of investment protection with a State’s public non-investment regulatory objectives. But the continued emergence of conflicting jurisprudence on ESI clauses, and the failure to engage with previous cases, impede significantly the ability of such clauses to fulfil their intended objective.
As this article has shown, there are few aspects relating to ESI clauses on which tribunals concur. A significant—and commendable—recent development on this front is a shared belief that tribunals must show a degree of deference to a State’s assessment of its ESI. However, a shared understanding of key elements of ESI clauses continues to elude the international investment community.
On the existence of ESI, a methodological flaw was evident in the reasoning of the tribunals in CC/Devas and DT . Both tribunals presumed, without explanation, that protection of critical infrastructure cannot constitute an ESI. More importantly, neither Tribunal queried whether the military’s need for resources automatically constitutes ‘essential’ security interests. This missed opportunity is all the more significant in the current global environment. The WTO is soon to grapple with a similar dilemma in the context of tariffs that the United States has imposed on aluminium imports. As justification, President Trump has invoked the ESI clause in Article XXI of the GATT, arguing that the tariffs are necessary to ensure the sustainability of domestic aluminium industries and to alleviate the ‘risk of becoming completely reliant on foreign producers of high-purity aluminium that is essential for key military … systems’.215
Equally disturbing is the continued lack of clarity on the required nexus between a State measure and its ESI. When elucidating the ‘necessary’ standard, for instance, the Tribunal in DT espoused a two-step test without citing Continental at all. Important differences are visible, as a result, in the content of the two-step test adopted in Continental and DT.
Not just the backlash against the investment treaty regime but the increased inclination among States to retreat towards nationalism have created a pressing need to identify clearly the extent of a State’s regulatory space. Quality reasoning and a robust interpretation of ESI clauses can go a long away towards providing that desired clarity. Given the challenges that tribunals have faced in interpreting and applying ESI clauses, future tribunals might benefit from requesting amicus curiae observations on the scope of ESI clauses from other (non-disputing) States parties to the investment treaty.216 To spur clarity in the interpretation of ESI clauses, States should also be exhorted to provide interpretive guidance, such as by way of interpretive notes, on what they intended the scope of ESI clauses to be.
The author thanks Dr Clair Gammage, Prof Eirik Bjorge Dr Martins Paparinskis, Dr Isabelle Van Damme and Jagdish John Menezes for their helpful suggestions and comments.